Noun Phrases in Creole Languages: A multi-faceted approach (Creole Language Library)

Noun Phrases in Creole Languages

Creole Language Library (CLL) A book series presenting descriptive and theoretical studies designed to add significantly to the data available on pidgin and creole languages.

Editors Miriam Meyerhoff

University of Edinburgh

Umberto Ansaldo

University of Amsterdam

Editorial Advisory Board Mervyn C. Alleyne

Pieter Muysken

Marlyse Baptista

Peter Mühlhäusler

Nicholas Faraclas

Shobha Satyanath

George L. Huttar

John Victor Singler

John Holm

Norval Smith

Sylvia Kouwenberg

Sarah G. Thomason

Susanne Michaelis

Tonjes Veenstra

Kingston, Jamaica

Athens, Georgia, USA Rio Piedras, Puerto Rico Dallas

Coimbra

Kingston, Jamaica Leipzig

Nijmegen Adelaide

University of Delhi New York

Amsterdam Ann Arbor Berlin

Salikoko S. Mufwene Chicago

Volume 31 Noun Phrases in Creole Languages. A multi-faceted approach Edited by Marlyse Baptista and Jacqueline Guéron

Noun Phrases in Creole Languages A multi-faceted approach

Edited by

Marlyse Baptista University of Georgia, Athens

Jacqueline Guéron University of Paris III

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Noun Phrases in Creole Languages. A multi-faceted approach / edited by Marlyse Baptista and Jacqueline Guéron. p. cm. (Creole Language Library, issn 0920-9026 ; v. 31) Includes bibliographical references and index. 1. Creole dialects--Noun phrase. 2. Creole dialects--Syntax. I. Baptista, Marlyse. II. Gueron, Jacqueline. PM7831.B37   2007 417'.22--dc22 isbn 978 90 272 5253 1 (Hb; alk. paper)

2007010196

© 2007 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Dedication For Roger, Zolan and Anton For Maurice, Claire and Sophie

Table of contents Acknowledgment



Part I An introduction



.



Noun phrases in creole languages: An introductory overview Marlyse Baptista and Jacqueline Guéron

Part II Portuguese-lexified Creoles



.

Bare nouns and the nominal domain in Santome Nélia Alexandre and Tjerk Hagemeijer



.

On the syntax and semantics of DP in Cape Verdean Creole Marlyse Baptista



.

Papiamentu and Brazilian Portuguese: A comparative study of bare nominals Ellen-Petra Kester and Cristina Schmitt



.

On the interpretation of bare noun phrases in Guinea-Bissau Portuguese Creole (Kriyol) Alain Kihm



Part III Spanish-lexified Creoles .

Some aspects of NPs in Mindanao Chabacano: Structural and historical considerations Anthony P. Grant

.

Bare nouns in Palenquero: A fresh consensus in the making Armin Schwegler

  

Part IV French-lexified Creoles



.



Bare nouns in Réunionnais Creole Robert Chaudenson

 Table of contents

.

The bare NP in Lesser Antillean Karl Gadelii

. Bare NPs and deficient DPs in Haitian Creole and French: From morphosyntax to referent construal Anne Zribi-Hertz and Herby Glaude Part V A Comparison of romance Creoles . Implicit determination and plural Viviane Déprez Part VI English-lexified Creoles





 



. Bare nouns and articles in Sranan Adrienne Bruyn



. Aspects of the syntax and semantics of bare nouns in Jamaican Creole Michele M. Stewart



Part VII African American English



. NPs in aspectual Be constructions in African American English Lisa Green



. Bare nouns in African American English (AAE) Arthur K. Spears



Part VIII Dutch-lexified Creoles . Bare nouns in Berbice Dutch Creole Silvia Kouwenberg Part IX A Synthesis and a postface

 



. Properties of noun phrases in creole languages: A synthetic comparative exposition Marlyse Baptista



. Functional deficiency, ellipsis or innovation in creole languages? A postface Marlyse Baptista and Jacqueline Guéron



Subject index



Acknowledgment The seeds to this volume were planted during a symposium that took place at the University of Paris III, Sorbonne Nouvelle in April 2000. The original participants to the symposium were Marlyse Baptista, Viviane Déprez, Karl Gadelii, Alain Kihm, Anne Zribi-Hertz and Herby Glaude. In 2002, we launched this volume with the final list of contributors representative of 15 different creole languages. We are thankful to all the authors for responding to and for tolerating gracefully our numerous queries and editorial demands. Thanks to their patience and willingness to explore with us the nominal domain of creole languages, we feel that this volume has achieved a more in-depth and accurate picture of creole nominal systems. We thank an anonymous reviewer for valuable suggestions that allowed us to clarify a number of theoretical and descriptive issues. We also thank three research assistants at the University of Georgia for their help in proof-reading the manuscript at various stages, Elise Amacker, Julia Tigner and Betsy Wang. We also wish to acknowledge Kees Vaes and Martine van Marsbergen for their continuing support and immense patience. Marlyse Baptista Linguistics Program University of Georgia Jacqueline Guéron Université de Paris III – Sorbonne Nouvelle

 

An introduction

 

Noun phrases in creole languages An introductory overview Marlyse Baptista and Jacqueline Guéron .

The premise of the volume and the questions it seeks to answer1

This volume grew out of a symposium hosted by Jacqueline Guéron and her research team (Jeune Equipe “Syntaxe anglaise et syntaxe comparative”) at the University of Paris III on April 21, 2000. The symposium’s original participants were Marlyse Baptista, Viviane Déprez, Karl Gadelii, Alain Kihm, and Anne Zribi-Hertz. A majority of the papers focused on the structure and semantics of NPs/DPs in creole languages, highlighting in particular their interpretive complexity. Baptista (on Cape Verdean Creole), Déprez (on Haitian) and Kihm (on Guinea-Bissau Creole) focused on the varied interpretations available to bare nouns and examined the interaction of null and overt determiners in yielding a specific or non-specific interpretation of the full Noun Phrase. In Cape Verdean and Guinea-Bissau Creoles, for instance, bare nouns may be interpreted as definite, indefinite, specific, non-specific or generic. The questions raised at this colloquium concerning the form and interpretation of bare NPs inspired us to invite fourteen other scholars to contribute to this volume by examining the semantic and structural properties of noun phrases in a variety of creole languages lexified by English, French, Dutch, Spanish and Portuguese. In the table of contents, creoles were grouped together based on their lexifier to make comparative analyses easier for the readers of this volume. In spite of the controversial status of African American English (AAE) as a creole, it was included in this volume due to the similarities in the nominal domain between AAE and creole languages. In Spears’ view, AAE is of interest for creole studies because it has not only lexical but also grammatical features that can be classified as Africanisms (features whose source can be traced to West African languages). In his contribution, Spears highlights a use of bare nouns in AAE that is quite distinct from that found in mainstream varieties. In her contribution on AAE, Green outlines the similarities between genitive and associative NPs in

. Abbreviations: sg. = singular, nb. = number, NbP = Number Phrase, Nb = Number node, NP = Noun Phrase, D or Det = Determiner, DP = Determiner Phrase.



Marlyse Baptista and Jacqueline Guéron

AAE and Creoles. As Spears defines AAE as a restructured noncreole variety, at most a semi-creole, and as Green does not take a categorical stance on its status, these two contributions were put together in their own section. This volume is thus devoted to studying the syntax and semantics of nominal expressions in a variety of creoles that have emerged from the contact between different European lexifiers and families of African languages. We have brought together descriptive and theoretical (generativist) studies that focus on similar issues in the domain of nominal expressions. We left the choice of framework to each contributor but in the spirit of comparative syntax, we proposed that each paper in the volume follow a similar format, and we provided the contributors with a specific set of guidelines. We requested that they present in their respective papers the complete overt determiner system (i.e. indefinite/definite, singular/plural overt determiners) while focusing on the distribution and interpretation of bare nouns. Some contributors chose to interpret bare as referring to nouns with no overt determiner but possibly endowed with a plural marker, while others chose to characterize as “bare” only determinerless nouns lacking plural marking. In either case, as editors, we asked the contributors to examine whether the bare nouns in a given creole yield different interpretations and if they do, what the exact range of interpretations is. We also invited the contributors to investigate the existence of specific grammatical factors which can help predict when a noun may or must be modified by a determiner, or when it may or must appear determinerless. We asked whether factors such as previous introduction of the NP in the discourse, animacy, number (singular versus plural), Tense (generic versus episodic), syntactic distribution (subject versus object position), uniqueness, and pragmatics could be shown to affect NP marking. Once the relevant distributional and interpretive properties of the NPs had been analyzed and presented, we encouraged contributors to investigate whether the same properties are found in the lexifier or the African languages that contributed to the formation of the creole, or whether they should be considered, rather, as language-internal innovations. For the papers developed within the generativist framework, we invited the authors to consult Chierchia (1998) in order to judge whether the creole they study verifies his Nominal Mapping Parameter and the resulting typology he develops in his seminal paper. In the next section, we consider the status of creole languages and the key research questions this volume raises.

.

The status of creole languages

.

Why group creoles together?

The title of this book Noun Phrases in Creole Languages begs a crucial question. Why are we grouping creole languages together in a comparative study? Usually, when distinct languages are studied as a group, they are genetically related. The Romance languages

Chapter 1. Noun phrases in creole languages

form a group because Fzrench, Spanish, Italian, Portuguese etc. all descend from Latin and are transparently related to Latin in both structure and vocabulary. Germanic languages, Semitic languages, and Slavic languages, likewise descend from one or a group of structurally-related mother languages. Creole languages are, however, a notorious exception, as they have no common ancestor. The genetic status of creole languages was, in fact, a matter of intense debate among historical linguists in the 19th century, when it was noted that contact languages in general, and creoles in particular, do not fit the traditional Stammbaum model (Schleicher 1853) that indicates how languages are genetically related to one another and traces every language to a single-parent source. Some historical linguists denied the existence of contact languages. Other historical linguists embraced the notion and documented language-mixture as characterizing both creole and non-creole languages. Schuchardt (1884) was among the first to show that grammatical mixture and changes through contact characterized not only pidgin and creole languages, but also affected non-creole languages, citing the interaction between Slavic and Germanic languages and between Slavic and Romance languages (see Thomason & Kaufmann 1988 and Winford 2003 for a good overview). The related issue of whether or not creoles are genetically related to their lexifiers is also a matter of controversy (see Thomason 2002 and Mufwene 2003). The editors of this volume adopt the position that creole languages did not develop from a single ancestor but resulted (gradually or not) from contact between at least two (and generally more) distinct languages during the catastrophic circumstances of slavery.2 They typically developed during a period of colonial conquest (15th–17th century), when African populations were deported from their home countries and put to work on plantations on lands conquered by Europeans. The first African plantation population presumably communicated with European overseers in a pidgin that mingled elements of their mother tongue with elements of a European language. From a heterogeneous source, the descendents of these first speakers developed a new and coherent natural language. What creole languages have in common then, is not a common ancestor, but rather the historical conditions of their birth (see Mufwene 2001 for a detailed historical account). And just as it is useful to study similarities between languages belonging to the same family, it is also useful to compare languages that emerged within similar socio-historical conditions, and that have in common that they are all contact languages par excellence. The editors of this volume do not assume that the creole languages constitute a typological class with respect to structure and interpretation (cf. McWhorter 1998, and counter-arguments in DeGraff 2001). As will become clear from the reading of this volume and from the preliminary conclusions we draw in the Postface, the determiner

. Note of course, that not all creoles developed under the conditions of slavery (i.e. Hawaiian Creole English).





Marlyse Baptista and Jacqueline Guéron

and number systems of the creoles examined in this volume are quite distinct from one another in specific ways that will be discussed below, and are not easily grouped into a single typological class. However, if one assumes the existence of a Universal Grammar, one can expect constraints in the linguistic forms of all natural languages, including those created by the contact between two or more independent languages rather than by descent from a single mother tongue. In the domain of the nominal system, a comparative study of creoles can be enlightening with regard to what kind of elements contact languages draw from their source languages while shaping their own innovative characteristics. The identification of the grammatical elements active in each language should provide us with insights on how the borrowed linguistic features are used by the contact languages in building a new and cohesive linguistic system. In the Postface of this volume, after having examined all the contributions of our authors, as editors, we will speculate on the possibility of characterizing how UG manifests itself in the formation of the nominal system in creole languages and how we may account for similarities in the grammars of distinct creoles. This discussion will lead us to relativize the failure of Cherchia’s classification and do justice to the nominal systems of creole languages by making finer distinctions in nominal paradigms and their variation over languages than have been found in recent theoretical proposals.

.

Some key research questions

Over the years, a number of authors have dedicated themselves to specifically examine plurality and the nominal domain in a wide variety of individual creoles. For instance, Tagliamonte, Poplack and Eze (1997) studied plural marking in Nigerian Pidgin English, Singler (1989, 1994) in Liberian Settler English and Sinoe Settler English respectively, Sankoff and Mazzie (1991) in Tok Pisin, and Dijkhoff (1983) in Papiamentu, to mention just a few. Alleyne (1980), Mufwene (1986) and Baptista (2003) are several attempts among many others to study pluralization strategies across creoles; while Holm (1990) and Bruyn (1995) conducted a systematic study of noun phrases from a comparative perspective. Though dealing with the same topic, such studies have had very different foci. For instance, although Holm offered a detailed account of the use of determiners and number in the Atlantic creoles he studied, he focused particularly on their use of the possessive pronoun as an emphatic form of the possessive adjective. In contrast, Bruyn’s comparative account was mostly concerned with showing how the loss of inflectional morphology in creoles affects the structure of the NP and the expression of reference. Adding one more dimension to the study of noun phrases, in this volume, the emphasis will be placed instead on the interpretive variation of overt determiners, bare nouns and plural marking across 15 creoles. Some key research questions emerging from this volume are the following: i.

In what respect do the relatively stable determiner systems of the creoles described in this volume differ from those of their lexifiers and/or from the African

Chapter 1. Noun phrases in creole languages

languages that contributed to their genesis? Can the input from the lexifiers and African languages be clearly identified? Does this input remain unchanged in the newly-formed creole? ii. Are the creole determiner systems under study similar or even comparable? The one feature they have in common is a use of bare nouns (or bare NPs) which is much more widespread than that of their respective lexifiers. Do the bare nouns in the creoles examined here cover the same semantic range or do the meanings of bare nouns vary from one creole to the next? Are the uses of overt determiners and plural marking similar to or distinct from those of the lexifier languages? This volume will attempt to bring reasonably informed answers to some of these questions by applying a multi-faceted approach to the analysis of creole determiner and number systems. The contributors profit from the wide variety of tools that syntax, semantics, discourse, pragmatics, historical and descriptive linguistics offer in an attempt to understand the full complexity of the creoles’ nominal systems. Natural languages contain two types of morphemes, lexical morphemes and grammatical morphemes. A grammatical morpheme may take the form of an inflectional affix like the plural morpheme -s on nouns or the past tense morpheme -ed on verbs in English, or that of a free-standing morpheme like the determiners a and the in English. Some creoles lack all or almost all of the inflectional system of their lexifiers, whereas others borrow lexifier morphemes but assign them modified or new functions. For instance, a plural morpheme may be inherited from the lexifier, like the inflection -s in Cape Verdean Creole or Guinea-Bissau Creole, while being used in basilectal varieties, only to modify human or animate and definite entities. French, Spanish and Portuguese nouns have Number and Gender features with which both adjectives and determiners must agree. But Portuguese, French, and Spanish creoles do not display such agreement in general. And if number agreement does occur between the determiner and the noun (as in Cape Verdean Creole (Baptista 2003, this volume)), it does not affect the adjective within the DP. The creoles typically lack gender or, if gender agreement can be found, it affects only a small set of adjectives (Baptista 2002). Some creoles lack inflectional number altogether and their use of determiners is also more restricted than in the lexifiers. While English, for example, has no gender or agreement for nb. in the NP, English obligatorily marks count nouns as sg. (zero inflection) or plural (-s). But English creoles can lack number inflection on N altogether (see Stewart on Jamaican, this volume, and Bruyn on Sranan, 1995 and this volume).3

. This is of course, not the case of all English-based creoles. Tagliamonte et al. (1997) and Singler (1989) report the presence of plural -s suffixation in Nigerian Pidgin English and Liberian Settler English, respectively. Animacy in the case of Nigerian Pidgin English and phonological environment in the case of Liberian Settler English are some of the factors playing a role in plural marking.





Marlyse Baptista and Jacqueline Guéron

The creole languages examined in this volume often (but not always) lack grammatical nb. inflection irrespective of their lexifiers, as illustrated by Lesser Antillean (French lexifier), Santome (Portuguese lexifier), or Jamaican (English lexifier). They often lack a full set of grammatical determiners as well, as in the case of Sranan, Haitian and Seychellois. It is clearly important to carefully examine the inflectional and determiner systems of a variety of creole languages in order to understand how they make up for lack or reduced use of number inflection and determiners. This situation raises other questions: do creole languages have another type of grammatical system that replaces the gender and number agreement suffixes and the determiners of the European lexifiers? Some creoles do not display number inflection on the noun; for those that do, are the function and nature of number inflection identical to those of their European counterparts? Did any of the creoles adopt the inflectional systems that characterize possible source African languages? For example, do any creoles have a classifier system, such as exists in the Bantu languages? Do any have focus marking on the verb and nominal inflections indicating argumental status like Philippine languages such as Tagalog? If we assume that some kind of inflectional system is necessary in the nominal structure of human languages, and if the creole languages have none or only fragments of the inflectional systems of the related European or African languages, then the hypothesis that creole grammars have recourse to principles of UG to fill in the gaps in their inflectional systems becomes reasonable (see Kihm, this volume). However, it is not easy to formulate and test the grammatical specificity of the creole languages studied here. We would need to examine the creole languages for properties they share which do not characterize either of their source languages. We cannot quite accomplish this task. For although we have information on the creoles, which are natural living languages, and although we have information on the 16th–17th century European source languages, we are uncertain of the African source languages. We refer to the native African languages of the first population in the colonies as the substrate languages and to the European language grafted on them as the superstrate or lexifier of the resulting creole. These terms suggest that the grammar of the creole is basically that of an African language and the vocabulary that of a European language. However, this hypothesis has not been proven to be correct. As Chaudenson (1994, 2001) has pointed out, it is often difficult to determine for creoles what languages were spoken by the deported Africans, who came from a variety of areas. Thus we do not dispose of all the pieces of the puzzle. We can, however, do part of the work. We can carefully examine the creoles, searching for those properties of their nominal structure which differ from the corresponding properties of the European lexifier and asking whether they share these properties. If our study shows that the creoles share properties not found in the plausible African source languages, then we have an argument for considering comparative work on creole languages as providing a direct window on the variety of grammatical systems which UG makes available. This task requires us to amass empirical data on a number of creole languages. We also need a model of grammar that clearly describes the components necessary in any

Chapter 1. Noun phrases in creole languages

grammar of a natural language, and, if possible, the constraints on sentence structures defined by these grammars. Our volume aims to further this task. Each chapter offers a detailed examination of the Noun Phrase grammar of one or more creoles. The reference grammar is Generative Grammar as developed by Noam Chomsky and his students since 1957, the latest development of which is The Minimalist Program (1995, 1999, 2000, 2001, 2002). This grammar posits a syntax with universal principles of sentence structure and combinatorial rules which Merge and Move syntactic constituents. The syntax also contains parameters whose value differs between languages. It has been proposed that all parameters are overtly realized by morphological items. It would be through overt morphology that children acquiring their language would set the syntactic parameters characterizing their language. For example, Romance languages have morphologically defective clitic pronouns, while English does not. A child learning its language can quickly set the relevant parameter by hearing sentences such as “John sees him” (English) or else “Jean le voit” (French). The hypothesis that parameters are purely morphological has very interesting consequences for the study of creole languages. If indeed inflectional morphology is missing or minimal in the nominal (or other) systems of creole languages, then the trigger for setting the inflectional parameters is also missing. Consequently, there should not be much parametric difference between the creoles in this (or any other) area of the grammar. As mentioned above, and as will be amply illustrated in the volume, this seems to be true to the extent that the use of bare nouns in the creoles studied here is more widespread than in their respective lexifiers. In the next section, we will briefly review the structure of a Generative Grammar and current hypotheses concerning the structure of the Noun Phrase. We will then briefly summarize the contribution of each author to the empirical study of one or more creole languages and the theoretical hypotheses he or she offers to explain the data. In a Postface, we will offer a hypothesis of our own concerning the specific resources UG provides to the nominal domain of creole languages that builds on the contributors’ empirical and theoretical work. We will also try to determine in what way the determinerless NPs which are common in the creole languages differ from those found in older languages such as Russian, Hindi, Japanese and Korean.

. .

On Universal Grammar and the structure of DP A general overview

The Grammar developed by Chomsky and his students from 1957 onwards has several distinct components. The Lexicon contains the basic building blocks of the sentence, which are lexical and grammatical morphemes. The Syntax is an autonomous computational mechanism that combines morphemes into words, words into phrases and phrases into sentences. The Phonology interprets the sounds of the terminal





Marlyse Baptista and Jacqueline Guéron

elements – words and morphemes – in terms of physical articulation. Logical Form assigns meanings to sentences on the basis of the denotation of their parts. This grammar feeds into two non-grammatical interface components of human competence, the phonetic articulatory component and the conceptual-intentional component. (1)

LEXICON

SYNTAX PHONOLOGY

INTERPRETATION

---------------------------------------------------------

(Interface)

conceptual

articulatory

A Noun Phrase (NP) typically denotes some object of the world. Proper Names (PN) like John refer to individuals directly. NPs containing common nouns refer to individuals indirectly, on the basis of the denotation of the Noun and the semantic function of the grammatical material that expands the Noun to form a Noun Phrase. An English Noun Phrase typically has three parts, a common Noun, a singular or plural number affix, and a definite or indefinite determiner. In English, Number is realized as -s for plural and 0 for singular. The structure of the Noun Phrases “the boy” and “the boys” is represented in (2): (2)

a.

DP (Determiner Phrase) Nb.P (Number Phrase) D

Nb.

NP (Noun Phrase) N

the

0/s

boy

[+def] [sg]/[pl]

b. c.

The boy The boys

The meaning of a syntactic phrase is made up of the meaning of its parts: N denotes a class of object, “boy”, but does not refer in isolation. Singular nb. + N denotes a class consisting of one member, while plural number +N denotes a class with a plurality of members. The definite determiner turns the class term into an object with a reference: the object is presupposed as existing in the discourse world of the speaker and hearer.

Chapter 1. Noun phrases in creole languages

If the speaker wishes to assert rather than presuppose the existence of the object in the discourse world, then s/he must use an indefinite rather than a definite determiner.4 (3)

a.

DP NbP

D Nb

NP N

a 0

boy

0 s

boy

b.

a boy

c.

0 boys.

Because the determiner is the highest syntactic node in the phrase, the nominal phrase NP has been renamed the Determiner Phrase or DP following Abney (1987). We hypothesize that in principle, the Nb. and Det. nodes are always present in an argument DP, although they may be absent in a nominal with a predicative function. However, there may be a mismatch between the syntactic structure, which contains several grammatical projections, and the phonological realization of the structure. A syntactic position or positions may contain no phonetically realized material. In English, for example, plural -s alternates in number with 0 singular, and the indefinite determiner 0 alternates with definite the. When a grammatical morpheme is not pronounced, how do we know it exists in the syntactic structure? i.

ii.

We assume that a null element exists if it alternates – is part of a paradigm – with one or more pronounced morphemes and if each element of the paradigm contributes a distinct semantic value. This is the case with 0 and -s under Number and “the” and 0 under D in English. Some member of a paradigm class must be chosen. Both the number morpheme and the determiner are obligatory in English.

Even when a position is lexically realized, we assume that there is not necessarily a one-to-one relation between a morpheme and a semantic value. Rather, it is possible . We have assumed that the indefinite determiner ‘a’ is generated in the Nb. head where it denotes singular number and then raises to the D node where it denotes, in addition, nonspecificity. Not all the contributors in this volume would agree with the representation of the indefinite noun phrase in (3).





Marlyse Baptista and Jacqueline Guéron

for a grammatical morpheme to raise from one position to another, accumulating in this way more than one semantically interpretable feature. For example, in English (3) above, “a” denotes sg nb. in Nb, and then raises to D, acquiring the feature nonspecific (-specific) in addition. On the other hand, we suppose that when -s denotes plural nb., the zero determiner is used for the -specific value of D. Chierchia (1998) proposed a Blocking Principle. This is a kind of Economy principle that stipulates that a zero morpheme for a given paradigmatic value is blocked whenever the grammar contains a phonologically overt morpheme with the same value (cf. Kester and Schmitt, Déprez, Alexandre and Hagemeijer, Gadelii, and Zribi-Hertz and Glaude, this volume, for analyses of Chierchia 1998). From the point of view of the Blocking Principle, since English has both a definite and an indefinite determiner in the singular, it is not surprising that it lacks a zero determiner for the indefinite or definite singular NP. (4)

a. Mary bought a book. b. Mary bought the book c. *Mary bought book.

Spanish, Portuguese, Italian, and English have bare plural noun phrases in which a zero indefinite determiner alternates with a definite determiner. But a bare singular NP is rare; it exists only in special cases like certain types of list coordinations and after specific prepositions. However, the bare singular is common and unmarked in many creole languages. Sentences like (5), from Portuguese-based Kriyol (Guinea-Bissau Creole), discussed at length by Kihm (this volume), do not exist in any of the European languages just mentioned. (5)

minjer kumpra kabra. woman bought goat ‘The woman bought the goat.’ or ‘The woman bought goats’

If the DP structure is universal, the nominals in (5) should have the structure in (6), where N is generated under the NP node, and a null morpheme is generated in both Nb. and D. In the absence of overt number and definiteness marking, “woman” and “goat” may be construed as definite or indefinite, singular or plural. (6)

DP D

0

NbP Nb

NP

0

N woman goat

Chapter 1. Noun phrases in creole languages

As Kihm shows, there is no insurmountable problem in interpreting such sentences given an appropriate context and grammatical cues the sentential position of the bare noun (whether it is in subject or object position) and its status as [+human] or [–human] all contribute to derive the correct interpretation. Interestingly, this is true of a number of other creoles examined in this volume, such as Cape Verdean Creole (Baptista, this volume), Santome (Alexandre and Hagemeijer, this volume), and Guinea-Bissau Creole (Kihm, this volume). If there is no alternation in a language between overt and null grammatical morphemes, unlike what pertains in English, then how do we know that D and Number exist? Given the ambiguity of the hypothesized null morpheme, are we justified in positing structure (6)? Should we not adopt the weaker hypothesis that the DP schema in (2) and (3) is not a universal, and that some languages, in particular creole languages, have bare NP structure? (7)

NP N woman goat

The articles in this volume supply empirical data and theoretical arguments which attempt to find answers to the two major questions raised here: i. ii.

.

Are there grammatical features common to creole languages of diverse origins? Is the DP structure universal or are there, alongside Determiner Phrases (DP), other nominal structures whose maximal projection is Number Phrase (NbP) or even a ‘bare’ lexical Noun Phrase (NP)?

Theoretical background

In this section, we review some specific theoretical proposals that were selected on the basis of their complementarity and also because they provide a theoretical foundation for a number of contributions in this volume. Reference to these proposals allows for a more accurate picture to emerge regarding how bare nouns behave in creole languages. More precisely, the studies by Chierchia and Longobardi focus on how syntactically comparable NPs can denote distinct interpretations within the same language and cross-linguistically. In contrast, Dayal’s proposal takes an opposite viewpoint and examines how NPs yielding the same interpretations can be associated with different structures. In other words, while Chierchia and Longobardi are concerned with semantic variation, Dayal examines syntactic variation. Guéron’s proposal complements Longobardi’s, Chierchia’s and Dayal’s accounts by going beyond the NP/DP levels. She stipulates that there is, at the clausal level, an interaction between grammatical heads (i.e. T and C) from which generic or episodic construals of nominal expressions emerge. Finally, Bickerton’s proposal is introduced here because





Marlyse Baptista and Jacqueline Guéron

it is the only one that is specifically concerned with the nominal systems of creole languages. A number of contributors to this volume examine his creole nominal system prototype, revealing its strengths as well as its limitations. We now consider each proposal individually.

..

Chierchia (1998)

In a cross linguistic perspective and on the basis of an examination of bare nominal arguments,5 Chierchia (1998) proposes a Nominal Mapping Parameter. Chierchia assumes the existence of semantic parameters that allow languages to vary the denotations of syntactically comparable NPs. He proposes that in languages like Chinese, endowed with a generalized classifier system and devoid of plural morphology, NPs are basically argumental. NPs denote names of kinds and are able to occur freely without a determiner in argument position. In the Romance languages, on the other hand, NPs are basically predicates. They cannot function as arguments unless the category D is projected and filled with a determiner to effect the semantic shift from Predicate to Kind. Finally, there are languages like the Germanic or Slavic families of languages that allow both predicative and argumental NPs. These languages being the ‘union’ of the previous two classes, they behave like Romance in certain aspects of their nominal system (singular count nouns) and like Chinese for others (mass and plural nouns). Ds contain semantic operators which effect a shift from one of the NP types defined by Partee (1987) to another. A Predicate shifts into a Kind via the “down” or NOM operation that takes the largest member of the extension of the property in any situation. For example, the “dog” kind can be identified with the totality of dogs. If a kind is the sum of its instances, then the function that creates kinds cannot apply to singulars. Existential readings involve the introduction of existential quantification over the instantiations of the kind in a given situation. Chierchia proposes that nouns coming out of the lexicon denote a kind 〈e〉, a predicate 〈e,t〉, or either of these, depending on the language. The possible meaning that a bare noun can have is a function of the setting of the Nominal Mapping Parameter. One of the crucial questions Chierchia addresses is how the definite interpretation arises for bare nouns. He proposes that for languages which do not have an overt definite article, the non-overt iota (ι) operator, a semantic type-shifter equivalent to a definite article, is available (Chierchia 1998: 359). However, if a language (such as English) has a definite article, ι is not available: it is “blocked.” This would reflect the last-resort nature of the type-shifting operation: “If there is a determiner D whose

. Bare nominal arguments are determinerless NPs occurring in canonical subject and object argument positions.

Chapter 1. Noun phrases in creole languages

meaning is a particular type shifting, then use of that operation as an automatic typechanging functor is blocked.” Chierchia’s proposal has one obvious empirical weakness: it fails to predict that in languages like Brazilian (Schmitt & Munn 1999, 2003; Munn & Schmitt 2001), or creole languages like Cape Verdean Creole (Baptista 2000), Papiamentu (Kester & Schmitt, this volume), or Haitian (Déprez 1999, this volume), bare singular nominals occur in argument position although these languages have plural morphology and lack a generalized classifier system. Chierchia’s stipulation that the nonovert typeshifting ι operator is available only for languages that do not have a definite article does not account for the data in Cape Verdean Creole or in Brazilian. On the other hand, Chierchia’s proposal seems validated by the study of creoles such as Lesser Antillean (Gadelii, this volume). We will return to this point in our Postface.

.. Longobardi (1994) Longobardi (1994) argues that bare nouns are not structurally bare but do in fact project a full DP, with an empty D head. Crucial to many papers in this volume is his theoretical claim that only DPs can serve as arguments. NPs are predicates or “propositions” and D turns them into arguments. The main function of D is to singularize or individualize: it can pick out a single instance of whatever is described by NP. The role of D is to mediate between the description arising form the NP and whatever specific entity the description refers to. In a nutshell, NP describes and D refers. Based on a comparative study of Romance and Germanic languages, one of the main generalizations in Longobardi’s (1994) proposal is that bare nouns with an indefinite interpretation can only be found in a lexically-governed position (object position). This would imply the presence of an empty D that is lexically governed, typically by V. The empty D is also responsible for the existential, indefinite interpretation of bare nouns in lexically-governed positions. In the domain of bare nouns, Longobardi makes a distinction between bare common nouns, generic nouns and proper names. He observes that distributionally, generics and proper names do not display the same restricted lexically-governed positions as other types of bare nouns. Longobardi accounts for the distributional difference by arguing that generics and proper names can raise from N-to-D overtly or covertly. They can fill the D position. This operation does not take place with bare common nouns. In this volume, the presence of bare singular count nouns in the creoles under study will allow some contributors to argue for N-to-D movement of bare common nouns. ..

Dayal (to appear)

Dayal’s paper focuses on ‘pleonastic’ determiners, with a particular emphasis on the plural definite generic common in Romance languages, illustrated in (8). She notes that generic kind-denoting terms vary between bare and definite forms within and across languages. (9a) is an example of a pleonastic determiner in English while (9b) is its bare counterpart.





Marlyse Baptista and Jacqueline Guéron

(8)

Les chiens aboient (French) I canni abbaiano (Italian) the[plur] dogs bark ‘Dogs bark.’

(9)

a. b.

The dog barks6 Dogs bark

The characterization of bare plurals as kind-denoting terms (‘dogs bark’), following Carlson (1977), is applicable to the singular definite generic in English (‘the dog barks’) and the plural definite generic in Romance (les chiens aboient) as well. Dayal raises the important question of how to account for the fact that the same meanings are associated with different NP structures across languages. She suggests that there are two approaches to the treatment of syntactic variation. The first approach is that apparent differences in form disappear at the level of representation, Logical Form, which feeds the semantics. Recall the hypothesis that languages that lack overt syntactic wh-movement have such movement on the level of LF. This means that there is no semantic variation across languages. The second approach, which Dayal defends, is that morpho-syntactic differences feed semantic interpretations. She draws part of her evidence from the Romance plural definite exemplified in (8) and the English bare plural in (9b) above. Dayal considers the domain of individuals to include atomic individuals as well as their sums. She makes the crucial assumption that singular morphology restricts the domain of quantification to the set of atomic individuals, while plural morphology includes plural individuals in the set defined by the noun. She takes the to denote a function iota which takes a set of entities and returns the unique maximal entity in that set. A maximal entity is one that includes all others. The maximality requirement is where the crucial difference between definites and indefinites lies: it ensures that plural definites select the whole group while singular definites denote the unique individual with the relevant property, or else are undefined. In a nutshell, indefinites differ from definites in not requiring maximality. Dayal’s study leads her to make three interesting generalizations applicable to the creoles examined in this volume: (10) (11)

(12)

Natural languages have no dedicated kind determiner, that is, a determiner used exclusively for NPs denoting kinds. (a) If a given language use bare nominals for contextually anchored/anaphoric readings, then it also uses them for plural kind formation; (b) if a language uses definites for plural readings, it also uses them for contextually anchored/ anaphoric readings. Contextually anchored/anaphoric readings and singular kind formation must agree in lexicalization. In a given language, they will either both be bare or both definite.

. As in many languages, the definite determiner seems to perform double duty as a standard definite and as a generic.

Chapter 1. Noun phrases in creole languages

The creoles examined in this volume seem to abide by these three generalizations. Like other natural languages, the creoles obey (10) above: none of them displays a dedicated kind determiner. Moreover, they seem, for the most part, to be subject to clause (a) of generalization (11): all creoles that use bare nominals for contextually anchored/ anaphoric readings, also use them for plural kind formation (i.e. Cape Verdean Creole and Guinea-Bissau Creole). Agreement in lexicalization outlined in (12) also obtains in the creole languages under study: whenever contextually anchored/anaphoric readings are associated with a bare NP, singular kind formation is also bare.

.. Guéron 2006 Guéron 2006 offers a useful proposal concerning the interaction between grammatical heads devoted to tense interpretation on the sentential level, essentially C and T, to yield generic and episodic construals of nominal expressions. Here are the main tenets of Guéron’s proposal that proved particularly useful for Cape Verdean Creole and that could easily be applied to other creoles in this volume (Guinea-Bissau Creole). (13)

Guéron’s basic assumptions:

According to Guéron, the constraint in (i) plays a pivotal role in LF structures: i.

ii. iii.

iv.

The situation a given sentence describes must be placed in the space and time of the discourse world, i.e. the world of the speaker or a world which the grammar makes accessible to the speaker. vP is a domain of spatial interpretation whereas TP/CP is the domain of temporal interpretation. A sentence is episodic if its tense node bears a positive tense index, and generic if its tense node carries a zero index or no index. This entails that any sentential syntactic structure may be ambiguous between an episodic construal and a generic construal. Differences in the tense index entail differences in the interpretation of (mainly subject) NPs. Thus, across languages, the same DPs refer to either real entities (‘real dogs’ for instance) or to the mental concept associated with the lexical item (‘the concept of dog’). Guéron proposes that the sentence structure contains a T(ense)-chain rooted in the complementizer node C associated with the Reference or Speech time, continuing with the tense node T associated with the Event time, and ending with V (or V plus an incorporated complement). If C is indexed 1 for Speech time and T is also indexed 1 for Event time (1 for Present and 2 for Past), then the Event time is construed as anaphoric to the Speech time and results in an episodic present. In Generic sentences, T has a zero index. In such sentences, no link is established between the event time in T and the discourse time in C. The basic hypothesis is that the on/off setting of the index on T is sufficient to distinguish episodic from generic sentences. If so, no generic operator is needed.





Marlyse Baptista and Jacqueline Guéron

v.

vi.

Aspect also plays a role and interacts with the output constraint mentioned in (i). The event time interval must be bounded in order to satisfy constraint (i). One way to bound a time interval is to place the event the sentence denotes in the scope of a bounded topic time. The topic time can be realized by an adverbial or a sentential adjunct. The subject of a sentence is linked to the discourse world by the anaphoric (or nonanaphoric, I would add) construal of its determiner.

In adapting Guéron’s proposal to the Cape Verdean case, Baptista adds two more stipulations. She proposes that the T-chain extends its index to DP in languages such as CVC where the use of bare nouns is prevalent. Thus if T carries a zero index and the chain ends within vP (the domain of spatial interpretation, see (13ii)), then D will also carry a zero index, whether D is filled or not, yielding a generic, non-anaphoric interpretation. If, in contrast, T carries a 1 or 2 index (for Present or Past), then D will match the index, deriving an episodic, anaphoric reading. In adopting (13vi) above, Baptista suggests that the subject of a sentence is linked to the discourse world by the anaphoric or non-anaphoric construal of its determiner (Lecarme (1996, 2004), Guéron and Hoekstra (1988), and Borer (2005)). The anaphoric construal of the determiner is obtained by matching D1 to T1, and the non-anaphoric construal by pairing D0 to T0.

.. Bickerton (1981) A number of studies in this volume (i.e. Bruyn, Gadelii, Kouwenberg) apply Bickerton’s (1981) determiner prototype to specific creole languages. In each case, they demonstrate that the language studied departs from Bickerton’s original model. Bickerton (1981: 22–24) takes a prototypical plantation creole such as Hawaii Creole English (HCE) and describes its determiner system as follows: i.

ii.

The definite article is used for all and only the specific reference of NPs that are familiar to the listener. The indefinite article is used for specific-reference NPs that are unknown to the listener. All other NPs have no article, including generic NPs, NPs within the scope of negation (which are non-specific), and in cases where, even though a specific referent may exist, the exact identity of that referent is either unknown to the speaker or irrelevant.

So, according to Bickerton (1981, 1984), there is a clear-cut opposition in creole languages between specificity and non-specificity reflected by the use (for specific entities) or absence (for non-specific) of an overt determiner. The non-specific bare nouns (characterized by absence of a determiner) include generic NPs, NPs in the scope of negation, and NPs with referents whose identity is unknown to the speaker or irrelevant. Naturally, a number of instances of bare nouns in the creoles examined here fall under some of Bickerton’s generalizations (i.e. a bare noun interpreted

Chapter 1. Noun phrases in creole languages

as generic). However, studies such as Bruyn’s and Kouwenberg’s show that contrary to the Bickertonian model, creole languages may use bare nouns to denote specific referents. It should be noted nevertheless that Bickerton’s model is not totally invalid, for it predicts that the occurrence of overt articles depends on a combination of factors, including referential properties (specific/nonspecific), familiarity (which is usually reflective of old as opposed to new discourse information), and syntactic contexts such as the presence of negation. Bickerton’s generalization holds, in fact, for several creoles studied in this volume. In the next section, we summarize the main findings of each contribution to this volume.

.

The contributions

Nélia Alexandre & Tjerk Hagemeijer The two main goals of this paper are to describe the nominal domain of Santome and to propose a DP structure which can account for the data examined. The authors introduce the complete determiner system of Santome, which includes both overt and null determiners, and focus more particularly on the properties of bare noun phrases. By bare noun phrases (BNPs), Alexander and Hagemeijer mean noun phrases without overt determiners. What sets this paper apart from a number of other contributions in this volume is that it stipulates two levels of interpretation for BNPs: sentence grammar, characterized by the interaction of syntax and semantics, and discourse, an extra-grammatical level where ‘information packaging’ takes place (Lambrecht 1994 and Vallduví 1993). BNPs can be found in all syntactic positions and appear to be underspecified for definiteness and number, since they lack overt grammatical and lexical markers identifying such features. The discourse level is thus necessary to allow for the full set of possible interpretations of BNPs. In their analysis, null determiners are analyzed as equivalent to the lexical determiners (wa and inen), and the BNPs they define are fully integrated in the nominal structure they propose. The authors claim that the two definite lexical determiners, wa and inen, carry number and definiteness specification. In their view, in the absence of overt plural affixation in Santome, wa and inen come from the lexicon with an inherent number feature. Definiteness, usually related to identifiability, links sentence level to discourse level, whereas number follows directly from their inherent number feature when determiners are overt. Therefore, the authors feel no need to assume a separate functional category NbP. This implies that Santome may be at an intermediate stage of development in which it may overtly express definiteness but not yet number. From a historical perspective, this fact is consonant with the observation of Wood (2003)





Marlyse Baptista and Jacqueline Guéron

according to which a language may develop definiteness before number. Wood showed that Old English differs from Modern English in realizing a Determiner Phrase but no Number Phrase. From a theoretical perspective, Woods’ observation seems to corroborate proposals such as that of Longobardi (1994) who assumes that number is checked in D0 and claims that “arguments but not predicates crucially rely on the D0 position in order to define their meaning with respect to number”. Alexander and Hagemeijer conclude that the determiner system of Santome uses minimal lexical resources (wa/inen) for two specific functions. Multifunctional null determiners fill in the remaining functions. In each case, the correct interpretations can be derived thanks to the existence of the grammatical and extra-grammatical levels discussed above.

Marlyse Baptista Baptista’s chapter provides an exhaustive description of the semantics and structure of the Cape Verdean (CVC) noun phrase, including both determined and determinerless nouns. Baptista argues that both the definite and indefinite determiner perform at least two functions in the language. The definite kel/kes functions as a demonstrative determiner and also fills the role of a definite determiner identifying specific/referential and non-specific/non-referential nouns. Indefinite un functions as a numeral, as a quantifier and as a determiner, while its plural counterpart uns acts as a quantifier. The indefinite determiner un may refer to a specific or non-specific entity. While overt determiners and their null counterparts are shown to cover the same range of meanings in yielding nouns interpretable as generic, definite, indefinite, singular, plural or mass, singular bare nouns display a subject/object asymmetry when construed as (in)definite. CVC bare nouns are subject to licensing conditions. Baptista examines factors that favor the use of bare nouns such as animacy, definiteness, the singular/plural contrast, generic versus episodic tense, uniqueness of entities and pragmatic conditions. Expanding on Guéron’s 2006 proposal, Baptista suggests that an indexing relation between D, T and C is needed in order to derive the desired interpretations of bare nouns. Baptista also examines adjective placement in the Cape Verdean DP in order to detect the presence of N-raising and to identify the nature and number of the structural layers within DP. Nominal projections are not assumed to be uniform across noun types. Baptista proposes that while count nouns involve DP, NumP and NP projections, generics and mass nouns involve only the DP and NP projections. She postulates N-to-D raising through NumP for definite specific/referential nouns. In such cases, NumP projects whether or not the plural marker is overt. Finally, Baptista compares CVC to both its European Portuguese (EP) lexifier and to Brazilian Portuguese (BP). She shows that while the use of bare nouns is much more prevalent in CVC, CVC and BP have much in common in several domains. They may both use a bare noun in the true sense of the term (no determiner, no plural

Chapter 1. Noun phrases in creole languages

marking) with generics, indefinite non-specific/non-referential plural nouns, and indefinite specific/referential singular and plural nouns. CVC and BP also use proper names in a similar fashion: in contrast to EP, no determiner is required.

Adrienne Bruyn For Bruyn, as for Zribi-Hertz and Glaude, bare nouns are simply determinerless nouns. Bruyn investigates the distribution of determiners in Sranan and discusses the factors determining their use and their absence. She also assesses the appropriateness of various proposals accounting for determiner distribution in creole languages, such as those of Bickerton (1981), Givón (1984) and Mufwene (1986). As discussed above, Bickerton (1981) argues that bare nouns in creole languages typically have non-specific reference; or, if the referent does have a specific identity, that identity is not relevant. Givón (1984) speaks of referentiality rather than of specificity, but essentially holds a view similar to that of Bickerton. Finally, Mufwene (e.g. 1986) considers bare nouns in Gullah and Jamaican Creole to be non-individuated; that is, they refer to a whole set rather than to its individual members. One of the main claims of Bruyn’s paper is that the role played by referentiality should not be overestimated vis-à-vis that of number and definiteness. The article system of Sranan cannot be properly understood if one does not take into account the fact that articles primarily express number and definiteness, a correlation that the proposals by Bickerton, Givón and Mufwene did not establish. Like Chaudenson (this volume), Bruyn offers both a synchronic and diachronic treatment of the use and distribution of overt determiners as well as of bare nouns in the creole she examines. She notes that the use of the indefinite singular article wan or of the definite articles (n)a () and den () is not generalized to the same extent as article usage is in English, for instance. She shows, in particular, that wan has not grammaticalized towards a more generally used indefinite article over time and that it is primarily a non-obligatory marker of singularity and indefiniteness. Bruyn argues that in Sranan, the marking of singular indefiniteness is the core function of wan, and that referentiality is not the primary distinction determining the use of wan versus zero, even though wan occurs more often with pragmatically salient participants. The occurrence of wan in Sranan can best be understood if one assumes that the function of wan is to make singularity and indefiniteness explicit, regardless of referentiality. The need to signal - individuation may vary. While the speaker’s intention constitutes the starting point, in some contexts an explicit marking is more appropriate whereas in other contexts a bare noun suffices. Although this corresponds to pragmatic referentiality more often than not, such is not necessarily the case. In brief, in some cases it is impossible to relate the occurrence of wan to anything other than the intentions of the speaker with regard to individuation. The rest of Bruyn’s paper introduces definite articles and demonstratives synchronically and diachronically. She also addresses the evolution of bare nouns and their





Marlyse Baptista and Jacqueline Guéron

possible interpretations. Finally, she compares bare nouns in Sranan with those in the lexifier and substrate languages, and concludes that given the differences between Sranan, English and Gbe languages, one must conclude that Sranan has developed its own system with respect to the presence or absence of determiners.

Robert Chaudenson Chaudenson’s paper examines the overt determiner system of Réunionnais Creole (RC), with a special focus on the use, interpretation and distribution of bare nouns. In an attempt to determine whether or not bare nouns in RC originate from earlier stages of French, Chaudenson compares the occurrence of bare nouns in RC, in Modern French, and in Middle French, while raising the question whether there are uses of bare nouns proper to RC. Bare nouns in RC occur not only with proper nouns, toponyms, professions, and kinship terms, but also with collectives and generics, anaphoric nouns, and both nominal and verbal complements, among others. Chaudenson’s study highlights similar and different uses of bare nouns in RC and in Modern French. Chaudenson also shows that a number of bare nouns in modern RC are actually inherited from XVII century French. Although the focus of his paper is on bare nouns in RC, Chaudenson makes interesting observations about Modern French. He brings ample evidence to show that while many bare nouns in Modern French proverbs are inherited from Old French, bare nouns are also frequent in Modern French, especially in expressions containing “light verbs” such as faire (make) or prendre (take). Examples are: faire mal ‘to hurt or to get hurt’, prendre forme, ‘to take shape’, donner vie ‘to give birth’, etc. In all these cases, the light verb combines with an abstract bare noun in Modern French. However, Modern French does not allow chercher herbe ‘to fetch grass’ in contrast to RC rod zerb. We must therefore distinguish light verb structures involving abstract nouns, which typically contain bare nouns in Modern French, and non-light verbs which combine with concrete nouns which require an overt determiner in Modern French but not in RC. Chaudenson notes that in the history of French between the XIV and XVII centuries, the use of the zero determiner regressed, while the use of definite and indefinite articles increased. He draws convincing evidence for this pattern from MarchelloNizia’s (1979) study of earlier stages of the French language (XIV–XV centuries). Echoing Valli (1994), Chaudenson proposes that Réunionnais Creole bare nouns are found in the same grammatical contexts as in XVth and XVIth century French. He cautions against the blind adoption of the dates that historical linguists provide with regard to specific grammatical features, as temporal gaps between literary usage, documented in grammars and dictionaries, and spoken French can be centuries long. Chaudenson notes that the linguistic situation is all the more difficult to describe, as a variety of French dialects were present when RC emerged. For this reason, one must remain cautious: while some grammatical features can be traced back to Middle French, others could be innovations proper to RC. Among possibly inherited features,

Chapter 1. Noun phrases in creole languages

Chaudenson notes a tendency in RC for overt determiners to appear before monosyllabic nouns where, in the same grammatical functions, a polysyllabic noun can remain bare. This tendency may go back to an older state of the French language, as up to the XIX century, the agglutination of the determiner to the noun was quite frequent. The current sociolinguistic situation of diglossia in la Réunion is also the source of many changes in the language, including changes in the use of determiners. Chaudenson’s careful descriptive study takes into account both acrolectal and basilectal varieties of RC and notes that “basilectal erosion” (Chaudenson 1981) occurs in a number of grammatical areas; this includes the disappearance of determiners agglutinated to the noun borrowed from older French by RC. As a result of erosion, a number of basilectal variants are disappearing.

Viviane Déprez Viviane Déprez’ paper begins by addressing two main questions: Is the range of meanings of bare nouns identical in creole and non-creole languages and is the same set of factors responsible for the choice among the possible meanings of bare nouns? Déprez applies methods of semantic analysis which have been elaborated for non-creole languages such as English and Italian to the study of four Romance creoles: Seychellois (or Seychelles Creole), Mauritian (or Morisyen), Haitian Creole and Cape Verdean Creole. Déprez limits her investigation of bare nouns to simple count bare nouns occurring in subject and object argument positions. Déprez shows that two two non-creole languages, English and Italian, differ with respect to the use of bare nouns for kind readings. English allows this while Italian does not. However, all of the creole languages in Déprez’ sample, do allow kind readings of bare nouns. They differ as to whether or not they allow a definite reading of a bare noun. Such variation can also be found in non-creole languages such as Russian, Hindi or Chinese. To account for the empirical observations in her study, Déprez proposes a theoretical analysis which centers on the role of number and its morpho/syntactic realizations in influencing the range of meanings bare nouns can have over languages. She formulates the Plural Parameter as an alternative to Chierchia’s (1998) Nominal Parameter, whose predictions were shown to empirically fail for a number of creole languages including Haitian Creole (Déprez 1999, 2001) and Cape Verdean Creole (Baptista 2000). Déprez correctly observes that in contrast to English and the Romance languages, where number is obligatorily expressed and conveyed by means of bound morphemes, number in creole languages is often optionally expressed and frequently (but not always) encoded with a free morpheme. At the core of Déprez’ theoretical proposal is that this difference in number marking plays a crucial role in determining the readings that argument bare nouns can have cross-linguistically. Her proposal leads her to formulate a number of interesting generalizations: Existential and generic readings tend to be available in languages that allow plural bare nouns. In her view,





Marlyse Baptista and Jacqueline Guéron

this derives from the universal availability of NumP and from standard existential closure and generic binding. A second generalization is that kind readings of bare nouns are available in all −PL languages while they may be missing from +PL languages. This would follow from the Plural Parameter: NumP projections can be missing only in –PL languages, in which the kind reading of a truly bare NP is the most basic one. In contrast, in +PL languages, the NumP projection derives the existential/ generic reading as the most basic construal of bare nouns. Kind readings must then be derived either constructionally or through the presence of an overt operator. Déprez’ third generalization is that systematic definite readings of bare nouns are in complementary distribution with overt definite determiners. This follows from her proposal that the null definite determiner occupies the same position in the nominal functional structure as its overt counterpart. The prediction is then that bare nouns with null definite determiners will be semantically and syntactically equivalent to nominal expressions with overt definite determiners (see also Déprez 2003, 2005 for more crosslinguistic comparisons on number and bare nouns).

Karl Gadelii Gadelii’s paper introduces the full inventory of overt determiners in Lesser Antillean (LA). He notes the absence of an indefinite plural determiner. Gadelii focuses more particularly on bare noun phrases. He shows that the bare NP in LA covers the same range of meanings as non-bare NPs, but that the use of bare nouns is regulated by a number of factors, in particular, by topicality. He notes that although topics typically appear in subject position, they may in fact also appear in object position. Other factors favoring the appearance of bare noun phrases include animacy, specificity, and number. Gadelii makes it clear, however, that the paramount factor facilitating the occurrence of a bare noun is topicality rather than specificity. Gadelii’s results, like those of Bruyn and Kouwenberg (this volume), thus falsify Bickerton’s claim that specific NPs require a determiner (definite or indefinite), whereas non-specific ones do not. One interesting generalization emerging from Gadelii’s paper is that a bare NP can always convey definite reference in LA. The challenge, however, is to account for the fact that non-bare NPs with the same interpretation of definite reference appear in his corpus as well. Gadelii tentatively proposes that the non-bare NPs with definite reference which appear in his corpus are not “topical enough” to allow their determiners to be omitted. In particular, they do not function as subsequent subjects of the sentences the story is woven around. In the last part of his paper, Gadelii presents a theoretical analysis following Gadelii (1997) in which he assumes that strong features in creole languages are checked by “lexicalisation” when the feature involved is a head feature, and by movement when it is a specifier feature. On the basis of these assumptions, he concludes that LA never exhibits head movement but does have more extensive phrasal movement than does its French lexifier.

Chapter 1. Noun phrases in creole languages

Anthony Grant Grant’s paper studies the structural properties and distribution of NPs in Mindanao Chabacano (MC), a Spanish-lexified creole. Among other things, the author notes that the use of overt determiners in MC is not entirely systematic: there are a number of instances where zero appears when a determiner, particularly an indefinite determiner, would be expected. In general, however, an overt determiner is used whenever the semantics of the MC NP requires one. The determiners in MC directly inherited from Spanish are masculine singular un and el. Neither feminine determiners nor plural determiners of either gender have passed into productive or systematic use as determiners in MC, as the language lacks both grammatical gender and overt plural marking on determiners. While not systematic, overt plural marking on otherwise bare nouns is frequent in MC. Bare nouns most frequently occur with copular predication and locative expressions. In the last section of his paper, Grant compares the MC system of NP marking to those of the languages that have most influenced it over the past few hundred years. The relevant Philippine languages that have come into contact with MC are Caviteño, Hiligaynon and Tagalog among others. Grant concludes that MC NPs and NPs in Philippine languages such as Tagalog show very different patterns and that under superficial similarities lie striking differences.

Lisa Green In contrast to Spears (this volume) who studies the distribution and interpretation of all types of bare nouns in African American English (AAE), Green’s paper examines more particularly the interpretation of indefinite and bare plural NPs in aspectual be constructions in AAE. One of her major findings is that bare plurals in aspectual be constructions reflect the same type of generic versus existential ambiguity associated with bare plurals in English first revealed in Carlson (1977) and Diesing (1992). Green notes, however, that the readings associated with bare plurals in aspectual be constructions are slightly different from those of bare plurals in English. More precisely, in aspectual be constructions, bare plurals with individual-level predicates are unambiguous, while bare plurals with stage-level predicates are ambiguous between a habitual and an existential reading. Green also shows that singular indefinite NPs get different types of readings in aspectual be constructions depending on the type of predicate they are associated with. Green offers a general overview of the syntactic and semantic properties of aspectual be, a verb which denotes habitual aspect rather than tense. She likens aspectual be constructions to simple tense generics in that they receive a habitual interpretation. She compares aspectual be constructions with characterizing sentences, on the one hand, and with sentences that express accidental properties, on the other. Green’s empirical evidence leads to her to conclude that aspectual be constructions indicate regular occurrences and express essential properties.





Marlyse Baptista and Jacqueline Guéron

Interestingly, Green notes that bare singular NPs are not permitted in AAE to convey singular indefinite or bare plural readings. Spears’ contribution (this volume) could on the surface go against Green’s generalization, as he characterizes a bare singular NP in AAE as being interpretable as generic (as well as definite). If generic is to be included in the category of bare plurals, then the differing grammatical judgements between these two native speakers of AAE may be an attestation to dialectal variation.

Ellen-Petra Kester and Cristina Schmitt Kester and Schmitt compare bare nominals in Papiamentu (a Portuguese-based Creole spoken in the ABC-islands Aruba, Bonaire and Curaçao) and in Brazilian Portuguese. Both these languages also have definite and indefinite determiners. The double focus on bare nominals and overt determiners, and more particularly the definite determiner, is based on the assumption, following Chierchia (1998) and Dayal (2002), that the existence and interpretation of bare nominals in a language can be correlated to the lexical options available in that language. The paper clearly demonstrates that bare nominals in Papiamentu and in Brazilian Portuguese are similar in many respects. However, there are some differences in the behavior of bare singulars and even more striking disparities in the behavior of bare plurals. The different distribution of bare plurals is attributed to a difference in the semantic features of the bare plural in the two languages. Bare singulars are argued to have the same semantics in the two languages, in spite of a different distribution: in Brazilian Portuguese, the definite determiner is required in many more contexts. The different distribution of Bare singulars in Papiamentu and Brazilian Portuguese would be related to the different semantic properties of the definite articles in the two languages, and not to the semantics of bare nominals as such. In terms of interpretation, the two authors treat bare singulars as names of kinds in the two languages, but they note that Papiamentu bare plurals have neither the same interpretations nor the same distribution as bare plurals in Brazilian Portuguese: in the latter language, bare plurals can receive generic and existential interpretations and can appear freely in both subject and object position, but this is not the case in Papiamentu. Furthermore, while the bare singular in both languages and the bare plural in Brazilian Portuguese denote names of kinds, the bare plural in Papiamentu actually represents a combination of a plural marker and a feature denoting specificity. Theoretically, the authors do not subscribe to the view that functional heads must have morpho-syntactic and/or semantic content in order to be projected in the syntax. They further assume that a single morpheme may carry features associated with more than one functional head. Kester and Schmitt propose that bare singulars in Papiamentu are syntactically to be treated as DPs with no number projection and semantically as names of kinds, as argued by Schmitt and Munn (1999, 2003) and Munn and Schmitt (2001) for Brazilian Portuguese bare singulars.

Chapter 1. Noun phrases in creole languages

Alain Kihm Kihm considers as bare only nouns appearing in their root form, lacking both a determiner and number marking. His description of the determiner system of GuineaBissau Creole can be summed up as follows: i. ii. iii.

There is no overt definite determiner (although the pronoun kil often seems to assume such function). There is a quantifier un(s) ‘one, a, some’ which functions as a singular or plural indefinite determiner. Noun phrases overtly pluralized with /-(i)s/ suffixed to the head noun are interpreted as referential; nonreferential nouns are not marked for plurality.

Consequently, in the particular variety of Kriyol that Kihm investigates, the sentence in (1) could be interpreted as ‘the woman bought the goat’ or ‘the woman bought goats’. (1)

Minjer kumpra kabra woman buy

goat

The first interpretation refers to a specific, definite goat whereas the second refers to an unknown quantity of goats. If the interpretation ‘the woman bought the goats’ were to obtain, the referential plural meaning of the NP “the goats” would require pluralization and yield ‘kabras’, according to clause (iii) above. The authors in this volume present distinct definitions of referentiality and definiteness. For Kihm, (following Lyons 1999), a noun phrase is referential if it denotes one token or a (potentially) countable set of tokens of a given type; it is definite if the token or token set is identifiable. Kihm extends the usual concept of definiteness when he argues that the goat in the goat is a mammal is definite as well as being generic because it refers to the kind “goat” and kinds are identifiable by definition, they belong to basic shared knowledge. However, the goat in that sentence is not referential as no particular member of the kind is being referred to. Kihm introduces the useful concept of “abstractability”, which could be viewed as a form of animacy hierarchy, according to which the interpretation of BNPs as mass terms depends on two factors. First, there is a scale on which humans are the least abstractable individuals (meaning that they cannot be viewed as an undifferentiated mass), individualizable animals are more abstractable, and inanimates (including plants) are most extractable, meaning that they can be easily massified. The second factor is syntactic: object BNPs are more likely to be interpreted massively than are subject BNPs. In other words, object BNPs can be interpreted individually or massively, whereas subject BNPs can only be interpreted individually. An important component of the animacy hierarchy is that nouns referring to inanimates, i.e. strongly abstractable entities, tend not to be pluralized unless they are very clearly topical. Topicality overrules abstractability; thus, even inanimate nouns may get an individual interpretation as long as they are in the topic (subject) position of the sentence. Kihm also proposes that BNPs are interpreted massively if they incorporate to the verb





Marlyse Baptista and Jacqueline Guéron

of which they are the internal argument. Thus kumpra kabra ‘buy goat’ is construed, when interpreted massively, as a kind of intransitive predicate (‘goat-buying’ similar to ‘deer-hunting’). This explains why subject BNPs do not receive mass interpretation, as it is a general property of subjects that they do not incorporate. At the syntactic level, Kihm assumes that the Kriyol lexicon includes a functional feature Number valued as <+> (more than one), < – > (singular noun), or (mass nouns and if number is totally absent or irrelevant). Summing up, Kihm proposes that referential, i.e. non-mass, noun phrases, singular or plural, definite or indefinite, are NumPs in Kriyol. Only mass nouns are true BNPs. Subjects are never interpreted as mass if they refer to humans, due to the inabstractability of humans. If subject BNPs refer to animals or inanimates, they cannot be interpreted as mass, as topicality overrules abstractability. In addition, Kihm accounts for the massive interpretation of BNPs in object position by having recourse to the concept of incorporation.

Silvia Kouwenberg Kouwenberg’s contribution examines the distribution and interpretation of bare nouns in Berbice Dutch (BD), a plantation creole that resulted from the contact between Dutch and Eastern Ijo, formerly spoken on the Berbice River and Canje river area in Guyana. She defines a “bare noun” as a determinerless NP, which may contain adjectival modifiers and plural markers. Following a similar line of investigation as Bruyn (this volume), Kouwenberg demonstrates that the BD determiner does not conform to the prototypical determiner system described in Bickerton (1981), even though some of the data do fit his model. Comparing determiner-taking BD nouns to their bare noun counterparts, Kouwenberg shows that the use of bare nouns is not random; there are several contexts where bare nouns are expected and even predictable. These include cases where the referent of the noun is inherently plural (collective) or designates an entity of indefinite size (mass). Bare nouns are also preferred for generic reference, or as predicates. They are also favoured when referring to “unique” referents (whether the “uniqueness” of the referent is semantic or pragmatic) or when the exact identity of the referent is irrelevant or cannot be determined. One of the most striking traits of BD nouns is the distinction they observe between class generics and individual generics. Whereas class generics are expressed by bare nouns, individual generics make use of a definite determiner: the referent denotes a typical representative of its class, the individual representative of the property predicated by the noun. The descriptive sections in Kouwenberg’s paper are completed by a theoretical account where she argues that for the most part, the bare nouns in her data have an extended projection headed by a null determiner; hence the bareness of the noun is only a surface property. She draws her evidence from distributional and functional

Chapter 1. Noun phrases in creole languages

properties of bare nouns, as well as from theory-internal considerations. Ultimately, Kouwenberg argues, the only nouns that are truly bare are those that are nonargumental and have predicate status.

Armin Schwegler Early investigations into Palenquero had all noted that the plural noun phrase of the creole differs in fundamental ways from that of its Spanish lexifier. There was general agreement that the (supposedly invariable) Palenquero zero form always functions as the singular definite article (e.g. ómbe ‘ man’, muhé ‘ woman’) or as a modifier of generic and mass nouns (ómbe ‘man [in general], mankind’, muhé ‘women [in general]’). In functional terms, the omission of article was thus said to correspond to Spanish el (masculine) or la ‘the’ (feminine). Earlier descriptions of Palenquero agreed, furthermore, that ma ‘definite particle’ and un ma ‘indefinite particle’ (lit. ‘one PL’) regularly mark the plural, thereby presumably yielding a system in which the singular alone allows bare nouns (1). ‘the man’ (singular)

(1)

Ø amígo [traditional analysis] *‘men’ (plural)

The present article argues that the traditional accounts of Palenquero’s article system are fundamentally flawed. Schwegler’s analysis is largely based on data he had collected in situ, but also relies on Moñino (in press), who similarly questions the traditional accounts of the Palenquero article system. Moñino and Schwegler have reached a consensus as regards two key points: in their view, (1) Palenquero has noun phrases consisting of just a bare noun; and (2) a striking property of these bare nouns is their non-specification for number (they are neither singular nor plural). The main objective of this study is to offer a comparison between Schwegler and Moñino’s analysis of Palenquero ma, and to outline why they – independently so – reject the traditional views on the matter. In so doing, Schwegler shows that the Palenquero lexicon emphasizes the role of context rather than overt morphology or lexical structure. This new approach represents a radical departure from earlier analyses, as it renders all Palenquero nouns “transnumeral.” Moñino’s and Schwegler’s findings coincide on several major points of grammar, but differ on the more theoretical issue concerning the genesis of ma.

Arthur Spears The main objective of Spears’ paper is to reveal the differences between African American English (AAE) and non-African-American dialects of English, i.e. other dialects of English (ODE), with respect to the distribution and interpretation of bare nouns. By bare nouns, Spears mean nouns without determiners or a plural suffix. Two striking features of bare nouns in AAE are that count nouns may take definite and generic





Marlyse Baptista and Jacqueline Guéron

interpretations and that mass nouns may also take definite interpretations, contrary to ODE. On this issue, as already mentioned, Green (this volume) notes that bare singulars NPs are not permitted in AAE to convey singular indefinite or bare plural readings. This could on the surface undermine Spears’ observation that count nouns can convey a generic interpretation if generic is to be included in the category of bare plurals. If such is the case, then the differing grammatical judgements between these two native speakers of AAE could just be a reflection of dialectal variation within AAE. To get back to Spears’ account of interpretive options in bare nouns, he observes that the definite interpretation of count and mass bare nouns can also be found in creole languages, as is attested by a number of studies in this volume. As a result, without taking a stand on the creole status of AAE, Spears views such a feature as a “creolism” within AAE. Just as in ODE, plurality must be marked in AAE, while the definite article is optional under a definite interpretation. In other words, for a noun to be definite and plural, it must be marked with the plural but does not need to be modified with the definite determiner. Unlike ODE, bare nouns may also be specific. An interesting generalization emerging from Spears’ data is that a bare singular noun is definite only in the subject position of a finite sentence. This could suggest that the bare noun may in fact contain a null determiner bound by a possibly null sentence topic, a situation that is reminiscent of what has been proposed for some creole languages (cf. Gadelii, this volume). Spears also observes that this solution does not hold for a bare singular in the scope of negation. In this case, it may be assumed that the null determiner is bound by a NEG/existential operator.

Michele Stewart Stewart makes several claims about the structure of the noun phrase in Jamaican Creole (JC). She draws a strong correlation between number and definiteness in JC and although these nodes are not phonologically realized, she argues for a functional structure above the noun phrase in JC. Upon analyzing the interpretive options of bare nouns phrases in this creole, she observes that they can receive a plural individual interpretation through a null functional head in the absence of overt plural marking. She also assumes a multi-layered structure for definite and indefinite bare nouns. In her proposal, only generics are structurally bare NPs. One of the claims that most radically departs from traditional descriptions of the grammar of JC concerns the nature of the post-nominal dem, usually described as a plural marker. Stewart argues that the primary function of dem is not to express plurality but inclusiveness. Based on her claim that there is no overt morphological plural marking in JC, she argues for an elaborate functional structure above NP allowing for the multiple interpretive options of bare nouns in JC, as they may convey a singular, plural, individuated or non-individuated reading.

Chapter 1. Noun phrases in creole languages

Stewart postulates a different set of projections depending on the reading of a given bare noun. She argues for instance that a bare noun with an indefinite reading projects NumP. Individuation, when it occurs, is achieved via a classifier phrase ClP (Borer 2005). Existential readings require only ClP and, as already mentioned, only generics are structurally bare in her view. Stewart also accounts for the overt determiners in JC and proposes that the definite article di is located in D but that the indefinite wan is really a numeral, and as such, occurs in Num(ber)P.

Anne Zribi-Hertz and Herby Glaude Anne Zribi-Hertz and Herby Glaude define bare NPs as noun phrases lacking a determiner, a situation they characterize as “functional deficiency”. They study functional deficiency in the determiner system parallel to number deficiency in both Haitian and French noun phrases. Zribi-Hertz & Glaude propose that the study of bare nouns should consider the phonological and syntactic modules of grammar separately. Their paper provides a thorough description of definiteness and number marking in simplex and relativized Haitian and French noun phrases and analyzes, in particular, the overt definite determiner and the plural marker. The authors conclude that the definite determiner has a different categorial nature in French and in Haitian. Using an adaptation of Kayne’s (1994) analysis of restrictive relativized noun phrases, they show that Haitian relativized noun phrases are characterized by two distinct functional levels, specified for number and/or for definiteness. A relativized DP may be deficient for some functional feature at one level but not at the other. They extend their analysis to French. The difference between the two languages would be that in French, a maximal DP is limited to one overt D head. In the Postface to this volume, we consider what kind of generalizations the contributions to this volume allow us to make regarding the nature of creole nominal systems.

References Abney, S. P. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Alleyne, M. 1980. Comparative Afro-American. Ann Arbor MI: Karoma. Baptista, M. 2000. Variabilité référentielle des syntagmes nominaux nus en créole cap-verdien. Paper presented at the University of Paris III Linguistics Symposium. April 7th, 2000. Baptista, M. 2002. The Syntax of Cape Verdean Creole: The Sotavento varieties [Linguistics Today 54]. Amsterdam: John Benjamins. Baptista, M. 2003. Inflectional plural marking in creoles and pidgins: A comparative study. In The Phonology and Morphology of Creole Languages, I. Plag (ed.), 315–332. Tübingen: Niemeyer. Bickerton, D. 1981. Roots of Language. Ann Arbor MI: Karoma. Bickerton, D. 1984. The language bioprogram hypothesis. The Behavioral and Brain Sciences 7: 173–221.





Marlyse Baptista and Jacqueline Guéron Borer, H. 2005. Structuring Sense, Vol. I: In name only. New York NY: OUP. Bruyn, A. 1994. Noun phrases. In Pidgins and Creoles: An introduction, J. Arends, P. Muysken & N. Smith (eds), 259–288. Amsterdam: John Benjamins. Bruyn, A. 1995. Grammaticalization in Creoles: The development of determiners and relative clauses in Sranan [Studies in Language and Language Use 21]. Amsterdam: IFOTT. Carlson, G. 1977. Reference to Kinds in English. PhD Dissertation, University of Massachusetts, Amherst (published 1980, Garland Press, New York). Chaudenson, R. 1981. Textes Créoles Anciens (la Réunion et Ile Maurice). Comparaison et Essai d’Analyse. Hamburg: Buske. Chaudenson, R. 1994. Créolisation et appropriation linguistique: De la théorie aux exemples. In Créolisation et Acquisition des Langues, D. Véronique (ed.), 171–190. Aix-en-Provence: Publications de l’Université de Provence. Chaudenson, R. 2001. Creolization of Language and Culture. London: Routledge. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton. Chomsky, N. 1995. The Minimalist Program. Cambridge MA: The MIT Press. Chomsky, N. 1999. Derivation by Phase [MIT Occasional Papers in Linguistics]. Cambridge MA: MITWPL. Chomsky, N. 2000. Minimalist inquiries: The framework. In Step by Step: Essays in minimalist syntax in honor of Howard Lasnik, R. Martin, D. Michaels & J. Uriagereka (eds), 89–155. Cambridge MA: The MIT Press. Chomsky, N. 2001. Beyond explanatory adequacy. In MIT Occasional Papers in Linguistics 20. Cambridge MA: MIT, Department of Linguistics. Chomsky, N. 2002. On Nature and Language. Cambridge: CUP. Dayal, V. 2002. Number marking and (in)definiteness in kind terms. Linguistics and Philosophy 27: 393–450. Dayal, V. To appear. Semantic variation and pleonastic determiners: The case of the plural definite generic. To appear in the proceedings of Asian GLOW 5. DeGraff, M. 2001. Morphology in creole genesis: Linguistics and ideology. In Ken Hale: A life in language, M. Kenstowicz (ed.), 53–121. Cambridge MA: The MIT Press. Déprez, V. 1999. De la nature sémantique des nominaux sans déterminant en créole haitien. Langues 2(4): 289–300. Déprez, V. 2001. On the nature of Haitian Bare NPs. In Current Issues in Romance Languages, T. Satterfield, C. Tortora & D. Cresti (eds), 48–61. Amsterdam: John Benjamins. Déprez, V. 2003. Determiner architecture and phrasal movement in French Lexifier creoles. In Romance Languages and Linguistic Theory 2001: Selected Papers from Going Romance, Amsterdam 6–8 December 2001, J. Quer, J. Schroten, M. Scorretti, P. Sleeman & E. Verheugd (eds), 49–74. Amsterdam: John Benjamins. Déprez, V. 2005. Morphological number, semantic number and bare nouns. Lingua 115(6): 857–883. Diesing, M. 1992. Indefinites. Cambridge MA: The MIT Press. Dijkhoff, M. 1983. The process of pluralization in Papiamentu. In Studies in Caribbean Language, L. Carrington (ed.), 217–229. St. Augustine, Trinidad: Society for Caribbean Linguistics. Gadelii, K. E. 1997. Lesser Antillean French Creole and Universal Grammar. PhD Dissertation, Göteborg University. (Göteborg: Gothenburg Monographs in Linguistics 15).

Chapter 1. Noun phrases in creole languages Givón, T. 1984. The pragmatics of referentiality. In Georgetown University Round Table on Languages and Linguistics 1984. Meaning, form, and use in context: Linguistic applications, D. Schiffrin (ed.), 120–138. Washington DC: Georgetown University Press. Guéron, J. & Hoekstra, T. H. 1988. T-chains and the constituent structure of auxiliaries. In Constituent Structures, A. Cardinaletti, G. Cinque & G. Giusti (eds), 35–99. Venice: Annali di Ca’ Foscari. Guéron, J. 2006. Generic sentences and bare plurals. In Proceedings of the Brussels Conference on Indefinites and Weak Quantifiers, S. Vogeleer & L. Tasmowski (eds), 219–234. Amsterdam/ Philadelphia: John Benjamins. Holm, J. 1990. Features in the noun phrase common to the Atlantic creoles. Linguistics 4: 867–81. Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge MA: The MIT Press. Lambrecht, K. 1994. Information Structure and Sentence Form – Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: CUP. Lecarme, J. 1996. Tense in the nominal system: The Somali DP. In Studies in Afroasiatic Grammar, J. Lecarme, J. Lowenstamm & U. Shlonsky (eds), 159–178. The Hague: HAG. Lecarme, J. 2004. Tense in nominals. In The Syntax of Time, J. Guéron & J. Lecarme (eds), 440–475. Cambridge MA: The MIT Press. Longobardi, G. 1994. Reference and proper names: A Theory of N-movement in syntax and logical form. Linguistic Inquiry 25(4): 609–665. Lyons, C. 1999. Definiteness. Cambridge: CUP. Marchello-Nizia, C. 1979. La Langue Française au XIVe et XVe Siècle. Paris: Bordas. McWhorter, J. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74: 788–818. Moñino, Y. In press. Convergencias lingüísticas iberocongolesas en palenquero: ¿integrarse a la sociedad mayoritaria o distinguirse de ella? To appear in Language Contact and Language Change in the Caribbean and Beyond — Lenguas en Contacto y Cambio Lingüístico en el Caribe y Más Allá, W. Mihatsch & M. Sokol (eds). Bern: Peter Lang. Mufwene, S. S. 1986. Number delimitation in Gullah. American Speech 61: 33–60. Mufwene, S. 2001. The Ecology of Language Evolution. Cambridge: CUP. Mufwene, S. 2003. Genetic linguistics and genetic creolistics: A Response to Sarah G. Thomason’s Creoles and Genetic Relationships. Journal of Pidgin and Creole Languages 18(2): 273–288. Munn, A. & Schmitt, C. 2001. Bare nominals and the morphosyntax of number. Current issues in Romance Linguistics, T. Satterfield, C. Tortora & D. Cresti (eds), 217–231. Amsterdam: John Benjamins. Partee, B. 1987. NP Interpretation and type-shifting principles. In Studies in Discourse Representation Theory and the Theory of Generalized Quantifiers, J. Groenendijk, D. de Jongh & M. Stokhof (eds), 115–143. Dordrecht: Foris. Sankoff, G. & Mazzie, C. 1991. Determining noun phrases in Tok Pisin. Journal of Pidgin and Creole Languages 6(1): 1–24. Schleicher, A. 1853. Die Ersten Spaltungen des Indogermanischen Urvolkes. Allgemeine Zeitung fuer Wissenschaft und Literatur. Schmitt, C. & Munn, A. 1999. Against the nominal mapping parameter: Bare nouns in Brazilian Portuguese. In Proceedings of NELS 29, P. Tamanji, M. Hirotani & N. Hall (eds), 339–353. University of Delaware.





Marlyse Baptista and Jacqueline Guéron Schmitt, C. & Munn, A. 2003. The syntax and semantics of bare arguments in Brazilian Portuguese. Linguistic Variation Yearbook 2: 185–216. Schuchardt, H. 1884. Slavo-deutsches und Slavo-Italienisches. Graz: Leuschner and Lubensky. Singler, J. 1989. Plural marking in Liberian Settler English, 1820–1980. American Speech 64(1): 40–64. Singler, J. 1994. Plural marking in Sinoe Settler English. Paper presented in a joint meeting of the Society for Pidgin and Creole Linguistics and the Society for Caribbean Linguistics, University of Guyana. Tagliamonte, S., Poplack, S. & Eze, E. 1997. Plural marking patterns in Nigerian Pidgin English. Journal of Pidgin and Creole Languages 11(1): 103–129. Thomason, S.-G., & Kauffman, T. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley CA: University of California Press. Thomason, S.-G. 2002. Creoles and genetic relationship. Journal of Pidgin and Creole Languages 17(1): 101–109. Vallduví, E. 1993. The Informational Component, PhD Dissertation. Philadelphia, University of Pennsylvania. Valli, A. 1994. A propos de l’emploi productif de la détermination zéro en moyen français et en créole réunionnais. In Créolisation et acquisition des langues, D. Véronique (ed.), 89–101. Aix-en-Provence: Presses de l’Université de Provence. Winford, D. 2003. An Introduction to Contact Linguistics. Malden MA: Blackwell. Wood, J. L. 2003. Definiteness and Number: Determiner phrase and number phrase in the history of English. PhD Dissertation, Arizona State University.

 

Portuguese-lexified Creoles

 

Bare nouns and the nominal domain in Santome1 Nélia Alexandre and Tjerk Hagemeijer2 This paper investigates the role of BNPs within the nominal system of Santome, the major Gulf of Guinea Creole. In the first part of the paper, we sketch an analysis of the DP focusing on the syntax and semantics of the modifiers that express definiteness, number and specificity and their interaction. In the second part, it will be shown that BNPs fulfil a wide range of semantic functions in virtually all syntactic positions, predicative and argumental. Therefore, we propose a unified hypothesis that accounts for both lexicalized and null determiners as an integrated system. It is argued that all determiners are inherently specified for definiteness and number. However, we consider number parasitic on definiteness (under D0) and, therefore, Santome’s syntax does not exhibit an independent projection for number. It is further shown that, especially in the case of BNPs, discourse anchoring and pragmatics are crucially in order to derive the correct interpretation.

.

Introduction

In what follows we will discuss the properties of bare noun phrases (BNPs), i.e. noun phrases without overt determiners. We will show that BNPs are at the core of the nominal domain in Santome (ST), one of the four Portuguese-based Gulf of Guinea Creoles, better known in previous work as São-Tomense. Especially since Bickerton’s (1981) Bioprogram, the article system of Creole languages has become an important aspect of Creole typology. Bickerton claimed that these languages typically exhibit a threefold determiner system based on the following

. We would like to thank Ana Lúcia Santos, Nuno Soares, both organizers of this volume and especially Inês Duarte for their comments on this paper. . The authors are respectively funded by doctoral grant BD/13536/2003 and BD/3159/2000 of the Foundation for Science and Technology and the European Structural Fund within the IIIrd Community Support Framework, Portugal.



Nélia Alexandre and Tjerk Hagemeijer

cognitive elements: presupposed-specific (definite article), asserted-specific (indefinite article) and nonspecific (null determiners, i.e. BNPs). Ever since, many influential syntactic and semantic studies on noun phrases (and more recently on BNPs as well) have been published. In syntax, Abney’s (1987) long-lasting proposal of the Determiner Phrase (DP) layer (until then NP) can be considered the most influential one and is crucial to the current research on BNPs. In several respects, ST’s nominal domain does not match Bickerton’s ideal system, as already noted by Lucchesi (1993). BNPs in ST (and other Creoles), for instance, can be found in all syntactic positions and appear to be semantically underspecified for definiteness and number: they lack overt grammatical/lexical information to identify these features. As so often, linguistic generalizations fall somewhat apart in the light of language-specific evidence. This paper therefore has a twofold aim: (i) to provide a description of ST’s badly known nominal domain and (ii) to propose a DP structure that accounts for the full range of data under analysis. Our first assumption is that from a typological point of view, it is tempting to consider BNPs reminiscent of the early stages of creolization, characterized by drastic language restructuring. Especially if we postulate an initial pidgin (or non-native precreole) stage in which nouns must have been the essence of identifiability (a less functional stage), it is not surprising that BNPs in ST still exhibit a range of interpretations such that they have to be central to the language’s nominal domain. It is not unlikely that BNPs became more restricted due to the appearance of new overt functional material at the time the pidgin developed into a full-blown language. Like other Creoles, particularly the ‘old’ ones, ST lacks historical records that could provide insights into the historical development of BNPs.3 Better documented Creoles, as for instance Sranan, a Surinam Creole, show that the number of bare nouns in the early stages of this language was much higher than it is nowadays (cf. Bruyn 1994, 1995). For the discussion of BNPs, we will set apart two interpretational levels: sentence grammar and the organization of a sentence within a discourse (universe) (cf. Lambrecht 1994: 7). We assume that sentence grammar is an autonomous grammatical level where syntax and semantics interact, whereas discourse is an extra-grammatical level concerned with ‘information packaging’ in the sense of Vallduví (1993). We assume with this author that “the successful interpretation of sentences requires not only the interpretation of logico-semantic meaning (…) but also, at least, the interpretation of informational meaning” (Vallduví 1993: 12–13). The following example illustrates how these two levels interact with each other: (1)

Alê d’e sode. Sode be. king give-3 soldier / soldier go ‘The king supplied him with soldiers. The soldiers went.’

. ST is a 16th century Creole (cf. Ferraz 1979; Lorenzino 1998; Hagemeijer 1999) but as far as we know the earliest records of written Santome date back to the second half of the 19th century.

Chapter 2. Bare nouns and the nominal domain in Santome

This example has two occurrences of a BNP sode ‘soldier(s)’. At the grammatical level, BNPs in object position in ST provide new information and are typically interpreted as plural indefinites, whereas BNPs in subject position are preferentially interpreted as singular definites (as exemplified by alê ‘king’ in subject position of the first sentence), if this definiteness was previously anchored in some way.4 Sode in subject position in (1) would thus typically read as a singular definite but in this case it is anaphorically anchored by the preceding sentence. To this discourse level, as well as pragmatics, we will refer as the extra-grammatical level that is required to obtain a correct interpretation of BNPs. For the present purpose we adopt Lambrecht’s (1994: 79) definition of the category of definiteness, which is “(…) a formal feature associated with nominal expressions which signals whether or not the referent of a phrase is assumed by the speaker to be identifiable to the addressee.” In this sense, we assume that identifiability is the cognitive counterpart of definiteness and interfaces the grammatical and extra-grammatical level. In our proposal for ST, definiteness is the nuclear formal feature, hosted under a DP. In addition to definite determiners, in many cases languages also have the means to overtly encode number marking. ST exhibits no inflectional and exclusive overt number marker, but given the importance of number for cognitive processing we will argue that in ST this feature is directly associated to definiteness, although it has to be somehow defined outside the grammatical level referred to above when we are dealing with BNPs. Assuming that languages may or may not exhibit split I (cf. Pollock 1989) or a split left Periphery (Rizzi 1997), the absence of an independent Number Phrase (NbP) would simply correspond to a non-split DP. Under our claim that null determiners should receive the same treatment as lexical determiners, BNPs can be fully integrated in the nominal structure we will propose. Section 2 and 3 focus on the encoding of definiteness, number and specificity and in Section 4 we will discuss several properties of BNPs. The data presented in this paper were drawn from a corpus of over 20 hours of tape-recorded and transcribed spontaneous speech (mostly folk stories), as well as elicited speech.

.

Definiteness and number

This section discusses the two determiners which have a phonetic counterpart: singular indefinite and numeral wa ‘a(n), one’ and inen, a definite plural marker corresponding roughly to English ‘the’.5 The feature specification of wa can be defined as [−definite,

.

This reflects, for instance, the Cape Verdean facts (cf. Baptista 2002; this volume).

. Va ‘a(n), one’ comes from Portuguese ‘uma’, with the same meaning. The etymology of inen has not been permanently settled but, considering that Edo (Nigeria) is the oldest and most significant substrate language which gave, for example, rise to an impersonal pronoun in the





Nélia Alexandre and Tjerk Hagemeijer

−plural], whereas inen has exactly the opposite feature specification, i.e. [+definite, +plural] (cf. also Schang 2000: 219–222). Descriptively, wa introduces new information in any syntactic environment (cf. Lucchesi 1993). Its plural counterpart is quantifier wa dôsu ‘some’. It is, however, worth noting that it is typically BNPs that yield plural indefinite reference, providing BNPs occur in a lexically governed position and constitute new information (cf. Section 4). (2)

a. b.

c.

Wa ome/bwê… ‘a/one man/cow…’ Wa dôsu ome/bwê… one two man/cow ‘some men/cows…’ Zon bê vunvu. Zon see bee ‘Zon saw bees.’

Unlike Upper Guinea Creoles (cf. Baptista 2002; Kihm 1994) and Portuguese, ST (and the remaining Gulf of Guinea Creoles) does not exhibit any overt plural markers suffixed to the noun. Just like wa, inen is inherently specified for number and its singular counterpart is a BNP. Examples of nouns that are exclusively modified by inen (i.e. without any other material modifying the noun, like possessives, relative clauses, etc.), as in (3–5), are relatively scarce in our corpus. (3)

~ Ua notxi inen kôlômba pega sun Faxiku one night . white catch Mr. Faxiku ku migu sun, sun Zoxi Mandjinga and friend Mr. Mr. Zoxi Mandjinga. ‘One night, the white caught Mr. Faxiku and his friend Mr. Zoxi Mandjinga.’

(4)

Inen funxônariu ska bi golo pixi blatu ô. . functionary ASP go search fish cheap  ‘The employees are searching for cheap fish.’

(5)

Ola ku mina be, mina da ku inen mosu. when  girl go girl give with . boy ‘When the girl left, she ran into the boys.’

The sequence of inen+noun always occurs with nouns specified as [+human]. Inen typically conveys a definite meaning with respect to a non-individualized (or

Gulf of Guinea Creoles, candidates are: (i) Edo 3rd person plural weak and strong pronoun iran ‘they’, since inen is homophonous with – and perhaps even derived from – the 3rd person plural pronoun, an overlap of functions also found in other Creole languages, (ii) Edo definite determiner (n)ene ‘the’, or (iii) Edo demonstrative pronoun ene/enena ‘these’, apparently related to (ii) (cf. Agheyisi 1986, 1990; Melzian 1937). Given our insufficient (historical) knowledge of Edo, we will leave this issue as it is.

Chapter 2. Bare nouns and the nominal domain in Santome

non-atomic) group familiar to hearer and speaker. This knowledge rests upon previously introduced information in Example (5) or on shared knowledge of the world in Examples (3–4). The plural marker shows a preference for nouns with the feature [+human], which we consider a consequence of its sensitivity to a semantic principle that takes [+human] as being more individuated than items that are [−human] (cf. Baptista 2003; this volume and Kihm, this volume for similar facts). An additional argument for the definite reading of inen comes from the fact that generic interpretations are blocked by this determiner. (6)

Inen (ngê) santome sêbê kanta. . (people) S. Tomé know sing Interpr. 1. ‘The people from S. Tomé sing well.’ (group of people familiar to the speaker; no generic interpretation allowed) Interpr. 2. *‘People from S. Tomé sing well.’ (each and every person from S. Tomé; generic reading)

Note further that the generic reading of (6b) is readily obtained in the absence of inen (Ngê santome sêbê kanta ‘People from S. Tomé sing well’). The corpus exhibits a much higher number of nouns modified simultaneously by inen and modifying items that contribute to specificity, like for instance relative clauses, adjectives or possessives, that all occur in the post-nominal field. Particularly frequent is the co-occurrence of inen with standard specific marker () se (cf. Section 3), which narrows down the referentiality. (7)

So inen ja se so sun Fenanji na ska so . day   Mr. F.   txila vin fa. extract wine  ‘So these days Mr. Fenanji isn’t extracting palm wine.’

(8)

Kê kwa ku bô mêsê pa a fe inen manu what thing  2 want for  do . brother se bô?   ‘What do you want them to do to your brothers?’

This higher degree of referentiality as the result of the co-occurrence of the definiteness marker and the specific marker is actually very common with [−human] nouns and therefore sharply contrasts with the absence of inen+noun[−human] without additional specific material (cf. (3–6)).6 Interestingly, one of the main uses of inen is found in direct speech addressing a number of hearers with the meaning ‘you’ (cf. non-standard English ‘them guys’, etc.), . In some cases, the high frequency of non-human nouns modified by inen and se resulted in crystallized expressions, which show the intrinsic relation between these two elements and the noun.





Nélia Alexandre and Tjerk Hagemeijer

as illustrated in (9–10). These two examples, together with the frequent co-occurrence of specific marker se (cf. (7–8) and Section 3), show that inen might have evolved from the demonstrative domain (cf. Footnote 5). (9)

(10)

Ola inen sun ka bi, n ga pô fla pikina when . r.  come 1  can talk little da inen sun. give . you ‘When you (formal) come back, I can talk a little bit to you.’ Inen

ke mu, sa pingada fan! house mine is gun  ‘My friends (people at my place), it’s a gun!’ DEF.PL

The referred asymmetries with respect to plural marking on [+/−human] nouns are confirmed by, for instance, dislocation tests. Pluralized [−animate] items are typically co-indexed with a 3rd person singular pronoun (cf. (11)), although there seems to be some variation among speakers and syntactic environments, whereas [+animate] dislocated constituents are obligatorily co-indexed with a 3rd person plural pronoun (cf. (12–13)). (11)

Inen zanela se, bô {fis’e/fis’inen/*fisa}. . window  2 {close-3/close-3/close} ‘The windows (in question), you closed them.’

(12)

Inen plôkô se, Zon {mat’inen/*mat’e/*mata}. . pig  {kill-3/kill-3/kill} ‘These pigs, Zon killed them.’

(13)

Inen ome se ala, Zon {bê inen/*bê’lê/*bê}. . man  there Zon {see-3/see-3/see} ‘The men (in question) over there, Zon saw them.’

Although the data above show ST’s sensitivity to degrees of animacy in the identification of nouns, we assume that its relevance is limited for the structural outline of the nominal domain in ST. The same goes for gender marking.7 (i)

a.

Inen

b.

. thing  ‘these things, this’ Tudu inen kwa se all . thing  ‘All these things, all this, everything’

kwa se …

. Not surprisingly, gender is actually dependent on animacy. Human nouns generally have a lexical gender distinction, as in tiu/tia ‘uncle/aunt’, donu/dona ‘grandpa/grandma’, ome/mwala ‘man/woman’. Professions show that gender is indeed available: infermêru/infermêra ‘male nurse/ nurse’, kuzinhêru/kuzinhêra ‘male cook/female cook’, dôtôlô/dôtôra ‘male doctor/female doctor’.

Chapter 2. Bare nouns and the nominal domain in Santome

In this section we have described the properties of the two lexical determiners wa and inen. These two items carry number and definiteness specification. Definiteness, as the formal instantiation of identifiability, links sentence level to discourse level, whereas number follows directly from this feature when determiners are overt. Therefore, we assume that a separate functional category for NbP is not required and not even warranted, given the absence of plural affixation. Our claim is that the two definiteness markers described in this section, wa and inen, come from the lexicon with an inherent number feature. The claim that a language may develop definiteness before number is argued for by, for instance, Wood (2003), who claims that Old English is different from Modern English in the sense that it exhibits a Determiner Phrase but no Number Phrase as yet. Furthermore, Longobardi (1994: 621) assumes that number is checked in D0 and claims that “(…) arguments but not predicates crucially rely on the D position in order to define their meaning with respect to number”.

.

Specificity

In addition to definiteness and number, ST disposes of one specificity marker se ‘the, this, these’8 with spatialized force (in the sense of Zribi-Hertz & Glaude, this volume) referring to specific objects in the world that represent shared knowledge of speaker and hearer.9 “Ferraz (1979: 73–4) refers to se as primarily a demonstrative that also fulfils the function of a definite article. A similar approach is argued for by Lucchesi (1993) and Schang (2000). Lucchesi additionally claims that se in its definite use serves to emphasize the noun and is therefore not a core element in Santome’s DP. These assumptions are contradicted by our analysis. Both authors further emphasize the spatialized force of se, which Schang splits up in an anaphoric (discursive) and deictic (situational) use of demonstrative se.

If a profession is typically feminine (modista/*modistu ‘dressmaker’) or masculine (flêlu/*flêla ‘smith’) the oppositions are not available. Animate nouns, on the other hand, may usually carry a male/female suffix (ome/mwala ‘male/female’), as in plôkô ome/plôkô mwala ‘male pig/female pig’. As expected, non-animate nouns do not exhibit any gender marking. . There are two possible sources for se’s etymology: (i) Portuguese demonstrative esse ‘that’, although, unlike se, in Portuguese it occurs canonically in prenominal position, or (ii) the contraction of the expression sa ai ‘is here’, which parallels sala ‘that’ < sa ala in ST ‘is there’ (cf. Ferraz 1979: 74) and say ‘this’ in closely related Fa d’Ambô (cf. Barrena 1957). . Santome has two other, less commonly used spatializing markers, xi ‘that (less proximate)’ and sala ‘that (distant)’. The nominalization of se, xi and sala yields respectively ise/isaki ‘this (here)’, ixi ‘that’ and isala ‘that’.





Nélia Alexandre and Tjerk Hagemeijer

Despite the well-attested historical development of definite articles from demonstratives across languages, we claim that se in ST should not be considered a definite article because it is able to co-occur with the indefinite and definite markers discussed in Section 2 (wa/inen  se) and, on the other hand, it will be shown in Section 4 that BNPs typically fill in the definite feature.10 In order to emphasize the particular status of se, we will refer to it as a specificity marker (). Se always occurs to the immediate right of the lexical item it modifies: pronouns11 (14), all types of nouns (15–17) and numerals (18). (14)

…punda non se na

tê

zêtê

doxi fa.

because 1   have olive-oil sweet  ‘…because we (in question) don’t have sweet olive oil.’ (15)

 mêsê pa men jê mina awa se da 1 want for mother fetch little water  give anzu se baby  ‘I want you (here: affective way to address a woman) to bring a bit of that water (in question) for the baby (in question).’

(16)

Kuma non lentla setembru se en sa non as 1 enter September   be 1 sa ni tempu suba za. be in time rain already ‘Since it’s September, we are already in the rainy season.’

(17)

San se ê, fe mu favôlô axi an. miss   do me favor such  ‘Oh miss, could you do me a favor…’

. Some languages typically exhibit definite articles and demonstratives in complementary distribution (cf. English *the this/*this the, Portuguese *o este/*este o, etc.). In Edo, a crucial substrate language in the making of ST, however, both determiners may co-occur.

(i)

Néne èbè mwén nii (Agheyisi 1990: 85)  book   ‘That particular book of mine.’

Notwithstanding some similarities with ST’s determiner system, Edo is for instance different in that possessive modifiers obligatorily precede any other determiner. Hence it is untenable that demonstrative nii ‘that’ in (i) is a nominal clitic like ST se. Nevertheless, it should be noted that Angolar, a maroon Creole closely related to ST, exhibits a less restricted distribution of se (cf. Maurer 1995). . Note that etymologically the 2nd person plural (i)nanse~(i)nense may have its origin in inen+se ‘they+se’, i.e ‘you (pl.)’, establishing proximity with respect to the point of view of the speaker (cf. Examples (9–10)).

Chapter 2. Bare nouns and the nominal domain in Santome

(18)

Sêxi se ku n fla bô. six   1 say 2 ‘The six I told you about.’

In the light of the available data, we consider se a clitic that requires a nominal host. Several arguments support this hypothesis. In the first place, the relative position of se with respect to numerals shows a strong adjacency requirement. Numerals may occur in several different syntactic positions in order to derive partitive-focus readings (19b–c). However, in no circumstances may numerals intervene between se and its host (19d). (19)

a.

b.

c.

d.

Inen dôsu mosu se… . two boy  ‘The two boys…’ Dôsu inen mosu se… two . boy  ‘Two of the boys…’ Inen mosu se dôsu… . boy  two ‘These/those two boys…’ Inen mosu *(se) dôsu (*se). . boy () two () ‘The two boys in question’

Furthermore, se requires an overt host with nominal features (20), whereas plural marker inen and quantifier wa allow for gapping of the noun and behave nominally. This can be related to the impossibility of stranding or fronting of the specific marker in (21b) and (21c) respectively. (20)

Zon paga *(mwala) se. Zon pay woman  ‘Zon payed the woman in question.’

(21)

a.

Mwala se, Zon pag’e. woman  Zon pay-3 ‘The woman, Zon payed her.’ b. *Mwala, Zon {pag’e/paga} se. woman Zon {pay-3/pay}  c. *Se so Zon paga mwala [-].   Zon pay woman

A final piece of evidence comes from compounding. The reanalysis of two X0 as a single one predicts correctly that it should not have internal structure. Hence, the clitic can only attach to the right – and crucially not to the intermediate position – of the newly coined word in the Examples (22a–c). (22)

a.

[[[boka] [sabi]] se] Mouth key  ‘The keyhole in question’





Nélia Alexandre and Tjerk Hagemeijer

b.

c.

[[[kwa] [kume]] se] Thing eat  ‘The food in question’ [[[kota] [bega]] se] Cut belly  ‘The last-born in question’

The relevant contrast is with (23a–c), which may resemble compounds due to juxtaposition of bare nouns without any overt mediating case marker. The crucial difference with the examples in (22) is that the clitic, as expected, may modify the head noun, its complement, or even both if one would like to. (23)

a.

b.

c.

[[opo] se] [(di) glavana] dust  (of) dry.season ‘The dry season’s dust in question’ [[Fenanji] se] [(di) Mate Ngola] Ferdinand  (of) Mateus Angolar ‘Ferdinand from Mateus Angolar’ [[klonveson] se] [doxi-doxi] talk  sweet.sweet ‘The very nice talk in question’

Whenever nouns are exclusively modified by se, they are typically interpreted as singulars, i.e. individualized items. This is illustrated in the following sequences for the subject and object position. (24)

 konsê dexi mosu. Ua ja, *(inen) 1 know 10 boy / one day (.) mosu se ba skola. boy  go school ‘I know 10 boys. A certain day, the(se) boys went to school.’

(25)

 konsê dexi mosu. Ua ja, sode 1 know 10 boy / one day soldier d’*(inen) mosu se plamatoya. give(-.) boy  ferule ‘I know 10 boys. A certain day, the soldier(s) hit the boys with the ferule.’

However, Example (26) shows that N+se may yield a taxonomic reading (i.e. referring to a subspecies) with non-human nouns due to the extension of the specific reading (26a) to a kind-denoting reading that quantifies over all the atoms of a type or subspecies (26b), but never over the whole species the subset belongs to (26c). (26)

Gweva se na bwa fa. guava   good  a. ‘The guava in question is no good.’ b. ‘The type of guava in question is no good.’ c. *‘Guavas are no good.’ (generic interpretation disallowed)

Chapter 2. Bare nouns and the nominal domain in Santome

This extension from a particular to a taxonomic reading (subkind interpretation) is possible for [−human] nouns because of their opacity to the kind of individualization generally associated to [+human] nouns. In the light of the systematic occurrence of the specific marker reinforcing the definiteness feature (e.g. inen++se, cf. (7–8)), we argue that the nominal domain requires the projection of a Specific Phrase (SpP). The interpretation of nouns modified by se provides yet another piece of evidence for the strong link between definiteness and number, since they typically trigger a singular reading. The presence of inen is required to turn this singular reading into a plural one. At this point, we will propose a syntactic structure based on the summarized description of the core elements with phonetic content in ST’s nominal domain. Examples (27a–b) yield the representation in (28).12 (27) (28)

a. b.

Ua mina se ‘a child in question’ Inen mina se ‘the children in question’ DP D’ D0

SpP ũa inen

Sp’ Sp0

minai

NP se

ti

Se is the head (a X0) of SpP, a node that only projects whenever there is lexical evidence. In the example, the noun mina left-adjoins to clitic se. As we saw, this clitic has a default singular reading associated to it. The properties of se appear to be the very same as those (optional) elements that reinforce identifiability in many other languages, for instance French (Ces livres ci ‘these books’), Cape Verdean (Kes omi(s)-li ‘the men in question’), perhaps also so-called double definiteness in Swedish with an additional enclitic definite article (Det store huset ‘the big house’) or non-standard English (These guys here). Like se, these elements contribute to deixis and are invariable.

. In the light of the available data, the following arguments lead us to claim that the DP structure in Santome is right-branching: (i) absence of postpositions, (ii) relative clauses, adjectives and possessives occur to the right of the noun, (iii) compounds are head-initial, (iv) economy principles of representation.





Nélia Alexandre and Tjerk Hagemeijer

In the referred languages, it is a well-known fact that these reinforcing elements typically co-occur with demonstratives (cf. examples above) but crucially not with determiners ({cette/*la} femme là13). Giusti (1993) proposes that demonstratives and reinforcers are in a Spec-Head configuration of a lower projection and that the former always raises to [SPEC, DP]. A slightly different hypothesis is proposed in Bernstein (1993), who argues that demonstratives and definite determiners are heads of DP, which would explain their mutual exclusiveness (*the this man / *this the man). We assume that wa and inen are both heads of DP. As mentioned, inen is homonymous with the 3rd person plural pronoun, which is further evidence for its D0 status. Besides, inen selects nominal complements, a property of D heads. As to wa, we assume that both the existential reading as the head of a Quantifier Phrase (cf. Giusti 1991) and the indefinite reading as the head of a DP are available. As determiners, inen and wa directly c-command their complement, NP and SpP if projected. Hence no additional noun movement is required. The proposed structure also has the advantage that in a bottom-up phrase structure specificity leads to definiteness, but not the other way around. We assume that the tree structure in (28) is a generalized structure for DPs in ST. Further evidence comes from predicates of copula clauses and vocatives. The defectiveness of these constructions in other languages has been taken as evidence that they are instances of NP due to the absence of, for instance, overt determiners and/or the impossibility to take relative clauses (Abney 1987 and later work by Longobardi, among others). Therefore DP would not project. In a language like ST, however, these restrictions do not apply. Example (29) shows a bare nominal predicate that takes a relative clause and (30) is a fully specified vocative for definiteness and specificity. (29)

Sabinu sa flêgadô ku tlaba ni Mansu. Sabinu is massager  worked in Mansu ‘Sabinu is a massager who worked in Mansu.’

(30)

Inen san se ê, toma kopu da mu. . lady   take glass gives me ‘Hey ladies (in question), bring me a glass.’

It follows that in ST non-arguments may project as high as DP, an issue we will not pursue in the present venue. On the other hand, it should be noted that proper nouns in ST never bear overt determiners but the DP projection is still required to interpret its definiteness (cf. Longobardi 1994), as for example in ST’s lexifier, Portuguese, where proper nouns typically have an overt counterpart of D0 (O João ‘John’, lit: the John). In the next section we will investigate whether and how this analysis can be extended to the properties of BNPs in ST.

. Marlyse Baptista (p.c.) believes however that la femme là is possible in certain colloquial registers of the French language.

Chapter 2. Bare nouns and the nominal domain in Santome

.

Bare nouns

The interpretational and distributional richness of BNPs in ST empirically underlines their key position in ST’s nominal domain. In this section we will provide an overview of this richness and set forth the hypothesis that BNPs can only be licensed on the interface between a grammatical and an extra-grammatical level that comprises discourse and pragmatics. The examples in (31a–d) show the interpretation of subject BNPs with different degrees of animacy (cf. Section 2). One should not forget that, despite the generalized availability of definite interpretations of BNPs, they have to be previously licensed in discourse to become part of the hearer’s knowledge. (31)

a.

b. c.

d.

e.

Sangê/Sungê kopla pixi. lady/mister buy fish ‘The lady/man (formal) bought fish’ (Definite sg.) Mwala kopla pixi. ‘The woman/women bought fish.’ (Def. sg./pl.) Bisu vwa ba liba po. bird fly go up tree ‘The bird(s) flew into the tree(s).’ (Def. sg./pl.) Floli môlê. flower die ‘The flower(s) died.’ (Def. sg./pl.) Inen {sangê/mwala/??bisu/*floli} sa glavi. . {lady/woman/bird/flower} be beautiful ‘The {ladies/women/birds/flowers} are beautiful.’

Considering that subjects are in a topic position, the definite feature that cuts through all classes of animacy can be straightforwardly explained away. Subject BNPs are not so well behaved with respect to number though. The respectful forms to address a woman or a man, sangê and sungê, are always interpreted as singulars (cf. 31a). Their special status in the language also follows from the fact that both have a specific anaphoric form (san and sun). Although mwala ‘woman’ in (31b), as well as other human BNPs can be interpreted as singulars or plurals, they generally occur with the singular interpretation in the corpus. The non-human BNPs in (31c–d) occur much more frequently with a plural reading. The fact that the singular and plural readings are easily obtained by non-human BNPs correlates nicely with their impossibility to take determiner inen (cf. (31e)), except in some very marginal cases with [+animate] nouns. As mentioned in Section 2, we suspect this is because inen is also the 3rd person plural pronoun and did make its way through into the non-human domain. It is not unlikely that inen is extending its properties to the non-human domain. The examples with BNPs in object position (32) crucially differ from the examples in (31) due to the fact that this position is typically associated to new information (also see Baptista 2002; this volume and Kouwenberg 1994; this volume). Whenever BNPs





Nélia Alexandre and Tjerk Hagemeijer

in object position constitute new information, they have a typical indefinite plural reading, except, once again, for the special cases of sangê and sungê (cf. (32a)). Just like in (31), definite readings have to be somehow previously licensed and this licensing determines whether object BNPs have a singular or a plural definite reading. (32)

a.

b.

c.

Zon bê sangê/sungê ni losa. Zon see lady/mister at plantation ‘Zon saw the lady/the man (formal) at the plantation.’ (Def sg.) Zon bê {mwala/bisu/floli} ni losa. Interpr. 1 ‘Zon saw {women/birds/flowers} at the plantation.’ (Indef pl.) Interpr. 2 ‘Zon saw the {woman/bird/flower} at the plantation.’ (Def sg.) Interpr. 3 ‘Zon saw the {women/birds/flowers} at the plantation.’ (Def pl.) Zon bê inen {sangê/mwala/*bisu/*floli} ni losa.

Crisma 1997 (apud Longobardi 2001: 585) states that if languages have BNPs for a specific function you do not expect an overt determiner to fill in this very same function. This correlation appears to be correct with respect to indefinite singular determiner wa, considering ST does not exhibit BNPs with a singular indefinite reading. As to inen, we have shown that things are a little more complicated but still Crisma’s observation seems to hold. In the absence of other modifying material, inen only selects human nouns. Given human nouns’ preference for singular readings, determiner inen fills in the plural definite function. Other environments where BNPs typically occur are generic sentences and light verb constructions14 in (33) and (34) respectively. (33)

Maji vinpema ka fe ome mali. but palm-wine  do man bad ‘But palm wine is harmful for {men/a man}.’

(34)

Sun da pema faka. Mr. give palmtree knife ‘The man (formal) cut an opening in the palmtree.’ (preparation for palm-wine extraction)

It could very well be the case that the determiner system in ST is still developing.15 If we assume that BNPs are more primitive than overt determiners, it would not be unexpected that, as part of information structure, they became initially associated to singular definite readings in subject position and to plural indefinite readings in governed positions. Languages that have not (fully) developed a lexical definite determiner would then automatically lend definite interpretations to BNPs, whereas languages that did develop a full-blown definite determiner may start dismissing the definite

. For an overview of the properties of light verb constructions we refer the reader to Butt & Geuder (2001). .

Cf. Baptista (2002; this volume) for the same claim with respect to Cape Verdean.

Chapter 2. Bare nouns and the nominal domain in Santome

reading of BNPs (e.g. Haitian Creole has a definite marker la (cf. Lefebvre 1998, among others) but no definite bare singulars (Déprez 2001)).

.

Mass and count nouns

Lexical items come from the lexicon with a [count] feature: [+count] nouns (countable, e.g. woman, bird, flower) and [-count] nouns (mass, e.g. water, furniture). The distinction between both nouns is related to number: count nouns can be atomized, whereas mass nouns cannot. We will assume that DP projects for both types of nouns because they occur in argument positions (cf. Longobardi 1994 and later work). Additional evidence for the fact that mass nouns are DPs comes from the availability of definite readings. (35)

Awa sa bwa. water is good a. ‘Water is good.’ b. ‘The water is good.’ (e.g. this bottled water)

(generic) (definite)

If mass nouns wouldn’t project higher than NPs, their interpretation should not be ambiguous because only the generic reading would be available. Hence, we assume that generic readings require D0 to carry a [−definite] value which is checked by the noun (cf. (36a)) in situ by long-distance agree.16 As to the possibility of the reading in (35b), it should be noted that the definite reading only comes about if there is prior identification of the referent. In this particular case, awa ‘water’ behaves like count nouns (cf. (36b)), where the [definite] feature is associated to number. Mass nouns with a definite reading are necessarily specified with a negative value for [number]. (36)

a.

DP D’ D0 [-def]

NP awa ‘water’ [-count] (generic)

b.

DP D’ D0

NP

[±def, ±nb] ome ‘man’ [+count] awa ‘water’ [-count] (definite reading)

. We assume with Longobardi (1994: 614) that “number specification (…) may sometimes be irrelevant. Number is obviously irrelevant in the case of mass nouns (…)”.





Nélia Alexandre and Tjerk Hagemeijer

Therefore the question is how definiteness and number interact in the case of null determiners under the hypothesis that number is parasitic on definiteness. Definite null determiners typically go with singular (the strong preferential interpretation of BNP subjects, cf. Example (31)), whereas indefinite null determiners typically go with plural (lexically governed BNPs, cf. Example (32)) when nouns are new information. Table 1 summarizes our findings so far. Table 1. Features of determiners in ST Determiners Va Inen Null

Definiteness

Plural17

− + + − −

− + − + n.a. (mass)

Table 1 provides a representation of the link between definiteness and number and accounts for the properties of (lexical or null) determiners in ST. The table shows that the determiner system of ST uses minimal lexical resources for two specific functions (wa/inen). Null determiners fill in the remaining functions. In the light of the multifunctional behavior of null determiners, the interface between the grammatical and extra-grammatical level is crucial to derive correct interpretations. Syntactic positions give BNPs preferred readings over others, but other mechanisms of interpretation may yield less expected readings and this is what we will try to explore in the next section.

.

Anchoring BNPs

From the examples in (31–32) it followed that in most cases we cannot obtain a correct interpretation of BNPs without making use of discourse and pragmatics, in addition to the purely grammatical level. This section sets out some cases that illustrate the importance of extra-grammatical information. Example (37) is taken from a conversation between a traditional doctor and the mother of an ill baby. (37)

A:

B:

Kê kwa ska da ku anzu? what thing  give with baby ‘What is wrong with the baby?’ Anzu ka sola. baby  cry ‘The baby cries.’

. We adopt ‘plural’ for number specification. Singular (−) corresponds to the default value and plural (+) to the marked value.

Chapter 2. Bare nouns and the nominal domain in Santome

In interlocutor A’s speech anzu ‘baby’ is found in a governed position, which is typically associated to plural indefinite readings when we are dealing with new information. The definite singular interpretation of anzu in interlocutor A’s speech is licensed because anzu constitutes shared knowledge of interlocutor A and B: the situation itself has already anchored the new information, which is therefore old information at the time of utterance. As a consequence, the definite singular reading of anzu carries over to interlocutor B’s speech, blocking the virtually possible generic reading ‘babies cry’. In Section 2 we have shown that singular non-previously identified items typically require the presence of indefinite article wa ‘a, an’. In (38) below, the speaker is aware that there are many folk stories but wants the storyteller to tell one with pretty women. Note that the indefiniteness of soya ‘story’ anaphorically anchors the indefinite reading of mwala glavi ‘pretty women’. (38)

Wa soya glavi ku mwala glavi. a story nice with woman pretty ‘A nice story with pretty women.’

There are, however, cases in which non-previously identified items are introduced as definite BNP because our knowledge of the world allows for this type of anchoring on the basis of inference. Part-whole relations are a good example of this type of anchoring. A previously identified whole anchors its “atomic” members, which become automatically identifiable at the grammatical level. When the whole of a part-whole relation (house-door in Example (39)) is previously introduced, the parts fare well as BNPs. This also extends to more specific constructions, such as inalienable possession (cf. (40)). (39)

Ke sa nglanji, maji poto na ka House is big but door   fisa fa. close  ‘The house is big, but the door(s) do(es)n’t close.’

(40)

San ka kluza mon pê pêtu. Lady  cross arm put chest ‘The lady crosses her arms over her chest.’

These part-whole relations can be conceptually linked to cases where one singled out element is taken to represent a (sub)species (A bird flies well vs. Birds fly well, or cf. Example (26)). If an entity is unique in the world (e.g. the sun in Example (41)), BNPs convey this definiteness (as does the obligatory determiner associated to sun as a countable entity in English, for instance). Although habitual aspect markers (ka) are often able to trigger pluralizing effects on object nouns, the unquestionable uniqueness of the sun in everybody’s reference world blocks any other interpretation.





Nélia Alexandre and Tjerk Hagemeijer

(41)

Ome ka pya solo. man  look.at sun Interpr. 1 ‘The {man/*men} look at the sun.’ Interpr. 2 ‘Men look at the sun.’ (generic)

Distributive readings are yet another case where the universe of discourse plays an important role. (42)

Bisu ka vwa ba liba po. bird  fly go up tree Interpr. 1 ‘The bird flies into the tree.’ Interpr. 2 ‘The birds fly into the tree/trees.’ Interpr. 3 ‘Birds fly into trees.’

If bisu is interpreted as a singular definite, as expected, this interpretation carries over to po ‘tree’ (cf. (38) for similar effects with indefinites), despite the claim that BNPs in object position have a preferred indefinite plural reading. In fact, the default interpretation of po is simply not available, unless the subject receives a plural interpretation. Our knowledge of the world is such that birds cannot fly to more than one tree at the same time.18 Interpretation 3 further shows that the presence of a habitual aspect marker (cf. (31c) for the same example without the aspect marker) also plays a role with respect to how sentences are read, an issue we will leave for future research. Consider once more subject BNPs (cf. (31)). In the sentences below, the subjects read either as definite singulars or as definite plurals, depending on prior discourse. (43)

a.

b.

Sode kopla pixi. soldier buy fish ‘The soldier(s) bought fish.’ Sode ba pega va ladlon. soldier go catch a thief ‘The soldier(s) went to catch a thief.’

When native speakers are confronted with (43a–b), they will preferentially assign a singular reading to the a. sentence and a plural reading to the b. sentence. It seems to be the case that the preferential plural reading of (43b) is not only related to the fact that it usually takes more than one soldier to catch a thief but crucially to the expectations concerning the activities of soldiers as a well defined collective. Catching thieves is what you expect from soldiers, rather than buying fish. The pragmatically markedness of the (a) sentence as being an activity of the collective of soldiers reduces the sentence to a preferential singular reading, say the one soldier that buys fish for his platoon. . Note that an iterative reading (flies from tree to tree) would only come about with verb reduplication.

(i)

Bisu ka vwa-vwa ba liba po. bird  fly-fly go up tree ‘The bird(s) fly from tree to tree.’

Chapter 2. Bare nouns and the nominal domain in Santome

So far, it followed that at the grammatical level ST shows a virtually very free distribution and interpretation of BNPs, which is restricted in the extra-grammatical component. This contrasts significantly with the restrictions on subject and object BNPs found, for instance, in Romance languages.19 At the level of information structure, some additional examples of the indefinite-definite shift are illustrated in (44–47). Example (44) shows the shift from an indefinite plural object to a definite plural subject, (45–46) show the shift from an indefinite singular object to a definite singular subject, and (47) the shift from an indefinite singular subject to a definite singular object. (44)

Alê d’e sode. Sode be. king give-3 soldier / soldier go ‘The king supplied him with soldiers. The soldiers went.’

(45)

 tê xa mosu ku xa mina mwala. 1 have a boy and a child woman Mina sa ni kwarenta tal. woman / girl is in 40 something Mosu sa ni xin kwenta. something / boy is in 50 ‘I have a boy and a girl. The girl is in her forties. The boy is fifty.’

(46)

Ê piji xa meza ku sêndê d’e. 3 ask a table  open give-3 / Meza plasê. table appear ‘He asked a table that would open for him. The table appeared.’

(47)

Wa floli pega mosu n’ope. Soku mina fada a flower grab boy in-foot / then girl say floli: … flower: … ‘A flower grabbed the boy’s feet. Then the girl said to the flower: …’

However, the step from an indefinite DP to a definite BNP often requires an additional and highly common “intermediate” step in the information structure, namely the anchoring of the indefinite by specific marker se (cf. Section 3), as illustrated in (48–49). (48)

Avia xa sungê ku mina sun. Sun se sa ve was a man with child man man  be old ketekete. 

. In Brazilian Portuguese, however, bare singulars and bare plurals in subject position can be kind-referring expressions (cf. Müller 2002). Oliveira (1998) presents similar facts for European Portuguese.





Nélia Alexandre and Tjerk Hagemeijer

Mina se sa ai, sun ka sam’e. child  be here man  call-3 ‘Once upon a time there was a man (formal) with his child. The man is very old. This child here, The man calls him.’ (49)

Kwa ku ê bi ku ê sa xa peli. thing  3 come with 3 be a skin / Sun fad’e: he (formal) tell-3: bixi peli se. Ê bixi peli se. dress skin  / 3 dress skin  ‘The thing he arrived with was a skin. The man (formal) told him to dress the skin. So he did.’

It follows from the relevant examples in the corpus that these two patterns are often in free distribution, although the presence of se assigns a higher degree of deictic force to identify the referent. Table 2 sums up how information structure is built up in ST. Table 2. Information structure

Singular

New information (indefinite)

Old information (definite)

Subject

Object

Anchoring

Anchored

va ome / va kabla ‘a man / a goat’

va ome / va kabla

Ome se / Kabla se Ome / Kabla

Ome / Kabla Ome / Kabla

Ome / Kabla

Inen ome se / Inen kabla se

Ome / Kabla

Inen ome / Kabla

Ome / Kabla

Plural

BNPs in lexically governed positions (V, P) may read as definites/indefinites and singulars/plurals. Only the indefinite singular reading isn’t available. In fact, indefinite singulars have to be preceded by wa in any syntactic position. Unlike the subject position, lexically governed positions typically generate new information and will thus often be the first step in the shift from indefiniteness to definiteness.

.

Final remarks

We have shown that definiteness and specificity are nuclear features of ST’s nominal domain. In the light of negative evidence for an independent functional category for number within DP, we assumed this feature is parasitic on definiteness and encoded under D0. This head hosts both lexical determiners (wa/inen) and null determiners (cf. Table 1). BNPs exhibit a wide range of interpretations, are able to occur in

Chapter 2. Bare nouns and the nominal domain in Santome

argumental and predicate positions and should therefore be fully integrated in the DP system outlined in (28). This richness of the null determiner system requires an extra-grammatical level of interpretation in order to capture the available nuances that cannot be exclusively assessed by the grammatical level. As the result of the availability of the former level, preferential readings associated to syntactic positions can be overruled. From a typological point of view, Longobardi’s (2001: 584) classification of BNP’s is insufficient to account for the data of ST where BNPs occur ‘freely’ in both argumental and predicate positions and are ambiguous between definite, indefinite and generic and singular/plural interpretations. Unlike Longobardi’s claim for Russian, Czech and Latin, this ‘ambiguity’ in ST cannot be assigned to the absence of articles. Chierchia’s Nominal Mapping Parameter (1998: 400), on the other hand, would integrate ST in the group of languages that distinguishes between mass and count nouns and would be of the type [+predicate, +argumental], like Slavic, without articles, and Germanic, with a richer article system. The inclusion in this type is motivated by the empirically supported claim that BNPs always project DP. It is crucial to bear in mind, though, that ST’s general typology sets this language far apart from the Indo-European languages Longobardi and Chierchia focus on. ST exhibits no overt number marking, no Case system and no inflectional morphology and does exhibit, for instance, serial verb constructions and preverbal TMA marking. Taken together, these features are likely to be an appropriate trigger for the widespread use of BNPs in this language. It also follows that ST bears several typological resemblances to Chinese, which Chierchia (1998) classifies as [−predicate, +argumental]. This translates into a system where nouns are always interpreted as mass and require a classifier system to derive countable interpretations. In other words, a very different nominal system from the one presently outlined. In sum, Longobardi’s syntactic and Chierchia’s semantic account prove insightful with regard to the discussion of the typology of BNPs but in this paper we hope to have shown that, in the case of BNPs in ST, grammar has to interface with an extragrammatical level where discourse and pragmatics play a crucial role in the anchoring of interpretation.

References Abney, S. P. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Agheyisi, R. N. 1986. An Edo-English Dictionary. Benin City: Ethiope Publishing Corporation. Agheyisi, R. N. 1990. A Grammar of Edo. Unesco. Baptista, M. 2002. The Syntax of Cape Verdean Creole: The Sotavento varieties [Linguistics Today 54]. Amsterdam: John Benjamins. Baptista, M. 2003. Inflectional plural marking in creoles and pidgins: A comparative study. In I. Plag (ed.) The Phonology and Morphology of Creole Languages, 315–332. Tübingen: Niemeyer.





Nélia Alexandre and Tjerk Hagemeijer Barrena, R. 1957. Gramatica Annobonesa. Madrid: Consejo de Investigaciones Científicas. Bernstein, J. 1993. Topics in the Syntax of Nominal Structure across Romance. PhD Dissertation, City University of New York. Bickerton, D. 1981. Roots of Language. Ann Arbor MI: Karoma. Bruyn, A. 1994. Noun phrases. In Pidgins and Creoles: An introduction, J. Arends, P. Muysken & N. Smith (eds), 259–269. Amsterdam: John Benjamins. Bruyn, A. 1995. Grammaticalization in Creoles: The development of determiners and relative clauses in Sranan [Studies in Language and Language Use 21]. Amsterdam: IFOTT. Butt, M. & Geuder, W. 2001. On the (semi)lexical status of light verbs. In Semi-Lexical Categories, H. Van Riemsdijk & N. Corver, 323–370. Berlin: Mouton de Gruyter. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Crisma, P. 1997. L’Articolo Nella Prosainglese Antica e la Teoria degli Articoli Nulli. PhD Dissertation. University of Padua. Déprez, V. 2001. Implicit determination constraining the meanings of bare nominals. Talk delivered at the Society for Pidgin and Creole Linguistics conference, Coimbra. Ferraz, L. 1979. The Creole of São Tomé. Johannesburg: Witwatersrand University Press. Giusti, G. 1991. The categorial status of quantified nominals. Linguistische Berichte 136: 438–453. Giusti, G. 1993. La Sintassi dei Determinanti. Padua: Unipress. Grimshaw, J. 1990. Argument Structure. Cambridge MA: MIT Press. Hagemeijer, T. 1999. As Ilhas de Babel: A Crioulização no Golfo da Guiné. Revista Camões 6: 74–88. Kihm, A. 1994. Kriyol Syntax: The Portuguese-based Creole language of Guinea Bissau. Amsterdam: John Benjamins. Kouwenberg, S. 1994. A Grammar of Berbice Dutch Creole. Berlin: Mouton de Gruyter. Lambrecht, K. 1994. Information Structure and Sentence Form – Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: CUP. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar. Cambridge: CUP. Longobardi, G. 1994. Reference and proper names: A theory of N-movement in syntax and logical gorm. Linguistic Inquiry 25(4): 609–665. Longobardi, G. 2001. The structure of DPs: Some principles, parameters, and problems. In The Handbook of Contemporary Syntactic Theory, M. Baltin and C. Collins (eds), 562–604. Malden MA: Blackwell. Lorenzino, G. 1998. Angolar Creole Portuguese. Munich: Lincom. Lucchesi, D. 1993. The article systems of Cape Verde and São Tomé Creole Portuguese: General principles and specific factors. Journal of Pidgin and Creole Languages 8(1): 81–108. Maurer, P. 1995. L’Angolar: Un Créole Afro-Portugaise Parlé à São Tomé. Hamburg: Helmut Buske. Melzian, H. J. 1937. A Concise Dictionary of the Bini Language of Southern Nigeria. London: Kegan Paul, Trench Trubner & Co. Müller, A. 2002. The semantics of generic quantification in Brazilian Portuguese. Probus 14(2): 279–298. Oliveira, F. 1998. Frases genéricas. Algumas especificidades do Português. Seminários de Linguística 2: 5–20. Pollock, J.-Y. 1989. Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry 20(3): 365–424.

Chapter 2. Bare nouns and the nominal domain in Santome Rizzi, L. 1997. The fine structure of the left periphery. In Elements of Grammar: Handbook of generative syntax, L. Haegeman (ed.), 281–337. Dordrecht: Kluwer. Schang, E. 2000. L’émergence des Créoles Portugais du Golfe de Guinée. Doctoral Dissertation, Université Nancy 2. Vallduví, E. 1993. The Informational Component, PhD Dissertation. Philadelphia PA: University of Pennsylvania. Wood, J. L. 2003. Definiteness and Number: Determiner Phrase and Number Phrase in the History of English, PhD Dissertation. Arizona State University.



 

On the syntax and semantics of DP in Cape Verdean Creole1 Marlyse Baptista In this chapter, I provide an in-depth description of the semantics and structure of the Cape Verdean noun phrase involving both determined and determinerless nouns. This study shows how overt determiners and their null counterparts interact and overlap in covering the same range of meanings. Cape Verdean bare nouns are interpretable as generic, definite, indefinite, singular, plural or mass. However, singular bare nouns display a subject/object asymmetry when it comes to their (in)definiteness status. Besides determined and determinerless nouns and pluralization strategies, this study of the Cape Verdean DP examines adjective placement in an attempt to identify the nature and number of structural layers reflecting the various types of nouns. In addition, I assume that an indexing relation between D, T and C takes place in order to derive the correct interpretation of bare nouns. Finally, I examine in what respects the determiner system of Cape Verdean Creole (CVC) differs from both European Portuguese (EP) and Brazilian Portuguese (BP). This comparative study will reveal that the use of bare nouns in CVC is much more widespread than in both EP and BP but that CVC and BP determiner systems also share a number of common points.

.

Introduction2

The objective of this paper is three-fold: first, I provide a detailed description of the Cape Verdean Determiner Phrase (DP),3 including common nouns and proper names and of the distribution of adjectives within DP. Second, I examine the semantics and . I am indebted to several scholars for their valuable feedback on one or several drafts of this chapter. I thank Anne Zribi-Hertz, Salikoko Mufwene, Ana Maria Brito and diverse conference audiences for insightful and useful comments that helped improve the quality and clarity of this paper. My special thanks go to Jacqueline Guéron who patiently read through several drafts and always provided quality feedback. All errors remain, of course, my own. . Most of the speech data in this paper are drawn from the corpus compiled during my field trips conducted in 1997, 2000, 2001 and 2003. The selected examples represent speech samples



Marlyse Baptista

structure of DP in Cape Verdean Creole (henceforth, CVC), with a special focus on bare nouns. By bare nouns, I mean determinerless nouns that may or may not carry the plural marker and are subject to a wide range of interpretive options. As a result, this study of bare nouns involves a close analysis of their interpretive variability as well as of the pluralization strategies affecting them. The analysis of the semantics and structure of DPs in CVC will postulate multiple layers. These will be considered at length in Section 2.4. The third objective is to confront CVC data with the determiner systems of European and Brazilian Portuguese4 revealing in which respects the three systems contrast and converge with one another. The organization of the paper is as follows: In an attempt to represent accurately the multiple layers of the Cape Verdean DP structure, the first section introduces the overt determiners in CVC as well as the pluralization strategies adopted within DP. The second section focuses on the licensing conditions of bare nouns and involves an exhaustive study of their distribution and interpretive variability. In this respect, factors such as animacy,5 definiteness status of the noun, singular/plural contrast, Generic versus Episodic Tense (c.f. Porterfield and Srivastav 1988, Guéron (2006)), uniqueness of entities and pragmatics will be considered; they will all be shown to affect DP marking and interpretation. The third section proposes a multi-layered structure for the Cape Verdean DP based on the empirical facts previously introduced. The fourth section demonstrates to what extent the determiner system of CVC resembles and differs from those of European Portuguese and Brazilian Portuguese. The fifth section presents a synthesis and concluding remarks.

. .

The full noun phrase The determiner system6

This section provides a full description of the Cape Verdean determiner system and examines the various interpretations of bare nouns. There are two types of overt determiners in CVC marking number but not gender, as a rule: the indefinite article un (sg.) (that can be traced back to the numeral un ‘one’) from all four Sotavento islands of the Cape Verde islands and have been assigned abbreviated labels identifying the informant and his/her native island. BR stands for the island of Brava, FO for Fogo, ST for Santiago and MA for Maio. The starred examples and the examples devoid of abbreviated labels are of my own devising. . Cape Verdean Creole is spoken in Cape Verde islands off the coast of Senegal and has European Portuguese as a lexifier. . I am greatly indebted to Fernanda Pratas and Fernanda Ferreira for providing me with the European Portuguese and the Brazilian Portuguese data respectively. I am also grateful to Ana Maria Brito and Tjerk Hagemeijer for taking the time to read the sections on EP and BP and for providing me with valuable input. All errors, remain of course, my own.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

and its plural counterpart uns (plur.). In the realm of definiteness, kel (sg.)/kes (plur.) may assume the role of a definite determiner in the language, although its primary function is that of a demonstrative. This evolutionary path is fairly common among the world’s languages: definite articles tend to originate from demonstratives through semantic weakening whereas indefinite articles generally derive from the singular numeral (Lyons 1999: 105). Given the evolution of definite determiners, it is not surprising that there exists an overlap in function between definite articles and demonstratives, a fact well illustrated by the Cape Verdean data, as will be demonstrated in Section 2.1.2.

.. Working assumptions and definitions Before going any further, I provide working definitions for my use of the concepts of (in)definiteness, (non)specificity, and (non)referentiality, so that any confusion may be avoided given the considerable shift and mutation these concepts have undergone in the field. Definiteness is generally viewed as being about identifiability (or familiarity to) by the speaker and hearer (Lyons 1999: 4). In contrast, an indefinite NP may be used to denote a particular entity (specific and familiar to the speaker) or to speak of an arbitrary member of the class described by the NP (non-specific) (Lyons 1999: 165). Thus, an indefinite NP may be specific or non-specific (Comrie 1981; Lyons 1999 and Croft 1990). The concept of referentiality is useful when dealing with the following type of examples. Consider (1) and (2) taken from Lyons (1999: 165): (1) (2)

I bought a car Pass me a book

“A book” in (2) is non-specific and describes but crucially does not refer whereas specific “a car” in (1) is a referring expression.7 Lyons (1999) noted that although the concepts of specificity and non-specificity have been mostly used in relation to indefinites (cf. Comrie 1981 and Croft 1990), this distinction is just as applicable to definites. Consider the following set of examples from Lyons (1999), featuring an indefinite determiner in (3) and a definite one in (4).

. In Baptista (2003), I show to what extent the factors of animacy and definiteness play a consistent role in plural marking in a variety of languages such as Cape Verdean Creole, GuineaBissau Creole (Kihm 1994), Angolar and Nigerian Pidgin English. . Abbreviations:  = aspect,  = Cape Verdean Creole,  = complementizer,  = copula,  = demonstrative,  = indefinite,  = negation,  = numeral,  = quantifier phrase,  = tense,  = tense, mood, aspect marker,  = reflexive,  = preposition,  = plural,  = specific. . Such a view of referentiality is not uncontroversial. Lyons (1999: 166) actually discusses three main positions on the issue. Some linguists argue that definites may (but need not) refer





Marlyse Baptista

(3)

a. b.

Peter intends to marry a banker – even though he doesn’t get on at all with her. Peter intends to marry a banker – though he hasn’t met her yet.

(4)

a.

Joan wants to present the prize to the winner – but he doesn’t want to receive it from her. Joan wants to present the prize to the winner – so she’ll have to wait around till the race finishes.

b.

The (a) examples show extensional readings which are specific and referential. “A banker” in (3a) and “the winner” in (4a) refer to specific individuals. In contrast, the (b) examples show intensional readings which are non-specific and non-referential. It is clear from the context that “a banker” in (3b) and “the winner” in (4b) have not yet been identified either by the speaker or the hearer; hence, they do not refer to specific entities. These two sets of examples illustrate Lyons’ observation that both indefinites and definites can be specific or non-specific.8 This particular insight is particularly relevant to our analysis of the CVC overt determiner system in Sections 2.1.2 and 2.1.3. There, it will be shown that in CVC, overt indefinites and definites can be interpreted as specific or nonspecific, hence they can be both extensional or intensional. When we consider bare indefinite and definite nouns, we will see that they are equivalent in interpretation to their overt counterparts. This is a significant refinement in our understanding of definite nouns modified by an overt determiner vs. no determiner, as the belief has been so far that nouns modified by kel can only receive a definite, specific and referential reading. At this point, I should add that in this chapter, I treat specific/referential and nonspecific/non-referential on a par. In other words, I consider “specific” to be synonymous with “referential” and “non-specific” as a synonym of “non-referential.” Working definitions of definiteness, specificity and referentiality having been provided, we may now examine the semantics of CVC determiners.

whereas indefinites do not refer precisely because the “referent” of a specific indefinite is not identifiable to the hearer. According to this view, referentiality and identifiability are correlated. The opposite and dominant view is that neither definites nor indefinites refer. Proper names, demonstrative and personal pronouns are considered to be the only referring expressions, their common point being that they all pick out a particular individual entity. In this view, the role of definites and indefinites is just to describe; they are analyzed as quantificational rather than directly referring. Although such a view concedes that definites have referential uses, such reference is considered to belong to the realm of pragmatics rather than semantics. Finally, Lyons (1999: 166–167) argues that both definites and indefinites are potentially referring and specific, see examples (3) and (4) above for discussion. This is the view I will adopt here, as it finds convincing evidence in the empirical data provided by CVC. This will be demonstrated in Section 2.1.3. . Some scholars would disagree with Lyons’ characterization and argue that Tense and predicate collaborate and combine with each other to give a flavor of specificity to the noun when the noun actually has no specificity on its own. They would argue that non-specific nouns become referential only if they have a truth-value.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

.. The semantics of indefinite and definite determiners This section examines the semantics of overt determiners. Here, I show that the indefinite determiner in both the singular and plural forms perform several functions in the language. As for the definite article, it performs a dual role in the language as a demonstrative and definite determiner. Let us start with the indefinites. Un in CVC can act both as a numeral, an indefinite determiner or a quantifier. The first two functions are illustrated in (5) and (6&7) respectively: (5)

a.

b.

(6)

a.

b.

c.

(7)

Numeral (F-Maio) N tene un matxu ki tene trezi. N tene un femia ki s’ I have one son who has thirteen I have one girl who  en faze des. come do ten ‘I have one son who is thirteen years old. I have one girl who is about to turn ten.’ Nu ta koba, nu ta pranta pe di mandioka, (AM-ST) we  dig we  plant foot of manioc dipos nu ten di puxa un anu, nu ta ten mandioka after we have to push one year, we  have manioc ‘We dig, we plant the roots of manioc and after one year has gone by, we get manioc.’ Indefinite non-specific/non-referential Si’ N atxa meiu di N sai pa un kau, ma ben inviste na if I find way of I leave for a place  come invest in nha tera (AM-ST) my country ‘If I find a way of leaving for a (different) place, I will return to invest in my country.’ O k’ u kuberta-l, e ta subi dentu d’ un bandeja, (AM-ST) when you cover-it it  raises inside of a pan Ora k’e subi, bu ta panha-l, bu abri-l k’ un garafa when it raised you  take-it you open it with a bottle ‘When you cover it, it raises inside of the pan. Once it is raised, you open it up with a bottle.’ Otu algen ta furta na rubera pa fase un tetu (AM-ST) other people  steal in river bed to make a ceiling pamo guvernu ka sa ta djuda-u fase kasa because government    help-you make house ‘Other people steal from the river bed to make a ceiling because the government is not helping you make houses.’

Indefinite specific/referential N panha un panu, N po na kabesa. (B-Fogo) I took  cloth N put on head ‘I grabbed a cloth, I put it on my head.’





Marlyse Baptista

In (6a), (6b) and (6c), un is used to indicate that the referent is non-specific/nonreferential: that is, it is new in the discourse and in the shared consciousness of the speaker and hearer. Un may also refer to a specific/referential entity already known to the speaker. This is the case in (7). Hence, it is important to emphasize that an NP introduced by un may be [indefinite, non-specific/non-referential], as in (6), or [indefinite, specific/referential], as in (7).9 This generalization is represented in the following schema: (8)

Indefinite un → [specific/referential or non-specific/non-referential]

There are also clear cases where singular un behaves like a quantifier, particularly when it modifies noncountable, abstract entities, as shown in (9). (9)

Quantifier a. E bota-m un ajudinha. (S-ST) she throw-me  little help ‘She gave me some little help.’ b. Nho kre faze un karidadi. (S-ST) you want make  charity ‘You wish to do some charity.’

Although one of the functions of un is to introduce a new referent (Lang 1990), its presence is not required for a given NP to be interpreted as new information. For instance, Ø can be used in place of the indefinite non-specific un in (6c), the indefinite specific un in (7) and even the quantifier in (9), yielding (10), (11) and (12) respectively: (10)

Otu algen ta furta na rubera pa fase Ø tetu other people  steal in river bed to make ceiling pamo guvernu ka sa ta djuda-u fase kasa because government    help-you make house ‘Other people steal from the river bed to make a ceiling because the government is not helping you make houses.’

(11)

N panha Ø panu, N po na kabesa. I took cloth N put on head ‘I grabbed a cloth, I put it on my head.’

(12)

E bota-m Ø ajudinha. she throw-me little help ‘She gave me some little help.’

In Section 2.1.3, we consider further cases where bare nouns may be interpreted as indefinite and either referential or non-referential. . The use of un in CVC is in contrast to that of Guinea Bissau Creole (GBC). According to Kihm (1994: 137; Kihm, this volume), un in GBC modifies NPs that can only be interpreted as non-specific non-referential.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

Uns, the plural counterpart to un, may modify a countable entity, in which case, it acts as a quantifier and not as a genuine plural determiner, as illustrated in (13). (13)

Quantifier Es faze uns batankinha di midju. (AA-BR) they made some rolls of corn ‘They made some corn rolls.’

Jacqueline Guéron (p.c.) notes that there may be no difference between a genuine plural determiner and a quantifier cross-linguistically. By extension, the singular un and its plural counterpart uns can both be regarded as quantifying the entity they modify, which in the case of un becomes quite clear in examples such as (9) above. We should emphasize, however, that while the distinction between a plural determiner and a quantifier such as uns is somewhat blurry, its singular counterpart un seems to clearly fulfill two distinct functions in the language: it acts as an indefinite determiner and as a quantifier. As a result, different sites for un the quantifier and un the determiner will be postulated in the structure of DP we offer in Section 2.4.3. We turn now to definite NPs. As a rule, CVC does not explicitly mark its NPs as being definite by means of overt determiners (this issue will be examined in Section 2.1.3). When an overt determiner expressing definiteness does appear, it has the form kel/kes. Kel/kes has a double function: on the one hand, it primarily performs the role of a demonstrative, as shown in (14):10 (14)

N ben faze kel kaza li ki pai di fidju da-m. (MCR-ST) I came make this house here  father of son gave-me. ‘I came to build this house that my son’s father gave me.’

Alternatively, it may mark definiteness when referring to a previously introduced, hence, known entity, as shown in (15) and (16).11 In (15a) and (16a), the expressions azagua and lutu are used for the first time in these particular passages (of the two interviews). When azagua and lutu recur in the following sentence, kel precedes it but involves no deixis:12 (15)

a.

Mas ami, N ta trabadja azagua. (S-ST) but me I  work rainy season. ‘As for me, I work during the rainy season.’

.

See Section 2.1 regarding the typical evolutionary path of demonstratives.

.

See Déprez, this volume, for a different position on the status of kel/kes.

. In Guinea-Bissau creole as well, Kihm (1994: 139; this volume) reports that the demonstrative is gradually gaining an anaphoric role, signaling that the entity denoted by the NP has already been mentioned in previous discourse. This function is fulfilled in Guinea-Bissau creole by the distal demonstrative kil.





Marlyse Baptista

(16)

b.

N ta munda kel azagua mi so. (S-ST) I  weed  rainy season myself ‘I weed during the rainy season on my own.’

a.

Padri ta bisti di pretu, pamo e po lutu di nosu Senhor priest  dress of black because he put mourning of our lord Jesus Kristu. Jesus Christ E ten ki guarda kel lutu. (C-Santiago) he has  keep the mourning ‘The priest wears black because he is in mourning of our Lord Jesus Christ. He must keep the mourning.’

b.

It should be emphasized that when azagua recurs in (15b) and lutu in (16b), the use of kel is not necessary. While it is difficult to predict with precision when the use of kel is obligatory, it is possible to define a set of conditions governing the appearance of definite bare nouns. These conditions will be outlined in Section 2.1.3, where it will be shown in accord with Kayne’s nonoptionality principle (Kayne 2004), that the competition between the overt definite determiner kel and its non-overt counterpart is regulated by a set of fairly well-defined factors. Note that kel can also modify abstract or non-countable entities, as illustrated in (17): (17)

Ma na kel tenpu, moda ki nhos ta panhaba agu, na rubera? (AM-ST) but in  time how  you  took water in river bed ‘But in that time, how did you use to collect water, in the river bed?’

The plural counterpart of kel, kes is illustrated in (18): (18)

N fika ku kes mininu, pai ba ranja rapariga (MM-ST) I stayed with these children father went arrange young woman ‘I stayed with these children, the father went and got himself a young woman.’

Finally, it is important to note that kel can be followed by the numeral un (or other types of numerals) (i.e. 19), a point that has implications for its position in the tree structure we propose in Section 2.4.3: (19)

N sta na poder di kel un fidju. (B-ST) I am in power of that one child ‘I depend on that one child.’

Although it is often difficult to decide whether one is dealing with the true deictic or its anaphoric counterpart,13 clear examples of the anaphoric, or definite, usage have

. This ambiguity originates in part from the overlap in function between the two due to the evolutionary path mentioned in Section 2.1.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

been provided in (15), and (16). The use of kel as a definite determiner cannot be considered systematic at this point in the development of the language due to its parallel usage as a demonstrative. As in the case of other languages, the systematic use will arise when the demonstrative eventually becomes morphologically distinct from the definite determiner avoiding the current functional overlap and ambiguity. Given the double function of kel, one could then argue that kel is a demonstrative when there is no earlier mention of a referent and an anaphoric determiner when there is an earlier mention. The situation is, however, more complex. Indeed, although it is often the case that the definite determiner kel refers to a familiar, salient entity or to an item previously introduced in the discourse, it can also modify a noun that involves no deixis or anaphoric reference. Observe for instance the example in (20), in which the speaker makes no mention of the corn prior to this first occurrence in the interview and is not pointing at it deictically either, and yet midju ‘corn’ is modified by kel. (20)

N ta labanta nha, tres ora madrugada, N po kel midju na pilon I  get up woman three hour early morning I put  corn in pestle, mi ku nha fidju. me with my child. (MM-ST) ‘I get up, woman, at three o’clock in the morning. I put the corn in the pestle, me and my child.’

The same situation obtains in (21), as kel dinheru ‘the money’, kel batata ‘the potatoes’, kel midju ‘the corn’ and kel fixon ‘the beans’ were never introduced in the prior discourse and are not being referred to deictically. Their interpretation is non-specific/ non-referential. (21)

Ta resebeba kel dinheru pa’ N kunpra kel batata,  received the money for I buy the potatoes kel midju pa nu po katxupa, the corn for we put katxupa ta kunpra kel fixon pa nu po riba (AM-Santiago)  buy the beans for we put on stove ‘We would receive the money so that I would buy the potatoes, the corn for the katxupa, we would buy the beans to cook on the stove.’

Such examples show that a noun modified by kel can be interpreted as specific/referential (i.e. (15b), (16b)) or non-specific/non-referential (20) and (21). This means that just like its indefinite counterpart un (i.e. (6), (10), (11) and (12) above), kel and Ø could be viewed as occurring in free variation, as they cover the same range of meanings. This is illustrated by the Cape Verdean sentences (22) and (23), counterparts to the English examples presented in (3) and (4) above: (22)

a.

Pedru kre kaza k’un/Ø bankeru – ma e ka ta da ku el. Pedru want marry with+a/Ø banker – but he Neg  get along with her ‘Pedru wants to marry with a banker – but he doesn’t get along with her.’





Marlyse Baptista

(23)

b.

Pedru kre kaza k’un/Ø bankeru – ma e ka inkontra-l inda Pedru want marry with+a/Ø banker – but he  meet-her yet ‘Pedru wants to marry with a banker – but he hasn’t met her yet.’

a.

Joana kre da kel/Ø jugador kel prezu – ma e ka kre J. want give the/Ø player the prize – but he Neg want rasebe-l d’el. receive+it from-her ‘J. wants to give the prize to the player – but he doesn’t want to receive it from her.’ Joana kre da kel/Ø jugador kel prezu – ma e ten ki J. want give the/Ø player the prize – but she must  spera te fin di jogu. wait to end of game. ‘J. wants to give the prize to the player – but she must wait till the end of the game.

b.

The example in (22) shows that the indefinite un may get specific/referential reading (22a) or a non-specific/non-referential one (22b). In a similar fashion, the overt definite kel can either get a specific/referential interpretation (23a) or a non-specific/ non-referential interpretation one (23b). In addition, both un and kel may be replaced by an empty category Ø. Given this seemingly free variation between the overt determiners and their null counterparts, a question that immediately comes to mind is whether it is possible to predict categorically when kel or un must be used and when Ø (with definite or indefinite meaning) must appear. In this chapter, I am not yet able to answer this question but do attempt to define the conditions favoring the occurrence of bare nouns. Indeed, although CVC may explicitly distinguish between definite (specific/ referential) and indefinite (specific/referential and non-specific/non-referential) NPs by means of overt determiners, we show in Section 2.1.3 that such determiners may be absent if sufficient information regarding the (in)definiteness status of the noun is provided in syntax by plural marking, DP distribution and what kind of tense lies in TP (generic or episodic). Such information may also be provided by the immediate context, prior discourse, or (in)animacy of the noun. The following section will illustrate the prevalent use of bare nouns in the language.

.. Interpretation and use of bare nouns14 In this section, the multiple possible interpretations of bare nouns in CVC will be presented, as well as the set of conditions underlying their occurrence. ... Interpretive variability of bare nouns. Bare nouns show a high degree of interpretive variability. They may be interpreted as generic (24), definite/specific15 plural (25), indefinite/specific plural (26), indefinite/non-specific plural (27), indefinite . See Baptista (1998, 2000) for an in-depth study of the various interpretations of bare nominals. . We make this more precise below when we actually argue that the (non)specificity reading emerges from an index agreement between D, T and C in association with the event in the sentence.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

specific singular (28), indefinite non-specific singular (29), definite/specific singular (30) or definite/non-specific singular (31). The relevant interpretation is captured through cotextual (found in the text itself) or contextual cues, pragmatics and syntactic clues, such as DP position and the nature of T. Indeed, in Section 2.1.3.2, we discuss how the right interpretation may be derived from the co-indexation of various heads in the sentence. (24)

Generic (AM-ST) a. Tradison di Badiu, ten txeu kuza: tradition of Badiu have lots thing kotxi midju, kotxi na pilon, bu kotxi, crush corn crush in pestle you crush bu bentia, bu tra farelu, you shake you remove bran bu po katxupa riba, i bu fase masa, you put katxupa on top and you make dough fase tenteren, ku pexi o ku leti. make tenteren with fish or with milk ‘The traditions of the Badius, there are a lot of things: you crush corn, crush it in the pestle, you crush, you shake it, you remove the bran, you put katxupa on the stove and you make dough, you make tenteren, with fish or milk.’ b. Kela k’ e nos kumida tradisional, fixon, ku xeren, ku tenteren this   our food traditional bean and xeren and tenteren ‘This is our traditional food: beans, and xeren and tenteren.’ (AM-ST)

(25)

Definite specific plural Kaza di es aldeia e baratu. house of this neighborhood is cheap ‘The houses in this neighborhood are cheap.’

(26)

Indefinite specific plural N odja pasaru riba di kaza. I see bird top of house ‘I saw birds on top of the house.’

(27)

Indefinite non-specific plural a. Nton, ten monti algen ki ka sabe ler na nos tera, so have lots people   know read in our land pa djuda mamai trabadja, ku papai trabadja to help mother work and father work pa kiria kes otu mininu. Ma dja gosi, mininus ta bai skola to bring up the other child but now children TMA go school mas txeu di ki un bes. (AM-ST) more a lot than  a time ‘So there are a lot of people who cannot read in our land, to help mothers work, to help fathers work, to bring up the other children. However, nowadays, children are going to school much more than in the old days.’





Marlyse Baptista

b.

O ki nu fase kasa, nu ta djunta mon, ku ke otu bizinhu (AM-ST) when we make home we  gather hand with the other neighbor ‘When we build houses, we help each other, with the other neighbors.’

(28)

Indefinite specific singular a. asi ki’ N bai, ben abridu portu. (B-ST) how  I went, came open harbor ‘That’s how I went, a harbor opened.’ b. N panha fratura na mon, fika ku el tortu (FLT-ST) I took fracture on hand stay with it bent ‘I received a fracture on my hand, now my hand remains bent.’

(29)

Indefinite non-specific singular Si bu ten terenu di fase, bu ta pidi djuda (AM-ST) If you have terrain to make you  ask help ‘If you have a terrain to make, you can ask for help.’

(30)

Definite specific singular a. Ma N ta trabadja gosi ku kanbra. (AM-ST) but I  work now with city hall ‘But I work now with the city hall.’ b. Ta ben txeu djenti ta ben buska-m  come lots people  come look for+me pa’N konta-l storia di kel organizason (AM-ST) for I tell them story of the organization ‘A lot of people come to look for me so that I tell them the story of the organization.’

(31)

Definite non-specific/singular (counterpart of (20) above) N ta labanta nha, tres ora madrugada, N po midju na pilon I  get up woman three hour early morning I put corn in pestle, mi ku nha fidju. me with my child. ‘I get up, woman, at three o’clock in the morning. I put the corn in the pestle, me and my child.’

In addition, mass nouns such as agu ‘water’ in (32) or lenha ‘wood’ (33) typically appear determinerless, and may appear in both subject (32b) and object (32a) positions: (32)

a.

b.

Nu ten txafaris pa nu panha agu. (AM-ST) we have well to take water ‘I have the well to collect water.’ Nu ten fonti, agu ta sai si, bu ta panha, we have spring water  come out like this you  take bu ta ben ku’el kaza pa bu sirbi ku el. (AM-ST) you  come with it home for you use with it

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

‘We have spring water, the water comes out like this, you take it and you bring it home so that you may use it.’ (33)

Po tres pedra fogon, bu po lenha, bu kuzinha. (AM-ST) put three stones stove you put wood you cook ‘Put three stones in the stove, you put wood, you cook.’

Proper names typically appear determinerless in CVC. They may occur in subject (34a) or object (34b) position: (34)

a.

b.

Brankinha ba sukundi. (RC-ST) Brankinha went hide ‘Brankinha went into hiding.’ N atxa Brankinha la vila ku si mininu. (RC-ST) I found Brankinha in town with her children ‘I found Brankinha in town with her children.’

The next issue at hand is to determine the exact nature of the Cape Verdean bare nouns introduced in the previous examples. Chierchia’s (1998) nominal mapping parameter would predict that the bare nouns appearing in argument position, in (25) and (26) for instance, denote kinds and emerge from the lexicon with mass denotations. However, evidence that bare nouns distinguish individuation from mass can be easily obtained through two basic tests (cf. Schmitt & Munn 1999). If bare nouns were treated as mass, the universal grinder (concept proposing that if one takes a count noun and put it through a conceptual grinder, the result will be an undifferentiated and spreadable mass) would not apply in contexts such as (35): (35)

Libru staba spadjadu na txon. (adapted from Schmitt & Munn 1999) book were spread on floor ‘Books were spread out all over the floor.’

Another piece of evidence for individuation in bare singulars comes from the acceptability of reflexives and reciprocals. This is shown in (36) with the reciprocal kunpanheru: (36)

Bizinhu ka ta daba ku kunpanheru. neighbor   give+ with each other ‘Neighbors didn’t get along with each other.’

The facts exemplified in (35) and (36) show that absence of both a determiner and plural marking does not entail that a bare noun has a mass denotation. Furthermore, if one adapts to CVC the test of discourse anaphora that Kester & Schmitt (this volume) apply to Papiamentu and Brazilian Portuguese, one can easily determine the atomicity of these bare nouns. The examples in (37) illustrate that count bare nouns can be referred back to with either a singular or a plural pronoun. In (37a) the atomic reading obtains whereas in (37b) the non-atomic interpretation emerges making it clear that





Marlyse Baptista

with count bare nouns such as djenti16 (a collective noun equivalent to ‘people’), the singular/plural distinction is not neutralized. (37)

a.

b.

Ta ben djenti di stranjeiru ki ta ben, (AM-T)  come people from abroad   come ta ben buska-m pa’ N konta-l storia di ….  come find+me for I tell-him/her story of Ta ben djenti di stranjeiru ki ta ben,  come people from abroad   come ta ben buska-m pa’ N konta-s storia di….  come find+me for I tell them story of ‘People would come abroad, they would come and fetch me so that I would tell him/them the story of…’

The next question is to determine the conditions and triggers associated with the emergence of bare nouns.

... Conditions and settings underlying the occurrence of bare nouns. A noun may appear bare (without the determiner un (marking indefiniteness, specificity or non-specificity) or the determiner kel (conveying definiteness and/or specificity)) under some conditions and settings. Those presented here are not meant to reflect the entire set under which bare nouns may occur but the objective of this section is to capture some of the most common instances. The marking of the definiteness/referential status of the noun is viewed as unnecessary if the collective awareness or uniqueness of the entity makes it familiar to the hearer and speaker and to the world around them. This would apply to natural phenomena such as txuba ‘rain’ (38b), or unique entities such as sol ‘sun’ (38a) that form their own class. The uniqueness of a given entity in a speaker’s knowledge of the natural world such as ‘sun’, ‘moon’, ‘earth’ makes it unnecessary to mark definiteness. (38)

a.

b.

Sol ta ben mas txeu, ta da kumida mas txeu. (AM-ST) sun  come more a lot,  give food more a lot ‘The sun comes out more, there is more food.’ Txuba ta txiga mes di agostu, txuba ta txobe. (D-ST) rain  arrive month of August rain  rain ‘The rain arrives during the month of August, the rain falls.’

A second reason is that there are individuals and other elements that are parts of the common experience and common knowledge of both the speaker and hearer making the use of a determiner unnecessary. For instance, in (39), xefri ‘the leader’ is an

. Some speakers use djenti while others use its plural counterpart djentis, both lexemes meaning ‘people.’

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

individual known both by the speaker and her audience, hence there is no need for a determiner: (39)

N ben ta ntende kuse ki xefri ta konbersa-m. (S-ST)  come  understand what  chief  talk+me ‘Little by little, I got to understand what the chief is talking to me about.’

It also seems that whenever a noun is first introduced by an overt determiner, its second occurrence is typically as a bare noun. Consider for this purpose the following text, in which the specific indefinite NP mudjer in (40) is introduced by the indefinite article un in its first occurrence and appears as a bare noun subsequently. Mudjer in the second sentence behaves like the English definite ‘the woman’. It refers anaphorically to the referent introduced in the previous sentence. (40)

Es gosta di Vieira na Praia, es ka xa-l bai, el fika k’ they like of Vieira in Praia they  let-him go he stay with un mudjer la. (TA-Brava) a woman there El fika ta kuzinha, ta lava ropa, el xina… he stay  cook  wash clothes he learn Di lago, N ka sabe kuma k’ el ba, mas es gostaba di rapazinhu, From there I  know how  he go but they liked of boy es fika ku el na Praia. Mudjer gosta d’ el, fika ku el ala. they stay with him in Praia woman like of him, stay with him there ‘They liked Vieira in Praia, they didn’t let him go, he stayed with a woman there. He kept cooking, washing clothes, he learnt…From there, I don’t know how he went but they liked the boy, they stayed with him in Praia. The woman liked him and kept him there.’

The same case obtains in (41) below: The indefinite NP un bolu is introduced by the indefinite article un in its first occurrence and recurs as a bare noun.17 (41)

Bu ta panha farinha, bu lorada na un bolu di forma, (AM-ST) you  take flour you dust in a cake of mould e ta fazedu didentu bolu di forma.   make in inside cake of mould ‘You take flour and you dust a cake mould with it, it is made inside a cake mold.’

The data in (40) and (41) corroborate an earlier observation by Lucchesi (1993: 81) that after an NP has been introduced in the discourse by the indefinite article un/uns, it may recur bare with no need for kel/kes or any other determiner.

. The examples in (40) and (41) are parallel with the difference that in contrast to un mudjer in (40) which is indefinite/specific, un bolu in (41) is indefinite/non-specific.





Marlyse Baptista

It should, however, also be noted that a noun appearing for the first time preceded by Ø may too be interpreted as indefinite specific or non-specific, referential or nonreferential. To Lucchesi’s observation that “once a discourse item has been introduced by an indefinite article, no marking of the noun by a definite article is necessary,” I would add that no prior introduction of the noun with the indefinite article un is even necessary or expected. If one wishes to differentiate un from Ø and predict its occurrence with respect to its null counterpart, then, following Bruyn’s observation (this volume) for Sranan wan, one could propose that the primary and primordial function of un is to make singularity and indefiniteness explicit (cf. Mufwene 1986; Patrick 2004). There is yet another oversight in the literature (cf. Lucchesi 1993) regarding the interpretation of bare nouns in CVC: Although Lucchesi (1993) correctly observed that nouns unmarked by an article may get a specific definite or specific/non-specific indefinite reading, it is, however, important to note that this generalization does not carry over to both plural and singular bare nouns. Bare nouns with no plural marker but interpretable as plural may get a specific/definite or specific/non-specific indefinite interpretation whether they appear in subject or object position. For instance, the object bare noun garafa di vinhu18 in (42) could be interpreted as specific definite, as in ‘João drinks/will drink the bottles of wine (stored in the cellar)’ or as non-specific indefinite, as in ‘João drinks/will drink bottles of wine (all night long)’. The specific indefinite interpretation also obtains as in ‘João drinks/will drink bottles of wine (that date back to 1985)’. The precise meaning of ‘bottles of wine’ varies according to the meaning of ta (habituality or future event), overall context or anchoring adverbs and modifiers (i.e. ‘all night long’, ‘stored in the cellar’). (42)

João ta bebe garafa di vinhu João  drink bottle of wine ‘João drinks the bottles of wine.’ will drink or ‘João drinks bottles of wine.’ will drink

The same interpretations obtain in the case of subject plural bare nouns (with no plural marker). (43)

Garafa di vinhu sta karu. Bottle of wine  expensive ‘The bottles of wine are expensive.’ or ‘Bottles of wine are expensive.’

However, a bare NP is construed as singular and may be interpreted as definite specific, as in (44), or indefinite (specific as in (45), or indefinite non-specific, as in (29),

. I chose on purpose an inanimate plural bare noun because, as will be discussed in Section 2.2, animate plural nouns without plural marking typically (though not always) receive a definite interpretation, see Ferreira (2004, 2006) for a convincing sociolinguistic treatment of the phenomenon.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

repeated here for convenience as (46)), depending on its position in the sentence. Definite singular bare nouns tend to be found in subject positions (old information) while indefinite singular bare nouns (new information) tend to appear in object positions. This is illustrated in (44) and (45)–(46), respectively: (44)

definite specific singular Mudjer gosta d’el, fika ku el ala. (TA-Brava) woman like of him stayed with him there ‘The woman liked him and stayed with him there.’

(45)

indefinite specific singular a. Nu tenha xefri ki ta leba libru ku nos. we had leader   take book with us ‘We had a leader who would take book(s) with us.’ (C-ST) b. Djedje atxa mudjer bibu, kuazi mortu lago. (BCR-BR) Djedje find woman alive almost dead there ‘Djedje found a woman alive, she was nearly dead over there.’

(46)

indefinite non-specific singular Si bu ten terenu di fase, bu ta pidi djuda. (AM-ST) If you have terrain to make you  ask help ‘If you have a terrain to make, you can ask for help.’

The state of affairs regarding bare nouns construed as singular is summarized in the schema in (47): (47)

Positions for new and old information (regarding singular bare nouns) Old information: subject position → definite New information: object position → indefinite

It is important to emphasize two points: first, one should keep in mind that these are just tendencies, as it is also possible to find indefinite specific/non-specific singulars or generics in subject position, as illustrated in (48). (48)

Amigu ka ta faze keli. [generic] friend   do this ‘A friend does not do such a thing.’

Second, episodic and generic tense also affects the interpretation of the bare noun whether it is in subject or object position. More precisely, other elements in the sentence such as Tense, Aspect or a topic time (i.e. adverbial, see below) can lead the hearer to correctly interpret whether the bare noun is referential (definite/indefinite) or generic. In sum, episodic versus generic tense can predictably affect the interpretation of bare nouns in ways that I will articulate below. In CVC, the bare noun is interpreted as referential when the tense is episodic; if the tense is generic, the bare noun is interpreted as non-referential/generic. This is illustrated by (49) and (50) respectively.





Marlyse Baptista

(NNNL-ST) (49) E trabadja sustenta familia, mudjer, fidju, te ki more (Episodic) he worked support family wife children until  died ‘He worked to support the family, the wife and the children until he died.’ (50)

Si mininu ten brinkedu, el ta fika ketu. (Generic) if child has toys s/he  stays quiet ‘If a child has toys, s/he stays quiet.’19 (adapted from Porterfield & Srivastav, 1988)

The same situation obtains in other languages, notably English. Porterfield & Srivastav (1988: 266) observed that the English (definite) generic in (51) gets a universal reading only when the tense is generic. In contrast, a noun is interpreted as specific when the tense is episodic, as shown in (52): (51)

If the child has a toy, she is happy. (universal reading) → generic tense

(52)

Yesterday, whenever the child had a toy, she was happy. (non-universal reading) → episodic tense

Then, the natural question that comes to mind is how the generic versus episodic interpretation emerges. In this respect, Guéron (2006) offers a useful proposal that examines what type of interaction is at work between sentential units ( and  for instance) in yielding the correct generic and episodic construals for nominal expressions. Here are the main tenets of Guéron’s proposal, keeping in mind that for the purpose of clarity, only the assumptions relevant to the particular case of Cape Verdean Creole have been selected and retained in this presentation: (53) i.

ii. iii.

Guéron’s basic assumptions: The constraint in (i) plays a pivotal role in LF structures: The situation a given sentence describes must be placed in the space and time of the discourse world, i.e. the world of the speaker or a world which the grammar makes accessible to the speaker. vP is viewed as the domain of spatial interpretation whereas TP/CP is the domain of temporal interpretation. A sentence is episodic if its tense node bears a positive tense index and generic if its tense node carries a zero index or no index. This entails that a sentence is ambiguous between an episodic construal and a generic construal. The same syntactic structures derive both episodic and generic sentences. In other words, across languages, the same DPs refer to either real entities (‘real dogs’ for instance) or to the mental concept associated with the lexical item (‘the concept of dog’).

. Naturally, if the entity mininu was introduced in prior discourse, it can get a definite/ specific interpretation (i.e. the child).

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

iv.

v.

vi.

To reflect the interaction at work between sentential units, Guéron proposes a -chain rooted in the complementizer node C associated with the Reference or Speech time, continues with the tense node  associated with the Event time and ends with V. If  is indexed 1 for Speech time and  is also indexed 1 for Event time (1 for Present and 2 for Past), then the Event time is construed as anaphoric to the Speech time and results in an episodic present. For Generic,  has a zero index. In a generic sentence, there is no link established between event space and discourse space. The basic hypothesis is that the on/off setting of the index on  is sufficient to distinguish episodic from generic sentences. If so, no generic operator is needed. Aspect also plays a role and interacts with the output constraint mentioned in (i) within both DP and TP. The event time interval must be bounded in order to satisfy constraint (i). One way to bound a time interval is to place the event the sentence denotes in the scope of a bounded topic time. The topic time can be realized by an adverbial or a sentential adjunct. The subject of a sentence is linked to the discourse world by the anaphoric (or non-anaphoric, as I will add below) construal of its determiner.

Given this set of assumptions, let us now examine how a given sentence may yield a generic or episodic interpretation. Let’s take for instance the generic sentence already introduced in (54) and consider its episodic counterpart in (55): (54)

Si mininu ten brinkedu, el ta fika ketu. if child has toys s/he  stays quiet ‘If a child has toys, s/he stays quiet.

(Generic)

(55)

(Gosi) mininu ten brinkedu, el ta fika ketu. now child has toys s/he  stays quiet ‘(Now) the child has toys, s/he will stay quiet.

(Episodic)

Given (53iii) and (53iv) above and keeping (53i) in mind as their anchoring foundation, the T chain for (54) connects the complementizer node  (associated with Reference or Speech time) to the tense node  carrying in this case a zero index and associated with the Event time and ends with V. In this case,  is indexed 1 for Speech time but  is indexed 0: The event time cannot be construed as anaphoric to the Speech Time, hence the generic interpretation arises. In contrast, in (55),  and  are both indexed 1. Given the matching index between Speech time and Event time, the Event time is construed as anaphoric to the Speech time and the sentence gets an episodic interpretation. The trees in (56) and (57) would then obtain, as representations of (54) and (55) respectively: (56)

[CP 1 [TP mininu T0 [VP ten brinkedu] (Generic interpretation)

(57)

[CP 1 [TP mininu T1 [VP ten brinkedu] (Episodic interpretation)





Marlyse Baptista

These two trees get the same syntactic structure but with differing indexing on T within the T-chain. Aspect (53v) also interacts with the output constraint mentioned in (53i) in TP (and DP, as will be discussed below), as the Event time interval must be bounded in order to satisfy that interface constraint. One way to bound a time interval is to place the event the sentence denotes in the scope of a bounded topic time. The topic time can be realized by an adverbial, as shown in (55) with the adverbial gosi ‘now’. However, the adverbial is not necessary, as indicated by the parentheses around gosi in (55). For this reason, the situation an episodic sentence describes must be placed in both discourse space and discourse time. I will add two more stipulations to Guéron (2006) and further propose that the T-chain extends its indexing to DP in languages such as CVC where the use of bare nouns is prevalent (Guéron & Hoekstra, 1988; Lecarme 1996, 2004). I propose that if T carries a zero index and ends within vP (the domain of spatial interpretation, see (53ii)), then D will also carry a zero index whether D is filled or not, yielding a generic, non-anaphoric interpretation for the sentence. If in contrast,  carries a 1 or 2 index (for Present or Past), then D will match the index and the episodic, anaphoric reading will then emerge. As a result, I would alter (53vi) above and suggest that the subject of a sentence is linked to the discourse world by the anaphoric or non-anaphoric construal of its determiner. The anaphoric construal of the determiner is obtained by matching D1 to T1 and the non-anaphoric construal by pairing D0 to T0. In concluding this section, an intriguing question should be addressed: Why do languages such as French and English use determiners in most of the cases listed above whereas CVC and other creoles in this volume display a predilection for bare nouns? The answer may lie in languages’ propensity for economy. If speakers judge that there is sufficient situational information, then they tend to abide by the principle of least effort. As van Loon (1986: 26) puts it: “The more predictable the content of a message thanks to contextual or situational elements, the fewer signs are required to encode it. Very frequent, grammatical words such as the, a, it, his have limited informational value because they are shortcuts for information that is sufficiently apparent from a context or situation.” Furthermore, from a historical perspective, languages seem to develop a systematic use of overt determiners only overtime. Chaudenson (this volume) presents a compelling case where in 15th century French dialects, the use of bare nouns was much more prevalent than in contemporary French.20 . Chaudenson (this volume) argues that the use of the zero determiner was prevalent in Middle French and only starts to regress between the XIV and XVII centuries. Convincing evidence for this pattern can be found in Marchello-Nizia’s (1979) study of earlier stages of the French language (XIV–XV centuries). This author states that in Middle French, in the majority of the examined texts, the noun is used without a determiner. The use of bare nouns seems correlated to the syntactic function of the noun, or to the semantic category to which it belongs: a noun referring to a more or less abstract general notion is not preceded by a determiner when it functions as a subject or in direct object position or as a prepositional complement

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

To summarize the main points of this section, the first generalization concerning the overt indefinite determiner/quantifier un is that it may introduce an indefinite specific/referential or non-specific/non-referential noun in its first occurrence. Its plural counterpart uns, acts like a quantifier. As a quantifier, un may also modify non-countable or abstract entities, as illustrated in example (9). Un may be replaced by a null determiner for marking specific/referential and non-specific/non-referential NPs. If one attempts to draw a distinction between un and its null counterpart and tries to predict the occurrence of the overt determiner, one could propose that the speaker uses un when s/he wishes to make singularity and indefiniteness explicit (following Bruyn, this volume and Mufwene 1986). The definite article kel and its plural counterpart kes may introduce specific/referential and non-specific/non-referential nouns but more often than not, the null determiner on bare noun fulfills this function. Furthermore, the appearance of the definite article may be inhibited by factors such as other specificity and referentiality-inducing modifiers (possessives, prepositional complements, adjectives, or relatives) because such modifiers already indicate referentiality and, in some cases, definiteness, as noted in Lucchesi (1993: 90). For instance, in (58) below, the relative clause modifying omi ‘man’ inhibits the use of the definite kel: (58)

Omi ki ben odja-bu e nha pai. man  come see+you is my father ‘The man who came to see you is my father.’

In an attempt to predict where bare nouns are favored and how the correct interpretation arises, we made several observations. Bare nouns occur in contexts of familiarity either because the entity is unique and forms its own class (i.e. sol ‘sun’) or because it is part of the speaker’s and hearer’s common experience and knowledge. We corroborated Lucchesi’s (1993) remark that once a noun has been introduced by an overt determiner, the occurrence of a bare noun is favored. We further noted, however, that the prior use of an overt determiner is not obligatory and that there exists a subject-object asymmetry when it comes to the interpretation of singular and plural bare nouns. Singular bare nouns tend to be interpreted as definite in subject position and as indefinite in object position whereas plural bare nouns display no preference in either position (either definite or indefinite in subject or object position). When we elaborated on how the correct interpretation may emerge, we examined the role of episodic and generic tense and adapted Guéron’s (2006) proposal. I modified the indexing in Guéron’s T-chain (involving  and ) and further stipulated that the index on T must match the index on D: The anaphoric construal of the determiner is obtained by matching D1 to T1 and the non-anaphoric construal by pairing D0 to T0. Table 1 reflects the CVC determiner system, as it has been described in this chapter. (Marchello-Nizia, 1979: 111). Furthermore, Marchello-Nizia statistically demonstrated that in the Fouke Fitz Warin, a novel from the XVI century, 51% of the nouns are without determiners, 21% appear with le and no evidence of indefinite determiner could be found.





Marlyse Baptista

Table 1. The Cape Verdean Creole determiner system21

Countables Plural Singular Noncountables

Non-referential

Referential

Indefinite

Definite

kes, Ø un, kel, Ø un, kel, Ø

kes, uns, Ø un, kel, Ø un, kel, Ø

uns, Ø un, Ø un, Ø

kes,Ø kel, Ø kel, Ø

In the next section, the focus shifts from nominal modifiers such as overt and null determiners to plural suffixation. More precisely, we briefly examine pluralization strategies in CVC with a particular focus on -s suffixation on noun stems, as they inform us about the internal structure of the Cape Verdean DP and its various layers. This analysis will reveal what type of information is encoded in -s suffixation in monolingual speech.22

.

Number marking: Pluralization strategies

A standard strategy in CVC is to mark plurality via a determiner (quantifier) (59b), a demonstrative (60b), a possessive (61), a numeral (62) or a floating quantifier (63) while the nominal stem remains bare. This makes plural marking economical. (59)

indefinite determiner a. N ben konxe un kazal. (AB-Fogo) I came know a couple ‘I came to know a couple.’ b. Otu, botadu na uns padja ki ta nese na mes others thrown on some grass   grow in month dazagua (APF-Fogo) of rain ‘Others were thrown on some grass that grows during the raining month.’

. This section has defined as accurately as possible the situational (uniqueness of entity, familiarity), discourse (previous reference to the entity) and syntactic (tense and distribution of the DP) constraints conditioning the interpretation of a bare noun as definite. The tools of accessibility theory (Ariel 1990 in Lyons 1999) may help determine whether the omission of a definite article can be fully predicted based on a statistical study of a given referent’s past occurrences. For instance, Jaggar (1985) shows that in Hausa, the omission of the article can be predicted to a large extent by the accessibility of its referent. Lyons (1999) explains that if the previous mention of a referent is too far back in the discourse and if other referents have been mentioned in the intervening discourse, a potential confusion may prevent the hearer from identifying the originally intended referent. This leads the speaker to resort to heavier coding by using an overt article. A bare noun can be predicted to be used only when the referent is judged to be easy to access.

. See Ferreira (2004) for a variationist treatment of pluralization strategies in Cape Verdean Creole and Brazilian Portuguese.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

(60)

demonstrative a. Nu ta ba mar, nu ta ben ku kel pexi. (AA-Fogo) we  go sea we  come with that fish ‘We go to the sea and come back with that fish.’ b. kes fidju, es sa na ses kaza ku ses fidju (IB-Brava) these children they are in their house with their children ‘These children, they are in their houses with their children.’

(61)

possessive N ta ba panhaba agu na paladron, dos bes pur dia, (IB-Brava) I  go take+ water in well two time per day pa pode txiga-m ku nhas fidju. for can be sufficient+me with my+ child ‘I used to go and take water from the well, twice a day so that I would have enough for me and my children.’

(62)

numeral N ten seti fidju, sinku matxu, dos femia. (FLT-ST) I have seven child five boy two girl ‘I have seven children, five boys two girls.’

(63)

quantifier a. Es ta ba anda tudu kabu. Pobason, kanpu, matu, vila, (AA-Brava) they  go walk all parts Pobason country bush city Len, tudu parti Len all part ‘They would walk to all sites, to Pobason, to the country, to the bush, to the city, to Len, to all parts.’ b. Tudu badju ki ten, N ba (AB-Fogo) all dance  have I go ‘All the dances, I attended.’ c. Tudu stranjeiru ki uji sa ta ben li, (NNNL-ST) all foreigner  today   come here tudu dja flaba ma ta binha. all  said that  came ‘All the foreigners who are coming here today, all of them had said that they would come.’

Furthermore, as already discussed in Section 2.1.3, a bare noun may yield a generic interpretation (64) or be interpreted as plural (65):

(64)

generic interpretation Nu konsa leba-l pa sidadi, pa odja pa valoriza pobri, we started take-it for city to see for valorize poor





Marlyse Baptista

pa valoriza mudjer domestika, pa valoriza mudjer privadu. (Ana Maria, ST) for valorize woman domestic for valorize woman private ‘We then took it to the city to see to it that it would valorize the poor, valorize women at home, valorize women in private.’ (65)

plural interpretation Kaza d’es rua ta parse bedju. house of+this street  look old ‘The houses in this street look old.’

In addition, the nominal stem may carry a plural suffix, as in (66) and (67). In Baptista (2002, 2003), I argue extensively that definiteness and animacy are the two chief variables in predicting plural suffixation.23 Not surprisingly, the nouns featured in examples (66) and (67) below are both [+animate], [+definite]. (66)

Ta bende un balai di pexi, nu ta kre pa kria fidjus. (MC-Brava)  sell a basket of fish we  want to raise children ‘We sell a basket of fish, we need it to raise the children.’

(67)

a.

b.

N ta spera nha kunpanherus tudu pa kunpanha-m. (FLT-ST) I  expect my companions all to accompany-me ‘I expect all my companions to accompany me.’ Nha fidjus tudu ta papia ku el (RC-ST) my children all  talk with him ‘All my children talk to him.’

In some rare instances, both the determiner and the noun may carry the plural marker. This occurs only when the referent is [+human], as in (68). (68)

Kes djentis, bedju, txeu ka ta baba skola (AM-ST) these people in the old days, a lot   go+ school p’ e djuda mamai kiria mininu for he/she help mother raise child ‘In the old days, these people did not go to school so that they would help their mothers help the children.’

The plural marking strategies in CVC are summarized in Table 2.

. This is in accord with Lyons (1999: 214) who notes that in a number of languages, only noun phrases which are relatively high in ‘animacy’ are marked with number. He also suggests that as human and definite referents are, in general, more salient in human perception than non-human ones, they are more prominent in our consciousness, hence marked. Definite referents are more to the fore in our minds than indefinites because they are familiar, hence salient. Typologists note that cross-linguistically, languages tend to grammatically mark whatever is most relevant, familiar and salient to speakers (i.e. human vs non-human, animate vs inanimate, definite vs indefinite). Lambrecht (1994) discusses in great detail the concept of salience in pragmatics (cf. Alexandre & Hagemeijer, this volume).

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

Table 2. Plural marking strategies Uns / kes + bare noun Kes + noun-s Nha (poss.)+ noun-s Bare noun (no -s suffixation) Quantifier (of various types) + bare noun Quantifier + noun-s Ø + noun-s

Let us now turn to adjective placement, as adjectives are relevant to the DP structure that we will propose in Section 2.4.

. Adjective placement Only a handful of adjectives inherited from the Portuguese lexifier may occur in a prenominal position, as shown in (69): (69)

Nu fase un bon trabadju, pa nu pode ten midju. (AF, Santiago) we did a good work for we can have corn ‘We did good work so that we could get corn.’

Otherwise adjectives typically appear in a post-nominal position, as witnessed in (70) and (71): (70)

Alem-li, kaza ki’ N ta moradu li e ka di meu, Here house    lived here is  of mine kaza bedju ma tudu mes, N ten ki paga house old but every month I have  pay tres kontu kinientu (Maria de Melo, ST) three conto five hundred ‘Here I am, the house where I am living is not mine, the house is old but every month, I must pay three contos five hundred.’

(71)

a.

Un mininu spertu ta trabadja riju. a child intelligent  work hard ‘a bright child works hard.’ b. *Un spertu mininu ta trabadja riju. a intelligent child  work hard

Adjectives occur after both common nouns, as seen in (70) and (71) and proper names (72): (72)

Praia antigu era sabi. Praia old was pleasant ‘Old Praia was pleasant.’





Marlyse Baptista

Now that overt determiners, bare nouns, plural marking strategies and adjectival placement have been examined, we are in a position to propose an analysis for the internal structure of DP.

.

An analysis

This analysis is divided into two subsections: the first proposes a multi-layered structure for the Cape Verdean DP that accommodates overt determiners, plural marking and bare nouns (countables and non-countables). These tree representations will stipulate N-raising to Num or to D in the case of bare nouns. The second subsection addresses adjective placement and brings further evidence for the N-raising analysis proposed.

.. Empirical facts: A synthesis To recapitulate, based on the empirical data examined earlier, the Cape Verdean noun displays the properties in (73): (73)

●

● ●

●

● ●

CVC is endowed with overt determiners expressing indefiniteness and definiteness. the use of bare nouns is pervasive. bare arguments are interpretable as generic, mass, definite, indefinite, singular or plural. Regarding bare singular and plural count nouns: i. bare singular nouns are preferably (but not invariably) interpretable as definite (specific/referential) in subject position and indefinite (non-specific/ non-referential or specific/referential) in object position. ii. bare plural nouns are interpretable as definite (specific/referential) or indefinite (non-specific/non-referential and specific/referential) in both subject and object positions. the interpretation of bare arguments is sensitive to Tense (episodic or generic tense). there is a count/mass distinction. morphological plural (no classifier system) i. plural inflection is sensitive to animacy (see examples (66), (67) and (68) above) ii. plural inflection is sensitive to definiteness (see (66) and (67))

In the light of these facts, a comprehensive analysis needs to answer the following questions: (74)

(a)

What are the layers of the Cape Verdean internal nominal structure regarding nouns with overt determiners, determinerless nouns, mass and count nouns? (b) Are nominal projections uniform or do they vary?

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

(c)

Are nominal projections uniform for both unmarked and marked plural bare nouns?24 (d) Does the CVC DP involve N-raising and if so, to what landing site(s)?

In the following sections, I present the assumptions underlying my analysis and propose a multi-layered DP for CVC.

.. A few background assumptions Longobardi (1994) claims that bare nouns are not really bare but are embedded in a full-fledged DP structure, with an empty D head. The idea of an empty D is partly motivated by the theoretical claim that only DPs can serve as arguments: NPs are predicates or “propositions” and D turns them into arguments. In addition, D has a singularizing or individualizing function: it has the ability to pick out a single instance of whatever is described by NP. Cheng and Sybesma (1999: 518) argue that D mediates between the description provided by the NP and whatever specific entity in the real world the description is applied to. They link this function of D to the individualizing-singularizing function of the determiner. The underlying assumption is that there is a division of labor between NP which describes, and D, which refers. Such a division of labor is a property of UG according to Cheng and Sybesma. In languages with articles/determiners, the deictic function in the nominal phrase is assumed by the article/determiner. If the describing/referring dichotomy is indeed part of UG, then, if a language has no articles/determiners, some other elements in the language must perform the deictic function, such as classifiers. It has been noted (Cheng & Sybesma 1999) that in some languages (like English), number morphology is the grammatical marker, whereas in languages lacking number morphology (like Chinese) the grammatical marker is the count classifier. There is indeed a universal tendency for languages without grammatical number to have classifiers (cf. Greenberg 1963; Cheng & Sybesma 1999: 517, Footnote 10).25 Interestingly, CVC represents a truly hybrid system in the sense that due to the availability of number morphology as seen in Section 2.2, it has not developed a classifier system (and is unlikely to do so) but CVC also uses bare nouns that are interpretable as plural (see examples (25) and (26) above), as in Chinese.

. By ‘unmarked’ and ‘marked’, I mean nouns that are interpretable as plural but carrying no overt plural marking versus nouns that carry an overt plural inflection. . For instance, as noted by Cheng and Sybesma, Ikoro (1994) sketched the development of the Nigerian Cross River Language Kana, which lost its number morphology and developed a classifier system, in contrast to related neighboring languages, which kept their number morphology and did not develop a classifier system.





Marlyse Baptista

.. The structure of non-bare and bare nouns ... DP layers. Based on the data introduced in Section 2.1 where we noted that the Cape Verdean determiners fulfill distinct functions, I propose that each function corresponds to a different site in the tree structure. Given the double life of kel/kes, I assume that it can be base-generated either in Spec-DP as a demonstrative or a non-specific/nonreferential determiner or in D as a definite specific/referential determiner.26 When the demonstrative kel/kes precedes a numeral, it is base-generated in Spec-DP (see tree in (76) below) and the numeral un in (75a) is in Spec-NumP: (75)

a.

b.

(76)

N sta na poder di kel un fidju. (B-ST) I am in power of that one child ‘I depend on that one child.’ N odja kel dos omi I saw that two men ‘I saw those two men.’ DP

Spec kel

D’ NumP

D Spec un dos

Num’ NP

Num

N’

Spec N omi

As a definite determiner, I propose that kel is generated in D. Hence, in a similar expression to (75), kel is interpreted as a definite determiner instead of a demonstrative. As such, it may appear in D and un remains in Spec-NumP. In (77), illustrated in (78) below, one can identify kel as a determiner rather than a demonstrative thanks to the relative clause modifying the NP omi. In this context, kel as a definite determiner with a specific/referential interpretation seems the most natural: (77)

.

Kel un omi k’ el kre xatia e bo. the one man  she wish annoy is you ‘The one man she wishes to annoy is you.’

See Lyons (1999: 332) for a proposal along these lines.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

(78)

DP D’

Spec

NumP

D kel

Spec un

Num’ NP

Num

N’

Spec N omi

In the same way that kel/kes have a double life as demonstrative and definite determiners (specific and non-specific), un may function as a quantifier, an indefinite determiner or a numeral. As a numeral, un is assumed to be in Spec-NumP, as in (76) and (78). As a specific/referential determiner, I propose that un is generated in D, as in (79) below; as a quantifier or indefinite non-specific/non-referential, it would be generated in Spec-DP (cf. (80)): (79)

DP D’

Spec

NumP

D un

Num’

Spec

NP

Num

N’ Spec N omi

DP

(80)

D’

Spec un

NumP

D

Num’

Spec

NP

Num

N’ Spec N omi

The plural counterpart to un, uns functions as a quantifier and can be assumed to be a QP in Spec-DP in the light of examples like (81) where the quantifier must be higher than the numeral dos ‘two’. (81)

Da-m uns dos kopu di grogu. give+me some two glass of hard liquor ‘Give me some two glasses of hard liquor.’





Marlyse Baptista

The ensuing generalization is that all quantifiers are generated in Spec-DP and this includes both un as a quantifier and its plural counterpart. (82)

DP D’

QP uns

NumP

D

Num’

Spec dos

NP

Num

N’ Spec N

PP kopu di grogu

Following Bernstein (1993), I postulate that the plural marker is generated as the head of Num and propose that Num can host overt or covert plural marking. If the plural marker is overt, as in (83) for instance, the definite plural subject omis is assumed to be base-generated in N and raises to Num to check its plural marker. Subsequently, as a definite specific/referential noun, omis is assumed to raise all the way to D. Indeed, in CVC, the fact that singular and plural definite bare nouns can have the same referential interpretation as constants and proper names, and may not be subject to lexical government, suggests N-to-D raising. Furthermore, we assume that DP is projected, as in accord with Longobardi (1994), we postulate that only DPs may function as arguments. (84) shows the derivation of the DP in sentence (83): (83)

Omis entra na kuartu. men came in room ‘The men entered the room.’

(84)

DP D’

Spec

NumP

D

Num’

Spec omi-si

Num ti-s

NP N’

Spec N ti

In the absence of an overt plural marker, as in (85), I postulate that the unmarked bare plural noun kaza still raises to Num to check a plural feature at LF and raises to D, yielding a definite specific/referential interpretation.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

(85)

Definite specific plural Kaza di es aldeia e baratu. house of this neighborhood is cheap ‘The houses in this neighborhood are cheap.’

The structure in (86) illustrates N-raising to D via the Num head for the sentence in (85). (86)

DP D’

Spec

NumP

D kazai

Num’

Spec Num [+Plur] t’i

NP N’ Spec N ti

In the case of definite singular specific/referential bare nouns, the same movement from N through Num (-Plural feature) to D is involved. The tree in (88) represents the DP in the sentence in (87). (87)

Libru sta riba di menza. book is top of table ‘The book is on top of the table.’

(88)

DP D’

Spec

NumP

D

Num’

Spec librui

Num [-Plur] ti

NP N’ Spec N

ti

In sum, the proposal is that, whenever definite, the plural determinerless nouns check features in Num and eventually raise to D whether or not they carry an overt plural marker. Hence, omis in (84) and kaza in (86) are assumed to undergo the same kind of movement within the same multi-layered DP structure. Definite singular bare nouns also involves N-to-D raising via Num where the noun checks a [-Plur] feature (i.e. libru in (88)).





Marlyse Baptista

As for indefinite nouns, I assume that they remain in N if they are non-specific/ non-referential but project DP and NumP which they may check at LF. (89)

Si bu ten terenu di fase, bu ta pidi djuda (AM-ST) If you have terrain to make you  ask help ‘If you have a terrain to make, you can ask for help.’

(90)

DP D’

Spec

NumP

D

Num’

Spec Num

NP N’ Spec N terenu LF checking

In contrast, I suggest that generic bare NPs are best analyzed as DPs27 with empty determiners, but lacking number. Zribi-Hertz (1999), (this volume) offers a similar analysis for Malagasy and Haitian respectively where she argues that NumP is optional, leading to the syntactic deficiency of such NPs. The tree in (92) featuring absence of NumP would then obtain, for generic nouns such as katxor ‘dogs’ in (91): (91)

Katxor ka gosta di gatu dog  like of cat ‘Dogs do not like cats.’

(92)

DP Spec

D’ D

NP N’

Spec N katxor

It is reasonable to assume that mass nouns such as agu ‘water’ in (93) also project DP and may undergo N-to-D raising when definite. However, as they are mass denoting,

. DP is postulated for generics as well due to their possible argumenthood status (only DPs can be arguments according to Longobardi (1994)).

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

they are devoid of NumP. Such an analysis accounts for why mass bare nouns can appear in argument positions, just like any other type of bare noun in CVC. (93)

Agu di es fonti e freska ma salgada. water of this spring is fresh but salty ‘The water in this spring is fresh but salty.’

(94)

DP D’

Spec

D agu Spec

NP N’ N ti

If the interpretation of agu were indefinite non-specific/non-referential, the head noun would remain in N, as in (32), repeated here as (95) for convenience: (95)

a.

Nu ten txafaris pa nu panha agu. (AM-ST) we have fountain to take water ‘I have the container to collect water.’

In summary, the tree representations in (76) through (94) reflect the structure of singular and plural definite and indefinite nouns endowed with overt determiners and overt plural marking. (86) and (88) represent the interpretation of definite bare nouns in the absence of overt determiners or plural marking. We postulate N-to-D raising through NumP for definite specific/referential nouns (i.e. (84) for instance). In these cases, the NumP projection is stipulated whether the plural marker is overt, as in (84) or not, as in (86). (92) conveys the structure of generic nouns and (94) that of referential mass nouns. Finally, the tree in (96) below summarizes the positions of the various determiners, QPs, and numerals:28 DP

(96)

D’ Spec

NumP D

Spec kel/s [dem] uns [QP] Num un [QP] kel/s [def] -s un [ind, spec.] un [numeral] dos

Num’ NP Spec

N’ N

. Alternatively to the tree in (96), we could postulate that the topmost projection is a QP in the case of quantificational nominals and DP for their referential counterparts. However, given that we maintain in this paper that only DPs can be arguments (following Longobardi 1994), only DP will be projected as the topmost projection.





Marlyse Baptista

We end this section by examining adjectival placement in CVC, as it lends further support to the N-to-D raising analysis we provided in the above sections.

... Adjective placement in CVC. As discussed in Section 2.3, adjectives in CVC are post-nominal. Following Cinque (1993) and Bernstein (1993), I assume adjectives in Romance and Romance creoles such as CVC are base-generated in a prenominal position. The post-nominal position is derived by N-movement across the adjective. Like Bernstein (1993), I assume that APs, like adverbs, are adjoined to XPs. In the presence of a determiner, a noun moves from N to Num (98) but when the determiner is absent, a definite noun may move all the way to D (100). (97)

kel mininu spertu. the child smart ‘The smart child.’

(98)

DP D’

Spec

NumP

D

Num’

Spec

mininui

(99)

mininu spertu child smart ‘The smart child.’

(100)

DP

NP

Num

kel

N’ AP spertu

N ti

D’

Spec D mininui

NumP Num’

Spec Num ti

NP N’ AP spertu N ti

In summary and in response to the questions raised in (74), this section showed that multiple layers are stipulated for the Cape Verdean DP and that nominal projections

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

are not uniform across noun types. We propose for instance, that mass and count nouns do not involve the same number of projections (mass nouns have no NumP). In contrast, nominal projections for both unmarked and marked plural bare nouns were assumed to be the same whether the plural inflection is overt or covert. Finally, we argued that bare specific/referential nouns involve N-raising to D (possibly through NumP if it is projected). In the presence of an overt determiner, the noun can potentially raise to Num. Having examined the structure and semantics of bare nouns in CVC, this chapter would be incomplete without comparing the data at hand with the determiner systems of European Portuguese (the lexifier of CVC) and of Brazilian Portuguese, a variety of Portuguese that has much in common with CVC. This is the topic of the next section.

.

A comparison with European and Brazilian Portuguese

This section is not meant to provide a thorough presentation of the determiner systems of European Portuguese (EP) and of Brazilian Portuguese (BP). Such systems, just like their CVC counterparts, are highly complex and will be the focus of an entirely separate paper in the future. The scope of this section is more modest and aims at highlighting in what respects the CVC determiner system differs from its lexifier (European Portuguese) and one other Portuguese variety, Brazilian Portuguese.

.

A comparison with European Portuguese and Brazilian Portuguese29

Both EP and BP make use of overt determiners to mark definiteness and indefiniteness. Both types of determiners agree in number and gender with the nouns they modify.30 The definite determiners are: o (masculine, singular), a (feminine, singular), os (masculine, plural) and as (feminine, plural). The indefinite counterparts are um, uma, uns, umas. Both European and Brazilian Portuguese have distinct markers in the area of demonstrative (marking proximity (101a) and distance (101b)), and definite determiners (102). Demonstrative and definite markers: (101)

a.

(Eu) comprei esta casa. (EP and BP) I bought this house ‘I bought this house.’

. I am greatly indebted to Fernanda Pratas and to Fernanda Ferreira for their generous and valuable help with the European Portuguese and Brazilian Portuguese data, respectively. Ana Maria Brito and Tjerk Hagemeijer were also extremely helpful with the last draft of this paper. . Ana Maria Brito (p.c.) informs me that number agreement in BP does not always take place as expressions such as os menino (with no plural agreement on menino ‘child’), or os meu menino (no plural agreement on the possessive or the noun) are possible.





Marlyse Baptista

b.

(102)

(Eu) comprei aquela casa. (EP and BP) I bought that house ‘I bought that house.’

(Eu) comprei a casa que estava à venda na tua rua. (EP and BP) I bought the house that was for sale in your street ‘I bought the house that was for sale on your street.’31

The following examples are counterparts to the CVC data presented in Section 1. They are meant to convey the same array of interpretations discussed in CVC in earlier sections of this paper while reflecting that the use of bare nouns is much more restricted in both EP and BP. Most of the examples that follow are representative of both EP and BP. In the few cases where BP behaves differently from EP, such examples will appear explicitly labeled as such. EP and BP Just like CVC, both EP and BP use the numeral um, morphologically identical to the indefinite determiner um.32 (103)

Numeral Ao fim de um ano, temos mandioca. At end of one year we have manioc ‘After one year goes by, we get manioc.’

In the case of non-specific/non-referential singular nouns, the determiners um/uma are used. (104)

Indefinite non-specific/non-referential singular Para preparar este prato, é preciso uma panela33 to prepare this dish is necessary a pan e um tabuleiro. and a tray ‘To make this dish, you need a pan and a tray.’

On the other hand, if a noun can obtain a mass or more abstract reading, the indefinite determiner seems optional, as witnessed in (105): (105)

Se tiveres (um) terreno, podes construir uma casa. if you have (a) land you can build a house ‘If you have land, you can build a house.’

. Interestingly, Fernanda Pratas (p.c.) notes that (Eu) comprei casa na tua rua using no determiner before casa is grammatical, as long as house has the meaning of ‘residence’ rather than house per se. .

Um can also modify a generic entity, see example (119) below.

.

Ana Maria Brito informs me that panela could also appear bare.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

In this particular case, the optionality of the determiner in (105) seems to be correlated to the fact that terreno can have a concrete, tangible meaning (a property, a terrain) or a more abstract, mass-like meaning (land, grounds).34 The indefinite determiner also typically modifies indefinite specific/referential singular nouns. (106)

(107)

Indefinite specific/referential singular Eu tenho um filho que tem 5 anos (de idade). I have a son who has 5 years ‘I have a son who is 5 years old.’ Indefinite specific/referential singular Tínhamos um líder, mas ele morreu. we had a leader but he died ‘We had a leader but he died.’

This is a first area where BP behaves distinctly from EP and converges with CVC, as it may use a bare noun to convey an indefinite specific/referential singular meaning. (108)

Indefinite specific/referential singular (BP) Eu comprei computador.35 I bought computer ‘I bought a computer.’

In the realm of indefinite non-specific/non-referential plural nouns, a dual system emerges (for both EP and BP) in which the noun appears bare (but carrying the plural marker) as in (109), or is modified by the determiner as/os (110), (111). (109)

(110)

(111)

Indefinite non-specific/non-referential plural Quando construímos casas, ajudamo-nos uns aos outros. when we build houses help-we ones with the others ‘When we build houses, we help each other.’ Indefinite non-specific/non-referential plural Naquele tempo as crianças não iam à escola. in that time the children  went to school ‘In those days, children did not go to school.’ Estas assistentes sociais ajudam as mães these assistants social help the mothers e os pais a manter as crianças na escola. and the fathers to keep their children in school ‘These social workers help mothers and fathers keep children in school.’

. Tjerk Hagemeijer (p.c.) observes that this type of noun incorporation seems favored with abstract nouns such as terreno ‘land’. . This particular example has been adapted from Kester & Schmitt (this volume).





Marlyse Baptista

This is another area in which BP radically departs from the EP determiner system, as in addition to this dual system, BP may opt for an alternative also available in CVC, that of using a totally bare noun, meaning with no determiner and no plural marker. This is exemplified in (112):36 (112)

Indefinite non-specific/non-referential plural Naquele tempo, criança não ia para a escola. (BP) in those days child  went for the school ‘In those days, children did not go to school.’

The same situation arises for indefinite specific/referential plural nouns, as they can appear as bare plurals (no determiner but presence of a plural suffix), as in (113) or completely bare (no determiner, no plural suffix), as in (114): (113)

(114)

Indefinite specific/referential plural Ela tem barcos. (BP) she has boats ‘She owns boats.’ Indefinite specific/referential plural Ela tem barco.37 (BP) she has boat ‘She owns boats.’

Fernanda Ferreira (p.c.) notes that in BP, the singular form is much more frequently used than the plural form in such cases. Let us note that the expression of generic in EP and BP has much in common with the various strategies English uses to convey genericity, although the English options represent a subset of those available in BP. English may use the definite determiner (as in (115)), the indefinite determiner (as in (116)), or a bare plural to express genericity (117): (115)

The lion is a noble animal.

(116)

A lion is a noble animal.

(117)

Lions are noble animals.

The same options are available to EP and BP where the definite determiner (both singular and plural) may be used (118a) and (118b), as well as the singular indefinite determiner (119), or bare plurals (120): (118)

Generic a. O leão é um animal nobre. the lion is an animal noble ‘The lion is a noble animal.’

. I thank Fernanda Ferreira for bringing these valuable examples to my attention. . According to Tjerk Hagemeijer, for some Brazilian speakers, this sentence could be interpreted as ‘she has a (one) boat.’

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole

b.

Os leões são animais nobres. the lions are animals noble ‘Lions are noble animals.’

(119)

Um amigo não faz uma coisa destas. a friend  do a thing of these ‘A friend does not do such a thing.’

(120)

Amigos não fazem coisas destas. friends  do things of these ‘Friends do not do such things.’

In addition to these options, BP uses a bare singular form,38 just like CVC.39 (121)

Amigo nao faz isso. friend NEG do this ‘Friends do not do this.’

(BP)

The definite determiner modifies both definite specific/referential singular nouns (122) and definite specific/referential plural nouns (123) and (124).

. The data in Kester & Schmitt (this volume) confirm Fernanda Ferreira’s intuitions about BP bare singulars. Kester & Schmitt characterize the use of bare singulars as conveying two distinct types of interpretations. They note that bare singulars in BP allow both generic and existential readings (corresponding to our indefinite specific/referential) depending on the predicates they appear with. Examples (1) and (2) illustrate generic readings in subject and object position; (3) and (4) illustrate existential readings (all these examples are taken from Kester & Schmitt (this volume)). (1)

(2)

(3)

(4)

.

Cabrito é muito comum em Curaçao. (BP) goat is very common in Curaçao ‘Goats are very common in Curaçao.’ Eu adoro gato. (BP) I love cat ‘I love cats.’ Tem computador na minha mesa. (BP) have computer on my desk ‘There is a computer/computers on my desk.’ Eu comprei casa/computador. (BP) I bought house/computer ‘I bought a house/houses/a computer/computers.’

Again, I thank Fernanda Ferreira for drawing my attention to this piece of data.



 Marlyse Baptista

Definite specific/referential singular (122)

a.

b.

Eu trabalho para a câmara municipal. I work for the chamber municipal ‘I work for the city hall.’ Muita gente quer ouvir a história da organização. lots people want hear the story of+the organization ‘A lot of people want to hear the story of the organization.’

Definite specific/referential plural (123)

As casas deste bairro são baratas. the houses of this neighborhood are cheap ‘The houses of this neighborhood are cheap.’

(124)

Os vizinhos não se davam bem uns com os outros. the neighbors   gave well ones with the others ‘The neighbors didn’t get along with each other.’

Both EP and BP use a variety of quantifiers, a limited set being exemplified in (125) and (126). (125)

(126)

Quantifier a. Preciso de uma pequena ajuda da tua parte. I need of a little help from your side ‘I need a little help from you.’ a. Preciso de alguma ajuda da tua parte I need of some help from your side ‘I need some help from you.’ Ela tem alguns barcos na marina. she has a few boats in+the harbor. ‘She owns a few boats in the harbor.’

Mass nouns or mass-type nouns typically occur with no determiner, as shown in (127) and (128). (127)

(128)

Mass nouns Precisas de um balde para trazer água. need of a bucket to bring water ‘You need a bucket to bring water.’ Para preparar este prato, móis o milho, to prepare this dish crush the corn móis o trigo, fazes massa com essa farinha crush the wheat make dough with this flour e comes com peixe ou leite. and eat with fish or milk ‘To make this dish, you crush corn, you crush wheat, you make dough with the flour and eat it with fish or milk.’

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole 

Proper names occur with a determiner in both subject and object positions in EP but no determiner appears in BP, just like in CVC.

(129)

(130)

Proper names a. O João tem duas vacas.40 (European Portuguese) the João has two cows ‘João owns two cows.’ b. (Eu) vi o João no mercado. I saw the João in+the market I saw João at the market Proper names a. João tem duas vacas.(BP) João has two cows ‘João owns two cows.’ b. (Eu) vi João no mercado. I saw João in+the market ‘I saw João at the market.’

Unique entities such as ‘the sun’ can appear bare (131a) or be preceded by the definite determiner (131b):

(131)

Unique entities a. Precisamos de sol e de chuva para cultivar fruta e vegetais. we need of sun and of rain to cultivate fruit and vegetables b. Precisamos do sol e da chuva para cultivar fruta e vegetais. we need of+the sun and of+the rain to cultivate fruit and vegetables ‘We need the sun and the rain to grow fruit and vegetables.’

Fernanda Pratas (p.c.) states that one can have both versions, without and with determiner, but she expressed a marked preference for the version without; she believes that

. According to Fernanda Pratas (p.c.), the definite determiner always modifies proper names in EP, except in cases of full names such as George Bush está a perder popularidade ‘George Bush is losing popularity’. However, in common discourse, the determiner can modify last names as well as in: O Bush quer atacar o Irão ‘The Bush wants to attack Iran.’ There is, however, absence of determiner in contexts like O meu primo chama-se João ‘My cousin’s name is João’. I thank Fernanda Pratas for providing these examples and for her insightful observations. According to Ana Maria Brito (p.c.), proper nouns should not all be lumped together: there seems to be differences in EP between person names, geographical names, country names. With person names, the presence of definite article is related to familiarity or degree of respect: it is possible to say o Bush / Bush. Tjerk Hagemeijer (p.c.) feels that there is also dialectal variation with regard to use of definite determiners before proper nouns in EP.

 Marlyse Baptista

the version with the determiner needs to be context-based, as in ‘I came to Algarve because I need the sun and the rain…’ meaning ‘the sun and the rain that exist here’.41 Just as in CVC (i.e. examples (49) and (50)) and English (51) and (52)), episodic and generic tense plays a role in building the correct interpretation of the noun in EP and BP. In (132), Guéron’s proposal (discussed in Section 2.1.3.2) would apply, as the units T, C and (we add) D interact in yielding the correct generic and episodic construals of nominal expressions. (132)

Generic/universal/episodic a. Em geral, se uma criança tem brinquedos, fica sossegada. in general if a child has toys stay quiet ‘In general, if a child has toys, s/he stays quiet.’ (generic) b. Se as crianças têm um brinquedo, estão felizes. if the children have a toy are happy If the child has a toy, she is happy. (universal reading) c. Ontem, sempre que a criança tinha um brinquedo, ficava feliz. yesterday whenever that the child had a toy stayed happy Yesterday, whenever the child had a toy, she was happy (episodic tense)

In terms of general conditions under which bare nouns occur in CVC, such conditions do not apply in either EP or BP. Indeed, contrary to CVC where the first mention of a noun modified by the overt determiner un allows it to recur bare in (40) and (41) for instance, such an option is not available in either EP or BP. In these varieties of Portuguese, the definite determiner is obligatory, hence, in the particular case of (133), a must appear in the second occurrence of the noun doença. (133)

O Pedro tem uma doença desconhecida e *(a) doença está a agravar-se. the Pedro has a disease unknown and the disease is  worsen- ‘Pedro has an unknown disease and the disease is getting worse.’

In summary, the use of bare nouns in CVC is much more widespread than in both EP and BP. In the latter, bare nouns (no determiner and no plural marking) are used in the case of abstract, mass or mass-type nouns and bare plurals (no determiner but plural marking is present) are used in the case of generics and indefinite non-specific/nonreferential plural nouns. However, BP departs from EP and behaves just like CVC in several respects. It uses bare nouns (no determiner, no plural marking) in the case of generics, as shown in (121) and in the case of indefinite non-specific/non-referential plural nouns (i.e. (112), with indefinite specific/referential singular (108) and plural

. Regarding example (131), Ana Maria Brito (p.c.) observes that sol ‘sun’ is in post-verbal position and post-preposition position in this example. She shrewdly points out that in subject position, the same options would not be available: O sol é a nossa estrela ‘the sun is our star’ would be the only acceptable utterance. * Sol é a nossa estrela ‘sun is our star’ would be ungrammatical.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole 

(114) nominals). Finally, in BP, proper names are never modified by the definite determiner, just as in CVC. The table in (3) highlights the domains in which bare nouns are possible in all three languages. Ungrammatical occurrences of bare nouns are marked by the symbol *Ø. Bare nouns with no determiners and no plural marking are signaled by the symbol Ø(-PL) whereas bare nouns with no determiners but carrying the plural morpheme -s are indicated by Ø(+PL). Table 3. A comparative overview of bare nouns in Cape Verdean Creole, European Portuguese and Brazilian Portuguese

Countables Plural Singular

Non-specific/nonreferential + generic

Referential Indefinite

Referential definite

CVC

CVC EP BP

CVC EP

BP

CVC EP BP

Ø Ø

*Ø *Ø

*Ø *Ø

Ø

*Ø Ø

Ø

Ø

Ø

Ø

NA

BP

Ø(-PL) Ø(+PL) Ø(-PL) Ø Ø *Ø Ø Ø

Noncountables Ø (abstract or mass or mass-type)

.

EP

Ø

Ø

Ø

*Ø Ø *Ø Ø Ø

Ø

Proper names

Conclusion

In this chapter, I have tried foremost to provide an accurate description of the semantics and structure of the Cape Verdean noun phrase involving both determined and determinerless nouns. In the domain of overt determiners, we have argued that both the definite and indefinite determiner perform at least two functions in the language. The definite kel/kes works as a demonstrative and also fulfil the role of a definite in the language modifying specific/referential or non-specific/non-referential nouns. The indefinite un functions as a numeral, quantifier and determiner. Its plural counterpart uns acts as a quantifier. The indefinite un may refer to a specific or non-specific entity. Both un and kel may be replaced by a null counterpart that has been shown to cover the same range of meanings. Bare nouns can be interpreted as generic, mass, definite, indefinite, singular or plural. Bare plural nouns are interpretable as definite (specific/referential or non-specific/ non-referential) or indefinite (specific/referential or non-specific/non-referential). In contrast, bare singular nouns tend to be (but not always) interpretable as definite (specific/referential) in subject position and indefinite (non-specific/non-referential or specific/referential) in object position. The interpretation of bare arguments was also shown to be sensitive to Tense (episodic versus generic). We adapted Guéron’s proposal (2006) and showed how the heads in T, C and D interact via an indexing system in yielding the correct interpretation

 Marlyse Baptista

of bare nouns. Various tests from Schmitt & Munn (1999) and Kester & Schmitt (this volume) helped establish that there is a distinction between bare count and mass nouns. In addition, the plural inflection was shown to be sensitive to animacy and definiteness. This study of the Cape Verdean DP also examined adjective placement in an attempt to detect N-raising and identify the nature and number of structural layers of DP. The nominal projections were not assumed to be uniform across noun types. For instance, I propose that count nouns involve DP, NumP and NP projections, while regular generics and mass nouns only involve DP and NP. Finally, I compared CVC to both its lexifier EP and to PB and showed that while the use of bare nouns is much more prevalent in CVC, CVC and BP had much in common in several domains. They may both use a bare noun in the true sense of the term (no determiner, no plural marking) with generics, indefinite non-specific/nonreferential plural nouns, and indefinite specific/referential singular and plural nouns. CVC and BP use of proper names is also identical (no determiner is involved).

References Ariel, M. 1990. Accessing Noun Phrase Antecedents. London: Routledge. Baptista, M. 2003. Inflectional plural marking in creoles and pidgins: A comparative study. In The Phonology and Morphology of Creole Languages, I.Plag (ed.), 315–332. Tübingen: Niemeyer. Baptista, M. 2002. The Syntax of Cape Verdean Creole: The Sotavento varieties. Amsterdam: John Benjamins. Baptista, M. 2000. Variabilité référentielle des syntagmes nominaux nus en créole cap-verdien. Paper presented at the University of Paris III Linguistics Symposium. April 7th, 2000. Baptista, M. 1998. The role of null morphemes in marking specificity and non-specificity of NPs: A comparative analysis. Paper presented at the Society of Pidgin and Creole Linguistics/Linguistic Society of America. Held in New York City. January 9, 1998. Bernstein, J. 1993. The syntactic role of word markers in null nominal constructions. Probus 5(1–2): 5–38. Cheng, L.L-S. and Sybesma R. 1999. Bare and not-so-bare nouns and the structure of NP. Linguistic Inquiry 30(4): 509–542. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6. 339–405. Cinque, G. 1993. On the evidence of partial N-movement in the Romance DP. Venice Working Papers in Linguistics 3(2): 21–40. Comrie, B. 1981. Language Universals and Linguistic Typology: Syntax and morphology. Oxford: Blackwell. Croft, W. 1990. Typology and Universals. Cambridge: CUP. Ferreira, F. 2004. Pluralization strategies in Cape Verdean Creole and Brazilian Portuguese: A variationist approach. Paper presented at the Society for Pidgin and Creole Linguistics conference in Boston, Jan. 2004. Ferreira, F. In preparation. Marcadores de Plural no Português Brasileiro e Crioulo Cabo-verdiano. In Português em Contato [Série Lingüística Luso-Brasileira], A.M. Carvalho (ed.). Madrid/ Frankfurt: Iberoamericana/Vervuert.

Chapter 3. On the syntax and semantics of DP in Cape Verdean Creole  Greenberg, J. 1963. Universals of Language. Cambridge MA: The MIT Press. Guéron, J. and Hoekstra, T.H. 1988. T-chains and the constituent structure of auxiliaries. In Constituent Structures, A. Cardinaletti, G. Cinque & G. Giusti (eds), 35–99. Venice: Annali di Ca’ Foscari. Guéron, J. 2006. Generic sentences and bare plurals. In Proceedings of the Brussels Conference on Indefinites and Weak Quantifiers, S. Vogeleer and L. Tasmowski (eds.), 219–234. Amsterdam: John Benjamins. Ikoro, S. 1994. Numeral classifiers in Kana. Journal of African Languages and Linguistics 15: 7–28. Jagger, P. 1985. Factors Governing the Morphological Coding of Referents in Hausa Narrative Discourse. PhD Dissertation, UCLA. Kayne, R. 2004. Case studies of non-optionality. Paper presented at the Susumu Fest, in honor of Susumu Kuno’s retirement. New York City. May 2004. Kihm, A. 1994. Kriyol Syntax: The Portuguese-based Creole language of Guinea-Bissau. Amsterdam: John Benjamins. Lambrecht, K. 1994. Information Structure and Sentence Form – Topic, Focus, and the Mental Representations of Discourse Referents. Cambridge: CUP. Lang, J. 1990. A categoria número no crioulo caboverdiano. Papia, 1: 15–25. Lecarme, J. 1996. Tense in the nominal system: The Somali DP. In Studies in Afroasiatic Grammar, J. Lecarme, J. Lowenstamm & U. Shlonsky (eds), 159–178. The Hague: HAG. Lecarme, J. 2004. Tense in nominals. In The Syntax of Time, J. Guéron & J. Lecarme (eds), 440–475. Cambridge MA: The MIT Press. Longobardi, G. 1994. Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry 25:4. 609–665. van Loon, J. 1986. Principles of Historical Morphology. Heidelberg: Universitätsverlag Winter. Lucchesi, D. 1993. The article systems of Cape-Verde and Sao-Tomé Creole Portuguese: General principles and specific factors. Journal of Pidgin and Creole Languages. 8: 81–108. Lyons, C. 1999. Definiteness. Cambridge: CUP. Marchello-Nizia, C. 1979. La langue Française au XIVe et XIe siècle. Paris: Bordas. Mufwene, S. S. 1986. Number delimitation in Gullah. American Speech 61: 33–60. Patrick, P. 2004. Jamaican creole: Morphology and syntax. In A Handbook of Varieties of English. Vol 2: Morphology and syntax, B. Kortmann, E.W. Schneider, C. Upton, R. Mesthrie & K. Burridge (eds), 407–438. Berlin: Mouton de Gruyter. Porterfield, L. & Srivastav, V. 1988. (In)definiteness in the absence of articles: Evidence from Hindi and Indonesian. WCCFL: 265–276. Schmitt, C. & Munn, A. 1999. Against the nominal mapping parameter: Bare nouns in Brazilian Portuguese. NELS 29. Zribi-Hertz, A. 1999. Towards a modular theory of linguistic deficiency: Evidence from Malagasy personal pronouns. Natural Language and Linguistic Theory 17: 161–218.

 

Papiamentu and Brazilian Portuguese A comparative study of bare nominals Ellen-Petra Kester and Cristina Schmitt* This paper presents a comparative study on bare nominals in Papiamentu and Brazilian Portuguese, taking into account also the distribution of definite and indefinite articles in these languages. Following Munn and Schmitt (2001) and Schmitt and Munn (1999, 2003), we argue that bare singulars in both languages are nouns phrases that lack interpretable number, which allows them to be interpreted both as singular or plural depending on the context. Semantically they can be interpreted as kinds as in Chierchia (1998). The differences are to be explained by differences among the items competing for the same position in the syntactic structure. Bare plurals in both languages, on the other hand, differ syntactically and semantically. While bare plurals in Papiamentu are disguised bare plurals in the sense that they have a definite feature, which dramatically restricts their distribution, bare plurals in Brazilian Portuguese behave like English bare plurals.

Introduction In this paper we extend the empirical domain of research on bare nominals by comparing bare plurals and bare singular nominals in argument position in two languages that also have definite and indefinite determiners. Empirically, our goal is to provide a comparative description of bare nominals in Papiamentu

* Authors are in alphabetical order. We would like to thank Alan Munn for comments on various versions of this paper and the Papiamentu speakers we consulted in the Netherlands: Walter Bakhuis and Ivor Curie. Special thanks to the Fundashon pa Planifikashon di Idioma in Curaçao for the cooperation and hospitality. We are grateful to the director of the Fundashon, Ronald Severing, and the staff who generously shared their judgements with us: Mariselle Bermúdez, Shudari Brenet, Cathleen Giterson and Ini Statia. This paper would have been impossible without their help. We also thank the reviewers and editors for their remarks and suggestions. Needless to say, we alone are responsible for any errors or misinterpretations.

 Ellen-Petra Kester and Cristina Schmitt

(a Portuguese-based1 Creole spoken in the ABC-islands Aruba, Bonaire and Curaçao) with bare nominals in Brazilian Portuguese.2 Since the existence and interpretation of bare nominals in a particular language is arguably sensitive to the lexical options available in that language (Chierchia 1998, Dayal 2002), it will also be important to compare the behavior of the definite determiner and, to a lesser extent, the behavior of the indefinite determiner in both languages. Our analysis will provide empirical support for a neo-Carlsonian analysis of bare nominals as names of kinds in the spirit of Chierchia (1998). We will argue, however, that, cross-linguistically, the distribution and interpretation of bare nominals in argument position is to be associated to functional elements and may have two different sources: (i) the variation can be associated with differences in the semantics of the features of functional heads associated to the bare nominals in different languages, or (ii) the variation can be associated not to the syntactic and semantic properties of the functional structure of the bare nominals themselves, but to the syntactic and semantic properties of other functional elements that compete for the same positions in the syntactic structure. In other words, identical semantics can be encoded in the syntax by means of distinct forms.3 By comparing Papiamentu and Brazilian Portuguese, we will try to demonstrate that these languages show many similarities, but also some differences in the behavior of bare singulars and clear differences in the behavior of bare plurals. We associate the differences in the behavior of the bare nominals to the different sources: the different distribution of bare plurals will be trivially associated to a difference in the features that compose the bare plural in both languages. The differences in the distribution of bare singulars, however, will not be attributed to a contrast in terms of different semantic features. Rather, we argue that bare singulars have the same semantics, but . The question of whether Papiamentu has originated as a Portuguese- or Spanish-based Creole has been a long-standing issue in the literature. However, as pointed out in Goodman (1987), it would have been impossible for Papiamentu to have begun as a Spanish-based pidgin. Rather, Papiamentu seems to be Portuguese-based diachronically and has undergone gradual relexification by Spanish due to the extensive business contacts between Curaçao and the Venezuelan coast and the immigration of Spanish-speaking Jews from Holland (also see the West African roots of Papiamentu in Martinus, 1996). Obviously, the lexicon of Papiamentu has also been influenced strongly by Dutch, which has been the official language of the ABC-islands almost uninterruptedly since 1634. . Needless to say, there is a fair amount of variation among speakers both of Papiamentu and Brazilian Portuguese. In this paper we will report converging judgements from an educated subset of speakers. The Papiamentu judgements are based on the variety spoken in Curaçao. We will report non-convergence whenever that is the case. . For example, in English, it is unlikely that there are semantic differences between the comparative -er and the comparative more in bigger and more interesting, for example, although there may be syntactic differences.

Chapter 4. Papiamentu and Brazilian Portuguese 

their distribution is more restricted in Brazilian Portuguese due to the fact that the definite determiner is required in many more contexts in this language. Hence, the different distribution of bare singulars in Papiamentu and Brazilian Portuguese is related to the different semantic properties of the definite articles in the two languages. This paper is organized as follows: first we provide an overview of the determiner system in both languages and we present the distribution of bare nominals in both languages. Section 2 compares the definite and the indefinite determiners in Papiamentu and Brazilian Portuguese. Section 3 proposes an analysis of the bare singulars in the spirit of Schmitt and Munn (1999, 2003) and Munn and Schmitt (2001) and section 4 sketches an analysis of the bare plurals in Papiamentu that can account both for the similarities and the differences in both languages. Section 5 discusses remaining issues and summarizes the conclusions.

.

Overview of the nominal system in Papiamentu and Brazilian Portuguese: similarities and differences

Both Brazilian Portuguese and Papiamentu have a definite and an indefinite determiner. In Brazilian Portuguese definite and indefinite articles inflect for gender and number, unlike their Papiamentu counterparts, which are invariable. As illustrated in Table 1, plurality is expressed in Papiamentu by the element -nan which in general attaches to the right of the noun. Both languages also make use of

Table 1. Bare, definite and indefinite noun phrases in Papiamentu, Brazilian Portuguese, Dutch, Spanish and English Bare singular

Bare plural

Definite

Indefinite

Papiamentu

buki

bukinan

un buki

Br. Portuguese

livro

livros

Dutch

*boek

boeken

Spanish

*libro4

libros

English

*book

books

e buki e bukinan o livro os livros het boek de boeken el libro los libros the book the books

um livro uns livros een boek un libro unos libros a book

. There is a restricted set of cases in which a bare nominal can appear in the object position of certain verbs. See Bosque (1996).

 Ellen-Petra Kester and Cristina Schmitt

bare count nouns in argument position (henceforth bare singulars) and also allow what we will call here bare plurals in argument position. The overview in Table 1 summarizes both the Brazilian Portuguese and Papiamentu options in argument position in comparison with English, Dutch and Spanish. The data from Dutch and Spanish, two of the lexifier languages of Papiamentu, are meant to show that we cannot associate the existence of bare singulars to an interference from these languages. At first sight, with the exception of the plural indefinite, it seems that both Papiamentu and Brazilian Portuguese have similar options and therefore one could hypothesize that bare nominals should display similar behavior in terms of distribution and interpretation, all things being equal. Indeed, a cursory inventory of examples with singular bare nominals suggests important similarities in terms of both interpretation and distribution, as will see below.

.

Bare singulars in both languages: similarities and differences

There are two main positions in the analysis of bare nominal arguments: bare nominals are treated as names of kinds (Carlson 1977, Chierchia 1998, Dayal 2002 and others) or as systematically ambiguous between a kind and an indefinite reading (Kratzer 1995, Diesing 1992 and others). In the following we describe the basic properties of bare singulars and we show that it is best to treat bare singulars as names of kinds and not as singular indefinites. Among the pieces of evidence presented by Carlson (1977) in favor of the treatment of bare plurals in English as names of kinds are the following: (i) bare plurals can appear as subjects of predicates that can only apply to kinds; (ii) bare plural interpretations vary depending on the context; and (iii) bare plurals cannot take wide scope over negation, intensional verbs and aspectual adverbs. When we examine bare singulars in both Papiamentu and Brazilian Portuguese, we find that they display the same behavior as bare plurals in English with respect to the properties mentioned above. First, bare singulars may appear as subjects of predicates that apply to kinds only, as in (1a,b). (1)

.

a.

Papiamentu5 Kabritu ta6 mashá komun na Kòrsou. Goat is very common in Curaçao Brazilian Portuguese

Henceforth P (Papiamentu) and BP (Brazilian Portuguese).

. The element ta appears in copula constructions, as in (1a), and also corresponds to the TMA-marker of the present tense, as in (2a). Although ta may be related etymologically to the Spanish/Portuguese verb estar, it has a different distribution in Papiamentu: ta is used both in nominal copula constructions (where Spanish uses ser) and with locatives (where Spanish uses estar). See Maurer (1988).

Chapter 4. Papiamentu and Brazilian Portuguese 

b.

Cabrito é muito comum em Curaçao. goat is very common in Curaçao ‘Goats are very common in Curaçao.’

Second, in both languages bare singulars allow generic and existential readings depending on the predicates they appear with. Examples (1a,b) and (2a,b) illustrate generic readings in subject and object position; (3a,b) and (4a,b) illustrate existential readings.7 (2)

a. b.

Mi ta gusta pushi. I  love cat Eu adoro gato. I love cat ‘I love cats.’

(3)

a. b.

Tin kòmpiuter riba mi mesa. Tem computador na minha mesa. have computer on my desk ‘There is a computer/computers on my desk.’

(4)

a.

kumpra kas/kòmpiuter. Mi a8 I  buy house/computer Eu comprei casa/computador. I bought house/computer ‘I bought a house/houses/a computer/computers.’

b.

P BP

P BP

P BP

Third and more importantly, only narrow scope readings are possible for the bare singular with respect to other elements such as negation, intensional verbs and durative adverbials, as illustrated in (5), (6) and (7).9 The contrast between (7) and (8) . An existential reading for an indefinite noun phrase simply means for us that this indefinite is under the scope of an existential quantifier, which does not necessarily come from the indefinite itself. (3) is an existential sentence. In English existential sentences are formed by ‘there be XP YP’. In BP and P, as in other Romance languages, existential sentences are formed with ‘have XP Y’, where XP is a DP that is a weak DP and not a presuppositional DP. .

a indicates past tense and ta present tense.

. The indefinite a in English in a sentence such as (i) has two readings (ia) and (ib), respectively. In (ia) the indefinite is outside the scope of negation and in (ib) it is under the scope of negation, as the paraphrases indicate. Any time the indefinite has scope over another operator we call this reading the wide scope reading of the indefinite, as in (ia). When the operator (in this case, negation) has scope over the indefinite, the indefinite is said to have narrow scope. This is the case in (ib). (i)

John didn’t kill a fly. a. There is a fly and John didn’t kill it. b. It is not the case that John killed a fly.

In this paper we are not committed to any particular analysis of the scopal properties of indefinites. The terms narrow and wide scope are being used descriptively.

 Ellen-Petra Kester and Cristina Schmitt

shows that the bare singular does not behave like the indefinite singular, since, while the bare singular allows a plural interpretation of iguana, the indefinite singular only allows a pragmatically odd reading of the predicate. If there is only one single iguana, in order to fill the two-hour interval, we have to imagine a very special scenario of perhaps being able to kill the same iguana many times. (5)

a. b.

(6)

a. b.

(7)

a. b.

(8)

a. b.

(neg < object and *object < neg)10 Mi no a mira mancha riba suela. I not  see spot on floor Eu não vi mancha no chão. I not saw spot on-the floor ‘I didn’t see spots on the floor.’ (want < object and *object < want) Maria kier kasa ku brasileño. Maria quer casar com brasileiro. Maria wants to-marry with Brazilian ‘Maria wants to marry a Brazilian.’ (any Brazilian) (adv < object and *object < adv) Mi a mata yuana pa dos ora largu. I  kill iguana for two hour long Eu matei iguana por duas horas. I killed iguana for two hours ‘I killed iguanas for two hours.’ (object < adv and adv < object) Mi a mata un yuana #pa dos ora largu.11 I  kill an iguana for two hour long Eu matei uma iguana #por duas horas. I killed an iguana for two hours ‘#I killed an iguana for two hours.’

P BP

P BP

P BP

P BP

While the variability in the interpretation of the bare singular cannot be used as evidence that bare singulars should be treated as names of kinds, since other indefinites may also receive different interpretations depending on the predicates they combine with, the first and third pieces of evidence present compelling evidence in favor of the analysis of bare singulars as kind-denoting expressions. The contrast between (7) and (8) constitutes evidence against the assimilation of bare nominals to indefinites, as proposed by Diesing (1992) and others, since the bare singular in this case has a reading that the indefinite singular does

.

Read Y< X as Y has scope over X.

. The symbol # is used here to indicate that the examples are not ungrammatical, but pragmatically ‘odd’ or infelicitous.

Chapter 4. Papiamentu and Brazilian Portuguese 

not have.12 If the bare singular were to be treated as a singular indefinite, then both (7) and (8) should have the same readings. The narrow reading should be either possible or impossible for both. However, (7) has a reading that is missing in (8), namely a reading in which the object takes scope under the adverb. Given that they are behaving differently, they must have different properties. In other words, an analysis of bare singulars as singular indefinites will not do justice this difference. The examples above show that bare singulars in Papiamentu and in Brazilian Portuguese seem to behave in the same way and therefore are amenable to a unified treatment. Moreover, it seems that they behave very much like bare plurals in English and could easily be analysed as names of kinds. We implement such an analysis in section 3. There are, however, differences in the distribution of bare singulars in the two languages. Bare singulars have a wider distribution in Papiamentu than they have in Brazilian Portuguese.13 They can appear in contexts of inalienable possession and also with unique and familiar objects, even if these objects have not been mentioned previously in the discourse. This is illustrated in (9a) and (10a). In Brazilian Portuguese the definite is required in both cases, as (9b) and (10b) illustrate. (9)

a. b.

(10)

.

a.

El a laba (su) kara. he  wash (his) face Ele lavou o rosto. he washed the face ‘He washed his face.’ Solo/*e solo ta kima sin miserikòrdia.14 sun/the sun  burn without mercy

P BP

P

The same point was made by Dayal (2002) for English bare plurals.

. Although bare singulars generally have a wider distribution in Papiamentu than in Brazilian Portuguese, this generalization does not hold for subjects of episodic sentences. In Papiamentu, bare singular count nouns are impossible for most speakers in the subject position of episodic sentences, whereas they are acceptable in Brazilian Portuguese with heavy focus/contrastive intonation. Focus does not seem to help the Papiamentu examples for most speakers we consulted. This is illustrated in the examples in (i). (i)

a. b.

*?Muhé ta lesando i homber ta skibiendo. Mulher tá lendo e homem tá escrevendo. woman is reading and man is writing ‘Women are reading and men are writing.’

P BP

In order to make (ia) acceptable, either the indefinite singular must be used or the noun must be pluralized by means of -nan. Although we don’t have an explanation for this fact, this pattern is not particularly surprising, being well-attested in other languages (see Stark 2002 for data from Old Italian and other references.) .

Example (10a) is partially based on Dijkhoff (1990: 16).

 Ellen-Petra Kester and Cristina Schmitt

b.

O sol/*sol está queimando sem misericórdia. The sun/sun is burning without mercy ‘The sun is burning without mercy.’

BP

The facts above raise various questions. If we treat bare singulars as names of kinds in the two languages, the question is what accounts for the differences in distribution in Papiamentu and in Brazilian Portuguese. A second question concerns the internal structure of these nominals: are they bare NPs, as Chierchia (1998) suggests for languages like Chinese, or do they have other layers of functional structure? In section 2 we show that there are important differences between the definite determiners in the two languages which may indirectly account for the differences described above, given particular assumptions about competition between items.

.

Bare plural differences

In the domain of bare plurals, a very different picture arises. Papiamentu bare plurals do not have the same interpretations nor the same distribution as bare plurals in Brazilian Portuguese. While the latter behave like bare plurals in English, bare plurals in Papiamentu have a much narrower distribution. First, while bare plurals can receive generic and existential interpretations in Brazilian Portuguese and can appear freely in both subject and object position, bare plurals in Papiamentu cannot. The generic interpretations are exemplified in (11) for subjects (a–d) and objects (e and f). Notice that the Papiamentu examples are unacceptable, unless they receive contrastive focus intonation or are modified, while the Brazilian Portuguese ones are perfect without any extra material. (11)

a. b. c. d. e. f.

*Muchanan ta inteligente.15 Crianças são inteligentes. ‘Children are intelligent.’ *Kabritunan ta mashá komun na Kòrsou. Cabritos são muito comuns em Curaçao. ‘Goats are very common in Curaçao.’ *Mi ta gusta pushinan. Eu adoro gatos. ‘I love cats.’

P BP P BP P BP

The existential interpretation is exemplified for subjects in (12). Again, the Brazilian Portuguese example in (12b) is perfectly acceptable, whereas example (12a) from Papiamentu is unacceptable.

. It should be noted that the bare plural is possible in this sentence with contrastive focus intonation.

Chapter 4. Papiamentu and Brazilian Portuguese 

(12)

a. *Tin kòmpiuternan riba mi mesa. b. Tem computadores na minha mesa. have computers on my desk ‘There are computers on my desk.’

P BP

The bare plural in Papiamentu is possible in subject position when the discourse allows it to refer to a pre-specified set. The following pair of sentences illustrates the contrast between the bare plural and the bare singular in Papiamentu.16 (13)

a.

b.

Despues ku *hende/hendenan a keha, after that person/persons  complain, nan a drecha e película. they  fixed the film ‘After some people complained, they fixed the film.’ Si hende/*hendenan ta keha, no wòri ku nan. if person/*persons pres complain, no worry with them ‘If people complain, don’t worry about them.’

P

P

Example (13a) shows that the bare plural can be used in the subject position of episodic sentences. However, in this case, the bare plural is not generic; i.e. (13a) is felicitous in a context where the speaker is reporting an episode he witnessed in the movie theatre; for instance, a case in which the projector had a problem and only after people in the audience complained, was the problem solved. Notice that in this case the bare singular is unacceptable. In (13b), on the other hand, only the bare singular is possible. Here the statement is generic. The example in (13b) is felicitous in a context where the manager of the movie theatre is explaining to the usher that he does not need to worry about complaints. The statement is generic and the bare singular is the only possibility. Using the bare plural is not an option in this case. The same sentences can be uttered in Brazilian Portuguese with a bare plural, although (14a) is somewhat odd and an overt determiner would be preferred. (14)

a.

b.

Depois que ?alunos reclamaram, after that ?students complained, eles consertaram o filme. they fixed the film ‘After students complained, they fixed the film.’ Se alunos reclamarem, não se preocupe com eles. If students complain, not  worry with them ‘If students complain, don’t worry about them.’

BP

BP

In object position bare plurals are not felicitous either in Papiamentu, unless they are modified. This is exemplified in (15) and (16) respectively. While the unmodified . We thank Alan Munn for discovering this fact.

 Ellen-Petra Kester and Cristina Schmitt

form is considered odd or rejected by most speakers, the modified form is perfectly acceptable for all speakers. The example in (17) shows that the bare plural in Brazilian Portuguese is not subject to a restriction of this kind. (15)

a.

b.

(16)

a.

b.

(17)

a.

b.

Mi ta mira *bukinan/buki riba mesa. I  see books/book on table ‘I see books on the table.’ El a bin ku *regalonan/regalo pa mi. he  come with presents/present for me ‘He came with presents for me.’

P

Mi ta mira bukinan na spañó riba mesa. I  see books in Spanish on table. ‘I see Spanish books on the table.’ El a bin ku regalonan mashá karu pa mi. he  come with presents very expensive for me ‘He came with very expensive presents for me.’

P

Eu vi livros na mesa. I saw books on-the table ‘I saw books on the table.’ Ele veio com presentes pra mim. He came with presents for me ‘He came with presents for me.’

P

P

BP

BP

The examples above show that, while the bare plural in Brazilian Portuguese behaves like the bare plural in English, the bare plural in Papiamentu does not allow kind readings and cannot appear in existential sentences, where it, of course, must have a nonpresuppositional interpretation, suggesting that a treatment of the bare plural in Papiamentu as a name of a kind is not a viable alternative. Moreover, it seems that the bare plural is only possible in Papiamentu when it can be treated as somehow anchored in the discourse. The line we are going to take is that the bare singular in both languages and the bare plural in Brazilian Portuguese denote names of kinds. The bare plural in Papiamentu, however, is not really a bare plural, but rather the combination of a plural marker and a feature that is only felicitous if interpreted in a context where a specific reading can be obtained. Before we move on to our analysis of bare nominals, we examine the definite and the indefinite determiner in both languages.

.

Definites and indefinites

The exact way to state the distinction between definites and indefinites is still a matter of controversy and our goal here is not to take sides on a particular analysis,17 but .

See Heim (1991) for a review.

Chapter 4. Papiamentu and Brazilian Portuguese 

rather to establish the descriptive similarities and differences we find between definite and indefinites in Papiamentu and Brazilian Portuguese.

.

Definites

With respect to interpretation, at first sight the definite determiners in both languages display a similar behavior.18 Below we present some data to show that definites in both languages (i) can have unique reference and are to be distinguished from the demonstrative; (ii) can be associated to an antecedent in the discourse via identity or a part-whole relation; and (iii) allow referential and attributive uses. First, the unique reference of the definite determiners is illustrated by the infelicitous examples in (18). In a context where there are two glasses on the table, one empty and one full, these sentences would be infelicitous, since the discourse situation violates uniqueness. In these cases we need a demonstrative, as illustrated in (19a,b).19 (18)

a. b.

(19)

a. b.

#E glas ta bashí i e glas ta yen. #O copo está vazio e o copo está cheio. the glass is empty and the glass is full ‘The glass is empty and the glass is full.’ E glas akí ta bashí i e glas ayá ta yen. the glass here is empty and the glass there is full Esse copo está vazio e este copo está cheio. This glass is empty and this glass is full. ‘This glass is empty and this glass is full.’

P BP

P BP

Second, in both languages the definite article can be anaphoric to an antecedent in the discourse, referring to a specific entity, mentioned previously, as in (20). (20)

a.

b.

c. d.

Mi a kumpra un bolo. I  buy a cake ‘I bought a cake.’ E bolo/*bolo a wòrdu kome den 10 minüt. The cake/cake  been eat in 10 minute ‘The cake/*cake was eaten in 10 minutes.’ Eu comprei um bolo. ‘I bought a cake.’ O bolo/*bolo recebeu muitos elogios. ‘The cake/*cake received many compliments.’

P

P

BP BP

. We are using the terms definite and indefinite in an almost pre-theoretical way. The term definite is used to mean familiar with respect to a discourse context. The term indefinite will be used as novel with respect to the discourse context. . This example is akin to Löbner’s (1985) test for consistency, which separates demonstratives from true definites.

 Ellen-Petra Kester and Cristina Schmitt

The definite article can also appear in the so-called bridging or associative contexts, as illustrated in (21) to (23), where the association to an antecedent in the discourse is not established under identity, but rather via a part-whole relation. Notice that constructions of this type require the presence of a definite article: the use of bare singulars (see (21) and (22)) or a demonstrative (consisting of the definite article e and a postnominal locative adverb like aki ‘here’ or ei ‘there’, as in (24)), is impossible in these contexts. (21) a.

b.

(22) a.

b.

(23) a.

b.

(24) a.

b.

Mi a kumpra un bolo. I  buy a cake E krema/*krema no tabata mashá bon. the frosting/*frosting not  very good Eu comprei um bolo. I bought a cake A cobertura/*cobertura não era muito boa. the frosting/*frosting not was very good ‘I bought a cake. The frosting was not very good.’ Mi ta dunando lès di un estudio independiente. I am giving lesson of a study independent E studiante/*studiante ta hopi bobo. the student/*student is very dumb Eu estou dando aulas particulares. I am giving lessons private O aluno/*aluno é muito burro. the student/*student is very dumb ‘I am teaching an independent study. The student is very dumb.’ Mi a drenta den e outo i mi a start I  enter in the car and I  start e motor. the engine Eu entrei no carro e liguei o motor. I entered in-the car and turned-on the engine ‘I got into the car and turned on the engine.’ #Mi a drenta den e outo i mi a start I  enter in the car and I  start e motor akí/ei. the engine here/there #Eu entrei no carro e liguei esse motor. I entered in-the car and turned-on this engine ‘I got into the car and turned on this/that engine.’

P

BP

P

BP

P BP

P

BP

A third important and expected property of the definites in both languages is that they allow both attributive and referential readings, behaving as expected in

Chapter 4. Papiamentu and Brazilian Portuguese 

intensional contexts. In (25) both a reading in which we know who the dean is, as well as a reading in which I am looking for the dean whoever he is, are possible. (25)

a. b.

Mi ta buskando e dekano/*dekano. Estou procurando o decano/*decano. ‘I am looking for the dean/*dean.’

P BP

The data presented so far show that the definite determiner in both languages is to be distinguished from the demonstrative (which requires the presence of a post-nominal adverb and is used in contexts where the referent is not unique), can be licensed in anaphoric contexts (including associative ones) and has referential and attributive uses. There are, however, important differences between the definite articles in the two languages under consideration. First, while the definite determiner followed by an NP can denote a kind in Brazilian Portuguese (as well as in Spanish, Dutch and English), as illustrated in (26b–e), the definite in Papiamentu does not allow kind readings, as the unacceptability of (26a) illustrates. (26)

a. b. c. d. e.

#E bayena ta un mamífero. A baleia é um animal mamífero. La ballena es un mamífero. De walvis is een zoogdier. The whale is a mammal.

P BP Spanish Dutch English

In Papiamentu the kind reading can only be expressed by a bare nominal, an option also available in Brazilian Portuguese, as shown in (27). (27)

a. b.

Bayena ta un mamífero. whale is a mammal Baleia é mamífero. whale is mammal ‘Whales are mammals.’

P BP

The definite in Papiamentu can never be used in generic contexts. Thus, while the (a) examples in (28) and (29) from Brazilian Portuguese can have generic readings, the examples from Papiamentu in (b) must be associated to non-generic entities previously anchored in the discourse in a way to be made precise by future research (see section 3 for a suggestion). In Papiamentu, to obtain a kind reading, the bare singular must be used, as illustrated in the (c) examples. Notice that a bare singular can also be used in Brazilian Portuguese as the (d) examples illustrate. The example in (30) shows that this generalization also holds for mass nouns. (28)

a. b.

O leão vive na África. the lion lives in-the Africa #E leon ta biba na Afrika. the lion  live in Africa ‘The lion lives in Africa.’

BP P

 Ellen-Petra Kester and Cristina Schmitt

c. d.

(29)

a. b. c. d.

(30)

a. b. c. d.

Leon ta biba na Afrika. lion  live in Africa Leão vive na África. lion live in-the Africa ‘Lions live in Africa.’

P BP

O cabrito é muito comum em Curaçao. #E kabritu ta mashá komun na Kòrsou. ‘The goat is very common in Curaçao.’ Kabritu ta mashá komun na Kòrsou. Cabrito é muito comum em Curaçao. ‘Goats are very common in Curaçao.’

BP P

?O leite é saudável. *E lechi ta sano. ‘The milk is healthy.’ Leite é saudável. Lechi ta sano. ‘Milk is healthy.’

BP P

P BP

BP P

The different distribution of the definite article in the two languages is made even clearer when we examine examples (31) to (33), in a context where they cannot have been previously mentioned. (31)

a. b.

(32)

a. b.

(33)

a. b.

Ninguém sabe quem inventou a roda/*roda. nobody knows who invented the wheel/wheel Niun hende no sa ken a inventa *e wil/wil. no person not knows who  invented the wheel/wheel ‘Nobody knows who invented the wheel/*wheel.’

BP

O sol/*Sol está queimando sem misericórdia. the sun/sun is burning without mercy #E solo/Solo ta kima sin miserikòrdia. the sun/sun is burning without mercy ‘The sun is burning without mercy.’

BP

A água/*água está fria. #E awa/awa ta friu. ‘The water/water is cold.’

BP P

P

P

Cases such as (31) require a definite DP both in Brazilian Portuguese and in English, but not in Papiamentu. If the same motivation for the definite in (31) can be extended to (32) and (33)20, then it is not surprising that, while a definite is required in Brazilian Portuguese, the bare noun is the only option in Papiamentu. We will argue below that . It should be noted that the definite in examples (32) and (33) is acceptable if the sun and the water have been mentioned previously in the discourse.

Chapter 4. Papiamentu and Brazilian Portuguese 

this is indeed the case: the bare nouns in (31) to (33) are kind-denoting expressions; the example in (32) involves a kind consisting of one unique member. The second difference has to do with the behavior of the definite in constructions of inalienable possession, as we have already seen in (9), repeated in (34): (34)

a. b.

El a laba (su) kara. he  wash his face Ele lavou o rosto. he washed the face ‘He washed his face.’

P BP

While the definite determiner must be used in constructions of inalienable possession in Brazilian Portuguese in unmarked readings, Papiamentu uses the possessive determiner or the bare nominal; the definite induces an alienable reading of a previously mentioned body part. A third important difference is that, although Brazilian Portuguese definites can precede proper names, this is impossible in Papiamentu. (35)

a. b.

(*E) Maria tabata malu. (A) Maria estava doente. (the) Maria was sick ‘Maria was sick.’

P BP

It is important to note that in all of the three cases above, involving kind readings (26–33), constructions of inalienable possession (34) and proper names (35), the definite determiner is not being used to refer to a previously mentioned entity in the discourse. In fact, these uses of the definite have been named expletive uses (Vergnaud and Zubizarreta 1992; Longobardi 1994 among others), and we use the term here merely to group these cases as non-referential uses of the definite.21,22 Whatever analysis we want to give to the so-called expletive determiners, the point here is just that while the definite in Brazilian Portuguese allows the

. In Vergnaud and Zubizarreta (1992) expletive determiners are to be correlated with kind/ type readings and non-expletive determiners with token readings: both definite kinds and definite DPs in constructions of inalienable possession are to be treated as types. It should be noted that we are not following their analysis in its details, just in its major claim. A full analysis of constructions of inalienable possession is outside the scope of this paper. . As we will see below, the so-called expletive determiner plays a very clear role in the distribution and interpretation of kind denoting elements and therefore the term expletive is a bit unfortunate. We will continue to use it here with the sense of non-referential only. See Guéron (2003) for arguments against the idea that the determiner is to be treated as an expletive in constructions of inalienable possession. Her proposal is that the definite should be treated as a classifier.

 Ellen-Petra Kester and Cristina Schmitt

so-called expletive uses, the definite in Papiamentu does not. If, as Vergnaud and Zubizarreta argue, the lack of expletive determiners in constructions of inalienable possession is to be correlated with the absence of definite kinds, then it is not surprising that the only option in Papiamentu for generic/kind/inalienable possession cases is the bare singular.23

.

Indefinites

In this section we will restrict ourselves to the singular indefinite and ignore the Brazilian Portuguese plural indefinite. Here, interestingly, we will not find apparent differences between the two languages. In both Papiamentu and Brazilian Portuguese, the indefinite singular can have wide and narrow scope, can be used in generic constructions and can be used to disambiguate between singular and plural readings, as illustrated below. First, as is well known, indefinites enter into scope interactions, such as (36) and (37). As these cases exemplify, Papiamentu and Brazilian Portuguese allow wide and narrow scope readings of the indefinite singular. (36)

a. b.

(37)

a. b.

(∃24 > ¬ and ¬ > ∃) Mi no a mira un mancha riba suela. I not  see a spot on floor Eu não vi uma mancha no chão. I not see a spot on-the floor ‘I didn’t see a spot on the floor.’ (want > ∃ and ∃ > want) Maria kier kasa ku un brasileño. Maria quer casar com um brasileiro. Maria wants marry with a Brazilian ‘Maria wants to marry a Brazilian.’

P BP

P BP

Second, the indefinites in these languages can also have a generic interpretation, but it seems that a distinction should be made between quantificational generic sentences and kind-denoting predicates. While the indefinite can appear with quantificational generics, this is not the case for predicates that can only be true of a kind, which are odd, unless we give them a taxonomic interpretation. This is illustrated in (38) and (39). The fact that the indefinite is acceptable in generic contexts shows that the indefinite must be separated from the homophonous numeral one, which does not allow the generic reading. . Guéron (2003) argues that the ‘non-referentiality’ of definite determiners correlates with the ability of the determiner to carry merely formal features (agreement features). The Papiamentu determiner, being invariable, would not be able to carry merely formal features and is always referential. This analysis leaves the definite generics in English unexplained. . ∃ represents here schematically the indefinite existential force.

Chapter 4. Papiamentu and Brazilian Portuguese 

(38)

a. b.

(39)

Un hulandes ta papia hulandes. a Dutchman  speak Dutch Um holandês fala holandês. a Dutchman speaks Dutch ‘A Dutchman speaks Dutch.’

P BP

a. #Un kabritu ta komún na Kòrsou. b. #Um cabrito é comum em Curaçao. ‘A goat is common in Curaçao.’

P BP

A third important feature of the singular indefinite is that it is often used to disambiguate between a clear singular reading and a non-singular reading, as illustrated in the two possible answers to the question in (40). While the bare singular is unspecified, the indefinite singular is unambiguous. (40)

Bo tin yu? Si, mi tin yu/un yu. You have child? Yes, I have child/a child ‘Do you have children? Yes, I have children-a child/a child.’

P

In sum, the indefinites in both languages seem to have a similar distribution and interpretation, which suggests a uniform treatment.

.

Summary

A few generalizations emerge from this discussion of the definite and indefinite determiners in both languages: first, the definite determiner in Papiamentu has a much more restricted usage than its counterpart in Brazilian Portuguese. In other words, the definite in Papiamentu must inherit a referential index from the discourse. It is therefore excluded from cases where the definite has a different function, the so-called expletive function. In such cases the only option is the bare singular, since the bare plural requires familiarity or some version of discourse linking. Second, indefinites behave in a similar manner in both languages. Table 2 summarizes the main contrasts between the two languages with respect to the distribution of bare nominals, as well as the definite and indefinite determiners.

Table 2. The distribution of bare, definite and indefinite nominals in Papiamentu and Brazilian Portuguese Nominal interpretation

Papiamentu

Brazilian Portuguese

true kind predications characterizing predicates

bare singular bare singular/indefinite

familiar and unique not previously mentioned inalienable possession

bare singular

bare nominals/definites bare nominals/definites/ indefinite singular definite

bare singular/possessive

definite/?? possessive

 Ellen-Petra Kester and Cristina Schmitt

.

The syntax and semantics of bare singulars

In this section we provide an account for the bare singulars in both languages. We propose that bare singulars in Papiamentu are syntactically to be treated as DPs with no number projection as argued by Schmitt and Munn (1999; 2003) and Munn and Schmitt (2001) for Brazilian Portuguese bare singulars. Semantically, based on the properties described in section 1, it seems very reasonable to treat bare singulars in both languages as names of kinds. The differences in distribution between Papiamentu and Brazilian Portuguese will be attributed to the differences between the definite determiners of both languages and not to the semantics of the bare nominals. This section is divided as follows: first we lay out our assumptions about the syntax and semantics of DPs in general. Then we present our proposal, the supporting evidence and the syntactic and semantic consequences of such an analysis.

.

Basic assumptions about the syntax and semantics of DPs

We depart from the general assumption that functional heads, in order to be projected in the syntax, must have morpho-syntactic and/or semantic import. Hence, the number and nature of functional projections depends entirely on morpho-syntactic and/or semantic properties. This assumption does not entail that there is a one to one correspondence between functional heads and morphological features. For example, a head may be fused and contain more than one feature (Bobaljik 1995, Marantz 1997). Likewise, there is no strict mapping between morphemes and features or functional heads. The same morpheme may carry features that must be associated to more than one functional head. In this paper, we will only consider a Number projection between DP and NP, but we do not exclude the presence of other functional heads on the path from NP to DP. On a semantic level, for simplicity we assume that the NP projection is interpreted as a predicate, inherently sorted as count or mass (contra Chierchia 1998). Determiners combine with predicates (NPs), deriving a DP whose semantics corresponds to an entity or a generalized quantifier. In other words, we assume that DPs, in the unmarked case, are to be treated semantically as individuals or generalized quantifiers. Since kinds correspond to individuals, they should be analysed as DPs, not NPs, under the assumptions made so far. Number Phrases are modifiers of NP in the semantic level. They host plural and singular morphology, as well as cardinals and other quantifying elements. We assume that cardinals can occupy two different positions within the NumP: they can be heads or specifiers. A final important assumption is that only DPs can introduce independent and transparent discourse referents.25 We assume that nominals that can appear as the

. Transparent discourse referents are discourse referents that can serve as antecedents for pronouns.

Chapter 4. Papiamentu and Brazilian Portuguese 

subject of an individual level predicate correspond to a DP. This assumption is in accordance with the Ladusaw (1994)’s proposal that individual-level predicates require their subjects to be independently identified entities, because sentences with individual-level predicates are interpreted as categorical statements about independently established referents. With these assumptions in place, we proceed to make a concrete proposal for the syntax and semantics of bare singulars.

.

The analysis of bare singulars

We follow the analysis proposed for bare nominals in Brazilian Portuguese in Schmitt & Munn (1999, 2003) and Munn & Schmitt (2001) who show that Brazilian Portuguese does not fit Chierchia’s (1998) language typology based on the Nominal Mapping Parameter. According to Chierchia, bare nominals are only found in argument position in languages like Chinese, which have a generalized classifier system and do not have plural morphology. As shown in Schmitt & Munn (1999, 2003) and Munn & Schmitt (2001), Brazilian Portuguese is problematic for this typology because it allows bare singulars in argument position, although it has regular plural morphology and it lacks a generalized classifier system. These observations also hold for Papiamentu, as we have shown in the previous sections. There is no generalized classifier system and there is productive plural morphology, at least with definites (see below). According to Schmitt & Munn (1999, 2003) and Munn & Schmitt (2001) bare singulars in Brazilian Portuguese correspond to DPs without a Number projection, a proposal we will also adopt for Papiamentu. Thus, the schematic structure of bare singulars in both languages is given in (41).26 DP

(41)

D

NP

Empty D has a semantic role in this structure: the NP predicate is shifted into a kind via an operation performed by an element in D. For concreteness, we can adopt Chierchia’s neo-Carlsonian view of kinds, except that the operation that creates kinds is a property of D elements and not a covert type shifting operation on NPs. Chierchia’s intuition is that a kind can be manufactured out of a property by taking the largest member of its extension (any given situation). In other words, a kind can be identified in any situation with the totality of its instances. A dog kind, for example, can be identified with the totality of dogs or the fusion of all dogs around. Thus a kind can be modeled as functions from worlds (or situations) into the sum of all its instances. Assuming that common nouns are of type 〈s,〈e,t〉〉, i.e., denote properties, the

. We will make a further refinement of this structure in section 4.

 Ellen-Petra Kester and Cristina Schmitt

nominalization function shifts the nominal into an individual 〈s,e〉 via a nominalization operation (nom), defined as in (42a). This function Nom is a function from properties to functions from situations to the maximal entity that satisfies that property in that situation. To go from kinds to predicates, a predicative function picks instances out of a kind. In other words, pred is the function that takes kinds and return their instantiation sets, as given in (42b). (42)

a. b. c.

nom∩: λs iλ [Ps (x)]27 pred∪: λx [x ⱕ ks] Derived Kind Predication (DKP): if P applies to objects and k denotes a kind then P(k) = ∃x [∪k(x) ^ P(x)]

If kinds are the totality of its instances, then the function that creates kinds cannot apply to singulars, which is not to say that a DP kind cannot refer to a kind that we happen to know has only one member. Existential readings involve the introduction of existential quantification over the instantiations of the kind in a given situation. We would like to suggest here that while the nom∩ function is encoded in the empty determiner, the pred∪ function, on the other hand, is a covert type shifting operation on the DP. The idea is that covert type shifting operations are a last resort operation. If DPs are not canonically mapped into predicates, their interpretation as predicates will be the result of a last resort operation to fix a mismatch.28 According to Dayal (2002), Derived Kind Predication as a means to obtain object readings of kinds has the advantage of predicting that the existential introduced by DKP will always take narrower scope than other operators since the kind term is a direct argument of the predicate. This is illustrated below for the Brazilian Portuguese ‘Mary didn’t buy computer’. (43)

Maria não comprou computador. Maria not bought computer ¬buy (m, ∩computer) DKP⇒¬∃x [∩∪computer (x) buy(x)]

BP

. s stands for situations, k stands for kinds and i stands for the iota operator. The iota operator can be used to interpret the definite article. When it applies to a set of pluralities, it will refer to the largest plurality in that extension. If it applies to a set of singularities, it also seeks the largest one, which yields a result only when the predicate has just one object in its extension. . Of course one could ask why can’t it be the other way around; i.e. why can’t the covert type shifting operation create a kind from a predicate DP (on the assumption perhaps that DPs with lexically empty Ds are predicates). At least for Brazilian Portuguese, one could argue that this cannot be the case since the definite determiner would block this covert operation, since the definite can also create kinds. For Papiamentu we cannot resort to this line of explanation since the definite determiner cannot create kinds. Given the similarities between both languages, we will assume that this is not a possible option.

Chapter 4. Papiamentu and Brazilian Portuguese 

The prediction that the existential always takes scope under other operators is interesting because in fact the bare singulars in both languages only allow narrow scope readings. In other words, the treatment of bare singulars as kinds accounts for their scopal properties and for their ability to appear in kind-level predications. The proposal that bare singulars do not have number projections and are indeed DPs makes very clear predictions. First, the bare singular count nominal should be able to be interpreted as singular or plural, but not as mass, given our assumption that NPs are sorted as count or mass. Second, being a DP, the bare singular should be able to appear as the subject of individual-level predicates. In the next section we will show that both of these predictions are borne out in both languages.

.

Lack of NumP and presence of a D projection: empirical support

.. Plural and singular interpretations In this section we show that the bare singular in Brazilian Portuguese and Papiamentu is unmarked for number. We first examine data that show that the bare singular allows atomic and non-atomic readings. As we have already seen in (7), repeated here as (44), in the context of a time-span aspectual adverbial, the bare singular behaves as a plural. Use of the singular indefinite is infelicitous, because the indefinite cannot take scope under the adverb. (44)

a. b.

Mi a mata yuana/#un yuana pa dos ora largu. I  kill iguana/#an iguana for two hour long Eu matei iguana/#uma iguana por duas horas. I killed iguana/#an iguana for two hours ‘I killed iguanas/#an iguana for two hours.’

P BP

Another case in which the bare singular behaves as a plural is when it is in the object position of predicates that impose a non-atomicity entailment on its internal argument, as is the case of verbs such as collect. (45)

a. b.

Pedro ta kolekshoná stampía. Pedro  collect stam Pedro coleciona selo. Pedro collects stamp ‘Pedro collects stamps.’

P BP

Notice, however, that singular readings for the bare singular are also possible. The verb to marry (at least in monogamous societies) imposes an atomicity entailment on its internal argument (in its most usual reading). The acceptability of (46) illustrates that an atomic interpretation is also possible. (46)

a. b.

Maria kier kasa ku brasileño. Maria quer casar com brasileiro. Maria wants marry with Brazilian ‘Maria wants to marry a Brazilian.’ (any Brazilian)

P BP

 Ellen-Petra Kester and Cristina Schmitt

A second piece of evidence for the unmarkedness of bare singulars for number comes from discourse anaphora. The bare singular in both languages can be resumed with either a plural or a singular pronoun, as illustrated in (47). (47)

a. b.

Mi tin yu. Bo ke topa nan/ top ’e? I have child. You want to-meet them/ meet him Eu tenho filho. Você quer conhecer eles/ conhecer ele? I have child. You want meet them/ meet him ‘I have children/a child. Do you want to meet them/meet him?’

P BP

The data above shows that bare singulars allow atomic and non-atomic readings and can be resumed by a singular or plural pronoun.29–30

.. No mass readings for bare singular count nouns We now show that the bare singular cannot be treated as a simple mass term,31 since it can be the argument of a predicate that requires individuation. Notice that mass nouns in general cannot be the subject of predicates that require individuation, as illustrated in (48). (48)

*Gold weighs two grams.

The examples below from Papiamentu and Brazilian Portuguese show no such restriction on bare singular count nouns. (49)

a. b.

Mucha ta pesa 20 kilo n’e edad aki. child  weigh 20 kilo in the age here Criança pesa 20 kilos nesta idade. child weighs 20 kilos in-this age ‘Children weigh 20 kilos at this age.’

P BP

. It should be noted that the preference in most contexts is for the anaphoric pronominal element to be plural in both languages, which may have to do with the information load of the bare singular. . The fact that Papiamentu also allows the bare singular to introduce a kind that has only one member such as the sun and the moon can potentially also be used to argue for the ability of the bare singular to be associated with atomic interpretations. . Obviously, mass terms can surface as bare arguments in Papiamentu, with an existential reading as in (33b) or a generic reading, as in (i):

(i)

Oro ta karu. ‘Gold is expensive’.

The issue at stake in this section, however, concerns the analysis of bare singular count nouns, which cannot be analysed as mass terms, because they are compatible with predicates requiring individuation, as in (49) and (50).

Chapter 4. Papiamentu and Brazilian Portuguese 

(50)

a. b.

Mucha ta bringa ku otro n’e edad child  fight with other in the age Criança briga uma com a outra nesta child fight one with the other in-this ‘Children fight with each other at this age.’

aki. here idade. age

P BP

The lack of Number has further consequences both semantically and syntactically. One consequence is that bare singulars will trigger default third person agreement in languages that have agreement. We can verify the agreement in Brazilian Portuguese but not in Papiamentu since the latter has no overt agreement of any sort. The second important consequence of lack of number is that the Nom function encoded in D (see (42) above) can take the morphologically singular predicate and shift it to a kind reading. This now follows from the fact that, although kinds cannot be semantically singular (see Chierchia 1998), bare singulars are not semantically singular in the languages under consideration.

.. Arguments for D The argument for a DP projection is more subtle and relies on a couple of assumptions we have made. The first assumption is that bare NPs are predicates, and the second assumption is that only Ds can introduce a discourse referent that is transparent (i.e., can be the antecedent to a pronoun). We also assume that only DPs can be subjects of individual-level predicates, as in (49) above, for example. As we have already seen, both assumptions are relevant for the analysis of bare singulars. First, bare singulars can be antecedents of pronominals, as in (47), and, second, they can be subjects of individual level predicates, as in (49), which forces them to be DPs, under our assumption that individual level predicates make categorical statements about independently established referents. Additional evidence for the analysis of bare singulars as DPs comes from their properties in coordinate structures. It is well-known that when two NPs are coordinated under the same determiner head, the only possible reading is one in which they must be associated with the same individual32 in Brazilian Portuguese and Papiamentu in (51a): (51) a. a’.

.

one single individual Mi a topa ku e I  meet with the Eu encontrei o amigo ‘I met the friend

amigu i kolega na Kòrsou. friend and colleague in Curaçao e colega em Curaçao. and colleague in Curaçao.’

In English this is not always the case: see Heycock and Zamparelli (2003).

P BP

 Ellen-Petra Kester and Cristina Schmitt

b. b’.

more than one individual Mi a topa ku amigu i kolega na Kòrsou. I  meet with friend and colleague in Curaçao Eu encontrei amigo e colega em Curaçao. I met friend and colleague in Curaçao. ‘I met friends and colleagues in Curaçao.’

P BP

Conjoined bare singulars, however, allow a reading in which each conjunct refers to potentially different sets of individuals, which is consistent with the possibility that we are coordinating two independent DPs in (51b).33

.

Differences in distribution

At this point, we have provided a syntax and a semantics for the general properties of bare singulars in both languages by assuming that they are DPs with no Number projection, which denote names of kinds. However, as we have seen, there are also some differences in distribution between bare singulars in both languages that should be accounted for: the acceptability of the bare singular in Papiamentu and the obligatoriness of the definite article in Brazilian Portuguese in two cases that we have subsumed as expressions involving the use of an expletive definite determiner: (i) cases such as the wheel or the sun where the kind has one unique member, and (ii) constructions of inalienable possession, which are arguably another instance of kind readings (Vergnaud and Zubizarreta 1992). There are at least two options: we can allow for a small difference between the two languages in terms of the definiteness of empty D, keeping constant the idea that it is to be treated as a kind with no number, or we can try to explain the distribution differences by differences in the definite articles found in both languages. In the following, we examine each of these possibilities in turn. As we have seen before, unlike Brazilian Portuguese bare singulars, Papiamentu bare singulars seem to be acceptable in contexts where the definite article is obligatory in Brazilian Portuguese. This generalization holds for constructions of inalienable possession (34), proper names (35) and certain kind-denoting expressions (31–32). One way to account for these differences would be to treat the bare singular in Papiamentu as systematically ambiguous between a definite and an indefinite kind but treat the Brazilian Portuguese as always indefinite. Since the bare singulars in both languages are acceptable in presentational contexts, it is plausible to assume they are/can be indefinite, indefinite understood as novel with respect to a discourse context. However, it is unlikely that Papiamentu bare singulars are ambiguous between a definite and indefinite interpretation, since the bare singular in Papiamentu does not have the properties that definite kinds have in languages that have both definite and . The predicative reading, which refers to only one set of individuals, is marginal or impossible and that may be associated with the lack of number.

Chapter 4. Papiamentu and Brazilian Portuguese 

indefinite kinds, as is the case of Brazilian Portuguese. In Brazilian Portuguese, the definite singular kind can appear as the subject of stage-level kind predicates, but the bare singular kind cannot, as illustrated below. (52)

a. b. c. d.

O dinossauro/*dinossauro está extinto. the dinosaur/dinosaur is extinct *Dinosaurus ta extingí. dinosaur is extinct Dinoussauro é (um) animal extinto. dinosaur is a animal extinct Dinosaurus ta un sorto/ un bestia extingí. dinosaur is a species/ an animal extinct ‘The dinosaur is an extinct species.’

BP P BP P

In order to say something like (52a) in Papiamentu, which is formed with the copula verb estar usually associated with stage-level predications, it is necessary to change the predicate whose subject is a bare singular into an individual-level predicate, as illustrated in (52c,d). In these cases, the predicate is a nominal projection and does not correspond to the stage-level predicate extingi. If the bare singular in Papiamentu could be interpreted as a definite kind, we would expect (52b) to be acceptable, but it is not. We therefore reject the ambiguity hypothesis. We consider now the hypothesis that the differences in distribution between bare singulars are to be associated to the differences between the definite articles in both languages. It is also clear from the examples above that the definite kind and the bare singular are not always interchangeable. Moreover, it seems that somehow the definite, in spite of being called an expletive, is not totally devoid of content. If that is the case, then it is not unexpected that in some cases the definite kind would win a competition with the less marked version, namely the bare singular.34 The differences in the bare singular distribution between the two languages would therefore be accounted for in terms of a competition between items in Brazilian Portuguese and no competition in Papiamentu. The idea of competition between items is not new and can assume many forms. In terms of lexical insertion, Halle & Marantz (1993) suggest there is an extrinsic ordering that dictates the winner. Noyer (1997), on the other hand, has proposed that the Vocabulary Item that uniquely has the highest feature in the hierarchy is inserted. Another version of competition is associated to Heim’s (1991) maximize presupposition principle, in which speakers choose the most informative option. Note that the two versions of competition are very different. In one case we are dealing with possibilities for insertion. In the second case, with speaker’s preferences. . We leave the precise characterization of the properties of definite kinds versus bare singular kinds in Brazilian Portuguese and the precise nature of the competition between the items for further research.

 Ellen-Petra Kester and Cristina Schmitt

In Chierchia (1998), the competition between items assumes a version in which the lexicalized option (if it exists) is to be preferred from a non lexicalized version; in other words, the definite will block wherever possible the bare version of the NP. This proposal is meant to explain the lack of bare plural kinds in Italian and the lack of definite plurals in English. Since in our analyses the nom∩ function is performed by the empty D, and therefore can be considered to be part of the inventory of functional elements, we need to say that the definite determiner has more features than the empty D, a not unreasonable solution, which both in Noyer’s version and in Heim’s version would win whenever the conditions are met. If the definite in both Papiamentu and Brazilian Portuguese were semantically identical, we would expect the same distribution of bare singulars and definite singulars. However, the definites are clearly distinct in that the Papiamentu imposes very clear restrictions that are incompatible with kind readings. This fact alone allows for the bare singular to show up any time the definite is impossible. In Brazilian Portuguese, since the definite is a lot less restricted, under either Noyer or Heim’s proposal it will be preferred.35 Summarizing, we have argued that bare singulars in both Brazilian Portuguese and Papiamentu should be treated syntactically as DPs with no Number projection and semantically as names of kinds.36–37

.

Bare plurals

As we have seen before, bare plurals in Papiamentu differ from their counterparts in English and Brazilian Portuguese, because they do not allow kind readings (11a, c, e) and they cannot occur in existential sentences (12a). The analysis we will propose here treats bare plurals in Brazilian Portuguese on a par with bare plurals in English and with bare singulars in both languages, whereas bare plurals in Papiamentu are analysed as disguised bare plurals, in the sense that the plural morpheme carries . Only further research on the feature composition of the definite determiner and the kinds of tests we can use to determine at what level there is a competition between items, will be able to determine the nature of the competition between definite kinds and bare kinds in Brazilian Portuguese. . The differences between Brazilian Portuguese/Papiamentu and English and Spanish can be accounted for in a similar vein as in Schmitt and Munn (1999, 2003). Roughly the idea is that English does not have the possibility of having DPs without Num because, for morphological reasons, number is always projected with the nominal. As for Spanish, the idea is that this is not a language that allows a free empty D with the ability to create kinds at all. Therefore kind readings with bare nominals do not arise. . Needless to say, we are assuming that empty D is not subject to government conditions in these languages, as has been proposed for Spanish by Contreras (1986); Longobardi (1994) and Chierchia (1998).

Chapter 4. Papiamentu and Brazilian Portuguese 

more than the plural feature. We argue that the plural marker in Papiamentu not only has number features but also has a D feature, while the plural marker in Brazilian Portuguese is purely a modifier of NP with no D properties. Before we present our proposal and the evidence in support of it, we describe the most important morphological differences between the plural marker -nan in Papiamentu and the plural marker -s in Brazilian Portuguese.38

.

Other important properties of plural morphology in both languages

Unlike the plural morphology in Romance languages, which appears with the determiner, the noun and its modifiers (53a), in Papiamentu there is only one overt marking of plural in the DP (53b). If -nan appears more than once, the result is unacceptable (53c). Thus while Romance -s can be a mark of agreement, Papiamentu -nan cannot. (53)

a.

b.

c.

as casas grandes the- house- big- ‘the big houses’ e kasnan grandi the house- big ‘the big houses’ *enan kasnan grandinan39 the- house- big-

BP

P

P

. It should be noted that the description of -nan in nominals will be somewhat simplified. -nan sometimes appears with postnominal adjectives, as in (i); attaches to the right of apparently left-headed compounds, as in (ii); and in so-called comitative constructions, illustrated in (iii). (i)

(ii)

(iii)

e stul kibránan the chair broken- ‘the broken chairs’ e buki di skolnan the book of school- ‘the schoolbooks’ su tanta ku tionan his aunt with uncle- ‘his aunts and uncles’

(Dijkhoff 1983:225)

(Muller 1989: 542)

(Dijkhoff 1983:223)

Furthermore, -nan also appears in associative constructions, and its use is generally required when the noun is modified, as we will see below. For a full discussion, see Kester (in progress) and also Schmitt & Kester (2005). . There is a construction in Papiamentu which consists of the definite article combined with -nan, as observed in Dijkhoff (1983). In these cases we find the form es, which presumably should be analysed as an allomorph of the definite article e. However, the derived form esnan is found in elliptical constructions only, such as esnan grandi ‘the big ones’, but it never combines with a full noun (see Dijkhoff 1983 for similar examples).

 Ellen-Petra Kester and Cristina Schmitt

(54)

a.

b.

c.

e dos kasnan the two house- ‘the two houses’ dos kas(*-nan) two house(*-pl) ‘two houses’ hopi kas(*-nan) many house(*-pl) ‘many houses’

P

P

P

In Brazilian Portuguese, at least in the educated dialects, the plural appears also on the noun in cases such as (54). If plurality is a feature of NumP, then it is plausible to analyze the plural features that are semantically interpretable in both languages under consideration as generated in Num, as the default assumption. We account for the incompatibility of the plural marker -nan and cardinals (on a par with other quantifiers) as follows.

.

The syntactic behavior of -nan

We would like to suggest that the plural marker -nan is always generated as the head of Num. Cardinals, on the other hand, can occupy two different positions in the Number projection: as heads of NumP or as specifiers, as illustrated in (55a) and (55b), respectively. In the DP is indefinite, we can hypothesize that cardinals may be generated as the head of Num, but must combine with D, since they also have a D feature.40 Together D+Num introduce a referential index in the discourse. This is illustrated in (55a) for example (54b). (55)

a.

DP D

b.

NumP

[-def] dos

DP D

NP kas

e

NumP dos

Num’ Num

NP

kasj+nan

tj

The incompatibility of -nan and a cardinal or a quantifier can be explained if both elements are competing for the same position, namely the head of NumP, as in (56), which illustrates an impossible situation.

. We will modify this proposal slightly below.

Chapter 4. Papiamentu and Brazilian Portuguese 

(56)

*

DP

D

NumP

{dos/nan} NP kas

When a definite determiner is present, the cardinal functions as a modifier and occupies the specifier position of NumP (on a par with other modifiers, like adjectives, that occupy the specifier positions of other functional projections on the path from N to D, according to Cinque 1993). In this scenario, the head of NumP contains -nan, which, as an affix, forces the realization of N+nan, kasnan ‘houses’ for example. This is illustrated in (55b) for example (54a), e dos kasnan. Below we make this intuition a bit more precise for the syntax of number in Papiamentu, implementing an analysis in terms of feature checking via Agree (Chomsky 2001, Frampton et al. 2000) and making use of some features of Distributed Morphology. We assume a system of interpretable and uninterpretable features. Uninterpretable features on a head enter the derivation unvalued and at the end of a phase must be deleted from the syntactic representation (see Frampton et al. 2000). The Agree operation provides the means to give a value to unvalued features under certain conditions, one of them the Minimal Link Condition. At the end of a phase, the features marked as having entered an Agree relation will be deleted. Let’s suppose that the definite determiner e is a head with a valued feature for definiteness and an unvalued feature for number.41 -Nan, on the other hand, has an interpretable number feature but an unvalued D feature. Schematically we then have the elements and features exemplified in (57), corresponding to the structure in (58). (57)

e

-nan

kas

the

-s

house

[ ]Num-F [plural]Num-F [def]D-F [ ]D-F42

(58)

DP D

NumP

[def]D-F Num NP [ ]Num-F | [plural]Num-F [ ]D-F

. We will ignore the feature CASE here for the purposes of this discussion since we are only interested in the DP internal Agree operations. . We will refine the D properties of -nan below.

 Ellen-Petra Kester and Cristina Schmitt

Under the assumption that Agree only obtains between elements that have active features, here both D and Num have active features (i.e., unvalued features) and therefore can enter an Agree relation. This is the case of a plural definite, with or without a cardinal (see below). In Brazilian Portuguese the plural morpheme has no D-features. Therefore a different relation will obtain in order for the definite determiner to have a copy of the features in Num (see Embick and Noyer (2001) for potential alternatives.) Now consider the case of indefinites, i.e., DPs with no definite determiner. Recall that the plural marker cannot co-occur with cardinals and quantifiers. We would like to suggest that indefinites and quantifiers are D/Num heads, rather than D heads that must agree in indefiniteness with an independent D head. The fused D/Num head has interpretable D and Num features, as in (59). In this case an Agree configuration does not obtain between the two, since the D/Num has only interpretable features and is actually a single head: singular or plural and indefinite. From this proposal we derive two facts: (i) the impossibility of a definite determiner in these cases, since the D position is already filled with an indefinite and (ii) the unacceptability of -nan with indefinite quantifiers, since Num is filled by the numeral/quantifier. In other words, there cannot be *e un kas ‘the a house’ or *dos kas-nan ‘two house-s’. This is schematically illustrated in (59). (59)

DP/NumP D/Num

NP

[sing/plural]Num [indefinite] D

We have seen in (54a) that a numeral can co-occur with a plural in a definite context. In order for a numeral to co-occur with a determiner, the numeral cannot be a head, but rather an XP. This XP will modify Num and must agree with it. The definite is therefore free to occupy the D position. Note that in fact if nan is the head of NumP, then the only determiner that can appear in D is the definite e. Agree will then obtain between -nan and e, under the assumption that -nan is closer to D than the cardinal, because it is the head of the projection. This case is illustrated in (60a), a definite plural, corresponding to the structure in (60b). (60)

a.

e [def]D-F [ ]Num-F

b.

dos nan kas [plural]Num-F [plural]Num-F [ ]D-F DP

D [ ]D-F [ ]Num-F

NumP dos

Num’

[plural]Num-F [ ]D-F

NP

Chapter 4. Papiamentu and Brazilian Portuguese 

Now that we have spelled out a way to deal with acceptability of the plural with a definite and its unacceptability with indefinite determiners, we need to verify how this proposal extends to both bare plurals and bare singulars. Our proposal will require a small revision of our schematic structures above. We need to assume that the D feature of -nan imposes a restriction on the value of the D-feature, such that it must be compatible with -nan’s familiarity requirement. As we know, definites will satisfy this requirement trivially (see Enç 1991). We could assume that when an empty D combines with -nan, it will have to be interpreted as backgrounded. Such a proposal poses two problems for the whole system: first we now have an unvalued D feature both in D and Num in (61a); and, second, and perhaps worse, we now have two empty Ds: an empty D for the ‘bare plural’ and an empty D for the bare singular (61b). (61)

a.

b.

D nan kas [ ]D-F [plural]Num-F [ ]Num-F [ ]D-F D kas [ ]D-F

(empty D has its value restricted)

(kind creator)

One way to solve this problem is to assume that there is no empty D. Rather we have a D/Num head where the plural is inserted. This would make the bare plural similar to the indefinites. The intuition here is that either -nan is inserted into a D/Num head, in which case it does not have unvalued features or it is selected by a definite D, in which case it must enter an Agree relation. For completeness we should go back to the bare singulars. Remember that the D of bare singulars is a D that creates kinds. It selects for [-singular] but crucially it does not have any unvalued D feature. Therefore it does not act as a probe for any goal and consequently a numberless projection is a possible complement for it. A plural would also be possible if -nan did not have D features, as is the case of plural -s in Brazilian Portuguese. Summarizing, nan is a morpheme with D and Num features, much like other indefinites in Papiamentu. However, unlike other indefinites, it can satisfy its D features via agree with a definite head. Cardinals cannot. They can appear as D/Num heads, when they are indefinite heads, and as specifiers of Num in definite DPs. Finally bare singulars have a morphologically empty D, whose semantic function is to create predicates. Interestingly, -nan differs from the other indefinites in forcing the DP to be interpreted as familiar. It is not surprising then that it can co-occur with the definite determiner. This may be correlated with the fact that it is homophonous and undoubtedly related to the third person plural pronoun, as illustrated in (62a), suggesting that they may be the same morpheme. Independent support for this connection between the plural and the third person plural is the fact that -nan can be an affix on proper names, deriving a so-called associative construction, as in (62b), suggesting that it either has a D feature or that it is compatible with D features.

 Ellen-Petra Kester and Cristina Schmitt

(62)

a.

b.

Nan ta kome sòpi. they  eat soup ‘They are eating soup.’ Marianan Maria-3 ‘Maria and her family/her group of friends.’

P

P

The dual featural composition of -nan, namely, plural and D feature, may perhaps be held responsible for its peculiar distribution and its D-linking requirements.43 In sum, our proposal attributes very different features to the plural -s and -nan. While the former is semantically a predicate modifier with no D features, the latter has a D feature as well. This difference in their semantics guarantees differences in distribution. If we consider the definite and the plural in both languages, it seems to be the case that these elements have a direct or indirect referential status in Papiamentu, but not in Brazilian Portuguese. In Brazilian Portuguese plural morphology carries no D features, while -nan in Papiamentu is always somehow D-linked and can never be used as an agreement marker. The definite article allows the so-called expletive uses in Brazilian Portuguese, but not in Papiamentu. The correlation between the definite and the plural in Papiamentu and their use as strong pronouns and not agreement markers suggests that the featural properties of strong pronouns can be preserved when they take overt complements.44

.

The specificity of -nan and unspecificity of -s

The bare plurals with -nan, as we have seen before, are excluded from existential sentences contexts (12) and from generic contexts as well (11a, c, e). They are also clearly not definite, since they can co-occur with the definite, as in (53b) and (54a). As we have seen, the distribution of bare plurals seems to be discourse dependent. More specifically, the ability of nouns pluralized by -nan to appear as subjects of episodic sentences suggest that some degree of specificity may be involved (see Haspelmath 2001). As is well-known, some languages mark specificity morphologically. Well-known cases are Spanish and Turkish where specificity seems to be encoded morphologically by case markers (see von Heusinger 2002 for an overview and references therein). But not only case markers may be associated with specificity. In certain Ewe languages, besides bare singulars, there are indefinite singulars that must be interpreted as specific. This is the case of the indefinite singular in Gen-Mina spoken in South Togo and Bénin. In this language the indefinite specific is translated roughly as ‘a certain’.

. It should also be noted that in various languages plural morphology is associated with quantificational properties with definites or specifics (Bole-Richards 1983 for Gen-Mina (Ewe); Aboh (1999) for Gungbe (Ewe); and Kihm, this volume for a Portuguese Creole). . We thank Jacqueline Guéron for bringing this generalization to our attention.

Chapter 4. Papiamentu and Brazilian Portuguese 

According to Bole-Richards (1983:221), a nominal marked by this element is not identified, but is selected among the possible ones, and hence it cannot be generic.45 Clearly -nan without a definite determiner is showing a distribution that is discourse dependent. Although it is beyond the scope of this paper to provide a fullblown analysis of the discourse function of -nan, we would like to end this section by suggesting an approach that may be promising to account for its properties. There has been a lot of work on specificity in recent years. Specificity is a cover term for a bundle of phenomena in which “the choice of a value for the introduced discourse referent is specified/restricted or varying.” (Stark 2002: 320). In other words, there are various levels of specificity. The most common definition of specificity is that in which the speaker has a particular referent in mind. However, as noted by various researchers, there are other types of specificity that do not seem to require a particular unique referent in the speaker’s mind. According to Enç (1991), for example, partitives (‘two of the books’) are specific because they are linked to the discourse via a partwhole relation. Drawing on this proposal, Geurts (2003) argues that specificity is to be associated with backgrounding. He defines the opposition between foregrounding and backgrounding in terms of informational prominence. Backgrounded information is less prominent, that is, less central to the concerns of the speaker than foregrounded information. As Geurts observes, backgrounded information can be given or new. Its most salient feature is that backgrounded information is of secondary importance in the utterance. Crucially, Geurts shows that indefinites are preferably interpreted in situ unless they are made ‘heavier’ by means of modification, as illustrated in the following pair of examples. (63)

a. b.

Several students reported that they had been harassed by professors. Several students reported that they had been harassed by professors wearing false beards and pink gowns.

The use of the modifier in (63b) enforces a specific reading of the DP (moving at LF to a position where it has wide scope), whereas the example in (63a) is more likely to be interpreted non-specifically and, hence, stay in situ. The main duty of the modifier is to help identify the intended referent and make backgrounding possible. Possibly the notion of backgrounding is relevant for the distribution of -nan in Papiamentu. As we have seen before, the use of bare plurals in Papiamentu is very restricted. They are only felicitous in contexts where (i) the subject can be interpreted as specific; (ii) the subject has a contrastive/focus reading; (iii) the object is modified by an adjective or relative clause. The specific reading is illustrated in (64) repeated from (13a). (64)

.

Despues ku *hende/hendenan a keha, nan a after that person/persons  complain, they 

See also Aboh (1999) for similar facts in another Ewe language.

 Ellen-Petra Kester and Cristina Schmitt

drecha e película. fix the film ‘After some people complained, they fixed the film.’

P

A contrastive/focus reading is involved in cases such as (65): (65)

a.

b.

Muhénan ta lesando i hòmbernan ta skibiendo. woman-pl are reading and man-pl are writing ‘Women are reading and men are writing.’ Muhénan a bisti shimís, hòmbernan a bisti dashi. woman-pl  wear dress, man-  wear tie ‘Women were wearing dresses, men were wearing ties.’

P

P

Finally, we see in (66) repeated from (16) that the overt plural marking of the object is possible when the noun is modified. (66)

a.

b.

Mi ta mira bukinan na spañó riba mesa. I  see book-pl in Spanish on table ‘I see Spanish books on the table.’ El a bin ku regalonan mashá karu pa mi. he  come with present- very expensive for me ‘He came with very expensive presents for me.’

P

P

Recall that in Papiamentu bare plurals without modifiers are unacceptable in object position for most speakers. We would like to suggest that the cases mentioned above can be analysed in a similar way, if we assume that nominals with -nan must be backgrounded, in a way to be made more precise in future research. As an initial step we suggest that the subject in (64) is backgrounded by contextual information regarding the people in the audience of a movie theatre; in (65) backgrounding is expressed by the contrastive/focus intonation; and finally, in (66), the objects are backgrounded by means of modification. In sum, what we are saying is that the element -nan in Papiamentu is a portmanteau morpheme, which expresses plurality and a D feature. Moreover it imposes a condition that the DP with -nan must be associated with a backgrounded element.46

. The idea that certain elements can mark background or foreground information is supported by the historical analysis on the evolution of determiners. Old Italian (Stark 2002) seems to show that in earlier stages of Italian, where we find bare singulars, bare plurals and nominals with preceding functional elements, the latter appear systematically marking foregrounded information, often with postverbal internal arguments. On the other hand, Chen (2003) argues that the classifier+one in Chinese can appear with proper names and, in this case, it marks backgrounded information.

Chapter 4. Papiamentu and Brazilian Portuguese 

.

Conclusion

In this paper we have presented a comparative study on bare nominals in Papiamentu and Brazilian Portuguese, taking into account also the behavior of the definite and indefinite articles in these languages. We have provided empirical support for a semantic analysis of bare singulars as names for kinds, whose syntactic structure corresponds to a DP which is headed by an empty determiner and not specified for number. The differences in distribution and interpretation of bare nominals in the two languages under consideration are assumed to be associated to variation in two domains: (i) there is variation in the features of the functional projections on the path from NP to DP, which accounts for the very different distribution of bare plurals in both languages; (ii) there is variation in the syntactic and semantic properties of other functional elements competing for the same position, such as the definite article. Put differently, for bare singulars, the linguistic variation is not explained by the different syntax or semantics of the bare nominals themselves, but by differences among the competing items for that position in the different languages. With respect to the plural, whereas the Brazilian Portuguese plural affix -s has no D features, Papiamentu -nan is better described as a portmanteau morpheme expressing two different and interpretable features of the noun: plurality and a D feature that imposes backgrounding of the DP. The different features of these plural markers account for the very restricted distribution of ‘bare plurals’ in Papiamentu when compared to Brazilian Portuguese bare plurals.

References Aboh, E. 1999. From the Syntax of Gungbe to the Grammar of Gbe. Sierre: Éditions à la carte. Bobaljik, J. 1995. Morphosyntax: The syntax of verbal inflection. PhD Dissertation, MIT. Bole-Richards, R. 1983. Systématique phonologique et grammaticale d’un parler éwé: Le gen-mina du sud-togo et sud-bénin. Paris: Editions L’Harmattan. Bosque, I. 1996. Por qué determinados sustantivos no son sustantivos determinados. Repaso y balance. In El sustantivo sin determinación. La ausencia de determinante en la lengua española, I. Bosque (ed.), 13–119. Madrid: Visor Libros. Carlson, G. 1977. A unified analysis of the English bare plural. Linguistics and Philosophy 1: 413–457. Chen, P. 2003. Indefinite determiner introducing definite referent: A special use of yi’one’+classifier’ in Chinese. Lingua 113: 1169–1184. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Chomsky, N. 2001. Derivation by Phase. In Ken Hale: A life in language, M. Kenstowicz (ed.), 1–52. Cambridge MA: The MIT Press. Cinque, G. 1993. On the evidence for partial N-movement in the Romance DP. Ms. University of Venice.

 Ellen-Petra Kester and Cristina Schmitt Contreras, H. 1986. Spanish bare NPs and the ECP. In Generative Studies in Spanish Syntax, I. Bordelois, H. Contreras & K. Zagona (eds), 25–49. Dordrecht: Foris. Dayal, V. 2002. Number marking and (In)definiteness in kind terms. Linguistics and Philosophy 27: 393–450. Diesing, M. 1992. Indefinites. Cambridge MA: The MIT Press. Dijkhoff, M. 1983. The process of pluralization in Papiamentu. In Studies in Caribbean Language, L.D. Carrington (ed.), 217–229. Trinidad: Society of Caribbean Linguistics. Dijkhoff, M. 1990. Gramátika Modèrno di Papiamentu: Artíkulo e Sustantivo. Curaçao: ILA/ KOMAPA. Embick, D. and Noyer, R. 2001. Movement operations after syntax. Linguistic Inquiry 32: 555–595. Enç, M. 1991. The semantics of specificity. Linguistic Inquiry 22: 1–26. Frampton, J., Gutmann, S., Legate, J. & Yang, C. 2000. Remarks on ‘Derivation by Phase’: Feature valuation, agreement, and intervention. Ms., MIT. Geurts, B. 2003. Specific indefinites, presupposition, and scope. In Presuppositions and Discourse, R. Bäuerle, U. Reyle & T.E. Zimmermann (eds). Oxford: Elsevier. Guéron, J. 2003. Inalienable possession and the intepretation of determiners. In From NP to DP. Vol. II. The expression of possession in noun phrases, M. Coene & Y. d’Hulst (eds), 189–220. Amsterdam: John Benjamins. Goodman, M. 1987. The Portuguese element in the American Creoles. In Pidgin and Creole Languages, G. Gilbert (ed.), 361–405. Honolulu HI: University of Hawaii Press. Halle, M. and Marantz, A. 1993. Distributed morphology and the pieces of inflection. In The View from Building 20: Essays in linguistics in honor of Sylvain Bromberger, K. Hale & S.J. Keyser (eds). Cambridge MA: The MIT Press. Haspelmath, M. 2001. Indefinite pronouns. Oxford: OUP. Heim, I. 1991. Artikel und definitheit. In Semantics: An international handbook of contemporary research, A. von Stechow & D. Wunderlich (eds), 487–535. Berlin: Walter de Gruyter. von Heusinger, K. 2002. Specificity and definiteness in sentence and discourse structure. Journal of Semantics 19: 245–274. Heycock, C. & Zamparelli, R. 2003. Friends and Colleagues: Plurality, coordination and the structure of DP. Ms. University of Edinburgh and University of Bergamo. Kester, E.-P. In progress. On Pluralization in Papiamentu. Ms. University of Amsterdam. Kratzer, A. 1995. Stage-level and individual-level Predicates. In The Generic Book, G. Carlson & J. Pelletier (eds), 125–175. The University of Chicago Press. Ladusaw, W.A. 1994. Thetic and categorical, stage and individual, weak and strong. In Proceedings of SALT IV, M. Harvey & L. Santelmann (eds), 220–229. Ithaca NY: Cornell University DMLL. Löbner, X. 1985. Definites. Journal of Semantics 4: 279–326. Longobardi, G. 1994. Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry 25: 609–665. Marantz, A. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. In Proceedings of the 21st Annual Penn Linguistics Colloquium [Penn Working Papers in Linguistics 4: 2], Alexis Dimitriadis et al (eds), 4–225. Philadelphia PA: University of Philadelphia. Martinus, F.E. 1996. The Kiss of a Slave: Papiamentu’s West-Africa connections. PhD Dissertation, University of Amsterdam. Maurer, P. 1988. Les modifications temporelles et modales du verbe dans le Papiamento de Curaçao (Antilles Néerlandaises). Hamburg: Buske.

Chapter 4. Papiamentu and Brazilian Portuguese  Muller, E. 1989. Inleiding tot de Syntaxis van het Papiamentu. PhD Dissertation, University of Amsterdam. Munn, A. & Schmitt, C. 2001. Bare nominals and the morphosyntax of number. In Current issues in Romance Linguistics, T. Satterfield, C. Tortora & D. Cresti (eds), 217–231. Amsterdam: John Benjamins. Noyer, R. 1997. Features, Positions and Affixes Autonomous Morphological Structure. New York NY: Garland. Schmitt, C. & E-P Kester. (2005). Bare nominals in Papiamentu and Brazilian Portuguese: An exo-skeletal approach. In Theoretical and Experimental Approaches to Romance Linguistics: Selected papers from the 34th Linguistic Symposium on Romance Languages (LSRL), Salt Lake City, March 2004, R. Gess & E. Rubin (eds.) 237–256. Amsterdam: Benjamins. Schmitt, C. & Munn, A. 1999. Against the nominal mapping parameter: Bare nouns in Brazilian Portuguese. In Proceedings of NELS 29, P. Tamanji, M. Hirotani & N. Hall (eds). 339–353. University of Delaware. Schmitt, C. & A. Munn. 2003. The syntax and semantics of bare arguments in Brazilian Portuguese. In Linguistic Variation Yearbook 2: 185–216. Stark, E. 2002. Indefiniteness and specificity in old Italian texts. Journal of Semantics 19: 315–332. Vergnaud, J.-R. and Zubizarreta, M.L. 1992. The definite determiner and the inalienable construction in French and English. Linguistic Inquiry 23(4): 595–652.

 

On the interpretation of bare noun phrases in Guinea-Bissau Portuguese Creole (Kriyol)* Alain Kihm Guinea-Bissau Portuguese Creole or Kriyol is one of those languages where bare nouns have free access to syntax. Bare nouns are given a restrictive definition: only nouns appearing in their root form, i.e. not marked for number, are considered bare. Kriyol is thus different from Portuguese or English insofar as so-called bare nouns in these languages must be pluralized. Another distinctive feature of Kriyol is the absence of a definite determiner analogous to English the. The conjunction of these features – bare nouns unmarked for number, no overt definite determiner – makes the interpretation of Kriyol noun phrases appear problematic in many cases. The aim of this article is to show that, once semantic (does the noun refer to a person or not?) and syntactic (is the noun a subject or an object?) factors have been taken into account, Kriyol actually is no more ambiguous in this domain than natural languages normally are.

Introduction Guinea-Buissau Portuguese Creole or Kriyol is one of those languages where bare nouns are given free access to the syntax.1 Bare nouns will be defined restrictively in this work: only nouns appearing in their root form, i.e. not marked for number, will be considered bare. Kriyol is thus different from Portuguese or English insofar as socalled bare nouns in these languages must be pluralized (cf. Tenho gatos/*gato na casa ‘I have cats/*cat at home’). Hence, Portuguese and English do not accept bare nouns under this strict definition. Another distinctive feature of Kriyol is the absence of a definite determiner analogous to Portuguese o(s)/a(s) ‘the’, whereas there exists an indefinite determiner un(s)

* I want to express my gratitude to Marlyse Baptista and to Jacqueline Guéron, whose extensive and careful comments greatly helped me to improve this article. That it still falls short of what it ought to be is entirely my fault. . On Kriyol in general, see Kihm (1994).

 Alain Kihm

‘a, some’, whose use is however partially unlike that of its Portuguese equivalent and etymon um(a)/uns/umas. The bundling of these features – bare nouns, no (overt) definite determiner, an indefinite determiner in need of clarification – makes the interpretation of Kriyol noun phrases problematic in many cases. The aim of this article is to explore and hopefully to unravel this problem to some extent. The first section below will be devoted to describing it. Then, in Sections 2 and 3, I will examine how the interpretations described in Section 1 are arrived at. In Section 4, I introduce the notion of “abstractability” which is necessary, I will argue, in order to explain how nouns not denoting intrinsically massive entities may nevertheless be assigned mass interpretation. A possible syntactic structure for bare noun phrases is then examined in Section 5, and the way to derive individual or mass interpretation in Sections 6 and 7, where a notion of lexical incorporation is entertained. Finally, I return to the root of the difference between Kriyol on the one hand, and Portuguese and English on the other hand, with a passing look at French which contributes to clarifying the issue, as it marks “partitive” meaning overtly. As a final warning to the creolist component of my readership, I want to say that this work is entirely a synchronic study of a language that happens to be a Creole. Therefore, the fact it is a Creole, i.e. a language that branched off from Portuguese in a different way than, say, the latter evolved from Latin, although certainly relevant, will not be taken into consideration.

.

The problem

Noun phrases in Kriyol present the following descriptive features:2 (i) There is no overt definite determiner (cf. minjer bonitu /woman/ ‘the handsome woman’); (ii) there is a quantifier un(s) ‘one, a, some’ which functions as a singular or plural indefinite determiner (cf. un minjer bonitu ‘a/some handsome woman’, uns minjeris bonitu ‘certain handsome women’);3 (iii) noun phrases overtly pluralized with /-(i)s/ suffixed to the head noun are interpreted as referential (cf. minjeris bonitu ‘the/some handsome women’, not *‘handsome women’ generically understood), and nonreferential, i.e. generic or mass noun phrases are not marked for plurality even though they necessarily denote sets having more than one member.4 . I use “noun phrase” as a cover term for the three levels the extended projection of nouns is commonly assumed to comprise, viz. NP, NumP, and DP. . Gender is inactive in Kriyol as in most Romance based Creoles (and beyond), so that (un) minjer bonitu is like (un) omi bonitu ‘a handsome man’ (compare Portuguese a/uma mulher bonita vs. o/um homem bonito). Also note that modifiers do not agree in Number with the head noun, to the difference of Portuguese (cf. (as) mulheres bonitas). .

See below for a definition of (non)referential and related concepts.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

Actually, this is an idealized description because Kriyol is not a coherent language in this domain. To put it more accurately, the social object (E-language) called Kriyol does not stand in a one-to-one relationship with a grammar (I-language), a perfectly common situation. In fact, three different grammars can be isolated from the observation of actual productions. First, there is what I will call the basilectal (sub)grammar, having the following properties:5 (a) [(un) N] noun phrases (un optional) are interpreted as [+R(eferential), −Def(inite), −Num(ber)]; (b) [N(-s)] noun phrases (/-s/) optional) are interpreted as [+R, +Num]; (c) [N] noun phrases (bare noun phrases, henceforth BNPs) receive all other interpretations, plus the interpretations induced by un and /-s/ when they are not spelled out. Clause (c) is a major source of ambiguities. Quite possibly, this grammar developed from an anteceding one, which had BNPs only (to the exception of the demonstratives es ‘this’ and kil ‘that’, for which see below). Optional insertion of un and /-s/ was thus a first step towards a more grammaticalized (less pragmatically dependent) expression of noun phrase referentiality. We are probably dealing with a phenomenon pertaining to the “naturalization” process through which pidgins, as possible instances of protolanguage, become creoles, i.e. real language (see Bickerton 1998; Kihm 2000a). Basilectal grammar gave rise to a different grammar through a minimal adjustment, viz. making un(s) and /-s/ insertion obligatory. This other subgrammar, which I call simply Kriyol, is the one described in (i)–(iii), and it is the most commonly observed (or at least was in the late 1970s and the 1980s when I did my fieldwork). In it, clause (c) only says: [N] noun phrases (BNPs) receive all other interpretations, thus becoming the Elsewhere case. Finally, there is a non-basilectal grammar, closer to Portuguese, where clause (b) says: [N-s] noun phrases are interpreted as plural, and /-s/ must be suffixed to every NP with a referent the cardinality of which is known to be greater than one. In this grammar, you cannot leave plurality unexpressed, and there is no connexion between overt plurality and referentiality. The problem for the describing linguist – and also perhaps for the acquiring child – is that these three grammars are not segregated behind fences raised by ethnicity, social class, age, etc., but they are represented as alternating variants in the minds of the native speakers as evidenced by the mixed character of the latter’s performances. No doubt such heterogeneity is not haphazard, as it never is, and careful sociolinguistic investigations would succeed in showing how it is determined. I will not address that problem, though. Rather, I will avail myself of the researcher’s indisputable right (and inescapable need) to idealization and consider only the grammar outlined in (i)–(iii), calling it Kriyol, since, after all, it does exist, it is dominant to a large extent, and it . In creolist tradition “basilect” refers to that variety of the language that is used by speakers who have very few or no contacts with the lexifier language when it is a component of the overall linguistic environment, as is the case with Portuguese in Guinea-Bissau.

 Alain Kihm

probably is the one easiest to isolate.6 The question I wish to ask, then, is how a grammar that generates the kind of ambiguity illustrated by the following sentence works: (1)

Minjer kumpra kabra woman buy goat

This sentence is ambiguous between the two following translations-interpretations: (1a) (1b)

The woman bought the goat7 The woman bought goats

The object BNP in (1) can be taken to refer to the particular goat the woman bought, or to an unknown quantity of goats she bought. Interestingly, given a context, meaning (1b) is compatible with a situation where the woman actually bought only one goat, so another proper translation would be ‘The woman bought a goat’. Yet, (1) would still not be synonymous with Minjer kumpra un kabra ‘The woman bought a (certain) goat’, so the availability of this interpretation does not warrant the conclusion that un might be optional in Kriyol, and in fact it is not. Accounting for this possibility will turn out to be a crucial step for understanding noun phrase semantics in the language. Also note that, for reasons to be clarified later on, (1a) cannot be understood as meaning that the woman bought “the goats”. The subject BNP, in contrast, can only be interpreted as indicated. How are these interpretations arrived at, and what are the conditions that make them available or unavailable? Such are the questions we have to answer if we wish to show that a grammar generating (1) is indeed functional. As I will try to show, meaning is in fact always retrievable given the interplay of two factors: (a) referentiality rather than definiteness is crucial; (b) plurality goes with referentiality. Given this, nonreferential and nonnumbered BNPs receive an interpretation analogous to the mass reading English and Portuguese reserve for nouns denoting entities such as “water”, but which all nouns, to the crucial exception of nouns referring to humans, can assume in Kriyol.8 In parallel to the semantic constraint just mentioned, there is a syntactic one: mass interpretation of non inherently mass nouns is only available if the said nouns are in a position such that they can incorporate to the predicate they are an argument of, in a special sense of “incorporation”. As we shall see, only the internal argument (object) position may have this capacity.

. Basilectal grammar would be very hard to study because of its multiple ambiguities. “Decreolized” grammar, on the other hand, is highly unstable. . Finite verbs not in the scope of an overt aspect marker are interpreted as perfective. As we shall see later on, aspect of the predicate is relevant to the interpretation of a BNP object. . Applying Chierchia’s (1998) typology, we would therefore say that Kriyol sides with languages like Chinese. I will not pursue along this line, however, because I am not convinced that such typology has any real validity.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

.

Deriving the correct interpretations

In this section, I wish to examine if and how the interpretations in (1a) and (1b) may be arrived at entirely from the properties of Kriyol grammar outlined above. No assumptions as to the syntactic structure of the expressions are entertained. In particular, I take BNPs at their face value, not supposing, at least for the present, there might be more to them than what shows. This, I think, is only sound method – if minimalist, so much the better. First, however, something must be said about the semantic features used to characterize noun phrases, and also, for the sake of clarity, about the overall shape of nominal expressions in Kriyol. Concerning the latter, it is enough for my purpose to give an example of a near “maximal” noun phrase, namely:9 (2)

es/kil ña sinku bon kabra this/that my five good goats these/those five good goats of mine

The following syntactic structure may be assigned to (2): (3)

[DP [D0 es][ModP [Mod0 ña][NumP [Num0 5][ModP’ [Mod0’ bon][NP [N0 kabra]]]]]]

This is a rather straightforward structure, no doubt much too simple-minded, but sufficient for my purposes. It assumes that es and kil, even though they are not determiners strictly speaking, as they always exert a demonstrative force that “pure” determiners like Portuguese o(s)/a(s) or English the lack, head the DP projection which, as we shall see later on, need not be projected in Kriyol.10 DP would also be the projection for un as suggested by the latter’s position vis-à-vis possessives (cf. un ña bon kabra ‘a good goat of mine’). DP thus dominates a projection I noncommittally call Modifier which hosts the possessive paradigm ña ‘my’, bo ‘your’, si ‘her/his/its’, no ‘our’, bos ‘your’, se ‘their’.11 This in turn dominates a Number Phrase (NumP) whose head is either the functional category [+plural] spelled out as /-(u)s/ or a numeral or a quantifier such as manga di ‘many’ excluding plural marking on the head noun.12 . One thing that could be added to (2) is an adjective (and/or a relative clause). Not to complicate matters unnecessarily, I leave the issue of adjectives in Kriyol aside (see Kihm 2000b). . In addition to being a distal deictic demonstrative, kil may be used with anaphoric force as in kil kabra ‘that goat [you were talking about]’ (see Kihm 1994: 138–139). That does not make it into a determiner, however, contrary to what seems to have happened with Cape Verdean kel/kes (see Baptista 1997; 2002: 23ff). .

Bo is for one possessor, bos for several.

. When Num0 is functional, N0 adjoins to it while the preposed modifier (see below) presumably moves to the specifier of NumP (cf. bon kabras ‘the good goats’). No movement is required, in contrast, when Num0 is lexical (see Beard 1995 for the morphological theory involved).

 Alain Kihm

Let us turn now to the semantic features of noun phrases. Following Lyons (1999), I make a distinction between referentiality and definiteness. A noun phrase is referential insofar as it denotes one token or a (potentially) countable set of tokens of a given type; it is definite if the token or token set is assumed by whoever generates the noun phrase to be identifiable in the situation where the noun phrase is proposed for interpretation. Different construals of the notions “referential” and “definite” are of course possible and in use.13 There is in particular the notion that referential noun phrases denote entities that exist independently from the truth value of the proposition that includes them, while definite noun phrases denote maximal sets in the currently described situation. It seems to me, however, that presupposition of existence as in “A goat entered the room” (which may be false qua event) is a necessary consequence of denotation, be it individual, generic, or even negative – since “No goat entered the room” means that it is not the case that one or several goats, whose existence is presupposed, entered the room.14 Referentiality and existence are thus two separate concepts. On the other hand, if definiteness was set maximally as in “The goats (i.e. all and the only ones that were supposed to be present in the current situation) entered the room”, then we would be hard put to explain how a singular definite DP like “the goat” may actually refer to the set of all goats in “The goat is a mammal”. There is a natural explanation, in contrast, if we assume that kinds are identifiable as such, that is they are accessible from any discourse situation. That said, it must be conceded that using a feature notation for such semantic constructs is actually a mnemonic, convenient device rather than a hypothesis about their ontology. Both features (if that is what they are) are mutually independent in principle: a goat in I bought a goat yesterday is referential as it refers to a particular (specific) token of the type “goat”; it is not definite, because I assume you had no previous idea about what I bought, or even that I bought anything. In contrast, the goat in The goat is a mammal is definite as well as being generically interpreted, because it refers to the kind “goat” and, as just proposed, kinds are identifiable by definition (they belong to basic shared knowledge), but it is not referential as no particular member(s) of the kind is/are meant.15 Actually, independence is not absolute, as there is an implicature from definiteness to referentiality (although not vice-versa), which explains the marginality, cross-linguistically and even within English, of The goat is a mammal, more commonly phrased as Goats are mammals, with a bare plural as the favourite means to express

. I am grateful to Jacqueline Guéron for comments on this point, which prompted me to attempt this necessary clarification. .

Such is indeed the magical force of language, which gives existence to what it evokes.

. Portuguese is like English in this respect: compare Comprei uma cabra ontem and A cabra é um mamífero.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

genericity in the language.16 Despite this and other complicating factors, however, a lot can be achieved by freely combining these two features, plus a Number feature. Being BNPs, minjer and kabra in (1) fall under the provisions of clause (c) above, i.e. the Elsewhere case. Let us consider the object BNP first. The reading given in (1a) (‘the goat’) is partially expected since kabra is not in the scope of un, hence it is not [−Def], it is not overtly plural, and goats belong to a category of animals I call “individualizable”, meaning they are linguistically treated like humans to some extent (see below). Therefore, all things being equal, kabra would be expected to be pluralized if it referred to more than one goat. The question is: where does it take its [+R] feature from? In fact, the (1a) reading seems to be only possible in a context such that the topicality of the referent is clear (e.g. Minjer bin fera. I oja un kabra. I kumpra kabra. ‘The woman came to the market. She saw a goat. She bought the goat.’), i.e. a context where the object noun phrase can be replaced by a pronoun (cf. I oja un kabra. I kumpra-l ‘She saw a goat. She bought it.’). I will therefore assume that referentiality is checked by the informational component (see Vallduví 1995) in a sense to be made clearer later on.17 Whether the assumption can be fully held remains to be seen. Meanwhile, it dispenses us from unwarranted suppositions concerning the syntax of the expression. What about (1b)? Here, the crucial factor seems to be that [+R] is not checked by the informational component. I take referentiality rather than definiteness to be the relevant feature here, which may deserve a few words of explanation. Clearly, kabra does not seem to be definite in (1) meaning (1b) because, if it were, either (1b) would be nondistinct from (1a) contrary to fact, or it would refer to the kind “goat”, which does not make sense in the context of (1) (see below), or it would have to be plural and show up as kabras (see below). But neither is it indefinite, i.e. to be translated as ‘a goat’ or ‘some goats’, depending on the value of [Num], because that would require the presence of un or uns. That is to say, it does not explicitly refer to a discrete, if unstated, quantity of goats, which is why (1) meaning (1b) is compatible with a situation where the woman actually bought one goat – as it turns out – although it does not imply anything about quantity by itself (see below). The value of [Def] thus does not seem to affect interpretation, which is, crucially, that kabra in (1) meaning (1b) has no individual denotation. I will return to this. This leaves two possibilities for the interpretation of kabra in (1) meaning (1b) given clause (c): [−R +Def −Num] or [−R −Def −Num]. The first reading corresponds

. Similarly in Portuguese: As cabras são mamíferos, except that bare plurals are not available for generic interpretation in (European) Portuguese, where they can only be understood as indefinite referential plurals, a meaning that is of course also possible for English bare plurals (cf. Goats crossed the street). . Note the referentiality of fera translated as ‘the market’ rather than ‘a market’ or ‘markets’ is also something for the informational component (alias “pragmatics”) to take care of.

 Alain Kihm

to the generic or kind interpretation of singular definite DPs in Portuguese or English in such sentences as A cabra é um mamífero ‘The goat is a mammal’ (see Footnote 15). Interestingly, this reading is available for BNPs in Kriyol, but it can only be assigned to the subject of (1) provided the tense-aspect features and the overall meaning of the sentence permit it. See for instance (4):18 (4)

Kabra ka ta kume karni goat Neg Ipf eat meat The goat / a goat / goats do(es) not eat meat (A cabra / uma cabra / cabras não come(m) carne) (NB: Ipf = Imperfective)

Note, however, that Kabra ka ta kume karni may also express, not a proposition about the caprine species, but a general observation about a particular goat (similarly in Minjer ta kumpra kabra – see Footnote 18). The same ambiguity is observed in English The goat (in question) does not eat meat or Portuguese A cabra não come carne. Given the innate vegetarianism of goats, which renders (4) rather otiose as an individual proposition, (5) may be a more convincing example: (5)

Kabra i amigu di kacur goat 3sg friend of dog

This is a quotation from a folktale where kabra and kacur are proper names (Kabra ‘Mr. Goat’ and Kacur ‘Mr. Dog’). Out of the blue, though, (5) can be understood as either a proposition about individual members of the caprine and canine species, who are described as friends, or a proposition about the species themselves, claimed to be generally friendly to each other (see the ambiguity of The goat is the friend of the dog or A cabra é a amiga do cão). Now, the individual (referential) interpretation of the BNP kabra appears to be correlated with its overtly topical position (and it is not excluded by the non-topical position of kacur).19 Kind interpretation must therefore be a particular case of individual interpretation, as various authors have argued (see Krifka et al. 1995). The two possible interpretations of kabra and kacur in (5) thus fall under the provisions of the grammar, if general interpretive principles are taken into account.

. Similarly (1) with imperfective aspect (Minjer ta kumpra kabra) could be understood in this way if it was a known fact that buying goats is a woman’s prerogative. I am grateful to Marlyse Baptista (pc) for drawing my attention to this point. Incidentally (4) shows that both definite and indefinite, singular and plural are compatible with generic interpretation in English and Portuguese. I will briefly return to this point in my concluding remarks. . That kabra is topicalized – although probably not dislocated – is shown by the occurrence of the resumptive pronoun i ‘s/he/it’. Topicalization of the subject is obligatory in Kriyol when the predicate is a noun phrase (see Kihm 1994; for a different analysis on Cape Verdean data, see Baptista 1997).

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

The question now is why this interpretation is not available for the object BNP in (1), even when the meaning of the sentence would seem to allow for it, as in: (6)

Jinti ta kumpra kabra (na fera) people Ipf buy goat (in market) People buy goats (in the market)

In this context, kabras would be odd, given that the referentiality feature attached to Number marking (property [iii] of Kriyol grammar) would mismatch with the nonreferentiality of the imperfective predicate referring to an habitual or iterative process. Yet, kabra cannot be understood as denoting the kind “goat”, except perhaps in an overtly contrastive context where goat-buying in the market is opposed to, say, rabbitbuying in stores. Even then, it is not clear that kinds as such are denoted, rather than types of animals (or meats) with respect to particular situations.20 And since kabra cannot mean ‘a (certain) goat’ which would be un kabra, it has to be understood as referring to an unspecified, perhaps unknown quantity of goats, as already pointed out, unless the informational context warrants the reading ‘the goat’. Moreover, as already mentioned, the unspecified quantity includes one as its lower bound, for reasons to be exposed presently.

. “Elsewhere” interpretations Let us review the ‘other interpretations” coming under clause (c), namely the possible meanings of unpluralized noun phrases not modified by un, i.e. BNPs. It seems we can point out three: (i) [+R, +Def, −Num] (kabra ‘the goat [in question]’); (ii) [−R, +Def, −Num] (kabra ‘the goat [as kind]’ – see (2) under one reading); (iii) [−R, −Def, uNum] (u = unspecified – see (1) = (1b)). As noted, interpretation (ii), i.e. kind interpretation, is heavily dependent on the type of proposition the BNP is an argument of, and we are not even sure (ii) should be formally distinguished from (i). I will therefore leave kind interpretation aside. Our task is then to lay out the conditions calling for (i) or for (iii). As mentioned above, (in)definiteness is assigned by the informational component, according to whether the speaker assumes the referent of the noun phrase to be identifiable or not by the hearer. In Kriyol, contrary to English or Portuguese, this assumption has no overt manifestation if the assumed value for definiteness is 〈+〉; if 〈–〉, un expresses it, but only if the noun phrase is also [+R]. Given this, the crucial feature of BNPs interpreted as in (iii), I will argue, is that they are unspecified for number.

. Indeed, this is a case where kind-denoting definite noun phrases do not sound felicitous: compare The goat is sold in the market, the rabbit in stores with Goats are sold in the market, rabbits in stores or Goat is sold in the market, rabbit in stores (meaning the meat).

 Alain Kihm

That is to say, Kriyol speakers have the option of generating noun phrases without having to specify whether the referent’s cardinality is one or more than one. This option is not open to English or Portuguese speakers because morphology forces interpretive decisions in these languages: singular if no /-s/, plural if /-s/.21 In Kriyol, in contrast, overt plural morphology is limited to referential noun phrases, so that non-pluralized [+R] noun phrases are indeed interpreted as singular. Conversely, non-pluralization of non-referential noun phrases does not imply uniqueness of the referent, nor does it imply the opposite. Now, not always having to specify number certainly is an advantage, considering the many cases where number, i.e. n = 1 or n > 1, is uncertain, unknown, or simply irrelevant. For instance, uttering (1) in English or Portuguese, I must make a choice about whether the woman bought one or several goats (The woman bought a goat / goats, A mulher comprou uma cabra / cabras). Yet, it may be the case that I do not actually know that the woman bought several goats or just one, or, more commonly perhaps, that I do not deem that information relevant in relation to what I wish to convey. A real example (Kihm 1994: 136) will help clarify what I mean by this: (7) N na disa pilon na porta pa sedu sinal kuma n sta ami son dentru I Prog leave pestle in door for be signal that I stay me alone inside I’ll leave a/the pestle(s) in the door as a signal that I’m alone inside

Consider the first noun phrase pilon (but the issue is the same with porta): translating the sentence into English or Portuguese (cf. Vou deixar um/o pilão / pilões…) forces us to take a stand as to whether the entity denoted by pilon is unique in the situation or one among several of the same. Moreover, (7) is also compatible with a situation where more than one pestle is left in the door, if that does not make a difference – say if the participants, a married woman and her lover, did not agree on a (somewhat redundant) code such that one pestle means ‘all clear’ and two pestles means ‘keep away’. That precisely is the point: for the message to be conveyed, viz. “pestle in the door = you may come in”, and given the world shared by the participants, it is simply not relevant how many pestles belong to the household, or how many will be left in the door. Kriyol grammar allows speakers to take stock of this irrelevance. This harks back to an old principle according to which languages (grammars) differ mainly (only ?) by what they force us to express or allow us not to express. To repeat, then, kabra in (1) or pilon in (7) being possibly translated as ‘a goat’ or ‘a pestle’ does not mean that un ‘a/one’ is optional in Kriyol. It is only consonant with the fact that singularity is a possibility when number is unspecified. What I want to argue now is that such a feature value bundle, viz. [−R, −Def, uNum], defines mass

. Even pluralia tantum like ‘scissors’ or ‘pants’ are plurals, as shown by agreement and the fact we use phrases such as ‘pair(s) of scissors / pants’ when we want to be clear about number.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

interpretation (on which see Chierchia 1998). That is to say, it is a property of Kriyol grammar that kabra, pilon, and all BNPs may be interpreted as only inherently mass terms such as ‘water’ are in English or Portuguese. We therefore have to explore the conditions that make Kriyol BNPs open to such an interpretation.

.

NPs and abstractability

Being able to interpret BNPs as mass seems to depend on two factors. First, given the noun’s denotation, there is the question of the appropriateness of the mental operation that consists in abstracting from the individuals that make up the extension of the concept, and viewing them as an undifferentiated mass.22 Here, there seems to be a gradient such that humans are the least abstractable individuals, individualizable animals are more abstractable,23 and inanimates (including plants) are easily massified. The effect of this factor appears clearly comparing the two following sentences: (8)

Arturu mata kabra tudu si bida Arthur killed goats all his life

(9) #Arturu mata omi tudu si bida Arthur killed men all his life

While (8) is appropriate as a description of the fact that Arthur is or was a goat-killer (working in a slaughterhouse, say), (9) is odd to mean that Arthur is a retired hangman. The reason of the oddity is that, given its meaning, omi ought to be pluralized (omis) in a context where more than one token of the concept is clearly implied. It would thus be assigned a [+R, −Def, +Num] reading, which is also compatible with the intended message.24 Without pluralization, however, and since the concept is not abstractable, omi in (9) can only be interpreted as [+R, +Def, −Num], i.e. ‘the man (in question)’, which is absurd in the context.

. With inherently massive concepts such as ‘water’ no abstracting operation is called for, and it is individuation on the contrary that requires a special operation (cf. ‘Klow and Chivy are the mineral waters I prefer’). Given this, it is not surprising that many languages, including English and Portuguese, only recognize inherently mass terms (although see below), and that many others do not readily lend themselves to mass-to-individual conversion, having to use explicit phrases of the ‘kind of water’ type. . As hinted above, individualizable are those animals imbued with economic and/or cultural significance which are easily considered to be like persons, e.g. dogs, goats, hyenas, monkeys, etc., as opposed to fish, ants, small birds, etc. . Compare Portuguese O Arturo matou homens toda a sua vida, and French Arthur a tué des hommes toute sa vie where the [+R, −Def, +Num] feature bundle is spelled out as des.

 Alain Kihm

Secondly, there is a syntactic factor: all things being equal, object BNPs are more likely to be interpreted massively than are subject BNPs. Take the following sentence: (10)

Kabra na kuri ba na fera goat Prog run Past in market

The most natural reading for it is ‘The goat (in question) was running in the market’, and the reading that some mass of goats was running in the market is not available. Overt pluralization would have to be resorted to (kabras), with a referentiality effect and ambiguity as to whether the referent is identifiable or not (‘The goats were running…’ or ‘Some goats were running…’, although the latter would more readily be expressed by Uns kabra na kuri ba na fera).25 Mass reading is available, however, if kabra is part of an impersonal construction such as (11): (11)

I na kuri ba kabra na fera it Prog run Past goat in market There were goats running in the market26

Given the definiteness constraint on impersonal constructions, the only alternative to (11) is I na kuri ba un kabra / kabras…, meaning that one or a countable number of goats the speaker does not assume to be identifiable by the hearer was / were running in the market. Note moreover that the interplay of the definiteness constraint with the abstractability factor brings it about that *I na kuri omi... is ungrammatical, whereas I na kuri ba un omi / omis… ‘There was / were a man / men running…’ is fine. What this evidence suggests is that the relevant syntax is indeed that of the informational component. That is to say, kabra in (10) cannot receive mass interpretation, although animals are abstractable, because it is in topic position in informational structure.27 On the other hand, the impersonal construction of (11) places kabra within the predicate, hence in focal (see Zubizarreta 1998) or, at least, non-topic position (see Vallduví 1995). The dual implication [−Topic]
⇒ [+mass] vs. [+Topic] ◊ ⇒ [–mass] is insufficient, however, because of the non-obligatory character of the relation between non-topicality and mass interpretation, so that kabra’s meaning cannot be predicted out of context in, e.g. I mata kabra ‘She/he/it killed the goat/goats’. We therefore need something else, not to clear up the ambiguity, which is bound to

. Compare Portuguese (As) cabras estavam a correr no mercado, and French Les/des chèvres couraient dans le marché. Also note that given the sentence’s aspect (progressive, i.e. not habitual or iterative) the kind [−R +Def −Num] interpretation of (4), to be carefully distinguished from the mass [−R −Def uNum] reading, is not available here. . Compare Portuguese Estavam a correr cabras no mercado, and French Il courait des chèvres dans le marché. . Recall that mass is distinct from kind, which is why kabra in (4), topical qua subject, may receive the latter interpretation, but not the former.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

resist in any event given the absence of surface clues, but to explain why there must be an ambiguity. In addition, as we shall see, there is a ranking of the semantic and the informational constraint as far as nouns denoting humans are concerned: non-abstractability tends to get the upper hand. This is why I took the precaution to provide the hedge “all things being equal” when introducing the informational constraint.

.

The syntactic structure of BNPs

The easy solution is to assume that BNPs referring to individuals actually are DPs with a phonologically null head, being thus syntactically distinct from mass BNPs, which are not DPs. This is not an implausible analysis. Since Kriyol grammar is historically related to Portuguese grammar, and the latter shows overt DPs (e.g. Comprou a cabra ‘S/he bought the goat’) in a subset of the contexts where the former has BNPs (cf. I kompra kabra meaning ‘S/he bought the goat’), the change from one to the other may conceivably have consisted in no more than loss of the phonological exponent of D, while the DP structure remained intact. While this assumption explains why Kriyol BNPs may be interpreted like Portuguese overt DPs, it reveals nothing about mass interpretation, however. In European Portuguese (Brazilian may be different) mass interpretation of BNPs is only for nouns denoting intrinsically massive substances (see Não bebe água ‘S/he does not drink water’, Come carne cada dia ‘S/he eats meat every day’, etc.), whatever their syntactic position (see Água está a correr na rua ‘Water is flowing in the street’). The fact that all BNPs, but those denoting humans, can be interpreted as mass in Kriyol may not be attributed, therefore, to a putative phonological deletion of Portuguese D0 , inasmuch as Portuguese overt DPs are never so interpreted (cf. Come carne ‘S/he eats meat’ vs. Come a carne ‘S/he eats the meat’, where the latter is assigned a kind rather than a mass reading). We come to the same result if we assume that Kriyol BNPs result from deletion of the plural morpheme of plural indefinites, which may be their counterparts in Portuguese (cf. Comprou cabras ‘S/he bought goats’). If it were so, Kriyol mass BNPs would actually be unmarked plural indefinites. I reject this assumption because it is contradicted by the fact that I mata omi ‘S/he killed the man’ does not mean the same as Matou homens ‘S/he killed men’, whereas I mata kabra may be equivalent to Matou cabras. Such a contrast cannot be explained, it seems to me, unless massivity is invoked in connection with a hierarchy where men are less abstractable than goats. Moreover, the assumption does not take into account the semantic change undergone by the plural morpheme, i.e. the presence of a referentiality feature in Kriyol /-s/ that Portuguese /-s/ does not include. Another problem with this assumption is that there are massively interpreted BNPs in Kriyol in any event, viz. inherently mass nouns such as yagu ‘water’, karni ‘meat’, and so forth. We are thus led to postulate two different structures for BNPs, one with a Number projection, the other without, that systematically collapse on the surface.

 Alain Kihm

Actually, this is not so bad in itself. It is a fact that Kriyol BNPs have two interpretations, individual and massive, and we need to ensure they are properly assigned. The obvious way to achieve this is to posit two different syntactic structures to be unambiguously interfaced with the corresponding semantic representations. Indeed, assuming one syntactic structure paired with two semantic representations would raise an interface problem. How, for instance, could we make sure that “mass” is never associated with the structure when it appears in subject position, even though the head noun is eligible for mass interpretation in itself, i.e. not taking into account the aspect of the predicate or the fact that it is a subject (cf. Kabra na kuri na fera ‘The goat is running in the market’, not *‘Goats [number irrelevant] are running…’, and see Footnote 25)? Yet, since the structural distinction we want to posit has no overt reflection, it seems mandatory we should be able to constrain it in some way. This, note, is a standard methodological point, raised to prevent uncontrolled burgeoning of structures for the same overt object. To this end, I propose to conform to the following principle: distinct structures may be assigned to the same phonological form provided (a) they involve no element the necessity of which is not established elsewhere; (b) no element is entirely virtual. This principle excludes the null D hypothesis a priori, independently from the empirical arguments given above. Indeed, null D is only required to tell apart [+R, +D, –Num] BNPs (kabra ‘the goat’) from mass ones (kabra ‘(a mass of) goats’). It is needed neither for un kabra ‘a goat’, since un has no covert equivalent with the same semantics in the Kriyol grammar under study, nor for kabras ‘(the) goats’ where, as we have seen and will further discuss below, the plural morpheme does office to ensure the interpretation. Moreover, there is no case in the language where definite D is not null, which makes it entirely virtual.28 Let me now propose an analysis that does not run afoul of the principle, while accounting for the data.

.

Deriving mass and individual interpretations

We must answer two questions: (a) Under which conditions are object BNPs interpreted individually or massively? (b) Why can subject BNPs not be interpreted massively, but may only refer to individuals or kinds, the latter a special case of the former, as suggested above? The hypothesis I will now attempt to support is the following: BNPs are interpreted massively if and only if they may (and must) incorporate to the verb of which they are an (internal) argument, in a special sense of the notion “incorporation”.

. It is crucial here that kil ‘that’, even when used with anaphoric force, maintains demonstrative (deictic) value, which makes it semantically different from a determiner such as English the or Portuguese o/a(s) whose sole function is to signal identifiability, even though we may choose to generate it in D0 position for the sake of parsimony.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

The Kriyol lexicon, I will assume, includes a functional feature Number, valued as 〈+〉, 〈–〉, or 〈u〉. [+Number] has /-s/ as a vocabulary exponent; [−Number] has no exponent; [uNumber] is invisible, that is to say it is merely a notation for absence of Number as distinct from [−Number].29 Number projects into syntax as the head of a NumP projection dominating NP (see [3]). On the other hand, we know that pluralized nouns are always referential (feature (iii) of Section 1), while non-pluralized nouns may be referential. Given previous assumptions, non-pluralized nouns stand for two very distinct entities: nouns associated with [−Number], i.e. singular count nouns, or nouns not associated with Number or, a notational variant, associated with [uNumber], i.e. mass nouns. Feature (iii), I propose, is related to a basic property of Number in Kriyol, namely that it is semantically “real” in the sense that marking it is never only a morphological requirement as it may be in English or in Portuguese. In these languages, there are, strictly speaking, no bare nouns since absence of plural marking (goat, cabra) is interpreted as covert [−Number] marking, i.e. singularity. It entails, as already noted, that nouns are always marked for Number, be it 〈+〉 or 〈–〉, even when the feature is semantically void (e.g. kind names as in ‘The goat/goats is/are a mammal/mammals’) or irrelevant (the pestle in Example [7]). In Kriyol, in contrast, genuine bare nouns exist, with no value at all for Number.30 A correlate of this state of affairs is that nouns in the scope of numerals or plural quantifiers are not marked for plurality (cf. sinku kabra(*s) ‘five goats’, manga di kabra(*s) ‘many goats’).31

. I am assuming a Distributed Morphology (DM) view of the lexicon and the organization of grammar. In a (very tight) nutshell, DM assumes the lexicon to be a list of features or feature bundles which are inserted into syntax as heads. The role of morphology is to linearize the heads and to associate them with exponents having a phonological form. For a short and clear exposition of the model, see Marantz 1997. Also see Noyer 1992; Halle & Marantz 1993; Embick & Noyer 1999. . Even inherently mass nouns such as water / água are morphologically singular. Part of the point I am making is that Kriyol yagu ‘water’ is neither singular nor plural. . I previously assumed there is a morphosyntactic reason for this, namely that numerals and plural quantifiers are lexical spell-outs of Num0 which do not require movement of N0. Here, I am pointing to the semantic correlate of that assumption which is that, since sinku kabra and like expressions are not necessarily referential (cf. Sinku kabra ka na ciga ‘Five goats [any goats] won’t be enough’), we may assume that nouns in the scope of numerals or quantifiers are interpreted as either referential indefinite or mass at their level. If true, Kriyol would count as counterevidence to Chierchia’s (1998) claim that languages where all nouns may be mass necessarily use classifiers in quantified expressions as in Chinese. Also note that the only way to express something like ‘the five goats’ is by using the deictic/anaphoric demonstratives mentioned above, which do not inflect for Number (es/kil sinku kabra ‘these/ those five goats’).

 Alain Kihm

Feature (iii) can therefore be generalized: noun phrases associated with Number, 〈+〉 or 〈–〉, are referential. And that is so because Number is a substantial category in Kriyol: it always implies countability, and countability implies reference to individual tokens. The syntactic structures of kabra meaning ‘the goat’ and of kabras ‘(the) goats’ are then as in (12) and (13) respectively: (12) (13)

[NumP [Num0 〈–〉][NP [N0 kabra]]] [NumP [Num0 〈+〉][NP [N0 kabra]]]

The functional head Num0 attracts N0 that adjoins to its left (see Kayne 1994). Num0 〈–〉 has no exponent in morphology; Num0 〈+〉 is associated with /s/ that morphologically merges with kabra. (Suffixhood need not be stated, since /s/ necessarily follows kabra given left adjunction of the head noun.) There is thus no need for a generalized DP level. It is not required for referentiality, since referentiality is a property of Number. Neither is it needed for definiteness: (12) may be considered definite by default, since indefinite referential non-plural noun phrases are obligatorily headed by un (un kabra), the only case where DP need be projected (apart from noun phrases including the demonstratives es and kil ). As for (13), it is genuinely ambiguous between two readings: ‘the goats’ and ‘(some) goats’. What these readings have in common is referentiality, and that comes from Number, as we assumed. Definiteness, on the other hand, is primarily informational as it rests on the speaker’s current assumption about identifiability of the referent by the hearer. Ambiguity in this respect for the hearer – since the speaker may be trusted to know what she means – is therefore no problem for the syntax. To put it differently, absence of a morpheme devoted to definiteness expression such as English the is something grammars can put up with at no great cost, provided the linguistically important category referentiality can be retrieved in some way. Summing up, my proposal is that referential, i.e. non-mass, noun phrases, singular or plural, definite or indefinite, are NumPs in Kriyol. Strictly speaking, then, they are no BNPs; only mass nouns are BNPs. Before we proceed to examine what this implies, however, a complication has to be mentioned. It stems from the fact that nouns referring to inanimates, i.e. strongly abstractable entities, tend not to be pluralized unless they are very clearly topical. Take, e.g. the sentence Minjer kumpra pratu ‘The woman bought /dish/’. Given an appropriate context, it may mean ‘The woman bought the dish (she was looking for)’. But that is a rather special meaning. In more run-of-the-mill contexts, and given that producing pratus ‘dishes’ is resisted in the variety we are studying,32 it is actually impossible for the hearer to decide whether the woman bought a specific, although unstated and

. It is produced, albeit with much variation, in so-called “decreolized” varieties. As mentioned at the very beginning of this study, Kriyol as a variety is of course an idealization, for which I do not apologize.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

perhaps unknown, number of dishes, or whether she engaged in what may be described, anticipating on arguments yet to come, as dish-buying. Again, the issue may be settled given a particular context, for instance if the sentence answers a question such as Ke ku minjer ta fasi ba na kil tempu? ‘What did the woman do at that time?’ – I ta kumpra pratu ‘She bought dishes’, meaning she traded in the stuff.33 Note the aspect of the predicate, here Imperfective, is also relevant. Interestingly, however, while there are means to disambiguate in the direction of mass interpretation, I cannot think of a way to bring to the fore the referential indefinite plural meaning ‘a certain number of dishes’ pratus could convey (in addition to ‘the dishes’) if it were acceptable with this value.34 On the other hand, as indicated, pratus is certainly generatable provided it is topical, as in Anta bu oja pratus? /Q 2sg see dishes/ ‘Did you see the dishes (we’re supposed to have in this kitchen)?’ – Pratus sta na mesa ‘The dishes are on the table’, where pratus refers to a particular set of dishes that both speaker and hearer are supposed to have present in their minds. But again, pratus does not seem to be interpretable in the referential indefinite mode, meaning ‘a certain number of dishes’.35 We thus observe a kind of complementarity between humans and inanimates as far as the distribution of ambiguities is concerned. With the former, ambiguity is about the definiteness value of referential plural expressions, since Minjeris bin na fera is translatable as ‘The women came to the market’ or ‘Women came to the market’ (As mulheres vieram no mercado or Mulheres vieram no mercado), and similarly N oja minjeris na fera may mean ‘I saw the women in the market’ or ‘I saw women in the market’ (Vi as mulheres no mercado or Vi mulheres no mercado). Mass interpretation is never available, which means that inabstractability is the key property, completely overruling the syntactic-informational factor, since position within or outside the predicate appears to make no difference for ambiguity or mass reading exclusion. The first is expected, since referentiality rather than definiteness is the crucial feature; the second non-effect is more interesting, as it shows the potency of the inabstractibility (or individuality) property that characterizes humans. With inanimates, on the other hand, what is ambiguous is whether non-pluralized expressions are interpreted as referential definite (Minjer kumpra pratu ‘The woman bought the dish’) or as mass (Minjer kumpra pratu ‘The woman bought dishes’). Here, however, the syntactic informational factor is not disconnected: non-pluralized subjects are not interpreted as mass, so that Pratu sta na mesa ‘The dish is on the table’ cannot mean ‘There are dishes on the table’, which is expressed as I ten pratu na mesa /it have dish on table/ (cf. colloquial Portuguese Tem pratos na mesa) or, since ten is a . The sentence is then said to be “all focus” (see Zubizarreta 1998). . Uns pratu is of course possible with this meaning, but that is a different construct where pratu cannot be pluralized. . Cf. French Est-ce que tu as vu des assiettes? which Kriyol would express as Anta bu oja pratu, to which the usual answer is Pratu ten na mesa ‘There are dishes on the table’ (see below).

 Alain Kihm

raising verb (see Kihm 1994), Pratu ten na mesa.36 Abstractability, which characterizes inanimates, thus appears to be a less potent property than its opposite, since it does not overrule topicality, but is subordinated to it. It remains indirectly active, however, insofar as referential indefinite plural interpretation is not available in subject position (see above), obviously as an upshot of its systematic conflation with mass reading. Individualizable animals such as goats are somehow intermediate. With them as well abstractability is overruled by topicality. But non-topics, i.e. nouns in object position, are not systematically skewed toward mass interpretation, because referential indefinite plurals are admissible in that position: Minjer kumpra kabras may mean ‘The woman bought (some) goats’ in addition to ‘The woman bought the goats’. And I will assume, although facts are not crystal clear, that kabras in subject position may, like human denoting nouns, but unlike inanimates, be understood with a referential indefinite value as ‘(some) goats’. This would be related to what we may call the “weak” abstractability of noun phrases denoting individualizable animals. Let us take stock of this rather complex array of facts. Subjects are never interpreted as mass, because of inabstractability when humans are denoted, which excludes it in any event; because of topicality overruling abstractability when animals or inanimates are denoted. As a consequence of these factors plus weak abstractability of individualizable animals, plural human and animal subjects are ambiguous between a [+R, +Def] (‘the Ns’) and a [+R, –Def] (‘a certain number of Ns’) reading. Inanimate plural and nonplural subjects (not modified by un), in contrast, can only be [+R, +Def] because of the indirect effect of (strong) abstractability. As for objects, on the other hand, a tabular presentation may help clarify matters (V = verb): (14)

V minjer ‘the woman’ V minjeris ‘the/some women’; *‘woman[mass]’ V kabra ‘the goat’, ‘goat[mass]’ V kabras ‘the/a certain number of goats’ V pratu ‘the dish’, ‘dish[mass]’ V pratus ‘the dishes’; *‘a certain number of dishes’

All in all, mass is inexpressible with humans; referential indefinite plural is inexpressible with inanimates and mass is restricted to object position; individualizable animals accept all interpretations in object position, and they also exclude mass in subject position. Inabstractibility explains the first state of affairs. What we are still in need of is a deep rationale for the other two. . Mesa ‘table’ is also potentially ambiguous. It is usually interpreted as ‘the (identifiable) table’, but it may be mass if the sentence asserts that, generally speaking, dishes are to be found on tables, a rare proposition to be sure. This is an area, let it be said in passing, where it may prove difficult to draw a neat line between mass and kind. Yet, it seems safe to stick to the notion that the latter ought to be restricted to sentences that assert something about the kind itself, as in definitions, rather than about collections of tokens of the kind, even if they amount to totality. In English, for instance, there is a clear distinction between “The dish/table is a material object” and “The dish is on the table” which cannot be used as a statement at kind level.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

.

Lexical incorporation

Let us look at the structure of Minjer kumpra kabra interpreted as ‘The woman bought the goat’. According to current assumptions (see, e.g. Collins 1998; Arad 1999) and given the present analysis of Kriyol noun phrases, the ‘lexical’ part of the structure is: (15)

[vP minjer [v0 v][VP [V0 kumpra][NumP kabra]]]

From (13) we derive (14) by raising V0 to v0 and NumP to Spec VP: (16)

[vP minjer [v0 kumpra+v][VP [NumP kabra][V0 t] t]]

Consider now the assumption that unergative verbs such as sleep actually are transitive verbs with a cognate object which is either lexically incorporated or realized in some cases, as in Haitian domi yon ti domi ‘to take a nap’, lit. ‘to sleep a little sleep’ (see Hale & Keyser 1993). The basic structure of Minjer durmi ‘The woman slept’ is thus: (17)

[vP minjer [v0 v][VP [V0 durmi][NP ]]]

where  is the head of the virtual cognate object that incorporates to durmi, forming a complex head with it, which raises to v : (18)

[vP minjer [v0 durmi.+v][VP [V0 t][NP t]]]

The point is that (18) looks very much like the structure we wish to assign Minjer kumpra kabra in the nonreferential interpretation of ‘The woman bought goats’, namely: (19)

[vP minjer [v0 v][VP [V0 kumpra][NP kabra]]]

It is indeed tempting to suppose that (19), just like (18), involves incorporating the object to the verb. This tallies well with the insight that kumpra kabra, massively interpreted, is a kind of intransitive predicate (‘goat-buying’ similar to ‘deer-hunting’). Moreover, it would straightforwardly explain why subject BNPs do not receive mass interpretation, as it is a general property of subjects that they do not incorporate (see Mithun 1986; Baker 1988: 81ff.). Two things must be done, then. First, we have to show that incorporation not only is possible, but is necessary. Secondly, we need an account of why the incorporated object still appears to the right of the verb, suggesting it is not syntactically adjoined to it (see Kayne 1994 for why adjunction is to the left), and the assumed incorporation is not Baker’s type resulting from head movement in syntax. Concerning the first point, BNPs in the strict sense, i.e. nonpluralized NPs, have to be considered defective arguments in the sense of Chomsky (1999) since they bear no referential index or number feature. In English, defective arguments show up in idioms such as give hell. They cannot be extracted (cf. *hell he gave to his colleagues) unless they are changed to overt DPs with variable felicity (cf. ?The hell he gave to

 Alain Kihm

his colleagues (surprised everybody)).37 Such DPs seem to be endowed with what one might call ‘pseudo-referentiality’. Similarly in Kriyol, kabra cannot be mass, but must be referential in Kabra ku minjer kumpra (bonitu) ‘The goat the woman bought (is fine)’. Extraction of inanimates yields the same result: pratu ku minjer kumpra only means ‘the dish the woman bought’. Even inherently mass nouns undergo the same type shift in this context: cf. Karni ku minjer kumpra ka bon ‘The meat the woman bought wasn’t good’. The fact that kabra with mass interpretation may be clefted as a potentially contrastive focus – cf. I kabra ku minjer kumpra ‘It’s goats that the woman bought’ – does not come as counterevidence, since defective arguments can be focused, either in situ (cf. I was given HELL by my colleagues) or, not so felicitously in English perhaps, through clefting (cf. ?It’s hell I was given by my colleagues).38 Note that in the Kriyol clefting example kabra is moved to a predicate-internal position compatible with mass interpretation.39 What seems to follow from such evidence is that defective arguments are unable to satisfy the selectional properties of transitive verbs (or of Chomsky’s 1999 v*). Incorporation would then come as the only way to license the predicate by making it intransitive. How does incorporation proceed? Obviously, we are not dealing with syntactic incorporation in the sense of Baker (1988), as it necessarily yields NV order given our general assumptions about phrase structure and adjunction. Moreover, according to Baker (1996), syntactic incorporation is an effect of the Polysynthesis Parameter, which Kriyol shows no sign of sharing. My proposal, then, is that we are dealing with lexical incorporation, ‘before’ mapping to syntax. Let us assume that every verb (or every root that can merge with v) includes an internal argument slot in its lexical representation. If that slot is free in the initial syntactic array, it must be filled with a syntactically legitimate argument, viz. a NumP or a DP.

. According to the native speaker I asked, this sentence is grammatical and makes sense, but “We wouldn’t use it”. Examples like ‘Despite the short shrift which Magnus Magnusson has given to the critic’ are attested, however (see Cowie, Mackin & McCaig 1983: 230), and the same informant finds it “less jarring to the ear”, although a bit odd. Here again short shrift has to be turned into a full DP. . In French the result of clefting defective arguments is usually perfect: cf. C’est raison qu’il m’a donnée (pas tort) ‘He sided WITH me (not AGAINST me)’ (donner raison à ‘to side with’, to be squarely distinguished from non-idiomatic donner la raison ‘give the reason (why)’). . Two analyses are a priori possible for I kabra ku…: either i is a 3sg expletive subject pronoun and a null copula is assumed; or i is a copula, homophonous with i ‘s/he/it’, and a null expletive subject is assumed. In either case kabra is in the predicate. For independent reasons, I think the first analysis is the right one for Kriyol (see Kihm 1994).

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

This gives us a transitive or unaccusative structure, depending on whether VP merges with v* or not. But the internal argument slot may also be filled already in the lexicon, with a cognate object (a copy of the root) as with sleep, or with an independent root. In both cases we get an intransitive unergative structure. Minjer kumpra kabra with nonreferential, mass interpretation of kabra should thus be represented as follows: (20)

[vP minjer [v0 v][VP [V0 kumpra [NP kabra]]]]

where [v0 kumpra [NP kabra]] comes ‘as such’ from the lexicon. Order is not a problem: since kumpra kabra is a regular VP once mapped into syntax, there is no choice for the complement but to appear in its canonical position, that is to the right of its head, Kriyol being a strictly SVO language.40 It can be moved for the same reason, provided it is to a position where mass interpretation remains an option, such as the higher predicate of a cleft construction (see above). Here the fact that Kriyol has no determiner is crucial. Take English give hell formed, we assume, by the same process of lexical incorporation as kumpra kabra in (20) (except for the process being limited to idioms in English). Extracting hell to relativize it, for instance, compels us to insert a determiner (?the hell he gave…). Because of this, ‘hell’ seems to become referential, the idiomatic sense is lost and it cannot be replaced by a sensible literal meaning, hence the oddness. The fact that short shrift fares better (see Footnote 37) may be due to its having no meaning (what is a ‘short shrift’?). Therefore, no literal interpretation at all, sensible or not, can be given to the short shrift he gave, and speakers are free (or at least freer) to treat short shrift as a kind of pseudo-noun meaning what the idiom means. Note further that the hell he caused sounds much better than ?the hell he gave. That is probably because the meaning of the verb cause allows one to interpret the hell as synonymous with, say, the trouble much more easily than when the verb is give.41 In Kriyol, none of these problems arises, because there is no determiner to assign pseudoreferentiality to the moved BNP. The only constraint is that the landing site should not be one that disallows mass interpretation. Note that linear adjacency of the verb and the complement is not strictly required, as long as the separating element does not stand in the way of lexical composition. In fact, two and only two kinds of elements may come between a verb and its massively interpreted object. One is the Past (or Anterior) morpheme ba as in Minjer kumpra ba kabra na fera ‘The woman had bought goats in the market’. Now V+ba is probably the result of inflectional morphology in present-day Kriyol, so the construction can be

. Of course the result is ambiguity since there is no clue in the sentence of whether lexical incorporation has taken place or not. The fact is, as repeatedly mentioned, that such sentences are ambiguous out of context from the hearer’s viewpoint. Exactly the same is true of, e.g. I love cats which may mean either that I am a cat-lover or that I love some cats, but not all. . Clefting does not force determiner insertion (cf. It’s hell he gave…), so the result should be better, even if not perfect.

 Alain Kihm

explained away by moving the verb to T after syntactic mapping.42 The other types of element are dative arguments, either clitic pronouns or full noun phrases: cf. Minjer kumpra-n kabra ‘The woman bought goats for me’; Minjer kumpra si amigu kabra ‘The woman bought goats for her friend’. Insofar as there is a position in Lexical Relational Structure for these arguments, it is governed by a different verb node in the verb shell than the one corresponding to kumpra (see Hale & Keyser 1993). Therefore, it does not prevent lexical adjacency of the latter with its direct object.

.

A comparative view and some concluding remarks

The difference between Kriyol, on one hand, and English and Portuguese, on the other hand, seems thus to reside in the unequal possibilities the languages have of exploiting lexical incorporation. Here, a look at French may be useful. In French there is a partitive preposition de whose function it is to overtly signal that a noun phrase is given mass interpretation provided its inherent semantics permits it: cf. boire du vin ‘to drink wine’ vs. boire le vin ‘to drink the wine’.43 As the example shows, de requires its complement to be a DP (du = de+le – cf. Kupferman 2000).44 As a consequence, lexical incorporation is precluded. It is only exploited to form idioms such as donner raison ‘to side with’, faire bonne figure ‘to put up a good show’, etc. Another consequence is that des N phrases where N denotes a non intrinsically massive entity (e.g. chèvres ‘goats’) are ambiguous: Hier j’ai acheté des chèvres may be understood as (i) ‘Yesterday I bought some goats’ (referential, individual), or (ii) ‘Yesterday I made an act of goat-buying’ (pseudo-English, nonreferential, mass). Notice that ‘ambiguity’ is itself an ambiguous term: either both interpretations are real, but not distinguished

. At a previous stage of the language ba was more like an adverbial and it could be separated from the verb by short complements, viz. bare nouns. (In modern Kriyol, only clitics can intervene between the verb and ba – cf. I kumpra-l ba ‘S/he had bought it’.) We can therefore suppose that Omi kumpra kabra ba was then the preferred construction when kabra was mass. This is hard to test, though. . Note that le vin may receive individual interpretation (‘the wine you bought’) or be interpreted as referring to the kind, as in Elle ne boit pas le vin, seulement la bière ‘She does not drink wine, only beer’. This again shows that individual and kind go together and contrast with mass. . This was not so in older French, perhaps as late as the beginning of the 20th century, where expressions such as J’ai de bon tabac ‘I have good tobacco’ were grammatical (Modern French, J’ai du bon tabac). In negative predicates, de N is still grammatical in free variation with du/de la N: cf. Elle ne boit pas de/du vin ‘She does not drink wine’, even with non intrinsically massive nouns: cf. Elle n’achète pas de/des chèvres ‘She does not buy goats’. The explanation may be that negation tends to impose nonreferentiality on the object. Note that Elle n’achète pas les chèvres may mean that goats, as a kind, are such that she does not buy them.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol) 

formally; or the choice is merely virtual, and French (or English, it seems) does not actually make the distinction. The latter seems to me to be more likely. In English and Portuguese, which seem to pattern alike, there is no partitive morpheme like French de. The crucial factor is then that noun phrases denoting entities the cardinality of which is superior to one must be pluralized, irrespective of whether number as such is relevant or not in the current utterance. When a noun denoting a not inherently massive entity (e.g. cabra ‘goat’) is interpreted as mass (i.e. abstracting from individuals), its referent’s cardinality is necessarily greater than one, hence comprar cabras / to buy goats, with cabras a NumP, and the same real or virtual ambiguity as in French. The meaning of the noun and the variable abstractability it implies may then help decide, so we may assume that comprar cerejas ‘to buy cherries’ is distinct from comprar cabras, which is probably distinct from comprar deputados ‘to buy off MPs’. Another thing to be noted is that, in theory, a mass of goats (or of anything) has one as its lower bound – pace the Sorites paradox!45 This is why the array of situations properly described in Kriyol by Minjer kumpra kabra, with kabra non-referential, includes the situation where the woman bought only one goat. Yet, this is no more than a logical implication. In the Kriyol variety we are studying where un is not optional, the acceptable way to describe a situation where only one goat was bought is to say Minjer kumpra un kabra. In Portuguese, in contrast, the logical implication can never be felt because the ‘greater than one’ property of the referent will obligatorily trigger overt pluralization. Only inherently mass nouns like vinho ‘wine’ may ‘stop’ at the NP level – from which it follows that absence of Number marking is interpreted as no count rather than singular (non-plural) – and lexically incorporate. The crucial factors for Kriyol are therefore the absence of a determiner and the fact that overtly plural NPs must be referential. Therefore, BNPs are liable to two analyses: either they are massively interpreted NPs, or they are NumPs whose head has no morphological exponent, being then interpreted as nonplural referential expressions. Comparing with Cape Verdean is revealing in this respect: Cape Verdean does have a definite determiner (kel/kes), and it seems that mass interpretation of non inherently mass nouns is only possible in idioms as in English, Portuguese, and French (see Baptista 1997). Both analyses / readings are always available for nouns in object position. Only the latter is for noun in subject position, because subjects must be syntactically merged and cannot incorporate. Thanks to these properties, Kriyol speakers have the option of not expressing Number, hence of conceiving entities as masses, whenever individuality and countability of the referents are not actually relevant, or are unknown, or even unknowable. . According to this revered sophistry, a pile of sand does not exist because, if you start removing one grain at a time, there is no definite point in time where you can say there is no longer a pile. How many grains does it take to make a pile? Of course, this is a play on words. The difficulty vanishes if we define a pile as an atomic semi-lattice having individual grains as its atoms.

 Alain Kihm

Finally, the crucial factor around which all this complex state of affairs organizes seems clearly to be the fact that plurality in Kriyol always is semantically real, in the sense laid out above, whereas its marking may be a mere morphological requirement in Portuguese and English. Why should that be so? As a concluding remark, for an elaboration of which I address the reader to other works of mine (see Kihm 2005; to appear), I would like to suggest that the ultimate reason probably is to be found in something that can properly be described as an accident, namely the disappearance of gender distinctions, which had the upshot that plural marking ceased to be an obligatory morphological process.

References Arad, M. 1999. On ‘little v’. In Papers on Morphology and Syntax, Cycle One [MIT Working Papers in Linguistics 33], K. Arregi et al. (eds), 1–25. Baker, M. C. 1988. Incorporation: A theory of grammatical function changing. Chicago IL: University of Chicago Press. Baker, M. C. 1996. The Polysynthesis Parameter. Oxford: OUP. Baptista, M. 1997. The Morphosyntax of Nominal and Verbal Categories in Cape Verdean Creole. PhD Dissertation, Harvard University. Baptista, M. 2002. The Syntax of Cape Verdean Creole: The Sotavento varieties. Amsterdam: John Benjamins. Beard. 1995. Lexeme-Morpheme Base Morphology. Albany NY: State University of New York Press. Bickerton, D. 1998. Catastrophic evolution: The case for a single step from protolanguage to full human language. In Approaches to the Evolution of Language, J. R. Hurford, M. StuddertKennedy & C. Knight (eds), 341–358. Cambridge: CUP. Chierchia, G. 1998. Plurality of mass nouns and the notion of ‘semantic parameter’. In Events and Grammar, S. Rothstein (ed.), 53–103. Dordrecht: Kluwer. Chomsky, N. 1999. Derivation by phase. MIT, Ms. Collins, C. 1998. Local Economy. Cambridge MA: The MIT Press. Cowie, A. P., Mackin, R. & McCaig, I. R. 1983. Oxford Dictionary of Current Idiomatic English. Oxford: OUP. Embick, D. & Noyer, R. 1999. Locality in post-syntactic operations. In Papers on Morphology and Syntax, Cycle Two [MIT Working Papers in Linguistics 34], V. Lin et al. (eds), 265–317. Hale, K. & Keyser, S. J. 1993. On argument structure and the lexical expression of syntactic relations. In The View from Building 20: Essays in linguistics in honor of Sylvain Bromberger, K. Hale & S. J. Keyser (eds), 53–109. Cambridge MA: The MIT Press. Halle, M. & Marantz, A. 1993. Distributed morphology and the pieces of inflection. In The View from Building 20: Essays in linguistics in honor of Sylvain Bromberger, K. Hale & S. J. Keyser (eds), 111–76. Cambridge MA: The MIT Press. Kayne, R. S. 1994. The Antisymmetry of Syntax. Cambridge MA: The MIT Press. Kihm, A. 1994. Kriyol Syntax: The Portuguese-based Creole language of Guinea-Bissau. Amsterdam: John Benjamins.

Chapter 5. Guinea-Bissau Portuguese Creole (Kriyol)  Kihm, A. 2000a. Are creole languages perfect languages? In Language Change and Language Contact in Pidgins and Creoles, J. McWhorter (ed.), 163–199. Amsterdam: John Benjamins. Kihm, A. 2000b. L’adjectif en portugais et en kriyoe: Essai de syntaxe comparée. Langages 138: 49–60. Kihm, A. 2005. Noun classes, gender, and the lexicon-syntax-morphology interfaces. In Handbook of Comparative Syntax, G. Cinque & R. Kayne (eds), 459–512. Oxford: OUP. Kihm, A. To appear. Le marquage de la définitude et du nombre dans les créoles de base française et portugaise. In Actes de la journée Conscila, D. Véronique (ed). Krifka, M. et al. 1995. Genericity: An introduction. In The Generic Book, G. N. Carlson & F. J. Pelletier (eds), 1–124. Chicago IL: The University of Chicago Press. Kupferman, L. 2000. Pour une analyse des groupes quantifiants. International Conference on the Syntax and Pragma-semantics of Noun Phrases, Anvers, 10–12 février 2000. Lyons, C. 1999. Definiteness. Cambridge: CUP. Marantz, A. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. In Proceedings of the 21st Penn Linguistics Colloquium [UPenn Working Papers in Linguistics, Philadelphia], A. Dimitriadis et al. (eds), 201–225. Mithun, M. 1986. On the nature of noun incorporation. Language 62: 32–38. Noyer, R. 1992. Features, Positions and Affixes in Autonomous Morphological Structure. PhD Dissertation, MIT. Vallduví, E. 1995. Structural properties of information packaging in Catalan. In Discourse Configurational Languages, K. É. Kiss (ed.), 122–152. Oxford: OUP. Zubizarreta, M.-L. 1998. Prosody, Focus, and Word Order. Cambridge MA: The MIT Press.

 

Spanish-lexified Creoles

 

Some aspects of NPs in Mindanao Chabacano Structural and historical considerations1 Anthony P. Grant This paper examines certain aspects of NP structure and use in Mindanao Chabacano (henceforth MC). This is a mixed creole of the Southern Philippines in which the bulk of the morphs derive from the Mexican Spanish which was the major language of intercommunication among the soldiers of the Philippines when these islands were under Spanish rule (soldiers having been mostly recruited from Mexico). Only two forms of MC are attested in any great detail. Firstly, there is Zamboangueño, the major language of the inhabitants of Zamboanga City and its surroundings. Secondly, there is the endangered Cotabateño, a language that is known to a minority of the inhabitants of Cotabato City in a creole community which linguistically is largely an early 20th century offshoot of that of Zamboanga. NP structures in these languages are identical, most of my examples are taken from Zamboangueño because I have more data from that. It will be seen that bare NPs are present in MC but are not any more frequent than they are in Spanish or in the Philippine languages among which MC was formed.

. Introduction Mindanao Chabacano is a name which has been given by linguists to the varieties of creolised Spanish which are spoken as a first language by a sizeable proportion of the population of Zamboanga City and the surrounding parts of the province of Zamboanga del Sur, in southwestern Mindanao, Philippines, and by a dwindling minority of . I would like to thank Marlyse Baptista and Jacqueline Guéron for the invitation to contribute to this volume and for comments on this paper, and the native speaker Joy Granados (and another native speaker with whom I worked and who wishes to remain anonymous), and the linguists Philip Baker, Clancy Clements, John Green, John Lipski, Salvatore Santoro and Deborah Sutherland for providing me with much of the Chabacano material which has been used in this work. I would also like to thank Paz Buenaventura Naylor for much material and many illuminating conversations on Tagalog and other Philippine languages. The spelling of the Chabacano forms conforms to Philippine orthographical conventions; primary stress is marked where the sources cited show it.

 Anthony P. Grant

aged speakers in Cotabato City, further east on Mindanao. It is also spoken by an expatriate community of people in the village of Kampong Air (Malay for ‘Water Village’), near Semporna, Sabah, Malaysia. A form of creolised Spanish was (or is?) also spoken by a few of the inhabitants of Davao City in southern Mindanao; this has not been described in any detail and its historical and structural relation to other varieties of MC is presently unknown. There are other speech varieties which are also known as Chabacano or, as more frequently spelt in the Philippines, as Chavacano (the word means ‘bungling, vulgar, uncouth’ in Spanish), but these are not spoken in Mindanao. (Nonetheless, all varieties of Chabacano are often classified together by some linguists, and they are held as constituting the forms of ‘Philippine Creole Spanish’.) The other forms of ‘Philippine Creole Spanish’ (a label which is unknown to most Filipinos and which is unsupported historically, since these creoles are not all descendants of a single ancestral creole) are spoken in dwindling communities in the Tagalog-speaking parts of southern Luzon, around Manila Bay. The speakers of these forms of Chabacano are bilingual (and increasingly dominant) in Tagalog. The languages in question are known to linguists as the Manila Bay Creoles. They consist of Caviteño, which is spoken by a minority of the inhabitants of Cavite City, near Manila, Ermiteño, which was heavily hispanised and was formerly spoken in the Manila suburb of Ermita but which is now extinct, and Ternateño, which is spoken in the Manila Bay town of Ternate and which shows evidence of having originated separately from all other forms of Philippine Creole Spanish including MC.2 To the best of my knowledge, no Manila Bay Creole is being passed onto children. These Manila Bay Creoles are alluded to on occasion but their NP structures are not dealt with in detail in this paper, for want of sufficient material on bare nouns and other types of nouns in these languages. But what little data I have on these varieties (Whinnom 1956, Llamado 1972) does not suggest serious structural differences between these and MC in these matters. Bare nouns, are rather infrequent in MC materials, and are even less common in the Tagalog data from which I will be drawing comparative parallels. Also, some MC nouns which look like bare nouns at first glance actually involve the use of markers which do not appear in the surface form of the sentence but which are nonetheless present in the underlying structure. This is notably true of the semantics of certain kinds of prepositional phrases as found in both Tagalog and MC, which involve na in MC and which I discuss in Section 4.

. This language and the community where it is spoken are named for the Moluccan island of Ternate from where the founders of this Philippine creole community are supposed to have fled in the 1650s. This language shows some features, such as the categorical change of earlier Spanish 〈ll〉 to /y/, and the possession of a few dozen lexical items which derive from languages with which the ancestors of the speech community were once in contact: Portuguese, Moluccan Malay and the non-Austronesian language Bahasa Ternate. These are not found in any other form of Philippine Creole Spanish.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

The varieties of MC are creoles that have remained in contact with the Philippine languages, which provided many features of their structures (and many items of their lexicon). The Manila Bay Creoles were only influenced in the Philippines by Tagalog. But the creole spoken in Zamboanga was influenced firstly by Tagalog (which was also present in the Manila Bay Creole component of MC3), then most decisively by the Western Bisayan language Hiligaynon. Later on, it also borrowed a certain amount of cultural lexicon4 from Tausug (the native language of the islands of the Sulu Archipelago to the immediate west of Zamboanga and an important regional lingua franca), by the Samalan languages Sinama and Yakan, which were spoken natively in the Zamboanga area before the arrival of the Spanish garrison. Then at the beginning of the 20th century MC was influenced by Cebuano (the language of many immigrants to Mindanao) and then, after Philippine independence in 1946, by Tagalog again, which was the medium of primary education and the official language. Philippine English also began to exert influence on the lexicon of MC. The MC variety of Cotabato contains all the elements that are found in the Zamboanga form of the creole, and items from the Mindanao Philippine languages Magindanaw and Tiruray too, languages local to Cotabato Province. Among creoles which have not also been predominantly pidgins (there are as many L2 speakers of Zamboangueño as L1 speakers, but they speak it as a non-pidginised language) this continuation of a creole’s contact with substrate languages is an unusual linguistic backdrop.

.

Documentation on MC

The main descriptive source for Zamboangueño, the Mindanao Creole of Zamboanga, is Forman (1972), a text collection with extensive grammatical and lexical analysis which regrettably still remains unpublished. The major source for Cotabateño data is Riego de Dios (1989). This is the work of a native speaker of this variety and a volume which also includes copious comparative lexical data on Zamboangueño and on the two Manila Bay Creoles which are still extant. McKaughan (1954) is an early text collection with some insightful grammatical notes, collected by a specialist in

. Although we cannot be certain, it is probable that not all the early tagalisms in MC are shared with Manila Bay creoles, however. MC hindi’ ‘not’ is closer to Tagalog hindi’ than to Hiligaynon ‘indi’ or Cebuano dili’, and it is not found in Manila Bay Creoles. . For instance, terms relating to Islam in MC derive from Tausug, where they are often Arabic loans mediated through Brunei Malay or written Malay. But since MC and its speakers were identified with intrusive Catholicism, and since Tausug was identified with pre-existing Islam, there was never an incentive for MC speakers to shift in large numbers to Tausug, so that the impact of Tausug remained at a purely lexical level.

 Anthony P. Grant

southern Philippine languages from a well-educated emigrant speaker who was in her mid-twenties. Many sentences and textlets in the Zamboanga variety of MC are also provided in the discussion of the MC verbal system in Frake (1980). Furthermore, a short text in IPA phonemic transcription is available in Ing (1967), which is the work of a native Sino-Filipina speaker of Zamboanga MC who wrote an unpublished synchronic pre-generativist phonological study of the creole as her PhD dissertation at University College, London (Ing 1968). A recently published typological sketch of Zamboangueño with information on issues and the effects of language contact, originally written in Spanish, is Grant (2002). Santoro (to appear) presents much grammatical information, some of it gathered by Santoro from a native speaker living in New York. Santoro’s treatment is presented in the comparative framework which is used in other sketches in the volume in which Santoro’s work is to appear (Holm and Patrick, to appear). Two collections of papers on Chabacano, each of them with determiners containing numerous sentential examples, have recently appeared (Fernández 2001; do Couto 2002). Our first written record of MC dates only from 1929 (Broad 1929: 142), and is supposed to represent the impromptu speech of a six-year-old boy whose L1 is American English. Structurally the sparse material which is offered there (slightly under three lines of text and a few stray phrases which are illustrative of the uses of typical MC words that are not known to Spanish) is very similar to the MC material which Forman recorded over 40 years afterwards. Indeed the Broad (1929) text contains two instances of bare nouns, for which I have provided the modern forms in parentheses: (1, MC)

Habla mi nana quiere kame come carne speak my mother want we.incl eat meat guisao kun patatas. roasted with potatoes (now: abla mi nana kyere kame kome karne gisaw kun patatas) ‘My mother says we want to have roast meat and potatoes.’

(2, MC)

No bos ulbida compra pescao. (no bos ulbida kompra peskaw) no 2sg forget buy fish ‘Don’t forget to buy (some) fish.’

As far as I can tell, there are no discernible differences between the morphological or syntactic behaviour of NPs in the two varieties of MC, those of Cotabato City and Zamboanga City, which have heretofore been described. (And as I have said, this paper does not discuss features of NPs in the Manila Bay Creoles.) Neither do the rules of behaviour of NPs, including bare nouns, appear to have changed diachronically in the 73 years which have elapsed since the appearance of our first record of MC. A single linguistic description of NP structure and behavior fits all attested diachronic and diatopic varieties of MC.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

.

Bare nouns and other nouns in MC

In order to understand the nature of bare and other kinds of nouns in MC we should first look at the determiner system. The determiner system in MC is represented in Table 1 (occasional forms only found in frozen Spanish phrases are omitted; X denotes the position of the noun in the various noun phrases). The use of overt determiners in MC is not completely categorical and there are instances where zero occurs when a determiner (especially an indefinite singular one) would be expected.

Table 1. Determiners in MC Determiner

Singular

Plural

Definite subject

El X The X Kon el X OBJ the X Del X Of the X Na X At X Si X The.PERSON X Kon el X OBJECT.the X Del X Of.the X Na X At X Un X, X (rare) An.X, X (Kon) un X (OBJECT) an X De un X Of an X Na un X At an X

El manga X The PLURAL X Kon el manga X OBJ the PLURAL X Del manga X Of the PLURAL X Na manga X At PLURAL X El manga5 X The PLURAL X Kon el manga X OBJECT the PLURAL X Del manga X Of.the PLURAL X Na manga X At plural X Manga X PLURAL X Kon manga X OBJECT PLURAL X De manga X Of PLURAL X Na manga X At PLURAL X

Definite object Definite possessive Definite oblique Personal subject

name

Personal object

name

Personal possessive

name

Personal oblique

name

Indefinite subject Indefinite object Indefinite possessive Indefinite oblique

. In the case of family names, which are the personal names most likely to be pluralized, the normal definite determiner is used, either with or without manga. Alternatively a singular definite determiner is used together with the Spanish plural morph -(e)s: thus one would find el (manga) Amado or el Amados ‘the Amados, the members of the Amado family’ in MC (cf. Santoro, to appear). There is also Konda Amado ‘the Amados and their crowd.’

 Anthony P. Grant

A similar remark applies to the use of the Philippine-derived plural marker manga [ma’ŋah]. (All sentential examples in this paper are from MC unless otherwise indicated). (3, MC)

Ya-mirá sila Ø sine. -look 3pl Ø movie ‘They saw a movie.’

(Forman 1972: 199)

But on the whole, in cases where a determiner is specified within the semantics of an MC NP, a determiner will be used overtly. The determiners in MC which formally match those in Spanish in terms of their phonological form are the masculine singular ones, un and el. Neither feminine determiners nor plural determiners of either gender have passed into productive or systematic use as determiners in MC, a language which lacks grammatical gender or overt plural marking on determiners. For instance we see the use of an article with a Spanish masculine phonological form as the article for what is clearly a feminine gender noun phrase (in terms of natural sex gender) in sentence (24a) further down. They do, however, occur in a number of fixed expressions: (4, MC)

a-la-manyana at-the-morning ‘in the morning, the following morning’

(5, MC)

a las syete at the seven ‘at seven o’clock’ cf. Spanish a las siete

(6, MC)

lo mísmo the same ‘the same (one)’ cf. Spanish lo mismo.

(Forman 1972: 154)

The general model of the division of labour of MC determiners is provided by the determiners of the Philippine languages, and in terms of the categories expressed overtly, the MC system resembles that of the Philippine languages more than that of its chief lexifier Spanish. Both the personal name determiner si and the pluralizer manga derive from phonologically identical forms in Philippine languages, although the patterns of use of these particles are not identical in Tagalog (and other Central Philippine languages) and in MC. In both MC and the Central Philippine languages manga has a sense of ‘plurality of type’: Tagalog manga aso and (perhaps to a lesser extent) MC manga iru both carry the

. In MC manyana has the sense of ‘tomorrow’, since morning is expressed by the philippinism ’ága (Frake 1971). An ambiguity which exists in Spanish is thus mostly absent in MC.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

connotation that several different kinds of dog are being referred to at the same time. A MC sentence such as: (7, MC)

Ya-mira yo manga iru.  see 1  dog ‘I saw (some) dogs.’

which of course uses an NP which is not preceded by a determiner, gives the MC hearer the background idea that the speaker saw (for instance) Great Danes, Weimaraners, poodles, Pomeranians and chows, and it further suggests that the speaker saw them all at the same time and in the same place.7

.

On the interpretative variability of bare nouns in MC

The presence of determiners (including zero) at the beginning of an NP usually makes it clear in MC how a bare noun is to be interpreted, whether it is definite or indefinite and whether it is meant to be singular or plural. But some instances of bare nouns in MC exhibit both structural and semantic ambiguity and therefore permit several possible readings. Take for example this sentence from Forman (1972: 158): (8, MC)

Tyene yo myedo na bisikleta. have 1 fear  bicycle

At first glance this looks as though it means ‘I’m/I was scared of bicycles’, or ‘I’m/I was scared of a bicycle’, or ‘I’m/I was scared [scared, that is, of riding] on a bicycle’, or even ‘I’m/I was scared [of riding, or when I was riding] on bicycles’. However, although the overt plural marking of otherwise bare nouns is not categorical in MC, and although the above readings are at least theoretically possible and valid, overt plural marking is frequent and is indeed the norm in MC. The likeliest English interpretation of the above sentence, if construed literally, is actually: ‘I’m/I was scared of the bicycle’ (for instance, because it has no brake chain). However this makes little sense pragmatically.

. It is just about possible that Spanish -(e)s was formerly productive as a plural marker in Mindanao Chabacano, since there are a few words which seem to have both singular and plural forms distinguished by the use or non-use of this pluralizer. (These may actually be preserved archaisms found only in set phrases.) However, it is also notable that most of the Chabacano words which bear Spanish pluralizers but which do not belong to the basic stratum of the vocabulary (for instance names of body-parts which normally occur in pairs) are words which relate to items introduced to the Philippines and which also occur as Spanish-style plurals in Tagalog and other languages, though with no overt sense of the referents being intrinsically plural. There are also some basic items which satisfy this criterion too, for instance Chabacano, Tagalog plores ‘flower(s)’, which in Tagalog contends with native bulaklak.

 Anthony P. Grant

Precisely why this reading should even be possible has to do with the operation of other syntactic rules which affect other parts of the MC grammar, given that bisikleta lacks a definite determiner, although it has one in a sentence such as (9, MC)

Ya-mira yo kon el -see-1  . bisikleta. bicycle ‘I saw the bicycle.’

This is a matter which has to do with determiner-omission in certain kinds of prepositional phrases. This will be explained and, we hope, sourced in Section 3.3.

. Various kinds of bare nouns in MC: an introduction to the types The analysis of the determiner system in MC, which includes zero, proceeds along semantic lines. As the schema of determiners given in the above section shows, personal names behave differently from all other nouns (but not place-names, for instance: we cannot call upon a ‘common noun’ versus ‘proper noun’ distinction). As I point out in 3.5, place-names do not take determiners unless they are part of the name itself, such as Las Palmas, so that they could be regarded as a subcategory of bare nouns, even though they are not structurally very exciting or illuminating. The interest in bare noun behaviour centres on common nouns, both mass nouns (including abstract nouns) and count nouns. From my analysis of the MC data, the rule of thumb appears to be this: if the bare noun has a mass noun as its head, the sense of the bare noun will be singular and it will represent ‘part of the whole’ (‘some butter’, ‘some advice’, etc.). If the noun is a count noun, the sense of the bare noun will be indefinite and also non-specific, and it will be construed as being plural rather than singular, although the number significance of the NP will not be crucial to the sense of the clause (an instance of this is Example 8). It should be noted (for the purpose of resolving the matter early on in this discussion) that the determiners in MC can modify more than one head noun, so that some apparently bare nouns are only superficially so, whereas in fact they are introduced by the same determiner as the noun which occurs first in the NP sequence: (10, MC)

Ta-akabá sale apuera kay ta -finish exit outside because  hala silá el piedra i arena por-káwsa pull.out 3 the stone and sand for-cause del gránde mariháda. of.the big waves (Forman 1972: 28; the bold face is mine)

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

‘They all come out because the stones and sand pull (them) out on account of the great wave.’

There are certain pragmatic situations in which MC, Spanish, Tagalog and English all forgo the overt use of determiners even though NPs are involved. Notable among these situations are imprecations and exclamations such as: (11, MC)

Iho de puta! son of whore ‘Son of a whore!’

(Frake 1980: 284)

in which the bare noun has a vocative sense. None of these languages would use an article or determiner with either noun here.

. Bare nouns in copular construction In order to express the idea ‘X is Y’ in MC where Y is a noun (and where X may be a noun or a pronoun), one of two different constructions may be used. The first uses what traditional Philippine grammars call an ‘inverse marker’ particle which is generally construed as an untensed copula. The second simply involves phrasal parataxis with the order COMMENT-TOPIC, or Y-X. Let us assume for the purpose of the next few sentences that we are talking about a young woman called Araceli who is a schoolteacher. maestra8 schoolteacher

(12a, MC)

el dalaga amo Ø the maiden  Ø

(12b, MC)

Ø maestra el dalaga Ø schoolteacher the maiden

Both sentences mean ‘the young woman is a schoolteacher.’ Comment precedes Topic in such sentences both in Central Philippine languages and in MC. The same would be the case with the following: (12c, MC)

Si Araceli es maestra  Araceli is schoolteacher

(12d, MC)

maestra si Araceli. Schoolteacher  Araceli

Both sets of sentences are typical of modern MC, though (12a) reflects a newer usage.

. Maestra is one of a small set of nouns, all of them borrowed from Spanish, which vary according to gender; a male schoolteacher is maestro. The word dalaga is a loan from one or more of several Philippine languages: it is common to Tagalog and the Bisayan languages of the Visaya Islands, central Philippines, which include Hiligaynon and Cebuano.

 Anthony P. Grant

In neither of these cases would one use an indefinite determiner with a professionword such as maestra ‘schoolteacher’9 as one does in the English equivalent of this sentence. The MC copula amo has been regrammaticalised within MC in the course of the twentieth century from its original role as an emphatic particle (where it meant ‘that’s it!’) of Hiligaynon origin, where it had the same sense as a particle as it did in Hiligaynon. It has come into greater use in MC in the course of the 20th century, possibly under the influence of Tagalog (itself a language which, as the result of Spanish influence in older schools, increasingly uses an inverse marker rather than relying on zero copula for the purpose of signalling Comment-Topic) and English. Instances of MC zero-copula have therefore decreased. (This new copula is discussed in greater detail in Aoto 2002.) This restriction upon the use of indefinite markers affects equative sentences involving professions and nationalities, but at least nouns referring to professions are not always disbarred from being used with indefinite determiners in existential contexts: (13, MC)

Tyene alyí un sultan. have there a sultan ‘There was a sultan there.’

(Forman 1972: 216).

. Bare nouns when the nouns are intrinsically and semantically non-plural It is the case that mass nouns, and also abstract nouns, can occur as bare nouns with an indefinite sense in MC: (14a, MC)

arros rice ‘some rice’, cf. Spanish arroz

. Or with a nationality: Lipski (1986) cites, in his hispanised orthography, the Chabacano sentence si Maning es cebuano ‘Maning is a Cebuano’ (compare the following patterns of the equivalents of this sentence in relevant languages):

Tagalog: si

Maning ay

sebuano,

Def. Maning  Cebuano Cebuano: si Maning mau sugbuhanon, Def. Maning  Cebuano Spanish: Maning es cebuano. Maning is Cebuano. The incorrect decision in the use or non-use of a bare noun is what lies behind John F. Kennedy’s famous pronouncement Ick [sic] bin ein Berliner ‘I am a jelly donut’ for a presumed Ich bin Berliner ‘I am from Berlin’.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

also: (14b, MC)

tyene arros have rice ‘there is rice’

(14c, MC)

nway arros not.be rice ‘there is no rice’

These last two sentences have direct or almost direct parallels in Tagalog: (14d, Tag)

may bigas there.is rice ‘there is rice’

(14e, Tag)

wala bigas not.be rice ‘there is no rice’

Note also sentences such as the following: (15, MC)

Yaase le danyo. - do he harm ‘He did some harm.’ cf. Spanish hizo daño.

I will show later that this ready employment of bare nouns is not always the case in Tagalog (we may compare the use of ang bigas in the sense of ‘some rice’ in the sentences from Schachter (1987) that are given in Section 6; this would correspond to the MC form given in 14a).

.

Bare locative nouns in MC

There is one context in which MC determiner use (or rather non-use) resembles that of Tagalog rather than that of the Spanish model which it usually follows, and moreover, determiner use resembles it in a marked manner which suggests substratal influence. This is true of definite locative phrases that are introduced by na, which are best regarded as a subset of prepositional phrases (MC, Spanish, English and Philippine languages are all prepositional). A comparative example from the four languages mentioned will make this clear: (16a–d)

MC: na tyenda Tagalog: sa tinda Spanish: en la/una tienda, but *en tienda does not exist. English: ‘at the/a store’

 Anthony P. Grant

On the other hand, it is possible to use an indefinite determiner with such phrases in MC: (17, MC)

Yaentra le na un tyenda. - enter 3sg  one store ‘He went into a store.’

Nonetheless, my Chabacano consultant stated that a form of this sentence (17) without un is far more common. If one said: (18, Tag)

sa isa- ng at one-

tiyenda store

in Tagalog it would carry the sense of ‘into one store’ (and only one, not two or three stores). The parallel of this marked structure and its resemblance to that which is found in Tagalog as a means of expressing the same idea brings us to the issue of MC structural parallels with Central Philippine languages. Since these structures and the morphs which they employ for structural purposes are remarkably similar (indeed they are usually identical) both in Bisayan languages and in the better-attested Tagalog, I have taken my examples from the latter, as I have more data on it. The rule for the omission of the definite determiner after the preposition na in MC is not found with de ‘of ’, para10 ‘for’, sin ‘without’, kontra ‘against’ or other prepositions in MC. These take the definite determiner freely. Only na does not do so. The preposition na is used in primary roles in MC much as the preposition sa is used in Tagalog. That is, it serves both as a general purpose locative and as a means of easing the transition from one form class to another when one is using adverbs or nouns to serve as more specific prepositions of rest or of motion. However, MC has more true prepositions than Tagalog, which has laban sa for MC kontra. Finally, we may note that those place names in MC which do not have a determiner as an intrinsic part of their phonological form do not take determiners: Ipil ‘(the town of) Ipil’, rather than *el/si Ipil.

.

The templatic structure of bare nouns in MC

The structure of a bare noun or of a bare prepositional phrase (one without an NP with a determiner in its specifier) in MC can be depicted using the following formula: (Preposition) # (Numeral/manga) (adjective) noun (adjective) (qualifying adverb) (possessive adjective).

. This preposition has been borrowed into Tagalog (and other Philippine languages) and is used freely there in the sense of ‘for, on behalf of ’.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

For instance, in MC one says: (19, MC)

para dos anak grande gayot dimyo for two offspring big very of.me ‘for my two very large children’

Optional elements are bracketed. The symbol # indicates where the determiner would be placed, were it present. Possessives usually precede the head noun, rather than following it as was shown in the syntagm above (although they may occur postnominally). An alternative syntagm is needed to account for the more common location of the possessives within the noun phrase: Determiner – Possessive Adjective – Noun.

In this syntagm it is impossible to omit the definite determiner. The following phrase would be grammatically correct: (20, MC)

el disuyu reyno the of-3sg kingdom ‘his kingdom’

But the following form, with the same intended sense, is not permitted: (21, MC)

*disuyu reyno of.3sg kingdom ‘his kingdom’

The question of whether an adjective in MC precedes or follows the head noun depends upon the adjective in question; as in Spanish, the vast majority of adjectives (of whatever etymological origin) are placed post-nominally. On the other hand, Central Philippine languages have Adjective-Ligature-Noun order (the two allomorphs of the ligatures, the first of which is used after adjectives ending in a vowel and the second of which is used after consonant-final adjectives) are exhibited here: (22a, Tag)

matanda-ng babae old woman ‘old woman’

but: (22b, Spanish) mujer vieja (22c, MC) muher byeho (23a, Tag)

pandak na babae short  woman ‘short woman’

 Anthony P. Grant

but: (23b, Spanish) mujer corta (23c, MC) muher pandak

Furthermore, Ms. Neile Martinez, Santoro’s MC consultant, contrasts two phrases in MC: (24a, MC)

el pobre muher the poor woman ‘the poor/unfortunate woman’

where the adjective precedes the noun, with: (24b, MC)

el muher pobre the woman poor ‘the woman who is living in poverty’

This same distinction between the two senses of ‘poor’ is identically coded both in Spanish and French.

.

Substrate parallels: the NP system in Philippine languages, with examples from Tagalog

The MC system of NP-marking should be compared with those of the languages which have influenced it most over the past few hundred years, namely the Philippine languages. The relevant Philippine languages all behave in much the same way in this regard. We may note that MC and Caviteño share a small number of lexical innovations which are not found in the Spanish lexical source language, and also that MC contains a stratum of loans which correspond to words that are found in Tagalog but not in Hiligaynon or other Bisayan languages. It is important to recognise that these Tagalog words were in use a long time before a knowledge of Tagalog (which underlies the official language of the Philippines, Pilipino) became widespread in southwestern Mindanao as a result of compulsory primary education in Tagalog. Such Tagalog-medium education in the province of Zamboanga del Sur predates Philippine independence in 1946 (education in Zamboanga had previously been in Spanish). All in all, the evidence is fairly strong that MC as we now know it is a transplanted and subsequently bisayanised version of a Manila Bay Creole. Tagalog was the only Philippine language with which the Manila Bay Creoles were in contact at any time and the only one from which they have borrowed any Philippine forms. I have thus chosen to illustrate typical Philippine NP usage with examples from Tagalog, which are intended, for comparative purposes, to parallel the categories which I demonstrated for MC in Table 1 of Section 3. The Tagalog count noun marking system is given in

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

Table 2 below, using the word bahay ‘house’ as a sample count noun and Juan ‘John’ as a personal name. (A good account, written in French, of the determiner system in Tagalog is given in Potet (1994), while an important account of NP structures and the uses of particles such as ang is provided in Bloomfield 1917.) Table 2. Determiners in Tagalog Determiner type

Singular

Plural

Personal subject

Si: si Juan ‘John’

Personal object

Ni: ni Juan ‘John’ (object) Kay: kay Juan ‘to/for John’ Ang: ang bahay ‘the house’

Sina: Sina Juan ‘John and friends’ Nina: nina Juan ‘John and friends’ (object) Kina: kina Juan ‘to/for John and friends’ Ang manga: ang manga bahay the PLURAL house Nang manga bahay The.OBJ PLURAL house Sa manga At PLURAL house

Personal oblique Non-personal subject

Non-personal object Non-personal oblique

Nang11: nang bahay ‘the.OBJ house’ Sa: sa bahay At house

We can see that the presence and use of the pre-nominal particle manga ‘plural’ turns what would otherwise be a bare MC noun into a non-bare one, though it is not strictly a determiner, but a free-standing number marker. Use of the pluralizer manga and its allomorphs is infrequent in Tagalog (and also in other Philippine languages) if there is also a numeral in the phrase. Such a usage is permissible in Tagalog but not, apparently, in MC. At any rate, I have come across no example of this usage of numeral plus pluraliser in my MC textual materials, which include a translation of the entire New Testament. Thus in Tagalog one would say: (25, Tag)

ang manga aso   dog ‘(the) dogs’

To express ‘three dogs’ Tagalog uses: (26, Tag)

tatlo-ng aso three- dog ‘three dogs’

. Usually written as 〈ng〉 in Tagalog; similarly, manga is usually written as 〈mga〉. This pluralizer has three allomorphs. Forman (1972: 96) points out that in Zamboanga, mana has a more Spanish connotation, maga has a more Filipino connotation and manga is more and more frequently used by younger speakers, although all these are equally readily used with nouns of Spanish, Philippine and English origin.

 Anthony P. Grant

But the following phrase structure would also be permissible though less frequent in Tagalog: (27, Tag)

ang tatlo-ng manga aso three-  dog ‘three dogs’

The same restriction is true of the equivalents of these sentences in MC. (28a, MC)

manga iru, plural dog ‘dogs’

In this sentence the word for ‘dog’ is a recent borrowing from Cebuano, and the sentence is fine in MC for ‘dogs (in general, or unspecified dogs)’. (28b, MC)

el manga iru the  dog ‘the dogs’

(28b) means ‘the dogs’ (and not just ‘dogs’ in general as its morphemic equivalent in Tagalog does). (28c, MC)

tres Ø iru three Ø dog ‘three dogs’

But we cannot say the following, because one does not use the numeral and pluraliser together: (28d, MC) *tres manga iru three  dog ‘three dogs’

(28c) is the only correct way of expressing ‘three dogs’ in MC. Determiner use in MC and in Philippine languages such as Tagalog shows very different patterns. The nature of the role of determiners in signalling the presence of NPs in the sentence and in secondarily indicating the purpose of NPs in the Tagalog sentence can be seen in terms of what Schachter (1987) calls ‘triggering’. By this Schachter means the focussing of emphasis upon particular non-agent arguments within the sentence. These are factors that mean that determiner-less nouns in Tagalog or other Central Philippine language sentences are rather infrequent. This infrequency of occurrence is exacerbated by the primary role of determiners in Tagalog and other Central Philippine languages. Their primary role is that of indicating argument relations within a sentence, rather than marking such features as definiteness or indefiniteness. This is the case even though Tagalog (unlike MC) lacks an indefinite determiner, and isa(ng) ‘one’ cannot be so used in Tagalog as un is used as an indefinite determiner in MC.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

The use of a determiner in a Tagalog clause is a marker which indicates that what follows is a noun phrase. Other Philippine languages, including MC’s historically secondary lexifier Hiligaynon, use determiner systems that are similar to, and that are structurally and semantically commensurate with, the system which is found in Tagalog. It is beyond the scope of this paper to go into a long description of the structure and workings of the Tagalog determiner system and the ways in which they accord with aspectual and voice marking on verbs, and with pragmatic considerations at sentence and paragraph levels. However it is essential for observers to realise that the choice of focuses, and the morphological decisions (choices of verbal affixes, determiners and so on) which obtain are not purely pragmatic. Even so, focus-based distinctions may sometimes be made for stylistic purposes, just as English-speakers can use a passive construction when an active one would suffice. The use of certain constructions demands that a particular focus form is used, and the use of other focuses to express the meaning of this construction would quite simply result in an ungrammatical sentence. People who learn Central Philippine languages have to master the focus system in all its ramifications simply in order to produce everyday grammatically acceptable sentences. The use of the different focus patterns is not an optional extra in everyday Tagalog (or Hiligaynon, or Tausug, or Cebuano) conversation. The father of creolistics, Hugo Schuchardt, said as much in 1883 and explored the topic in his discussion of Manila Kitchen Spanish, as I show later. But none of this complex mechanism has been passed onto MC, as we will see. Not all noun phrases in Tagalog require determiners; there are certain syntactic constructions where they do not take determiners. For instance, when an NP is in predicate position, a determiner is not used: (29, Tag)

Maestra ang dalaga. teacher the maiden ‘The young lady is a teacher.’

As we saw in 3.3, this particular sentence would be almost identical in MC, both lexically and syntactically (apart from creole el replacing Tagalog ang), though it would be phrased somewhat differently in Spanish. We see this in (12a) used earlier in this article. But Spanish has the following for the same sentence: (30, Span)

La doncella es maestra. the.Fem. maiden is teacher-

In both MC and Tagalog a form of the sentence is available that uses an overt inverse marker. The MC equivalent was given as Example (12b). In Tagalog the form is: (31, Tag)

ang dalaga ay maestra the maiden  schoolteacher

This discussion of certain clause-level syntactic similarities between MC and the Central Philippine languages which helped shape it leads us to a closer consideration of substrate influences on the creole.

 Anthony P. Grant

.

MC NP structure in a historical context: substrate, superstrate, adstrate and innovation

The history of the development of MC varieties and of the communities which use them natively is complex and is still not perfectly understood. Furthermore, the exact nature of the relationship between one or more of these varieties and the Manila Bay Creoles12 is still uncertain. This uncertainty is partly a consequence of the dearth of more basilectal records for these varieties, especially for Ermiteño (which was strongly hispanised in its later decades: see the text in Whinnom 1956). But partly it is because there are a number of features of Ternateño, including tranches of lexicon from several languages spoken in the Moluccas (now Maluku) in what is now eastern Indonesia, which are not found in Caviteño or Ermiteño – or in Mindanao Creole Spanish. It is at least possible, if not probable, that Mindanao Creole Spanish has its origins in a form of Manila Bay Creole (for example a form of Caviteño) which was transported to Zamboanga by Manila Bay Creole-speaking garrisons whose more junior or lower-ranking members would also have known more or less Spanish and/or Tagalog, and which were led by Spanish-speaking officers.13 Subsequently the creole underwent profound influence from Philippine languages, but this time the influence came especially strongly from Hiligaynon. This influence was a result of being diffused as a language of the home, at least in the first generation of the existence of a Zamboanga creolophone community, by people who natively spoke Central Philippine languages but who spoke the nascent creole with their hispanophone or creolophone husbands who had not learned Central Philippine languages. The impact of Central Philippine languages (the group including Tagalog) on MC has been stronger than that of Tagalog has been upon the Manila Bay Creoles (this is a geographical rather than a genetic term). Nonetheless, the various Central Philippine languages, which constitute a valid genetic group, are typologically very similar to one another. In addition, these languages share numerous inherited lexical and bound morphemes, also sharing many others which they have acquired from Malay, Hokkien, Spanish and English, and in many cases from the more prestigious languages among their own number. This great degree of near-identity is certainly true of the main typological (and in some cases morphological) features of their NP structures. The Central Philippine language which has shaped MC most strongly and deeply is Hiligaynon, which as I said in the introduction is a western Bisayan14 language that is . As stated on p. 179, the Manila Bay Creoles are Ternateño, Caviteño and the extinct Ermiteño, which were popularly believed to have their origins in a hispanisation of a form of Creole Portuguese which was spoken by 200 Christian refugees fleeing Muslim rule in Ternate in the Moluccas. . Many young men from the Manila Bay Creole communities entered the Spanish navy as a career. . Bisayan is the general spelling used for the cover term for a subgroup of Central Philippine languages which are spoken in the central islands of the Philippines, a collection of islands known as the Visayas, which were themselves named after the Indic kingdom of Vijaya.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

spoken on the islands of Panay and Negros. Most importantly, it is the native language of Iloilo City on the island of Panay, the most important staging-point between Manila (where the Spanish-speaking soldiers who helped found Zamboanga as a permanent garrison in 1719 sailed from) and Zamboanga itself. The first locally-born generation of the permanent garrison community of Zamboanga appears to have arisen through the intermarriage of Hiligaynon-speaking women from Iloilo City with the soldiers and mariners from Manila Bay. We know that many families of Hiligaynon origin in Zamboanga maintained connections with Iloilo City for a long time, and that, in addition, many labourers from Panay went to Zamboanga in the 19th century in order to work, and settled there permanently (Lipski 2001). Hiligaynon is probably not the first Philippine language to have left its imprint on MC. If the language does have its roots in a form of Manila Bay Creole, and I think that all the evidence that points in any direction suggests that it does, then the Tagalog influence on such creoles will have historical precedence. But Hiligaynon is the source of the most profound Philippine lexical influence on the creole. The other Central Philippine languages which influenced MC, namely Tausug and Cebuano, largely just donated vocabulary rather than both lexicon and structural features. Some speakers of MC would have known Tausug in earlier decades, as it played the role of a lingua franca in extreme southwestern Mindanalo, but MC speakers would have been unlikely to shift to Tausug because it was closely associated with Islam. By contrast, MC was the language of intrusive soldiers and their families, who were Catholics in a sea of Islam. Other minor lexical sources that gave lexicon to MC were the non-Central Philippine languages Sinama, Yakan and in the case of the Cotabato variety of the creole, also Magindanaw and Tiruray (which are both Philippine languages but not Central Philippine ones). The role of Tagalog in the speech economy of Zamboanga City and environs has increased over the past few decades. This is because of its increasing use in primary education and the popular media. But it is also because Zamboanga has been the site of increased military strife between armed groups of local Muslims (largely Tausugspeakers, though the local ones are often with a second-language competence in MC) and units of the Philippine army, who use Tagalog as the language of command. Zamboangueño serves as a lingua franca in much of the Zamboanga area. It is often used in intercultural encounters between Christians (who are either L1 speakers of Chavacano, of Central Philippine languages such as Cebuano, Tagalog or Hiligaynon, or occasionally of Chinese languages or of Philippine English), Muslims (who speak the Central Philippine language Tausug or the Samalan language Yakan as their first language) and pagans, who speak other Samalan languages, such as Sinama, or the Central Philippine Subanen (or Subanon) languages, as their native languages.15

. Zamboanga was originally a Samalan-speaking area (Sama[l] people speak S-in-ama) and its name is Sinama for ‘landing place, jetty’ (in Sinama the word for this is sambwanggan). The Yakans are Samals who embraced Islam under the influence of speakers of Tausug (a Central

 Anthony P. Grant

Cebuano is a useful lingua franca which has been used in the region for about a century and which has an increasingly large body of native speakers in the area. Tagalog, in its Pilipino/Filipino guise, has been diffused through universal primary education and the mass media, as well as by Tagalog-speakers who have moved to Zamboanga. English is also increasingly well-known and is the medium of higher education throughout the Philippines. But Spanish has few native speakers in the area and its influence, once stronger here than anywhere else in the Philippines (and it was the medium of education in Zamboanga City until the 1930s), has largely waned. Hiligaynon, however, has never been used as a lingua franca or as a second language in the Zamboanga area except among families originating from the parts of Panay or Negros where it is spoken. Consequently the period of its impact on the molding of MC was early, probably the 18th century, but it was probably not prolonged. The various strata in MC played the following roles. Inasmuch as MC derives from a form of Manila Bay Creole, Hiligaynon and probably Tagalog were substrate languages (as the indigenous language Sinama may have been if some of its speakers shifted to MC in the early 18th century), providing much of the typology of the language as far as the reduced creole morphology would permit this to be perpetuated. Spanish and very probably its restructured and tagalised offshoot, Manila Bay Creole, were the superstrate languages.16 (Spanish, Tausug, Sinama, Yakan and Cebuano, and also Tagalog, and latterly English, were adstrate languages.) In short, some languages may have been both substrate and superstrate langages, or superstrate and adstrate languages, because of the nature of the history of language contact in the Zamboanga area, and also the situations of prolonged societal bilingualism which shaped and maintained MC. MC is in an unusual sociolinguistic situation since it is a creole whose native speakers were often in intimate contact with (and had at least a passive knowledge of) most of its crucial components for a period stretching over two centuries. It is also a language which has had a role as a lingua franca for an undetermined but certainly

Philippine language which was itself considerably influenced by Samalan languages). It is probable that the ancestors of many speakers of Zamboanga Chabacano were Samals who became Catholics and who adopted the language of the prestigious and powerful people who monitored the area from the safety of Fort Pilar, the Spanish fort, thereby expanding the creole-speaking community, but we cannot be sure. Other Samalan-speakers who became Catholics and moved from western Mindanao are found on Capul Island in the north-east of Mindanao; they are speakers of a language called Abaknon. . It is possible, though, that Manila Bay Creole and maybe also Spanish would have been a substrate language in those families where one parent spoke Manila Bay Creole as a first language, so that Hiligaynon may have been a superstrate language providing features from the mother’s language to the creole developing in a family household where the parents spoke different languages but where the father’s Manila Bay Creole speech formed the basis for the family language).

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

long time. Indeed there are a few loans from MC in several of the local languages, including some languages that are spoken in areas, such as those Tausug-speaking parts of Sabah and Malaysia where MC speakers never settled. For the rest, there has been rather little innovation in the structure of MC NPs; instead, the emphasis has been on the perpetuation of features which are common to most or all the important languages that helped in the growth of the creole, and of those features which occurred in languages in which significant numbers of MC speakers were bilingual.

.

MC NPs and the Philippine structural template: parallels between MC and Tagalog

A brief outline of the way in which NPs are interconnected with other structures and semantic systems in Philippine languages is in order. Tagalog, in common with other Philippine languages, has a topic-focus system which is explained in brief in Schachter (1987) and which is demonstrated in greater detail in Naylor (1975). The focus system allows the speaker to give emphasis to any one of several NP actants in a sentence, with certain restrictions. It operates at both a syntactic and a morphological level. It permeates the entire structure of the language, affecting choices in verb morphology, including voice and valency, as well as the use of determiners, and at its very core is the use of determiners presented in Rubino (1998: xiv). Sets of relationships which indicate the role of NPs in a sentence are indexed on the Tagalog verb using a complex morphological system of affixes. The morphological strategies which are used in this system include prefixation, suffixation, infixation and partial stem reduplication, and we often find a combination of some of these used within the same verb form in order to project a particular meaning (as the examples from Schachter 1987 below will show). Concepts such as ‘subject’, ‘object’ and to a lesser extent ‘agent’ and ‘patient’ are not really relevant to a discussion or elucidation of these focus system features in Philippine languages. Conversely, these same concepts are readily applicable to the analysis of MC phrase and sentence structure, since MC is a straightforward nominative-accusative language which lacks both the patterns of a Philippine-style focus system and the morphological apparatus that is needed to make such a system work. An overt example will make some of this clear. The following five sentences are sentences with differing emphases in Tagalog, even though all but the last can be translated into English in the same way (Schachter 1987: 123). I have slightly adjusted the abbreviations of the glossing; AT indicates Schachter’s ‘actor-trigger’ (where a ‘trigger’ is an overt morphological realisation of an underlying semantic role), BT ‘benefactive trigger’, PT ‘patient trigger’, DT ‘directional trigger’, IT ‘instrumental trigger’, ‘drc’ indicates what Schachter calls the ‘directional determiner’.

 Anthony P. Grant

(32a, Tag)

Mag-aalis17 ang tindero ng18 -take.out trigger storekeeper patient bigas sa sako para sa babae. rice directive sack for drc. woman ‘The storekeeper will take some rice out of a/the sack for a/the woman.’

(32b, Tag)

Aalisin ng tindero ang Cont.-take.out-PT actor storekeeper trigger bigas sa sako para sa babae. rice directional sack for drc. woman ‘A/the storekeeper will take the rice out of a/the sack for a/the woman.’

(32c, Tag)

Aalisan ng tindero ng Cont-take.out- acr. storekeeper patient bigas ang sako para sa babae. rice trigger sack for drc. woman ‘A/the storekeeper will take some rice out of the sack for a/the woman.’

(32d, Tag)

Ipag-aalis ng tindero -cont.-take.out patient storekeeper ng bigas sa sako ang babae. patient rice directional sack trigger woman ‘A/the storekeeper will take some rice out of a/the sack for the woman.’

(32e, Tag)

Ipangaalis ng tindero -cont.-take.out patient storekeeper ng bigas sa sako ang sandok. patient rice directional sack trigger scoop ‘A/the storekeeper will take some rice out of a/the sack with the scoop.’

As we can see, several of these sentences are open to several interpretations in English, with the variation especially apparent in the use of different determiners in English. Sometimes NPs in these Tagalog sentences are categorically definite, sometimes they are categorically indefinite, and sometimes both interpretations are possible. This ambiguity would also be true of the semantically equivalent sentences in other Philippine languages, such as Hiligaynon. The denotation of definiteness depends upon the particular combination of prenominal and intraverbal markers. Yet all these sentences

. The digraph 〈aa〉 does not indicate a long vowel. Phonetically this word is [magaa:lis] with the stress on the penultimate syllable and with a stem [a:lis]. . This is pronounced (and occasionally written) nang. Similarly, in Tagalog the pluralizer marker manga is generally written mga.

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

can and must be translated by the MC equivalents which emphasize different features from those emphasized in the Tagalog versions: (33a, MC)

Ay-saka el/un tendero (el) arros -take the/a storekeeper (the) rice del sako para el/in muher. out.of.the sack for the/a woman ‘The/a storekeeper will take the rice from the sack for the/a woman.’ (Spanish: El tendero tomó el arroz del saco para una/la mujer.)

And again: (33b, MC)

Ay-saka el/un tendero kon arros del  take the/a storekeeper  rice of.the sako kon el sanduk. sack with the scoop ‘The/a storekeeper will take the rice from the sack with a scoop.’ (Spanish: El tendero tomó el arroz del saco con un cucharón.)

It should be explained that in this second translation, kon plays two different semantic or semantactic roles in these sentences. It denotes the direct human object in the first instance (thereby corresponding to the Castilian Spanish use of ‘personal a’) and it refers to the instrumental and comitative adposition ‘with’ in the second. As we can see from the above sentences, especially sentences (33a) and (33b), MC has distinct definite and indefinite determiners. What is more, it can use them in ways which resemble their use in Spanish or English more than the ways they are employed in Philippine languages. This is true at least of the ‘definite determiner’, there is no true indefinite determiner in Tagalog or other Philippine languages and the Tagalog numeral isa ‘one’ is not used in this way. This is demonstrated in the differing Tagalog forms of the above sentences. MC generally uses morphological techniques which make distinctions (however they may be encoded) between definite and indefinite NPs when its speakers use determiners, although definiteness is more often signalled overtly than is the case with indefiniteness. The feature of plurality or otherwise of an NP is also generally indicated as the result of using such determiners. Conversely, in Tagalog the semantactic feature of definiteness of an NP as opposed to indefiniteness is not overtly manifested in the determiner being used, but it is signalled more precisely in the use of a particular combination of types of determiner and of certain kinds of verbal morphology.

.

Considerations for creolistic and general linguistic theory

In regard to the use of bare nouns, MC is somewhat anomalous or exceptional when it is compared with other creole languages, even those with Iberoromance languages as their major lexifiers. The language also appears anomalous when its bare noun usage is compared with those of Central Philippine languages (and in this respect the structure of the Central Philippine languages which have influenced MC are in any case

 Anthony P. Grant

extremely similar to one another). This is largely because of the proliferation of determiners in use. Philippine languages use the prenominal particles that we designate as determiners as a means of indexing nominal arguments and of keying them into their roles in the focus system. On the other hand, MC uses them (including the ‘empty determiner’ zero) to indicate the different kinds of semantic status of the noun in question: definite versus indefinite, human/personal versus other kinds, plural versus singular. I am and always have been a believer in the primacy of data over theory, and in the power of new data to shape and remodel pre-existing theories, which need to be accountable to these data. I would maintain that the theoretical considerations that are raised by bare nouns and other kinds of nouns in MC belong largely to the development of general creolistic theory, and especially that part of it which asks how creole languages arose and how they (or rather their earlier generations of speakers) came to select and embody the features which they have. But it does seem to me that one of the areas in which the data in this paper can make a contribution to theory is the elaboration of topics and ideas concerning the nature and interplay of certain structural features: the overt expression of number and definiteness in NP systems, and the relative privileging of the overt marking of definiteness rather than of number in such phrases. One such subtopic would be that of the study of what is called the singulative and of number-neutrality within a marked system. Quite a few languages (including some Celtic and many Afroasiatic languages, including Egyptian Arabic and many Cushitic languages) have several morphologically discrete numbers among the morphological categories found in count nouns. There is an overt plural form, there is also an overt and distinctively singular form (sometimes known as a singulative), both of which are morphologically marked. Both of these numbers are in many cases opposed to a generic or ‘number-neutral’ form of the noun which can be used in those contexts where number is not salient. For instance the ‘number-neutral’ form is the form of the noun which would be employed in a sentence such as ‘a horse has four legs’, which is largely interchangeable with ‘horses have four legs’. This form would also be used in those cases where number is overtly expressed by the use of a numeral. MC usage is not too different from this. The number-assignment of an NP in MC is usually determinable from the use of a particular determiner or more readily from the use of manga or a numeral. But this is not always so, and the facultative and free use of zero-marking and number-neutrality within a system where number and definiteness can both be readily overtly marked without much trouble is a matter which requires further examination.

.

The Philippine impact on MC structure: transfer of fabric and transfer of pattern

The impact of Philippine languages upon the structure of MC can been seen both in terms of transfer of pattern and what I call transfer of fabric. Pattern transfer in MC,

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

which involves the introduction of syntactic or typological changes which help to differentiate MC from its Spanish lexifier, includes basic VSO word order, the structure and order of the elements in the verb phrase (Aspect-Stem complex,19 then Pronoun or at least Verb-Pronoun), the post-nominal placing of pronominal possessors, and the frequent use of apposition to express copular relations. (As I stated above, the overt copula which is normally used, amo, has been regrammaticalised within Mindanao Chabacano in the course of the twentieth century from a Hiligaynon emphatic highlighter particle which does not serve as a copula in Hiligaynon, though it also serves as an emphatic particle in MC.20) What I call ‘transfer of fabric’ from Hiligaynon to MC involves the borrowing of actual morphs from Hiligaynon into MC, with or without the simultaneous borrowing of their distinctive syntactic patterns. The concept of metatypy, which was first aired in Ross (1996) is essential to an understanding of the relationship of MC to its substrate languages and to the operation of many features of MC NPs, although some underlying features of this were adumbrated in Frake (1971). In the case of what Ross calls metatypy, the syntactic or typological structure of a language whose speakers are under the influence of a more powerful language is modified typologically rather than by borrowing. The effects of this make the syntactic structure of the less powerful language resemble the syntactic or typological structure of the influencing language more closely. However, in cases of true metatypy, this directionally-driven modification of a language is accomplished without borrowing morphs from the language whose typological patterns are to be emulated. Instead the modification is carried out by changing or extending the uses of morphs which have been inherited from an ancestral form of the language whose structure is being modified, and by making these uses fit those of the language in whose direction the modification is taking place. We can see occurrences of this in MC NPs, in issues such as pluralisation where both transfer of fabric and transfer of pattern occur within the same structure. There are striking similarities in the NPs of MC with features of NPs in Philippine languages (of which data from Tagalog, documented in Schachter 1987, will serve as an entirely typical example), but these similarities mask some of the startling differences, which are much more profound and which affect some core characteristics of MC NP and VP structures. The action of metatypy between Central Philippine

. But MC and Central Philippine languages are different morphologically. MC uses only prefixed particles and stems, whereas Central Philippine languages use stems shapes which may include infixes inserted after the first consonant of the stem, and also suffixes, in addition to prefixes (though they are different particular prefixes) as used in MC. . It should not be necessary for me to note that although Tagalog elements are found in all forms of Chabacano, elements that derive from languages of the Visayas and Mindanao are certainly not found in the Manila Bay Creoles.

 Anthony P. Grant

languages and MC, with MC syntax remodelling itself little by little on that of Central Philippine languages, has carried on apace, as (we assume) has the borrowing and subsequent productive use of some significant morphs from Central Philippine languages into MC. There are probably more productive morphs of Philippine origin used in MC structure than there are morphs of Spanish derivation. Nonetheless, wholesale reliance upon a substratist argument of this sort will not explain everything about the rules and categories of MC NP structure and about restrictions on the use of bare nouns in the language, though it accounts for a great deal. This chapter has been about NPs, but deliberate attention has been paid to bare nouns. Bare nouns in MC are not as frequent in occurrence as they are in many other creoles, and the reasons for their relative scarcity are morphological. Bare nouns are less frequent because MC has both definite and (singular) indefinite determiners, and also a prenominal free-standing NP pluralizer, and these are widely used. Many other creoles lack one or both of these morphological features, and both the number and frequency of bare nouns in such creoles are thereby increased. Indeed the use of one or another of the determiners (but this is less the case with the pluralizer) appears to be all but categorical on the occasions when it is the speaker’s intention to impute particular senses such as definiteness in regard to a certain noun phrase. One cannot logically signal overt definiteness on a MC noun phrase just by using a bare noun, as the use of one of a small set of determiners is the only means available for signalling definiteness. Bare nouns and other kinds of nouns in MC can be subcategorised according to use, as there are several different types, each having their distinctive patterns of determiner use. The context which is based on certain semantic properties of the NP in question defines the sense and number-allocation of the bare noun in MC. In general, if the noun is a mass noun or an abstract noun, then a bare noun using such a noun as its head will naturally be construed as being singular (or it will be number-insensitive, in the case of abstract nouns) and non-specific (‘some X’), and indeed such nouns are used as bare nouns. This usage is found especially frequently when they are not subjects. (34, MC)

Ta-kome yo arros. -eat 1 rice ‘I am eating (some) rice’ (see also Examples (1), (2), (3))

If the noun is a count noun, then the bare noun can be interpreted by default as a plural indefinite noun (although it is possible to use manga before otherwise bare nouns in order to make it clear that the NP in question is plural rather than singular): (35, MC)

Ta-buta lang urinola  throw only urinal ‘He just empties urinals’ (Frake 1980: 284, see also Example 8).

If the bare noun in question is a mass noun, then it is going to be construed as a singular (or at least a nonplural) in any case (Section 3.4 and Examples (14) and (15)).

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

If the noun is a personal noun, it will not be involved in a bare noun construction as the head, as personal names of whatever origin require the determiner si as subject and kon el, del as the object and possessive determiners, and do not appear without determiners (see Examples (12a) and (12b) above). It is a morphological (rather than more purely semantic) consideration, namely the obligatory use of a determiner, which rules out personal NPs as potential candidates for bare noun status. These sentences also show the structure and order of NPs in copular sentences. If the noun in question is a place-name, then it will be determiner-less as a matter of course, so that place-names constitute yet another semantically-defined determiner category (Section 3.5). Names of professions and nationalities occur without determiners in equative sentences (see Examples (12a), (12b) and Footnote 7). Certain idioms, such as: (36, MC)

entra eskwela enter school ‘to enter school’

(Frake 1980: 293),

imprecations (the sentence in Frake 1980: 284, which is also our Example 11) and other fixed expressions comprise the rest of the true bare nouns in MC. So far, all is like Spanish. As far as I can tell, there are no restrictions in MC on the co-occurrence of different kinds of determiners in a sentence in the way in which there are in Tagalog, just as long as the determiners that are used are semantically appropriate to their noun phrases. However, in common with the rule in Philippine languages governing the structure of phrases introduced by locative prepositions such as sa, MC does not use definite determiners before nouns in prepositional phrases which are introduced by the utility preposition na. (See Example 8.) There is no parallel for this construction and this ‘determiner gap’ in Spanish, where the use of a or en does not license the omission of an otherwise expected definite article. Therefore this MC feature appears to be a relic of the Philippine substrate (which is also an adstrate in the form of Tagalog – which constituted an element in Manila Bay Creoles - and Cebuano). It is a rule which affects the surface structure of MC and it has been apparently taken over from the grammar of one or another Central Philippine language. Yet, it is plausible to assert that such nouns are marked for definiteness in their underlying structure but that the definiteness is not shown in the surface structure. The absence of a determiner in such phrases has next to nothing to do with the semantic content of the noun in question; it is a syntactic matter. And yet, tantalisingly, there is a parallel case with the use of na in Papiamentu. The obligatory absence of the definite article after na in MC, but the continuing use of na before indefinites and possessive pronouns, and the use of the definite article after all other prepositions (where appropriate), are features on which MC and Papiamentu in all its stages of recording agree completely. I ascertained this recently by checking all

 Anthony P. Grant

the Papiamentu texts (from 1775 onward) that I could find. Similarly the equivalent locative preposition sa is used before indefinites in Tagalog. Overall, the general nature of determiner usage (and of bare noun occurrence) in MC is more reminiscent of what is found in Spanish or even English than what is found in Tagalog, Hiligaynon and other Central Philippine languages, where determiners are more like generalised NP markers. If a noun or noun phrase is kept bare in a Spanish sentence, it is very likely to be bare in the MC equivalent of that sentence. Yet the bare locatives that I mentioned above are an interesting exception (and a rare example of zero as a syntactically marked option!). The biggest difference between MC and Philippine languages in this respect is that determiner usage in Tagalog, Bisayan languages and other Philippine languages needs to be seen as part of the morphosyntactic framework of determiner use that has been brought about by the nature of the Philippine verb phrase and its focus system. The use of certain triggers in an initial verb phrase licenses (or at least compels) the use of certain kinds of NP, which have different roles and which are distinguished from one another by the use of particular determiners. In this system the overt marking of (for instance) definiteness against indefiniteness is a decidedly secondary consideration within the sentence. None of the syntactic interplay between NPs and VPs, characteristic of Tagalog and other Central Philippine languages, which determines the form which different kinds of NPs may take when specific kinds of VP structures are used, has been transferred into MC. This is not to forget that there are a number of features of Philippine sentential, NP and indeed VP semantics which are crosslinguistically highly infrequent and thus ‘marked’, and which have been transferred from Philippine languages into MC. We may note, for instance, that the use of a bare noun in copular sentences, shown above in Example 12a, is one such example. But the syntactic consequences of the use of a complex focus system (whose primary complexity is manifested at the morphological level) are not reflected in MC, which has never had such distinctions, as far as we can tell, and which operates on a strictly Subject-Object case-marking principle. Schuchardt (1883: 130) had made this much quite clear in his study of Manila Kitchen Spanish, a pidginised Spanish that was influenced by Tagalog and that was used by uneducated Filipinos and Chinese who interacted with Spaniards. He provides a couple of examples in his discussion of this register. In doing this he shows that there are overtly morphologically-marked focus-changes which are required in the verbal and determiner systems in Tagalog in order to distinguish between the Tagalog equivalents of the imperative sentences ‘kill a chicken’ and ‘kill this chicken’. These two apparently clearly related meanings cannot be expressed with the same focus markers and determiners. But these are not perpetuated in their equivalents in 19th century Manila Kitchen Spanish, or for that matter, in 20th and 21st century MC, as the equivalents show: (37a, Spanish)

mata (tú) una gallina kill thou one.. chicken

Chapter 6. Some aspects of NPs in Mindanao Chabacano 

(37b, Tagalog) (modern spelling)

matay ca nang isáng manúc matay ka ng isa-ng manók kill 2  one- chicken

(37c, MC)

mata kon-un galyina kill -a chicken ‘kill a chicken’

(38a, Spanish)

mata (tú) esta gallina kill thou this.. chicken

(38b, Tagalog) (modern spelling)

pataín mo itóng manúc 21 pa-matay-ín mo itó-ng manók -kill- 2 this- chicken

(38c, MC)

mata kon-este galyina kill -this chicken ‘kill this chicken’

The Philippine focus system, combining verbal morphology and the subsequent operation of noun-phrase syntax, is the place where the operation of metatypy between MC and Central Philippine languages gives up the ghost. There is an important consequence of this which comes out most clearly in a cross-linguistic analysis. The use or non-use of determiners in MC, Spanish, Malay, English, and most other languages is semantically defined in the first instance, according to the meaning of the noun phrase (although certain pragmatic conditions may also come into play). The rules governing the use of determiners in some other languages may be morphologically defined (and this is the case with many of the languages that are examined in Chierchia (1998), and in many other languages too). But the system of rules governing the use and nonuse of determiners in Central Philippine languages is best seen as being syntactically conditioned, though the matter is rather more complex than that, and the selection among individual determiners in a Central Philippine sentence depends upon the nature of the arguments which the verb in a sentence may take. It would further be possible for us to draw up a cline of obligatory and optional determiner use for many languages, using a metric which is based on such considerations as (for example) whether abstract nouns always required the use of a determiner. Thereafter we could hypothetically arrange the determiner behaviour of numerous

. This form appears irregular because the first syllable of Tagalog matay ‘kill’ has been construed as being identical at a synchronic level with the common stative verbal prefix ma-. In fact, the alternation of the stem /-atay/ ‘to die’ with the causative prefix pa- and the stative prefix ma- goes back to Proto-Austronesian, where the stem form was /-aCay/ (Blust 1999). Tagalog has preserved an alternation which has been ‘regularised’ by most other Austronesian languages. The stem has nothing to do with the Spanish verb matar ‘to kill’, whose origins are still uncertain. The PT in the gloss of this sentence stands for ‘patient trigger’.

 Anthony P. Grant

languages on a cline according to this. But it makes little sense to include Central Philippine languages (including Tausug in this sense) in this cline. Their conditions constraining and governing determiner use are motivated by quite different considerations from those of most of the other, non-Central Philippine, languages whose speakers have played a part in the history of the development and the growth of the lexicon of MC. The relevant languages include Arabic, Malay, Samalan languages,22 Hokkien Chinese, Spanish, Manila Bay Creoles, and of course English. Determiner use or non-use in those languages listed above which have determiners (not Samalan languages, Malay or Hokkien) is a semantically defined matter, just as it is in MC, whereas in Tagalog and in the other Central Philippine languages determiner use is syntactically defined. The impact of Philippine languages on the determiner system in MC, including that on the rules for use of bare nouns, has been surprisingly strong and subtle. However, the focus system which underpins so much of the morphological, semantic and syntactic structure of Philippine languages, has not been replicated into the creole, which paradoxically seems more ‘European’ in determiner semantax than most other creoles.

References Aoto, S. 2002. La cópula AMO del chabacano de Zamboanga. PAPIA: Revista de Crioulos de Base Iberica 12: 84–106. Bloomfield, L. 1917. Tagalog texts with grammatical analysis. University of Illinois Studies in Language and Literature II: parts 2–4. Blust, R. A. 1999. Subgrouping, circularity and extinction of some issues in Austronesian comparative linguistics. In: Selected Papers from the Eighth International Conference on Austronesian Linguistics, edited by Elizabeth Zeitoun and Paul Jen-kuei Li, 31–94. Taipei: the Institute of Linguistics (Preparatory Office), Academia Sinica.

. The Samalan languages are spoken in the southern Philippines and in surrounding parts of Sabah and Borneo, and they have exerted considerable influence upon Tausug, the local lingua franca among Muslims (as has Brunei Malay). But both genetically and typologically they are closer to Malay and the Dayak languages than to Philippine languages such as Tausug, Hiligaynon and Tagalog, and they lack the focus system and other morphological characteristics which Philippine languages have retained (as well as lacking many morphs which Philippine languages have developed through innovation). Consequently they are normally omitted from classifications of ‘Philippine languages’, despite being spoken extensively in several locations in the southern Philippines. The Samalan languages in this category which have donated lexemes to Mindanao Chabacano are Sinama and Yakan. It is probable that many of the modern speakers of Mindanao Chabacano are descendants of people who spoke Samalan languages and who subsequently underwent language shift to Chabacano, together with ideological shift from paganism or possibly Islam to Catholicism.

Chapter 6. Some aspects of NPs in Mindanao Chabacano  Broad, H. P. 1929. Chabacano. Philippine Magazine XXVI: 142–160. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Do Couto, H. H. (ed.). 2002. [Special issue devoted to Chabacano.] PAPIA: Revista de Crioulos de Base Iberica, 12. Brasilia: Universidade National de Brasil, Associaçião Brasileira de Estudios de Crioulos e Similares. Fernández, M. (ed.). 2001. Shedding Light on the Chabacano Language [Estudios de Sociolingüística 2: 2]. Vigo: Universidade de Vigo. Forman, M. L. 1972. Zamboangueño Texts with Grammatical Analysis. hD dissertation, Cornell University. Frake, C. O. 1971. Lexical origins and semantic structures in Philippine Creole Spanish. In Pidginization and Creolization of Languages, D. Hymes, 223–242. Cambridge: CUP. Frake, C. O. 1980. Zamboangueño verbal expressions. In Language and Cultural Description: Selected essays by Charles O. Frake, A. S. Dil, 275–310. Stanford CA: Stanford University Press. Grant, A. P. 2002. El chabacano zamboangueño, una lengua mezclada. Papia 12: 7–40. Holm, J. A. & Patrick, P. L. To appear. Comparative Creole Syntax. London: Battlebridge. Ing, R. O. 1967. A short account of Chabacano phonology. Le Maître Phonétique 128: 24–33. Ing, R. O. 1968. A Phonological Analysis of Chabacano. PhD Dissertation, University College London. Lipski, J. M. 1986. Modern Spanish once removed in Philippine Creole Spanish: The case of Zamboanga. Language in Society 16: 91–108. Lipski, J. M. 2001. The place of Chabacano in the Philippine linguistic profile. In Shedding Light on the Chabacano Language [Estudios de Sociolingüística 2: 2], M. Fernández (ed.), 119– 163. Vigo: Universidade de Vigo. Llamado, L. 1972. Phrase structure rules of Cavite Chabacano. Philippine Journal of Linguistics 3: 65–93. McKaughan, H. P. 1954. Notes on Chabacano grammar. Far Eastern Faculty Studies 3: 205–226. Naylor, P. B. 1975 Topic, focus and emphasis in the Tagalog verbal clause. Oceanic Linguistics 14: 12–79. Potet, J- P. 1994. Les marqueurs nominaux en tagal. Cahiers de Linguistique Asie Orientale 23: 279–292. Riego de Dios, M. I. 1989. A Composite Dictionary of Philippine Creole Spanish. Linguistic Society of the Philippines. Manila: Summer Institute of Linguistics. Ross, M. D. 1996. Contact-induced change and the comparative method: Cases from Papua New Guinea. In The Comparative Method Reviewed, M. D. & M. Ross (eds), 180–217. New York NY: Academic Press. Rubino, C. A. G. 1998. Tagalog-English, English-Tagalog Dictionary. NewYork NY: Hippocrene. Santoro, S. To appear. Zamboangueño. In Comparative Creole Syntax, J. Holm and P. Patrick (eds). London: Battlebridge. Schachter, P. 1987. Tagalog. In The Major Languages of East and South-East Asia, B. Comrie (ed.), 208–230. London: Croom Helm. Schuchardt, H. 1883. Kreolische Studien IV: Über das Malaiospanische der Philippinen. Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften zu Wien 105: 111–150. Whinnom, K. 1956. Spanish Contact Vernaculars in the Philippine Islands. Hong Kong: Hong Kong University Press.

 Anthony P. Grant

Appendix 1: Determiner choices in MC, Tagalog, Spanish and English according to different types of NPs Personal name (‘Carlos’) Place names (eg Manila) Indefinite mass noun (‘some rice’) Count noun in the singular (‘a man’) Count noun in the plural, used indefinitely (‘some men’) Count noun in the plural used definitely (‘the men’, ‘men’ (in general) Abstract noun as a totality (‘truth’) Abstract noun as a non-totality (‘some hope’)

MC

Tagalog

Spanish

English

Si 0 0

Si 0 Ang

0 0 0

0 0 0/‘some’

Un/0

Ang/isang

Un/una

a/an

Manga/0

(ang) manga

(unos/unas)/0

some/0

El manga

Ang manga

Los/las

the/0

El

Ang

el/la

0

0

Ang

0

0

 

Bare nouns in Palenquero A fresh consensus in the making Armin Schwegler The present article argues that the traditional accounts of Palenquero’s article system are fundamentally flawed. Schwegler’s revisionist analysis is largely based on data he had collected in situ, but also relies on Moñino (in press), who similarly questions the traditional accounts of the Palenquero article system. Moñino and Schwegler have reached a consensus as regards two key points: in their view, (1) Palenquero has noun phrases consisting of just a bare noun; and (2) a striking property of these bare nouns is their non-specification for number (they are neither singular nor plural). One of the objectives of this study is to offer a comparison between Schwegler and Moñino’s analysis of Palenquero ma, and to outline why they – independently so – reject the traditional views on the matter. In so doing, Schwegler shows that the Palenquero lexicon emphasizes the role of context rather than overt morphology or lexical structure. This new approach represents a radical departure from earlier analyses, as it renders all Palenquero nouns “transnumeral.” As will become apparent, Moñino’s and Schwegler’s findings coincide on several major points of grammar, but differ on the more theoretical issue concerning the genesis of ma.

.

Introduction*

Faingold (1994), Friedemann/Patiño (1983: 138–149), Megenney (1986: 149–150) and other early investigations into Palenquero (Pal.) had all noted that the plural noun phrase differs in fundamental ways from that of the Spanish lexifier. These authors * This article is a slighly revised version of Schwegler (in press b), originally edited by Wiltrud Mihatsch and Monika Sokol. I thank them as well as the publisher (Peter Lang) for giving me permission to include my study in the present volume. I am equally appreciative of Marlyse Baptista’s and Jacqueline Guéron’s extraordinary patience: although I was invited early on to contribute to this volume, long-standing commitments to other projects (e.g. Schwegler/Kempff 2007) kept me from submitting my contribution in a more timely manner. Thomas Morton and Yves Moñino made very constructive comments to a draft version of this article. The usual disclaimers apply.

 Armin Schwegler

were in agreement that the (supposedly invariable) zero form always functions as the singular definite article (e.g. ómbe ‘THE man’, muhé ‘THE woman’) or as a modifier of generic and mass nouns (ómbe ‘man [in general], mankind’, muhé ‘women [in general]’). In functional terms, the omission of articles was thus said to correspond to Span. el (masc., ex. 1a) or la ‘the’ (fem., ex. 1b):1 (1a)

Ø ómbe tá aí. (zero) man be (over) there ‘The man is over there.’ ‘El hombre está allí.’

(A.S.)

(1b)

Ø muhé tá aí. (zero) woman be (over) there ‘The woman is over there.’ ‘La mujer está allí.’

(A.S.)

These same sources uniformly reported, furthermore, that ma ‘definite particle’ (2) and un ma ‘indefinite particle’ (lit. ‘one PL’) regularly mark the plural (3), thereby presumably yielding a system in which the singular alone allows bare nouns (4). (2)

Ma nimá mí a- tá kumé yébba. (A.S.) PL animal my T/A T/A eat grass ‘The animals of mine (= my animals, i.e. donkeys) are eating grass.’ ‘Los animales míos (= mis animales, i.e. burros) están comiendo hierba.’

(3)

Ma kúsa tá aí mémo. PL thing be there right ‘The things are right over there.’ ‘Las cosas están allí mismo.’

(4)

Ø amígo

(A.S.)

‘the man’ (singular) *‘men’

(plural)[traditional analysis]

Scholars’ early and repeated attention to Pal. ma can readily be understood: on one hand, this particle is a prominent, highly frequent element of the creole; on the other hand, it was identified early on as an “exotic”  feature, one that I and others have deemed especially useful for tracing the substratal (Kikongo) origins of Palenquero.2

. Readers will note that stressed syllables are routinely marked with an acute accent (e.g. ómbe ‘man’, muhé ‘woman’). Contrary to my own earlier practice, this system of transcription closely mirrors that adopted by Moñino (in press). This paper will draw on data samples from multiple sources. Examples from my own corpus (described in Section 2) are labeled “A.S.”. . Overtly African elements – lexical or otherwise – are surprisingly rare in Palenquero, numbering less than a dozen or so in everyday language. For details, consult Hualde and Schwegler

Chapter 7. Bare nouns in Palenquero 

So great was the consensus about the supposed behavior of ma (and, with it, the overall nature of Palenquero’s article system) that scholars subsequently felt little or no need to return to the topic under discussion. Soon after the last of my many intermittent stays in Palenque (1985–1997), I began to suspect, however, that the traditional accounts of Palenquero’s article system are fundamentally flawed. A subsequent re-examination of data I had collected in situ (between 1985 and 1995) further strengthened my suspicion, and eventually formed the basis of a conference paper on the topic (Schwegler 2002a). These initial findings were then incorporated into a forthcoming article (Schwegler, in press a)3 that offers a panoramic assessment of Palenquero grammar. Working independently from me, the Africanist and Palenquero scholar Yves Moñino also began to question the traditional accounts of the Palenquero article system, and, more specifically, how Lengua (the creole’s local name) encodes ‘plural’ on nouns. Basing himself on a total of over two years of fieldwork experience (accumulated in Palenque during intermittent stays between 1994 y 2001), he too began to observe that, contrary to earlier claims, the ‘definite’ plural is not always marked by the particle ma. He too noticed that speakers of Palenquero could, in other words, be shown to rely, with more than just sporadic frequency, on a system in which bare nouns allow both singular and plural interpretation. Moñino and I have thus reached a consensus as regards two key points: in our view, (1) Palenquero, like many creoles and other languages, has noun phrases consisting of just a bare noun without any functional elements such as determiners, classifiers, or number morphemes; and (2) a striking property of these bare nouns is that they are neither singular nor plural, but somehow “neutral” or “unspecified” for number. Borrowing terminology from Corbett (2000) and Rullmann/You (2006: 175), such bare nouns can thus be said to have GENERAL NUMBER. As shown in the remaining sections of the present study, our recent observations and other considerations have led Moñino and me to reject the traditional analysis, and to offer a view of the Palenquero lexicon that emphasizes the ROLE OF CONTEXT rather than overt morphology or lexical structure. This new approach represents a radical departure from earlier analyses, as it renders all Palenquero nouns transnumeral.4

(in press), Moñino (1999, 2002), Morton (2005), Schwegler (2002b, 2002c, 2006), and relevant sources cited therein. References to etymological inquiries about ma are given in Schwegler (2002b: 198) and also in Section 5 of Moñino in press. . Being panoramic in nature, the forthcoming study does not concentrate specifically on pluralization. Its publication (and that of the collective volume in which it is slanted to appear) has been delayed for several years for reasons beyond my control. Its scheduled date of publication is now early 2007. . In (Romance) creoles, the use of bare nouns or transnumeral systems are not unusual. See, for instance, Stolz (1998: 632) for Krîol, Baptista (this volume) for Cape Verdean, or Alexandre and Hagemeijer (this volume) for São Tomé. Also relevant is Maurer (2002).

 Armin Schwegler

. Goals and data sources As explained above, to date, virtually all available printed sources continue to portray the Palenquero article system in ways that Moñino and I no longer find convincing. Until most recently, several research commitments (e.g. Fuentes/Schwegler 2005, Schwegler/Kempff 2007) had prevented me from re-examining the Palenquero article system in print. Personal communications (in persona or via e-mail) with Yves Moñino, and especially his captivating study of Palenquero ma (Moñino, in press), now present me with a welcome and most timely incentive to return to the topic in this contribution. The goal of this article is to offer the scholarly community a brief comparison between Moñino’s analysis of Pal. ma and my own. As will become apparent, our findings coincide on several major points of grammar, but differ (amicably so) on the more theoretical issue concerning the genesis of ma. It is hoped that the ensuing discussion will serve as point of departure for future, more ambitious investigations into the grammar of Palenquero.5 My data sources are essentially those already mentioned: I will here draw on recordings made during my fieldwork in Palenque, and combine them with pertinent examples from Moñino (in press) and other relevant sources. My Palenquero informants are all fully fluent speakers of the creole. Language attrition and/or especially language abandonment – much advanced today among many young(er), mostly monolingual Palenqueros6 – are, therefore, not factors that conditioned my data set in noteworthy ways. The recordings are consistently CASUAL and INFORMAL in nature, and typically consist of dialogues between two or more Palenqueros. As such they are highly representative of “normal” everyday speech, and can be viewed as valid examples of how Palenqueros routinely encode “plural” in their noun system.

.

Plural ma and the value of bare nouns

Ma, we recall (Section 1 above), has traditionally been described as the nominal pluralizer par excellence, one on which the Palenquero nominal system has been said to depend for signaling both definite and indefinite plural. This account thus yielded a system in which only the singular allows the omission of an article. Schematically, this

. A book-length revision of Palenquero grammar is long overdue. The standard reference remains Friedemann and Patiño (1983), a work that is in urgent need of revision. . Today, many younger Palenqueros have no, or only very limited, speaking knowledge of the creole, though their passive knowledge (comprehension) is often quite extensive. For further information on this topic, see Morton/Schwegler (2003: 104–111), and especially Pfleiderer (1998).

Chapter 7. Bare nouns in Palenquero 

conventional analysis can be represented as follows (the accompanying examples 5a–d are from Schwegler, in press a): Table 1. Traditional schematization of the Palenquero article system definite, or generic and mass nouns (always pre-nominal)

indefinite (always pre-nominal)

singular

Ø (= omission of article)

ún

plural

ma

ún ma

(5a) (5b) (5c) (5d)

Ø Ún Ma Ún ma

pélo pélo pélo pélo

asé asé asé asé

ndrumí ndrumí ndrumí ndrumí

múcho. múcho. múcho. múcho.

‘The dog sleeps a lot.’ ‘A dog sleeps a lot.’ ‘(The) dogs sleep a lot.’ ‘Some dogs sleep a lot.’

In a footnote to a study on Palenquero genitive constructions, Moñino (2002) offered first hints that, in his view, context rather than strictly overt morphological marking may, at times, signal plural. He expressed his preliminary thoughts on the matter as follows: Patiño opone este ‘plural definido’ a un ‘plural indefinido’, que preferimos llamar ‘plural singulativo’ porque aisla unos ejemplares dados de su conjunto total. El plural singulativo se marca en palenquero con ún ma ‘unos, algunos’, compuesto de ún ‘artículo singulativo’ y de ma ‘artículo colectivo’. La oposición definido/in definido, propia del español, no da cuenta del sistema palenquero de los determinantes nominales, que se estructura alrededor de la forma no marcada del nombre: pekáo ri lóyo puede según el contexto significar ‘el pez del río’, ‘un pez del río’, ‘un pez de río’ o ‘los peces del río’. La presencia de un determinante es facultativa, pero trae una información adicional que elimina la ambigüedad: ún pekáo ri lóyo ‘uno de los peces del río’ (singular singulativo), ma pekáo ri lóyo ‘el conjunto de los peces del río’ (plural colectivo), ún ma pekáo ri lóyo ‘unos entre el conjunto de los peces del río’ (plural singulativo). (Moñino 2002: 246, n. 23; the emphasis is mine)

In January 2000, I had made a similar point when writing the following (Schwegler, in press a):7 Furthermore, and more importantly, it is also true that context rather than overt morphology often serves to make the definite/indefinite distinction. The importance of contextual input can be appreciated in the following excerpt of a dialogue

.

The accompanying example has been renumbered for the purpose of this article.

 Armin Schwegler

between V. and A., where the overt plural marking of ma kuágro ‘the cuadros’8 by the first interlocutor makes possible the non-overt plural kuágro re aóra ‘the “cuadros” of nowadays’ in the second utterance. (6)

V: ¡Ablá mí kúmo éra ma kúsa ri ma kuágro akí! (A.S.) tell me how be-PAST PL thing of PL “cuadro” here ‘Tell me how these things were with the cuadros here (in Palenque)!’ A:

Ø kuágro ri aóra, éso tá desoddenáo. (A.S.) “cuadro” of now this be disorganized ‘The cuadros of nowadays, this is disorganized (= today’s cuadros are disorganized).’

In the same article, I also drew attention to the fact that there exists a very common mechanism of signaling the definite/indefinite distinction that has been altogether overlooked in previous studies. I am here referring to locative constructions involving the ubiquitous prepositions akí, aí, ayá and ándi (variants ónde, dónde), whose very wide meaning approximates ‘in, at, on, by, etc. + locative deixis’. As exemplified in greater detail in Section 19 of Schwegler (in press a), nouns preceded by prepositional akí, aká, aí, ayá are inherently definite whereas those preceded by prepositional ándi can be definite or indefinite (cp. akí iglésia lit. ‘here church’ = ‘in, at, to + THE church’, but never *akí ún iglésia ‘in, at, to + A church’). This fact of Palenquero grammar explains, among other things, why a) prepositional constructions like *akí ún iglésia, *aká ún iglésia, or *ayá ún iglésia are ungrammatical (i.e. never used), and b) the indefinite article ún is allowed to combine with ándi (cp. ándi ún iglésia ‘in, at, to + A church’) but not with akí, aká, ayá, aí.9 As the preceding examples suggest, the Palenquero noun system ultimately operates in ways that are quite different from what has traditionally been reported (see Table 1 above). In my view, in Pal. bare nominal forms are entirely neutral in terms of number or (in)definiteness (readers will recall that in the traditional view, bare neutral forms are said to be singular and definite so that kása equals ‘the house’]). Taking a decidedly (and much welcome) Africanist perspective, Moñino (in press) similarly concludes that bare nouns – the base form – can have singular or plural, and definite or indefinite meanings. Under his (as well as my) revised analysis, THE INTERPRETATION OF A BARE NOUN AS SINGULAR, PLURAL, DEFINITE, AND INDEFINITE IS OFTEN DRIVEN

.

Cuadros are local youth organizations.

. In their locative adverbial (rather than prepositional) meaning, akí, aká, and ayá can of course combine with the definite article to form constructions like akí ún iglésia, aká ún iglésia, or ayá ún iglésia. But in such instances the adverbs akí, aká, and ayá have the original lexical meanings ‘here/there’ (rather than ‘in’). Consequently, akí ún iglésia, for instance, predictably denotes ‘here, a church …’ rather than ‘in a church’ (cp. akí ún iglésia é ún kúsa múy impottánte ‘here, a church is something very important’).

Chapter 7. Bare nouns in Palenquero  ENTIRELY BY CONTEXT.

This would explain, among other things, why decontextualized expressions like puétta ri kása lit. ‘door(s) of house(s)’ can have any of the following meanings (Schwegler, in press a): (7a) (7b) (7c) (7d)

puétta ri kása ‘the door’ ‘a door’ ‘the door’ ‘a door’

or or or or

‘doors of the house’ ‘doors of the house’ ‘doors of a house’ or ‘houses’ ‘doors of a house’ or ‘houses’

Similarly, in answer to a question like “what do you see over there?”, the reply “puetta” (lit. ‘door’) can have any of following meanings (context alone drives definiteness and number): (8)

¿Ké bó a-tá miná aí? — Puétta. ‘What do you see over there?’ — ‘A door; the door; the doors, doors / some doors.’

(singular) (plural)

In Moñino’s and my judgment, the markers ún ‘singular/indefinite’, ma ‘plural/definite’, and ún ma ‘plural/indefinite’ merely add optional information in that they eliminate possible ambiguity (i.e. singular vs. plural). This system differs profoundly from that of Spanish, where number and (in)definiteness are overt, obligatory, and, therefore, predictable components of the overwhelming majority of noun phrases.10 It should be noted, however, that this does not mean that in the majority of cases number and (in)definiteness go unmarked: my corpora suggest that Palenqueros indeed disambiguate this grammatical information quite regularly via ún, ma, ún ma, and so on. In a randomly selected data sample of continuous speech (my corpus), I have examined one hundred noun phrases that, based on discourse content alone, clearly possess a plural meaning, and found the following distribution:11 39% 61%

Ø ma

+ +

noun12 noun

In our revised analysis of non-obligatory number marking there lies a fundamental difference: Moñino and I believe that the motivation for the behavior of ma (or un,

. In some instances, Spanish does allow constructions without an overt expression of indefiniteness: cp. no tengo coche ‘I don’t have a car’ or coche no tengo lit. ‘[topic] car, I don’t have’. . See also the quantitative data in Moñino (in press, note 7); in his sample of 1000 Palenquero noun phrases, 423 are semantically plural, and exhibit the following configuration (readers will note that Moñino’s results differ only slightly from mine): 45% 55%

Ø + ma +

noun noun

. Palenquero verbs lack number inflections, which naturally contributes to the frequently ambiguous nature of number marking.

 Armin Schwegler

un ma) is unrelated to functional, i.e. strictly grammatical considerations. As a result, there exists no predictable correlation between the co-occurrence of bare nouns and reference type. Moñino’s description of ma (in press) goes considerably beyond my own earlier analysis (Schwegler, in press), both in terms of the depth of its analysis and wealth of the examples provided. He too observes that from a semantic perspective, Palenquero bare nouns often have an unmistakably plural meaning, GENERIC (9) as well as DEFINITE (10): (9)

(10)

pa Ø hénde ablálo ké kúsa huéba for people say+it what thing be+PAST ‘so that people say what it was’

(ex. 15 in Moñino)

¡Sí, Ø abíppa á pegá trá ri éle! yes wasp T/A stick after of him/her ‘Yes, the wasps stuck to him (= pursued him)!’

(ex. 3 in Moñino)

Seen from this revised perspective, alternations (common in everyday speech) between hénde (11) and ma hénde (12) below, both meaning ‘(the) people’, may be viewed as simple cases of free variation (when queried, Palenquero creole speakers consistently qualify both constructions as fully grammatical, and they often note, correctly so, that ma hénde “is more common than hénde alone”).13 (11)

(12)

Hénde a miní toabía nu. not people14 T/A come yet ‘People haven’t come yet’ ‘The(se) people haven’t come yet’ Ma hénde a yegá ayé. PL people T/A arrive yesterday ‘People arrived yesterday.’ ‘The(se) people arrived yesterday.’

(A.S.) [indefinite] [definite]

OR (A.S.)

[indefinite] [definite]

OR

. As Moñino (in press) points out, in Spanish, plural nouns (e.g. siglos ‘centuries’, personas ‘persons, people’) are routinely accompanied by a determiner, except when they have an existential meaning, as in hace siglos ‘centuries ago’, faltan personas ‘people are missing’). This is tantamount to saying that a Palenquero expression like hénde á miní toabía nu must be rendered  the definite article: cp. las personas (la gente) no han llegado todavía vs. ungrammatical *Personas no han llegado todavía. . As explained in Schwegler (1993, 2002c), hénde (and ma hénde) can also have the pronominal meaning ‘we’. In Examples (10) and (11), context (not provided here) makes clear, however, that the intended meaning is ‘people’ and not ‘we’.

Chapter 7. Bare nouns in Palenquero 

Examples (11) and (12)15 above naturally raise the question as to what factors currently trigger (or constrain) the inclusion or exclusion of plural ma. On this point, I am in agreement with Moñino (in press) that a.

b.

the OMISSION OF ma (in bare nouns) is linked to the availability of verbal or nonverbal CONTEXTUAL cues (in Palenquero, as in other languages, preceding discourse and context routinely clarify whether a singular/plural and/or definite/indefinite meaning is intended; for a sampling of Palenquero bare nouns with a patently plural meaning, see Moñino’s Examples 8–19); the primary reason for the frequent use of ma + NOUN is almost certainly unrelated to speakers’ need for disambiguation.16

My rationale for the observation in (b) above is somewhat complex, but can in essence be simplified to the following: If the need to disambiguate number were the primary reason for the use of ma, then its relative frequency should predictably become reduced whenever preceding discourse has made explicit the  value of a bare noun. This is, however, not the case, as Moñino’s and my own data clearly demonstrate. Witness, for instance, the excerpt of the Palenquero dialogue reproduced on pp. 67–68 of this study. There, the interlocutors employ a total of eight bare nouns, only one of which is NOT accompanied by the pluralizer ma (see line 9, Ø besíno ‘neighbors’). The flow of the discourse (and the contextual information generated by it) is such, however, that only the opening ma besíno in Ma besíno sí, ¿kúmu sé ñamá? ‘Your neighbors, what are their names?’

(Line 1 in Palenquero dialogue)

is in need of the explicit plural marker ma (without it, the bare besíno sí would be ambiguous, meaning either singular ‘your neighbor’ or plural ‘your neighbors’). Throughout the remainder of the sample dialogue, the discourse content invariably FORCES a plural reading of besíno (or hénde ‘people’), thereby disallowing its singular interpretation. One can argue, therefore, that the presence of ma in lines 2–11 is altogether superfluous (the omission of ma would not have caused ambiguity). These observations lead to an important conclusion: if the primary role of ma were indeed to disambiguate number (i.e. singular vs. plural), then its repeated use in lines 2–11 would be inexplicable from a strictly functional (grammatical) perspective.

. For parallel cases of Palenquero bare nouns, see Moñino (in press), examples (14)–(15) and (28)–(29). . Note that I am, of course,  claiming here that ma does not disambiguate number. I am merely asserting that there must be other causal factors (beyond the need for disambiguation) that trigger the frequent use of ma.

 Armin Schwegler

Palenquero dialogue (excerpt) Note: To further contextualize the data below, see also Moñino’s (in press) useful Table 4, which lists the quantitative distribution of determiner + noun in 1000 Palenquero noun phrases. Three Palenqueros friends (V., A. and R.) participated in the dialogue that is partially reproduced hereafter. Recorded in 1988 at V.’s house, A. opens the segment transcribed below by asking V. and R. to talk about their many neighbors; this then becomes the main topic of their conversation. The excerpt is meant to illustrate that, more often than not, ma is used even when the context amply clarifies the plural reading of the noun. In light of this and similar evidence, it cannot be argued convincingly that the primary trigger for the repeated presence of ma in lines 2–11 is found in speakers’ need to disambiguate number. To facilitate the interpretation of this Palenquero text, I offer English as well as Spanish word-for-word glosses (the Spanish glosses are intended, in part, to assist readers in capturing the predominantly Spanish origin of the Palenquero lexicon). When evaluating the data below, it should kept in mind that all of the highlighted nominal constructions (e.g. ma besínu, or ma hénde) would also be grammatical without the presence of ma. Code-switched Spanish segments are given in “〈…〉”. A:

MA BESÍNO sí, ¿kúmu sé ñamá? vecino tuyo cómo T/A llamar PL neighbor yours how T/A call/name ‘¿Your neighbors, what are their names?’ PL

V:

mí asé ñamá …, 〈la úna〉 asé ñamá MA BESÍNO vecino mío T/A llamar … la una T/A llamar PL neighbor mine T/A call/name one (of them) T/A call/name ‘My neighbors are called … one of them is called PL

Nadzíra. Narcida Narcida “Narcida”.’ R:

Ahá.

V:

〈La ótra〉 asé ñamá Rrósa Miránda. Rosa Miranda la otra T/A llamar the other T/A call/name Rosa Miranda ‘The other is called “Miranda”.’

[       ] R:

Tambié la Méya. también la Mella

Chapter 7. Bare nouns in Palenquero 

also the Mella ‘[There is] also “Mella”.’ V:

MA BESÍNO mí. MA HÉNDE e La Méya; tú ése ser PL vecino mío la Mella todo ese PL gente the Mella all this PL people be PL neighbor mine ‘ “Mella”; all these people are my neighbors.’

[        ,    MÍ P CÑ] V:

MA BESÍNA … María Primítiba Kañáte; ése é mí. mío María Primitiva Cañate eso ser PL vecina Maria Primitiva Cañate this be PL [female] neighbor mine ‘… María Primitiva Cañate; [La Mella, María, etc.] … these are my [female] neighbors.’

R:

Tú ’se; Chó- Fáo. todo ese Cho- Fausto ‘All these [people]; Cho-Fausto [included]. [        ]

R:

Batólo Juliá Teeráng; Minébba Agilá; tú és’ Bartolo Julián Teherán Minerva Aguilar todo ese ‘Bartolo Julián Teherán; Minerva Aguilar; all of them

Bare plural é

BESÍNO

mí.

ser be

vecino neighbor

mío mine

The preceding context clarifies (and indeed “forces”) the plural interpretation of besíno.

are my neighbors’. [        ] R:

Éso é. eso be ‘That’s it [= these are all my neighbors]’.

V:

MA BESÍNO mí. Ma éso é PL eso ser PL vecino mío PL that be PL neighbor mine ‘These are all my neighbors’.

R:

MA BESÍNO ¡Súto é batánte i tá pu akí! vecino nosotros ser bastante que estar por aquí PL neighbor we be numerous that be around here ‘[These are] the neighbors. We are quite a few [neighbors] around here!’ PL

 Armin Schwegler

If we can agree that functional (grammatical) considerations are  the primary trigger for ma, what then prompts its frequent use in unambiguously ‘plural’ noun phrases like those in our sample dialogue? Favoring a predominantly historical and ethnolinguistic approach, Moñino (in press) offers the following hypothesis: during the formative phase of the creole, ma entered the language as a result of Palenqueros’  decision to render their speech more opaque, i.e. less intelligible to Spanishspeaking outsiders. Moñino further reasons that the patently “Africanizing” flavor of ma17 fostered its acceptance in the community, as constructions involving this particle afforded Palenqueros an ever greater sense of social and ethnic identity. The primary function of ma was thus not to indicate ‘plural’, but rather to signal “I am speaking Lengua rather than Spanish”. The growing communicative function of ma then presumably contributed to rendering the creole progressively less intelligible to outsiders, thereby converting it into a secret (or secretive) code – one that had the potential of becoming an outward symbol of cultural as well as linguistic resistance.18 I myself have written on the historical importance of Palenqueros’ penchant for linguistic camouflaging and ethnolinguistic identity building (cp. Morton/Schwegler 2003: 110, 117–118). I am, therefore, especially sensitive to approaches that value the effects of (conscious) linguistic choices, particularly those driven by socio-ethnic and/ or identity considerations.19 That said, I find Moñino’s hypothesis astutely argued, but speculative in nature. To be sure, I can at present offer no convincing evidence to invalidate his hypothesis about the early function(s) of ma, nor can I currently advance an explanation for its vexing behavior. But I do have reservations and questions that, in my estimate, future research will need to address in order render his scenario more compelling. Among these reservations and questions are:20 a.

b.

The historical circumstances surrounding Palenque’s formation continue to be so murky that any attempts at reconstructing the early psycholinguistic disposition of this former maroon community must a priori be deemed highly speculative. Partly because of its structural transparency, ma +  tends to be among the very first (and easiest) grammatical constructions that L2 learners of Palenquero

. Section 5 of Moñino’s article studies the African rather than Spanish origin of ma. The author argues, convincingly so in my opinion, that the Kikongo class prefix ma rather than Spanish más ‘more’ (Megenney 2002: 109–111) is its likely source. . According to Moñino, the use of ma as a “secret” and socio-ethnic particle is also effective in constructions where it combines with other nominal determiners, as in ún ma kása lit. ‘one PL house’ = ‘some houses’, ése ma kása lit. ‘this PL house’ = ‘ these houses’, etc. . For three classic studies on the interrelation between identity, variant choice and language change, see Gal (1979), Hamel (1996), and Labov (1962). . Additional reservations are raised by Salikoko Mufwene in Moñino (in press), Note 6.

Chapter 7. Bare nouns in Palenquero 

d.

acquire.21 As such, the construction seems ill-suited for safeguarding communicative secrecy. If secrecy and social (linguistic) identity were indeed a primary concern in the early formation of the community, would the maroons not simply have resorted to their African languages, especially Kikongo (or a variety thereof) – Palenque’s most important substrate?22 To what extent can (or should) we assume that, in terms of mutual intelligibility with Spanish, 17th-century Palenquero was even remotely similar to its modern counterpart? The question is relevant for the following reason: if the early lexicon and grammar of the creole were, as I tend to believe, considerably more ‘African’ (and, therefore, ‘non-Spanish-like’), then the concomitant unintelligibility between Lengua and Spanish would have made it altogether unnecessary to camouflage the creole with ma.

.

Africanicity of 

c.

A principal merit of Moñino’s paper (in press) is his justified insistence on the fact that ma, in spite of its African origins (cp. ma < Kikongo ma) has a strikingly Spanish-like morphosyntactic behavior. Its word order, for instance, closely matches the sequence in which Spanish aligns its (synthetic) plural morpheme -s whenever a second determiner is present, as can be seen in Examples 13–15 (Spanish word-for-word glosses are provided for ease of comparison). (13)

¿Bó asé komblá ése ma tabáko aónde? tú T/A comprar este PL cigar a+dónde this PL cigar where you T/A buy ‘¿Dónde compras esto+s cigarros?’ ‘Where do you buy these cigars?’

(14)

¿Bó a miná tó ése ma kúsa aí? tú T/A mirar todo ese PL cosa allí all this PL thing there you T/A see ‘¿Has visto todas esta+s cosas allí? ‘Have you seen all these things over there?’

(ex. 42 in Moñino)

(A.S.)

. I here base myself on personal observations, gathered mostly after the mid 1990s, when a more or less constant stream of academic visitors began to flock to Palenque. L2 learners of Palenquero tend to overextend ma by using it (almost) categorically with definite plural nouns. . It is not necessary to presuppose that  or even MOST members of the community shared the same language background. Non-Bakongo members could well have acquired Kikongo as L2, thus rendering the community progressively more uniformly Kikongo-speaking.

 Armin Schwegler

(15)

Kú tú ésema kúsa súto asé- ba salí (A.S.) con todo esePL cosa nosotros T/A T/A salir PL thing we T/A T/A go out with all this ‘Con todas esta+s cosas salíamos.’ Lit. ‘With all these things we used to go out = We used to go out wearing all these things [= clothing]’

Referring to examples like those shown above, Moñino makes the following pertinent observations: Es impactante la observación de que ma, cuando es presente, aparece donde uno esperaría los, las o el plural sin determinante en castellano. Comparte con el má kikongo su forma fónica y un sema parcial ‘plural’, pero ningún otro rasgo morfosintáctico, como, p.ej., el acuerdo. […] Empezamos a sospechar que detrás de la máscara extraña y exótica que exhibe ma, se esconde una cara viva que se parece mucho más a la de los que a la del má- kikongo. […] Aquí también la forma fónica [de ma] despista por su exotismo, pero SU COMun-os, un-as Y LOS EMPLEOS SON PUROS CALCOS SEMÁNTICOS DEL ESPAÑOL. Lo mismo pasa con la combinación ése + ma (42), calco de es-os, es-as.

POSICIÓN ES UN HOMÓLOGO EXACTO DE LA FORMA CASTELLANA

(Moñino, in press; my emphasis) In my corpus, the overwhelming majority of constructions involving ma support Moñino’s claim that the word order of this plural particle appears to have been calqued on Spanish word order. At the same time, my data also sporadically present seemingly “deviant” examples that cannot be readily explained as calques on Spanish constructions. Witness (16) below, for instance, where one would expect ése ma muhére (cp. Span. esa+s mujeres …) rather than ma ése ma muhére (with its reiterated ma … ma). A parallel example is found in (17), where ma ése ma hénde and ma ése ma muhére similarly exhibit a pattern that is not immediately reminiscent of Spanish. (16)

Ma23 ése ma muhére éra. PL this PL women be () ‘It was these women.’

(A.S.)

. I should clarify here that neither external circumstances nor the discourse surrounding this example gave any evidence that the insertion of this (unexpected) initial ma is in any way the result of the speaker’s attempt to reduce the intelligibility of the phrase. As Moñino (Section 6, p. 51–53) correctly mentions, Palenqueros sometimes do engage in intentional camouflaging of their creole speech through the repeated addition of parasitic particles, thereby attempting to exclude outsiders from their communicative interactions. But in Example (15) this was clearly not the case.

Chapter 7. Bare nouns in Palenquero 

(17)

Ándi Inásio, Inásio Tórre, ma24 ése ma hénde, this PL people to Ignacio Ignacio Torres PL ‘We used to go to Ignacio, Ignacio Torres [that is], to all these people, sí, é, Sipriána, prím’ emmána mí, mamá Míngo; yes eh Cipriana first sister/cousin my mother Míngo yes, … eh, to Cipriana’, my (first female) cousin, [that is] Mingo’s mother; ma ése mamuhére […] súto a- sé- ba dí. PL this PL women we T/A T/A T/A go to all these women [we used to go].’

(A.S.)

These and similar examples do not invalidate Moñino’s claim that Spanish rather than Kikongo grammar must have profoundly conditioned the behavior of Pal. ma in its formative phases. But at the same time they also point to the need for additional, more detailed analyses of plural marking in Palenquero. These future studies will want to explain what, if any grammatical, semantic and/or pragmatic differences exist between variable expressions like the sporadic ma ése ma + NOUN and the usual ése ma + NOUN.

.

Conclusions and suggestions for further research

As regards the general nature of the Palenquero article system, Moñino and I have reached a consensus, independently so: in our view, earlier treatments of Palenquero are correct in claiming that the creole can and generally does express number and ±definite in nouns via un (singular indefinite), Ø (singular definite), un ma (plural indefinite), and ma ‘plural definite’. Contrary to earlier claims, we both believe, however, that the use of these articles is conditioned not by grammatical but by other principles. Moñino and I are also in agreement that these articles are non-obligatory, and, therefore, seemingly non-predictable components. Our joint observations also converge in another respect: when ma is present, it can mark generic plurals,25 definite plurals,26 or indefinite plurals27 (contrary to earlier claims, ma +  is thus not

.

The observations in the previous note also apply to this example.

. Cp. ma moná tá má ngánde ku riesiséi ‘youngsters (in general) are bigger once they are sixteen’ (A.S.). . Cp. i tába miná un póko ma muhé aí lóyo ‘I was looking at the women (down there) by the creek’ (A.S.). . Cp. kuand’ úto sé-ba kotá ma ngahíto plánda ‘when we used to cut bunches of bananas’ (Example 37 in Moñino).

 Armin Schwegler

always ‘+definite’). As such, Palenquero bare nouns have general number. They can thus also be labelled as transnumeral. Although Moñino’s and my analysis differ rather profoundly from earlier descriptions, ours does coincide with them on a number of points. Among these is, for instance, the recognition that ma – whenever it is present – explicitly marks the plural, and, in contrast to Spanish, does so without differentiating gender. As Section 3 above explained, Moñino’s and my views diverge in at least one respect: he believes that, historically, the spread of ma was triggered by Palenquero’s early conscious desire to make the language less intelligible for outsiders. Moreover, according to Moñino, the “Africanizing” flavor of ma contributed to making Lengua an outward symbol of cultural as well as linguistic resistance. I have advanced arguments to suggest that this hypothesis, although not implausible, is too speculative without additional evidence.28 And there is also another limitation that will make it necessary for future investigations to closely reexamine the behavior of ma: although it now seems clear that, in the majority of instances, the primary trigger for ma is not related to a need to disambiguate number and/or ±definiteness, it remains altogether unclear what primary factor (or factors) motivates its use in noun phrases that, on account of preceding discourse or context, are already unambiguously plural.29

Abbreviations A.S. ex. lit.

Armin Schwegler (data from his corpus of recordings) example literally

. I should also note here that, from a strictly synchronic perspective, Moñino’s hypothesis – even if eventually proven correct – has no explanatory force: it seeks to describe how ma may originally have acquired its prominence, but says nothing about the decision processes that TODAY account for the frequent presence (or recurrent absence) of ma in unambiguously plural Palenquero noun phrases. . The fact that a FUNCTIONA explanation does not seem to explain the complex behavior of ma ought to be less surprising than it may appear at first. In this context it is useful to recall Labov’s (1994) book-length examination of the principles of language change and the role of functionalism. He finds that functional hypotheses often fail to explain linguistic variation like the one I have assessed critically in this study: [g]iven phonological and morphological variation, the functional hypothesis predicts a tendency for speakers to choose one variant or the other in a fashion that will preserve information. [BUT] MOST OF THE RESULTS CITED SHOW THE OPPOSITE: IN THE STREAM OF SPEECH, ONE VARIANT OR THE OTHER IS CHOSEN WITHOUT REGARD TO THE MAXIMIZATION OF INFORMATION. (Labov 1994: 568; my emphasis)

Chapter 7. Bare nouns in Palenquero 

Pal. PL Span. T/A

Palenquero plural Spanish Tense/Aspect marker

References Corbett, G. G. 2000. Number. Cambridge: Cambridge University Press. Faingold, E. D. 1994. The genesis of the article system in creolization and historical change. Papiere zur Linguistik 50: 52–63. Friedemann, N. S. de & Patiño Rosselli, C. 1983. Lengua y Sociedad en El Palenque de San Basilio. Bogota: Instituto Caro y Cuervo. Fuentes, J. & Schwegler, A. 2005. Lengua y ritos del Palo Monte Mayombe: dioses cubanos y sus fuentes africanas. Frankfurt: Vervuert/Madrid: Iberoamericana. Gal, S. 1979. Language Shift. Social Determinants of Linguistic Change in Bilingual Austria. New York NY: Academic Press. Hamel, R. E. 1996. Language, discourse, and cultural models: Three levels of language shift and maintenance. Southwestern Journal of Linguistics 15: 63–88. Hualde, J. I. and A. Schwegler. In press. Intonation in Palenquero. To appear in Journal of Pidgin and Creole Languages. Labov, W. 1962. The social history of a sound change on the island of Martha’s Vineyard, Massachusetts. Master’s essay. Columbia University. Labov, W. 1994. The Principles of Linguistic Change: Internal factors. Cambridge: Blackwell. Maurer, P. 2002. La pluralización nominal en los criollos atlánticos con base léxica ibérica. In Palenque, Cartagena y Afro-Caribe: historia y lengua, Y. Moñino & A. Schwegler (eds.), 129–136. Tübingen: Niemeyer. Megenney, W. W. 1986. El palenquero: Un lenguaje post-criollo de Colombia. Bogota: Instituto Caro y Cuervo. Megenney, W. W. 2002. Sobre los orígenes de algunas estructuras gramaticales del palenquero. In Palenque, Cartagena y afro-Caribe: historia y lengua, Y. Moñino & A. Schwegler (eds.), 103–115. Tübingen: Niemeyer. Moñino, Y. 1999. L’aspect en palenquero: une sémantaxe africaine. Actances 10: 177–190. Moñino, Y. 2002. Las construcciones de genitivo en palenquero: ¿una semantaxis africana? In Palenque, Cartagena y afro-Caribe: historia y lengua, Y. Moñino & A. Schwegler (eds.), 227–248. Tübingen: Niemeyer. Moñino, Y. & Schwegler A. (eds.). 2002. Palenque, Cartagena y Afro-Caribe: historia y lengua. Tübingen: Niemeyer. Moñino, Y. In press. Convergencias lingüísticas iberocongolesas en palenquero: ¿integrarse a la sociedad mayoritaria o distinguirse de ella? To appear in Language Contact and Language Change in the Caribbean and Beyond — Lenguas en contacto y cambio lingüístico en el Caribe y más allá, W. Mihatsch & M. Sokol (eds.). Bern: Peter Lang. Morton, T. 2005. Sociolinguistic Variation and Language Change in El Palenque de San Basilio (Colombia). PhD Dissertation, University of Pennsylvania. Morton, T. & Schwegler, A. 2003. Vernacular Spanish in a microcosm: Kateyano in El Palenque de San Basilio (Colombia). Revista Internacional de Lingüística Iberoamericana (RILI) 1: 97–159.

 Armin Schwegler Pfleiderer, B. 1998. Sprachtod und Revitalisierung der Spanisch Basierten Kreolsprache Palenquero (Kolumbien). (Romanistische Linguistik 7). Berlin: Freie Universität Berlin. Rullmann, H. & You A. 2006. General number and the semantics and pragmatics of indefinite bare nouns in Mandarin Chinese. In Where Semantics Meets Pragmatics: Current Research in the semantics/pragmatics interface, Heusinger, K. von & Turner, K. (eds.), 175–196. Oxford: Elsevier Science. Schwegler, A. 1993. Subject pronouns and person/number in Palenquero. In Atlantic Meets Pacific: A Global View of Pidginization and Creolization, F. Byrne & J. Holm (eds.), 145–161. Amsterdam: John Benjamins. Schwegler, A. 2002a. Reconsidering the evidence: Bare nouns in Palenquero and what they really mean. Paper presented at the Annual Meeting of the Society for Pidgin and Creole Linguistics, San Francisco, Jan. 4–5, 2002. Schwegler, A. 2002b. El vocabulario africano de Palenque (Colombia). Segunda Parte: Compendio de palabras (con etimologías). In Palenque, Cartagena y afro-Caribe: historia y lengua, Y. Moñino & A. Schwegler (eds.), 171–227. Tübingen: Niemeyer. Schwegler, A. 2002c. On the (African) origins of Palenquero subject pronouns. Diachronica 19(2): 273–332. Schwegler, A. 2006. Bantu elements in Palenque (Colombia): Anthropological, archeological and linguistic evidence. In African Re-genesis: Confronting Social Issues in the Diaspora, B. Haviser & K. C. MacDonald (eds.), 204–222. London: University College London Press. Schwegler, A. In press a. Palenquero (Creole Spanish). In Comparative Creole Syntax, J. Holm & P. Patrick (eds.). London: Battlebridge. Schwegler, A. In press b. A fresh consensus in the making: Plural MA and bare nouns in Palenquero. In Language Contact and Language Change in the Caribbean and Beyond — Lenguas en contacto y Cambio lingüístico en el Caribe y más allá, W. Mihatsch & M. Sokol (eds.). Bern: Peter Lang. Schwegler, A. & Kempff, J. 2007. Fonética y fonología españolas. 3d edn. New York NY: Wiley and Sons. Stolz, T. 1998. Romanische Kreolsprachen II. Portugiesische Kreolsprachen. In Lexikon der Romanistischen Linguistik, G. Holtus (ed.) (in collaboration with M. Metzeltin and C. Schmitt), 618–637. Tübingen: Niemeyer.

 

French-lexified Creoles

 

Bare nouns in Réunionnais Creole Robert Chaudenson This paper examines the determiner system of Réunionnais Creole (RC) with a special focus on the use, interpretation and distribution of its bare nouns. The synchronic part of the study involves a comparison of bare nouns between RC and Modern French whereas the diachronic section considers RC and Middle French in an attempt to determine the origins of RC bare nouns and whether there are uses proper to RC. On the one hand, this paper uncovers convincing evidence tracing the use and distribution of bare nouns in RC to Middle French. On the other, it highlights the full complexity of the situation by taking into account the many variants spoken at the time when RC emerged, as well as the current situation of diglossia with French.1

The nature and situation of creole linguistic systems make it difficult to give a precise description of bare nouns in creoles. Although these languages are autonomous with respect to the European varieties from which they borrowed many of their linguistic features, they often coexist with these same European languages. Such situations of diglossia tend to favor changes and evolutions that result in complex and unstable linguistic facts, as this paper will illustrate. Although I will briefly examine the overt determiner system of Réunionnais Creole (henceforth, RC), the main focus of this paper is the use, interpretation and distribution of its bare nouns. This three-layered examination will include a comparison with respect to the occurrence of bare nouns between RC and Modern French and RC and Middle French and it will also try to reveal whether there are uses of bare nouns proper to RC.

. I thank Marlyse Baptista and Jacqueline Guéron for their comments on this paper. I also thank Marlyse Baptista for translating this chapter.

 Robert Chaudenson

.

Determiners in Réunionnais Creole

Before tackling the question of bare nouns, let us briefly present the determiner system in RC. This presentation is limited to ‘articles’, including the ‘zero determiner’. It does not take into account possessives, for instance, the use of which would exclude the zero determiner. Together with Louisianais, RC is undoubtedly the French creole that includes the most material inherited from the French system of grammar. The singular definite denotation is realized sometimes with le (close to French), sometimes with lo. In the latter case, the choice of the orthography can be accounted for, originally, by the deliberate attempt on the part of creolophones to deviate from French. The real contrast resides between le and la. These two determiners are in complementary distribution; la appears in front of words that used to be feminine in French.2 It is worth noting that some words have switched gender from feminine to masculine in Modern French while the original feminine gender present in 17th century French was retained in RC, for instance, la sab (now le sable) ‘the sand’; la poizon (now le poison) ‘the poison’, etc.… Non-French words may receive the feminine marker due to paronymic resemblance to other French words. For example, tant, which refers to a kind of basket made of braided vacoa and originates from Malagasy tanty, is feminine due to paronymic affinities with French tante ‘aunt’ or tente ‘tent’; the same goes for the word touk ‘hearth’ originating from Malagasy toko but associated with French ‘touque’. The definite plural lé is a rather acrolectal determiner. There are in RC two varieties, one acrolectal, closer to French phonetic and morphosyntactic features, and the other basilectal. The lexicon is, however, common to both. The dual determiners (singular and plural, le koté and lé dé) can only be used for entities that come in pairs (ears, eyes, socks, etc…); if one is dealing with things that do not form pairs, even if there are two of these items, one uses ordinary markers. The use of indefinite determiners is fairly straightforward, as is illustrated in Table 1. Table 1 summarizes the essential facts about the Réunionnais determiner system. The bare noun is not represented in this table but will be the focus of the next section. Table 1. The determiner system of Réunionnais Creole

Singular Plural The/one of 2 Both

Definite

Indefinite

le/lo la le bane, lé le koté (d) lé dé

in in bane in koté (d) dé

. However, the la determiner does not involve, except in frozen expressions, agreement with the adjective; as a result, one cannot really speak of gender, as in French.

Chapter 8. Bare nouns in Réunionnais Creole 

. .

Bare nouns in RC Some observations

Before providing an exhaustive list of the contexts in which RC uses bare nouns and pointing out the parallelisms with Modern French whenever relevant, I will begin with a comment or two on the sociolinguistic situation of la Réunion. The sociolinguistic situation of la Réunion has always been complex due to the existence of two varieties of creole that can either be viewed as sub-systems or as the acrolectal and basilectal poles of a continuum. Moreover, for the past forty years, economic and social developments have triggered significant population shifts and strong urbanization on the island. Such trends have blurred distinctions between varieties of creoles and have contributed to increasing the internal variation of the linguistic system. This variation is made all the greater by the strong influence of French through massive (albeit relatively inefficient) schooling and above all by the general and intense consumption of French TV programs. As a result, linguistic descriptions such as those of Chaudenson (1974) or of Cellier (1985) (whose corpus is primarily based on the inquiries of the Linguistic and Ethnographic Atlas of the Reunion) have become historical records of Réunionnais Creole which do not necessarily account precisely and in detail for the current situation and uses. Consider (1): (1)

a.

b.

c.

le pyé d bwa the foot of tree ‘the tree’ in pyé d bwa a foot of tree ‘a tree’ pyé d bwa foot of tree ‘trees, a tree, the tree (in general)’

While (1a) gets a definite reading by virtue of the definite determiner and (1b) an indefinite interpretation due to the presence of the indefinite article, the bare nominal in (1c) may be interpreted as generic or specific singular or plural. Before moving onto the classification of bare nouns, I would like to echo Cellier (1985) in making a crucial distinction between the ‘zero determiner’ and the ‘absence of determiner’: while (2) illustrates a zero determiner, the disappearance of the determiner in (3) via an erasing rule is triggered by the presence of a postposed deictic element. (2)

mi manz pwason (zero determiner)

I eat fish ‘I eat some fish or I eat fish.’3 . The second translation obtains when the speaker implies that s/he eats fish in general because s/he does not have any allergies or food prohibition.

 Robert Chaudenson

(3)

mi manz pwason la (absence of determiner due to deictic)

I

eat

fish

this

‘I am eating this fish.’

The presence of another determiner, such as the possessive, also triggers the disappearance of the marker originating from the French article except, of course, in cases in which this marker is agglutinated to the noun, as in mon loto (my car), or mon dri (my rice). In the nine-category classification offered below, one should keep in mind that within the same utterance, a noun may appear with or without a determiner.4 . While I classify bare nouns into nine distinct categories, Cellier identifies seven types of uses of the zero determiner and Staudacher-Valliamée notes three of them. Their respective categorization can be summarized as follows. Cellier (1985: 315) notes that the zero determiner often appears in prepositional frozen expressions of time, of place, of manner, and of matter. Bare nouns also occur in verbal expressions as in (i), adverbial expressions as in (ii), enumerations of partitives as in (iii), in interpellations as in (iv), in characterizing oppositions as in (v) and in proverbs or sayings as in (vi): (i)

(ii)

(iii) (iv)

(v)

(vi)

koz an kamarad talk as friend ‘to talk as friends’ gran matin early morning ‘early in the morning’ boi d golèt ‘wood of gaulette’ konpèr get sa pal, look at this ‘pal look at this’ lanp pétrol lamp oil ‘oil lamp’ mariaj lé pa badinaj marriage is NEG joke ‘marriage is no joke.’

In contrast, Staudacher-Valliamée classifies the use of bare nouns in three main categories that include names of stars referring to a known reality, as in (vii), proverbs in which the noun has a general or truth value, as in (viii), and cases in which the referent can be easily identified, as in (ix): (vii)

(viii)

(ix)

Soley i pwak zordi sun is scorching today ‘the sun is scorching today.’ Baton i menn par le bout stick is led by the end ‘one respects the elders’ Bisik la ponkor monté ‘It is not yet time for the bisiks.’

Chapter 8. Bare nouns in Réunionnais Creole 

.

Types of bare nouns

.. Proper nouns, anthroponyms and toponyms Proper nouns as well as common nouns occur bare when they are anthroponyms, toponyms, kinship terms, or profession labels. Nouns of beings or unique places appear bare, irrespective of syntactic position and function.5 In listing these examples, I will point out whether or not the same use of bare nouns obtains in Modern French. In (4), both the proper noun Orélyin and the occupation name mason appear bare. The same holds for Modern French. In (5), the kinship term gran mèr is bare, just as in Modern French. However, in (6), in the expression lakès agrikol both the preposition à ‘to’ and the determiner la ‘the’ necessary in Modern French are missing. RC lakès is generally not considered to correspond to Modern French la caisse. In (7), the preposition and determiner required in Modern French are also absent. From these data, it would seem that all nouns governed by V, as in (4), (6) and (7), may appear bare in RC, contrary to Modern French. This observation will be shown to be generalizable to nouns governed by P when we consider bare nouns preceded by a preposition in Examples (26), (27), (28) and (29) further down: (4)

Orélyin lété mason Aurélien was mason ‘Aurélien was a mason.’

(5)

Gran mèr navé touzour en ti friandiz Grandma had always a little treat ‘Grandma always had a little treat.’6

(6)

Lété ki travay lakès agrikol he that work bank agricultural ‘He worked at the Agricultural Bank.’7

(7)

alé bazar go market ‘to go to the market.’

This distribution of bare nouns was already mentioned in Chaudenson (1974), where a brief “grammatical note” pointed out that bare nouns appear more frequently as direct object complement, prepositional or non-prepositional NPs.

.

Bare nouns are in bold face.

. In the case of kinship terms, common nouns such as Uncle, Grandma, etc. can also be treated as anthroponyms. . Agricultural Bank is a translation for Caisse Agricole, a French chain of banks.

 Robert Chaudenson

.. Appelatives Appelatives also appear bare in RC, contrary to Modern French that requires a determiner. The Modern French equivalent of (8a) is Les enfants!8: (8)

a.

b.

Marmay! kids ‘kids!’ Vou, gran moune lontan! you great people long ago ‘you, the old people from long ago!’

.. Collectives and generics Collective and generic nouns may appear bare, as shown in (9) and are particularly prevalent in proverbs, as illustrated in (11). With regard to (10b) and (11d), it is worth noting that RC seems to favor a bare noun whenever the noun is in a topic position signaled by the predicate marker i. Such a marker is also found in a variety of other creole languages (i.e. Vanuatu). The same goes for mass nouns like lo sal ‘dirty water’ in (39) or generic lera ‘the rat/rats’ in (40) respectively, both of which are in topic positions. (9)

a.

b. (10)

a. b. c.

(11)

a. b. c.

Gran diférans ant marmay lontan é marmay koméla great difference between children past and children now ‘There is a great difference between today’s youth and the youth from long ago.’ ou sa la antèr zesklav nana bébèt ‘The place where slaves/the slaves were buried, there are ghosts.’ rod zerb pou zanimo ‘to fetch grass for animals’9 koson i manz pa lswar ‘the pig/pigs do(es) not eat in the evening.’ dan l tan soulyé navé poin ‘in the old days, shoes, there were none.’ gran moune lontan mantèr people in the old days were liars. maryaz napa badinaz marriage is no joke baton la sanz de bout stick  change of end the stick changed end

. It would seem, however, that bare appelatives may be inherited from Middle French where it was common to use appelatives such as bonnes gens! ‘good people!’ .

Note that zanimo can get a specific or non specific interpretation according to the context.

Chapter 8. Bare nouns in Réunionnais Creole 

d.

fourgon i mok la pèl poker it make fun the shovel ‘The poker makes fun of the shovel.’

Modern French exhibits bare nouns in a number of proverbs (inherited from Middle French), such as Pierre qui roule n’amasse pas mousse ‘Rolling stone gathers no moss’ or chat échaudé craint l’eau froide ‘once bitten, twice shy’. However, collective and generic bare nouns as they appear in the RC examples in (9) and (10) are not grammatical in Modern French.

.. Use of anaphoric nouns Anaphoric nouns, that is, nouns that have been mentioned in the previous discourse, also appear bare, as shown in the following examples. This use is distinct from Modern French which requires an overt determiner. (12)

dokter la di ael doctor told him ‘The doctor [that we already talked about] told him…’

(13)

la zourné demoune, le swar koson the day human the evening pig ‘during the day a human, during the night a pig’10

(14)

i koup in pyé d bwa…pyé d bwa i tonb he cuts a foot of tree foot of tree it falls ‘He cuts a tree… the tree falls.’

.. Attributes Attributive predicates also appear bare. Depending on the type of predicate, the use may be similar or different from Modern French. For instance, attributive predicates involving names of occupations/professions such as mason ‘mason’ in (15) and métrès d ékol ‘primary/elementary school teacher’ in (16) appear bare both in RC and Modern French. Note, however, that nominal predicates involving other types of common nouns such as marmay malin ‘clever kid’ may appear bare in RC, as shown in (17), contrary to Modern French where a determiner would be required. (15)

son garson lété mason his son was mason ‘His son was a mason’

(16)

lété métrès d ékol was teacher of school ‘She was an elementary school teacher.’

.

This is the story of a woman who would change into a pig at night.

 Robert Chaudenson

(17)

moin lété marmay malin I was kid clever ‘I was a clever kid.’

.. Complement of the noun Complements of nouns typically appear bare, contrary to Modern French that would require in these particular cases both a preposition and a determiner (for instance, the Modern French equivalent of (18) is sur le bord du (de + le) chemin): (18)

Su l bor semin on the side road ‘On the side of the road.’

(19)

Rod mazé bèf fetch food ox(en) ‘To fetch food for oxen.’

.. Adjectivized or epithet nouns11 Cases of a nominal functioning as an epithet modifier to another noun are quite frequent in RC (cf. Chaudenson (1974: 1004–1005), as illustrated in (20)–(23). This does not occur in Modern French. (20)

bal maryaz ball wedding ‘a wedding ball’

(21)

batay kok fight cock ‘a cock fight’

(22)

bébèt koko insect coconut ‘coconut insect’12

(23)

kapisin kato capucin cateau ‘cateau fish’13

.

This expression is borrowed from Noailly (1990).

.

Insect that is found in coconut trees.

. The capucin cateau fish is called this way due to its colors, as cateau is the Old French name for parrot.

Chapter 8. Bare nouns in Réunionnais Creole 

One also finds expressions featuring the determiner with the noun as in: (24)

baton la kol stick the birdlime ‘birdlime stick’14

(25)

pès la kot fish the coast ‘to fish near the coast’ or ‘to fish from the coast’15

.. Prepositional complements Complements following a preposition typically occur without a determiner in RC, contrary to Modern French. (26)

rod manzé pou kabri fetch food for goat ‘to fetch grass for the goat(s)’16

(27)

dan bitasyon in field ‘at the fields’

(28)

i koud ek fil kadèr one sew with thread aloe ‘one sews with an aloe thread.’

(29)

amène isi dan sal vert17 bring here in trellis ‘to bring in the green trellis.’

.. Direct verbal complements One finds bare nouns in the position of direct object complements as well, contrary to Modern French. (30)

I falé nou plant mai, manyok, patat it needed we plant corn manioc sweet potato ‘We needed to plant corn, manioc, sweet potatoes.’

. A small stick covered with birdlime for capturing small birds. . The primary meaning of this expression is “to fish near the coast or from the coast”, the secondary meaning is, however, more blurry, as kot may come from “cotte” (piece of clothing), in which case, one means that one is fishing with the cotte, a piece of clothing with which one catches larvae (cf. Chaudenson 1974: 1004–1005). . In this example, kabri could be interpreted as generic or specific, singular or plural, according to the context. .

Sal vert is a green trellis built for weddings.

 Robert Chaudenson

(31)

fé kwui manzé make cook food ‘to prepare food’

(32)

nou té pran bwat sardine we  take can sardine ‘we would take a sardines can’

(33)

i bat tanbour… he beat drum ‘He beats the drums.’ li fé paryaz he make bet ‘He makes a bet.’

(34)

Table 2 highlights the similarities and differences in the uses of bare nouns between RC and Modern French: Table 2. Similarities and differences in the uses of bare nouns between RC and Modern French

Proper nouns Professions Kinship terms Toponyms Appelatives Collectives and generics Anaphoric nouns Nominal predicates (other than professions) Complements of the noun Adjectivized or epithet nouns Prepositional complements Direct object verbal complements

RC

Modern French

yes yes yes yes yes yes yes yes yes yes yes yes

yes yes yes yes no no no no no no no no

This table is a first step towards distinguishing between the similar and different uses of bare nouns in RC and Modern French. The next steps will be to identify the use of bare nouns inherited from XVII century French and to raise the question whether there are uses proper to RC. But first, I will propose a classification and interpretation of the facts.

. .

Classification and interpretation of the facts The generic in RC

The most obvious case of the generic is usually found in general formulas including proverbs or frozen expressions with a general value. For instance, in my dissertation

Chapter 8. Bare nouns in Réunionnais Creole 

(1974: 310–328) where I collected 202 formulas of this type,18 a quick survey of these formulas leads to some conclusions that could not have otherwise been drawn from the reduced corpus offered in this paper. Out of a hundred frozen expressions starting with a nominal, more than two thirds (exactly 68 out of 96) feature a bare noun: (35)

papang i manz tektek the papangue eats the tectec19

(36)

gro pwason i bèk sir le tar big fish(es) bite late

The other cases (28 out of 96) are not as straightforward. The indefinite is present in five cases: Examples 101, 151, 150, 168, 194 of my dissertation. Case 101 ène zèf in bousé, di zèf in gouté ‘one egg one mouthful, ten eggs a meal’ is only apparently relevant as, in fact, ène in Réunionnais is not an indefinite determiner but a numeral. Out of the four remaining relevant cases, only two clearly illustrate indefinites: (37)

in békèr d klé na pwin kamarad a poor has no friend ‘A poor has no friends.’

(38)

in mus vèr i gat in fer blan/ barik d salé a fly worm it spoils a can / drum of salted ‘A worm fly spoils a can or drum of salted meat.’

Examples of nouns starting with l- or preceded by la or le are more difficult to analyze. The analysis of these elements as definite determiners presents a problem above all in frozen expressions that can be traced back to an old stage in the language. In other words, it is risky to analyze lo in (39) and le in (40) as a determiner + noun, as a definite interpretation does not obtain: (39)

lo sal i sali lo prop water dirty it dirties water clean ‘Dirty water dirties clean water’

(40)

le ra/lera i mont pas sir ferblan the rat it climbs not on can ‘The rat does (or rats do) not climb on cans.’

.

In my dissertation, they are numbered from 1 to 202.

.

The papangue is a bird of prey and the tectec a small bird.

 Robert Chaudenson

One may hesitate about how to define and analyze the members of the nominal phrase in (39) and (40) above, since such expressions are in fact comparable to (41), which contains bare nouns: (41)

ra miské i mont pas sir ganblon rat musk it climbs not on ganblon ‘The muskrat does not climb on the ganblon.’20

I noted in my dissertation that the agglutination of the determiner to the noun seems to have mostly disappeared from RC in the course of the XIX century due to what I called “basilectal erosion” (Chaudenson 1981). Given the continuum situation of la Réunion, I qualify as such linguistic changes that trigger the disappearance of some basilectal variants. There is still some residue of such agglutination in numerous words of French origin that are endowed in creole with a prosthetic z: zanfan ‘children’, zariko ‘beans’. Some take a prosthetic l : loto, ‘car’. Some residual traces can also be seen in words such as dri ‘rice’, dlo/delo ‘water’, dlé ‘milk’, dpin ‘bread’, debwa ‘wood’; there are, as well, rare uses of forms such as mon lakaz ‘my house’, mon lepyé ‘my foot’, mon latèt ‘my head’ (Chaudenson 1974: 349). I also noticed (ibidem: 357) a certain tendency to maintain la or le before monosyllables, as in (42a) as opposed to (42b): (42)

a.

b.

mi manz le sou I eat the cabbage ‘I eat the cabbage.’ mi manz manyok I eat manioc ‘I eat manioc.’

The sensitivity of the agglutination process to the syllabic structure of a given word is further illustrated by the following expressions: la tèr, la mor, l dyab (‘the devil marries his daughter’), la ké (d syin) ‘the tail (of the dog)’, le syin ‘the dog’ and la sans ‘luck’. The agglutinated determiner appears in all cases on monosyllabic nouns. The same syllabic variable applies to proverbs as well. Consider (43) and (44), as they illustrate interesting facts: (43)

pasyans i géri la gal patience it heals the plague ‘Patience cures the plague.’

(44)

a.

b.

fourgon i mok lapèl poker it make fun shovel ‘the poker makes fun of the shovel.’ mang karot la di bilimbi lé èg mango carrot said bilimbi is tart ‘The carrot mango21 says that the bilimbi [tart fruit] is tart.’

. A gamblon or gamblo is a metal container that is derived from the French word gamelle. .

The carrot mango is a mango of poor quality and not very sweet.

Chapter 8. Bare nouns in Réunionnais Creole 

In (43), la gal must be used and not gal. The same goes for la pèl and not pèl in (44a). The meaning of the proverbs in (44a) and (44b) is equivalent but what (44b) illustrates is that a word such as bilimbi is not preceded by an article because it is not monosyllabic. I tend to believe that these formulas can nearly all be viewed as involving a noun without a determiner. Some of them clearly come from the French tradition, as in la pelle se moque du fourgon; not surprisingly, the word fourgon does not appear anywhere else in the creole lexicon. The elements that precede the noun are in fact pseudo-determiners that may have become agglutinated to the noun in the early stages of creolization. Of course, one may still have questions regarding the true status of the agglutinated determiner: is it a true determiner or is it a prefix without any semantic import?

.

Evidence of the relationship with the referential context

Cellier quotes a general observation from Galmiche that deserves attention: “Ce ne sont pas les expressions linguistiques qui “réfèrent”, mais ce sont les locuteurs eux-mêmes”, translatable as “It is not the linguistic expressions that ‘refer’ but rather the speakers themselves” (Galmiche, in Sémantique générative 1975: 276). In fact, as already noted by Cellier in a general remark, in RC, one also has the zero determiner when, for the speaker, the relationship between the lexical unit and the referential context is obvious. This idea can be found, with an analogous formulation, in Staudacher-Valliamée (2004), as she notes that one of the cases where the zero determiner occurs is when the referent can easily be identified. It should be emphasized, however, that the observation that the noun may appear bare when reference is provided by the relevant context does not account for why the pragmatic situation justifies the use of a bare noun in RC and not in Modern French.

.

Towards an explanation: RC bare nominal expressions originating from Middle French?

At the risk of being labelled a ‘superstratist’, I would say that a number of points can be clarified by examining the French determiner system (its history and evolution), without of course, questioning for a single moment the linguistic autonomy of RC. Historically speaking, the frequency of bare nouns in French proverbs is inherited from Old French; however, one also finds them in a large number of expressions that became frozen and still appear as bare nouns in Modern French: faire mal ‘to hurt or to get hurt’, prendre forme, ‘to take shape’, donner vie ‘to give birth’, donner lieu ‘to give way’, prendre conscience ‘to become aware’, faire confiance ‘to trust’ etc… In all cases, these ‘light’ verbs are combined with abstract bare nouns in Modern French.

 Robert Chaudenson

However, Modern French does not allow chercher herbe ‘to fetch grass’ contrary to RC rod zerb (as in (10a)). Hence, one should make a distinction between ‘light’ verb structures involving abstract nouns that will typically appear bare in Modern French and non-light verbs combined with concrete nouns that require an overt determiner in Modern French but none in RC. André Valli who took his Réunionnais examples from Cellier’s dissertation, has written an excellent article on the productive use of the zero determiner in Middle French and in RC. His analyses converge, for the most part, with my own viewpoint elaborated in Chaudenson (1981, 1985). For the French spoken in the 15th and 16th centuries, the main uses of the zero determiner can be characterized as follows, according to Valli (1994: 90–92): “the zero determiner competes with the definite article before concrete or abstract nouns of unique entities […] It does not matter whether the noun appears in subject or object position.” The same goes for plural nouns, as in (45): (45)

De quelque manière que on combate gens… of whatever way that we fight people ‘in whatever way we fight people’.

The general consensus is that this determiner generally appears in frozen expressions that are like proverbs: (46)

Necessité fait gens mesprendre. necessity make people mislead ‘Necessity misleads people’

(47)

Faim saillir le loup du bois. hunger go out the wolf from woods ‘Hunger makes the wolf come out of the woods.’

It has been noted that some syntactic environments favor the zero determiner: coordination, apposition, restriction, comparison, attribution, as well as the position after prepositions. Valli (1994: 90–91) notes that the alternation between the zero determiner and the indefinite or definite determiners is neutralized and the bare noun is of significant frequency in the syntactic position of direct object of ‘light’ verbs such as avoir ‘to have’, faire ‘to do’, mettre ‘to put’, prendre ‘to take’, tenir ‘to hold’… Thus, the competition between overt determiners and their null counterpart does not take place with every kind of verbs and in every syntactic context. Indeed, regarding the latter point, Valli notes that the zero determiner appears with modifying elements: the zero determiner typically occurs with preposed adjectives much more often than with postnominal adjectives. He notes, however, that when the modifying element is a relative clause or a superlative clause, the determiner is obligatory. In summary, the structures involving light verbs in the 15th century make use of the zero determiner overwhelmingly. Although the zero determiner is in competition

Chapter 8. Bare nouns in Réunionnais Creole 

with overt determiners in other areas of the grammar, Valli notes that the type of verbs and the type of complements condition the use of bare nouns (i.e. light verbs select bare nouns). The subsequent widespread use of the determiner gradually eroded such a distinction that only survived in a limited set of ‘fossilized’ structures. It is clear that in the history of French between the XIV and XVII centuries, the use of the zero determiner is regressing, perhaps due to the increase in the use of definite and indefinite articles. Convincing evidence for this pattern may be drawn from Marchello-Nizia’s (1979) study of earlier stages of the French language (XIV–XV centuries). The author states that in Middle French, there are a number of cases where, in the majority of texts, the noun is used without a determiner; this phenomenon is linked either to the syntactic function of the noun, or to the semantic category to which it belongs. Quite often, a noun referring to a more or less abstract general notion is not preceded by a determiner when it functions as a subject: et sembloit que foudre cheist des ciex, “and it appeared that lightning was falling from the skies.” It is also the case that bare nouns appear in direct object position or as a prepositional complement (Marchello-Nizia 1979: 111). Marchello-Nizia did provide some statistical results that I cannot reproduce in detail here. If one sets aside the texts in verse, in the Fouke Fitz Warin, a novel from the XVI century, 51% of the nouns are without determiners, 21% appear with ith le and no evidence of indefinite determiner could be found (ibidem: 112). The author adds, regarding the indefinite determiner un, that some texts (including the one I have just mentioned) ignore it completely (ibidem: 125). With the indefinite plural, just like Old French, the most frequent case is the absence of a determiner, but de/des start to appear to express indefinites; it is, however, “confined to being used before direct object complements.” (my translation) (Marchello-Nizia 1979: 125). The situation Marchello-Nizia describes is, therefore, not very different from that in RC, as confirmed by Valli’s observation. As mentioned above, even if the article has become obligatory in a number of modern uses, certain frozen expressions have been preserved to this day in Modern French: prendre forme ‘to take shape’, perdre pied ‘to lose footing’, faire mal, ‘to hurt’ etc. I will content myself with quoting the last paragraphs of Valli’s conclusion regarding bare nominals: “Ce que les faits relevés par P. Cellier, qu’une étude plus détaillée devra confirmer, indiquent c’est que le créole réunionnais maintient l’emploi du déterminant zéro dans les mêmes contextes grammaticaux que le français des XVe et XVIe siècles” (1994: 100).

This is translatable as follows: “What the facts collected by Cellier reveal, and that a more detailed study will need to confirm, is that Réunionnais Creole maintains the

 Robert Chaudenson

use of the zero determiner in the same grammatical contexts as the French spoken in the XV and XVI centuries.” In other words, bare nouns appear when they are directly governed by a preposition or a verb (including ‘light’ verbs). I would add two remarks. The first, that I have often made in the past, is that one must be cautious with the dates that historical linguists give us, as they unavoidably refer exclusively to written and even literary language. Numerous examples prove that in some regions there exist, between literary usage and ordinary French, temporal gaps that may be centuries long. I will mention here only one example, as a lot has been written on the topic in the field of creolistics: the reflexive marker corps ‘body’. For instance, tuer son corps ‘to kill one’s body’ means ‘to kill oneself ’ in the literal or figurative sense. This expression may be found in all French-based creoles and is supposed to be of African origin as, according to historical linguists, it had already disappeared at the moment of colonization in the XVII century. Indeed, if one believes Brunot, “[au XVIe siècle] la vieille locution “son corps” cesse peu à peu d’être usitée comme substitut des pronoms ”. (“[in the XVI century], the old expression “one’s body” ceased little by little to be used as a substitute for pronouns” (Histoire de la langue française, II: 414). However, dictionaries dating back to the end of the XVII century (Furetière 1690, Académie 1694) in fact cite this type of expression; even better, Wartburg (1928: II/2) shows that this expression still appears in modern times in regional varieties of Western France. These facts are particularly interesting for creole genesis as the colonizers came, for the most part, from those areas. For instance, in Saint Seurin (Lower Charente), speakers say porter son corps, “to carry one’s body”. Hence, Brunot was misled not only about Standard French, as the dictionaries of the end of the XVII century prove, but also about the regional varieties of the language (cf. Wartburg 1928). One can reasonably think that in the language of the settlers, the use of bare nominals lasted (while obeying specific rules) much beyond the dates that the official history of the French language gives us. The second remark is that Modern French itself displays a tendency to use bare nouns. Noailly (1990), studied this issue, focusing on the construction N1 + N2 of the type pause café, ‘coffee break’ or question loisirs, glossable as ‘question leisure’ but translatable as ‘about leisure’. The relevant point here is that no preposition intervenes between N1 and N2 and that no determiner appears before N2. This construction is so common that the author found and collected several thousands of examples. Some N1 tend to function like prepositions (question travail, “about work”). Noailly rejects the idea that there could be an Anglo-Saxon influence and quotes old examples from French; she argues that the great period of expansion of the construction N1 + N2 occurs in the middle of the 19th century and that it is found as frequently in the literary language as in popular French (1990: 209). I do not wish to stretch the comparison too far but I simply want to point out that the French language has found in this manner a new way of exploring the possibilities that the bare noun offers.

Chapter 8. Bare nouns in Réunionnais Creole 

.

Conclusion

The use of the bare noun in RC is obviously not the same as in Modern French, as discussed above.22 In contrast, the Réunionnais system seems relatively close to the state of older varieties of French (XIV–XVI centuries) which is likely to have persisted in ordinary French beyond the dates proposed by the ‘official’ history of the French language which is based exclusively on the literary language. The use of the bare nominal depends less on semantic criteria than on syntactic position or morpho-phonetic constraints; for example, as discussed in Section 3.1, the zero determiner is scarce before monosyllabic nouns but prevalent before polysyllabic nouns. Here too, one must remain cautious, as one is in some cases in the presence of obvious residues of the Middle French system. Indeed, one encounters facts that could be interpreted either as residues or as innovations, no one knows for sure. Hence, the tendency, noted above, towards displaying overt determiners on monosyllabic nouns whereas, in the same function, a polysyllabic noun can remain bare23 may attest in fact to an older state of the French language. It has actually been observed that up to the XIX century, the agglutination of the determiner to the noun was much more common than it is today. Valli (1994) provided a thorough comparison between 15th century French and RC and could not conclude categorically that the two systems were distinct from each other, quite the contrary. What adds to the complexity of such an endeavor is the number of regional variants spoken at the time. Not surprisingly, the study of the evolution of the zero determiner from a historical perspective, displays, particularly in the 14th and 15th centuries, a lot of variation reflected in the descriptions of the grammarians of the time. The current sociolinguistic situation in la Reunion also triggers language evolution due to more and more intensive contact with French, whereas a few decades ago, the use of the determiner examined in this paper, had not evolved since the preceding century. Recall, as was already mentioned, that Cellier’s (1985) thesis was based on the investigations of the Atlas linguistique et ethnographique de la Réunion. The Atlas started in 1972 and was conducted using monolingual creolophone informants the majority of whom had been born during the first quarter of the XX century.

. Note also that Modern French has a set of expressions that alternate between a bare form and a non bare form: prendre forme ‘to take shape’ vs. prendre une forme ‘to take a shape’; avoir loisir de ‘to have leisure to’ vs. avoir le loisir de… ‘to have the opportunity to’, etc. Attributive predicates also display such an alternation: Il était maçon, literally, ‘he was mason’ vs. C’était un maçon ‘it was a mason’. . Such a tendency is displayed by the following examples: mi manz le sou vs mi manz ziromon ‘I eat cabbage’ vs ‘I eat giraumont’; pasyans i géri la gal, ‘patience heals the plague’.

 Robert Chaudenson

References Académie 1694. Dictionnaire de l’ Académie Française. Paris. Cellier, P. 1985. Description Syntaxique du Créole Réunionnais. Essai de Standardization. Thèse d’Etat. Aix-en-Provence: Université de Provence. Chaudenson, R. 1974. Le Lexique du Parler Créole de la Réunion. 2 vols. Paris: Champion. Chaudenson, R. 1981. Textes Créoles Anciens (la Réunion et Ile Maurice). Comparaison et Essai d’Analyse. Hamburg: Buske. Chaudenson, R. 1985. Français avancé, français zéro, créoles. In Actes du XVIIe Congrès International de Linguistique et de Philologie Romanes, vol 5: 165–180. Aix-en-Provence: Presses de l’Université de Provence. Furetière, A. 1690. Dictionnaire Universel. La Haye et Rotterdam: Leers. Galmiche, M. 1975. Sémantique Générative. Paris: Larousse. Marchello-Nizia, C. 1979. La Langue Française au XIVe et XVe Siècle. Paris: Bordas. Noailly, M. 1990. Le Substantif Épithète. Paris: PUF. Staudacher-Valliamée, G. 2004. Grammaire du Créole Réunionnais. Paris: SEDES. Valli, A. 1994. A propos de l’emploi productif de la détermination zéro en moyen français et en créole réunionnais. In Créolisation et Acquisition des Langues, D. Véronique (ed.), 89–101. Aix-en-Provence: Presses de l’Université de Provence. Véronique, D. (ed.). 1994. Créolisation et Acquisition des Langues. Aix-en-Provence: Presses de l’Université de Provence. Wartburg von, W. 1928. Französisches Etymologisches Wörterburch. Bonn: Klopp.

 

The bare NP in Lesser Antillean Karl Gadelii This paper examines the structure and behaviour of noun phrases in Lesser Antillean French Creoles. It is argued that bare NPs have an indefinite plural reading when appearing in isolation, but that [INDEF, SG], [DEF, PL] and [D, SG] readings are possible in case a noun phrase with the relevant properties has been previously introduced in the discourse. Lesser Antillean thus displays sensitivity to discourse anaphora in a way that French, the superstrate language in question, does not. When definiteness and number are spelled out in the Lesser Antillean noun phrase, this is done by means of independent particles, leading to a syntax that is superficially more complex than that of an agglutinative/fusional language like French. Post-nominal determiners are shown to be derived via XP-movement, not head movement. Certain contrasts between Guadeloupean and Martinican are shown to be due to different natures of XP-movement.

.

The bare NP in Lesser Antillean*

This paper analyzes the bare NP in Lesser Antillean French-lexified creole (henceforth LA) using examples from a Guadeloupean folk tale, Lapin ki vlé mandé Bondyé tiboin lèspri ‘Rabbit who wanted to ask some smartness from the Lord’, accessible on the web page 〈creoles.free.fr〉, maintained by Marie-Christine Hazaël-Massieux in Aix-enProvence. The examples are rendered in frenchified orthography containing some minor inconsistencies, which however does not affect the analysis of the data. The main idea behind the paper is that the bare NP in LA can express all meanings that non-bare NPs can, but that its presence is regulated by certain factors, among which the topicality of the referent is the most important one. Animacy, specificity and number are also relevant to the extent that the canonical topic is animate, specific and singular. By ‘specific’ I mean ‘particular’ and by ‘non-specific’ ‘any’, i.e. roughly the sense in which Bickerton (1981) uses these terms. As far as I can see, the distinction hinges on the predicate to the extent that “a shirt” in “I bought a shirt” is necessary since the act of buying implies that you become the possessor of a particular shirt, * I would like to thank Jacqueline Guéron and Marlyse Baptista for extensive constructive comments on this paper. Remaining inadequacies are my own responsibility.

 Karl Gadelii

whereas in an utterance like “I would like a shirt”, the speaker does not refer to a particular shirt, but to any shirt. See further Bickerton’s (1981:56) examples below. Grammatical function plays a certain role since “continued” topics (i.e. a topics that reappear in a sequence of sentences) tend to occupy subject positions, but topics can be introduced in both subject and object position. Before investigating the bare NP in Lesser Antillean, let us look at NPs in the language containing overt determiners.

.

Overt determiners in Lesser Antillean

In standard descriptions of LA, the NP in the Guadeloupean variety is said to display the overt determiners illustrated below: (1)

on

liv book ‘a book’ INDEF

(2)

livla book- DEF ‘the book’

(3)

liv- la- sa book DEF DEM ‘this/that book’

(4)

sé liv- la PL book DEF ‘the books’

(5)

sé liv- la- sa book DEF DEM ‘these/those books’ PL

It could be noted in this context that only indefinite determiners (assuming that plural sé is also indefinite to some extent) can appear prenominally in LA. Other determiners show up in post-nominal position, something which creates a very strong contrast to the French NP, where all determiners are prenominal. Note also that LA does not make the proximate ~ distant distinction, as illustrated in (3), though many creole languages do (see also further below). The Martinican variety of Lesser Antillean largely resembles that of Guadeloupe with the main difference that the demonstrative item in Martinican is ta instead of sa, and occurs before la: (6)

sé

liv- ta- la book DEM DEF ‘these books’ PL

Chapter 9. The bare NP in Lesser Antillean 

This difference in distribution between the demonstrative and definite determiner in the Martinican vs. Guadeloupean variety of Lesser Antillean is intriguing. I will return to it towards the end of this paper, where I will argue that it is due to parametric variation between these two LA varieties concerning the nature of the moved XP. For an alternative analysis, cf. Déprez (2005), who argues that the Guadeloupean strucure is unusual as compared to the Martinican one, which conforms to what we find in other French creoles. Other French-lexified creoles in the Lesser Antilles, such as those of Dominica and Saint-Lucia, resemble Guadeloupean in all important respects. The difference between proximate and distant demonstrative is not coded in the Lesser Antillean NP, but it appears in the independent pronominal form, illustrated in (7) below: (7)

sila

pli piti ki tala more small than TALA ‘this one is smaller than that one’ SILA

It should further be noted that *sé liv ‘books’ is ungrammatical in Lesser Antillean, i.e. the occurrence of sé requires the presence of la. This curious fact will also be returned to in the analysis section at the end of this paper. Similarly, the demonstrative cannot appear alone as in *liv-sa or *liv-ta, but needs to be derived from a definite NP. It should thus be noted that the determiner system of Lesser Antillean strongly sticks to the one-to-one principle between form and meaning. LA differs in this respect not only from French but also from various other French-lexified creoles. French exhibits portmanteau determiners such as les [PL, DEF], cette [SG, DEM, FEM], etc. In Frenchlexified creoles, we find examples such as the following: (8)

liv yo book DEF.PL ‘the books’

(Haitian)

(9)

dile la milk DEM.SG ‘this milk’

(Louisianais, Goodman 1964: 46)

(10)

(11)

liv yé (ibid.) book DEM.PL ‘these books’ sa

zonm (Seychellois, Holm 2000: 212) man ‘this man’

DEM.SG

(12)

famya (Guyanais, Goodman 1964: 45) woman DEM.PL ‘these women’

 Karl Gadelii

The necessity in Lesser Antillean of factoring out meanings on separate morphemes in the DP thus actually makes its surface nominal syntax more articulated than that of the languages illustrated above. Cf. the reasoning concerning Haitian by Zribi-Hertz & Glaude (this volume).

.

Determiners as a defining feature of creole languages

Determiners constitute one of Bickerton’s (1981) 12 defining features of creole languages. He claims (1981:56) that determiners display the following three types of behaviour in (virtually all) creole languages: • a definite article is used for presupposed specific NP as in Jan bai di buk ‘John bought the book (that you already know about)’ • an indefinite article is used for asserted-specific NP Jan bai wan buk ‘John bought a (particular) book’ • zero occurs for non-specific NP: Jan bai buk ‘John bought a book/books’, bok dia fi tru ‘books are really expensive!’ (examples from Guyanese) Bickerton’s claims above can be summarized as follows: specific NPs require a determiner (definite or indefinite), whereas non-specific ones do not. I will argue in this paper that specificity is not the crucial parameter at stake here, but rather topicality. On the other hand, these are closely interrelated since it would be very strange to have a non-specific referent as topic. Bickerton does not say anything about the grammatical function of the NP, but as I pointed out before, it is argued in the present paper that bare NPs tend to be subjects rather than objects. Green (1988: 444) observes that bare NPs are possible in creoles in both positions, but tend to be interpreted as definite (singular or plural) in subject position (i.e. preverbal position). In postverbal position, yet more interpretations are possible. We will return to this topic later on. I will now investigate the consequences of Bickerton’s proposal that specific nounphrases display overt determiners and apply it to singular and plural noun phrases in object position. Below are some constructed example sentences using French and English: Singular: (13) (14) (15) (16)

Indefinite non-specific: Indefinite specific: Definite specific: Definite non-specific:

Plural: (17) (18) (19)

Indefinite non-specific: j’aimerais des chemises Indefinite specific: j’ai vu des chemises Definite specific: j’ai acheté les chemises

j’aimerais une chemise j’ai vu une chemise j’ai acheté la chemise j’aimerais voir le gardien

‘I would like a shirt’ ‘I saw a shirt’ ‘I bought the shirt’ ‘I would like to see the caretaker’ ‘I would like (some) shirts’ ‘I saw (some) shirts’ ‘I bought the shirts’

Chapter 9. The bare NP in Lesser Antillean 

(20)

Definite non-specific:

j’aimerais voir les gardiens

‘I would like to see the caretakers’

Note that all French NPs have overt determiners in the above examples, also with plural indefinite NPs as in (17) and (18), where English counterparts can be bare. Now if Bickerton is right, (13), (16), (17) and (20) would not involve determiners in creole languages, whereas (14), (15), (18) and (19) would. That is, a French-lexified creole like LA would exhibit zero determiners in the following cases: (21) (22) (23) (24)

“j’aimerais Ø chemise” “j’aimerais voir Ø gardien” “j’aimerais Ø chemises” “j’aimerais voir Ø gardiens”

(I would like Ø shirt) (I would like to see Ø caretaker) (I would like Ø shirts) (I would like to see Ø caretakers)

(=13) (=16) (=17) (=20)

but overt ones in the following ones: (25) (26) (27) (28)

j’ai vu une chemise j’ai achete la chemise j’ai vu des chemises j’ai acheté les chemises

(I saw a shirt) (I bought the shirt) (I saw some shirts) (I bought the shirts)

(=14) (=15) (=18) (=19)

Now elicited equivalents to (13) through (20) in LA do not conform to Bickerton’s hypothesis. Cf. the sentences below (definite non-specific NPs have been left out due to elicitation difficulties, but there is reason to believe that they would display a null determiner). (29)

Indefinite non-specific: an té

ké enmé on chimiz FUT like INDEF shirt ‘I would like a shirt’ I

ANT

(30)

Indefinite specific:

an vwè on chimiz I see INDEF shirt ‘I saw a shirt’

(31)

Definite specific:

an achté chimiz la DEF I buy shirt ‘I bought the shirt’

Plural: (32)

Indefinite non-specific: an té ké vlé Ø chimiz shirt I ANT FUT want ‘I would like some shirts’

(33)

Indefinite specific:

an vwè Ø chimiz I see shirt ‘I saw some shirts’

(34)

Definite specific:

an achté sé chimiz la DEF I buy PL shirt ‘I bought the shirts’

 Karl Gadelii

As we can see, Bickerton’s hypothesis is not fully borne out by the above data. For example, we find an overt determiner in (29) although its referent is non-specific, and we find a null determiner in (33) in spite of the referent being specific. In subject position in isolated sentences, the situation would seem to be the same: Singular: (35) Indefinite non-specific:

on

chimiz disparèt shirt disappear ‘some shirt disappeared’ INDEF

(36)

Indefinite specific:

on

chimiz disparèt shirt disappear ‘a (particular) shirt disappeared’ INDEF

(37)

Plural: (38)

Definite specific:

chimiz- la disparèt DEF disappear shirt ‘the shirt disappeared’

Indefinite non-specific: Ø chimiz disparèt shirt disappear ‘some shirts disappeared’

(39)

Indefinite specific:

Ø chimiz disparèt shirt disappear ‘some (particular) shirts disappeared’

(40)

Definite specific:

sé chimiz- la disparèt PL shirt DEF disappear ‘the shirts disappeared’

It would thus seem that the parameter of specificity does not fully explain the presence or absence of determiners in Lesser Antillean. I will in the following further develop the idea that, although they are interrelated, topicality and not specificity, is what decides on the presence or absence of determiners in Lesser Antillean. To this end I will also use the concept of discourse anaphora, arguing that Lesser Antillean is sensitive to this phenomenon, whereas its lexifier, French, is not. As previously mentioned, Green (1988) suggests that the more topical a referent is, the greater the likelihood that it can be referred to by a bare NP in creole languages. Proper names are topical by definition and should thus never appear with a determiner in creoles, as opposed to what is the case in languages like Italian and Portuguese. Topicality is also related to uniqueness, which according to Baptista (2000) leads to the absence of determiners in creole languages. Croft (2003: 111) observes that topics have a tendency to be definite and specific, and maybe also singular. The notion of topicality is proposed with respect to Haitian by Joseph (1988), who suggests that when an NP has a known referent (within or outside the discourse), it can be bare.

Chapter 9. The bare NP in Lesser Antillean 

Bickerton (1981) also suggests that creole NPs are bare under the scope of negation and in generic uses. Interestingly, in these cases NPs are non-referential (in the case of generics at least non discourse-referential), i.e. fundamentally different from topical NPs. Bickerton’s general point is thus that creole languages do not mark NPs which have null or low referentiality. However, he overlooks the fact that highly referential NPs (= topics) are not overtly marked either, but for a totally different reason, namely that their identity is self-evident. The paradox that NPs with irrelevant as well as selfevident reference are not marked in creoles goes unnoticed if one only considers the specificity, not the topicality of referents.

.

Bare NPs in a Lesser Antillean corpus

I will now look at bare NPs in an LA corpus arguing that not only are indefinite plural NPs in Lesser Antillean bare (i.e. cases such as (32) and (33) above), but in fact, all exemplified cases of non-bare NPs in the elicited isolated sentences above will also have bare counterparts in narrative data. This hypothesis implies a remarkable difference between Lesser Antillean and French: combining the parameters of number, definiteness and specificity we get eight outcomes (including definite non-specific cases), where none can be bare in French, but all can in LA. The data considered so far result from elicitation tasks aimed at producing isolated sentences, and they will therefore not reveal discourse-related effects. It is thus important to complement such elicited data with stretches of running text, which is what I intend to do in the following. The corpus used here contains 100 noun phrases (not counting proper names), of which as much as 70 constitute bare NPs. The difference to French, the lexifier, is striking since, as we just noted, argumental NPs in French in principle have to display overt determiners. There are three examples of bare argumental NPs with indefinite singular reference in the data. These are illustrated in (41) through (43): (41)

ou ké rapòté moin dan a Zanba tooth from Zanba 2SG FUT bring back 1SG ‘you will bring back a tooth from Zanba to me’

(42)

ou ké rapòté moin kaka tig poopoo tiger 2SG FUT bring back 1SG ‘you will bring back a tiger poopoo to me’

(43)

moin ké voyé on moun chèché kaka- tig mème FUT send a person get poopoo tiger very 1SG ‘I will send someone to get a real tiger poopoo’

(The NPs in question are translated in French as ‘une dent de Zanba’ and ‘une (vraie) crotte de tigre’, respectively). It should be pointed out that these three instances of bare NPs with indefinite singular reference also refer to non-specific items (any tooth of

 Karl Gadelii

Zanba, any tiger poopoo). Note that the nouns in (41) through (43) are not mass terms, although kaka ‘poopoo’ could have such a reading, but in (42) and (43) the intended referent is countable (‘une crotte de tigre’). I am not treating mass nouns in this paper. All specific indefinite singular referents in the text are referred to by an NP including the indefinite article, such as (44)

i tonbé douvan on gran pyé- koko in front of INDEF big tree coconut 3SG fall ‘he fell down in front of a big coconut tree’

So when it comes to indefinite singular noun phrases, Lesser Antillean corroborates Bickerton’s (1981) proposal that these do not take a determiner when they refer nonspecifically (as in (41) through (43) above), but require one when their referent is specific, as in (44). This is in contrast to example (29), where the overt determiner on ‘a’ was provided also for indefinite non-specific NPs. I think Bickerton is basically correct in this respect (which also (41) through (44) show, which are taken from an authentic narrative), and I regard (29) as an atypical example, which maybe results from superstratal influence. Let us now turn to definite NPs. The data contains various examples of non-bare NPs expressing definiteness, as illustrated in (45) and (46) below: (45)

konpè Zanba tand muzik- la friend Zanba hear music DEF ‘Zanba heard the music’

(46)

gay sé makak- la look at PL monkey DEF ‘look at the monkeys’

However, we also find numerous examples where definite reference is achieved through bare NPs, in contrast to Bickerton’s (1981) proposal. In these we include NPs with an agglutinated definite determiner such as listoua ‘the story’, latè ‘the earth’, etc. (cf. French l’histoire, la terre). (This is not to say that preposed determiners à la française never occur in Lesser Antillean. However, this is a topic we cannot pursue here, and in the examples just mentioned it is clear that the determiner has completely merged with the noun and does not express definiteness by itself). The vast majority of bare NPs with definite reference in the text, 51 out of 70, are constituted by possessive phrases such as in (47): (47)

Lapin woutouné douvan kaz Rabbit return to house ‘Rabbit returned to Zanba’s house’

a Zanba of Zanba

It is possible that possessive NPs constitute a somewhat special case in that they “absorb definiteness” and are “inherently definite” (see the English translation of (47) above, and cf. the impossibility of saying *the my book in English or *le mon livre

Chapter 9. The bare NP in Lesser Antillean 

in French). In the generative tradition, this definiteness of possessive NPs is accounted for by assuming that D0 is filled in the phrases in question. Thus, there would be no need to code definiteness on possessive NPs in creoles since they are definite by nature. We could then tentatively formulate the following hypothesis: (48)

Possessive noun phrases in Lesser Antillean are bare.

However, this hypothesis is contradicted by two argumental possessive NPs in the story which display overt determiners. This is not a phenomenon typical of creole languages – it was just noted that French disallows *le mon livre, but the equivalent construction is grammatical in such close relatives as Italian and Portuguese. The non-bare NPs in our data are given in (49) and (50): (49)

sé timoun an moin- la ké joué avè sa PL child of me DEF FUT play with that ‘my children will be playing with that’

(50)

i mété- y anba mamèl a vach- la he put him under nipple of cow- DEF ‘he put him under the nipple of the cow’

On closer scrutiny, it turns out that what these two contexts have in common is that the children and the nipple are here mentioned for the first time in the story. It would thus seem that in a language like LA which is sensitive to discourse anaphora, overtly determined NPs appear when a referent is introduced for the first time, even when the NP at hand is possessive. Once introduced, the individual can subsequently be referred to by a bare NP. Notice that subject vs. object position is not decisive here, since the topic is introduced in subject position in (49) but in object position in (50). (48) thus has to be reformulated as (51): (51)

Possessive noun phrases in Lesser Antillean have overt determiners only when they introduce a new referent.

Let us now turn to non-possessive bare NPs referring to definite entities. Examples of those can be found in (52) and (53) (52)

i fini pa touvé koulèv snake 3SG finish by find ‘he finally found the snake’

(53)

pwal ka voltijé toupatou all-over fur NONP fly ‘the fur was flying all over’

In contrast to (49) and (50), (52) and (53) constitute counterexamples to Bickerton’s hypothesis, since the referents in question are clearly specific, yet the NPs are bare.

 Karl Gadelii

Here too, we adopt the initial hypothesis that a bare NP is always possible when definite reference is intended. This leaves us with the task of explaining when also nonbare NPs can refer in this way in our corpus. A particularly clear example is provided in (54), where a topic is introduced by means of a non-bare NP in object position, whereafter the same NP appears in bare form in subject position: (54)

I di: “Gay sé makak- la: he say look PL monkey DEF Alor, makak pa té so monkey not ANT

oui, zò lèd…! Ka zòt ka santi!” yes they ugly how they NONP smell

kontan! happy

‘He said: “Look at those monkeys, aren’t they ugly!…How they smell!” So, the monkeys were not very happy!

There are 27 tokens of non-bare NPs with definite reference in the corpus (not counting possessive ones). 22 of those are used to introduce referents, and are thus unproblematic for the present thesis. However, five NPs with determiners require explanation. Those five instances comprise two occurrences of muzik-la ‘the music’, two of dan-la ‘the tooth’ and one occurrence of sé kochon-mawon-la ‘the wild pigs’. These noun phrases occur in contexts where they are not expected, since the referent has already been introduced with an identical NP. We would thus expect subsequent occurrences to be determinerless. I quote the data in full below, and indicate the first appearance of muzik-la and dan-la in bold, and the following ones underlined: (55)

Konpè Zanba ki té atab, ka manjé avè madam-a-y, tand muzik-la; i di: “Aaa… ki bèl muzik è sa.” I vansé tèt a-y é i vouè sé Konpè Lapin ki té ka joué muzik-la. I di: “Aaa konpè, oui, bèl plézi ou ka fè moin!” é i sòti, é i mété-y ka dansé. Kan i mété-y ka dansé, ti dlo a gonbo-la ki té glisant, - pas ou sav gonbo ka glisé an gòj a maléré - ébin, anlè latè a Zanba, sa fè parèy: Zanba inki glisé, épi i tonbé, é i kasé on dan a-y. Aaa Konpè Lapin di: “Aaa Zanba, ki malè! Koman sa té rivé ou, non!!!” Alòr, Zanba di: “Ebin, monchè, sé arivé: an té kontan muzik-la, dan-la kasé; manfou a sa!” Alòr Konpè Lapin di: “Mé ka ou ké fè épi dan-la ? Pa ni ayin pou fè avè sa. Zanba who was sitting at the table eating with his wife, heard the music; he said: “Aaah…such sweet music!” He bent his head and saw that it was Rabbit who was playing the music. He said: “Aaah, my friend, yes! Such joy you bring me !” and he got out, and he started to dance. When he started to dance, the gumbo sauce which was slippery – because you know, gumbos slip down the throat of the poor – well it did the same thing outside Zanba’s door: he slipped, fell, and broke a tooth. “Aaah, said Rabbit, Aaah Zanba, how unlucky you are! How did that happen to you!!!” Then Zanba said: “Well, my friend, it just happened: I was enjoying the music, my tooth broke,

Chapter 9. The bare NP in Lesser Antillean 

I don’t care!” Then Rabbit said: “But what are you going to do with that tooth? There is nothing you can do with it.”

The section containing sé kochon-mawon-la ‘the wild pigs’ is cited in (56): (56)

I komansé fouté sé kochon-mawon-la chak kou-d-sab: wach, wach, wach, wach… Chak kou-d-sab-la, pwal ka voltijé toupatou; alò Konpè Lapin lésé Zanba démèrdé-y avè sé kochon-mawon-la. (He started to hit the wild pigs with a sabre: wach, wach, wach, wach…With every blow of the sabre, the fur was flying all around: so Rabbit let Zanba handle the wild pigs.)

I do not have a really good explanation as to the presence of determiners in the underlined NPs above. They do have some things in common: they are all objects or objects of preposition, except dan-la kasé ‘the tooth broke’, where dan-la is however not a canonical subject; its thematic role is Theme. Muzik-la ‘the music’ and dan-la ‘the tooth’ are also inanimate, and they do not appear in the immediate vicinity of their antecedents. But most importantly: presumably the music, the tooth and the wild pigs are not “topical enough” to allow their determiners to be omitted, i.e. they do not become subjects of sentences moving the story along (with one possible exception, dan-la kasé ‘the tooth broke’). So to summarize the paper so far, we have found that Lesser Antillean displays overt determiners in sentences elicited in isolation with the exception of indefinite plural NPs (corresponding to NPs which in French require the partitive determiner des as in des livres ‘some books’). However, in context, bare NPs can be used to refer also to indefinite singular and plural individuals as well as to definite plural ones, as long as the referent is topical and has been previously introduced by means of a non-bare NP. I will now briefly look at some superstratal and substratal facts relevant to the emergence of NP structure in Lesser Antillean before I devote the rest of the paper to an analysis of bare and non-bare NPs in LA.

.

Superstratal facts: bare NPs in French

We have already noticed on several occasions that French NPs require the presence of a determiner, whether definite or indefinite. Cf. the following example: J’ai vu une fille et un garçon. La fille dormait tandis que le garçon faisait la vaisselle ‘I saw a girl and a boy. The girl was sleeping while the boy was washing the dishes.’ There are certain exceptions to this rule: determiners of argumental NPs can be omitted in very special cases such as headline language or commercial slogans, a well-known fact in many languages. Cf. a headline such as Nouvel attentat à Bagdad (Libération Jan. 29, 2004) ‘new attack in Bagdad’. In ordinary contexts, nouvel attentat without

 Karl Gadelii

the determiner is not grammatical in French (or in English for that matter): *nouvel attentat s’est produit ~ *new attack took place; *nous avons vécu nouvel attentat ~ *we experienced new attack. Also, proper names in French do not have determiners (as they do in e.g. Italian and Portuguese). However, certain toponyms display a determiner in French (la France ‘France’, la Grande-Terre ‘Grande-Terre’ (cf. Lesser Antillean: Fwans, Grantè). When it comes to mass nouns, they require a determiner in French but not in English: (cf. j’ai trouvé de l’or vs. ‘I found gold ’). As to nonargumental NPs, predicative ones can appear determinerless in French (il est médecin ‘he is a doctor’), as can vocative ones. Kihm (2000) further points to idioms such as donner raison ~ tort ‘corroborate ~ refute’, literally “give right/wrong”, and faire bonne figure ‘look nice’, literally “make nice face”, where the object is tightly integrated with the verb. Probably, the phenomenon is more general than this in french – cf. examples such as j’ai cours à six heures ‘I have class at six o’clock’, cela pose problème literally “this poses problem”, etc. See further Bouchard (2003). In spite of these counterexamples, it does not seem reasonable to propose that the presence of determinerless NPs in French could in any way have influenced the emergence of this NP-type in Lesser Antillean. On the contrary, French argumental NPs require overt determiners by default.

.

Substratal facts: bare NPs in African languages

Without going too deeply into the debate on creole genesis, we would still like to say a few words about determiners in certain African languages. In the case of Lesser Antillean, Bantu languages (in particular from the Loango area) probably constituted a more important substratum than in other French-lexified creoles (see Arends 1995). Bantu languages do not exhibit determiners, neither definite nor indefinite. When it comes to Niger-Congo languages in West Africa, Lefebvre has on several occasions (e.g. 1998) proposed a profound parallel between the definite determiner la in Haitian and o in Fongbe. Both these items appear post-nominally, in fact post-NP, and they are absent in generic contexts (for an opposing point of view, see however e.g. Zribi-Hertz (2000)). Holm (2000) brings up the case of Yoruba, where facts are similar to those in Fongbe. Creissels (2000: 243) generalizes the picture by saying that “languages with or without definite determiners are present in almost all [African] language families and all over the African continent” but observes that indefinite determiners are much less common than definite ones. When it comes to definite determiners, Creissels adds that they are not infrequently Ø under the scope of negation, as in “they have not seen the boat”, where we thus see a parallel with creole languages. On the other hand, zero determiners in non-referential contexts may also be a universally common phenomenon, and thus not necessarily an example of substratal influence in creoles. Creissels (2000: 245) also notes that the definite determiner has

Chapter 9. The bare NP in Lesser Antillean 

eroded in many African languages to the extent that it only remains as a tonal effect. Cf. básá ‘a lizard’ vs. básà ‘the lizard’ (Kita Maninka). These scattered observations do not allow us to make any great claims about possible parallels between determiners in Niger-Congo languages and Lesser Antillean. We could however surmise that the general absence of determiners in Loango Bantu languages, the most important substratum of LA, may have converged with universal semantic/discourse-related phenomena to promote the presence of bare NPs in Lesser Antillean.

.

Analysis of the DP in Lesser Antillean creole

Baptista (2000) suggests that bare NPs with definite interpretation which appear in preverbal position have undergone movement from N0 to D0 (cf. Longobardi 1994). Once the noun is in D0, the NP does not have to appear in a lexically governed position (in pre-minimalist terms), a suggestion which by the way goes well with the subject–object asymmetry tendencies observed for Lesser Antillean above. Chierchia (1998) surmises that languages with null D0 have a iota operator in this position, doing the same job as a definite determiner. I will here only add the observation that tests for overt head movement within the DP are less developed than those concerning the extended VP, where the position of the verb can be determined with great precision with the help of negation and adverbial insertion tests, as was first shown in the seminal work by Pollock (1989). The best test for head movement within the DP presumably remains that of Cinque (1994), where the author proposes that preadjectival nouns as in French la femme intelligente ‘the intelligent woman’, literally “the woman intelligent” result from movement of the noun around the adjective. It will be seen below that this kind of movement is not allowed in Lesser Antillean. Rather, it is proposed in Gadelii (1997) that strong features in creoles are checked by “lexicalisation” when the feature involved is a head feature, and by movement when dealing with a specifier feature. It is proposed in Gadelii (1997) that Lesser Antillean never exhibits head movement, but on the other hand has more extensive phrasal movement than its lexifier, French. Therefore, Baptista’s proposal is not applicable to the bare NP in LA, whereas Chierchia’s is. Turning to DPs with overt determiners in Lesser Antillean, the structure consisting of a noun + postposed determiner (such as liv-la book-DEF ‘the book’) however looks like a typical case of head movement, where the noun has moved around the determiner, head-adjoining to the left of it. However, this contradiction is only apparent, since the determiner is in final position also when the NP is complex. Consider the below example: (57)

siklis Gwadloup-la cyclist Guadeloupe-DEF ‘the cyclist from Guadeloupe’

 Karl Gadelii

If we move the head noun only to the front as suggested above, the result will be ungrammatical: (58)

*siklisi la ti Gwadloup

The examples in (57) and (58) show that the whole phrase [siklis Gwadloup] must raise above the determiner (given that DP is an extended projection of NP and phrases are left-headed). Since we are dealing with phrasal movement, the moved NP must have an accessible target, which is of course Spec-DP: DP

(59)

SPEC [siklis Gwadloup]i

D'

D0

NP

la

N0 [

t

XP ]i

The movement which takes place in lèt-la milk-DEF ‘the milk’ is thus analyzed in the same way: DP

(60) SPEC [lèt]i

D'

D0

NP

la [t]i

Turning to the determiners on ‘a’ and sé ‘PL’ , these can be supposed to sit in the head of NumP, which is positioned below DP, in conformity with the generally accepted

Chapter 9. The bare NP in Lesser Antillean 

view of the articulated DP structure (cf. e.g. Platzack (1998)). By further analogy, we suggest that the bare noun is also a maximal projection. We derive indefinite and definite singular and plural DPs in Lesser Antillean as illustrated in (61): DP

(61) SPEC

D'

D0

NumP

SPEC

Num’ Num0

(a) (b) (c) (d)

[on fanm]i [fanm]i [fanm]i [sé fanm]i

Ø la Ø la

[on Ø Ø [sé

fanm] [fanm] [fanm] fanm]

NP

(‘a woman’) (‘the woman’) (‘(some) women’ (des femmes)) (‘the women’)

The idea is thus that on, sé and la are merged in Num0 and D0, respectively, to check strong head features. The position of NumP with regard to la shows that NumP has moved to Spec-DP. By default we suggest that this movement also takes place when D0 is null, i.e. in (61a) and (61c). There is no syntactic diagnostic available proving that movement has taken place in these cases, but since it visibly takes place in (61b) and (61d), there is no reason to believe it doesn’t in the other two cases, although it is string vacuous. This is yet another illustration of the fact that Lesser Antillean displays more XP-movement than French does, contrary to the commonly held view that creole languages exhibit less movement than their lexifiers. I would also like to suggest that when D0 is null we get the unmarked reading, which is the indefinite one. However, there are a few remaining problems with the analysis presented in (61), such as the following: • Why cannot la and on cooccur, yielding *fanm la on, that is, a definite singular noun phrase? • Why does the absence of an element in Num0 yield a singular reading in (61b) but a plural one in (61c)? • Why can’t one say *sé fanm (plural indefinite), when on fanm (singular indefinite) is fine?

 Karl Gadelii

I would like to propose a solution to the above problems along the following lines. Implied categories do not have to be overtly marked, and thus cannot be. Definite implies singular and singular therefore does not have to be marked, and thus cannot be (*fanm la on). Instead, fanm-la without number marking is interpreted as singular ‘the woman’. If a plural reading is intended it has to be marked overtly: sé fanm-la ‘the women’. Indefinite implies plural and plural therefore does not have to be marked, and hence cannot be. There is no specific morpheme functioning as indefinite determiner in LA (recall that on and sé are Num0 heads), which is why null D expresses the indefinite (as opposed to filled D0, which expresses the definite). Plural cannot be marked in this case. fanm means ‘some women’, *se fanm cannot be generated. If a singular reading is intended, it has to be overtly marked : on fanm. Let us now also include an analysis of demonstrative DPs in Lesser Antillean, as announced at the beginning of this paper. These DPs exhibit an interesting variation involving Guadeloupean vs. Martinican. The relevant examples are here repeated as (62) and (63): (62)

(63)

sé fanm la PL woman DEF ‘these women’ sé

sa

(Guadeloupean)

DEM

fanm ta la woman DEM DEF ‘these women’

(Martinican)

PL

To account for these examples, we need an extended projection DemP, which I will suggest sits on top of DP. Its head is filled by either sa or ta. The previous DP-tree will thus be further articulated and now includes DemP–DP–NumP–NP. One problem with this analysis is that it does not take into account the fact that Num selects D in LA (that is, *sé liv is not grammatical whereas liv la is), nor the fact that Dem selects D (cf. the ungrammaticality of *liv sa and vs. the grammaticality of liv la) Cf. further Zribi-Hertz (2005) and Déprez (2005). Given these facts, NumP and DemP should dominate DP in the extended NP in LA. I cannot develop this alternative here, but would like to add that it does not affect the main reasoning in this paper. There are good reasons for suggesting that DemP should be the topmost projection of N. For one thing, it is commonly assumed that the higher up in the tree a certain projection sits, the more is it related to the world. Cf. Platzack’s (1998) observation that the finiteness feature in AgrsP or CP makes it possible to assign a truth value to the sentence, whereas this cannot be done at lower levels such as TP or VP. Cf. also the commonly made assumption that discourse-related projections such as FocusP or TopicP, which have to do not only with the world but also with language use, are even higher up than CP (see Rizzi 1997). In the following, I will suggest the presence of a Topic phrase in Lesser Antillean, situated at the very top of the extended NP, under which DemP is positioned (see (65) below). The demonstrativity feature in DemP can be regarded

Chapter 9. The bare NP in Lesser Antillean 

as being closely related to the extra-linguistic context, thus constituting a parallel to AgrsP or CP. Assuming the extended NP-tree sketched above, we explain the variation between Guadeloupean and Martinican by suggesting a parametric variation such that the DP in the former language “snow-balls” (using a term of Enoch Aboh’s 2004) its way to Spec-DemP, whereas NumP in Martinican moves in successive cycles to SpecDemP. Graphically this would look like below: (

6

4

)

DemP SPEC

Dem' Dem0

DP SPEC

D’ D0

NumP SPEC

Num’ Num0

(a) [[sé fanm]i laj]k sa (b) [sé fanm]i

ta

[[ti] ti

NP

tj ]k

[

ti

]

(Guadeloupean)

la

[

ti

]

(Martinican)

The Guadeloupean structure is thus derived in successive cycles such that NumP first moves to Spec-DP. There it picks up la, and the DP sé fanm la moves as a unit to Spec-DemP. In Martinican, NumP moves to Spec-DP, leaves la behind and continues to Spec-DemP. These examples provide further evidence of the omnipresence of XPmovement in creoles. Both Guadeloupean and Martinican display phrasal movement in the DemP, the only difference being that Guadeloupean does it in a snowballing way, Martinican by successive cyclic XP-raising. Finally, to account for the fact that the bare NP can also express all the meanings that non-bare NPs can in (61a, b, d), we would need to integrate a Topic phrase in the tree, whose head would host a topic operator. In conformity with the reasoning above that discourse-related projections should be high up in the tree, the Topic phrase should constitute the topmost node of the tree, i.e. it should sit above DemP. The bare noun, which is in fact a phrase, would move successively through the specifier positions of NumP, DP and DemP, driven by the need to check its topicality feature, and would finally end up in Spec-TopP. In the case of non-topical noun phrases, we have seen throughout this paper that the head positions in the extended NP must be filled (by on

 Karl Gadelii

‘INDEF’, la ‘DEF’ sé ‘PL’ or sa ‘DEM’). When the noun phrase is topical, the head positions in question are not filled, but get their interpretation by virtue of being discourse-linked to previous NPs having the properties in question. On its way to the top, the bare NP checks these invisible discourse-induced Number, Definiteness and Demonstrativity features when in a Spec-Head relation with the head of the extended projection in question. This ensures that the bare NP has received its correct interpretation when it finally ends up in Spec-TopP. The structure in question would look as in (65): (65)

.

[TopP XPi Op[DemP ti Ø [DP ti Ø[NumP ti Ø [NP ti ]]]]]

Conclusions

The tree in (61) illustrates the presence of overt determiners in Lesser Antillean, but conversely, the bare NP in Lesser Antillean creole can express all meanings conveyed in (61) by non-bare NPs, i.e. not only the indefinite plural (“some women”) but also [INDEF, SG] “a woman”, [DEF, SG] “the woman”, and [DEF, PL] ‘the women’. The mini-corpus investigated in the main part of this paper indicates that the bare NP expresses [INDEF, SG/PL] or [DEF, PL] in case the referent is topical. We found parallels between topicality, animacy, specificity, uniqueness and subjecthood, claiming that the canonical topic is an animate subject in the singular with specific reference, which appears in the form of a bare NP. This remark is in total opposition to Bickerton (1981), where it is claimed that increased specificity results in overt marking of definiteness. It was further suggested that the Lesser Antillean facts could be explained by its being sensitive to discourse anaphora: once a referent has been introduced, it can subsequently appear in the form of a bare NP. In contrast, French requires an overt definite determiner on the NP also when anaphoric reference is intended. Possessive NPs in Lesser Antillean turned out to be almost always bare, maybe for the reason that definite marking is “absorbed” in possessive constructions. The few exceptions that were noted could be explained by the NPs in question being used to introduce new referents. Finally we have shown that the structure of the Lesser Antillean DP can be derived assuming that on, sé and la “lexicalize” functional head projections, whereas the apparent raising of the nominal head is actually an instance of NP movement, thus phrasal movement.

References Aboh, E. 2004. The morphosyntax of complement-head sequences. Clause structure and word order patterns in Kwa. New York: Oxford University Press. Arends, J. 1995. The socio-historical background of creoles. In Pidgins and Creoles. An introduction, J. Arends, P. Muysken & N. Smith (eds), 15–24. Amsterdam: John Benjamins.

Chapter 9. The bare NP in Lesser Antillean  Baptista, M. 2000. Typologie du DP et déterminants vides en créole cap-verdien. Paper presented at la Journée d’Etude sur les Créoles, April 21 2000, University Paris-3, France. Bickerton, D. 1981. Roots of Language. Ann Arbor MI: Karoma. Bouchard, D. 2003. Le SN sans déterminant. In Essais sur la grammaire comparée du français et de l’anglais, P. Miller & A. Zribi-Hertz (eds), 55–95. Paris/Saint-Denis: Presses Universitaires de Vincennes. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Cinque, G. 1994. On the evidence for partial N movement in the Romance DP. In Paths Towards Universal Grammar. Studies in honor of Richard S. Kayne, G. Cinque, J. Koster, J.-Y. Pollock, L. Rizzi & R. Zanuttini (eds), 85–110. Washington DC: Georgetown University Press. Creissels, D. 2000. Typology. In African Languages: An introduction, B. Heine and D. Nurse (eds), 231–258. Cambridge: CUP. Croft, W. 2003. Typology and Universals. 2nd edn. Cambridge: CUP. Déprez, V. 2005. Nominal constituents in French lexifier Creloes: probing the structuring role of grammaticalization. Paper presented at the conference ‘Creole languages between substratum and superstratum’, Max Planck Institute, Leipzig, June 3–5, 2005. Gadelii, K. E. 1997. Lesser Antillean French Creole and Universal Grammar. PhD Dissertation, Göteborg University. (Göteborg: Gothenburg Monographs in Linguistics 15). Goodman, M. F. 1964. A Comparative Study of French Creole Dialects. The Hague: Mouton. Green, J. N. 1988. Romance Creoles. In The Romance Languages, M. Harris & N. Vincent (eds), 420–474. London: Croom Helm. Holm, J. 2000. An Introduction to Pidgins and Creoles. Cambridge: CUP. Joseph, F. 1988. La Détermination Nominale en Créole Haïtien. PhD Dissertation, University Paris-7. Kihm, A. 2000. Les possibilités référentielles des groupes nominaux en kriyol (créole à base lexicale portugaise, Guinée-Bissau). Paper presented at la Journée d’Etude sur les Créoles, April 21 2000, University Paris-3, France. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The case of Haitian Creole. Cambridge: CUP. Longobardi, G. 1994. Reference and proper names: A theory of N0-movement in syntax and logical form. Linguistic Inquiry 25(4): 609–65. Platzack, C. 1998. Svenskans Inre Grammatik – Det Minimalistiska Programmet (The mental grammar of Swedish – the minimalist program). Lund: Studentlitteratur. Pollock, J.-Y. 1989. Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry 20(3): 325–365. Rizzi, L. 1997. The fine structure of the left periphery. In Elements of Grammar, L. Haegeman (ed.), 281–337. Dordrecht: Kluwer. Zribi-Hertz, A. 2000. La syntaxe du déterminant postnominal LA dans les créoles à base française face aux théories de la créolisation. Paper presented at la Journée d’Etude sur les Créoles, April 21 2000, University Paris-3, France. Zribi-Hertz, A. 2005. From morphosyntax to reference construal: noun phrase structure and interpretaton in Haitian. Course given at the École d’automne de linguistique (EaLing), Paris, September 2005.

 Karl Gadelii

Appendix Lapin ki vlé mandé Bondyé tiboin lèspri ‘Rabbit who wanted to ask some smartness from the Lord’

(1) (2) (3) (4) (5) (6)

creole phrase containing bare NP

equivalent in English

rakonté nou listoua a Konpè Lapin ou mété moin asu latè ban-moin tiboin lèspri ou ka kouyoné tout moun ou ké rapòté moin dan a Zanba ou ké rapòté moin pwal a kochon mawon

tell us the story of Rabbit you put me on Earth give me some smartness you are fooling everybody you will bring back a tooth from Zanba you will bring back some fur from wild pigs you will bring back some milk from wild cows you will bring back a tiger poopoo you will make the snake and its seven children enter into a Spanish coconut he said: Look at those monkeys: they are ugly, right…! How they smell!” Well, the monkeys were not happy! having come to Zanba’s house… he spread it out before Zanba’s door he took his mandolin he ate with his wife he stuck his head out you are giving me some good pleasure gumbos are sliding down the throats of the poor the gumbo sauce which was slippery did the same thing she did the same thing in front of Zanba’s door Rabbit took Zanba’s small tooth he put it in the Spanish coconut those are big pigs… Zanba’s tooth broke Zanba wanted to put on a show (“make his big man”) his tooth broke Rabbit returned to Zanba’s house

(7) ou ké rapòté moin lèt a vach sovaj (8) ou ké rapòté moin kaka tig (9) ou ké fè rantré koulèv é sèt pitit a-y adan on ti koko èspanyol (10) I di: “Gay sé makak-la: oui, zò lèd…! Ka zòt ka santi!” Alor, makak pa té kontan! (11) (12) (13) (14) (15) (16) (17)

Rivé douvan kaz a Konpè Zanba… i simé-y douvan pòt a Konpè Zanba i pran mandoline a-y i té ka manjé avè madam a-y i vansé tèt a-y bèl plézi ou ka fè moin gonbo ka glisé an gòj a maléré

(18) ti dlo a gonbo-la ki té glisant fè parèy (19) sa fè parèy anlè latè a Zanba (20) (21) (22) (23) (24)

Konpè Lapin pran ti-dan a Konpè Zanba i mété-y an koko dèspagne sé gwo kochon… dan a Zanba kasé Zanba té vlé fè gran misyé a-y

(25) dan a-y kasé (26) Konpè Lapin woutouné devan kaz a Konpè Zanba (27) i kasé dizuisan zasyèt san konté léfélé (28) i pran sab a-y (29) i alé avè Lapin a chan a kochon-mawon

he broke 1800 plates not counting those who got cracks he took his sword he went with Rabbit to the field of the wild pigs

Chapter 9. The bare NP in Lesser Antillean 

Appendix: Continued creole phrase containing bare NP (30) i fouté sé kochon-mawon-la chak kou-d-sab (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43) (44) (45) (46) (47) (48) (49)

pwal ka voltijé toupatou i mété-y an koko-dèspagne a-y mamèl a-ou gwosè a on zéléfan Pitit a-ou pa ka tété? pitit an-moin tèlman fèb pa mème ni fòs vini tété fo moin maré katpat a-ou byen tété a-ou tèlman gwo i pran ti koko-dèspagne a-y i pati pou alé chèché kaka-tig Kaka-tig! i pran kaka a-y mété-y an koko dèspagne a-y i ka rèsté moin koulèv é sèt pitit a-y i fini pa touvé koulèv Rivé douvan twou a koulèv… Koulèv lonji tèt a-y Ka ki douvan pòt an-moin ou pé ké jin rantré avèk sèt pitit a-ou

(50) (51) (52) (53) (54) (55) (56) (57)

timoun a Konpè Koulèv té ka manjé on ti van ka pasé an bouch a-y Mézanfan… i glisé tèt a-y i fè plas pou tout lésèt pitit Lapin alé fèmé ti kalbas a-y Bondyé, mi komisyon a-ou Koulèvsoti pwèl a kochon-mawon, lèt a vach sovaj, kaka-tig

(58) (59) (60) (61)

ou ké mandé moin lèspri sa pa kaka-tig sé kaka-tig mème moin kè voyé on moun chèché kaka-tig méme (62) apa kaka-tig… (63) i bay on kalòt déyè tèt (64) Lapin tonbé su latè

equivalent in English he started to beat the wild pigs with his sword (give the wild pigs many beatings with his sword) the fur whirled around he put it in his Spanish coconut your nipples are big as an elephant doesn’t your calf suck? my calf is so weak he can’t even come here and suck I have to tie your four pats well your nipples are so big he took his small Spanish coconut he left to look for the tiger poopoo A tiger poopoo! he took his poopoo he put it in his Spanish coconut I still have the snake and its seven children he finally found the snake Having arrived before the snake’s hole… Snake put out his head Who is at my door you will never be able to enter with your seven children Snake’s children were eating a light wind passed down his throat My children… he slid his head in she made room for all the seven children Rabbit went to shut his small calebasse Lord, here are your commissions Snake took out the furs from the wild pigs, the milk from the wild cow, the tiger poopoo you ask some smartness from me that is not a tiger poopoo yes, it is a tiger poopoo I will send someone to get a real tiger poopoo if that is not a tiger poopoo… he gave him a big blow on the head Rabbit fell back to Earth

 

Bare NPs and deficient DPs in Haitian and French From morphosyntax to referent construal1 Anne Zribi-Hertz and Herby Glaude Based on a comparative description of Haitian and French noun phrases, this study focuses on the notion of bareness which characterises so-called bare NPs. Lack of determiner is identified as one type of syntactic deficiency, which must be distinguished from Number deficiency. For any functional feature F, phonological deficiency is argued to be ambiguous between feature unmarkedness, feature deficiency, and spell-out deletion. The study of bare NPs must therefore crucially separate the phonological and syntactic modules of grammar. Surveying the main determiners of Haitian in simplex (nonrelativised) noun phrases, we show that these morphemes all have a deictic value, either because of their inherent features, or because of their selectional properties, and that ‘bare NPs’ correlatively occur as default options which take up whatever interpretations are not available for overt determiners. We next turn to complex relativised noun phrases. Adapting Kayne’s (1994) theory of relativisation, we argue from Haitian evidence that they include two layers of functional structure which are independently specified for number and/or definiteness, each distribution of features triggering its own set of semantic effects. We suggest that the distribution of number and locative determiners in relativised DPs, and the correlated semantic effects, might be essentially similar in French and Haitian, in spite of the crucially different properties of ‘D’ heads in these two languages.

. We have a huge debt of gratitude towards Marlyse Baptista, Jacqueline Guéron, and Makoto Kaneko, for their helpful comments on earlier drafts of this text. We also thank Karl Gadelii for sharing with us some typological data, and the audience of the Paris ‘DP group’ for their precious stimulus and feedback: N. Boneh, P. Cabredo, A. Daladier, C. Dobrovie, A. Kihm, M.-L. Knittel, S.-N. Kwon, B. Laca, O. Matushansky, J. Roodenburg. The acceptability judgements on Haitian are those of H. Glaude, who was born in 1977 in L’Estère, in the Northern department of Artibonite (where definite LA remains undeleted to the left of plural YO), but lived for ten years in Port-au-Prince (where LA undergoes deletion when followed by YO).

 Anne Zribi-Hertz and Herby Glaude

.

Bare NPs and Number-deficient DPs

Bare NPs are usually understood (cf. Chierchia 1998, Longobardi 1999, 2001) as noun phrases which lack a determiner, their semantics being of special interest in languages which otherwise have determiners. For example, English licenses bare NPs where French does not,2 as exemplified in (1)–(2): (1)

a. b.

(2)

a.

The whales are cute. Whales are cute.

Les baleines sont mignonnes.  whale- are cute- b. *Baleines sont mignonnes.

[specific reading] [generic reading] [specific or generic]

The lack of determiner, however, is not the only type of functional deficiency within noun phrases which deserves to be studied. Another relevant case is number deficiency, which we illustrate in (3b) and (4b,c) by French examples: (3)

a.

b.

(4)

a.

b.

c.

d.

Achetez mes jolies tomates! buy-2 my- pretty- tomato- ‘Buy my pretty tomatoes!’ Achetez ma jolie tomate! buy-2 my- pretty- tomato lit. ‘Buy my pretty tomato!’ J’ ai mangé une tomate au petit déjeuner. 1 have eaten a/one- tomato at breakfast ‘I ate {a/one} tomato for breakfast.’ J’ ai mangé de la tomate au petit déjeuner. 1 have eaten de - tomato at breakfast ‘I ate tomato for breakfast.’ Je me suis cassé la jambe en glissant sur de la tomate. 1 broke my leg as I slid on de - tomato ‘I broke my leg as I slid on tomato (pulp).’ J’ ai mangé des tomates au petit déjeuner. 1 have eaten de+- tomato- at breakfast ‘I ate tomatoes for breakfast.’

The two examples in (3) may be naturally uttered by a Paris vegetable grocer. In sentence (3a), plural marking on the object noun phrase unambiguously triggers a plural interpretation of the referent, i.e. forces us to construe an atomised set of tomatoes minimally including two items. Sentence (3b), where the object is not pluralised, is on the other hand ambiguous between a singular construal (only one tomato-item on sale), and what we shall call a Sample reading, pointing to an unspecified quantity . Determinerless nominal arguments are in fact licensed in French, but only under special conditions, for example under coordination, as shown by Roodenburg (2001, 2003).

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

of discontinuous items belonging to the  species.3 This latter interpretation is similar to that of noun phrases which have been or may be characterised as involving number deficiency in various other languages.4 In sentences (4a) and (4d), the determiners une and des in the object noun phrase, which are respectively specified for the singular (une = ‘one’) and for the plural (des = de+les), trigger a so-called +Count construal of the object referent: one tomato-item in (4a), at least two tomato-items in (4d). In (4b), on the other hand, the so-called ‘partitive’ determiner de+la triggers, in the same context, a –Count interpretation involving a continuous construal of the referent (tomato substance or pulp).5 A subtle semantic difference between the Sample reading in (3b) and the Pulp reading in (4b,c) is that the Sample reading does not involve any grinding of the referent in the sense of Pelletier (1979), i.e. its being reduced to a pulp/a powder/a stew while the Pulp reading involves Grinding. A crucial observation regarding these examples is that the Sample reading available in (3b), and the Pulp reading available in (4b,c) cease to be available whenever a number specification (−plural, or +plural) is present in the noun phrase: thus, sentence (3a) is infelicitous if the salesman only has one tomato-item to sell, while (3b), in its Sample reading, is felicitous in this pragmatic context; and the Pulp reading is unavailable in (4a) and (4d), contrasting with (4b,c). To account for these data, we assume that an argument noun . The Sample reading is introduced here in contradistinction with the Kind label, which is used in the semantic literature in reference to the interpretations of nominals such as those italicised in (i): (i)

a. b.

Iguanodons are extinct. The iguanodon is extinct.

These noun phrases are however, arguably, syntactically specified for number — [+] in (ia), [−] in (ib) — and are correlatively construed semantically as plural in (ia) (i.e. as denoting a(n open) set of more than one atomised items) and singular in (ib) (i.e. as denoting a single species-item). What we call the Sample reading crucially correlates, like the Pulp reading, with number deficiency. Like the Pulp reading, the Sample reading ceases to be available if number is syntactically specified in the noun phrase. . Cf. Zribi-Hertz & Mbolatianavalona (1999) on Malagasy, Schmitt & Munn (2000) and Müller (2001) on Brazilian Portuguese, Song (1975), Jun (1999), Kwon & Zribi-Hertz (2004) on Korean, Chung (2000) on Indonesian. . The French partitive determiner is also open to what we have called the Sample reading, in appropriate contexts, as witnessed by (i): (i)

Aujourd’hui je vends de la jolie tomate. Today 1 sell de+- pretty- tomato lit. ‘Today I have pretty tomato to sell.’ = ‘Today I have (some) pretty tomato produce to sell.’

In our view, the Pulp and Sample readings ultimately boil down to a single (Mass) semantic construal, characterised by number deficiency.

 Anne Zribi-Hertz and Herby Glaude

phrase may be syntactically unspecified for number, and that number deficiency is the source of Mass readings which include the Sample reading exemplified in (3b) and the Pulp reading exemplified in (4b).6 Under our assumption, the Mass/Count contrast is therefore not a lexical distinction, but a semantic effect derived from number specification in syntax.7 Any noun, such as tomato, may a priori be combined with a number specification, as exemplified above — although some nouns favour one type of syntax on account of the type of referent they serve to denote: thus, since liquid referents are physically perceived as nonatomised substances, they naturally call for a Pulp-type reading, so that the nominals which denote them call for [−number] syntax, unless they are combined with a Count classifier (one water = ‘one unit/serving/bottle of water’); conversely, since human referents are likely to be treated as separate individuals, the nominals which denote them tend to call for [+number] syntax. A correlate of our analysis is that the absence of overt plural marking, in a functional-number language, is a priori ambiguous between the negative value of number (−plural, i.e. ‘singular’) and number-deficiency (−number). This distinction is formalised in (5):8 (5a)

negatively specified for Number DP D0

NumP Num0

NP

ma -PL

tomate

= ‘my (single) tomato’

. These two readings are likely not to exhaust the list of possible semantic effects of number deficiency — an open question. Interestingly, while the Pulp semantic effect associated with French (4b,c) is similarly available in the English translation (I ate tomato for breakfast – this is why my allergic rash has come out), the Sample semantic effect associated with (3b) in French is not available in Come and buy my pretty tomato! — which in the mouth of an English-speaking vegetable grocer seems to require a singular construal of the tomato referent. . Our analysis is but a syntactic phrasing of various ideas put forward in the semantic literature. What Pelletier (1979) calls the Universal Grinder boils down — in our terms — to the fact that the Pulp reading is in some languages (e.g. English and French) a productive semantic effect of syntactic Number deficiency. The fact that in such languages, most lexical nouns are a priori open to a Pulp reading (cf. Ware 1979) similarly boils down to the fact that number deficiency is a priori available in syntax, regardless of lexical selection (cf. I ate little girl for breakfast, I stepped in (mashed up) machine-gun, etc.). The assumption that the Mass/Count distinction originates in syntax is also explored and formalised by Doetjes (1997) and Borer (2005). . As pointed out by one referee, the representation given in (5b) runs against two current ideas: (a) the assumption that, for cognitive reasons, number is a necessary ingredient of any

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

(5b)

unspecified for Number DP D0

NP

ma tomate = ‘my tomato produce or pulp’

This ambiguity involving number may be regarded as a specific manifestation of the linguistic principle phrased in (6): (6)

The ambiguity of feature deficiency For any functional feature F, the lack of phonological spell-out for F may a priori indicate either the nonspecification of F, i.e. syntactic deficiency, or the unmarked value of F, i.e. phonological deficiency.9

Our analysis of syntactic number leads us to revise the notion of bareness, in noun phrases. The English bare plural which occurs in (1b) is not functionally bare since it is specified for number, whereas the noun phrase tomato which occurs in the English translation of (4b) may be assumed to be bare if this should mean deficient for both determiner and number. Under our assumptions, English argument noun phrases may also include a determiner but be left unspecified for number, as shown by (7): (7)

a. b.

[Mary had a dog and a cat.] The cat was disgusting. [−plural], Count reading > single cat-creature [Mary had cat and rice for breakfast.] The cat was disgusting. [−number], Mass (Pulp) reading > cat meat10

Our main starting assumption is that the current bareness (bare NP) concept should be replaced by a more fine-grained approach to feature deficiency. First, the distinction denoting nominal argument (cf. Bouchard 2002, 2003); (b) the assumption that so-called mass nouns are pluralised in the lexicon (Chierchia 1998). Under our own analysis, Number deficiency does not stop a noun phrase from denoting; and Mass readings stand as possible semantic effects of syntactic Number deficiency. . This is a very old and basic linguistic idea, exemplified by a subclass of neutralisation situations, in the sense of the Prague School: unvoicedness does not have the same status in a context where the [+voiced] specification is otherwise available, and in one where it is not. The construal of zero morphology as an unmarked value — a negatively-specified value — could be the crucial distinction between inflectional features, for which this property obtains, and lexical features, for which it does not. . One might want to object that the cat is actually specified as [−plural] in (7b), because it is D-linked to a preidentified  referent which is ‘singular’ in the sense that it denotes a single occurrence of cat meat. Note, however, that even in a D-linked context, number specification blocks the Pulp reading; thus, in (i) below, the pluralised noun phrase both cats forces us to

 Anne Zribi-Hertz and Herby Glaude

between lack of determiner and lack of number should be crucially taken into account for the analysis of noun phrases. Second, phonological deficiency (lack of phonological spell-out) should be crucially distinguished from syntactic deficiency (lack of feature specification).

. ‘Definiteness’ and number marking in simplex Haitian noun phrases .

The definite determiner LA

Haitian,11 is a ‘determiner language’, since its lexicon includes a morpheme commonly identified as a definite determiner which we shall gloss as  for convention’s sake. This morpheme is basically spelt out as la and exhibits a good deal of allomorphy.12 In what follows the capitalised transcription LA will be meant to refer to the Haitian definite determiner regardless of its context-dependent phonological spell-out. From a historical perspective, Haitian LA is a recycled development of the French locative adverb là (‘here/there’) (cf. Hazaël-Massieux 1999). The examples given in (8) and (9) illustrate the allomorphic variation of LA, and show, through their English translations, why this morpheme has been labeled a definite determiner:13 (8)

a.

b.

Pòl ekri lèt la. Paul write letter  ‘Paul wrote the letter.’ Pòl ap achte vach la. Paul  buy cow  ‘Paul will buy the cow.’

construe in the second sentence at least two (Count)  referents from which the (Pulp) catmeat mentioned in the first sentence was taken: (i) For breakfast, Mary ate both Siamese cat and Angora cat. Both cats were disgusting. (7b), on the other hand, tells us nothing about the number of cat-creatures which were used to produce the cat-meat. Under our assumption, this contrast directly derives from the fact that cat is unspecified for number in (7b) while both cats is specified as [+plural] in (i) above. We hence maintain our claim that the Pulp construal of the referent is crucially correlated with number deficiency in syntax. The fact that the cat referent is construed in (7b) as a ‘single instance’ of cat-meat is due, we believe, to DP-external factors (verbal aspect, in particular). . On the syntax of French-lexifier Creoles and of Haitian in particular, see Sylvain 1936, Pompilus (1976), Bernabé (1983, 1987), Bickerton (1981), Chaudenson (1992), Damoiseau (1999), DeGraff (1992, 1999b, 2002), Deprez (2000), Gadelii (1997), Germain (1995), Janson (1984), Joseph (1988), Lefebvre (1998), Valdman (1978), among others. .

Depending on phonological context, la is spelt out [la], [a], [lã], [nã] or [ã].

. Abbreviations used in our glosses:  = definite determiner;  = demonstrative;  = feminine gender;  = future tense;  = intensifier;  = locative;  = masculine gender;  = locative preposition;  = past tense;  = plural;  = singular; 1, 2, 3 = 1st, 2nd, 3rd person.

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

(9)

a.

b.

c.

Pòl ap achte chyen an. Paul  buy dog  ‘Paul will buy the dog.’ Pòl ap achte labalèn nan. Paul  buy whale  ‘Paul will buy the whale.’ Pòl leve bebe a. Paul wake up baby  ‘Paul woke up the (aforementioned) baby.’

Two syntactic representations have been considered in the linguistic literature to account for these data: the one sketched in (10), proposed by Gadelii (1997) and Lefebvre (1998), assumes that DP is head-final in Haitian; the one in (11), proposed by Déprez (2000) and Lyons (2000), assumes that DP is universally head-initial and that D-final noun phrases involve NP-raising to spec, DP: (10)

Gadelii (1997), Lefebvre (1998) DP

(11)

NP

D0

vach

la

Déprez (2000), Lyons (2000): NP-raising DP spec

D’ D0

vach

NP

la

As already mentioned above, Haitian LA was historically recycled from the French locative adverb là, which also contributes to noun-phrase functional structure in French as a correlate of demonstrative determiners, cf. cette vache-là, lit. ‘that cow-there’. To derive such correlative structures within the X-bar framework, it may be argued that the NP (or NumP) is raised to its surface position — the specifier of the phrase headed by là. Diagram (11) suggests that this raising rule still applies in the derivation of Haitian vach la; but it further states that the head which hosts LA in Haitian has become a ‘D’ head since the topmost DP layer of French has been discarded. Diagram (10) says like (11) that the head which hosts LA has become a ‘D’ head in Haitian, but it suggests that the NP-raising rule which applies in French does not apply in Haitian. One argument in favour of (11) is of a purely theory-internal nature — it is compatible with Kayne’s (1994) assumption that phrases are universally head-initial. Further empirical

 Anne Zribi-Hertz and Herby Glaude

evidence in support of (11) is that Haitian is otherwise a head-initial language (V, P and C precede their complement).

.

The plural marker yo and its relation to LA

Haitian has a plural marker, yo, which is linearly ordered to the right of the definite determiner: (12)

a.

b.

Pòl ekri lèt (la) yo. Paul write letter   ‘Paul wrote the letters.’ Pòl ap achte chyen (an) yo. Paul  buy dog   ‘Paul will buy the dogs.’

[compare (8a)]

[compare (9a)]

When the plural marker yo occurs, the definite marker LA is deleted in most Haitian dialects, but it remains overtly realised in at least one (Northern) dialectal variety. The most straightforward way to account syntactically for the LA-yo linear ordering is the representation given in (13), proposed by Gadelii (1997) and Zribi-Hertz (2002): (13)

Gadelii (1997), Zribi-Hertz (2002): no movement NumP Num0

DP NP

D0 yo

vach

la

A first problem with this representation is that it is D-final (see above), but it may easily be translated into its D-initial equivalent (13’): NumP

(13’) spec

Num’ DP

Num0 spec yo

D’ D0

NP

la

vach

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

A second and more serious problem with (13)–(13’) is that it conflicts with a widespread understanding of the ‘DP Hypothesis’, according to which D crucially pertains to the left periphery and DP should hence be the topmost projection of the noun phrase.14 Two proposals have been made to fit such data as (12) into the DP Hypothesis: one is by Lefebvre (1998) and is represented in (14); the other is by Aboh (2002), whose assumption is put forward for Gungbe but is adapted here to Haitian: (14)

Lefebvre (1998): plural adjoins to D DP NumP

(15)

D

NP

Num0 +df

+pl

vach

la

yo

Aboh (2002, for Gungbe): cyclic NP-raising to spec, DP DP spec

D’ D0

NumP spec

Num’ Num0

vach la

WP

yo

The structures in (14) and (15) rescue the conventionally-understood DP-Hypothesis but face some other problems. Lefebvre’s assumption that the plural adjoins to D incorrectly suggests that the Haitian definite marker is inflected for number, while it seems on the contrary quite clear that Haitian has characteristically done away with inflected determiners. Aboh’s proposal doesn’t run into this serious problem, but includes a D head whose feature content is unclear, since LA is generated here in the WP phrase. In order to straighten up this problem, we might consider generating LA in D in (15), with the NP raising cyclically to spec, DP.

.

Cf. for instance Cardinaletti & Starke (1999), Longobardi (1999, 2001).

 Anne Zribi-Hertz and Herby Glaude

However, a further important property which must be expressed by whatever analysis we should select is that Plural selects Definiteness in Haitian. This is exemplified in (16) and (17):15 (16)

a.

Pòl achte de vach. Paul buy two cow ‘Paul bought two cows.’ b. Pòl achte de vach yo. Paul buy two cow  ‘Paul bought the(se) two cows.’ ≠ ‘Paul bought two cows.’

(17)

a.

Pòl ekri lèt. Paul write letter ‘Paul is writing mail.’ b. Pòl ekri lèt yo. Paul write letter  ‘Paul is writing the(se) letters.’ ≠ ‘Paul is writing letters.’

The examples show that the occurrence of the plural marker yo automatically calls for the so-called definite reading of the referent, glossed in English by means of the definite article the. In other words, a pluralised noun phrase such as vach yo is semantically equivalent to its dialectal variant vach la yo, which overtly contains the definite marker. But conversely, the occurrence of the definite marker LA does not necessarily involve number specification in the noun phrase, as revealed by the ambiguity of (18a,b): (18)

a.

b.

Pòl ap achte ze a. Paul  buy egg  (i) ‘Paul will buy the (single) egg-item.’ (ii) ‘Paul will buy the egg-material.’ (i.e. that unspecified quantity of  produce which is needed for some purpose or other) Pòl achte mori a. Paul buy codfish  (i) ‘Paul bought the (single) codfish animal.’ (ii) ‘Paul bought the codfish {meat/produce}.’

Under the assumption introduced in Section 1, we claim that the interpretations glossed under (18a(ii)) and (18b(ii)) correlate with Number deficiency in syntax. Hence, definite LA does not require number specification, whereas as shown in (16)–(17), . From now on, we shall systematically delete LA in our Haitian examples whenever it is followed by yo. In this respect, our data instantiate the most ‘advanced’ (also the most widespread) dialectal variety of Haitian.

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

overt number specification requires definiteness. This asymmetrical dependency is not captured by the diagram in (14), while it is by the structures proposed in (13)–(13’), as well as by the structure in (15). Since (15), but not (13)–(13’), includes an ad hoc D head, the structural representation which appears to us as optimally adequate to account for the described data is (13’), a head-initial structure where Number (yo) dominates D (la). For purely typographical reasons, however, we shall adopt below head-final representations similar to (13), every one of which may easily be translated into its D-initial counterpart as in (13’). The diagrams in (19) represent the ambiguity of (18b): (19)

mori a ‘the codfish’ a. unspecified for Number: ‘the codfish meat/produce’ DP

b.

NP

D0

mori

LA

negatively specified for Number: ‘the (single) codfish animal’ NumP Num0

DP NP

D0

mori

LA

-PL

The structure in (19b) leaves us with the DP-Hypothesis twist mentioned above: how can DP – the ‘left periphery’ — be located below NumP — an inflectional projection – in syntax? Our belief is that the Haitian data create neither Twist nor Paradox: the key to our apparent problem is that the so-called definite marker is of an inherently different nature in Haitian and in French. .

How can NumP stand above DP in syntactic structure?

The Haitian definite determiner has a strong deictic value, which we may correlate to its locative nature, and which makes it contrast sharply with the French definite article.16

. The semantic constrasts between French and Haitian definite determiners are further developed in Zribi-Hertz (2002) — from a synchronic perspective. For a diachronic approach to Creole determiners, see Hazaël-Massieux (1999). On the bound variable character of the French definite article, see Guéron (1985, 1992, 2003), who focuses on inalienable-possession constructions, and Zribi-Hertz (to appear), who explores the general assumption that French definite articles are topic-bound pronouns.

 Anne Zribi-Hertz and Herby Glaude

Thus, while French le is associated in (20a) with a generic reading, Haitian LA only allows a specific reading in (20b): (20)

a.

Paul aime le riz. Paul likes  rice ‘Paul likes rice.’ b. Pòl renmen diri a. Paul like rice  ‘Paul likes the rice.’ ≠ ‘Paul likes rice.’

While French la is open to a variable reading in (21a), only the referential reading is allowed in (21b): (21)

a.

Paul a éteint la lumière, et moi aussi. Paul turned off  light and so did I. b. Pòl etenn limyè a, ak mwen menm tou. Paul turned off light  and 1  too ‘Paul turned off that light, and so did I.’ ≠ ‘Paul turned off the light (wherever he was), and so did I (wherever I was).’

While French la is open to the ‘inalienable-possession’ reading in (22a), only the alienable reading is allowed in (22b): (22)

a.

Paul a levé la tête. Paul has raised  head lit. ‘Paul raised the head.’ = ‘Paul raised his head.’ b. Pòl leve tèt la. Paul raised head  ‘Paul raised the head.’ ≠ ‘Paul raised his head.’

While French le is open to a variable reading in (23a), only a referential reading is available in (23b): (23)

a.

Sonnez: le boucher va vous servir. ring  butcher will 2 serve ‘Ring the bell: the butcher [whichever one is on duty] will come and serve you.’ b. Sonnen: bouche a ap vin sè vou. ring: butcher   come serve 2 ‘Ring the bell: {this/that} butcher will come and serve you.’ ≠ ‘Ring the bell: the butcher [whichever one is on duty] will come and serve you.’

The semantic contrasts brought out by (20)–(23) may be subsumed under the descriptive assumption that the French definite article is construed as a bound variable, whereas Haitian LA is not. Following Zribi-Hertz (to appear), we derive the variable

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

semantic behaviour of the French definite article from its pronominal nature, which crucially involves topical binding — a property assumed to be anchored in the left periphery of DP. Haitian LA, on the other hand, is not a pronoun, but a deictic locative morpheme, categorially akin to sa – the so-called demonstrative, a recycled development of French ça (the spoken variant of the neuter, demonstrative pronoun cela). In Modern Haitian, sa is selected by LA, as shown by (24c): (24)

a.

Pòl ekri lèt la. Paul write letter  ‘Paul wrote the/this/that [no ostensive signal] letter.’ b. Pòl ekri lèt sa a. Paul write letter   ‘Paul wrote this/that [ostensive signal] letter.’ c. *Pòl ekri lèt sa.

The contrast between (24a) and (24b) is that (24b) involves two layers of deixis marking, while (24a) only involves one. Correlatively, (24b) suggests strong deixis (ostension), while (24a) suggests weak deixis (presupposed spatialisation). As witnessed by the ungrammaticality of (24c), demonstrative sa must be selected by definite LA. The Haitian plural marker yo, on the other hand, is a recycled development of eux, the French nonclitic 3 pronoun. The LA yo linear ordering observed in Haitian may be related to the [LA, eux] linear sequence observed in French (25), where là, as a syntactic correlate of the demonstrative determiner, is positioned inside the DP while eux, as a dislocated DP-adjunction, is positioned outside the DP: (25)

Ces chiens -là, eux, aboient rarement.17 - dog-  3 bark-3 rarely ‘Those dogs rarely bark.’

In short, the syntactic representation proposed in (13)–(13’) does not lead us to discard the ‘DP Hypothesis’, provided we realistically accept that the conventional labels . As pointed out to us by K. Gadelii (p.c.), who draws his data from Goodman (1964), the LA yo linear order exhibited by Haitian is not the only attested one in those French-lexifier creoles which recycled a 3 pronoun as a plural marker; thus, where Lousisianais creole has (i-a) (like Haitian), (i-b) is found in Guyanais: (i) a. Lousianais madam la ye woman   ‘the women’ b. Guyanais fam ye la woman   ‘the women’ This dialectal variation may be correlated with the fact that French locative là may not only be the DP-internal correlate of the demonstrative determiner, as in (25), but also a DP-external

 Anne Zribi-Hertz and Herby Glaude

DP and NumP identify positions rather than categories (cf. Milner 1989). NumP dominates DP in Haitian while DP dominates NumP in French because the positions labeled Num and D do not host the same categories in the two languages.

.

Doubly-‘definite’ DPs

Gadelii (1997: 142) further notes that a Haitian DP may contain two distinct occurrences of LA, positioned below and above Number, and which he respectively labels DF (definite) and DEIX (deixis), as in (26): (26)

liv mwen sa a yo (a) book 1     lit. ‘those books of mine over there’

However, as shown above, even the morpheme which Gadelii labels definite has a deictic force in Haitian. The two instances of LA (> a) which occur in (26) are historically derived from the same lexical source (French là). From a categorial viewpoint, there is only one LA morpheme in Haitian, but as shown in (26), it can occupy two distinct and combinable functional heads to which Gadelii assigns different labels for convenience’s sake. These facts suggest that what is glossed as  (definiteness) in our examples does not identify a feature, a category, or even a structural position, but only a conventional label inspired by the semantics of French-type definite articles. The double occurrence of LA observed in (26) is perfectly consistent with the deictic nature of this morpheme, if we assume that deixis may recursively contribute to referent identification. In this respect, Haitian LA sharply contrasts with the French definite article, whose number of occurrences is limited to one in a maximal noun phrase.

. ‘Bare NPs’ The three functional morphemes LA, yo and sa of Haitian considered above, all trigger a referential, spatialised construal of the referent: LA and sa because of their deictic character, and yo because it always selects deictic LA as its complement. As a result, none of the Haitian determiners considered above allows a nonspatialised reading, i.e. an interpretation which does not link the referent to a specific portion of space.

adjunction, as in (ii) below, where we see that it is no longer selected by the demonstrative. In such a case, the two DP-adjunctions, eux and là, are freely ordered in French: (ii)

a.

b.

{les/ces} chiens, eux, là, ils aboient trop. ⁄- dog- 3  3 bark too much lit. ‘The/those dogs, them, there, they bark too much.’ {les/ces} chiens, là, eux, ils aboient trop. df/dm-pl dog-pl loc 3pl 3mpl bark too much lit. ‘The/those dogs, there, them, they bark too much.’

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

It is thus not very surprising to find that Haitian licenses bare NPs, which take up whatever interpretations do not involve spatialisation: (27)

a.

b.

c.

d.

Pòl renmen vach. Paul like cow ‘Paul likes cows (any member of the  species).’ Pòl etenn limyè, ak mwen menm tou. Paul turn off light and 1  too ‘Paul turned off {the light/lights} and so did I [variable reading only].’ Sonnen: bouche ap vin sè vou. ring butcher  come serve 2 ‘Ring the bell: the butcher(s) (whichever one(s) is/are on duty) will come and serve you.’ Soup bon pou bouche. soup good for butcher ‘Soup is good for {butchers/the butcher (whoever he is)}.’

‘Bare NPs’ are unspecified for number not only in morphology, but also in their interpretation, as witnessed by the strongly context-sensitive construal of their referent (cf. (27b,c)). We thus assume that they lack a Number projection in syntax. To capture the fact that their referent is construed as non-spatialised, we assume that they include a D head hosting a [–locative] feature value, contrasting with the [+locative] value spelt out by LA, which, as pointed out above (Section 2.2), undergoes phonological deletion in the presence of yo in most Haitian varieties: (28)

a.

vach (= (27a)) DP

(28)

b.

NP

D

vach

–loc ⇓ ø

vach (la) yo (= (13)) NumP

DP NP

Num D +pl

vach

+loc ⇓ LA (⇒ø)

⇓ yo

 Anne Zribi-Hertz and Herby Glaude

These diagrams instantiate two different cases of phonological deficiency: in (28a), the zero determiner spells out the unmarked value of the [locative] feature; in (28b), D hosts the marked value of the [locative] feature in syntax, prior to phonological deletion.

.

Summary

The main descriptive results of this section are listed below: (i)

The so-called definite determiner has a different categorial nature in French and in Haitian. Because the French definite article is a topic-bound pronoun, it is located in the left-periphery of the noun phrase, hence above the Number projection which pertains to the inflectional domain, and its number of occurrences is limited to one. Because the Haitian plural marker yo is recycled from a rightadjoined 3 pronoun, its syntactic projection — labeled Number Phrase — stands above LA, itself recycled from a DP-internal locative correlate of the French demonstrative. We have attempted to show why this state of affairs causes no unsolvable paradox as far as the DP-Hypothesis is concerned. (ii) In Haitian as in French, a nominal argument is not necessarily specified for number. Number deficiency correlates with Mass semantic effects which include those we have labeled above Pulp and Sample. (iii) Because all overt determiners (sa, LA, yo) have a deictic effect in Haitian, the occurrence of an overt determiner in this language cannot be associated with a nonspatialised construal of the referent. Hence, ‘bare NPs’ — i.e. noun phrases lacking an overt determiner — freely occur as arguments in this language, and they take up whatever semantic interpretations are not available for overt determiners. In this respect, bare NPs may be described as default nominals in this language. We have proposed to analyse Haitian ‘bare NPs’ as number-deficient DPs whose D head hosts the unmarked feature value [–locative].

. .

Definiteness and Number in Haitian relativised noun phrases The complex structure of relativised noun phrases

In what follows we shall be calling relativised DP a noun phrase which includes a restrictive relative clause. An aspect of Haitian relativised DPs which seems at first glance remarkable to a French or English speaker is that they may include determiners and plural markers in two distinct and combinable positions. Thus, the English noun phrase the codfish which Paul buys/bought18 may translate as (29a), (29b), (29c) or (29d); and . In Haitian as in English, the noun mori ‘codfish’ easily combines both with [+number] syntax (> ‘one or several codfish (item(s)’), and with [−number] syntax (> ‘codfish meat, pulp or produce’).

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

the (various) codfish (items) which Paul bought, as (30a), (30b) or (30d) — though (30c) is ungrammatical: (29)

(30)

the codfish which Paul bought a. mori Pòl achte codfish Paul buy b. mori a Pòl achte codfish  Paul buy c. mori Pòl achte codfish Paul buy d. mori a Pòl achte codfish  Paul buy

a  a 

the codfish (items) which Paul bought a. mori yo Pòl achte codfish  Paul buy b mori Pòl achte yo codfish Paul buy  c. *mori yo Pòl achte yo codfish  Paul buy  d mori yo Pòl achte a codfish  Paul buy 

To account for these facts we shall adopt Kayne’s (1994) general approach to restrictivelyrelativised noun phrases. His leading idea is that they crucially involve two layers of noun phrase structure: a larger DP which contains a CP projection, and an argument noun phrase which raises out of the embedded clause to the spec of CP: (31)

Definite relativised noun phrases according to Kayne (1994)

DP CP

D0 spec

C' C0

the

IP

codfishz

XP that

Mary bought

tz

Kayne’s leading assumption is that in a complex noun phrase such as (31), the string the codfish does not form a syntactic constituent, but results from a nondefinite noun phrase (codfish) raising up to spec, CP. The issue of the categorial label of the embedded noun phrase tagged ‘XP’ in (31) is left unsettled by Kayne, who only suggests that it must not be ‘definite’ — the definite determiner belonging upstairs. Kayne’s analysis is strongly criticised by Borsley (1997), whose main objection is that the raised expression in structure (31) must be a maximal projection, therefore a

 Anne Zribi-Hertz and Herby Glaude

‘DP’, since it stands as an argument within the embedded clause. It follows that Kayne’s grammar generates ungrammatical strings such as *the the codfish that Mary bought. We do not believe, however, that this overgeneration problem is a necessary consequence of Kayne’s general theory. First, empirical evidence suggests that in such languages as English or French, the raised argument in structure (31) is  ‘definite’ (cf. Smith 1964, Vergnaud 1985, Kayne 1994). Second, assuming that XP in (31) must be a maximal projection (whatever its categorial label) is not conceptually problematic if the D-CP relation is seen as adjunction rather than government (cf. Zribi-Hertz to appear). The analysis of relativised noun phrases sketched in (31) supplies a convenient means of accounting for the double functional specification exhibited in (29) and (30). Assuming that a relativised structure such as (31) actually involves two noun phrase projections, an  noun phrase, and a  noun phrase, it is not unexpected to find doubly-determined relativised noun phrases in Haitian, as illustrated by (29d), which we may roughly represent as in (32) (leaving out Number, discussed below): (32)

relativised noun phrase with two definite determiners, e.g.: mori a Pòl achte a (28d): DP

CP spec

D0 C'

C0

IP DP Pòl achte

ø

NP mori

D0 +loc LA

+loc LA

As a result of Relativisation, the downstairs determiner surfaces in an internal position within the larger DP, while the upstairs determiner surfaces in the phrase periphery. The result is a double-D noun phrase, not unsimilar to that in (26). We shall further continue to assume, as we did above in Section 2, that nonpluralised noun phrases, in Haitian, are syntactically ambiguous between a [–plural] reading correlating with the negative value of Number (a ‘singular’ construal of the referent), and a [–number] reading correlating with the absence of the Number projection (‘Pulp’ or ‘Sample’ construal of the referent). This means that each one of the projections labeled DP in diagram (32) could a priori be either specified for the singular, or unspecified for number. To sort out this issue, we shall first look at relativised noun phrases which include an overt number specification, and shall return to nonpluralised noun phrases afterwards.

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

.

The distribution of Number in relativised noun phrases

.. Plural relativised noun phrases As witnessed by paradigm (30), repeated below, the [+plural] value of Number may be specified downstairs (30a) or upstairs (30b), but it cannot be specified at both levels (30c): (30)

the codfish (items) which Paul bought a. mori yo Pòl achte codfish  Paul buy b mori Pòl achte yo codfish Paul buy  c. *mori yo Pòl achte yo codfish  Paul buy  d. mori yo Pòl achte a codfish  Paul buy 

The ill-formedness of (30c) could derive from a general restriction to the effect that number specification, unlike deixis, may contribute only once to identify a single referent. We argued in Section 2 that Number selects DP in Haitian, and that noun phrases which surface as ‘bare’ are number-deficient, [−locative] DPs. It follows that the noun phrase in (30a) combines a fully saturated NumP downstairs with a numberdeficient, [−locative] DP upstairs, while the noun phrase in (30b) combines a fully saturated NumP upstairs with a ‘bare noun phrase’ — i.e. a number-deficient [−locative] DP — downstairs: (33)

a.

mori yo Pòl achte (= (30a)) DP

CP

D

spec

C'

[mori yo]z

b.

C0

IP

ø

Pòl achte

NumP –loc tz

mori Pòl achte yo (= (29b)) NumP Num0

DP D0

CP C'

spec C0 moriz

ø

IP

Pòl achte

DP – loc tz

+loc (LA⇒ø)

yo

 Anne Zribi-Hertz and Herby Glaude

The result of these combinations is a complex noun phrase which is number-deficient (or ‘bare’) at one level but not at the other. The well-formedness of (30d) further reveals the availability of the combination represented in (34): (34)

mori yo Pòl achte a (= (30d) DP

CP

D0 C'

spec C0

IP

[mori yo]z

NumP ø

Pòl achte

tz

+loc LA

In this case, the upstairs determiner cannot be associated with its own Number projection. That Number cannot be projected both upstairs and downstairs is straightforwardly revealed by the ill-formedness of (30c). As regards (34)/(30d), if Number were projected upstairs, its value would be [–plural], which would conflict with the plural downstairs. We thus assume that (34)/(30d) involves Number (hence Definiteness19) downstairs, but bare Definiteness (i.e. Definiteness without Number) upstairs. As naturally expected, the three structures in (33a), (33b) and (34) are not associated with the same semantic interpretations. Upstairs plural marking, as in (33b), triggers a spatialised and plural construal of the referent in which the plural has scope over the entire noun phrase: in other words, an antipartitive semantic effect. Downstairs plural marking with a ‘bare noun phrase’ upstairs, as in (33a), triggers a partitive effect implying that the several items that Paul bought were part of a larger set of available codfish animals. Combining downstairs plural with upstairs definiteness, as in (34), triggers a discourse-linked reading. In (35) below we propose an English translation for each of these three patterns: (35)

a.

mori Pòl achte yo

:

b

mori yo Pòl achte

:

c.

mori yo Pòl achte a :

‘(all of) the codfish items which Paul bought’ ‘those of the codfish which Paul bought’ ‘these [aforementioned] codfish which Paul bought’

[= (33b)] [= (33a)] [= (34)]

. We use the term Definiteness here as a convenient label referring to whatever functional feature is hosted by the D head. As shown above in Section 2, Definiteness has a different feature content in French and in Haitian: in Haitian, Definiteness is expressed by a [locative] feature.

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

The three semantic interpretations glossed in (35) are felicitous in different discourse contexts, as shown in (36)–(38): (36)

Lè Pòl al nan mache, when Paul go P market ‘When Paul goes to the market, a. mori li achte yo, se pou Elsi. codfish 3sg buy   for Elsi ‘all the codfish animals he buys are for Elsi.’ b. %mori yo li achte, se pou Elsi.20 codfish  3sg buy  for Elsi ‘those of the codfish which he buys are for Elsi.’

(37)

Mèkredi, gen anpil bèl mori nan mache. Wednesday have a lot nice codfish P market ‘On Wednesday, there {was/were} lots of nice codfish at the market.’ a. Mori yo Pòl achte te pli gwo pase tout. codfish  Paul buy  more big than all ‘Those of the codfish which Paul bought were the biggest of all.’ b. *Mori Pòl achte yo te pli gwo pase tout. codfish Paul buy   more big than all lit. ‘All the codfish items Paul bought were the biggest of all.’

‘

(38)

Mèkredi, Pòl achte de mori nan mache. Wednesday Paul buy two codfish P market ‘On Wednesday, Paul bought two codfish (items) at the market.’ a. %Mori yo li achte, se te pou Elsi. codfish  3sg buy   for Elsi ‘Those of the codfish which he bought were for Elsi.’ b. Mori li achte yo, se te pou Elsi. codfish 3sg buy ,   for Elsi. ‘The codfish (items) which he bought were for Elsi.’ c. Mori yo li achte a, se te pou Elsi. codfish pl 3sg buy    for Elsi ‘Those codfish (items) which he bought, they were for Elsi.’

Summarising, we assume on empirical grounds that a pluralised relativised noun phrase, in Haitian, involves two levels of functional specification, allowing three different feature combinations: ● ● ●

Number upstairs, ‘bare noun phrase’ downstairs Number downstairs, ‘bare noun phrase’ upstairs Number downstairs, Definiteness upstairs

. Examples marked by ‘%’ are syntactically well-formed but infelicitous in the given discourse context.

 Anne Zribi-Hertz and Herby Glaude

Each pattern triggers a different construal for the referent: ● ● ●

Number upstairs has scope over the entire noun phrase Number downstairs has scope over the downstairs noun phrase only Definiteness upstairs triggers a discourse-linking effect

..

Nonpluralised relativised noun phrases

For nonpluralised relativised noun phrases, we find the four patterns exemplified in (29), repeated below in (39) with a more accurate English translation: (39)

a.

b

c.

d.

[Mori Pòl achte], se pou Elsi. codfish Paul buy  for Elsi ‘Whatever codfish Paul buys is/are for Elsi.’ [Mori Pòl achte a], se pou Elsi. codfish Paul buy   for Elsi ‘This codfish {meat/produce/item} which Paul bought is for Elsi.21 [Mori a Pòl achte], se pou Elsi. codfish  Paul buy  for Elsi ‘The one codfish which Paul bought is for Elsi.’ [Mori a Pòl achte a], se pou Elsi. codfish  Paul buy   for Elsi ‘This single codfish which Paul bought, it is for Elsi.’

In (39a), the bracketed noun phrase includes no overt occurrence of the definite determiner and is correlatively unspecified for Number. The semantic effect is rendered in the English translation by a universally quantified nonspecific relativised noun phrase whose referent may be construed either as Pulp or as Sample. In (39b) and (39c), the bracketed noun phrase contains one definite determiner positioned at the phrase periphery in (39b), and in an internal position in (39c). The right periphery determiner triggers a discourse-linking effect, glossed here in English by this, implying that the buying of one or some codfish by Paul is given information. Interestingly, the phrase mori (‘codfish’) may be construed in (39b) either as Pulp (codfish meat) or Sample (unspecified sample of codfish items), or as singular (single codfish creature): we hence assume that the upstairs DP in this example is ambiguously construed either as [−number] (unspecified for Number) or as [−plural] (singular). Contrasting with the external determiner of (39b), the internal determiner in (39c) only allows for a singular construal of the phrase mori, as translated by one, which suggests a partitive implication (i.e. the one codfish which Paul bought was a member of a larger set of available codfish items). We hence assume that in a Haitian complex noun phrase containing a restrictive . The morphologically unspecified tense in the relative clause may a priori translate into English as a generic present (‘buys’), a present perfect (‘has bought’) or a preterite (‘bought’). We have chosen to associate a generic-present translation with unspatialised relativised noun phrases, and a preterit or present perfect translation with noun phrases including a spatialiser at one level and/or the other. Hence, the translations we propose for the Haitian examples hardly ever exhaust all possible interpretations.

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

relative clause, downstairs LA, unlike upstairs LA, is always selected by Number. We thus predict that those nouns whose semantic content calls for [−number] syntax (i.e. strictly Mass nouns, such as ‘mud’ (labou)) do not combine with the feature patterns exemplified by (39c,d): this prediction is confirmed below by the ungrammaticality of (40c,d): (40)

a.

[labou Pòl ranmase], se pou Elsi. mud Paul pick up  for Elsi ‘Whatever mud Paul picks up is for Elsi.’ b. [labou Pòl ranmase a], se pou Elsi. mud Paul pick up   for Elsi ‘This mud which Paul has picked up is for Elsi.’ c. *[labou a Pòl ranmase], se pou Elsi. mud  Paul pick up  for Elsi lit. ‘The one mud which Paul has picked up is for Elsi.’ d. *[labou a Pòl ranmase a], se pou Elsi. mud  Paul pick up   for Elsi lit. ‘This single mud which Paul has picked up is for Elsi.’

In Haitian as in the English translations, such examples as (40c,d) would only be acceptable if construed as including an implicit classifier allowing for [−plural] syntax, which is pragmatically out of order with  (‘one bottle of mud’?). As suggested by our English translations, (39b) and (39d) share one interpretation. According to our own intuition, the two examples crucially contrast in that (39b) allows either for a [−number] or for a [−plural] construal of the noun phrase, whereas (39d) only allows for the [−plural] reading: we thus suspect that the syntax exemplified by (39d) should be more frequently chosen over that of (39b) for discourse-linked singular readings, while the syntax of (39b) should be selected for discourse-linked [−number] (e.g. Pulp or Sample) readings. Pursuing the reasoning introduced in Section 2.1, we assume that in their [−plural] readings, the Examples in (39) are the singular counterparts of (30), while in their [−number] readings, they involve syntactic number deficiency both upstairs and downstairs: (41)

a.

(= (39a))

mori Pòl achte ‘whatever codfish Paul buys’ DP

CP spec

D C'

C0

TP spec

T' VP

T0 spec

moriz ø Pòlk

V' V0

DP

achte

tz

ø tk

– loc

 Anne Zribi-Hertz and Herby Glaude

b.

(= (39b) with Mass or Sample reading) mori Pòl achte a ‘this codfish meat/produce which Paul bought’ DP D0

CP spec

C' C0

TP

spec

T' VP

T0 moriz ø

spec

V'

Pòlk ø tk

c.

V0

DP

achte

tz

+loc LA

(= (39b) with count reading)

mori Pòl achte a

‘this codfish item which Paul bought’ NumP Num0

DP

D0

CPD spec

C' C0

TP spec

moriz

T'

ø

spec Pòlk

d.

VP

T0

ø

tk

V' V0

DP

achte

tz

+loc LA

– pl

(= (39c))

mori a Pòl achte

‘the one codfish which Paul bought’ DP CP spec

D C'

C0

TP

spec

T' VP

T0 spec [mori-LA-ø]z ø

V' V0

Pòlk ø tk achte

NumP tz

– loc

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

(41)

e.

(= (39d))

mori a Pòl achte a

‘this one codfish which Paul bought’ DP

CP spec

D0 C'

C0

TP spec

[mori-LA-ø]z

T' T0

VP spec

ø Pòlk ø

tk

.

V' V0

NumP

achte

tz

+loc LA

French relativised DPs: the teachings of Haitian

The data presented in Section 3 have revealed that Haitian relativised noun phrases may include no overt functional specification (surfacing as ‘bare DPs’), an upstairs bare DP with a downstairs NumP or bare DP, an upstairs NumP with a downstairs bare DP, or a downstairs NumP with an upstairs bare DP. We have also assumed on empirical grounds that number cannot be independently specified both upstairs and downstairs within a single complex noun phrase. As recalled in Section 2, the assumed French analogue of Haitian LA is the so-called definite article, spelt out as le (nonplural, masculine), la (nonplural, feminine), les (plural), and which for simplicity’s sake we shall from now on designate as LE. As also recalled in Section 2, French LE differs from Haitian LA in both its inherent feature content (LE is an inflected bound pronoun while LA is an uninflected deictic locative) and its structural position (LE dominates NumP while LA is dominated by NumP). Nevertheless, both French LE and Haitian LA are commonly given the same categorial label — definite determiner — in the linguistic literature. One likely reason is that both seem to occupy similar paradigmatic positions within their respective determiner systems: each of them instantiates a functional item whose feature content contributes to referent-identification but is morphologically distinct from both the ostension marker (ce+ci/là in French, sa in Haitian) and the plural marker (-(e)s in French, yo in Haitian). Another interesting similarity between Haitian LA and French LE is that their occurrence is somehow required whenever functional number is specified: in Haitian, this is due to the Plural>Definiteness selectional restriction discussed in Section 2; in French, it is due to the fact that gender and (functional22) number are generally spelt out on D.23 Let us now consider the functional structure of French relativised noun phrases. An important morphological constraint, pointed out above, is that their phi-features

 Anne Zribi-Hertz and Herby Glaude

(gender and functional number) must be externalised on the definite article, generated in the D-head. A central assumption of Kayne’s (1994) analysis, which echoes both diachronic facts (cf. Meillet & Vendryès 1979, Muller 1996) and an idea informally expressed by Vendler (1967), and variously formalised by, e.g. Smith (1964), or Vergnaud (1985), is that within a French definite relativised DP, the definite article partakes in a head-head correlation with the complementiser, i.e. it heads a DP-CP shell within which argument raising (Relativisation) occurs. Thus, the syntactic structure of les livres que Paul a lus (‘the books which Paul read’) may be roughly represented as in (42): (42)

les livres que Paul a lus DP D0

CP spec

C' C0

les livresz

IP spec

que Paul

I' XP a lus tz

A question immediately raised by this tree diagram is that of the categorial identity of the phrase labeled XP — the argument raised to spec, CP. Kayne (1994) labels it NP, but assuming that the plural number must be syntactically represented somewhere in the DP under consideration, XP in (42) could also be NumP. We on the other hand assume, following Kayne, that XP in French must be nondefinite, hence cannot be ‘DP’.24 . Functional number (the ±plural contrast) must be distinguished from lexical number, as expressed by cardinals. If a cardinal occurs, D ceases to be required, as witnessed by the Haitian examples in (16), which are paralleled in French: (i)

a. b.

Paul a acheté deux vaches. ‘Paul bought two cows.’ Paul a acheté les deux vaches. ‘Paul bought the two cows.’

. This is a consequence of the phonological deletion of word-final consonants which started out in Old French and was finalised in the 17th century. Due to this phonological event, the plural ending – s fails to be pronounced on most nouns in Modern French, whence plural marking is only audible on determiners — e.g. les, pronounced [le] or [lε]. This phi-feature externalisation requirement contributes to account for the fact that the distribution of determinerless noun phrases is far more restricted in French than in, e.g. English, or other Romance languages. Although some modern authors seem to attribute it to Delfitto & Schroten (1991), this correlation goes back to French traditional grammarians (cf. Wagner & Pinchon 1962). . The raised XP may be Number Phrase, as in (i), (ii), or a [−number] phrase construed as a Mass existential indefinite, as in (iii). Whether the necessary adjustments made explicit in (ii-b) and (iii) pertain to syntax or morphology is an open issue:

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

It follows that unlike their Haitian homologues described in Section 3, French definite relativised DPs may only include one D-head, upstairs, which both supports the D-C correlation and provides a unique locus for phi-feature externalisation. We shall now explore the idea that despite the uniqueness restriction on the Dhead which prevails in French, the distribution of number and locative markers in relativised DPs and the correlated semantic effects are largely similar in French and in Haitian. Consider the French relativised DPs which are bracketed in (43) and their various possible interpretations:25 (43)

a. (i) (ii) (iii) b. (i) (ii)

[La morue que Paul achète] est pourrie.  codfish that Paul buys/is buying is rotten ‘Whatever codfish {meat/produce} Paul buys is (always) rotten.’ ‘The codfish {meat/produce} which Paul is buying is rotten.’ ‘The one codfish which Paul is buying is rotten. [Cette morue que Paul a achetée] est pourrie. ‘This codfish {meat/produce} which Paul bought is rotten.’ ‘This single codfish which Paul bought is rotten.’

In French as in Haitian above, a nonpluralised noun phrase headed by the noun morue (‘codfish’) may a priori be read as Pulp or Sample (cf. (43a–i, ii), (43b–i)) or as singular (−plural) (cf. (43a–iii), (43b–ii)), an ambiguity which we propose to derive, under the assumption in (5)–(6), from the intrinsic ambiguity of plural deficiency. Both the [−number] and the [+number] readings are compatible with the definite article, which — due to the restrictive relative clause — triggers a non-discourse-linked reading (cf. (43a)), and with the demonstrative determiner, which triggers a discourse-linked reading (cf. (43b)).26 From a semantic perspective, these French data are essentially parallel to the Haitian data in (29). However, where four different strings are spelt out (i)

a. b.

Marie a acheté deux chiens. ‘Marie bought two dogs.’ les [deux chiens] que Marie a achetés t

‘the two dogs (ii)

a. b.

(iii)

a. b.

that Marie bought

Marie a acheté un chien. ‘Marie bought a dog.’ le [(*un) chien] que Marie a acheté t the (*a) dog that Marie bought Marie a eu du mal à le trouver. ‘Marie had trouble finding it’ le [(*du) mal] que Marie a eu t (à le trouver) the DU trouble that M. had finding it

. The interpretation of the bracketed noun phrase in (43a) is sensitive to the aspectual construal of the present tense, which in French may or may not be anchored to the Utterance Time (cf. English buys vs. is buying). . Following Zribi-Hertz (2002), we assume that the French demonstrative determiner is, like the definite article, generated in the D-head, rather than in some specifier position.

 Anne Zribi-Hertz and Herby Glaude

in Haitian, only two surface in French — (43a) and (43b) — distinguished by the content of the D-head: la (the  definite article, unspecified for spatialisation and glossed below as [−]) vs. cette (the  demonstrative determiner, glossed as [+]). (44) Haitian

French

English translation

a.

mori Pòl achte

la morue que Paul achète

b.

mori Pòl achte a

cette morue que Paul achète/a achetée

c.

mori a Pòl achte

d.

mori a Pòl achte a

la morue que Paul {achète/a achetée} cette morue que Paul {achète/ a achetée}

{the/whatever} codfish (meat/ produce) Paul buys {this/that} codfish (meat/produce) Paul {is buying/ bought} the one codfish Paul {is buying/bought} {this/that} single codfish Paul {is buying/bought}

The interpretive options which are provided in Haitian by the syntactic patterns distinguished in (41) are also available in French and English. In all three languages the semantic construal of the relativised noun phrase is sensitive to the aspectual properties of the clause — but we shall leave this issue out of the present study and focus on the semantic effects of DP functional structure. Let us assume that each construal of the noun phrase is triggered by the same feature combination in both French and Haitian: the Pulp and Sample readings result from syntactic Number deficiency (i.e. the lack of the Number projection), the singular reading is triggered by the negative value of the Number feature, and the discourse-linking effect is triggered by an upstairs locative spelt out as LA in Haitian, and by cette in French: (45)

a. Number deficiency > Pulp or Sample reading DP CP

D0 spec

C' C0

–LOC moruez

que +LOC moruez que

IP Paul achète Paul a achète

XP[-number] > la morue que P. achète tz ‘whatever codfish Paul buys’ tz > cette morue que P. achète ‘this codfish meat/produce P buys ’

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

b.

Number upstairs > singular with wide scope DP NumP

D0 Num0

CP spec

C' C0

– LOC –PL moruez

que

+ LOC –PL moruez

que

c.

IP XP[-number] Paul a achetée tz > la morue que P. a achetée ‘the codfish item P bought’ Paul a achetée tz > cette morue que P a achetée ‘this codfish item P bought’

Number downstairs > singular with narrow scope DP CP

D0 spec

C' C0

– LOC moruez

que

+ LOC moruez

que

–PL –PL

IP XP[-plural] Paul a achetée tz > la morue que P a achetée ‘the one codfish P bought’ Paul a achetée tz > cette morue que P a achetée ‘this single codfish Paul bought’

If the relativised phrase is specified for the plural, the resulting noun phrase may, in French as in Haitian, be associated with three different construals of the referent: (46)

a. French:  : a.’

Haitian:

b. French:  : b’. Haitian:

les morues que Paul a achetées ‘(all of) the codfish items which Paul bought’ mori Pòl achte yo les morues que Paul a achetées those of the codfish animals which Paul bought mori yo Pòl achte

c. French: ces morues que Paul a achetées - : these/those codfish animals which Paul bought c’. Haitian: mori yo Pòl achte a

 Anne Zribi-Hertz and Herby Glaude

Our assumption is that each of the three readings glossed in (46) is triggered by the same feature pattern in both languages: the upstairs plural has scope over the entire noun phrase and triggers an antipartitive reading; the downstairs plural has scope over the internal noun phrase only and triggers a partitive reading if the upstairs D is [−]; and a [+] feature in the upstairs D triggers a discourse-linking effect: (47)

a.

plural upstairs DP

D0

NumP Num0

CP Spec

– LOC +PL

C' C0

IP

moruez que

Paul a achetées

French: les morues que Paul a achetées: cf. (46a) Haitian: mori Pòl achte yo: cf. (29b), (32b)

b. plural downstairs DP D0

CP Spec

–LOC

C' C0

IP

moruesz que Paul a achetées

XP [+plural] tz

French: les morues que Paul a achetées: cf. (46b) Haitian: mori yo Pòl achte: cf. (29a), (32a)

c.

plural downstairs, [+locative] upstairs DP

D0

CP Spec

+ LOC

C' C0

IP

moruesz que Paul a achetées

XP [+plural] tz

French: les morues que Paul a achetées: cf. (46c) Haitian: mori yo Pòl achte a: cf. (29d), (32d)

XP[-number] tz

Chapter 10. Bare NPs and deficient DPs in Haitian and French 

Because only one determiner (variable LE or deictic-locative CE) may be spelt out in French within the maximal DP domain, these data contrast with those of Haitian, whose definite determiner LA may overtly occur both upstairs and downstairs. Once passed through the morphological component, Number is externalised upstairs in French whatever its syntactic source (downstairs, or upstairs), hence the homonymous spell-outs of (46a) and (46b), which are morphologically distinguished in Haitian.

.

Conclusions

Our starting point is the assumption that for any functional feature [α ±F], lack of morphological specification is syntactically ambiguous between feature deficiency [−F] and feature unmarkedness [−α]. Moreover, zero morphology may also result from the phonological deletion of a marked feature spell-out. This has crucial bearing on the analysis of ‘bare’ phrases. We have argued on empirical grounds that in Haitian, a phonologically null D head may instantiate either the zero spell-out of a [−locative] feature value, or the deleted spell-out of a [+locative] feature selected by yo. It follows that ‘bare noun phrases’, in Haitian, are all DPs. DP structure however varies as to number specification, since the lack of overt plural marking may instantiate the negative value of functional number — [−plural], i.e. ‘singular’ — or number deficiency, which triggers Mass semantic effects. French crucially differs from Haitian in the feature content of its definite determiner — a locative item in Haitian, a bound variable in French. As regards number, we have argued that the syntactic ambiguity of the nonplural correlates with the same semantic effects in both languages. Adapting Kayne’s (1994) analysis of restrictively-relativised noun phrases, we have shown that those of Haitian overtly stand as complex structures involving two distinct levels of functional specifications, with each pattern of features triggering its own construal of the referent. Correlatively, a relativised DP may be deficient for some functional feature at one level, but not at the other. We have argued that a similar analysis may be extended to French, although in this language, contrasting in this respect with Haitian, each maximal DP is limited to one overt D head. Basing ourselves on the French-Haitian comparative evidence presented above, we feel entitled to conclude that definiteness is neither a feature nor a syntactic category and that the conventional label D identifies neither a syntactic category nor a unique structural position. Furthermore, so-called bare NPs do not form a relevant or coherent class from a syntactic perspective, and their description should at least separate determiner deficiency from number deficiency, and syntactic deficiency from phonological deficiency.

References Aboh, E. O. 2002. La morphosyntaxe de la périphérie gauche nominale. Recherches Linguistiques de Vincennes 31: 9–26.

 Anne Zribi-Hertz and Herby Glaude Bernabé, J. 1983. Fondal-Natal: Grammaire basilectale approchée des Créoles Guadeloupéen et Martiniquais. Paris: L’Harmattan. Bernabé, J. 1987. Grammaire Créole – fondas kréyol-la. Paris: L’Harmattan. Bickerton, D. 1981. Roots of Language. Ann Arbor MI: Karoma. Borer, H. 2005. Structuring Sense, vol.1, In name only, Oxford: OUP. Borsley, R. 1997. Relative clauses and the theory of phrase structure. Linguistic Inquiry 28(4): 629–647. Bouchard, D. 2002. Adjectives, Number, and Universal Grammar. Oxford: Elsevier Science. Bouchard, D. 2003. Les SN sans déterminant en français et en anglais. In Essais sur la grammaire comparée du Français et de l’Anglais, P. Miller & A. Zribi-Hertz (eds), 55–95. Saint-Denis: Presses Universitaires de Vincenne. Chaudenson, R. 1992. Des iles, des hommes, des langues: Essai sur la créolisation linguistique et culturelle. Paris: L’Harmattan. Cardinaletti, A. & Starke, M. 1999. The typology of structural deficiency: A case study of the three classes of pronouns. In Clitics in the Languages of Europe, H. van Riemsdijk (ed.), 145–233. Berlin: Mouton-De Gruyter. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Chung, S. 2000. On reference to kinds in Indonesian. Natural Language Semantics 8(2): 157–171. Damoiseau, R. 1999. Eléments de grammaire comparée Français-Créole. Petit-Bourg, Guadeloupe: Ibis Rouge Editions. DeGraff, M. 1992. Creole Grammars and the Acquisition of Syntax: The case of Haitian. PhD Dissertation, University of Pennsylvania. DeGraff, M. 1999a. Language Creation and Language Change: Creolization, diachrony and development. Cambridge MA: The MIT Press. DeGraff, M. 1999b. Creolization, language change, and language acquisition. In DeGraff (1999a), 1–46. DeGraff, M. 2002. Morphology in creole genesis. In Ken Hale: A life in language, M. Kenstowicz (ed.), 53–121. Cambridge MA: The MIT Press. Delfitto, D. and Schroten, J. 1991. Bare plurals and the number affix in DP. Probus 3(2): 155–185. Déprez, V. 2000. Parallel (a)symmetries and the internal structure of negative expressions. Natural Language & Linguistic Theory 18(2): 253–342. Doetjes, J. 1997. Quantifiers and selection. The Hague: HAG. Gadelii, K. E. 1997. Lesser Antillean French Creole and Unversal Grammar. PhD Dissertation, Göteborg University. Germain, R. 1995. Grammaire Créole. Paris: L’Harmattan. Goodman, M. F. 1964. A Comparative Study of Creole French Dialects. The Hague: Mouton. Guéron, J. 1985. Inalienable possession, PRO-inclusion and lexical chains. In Grammatical Representation, J. Guéron, H. Obenauer & J.-Y. Pollock (eds), 43–86. Dordrecht: Foris. Guéron, J. 1992. La possession inaliénable et l’aspect locatif. In De la Musique à la Linguistique. Hommages à Nicolas Ruwet, L. Tasmowski & A. Zribi-Hertz (eds), 255–272. Ghent: Communication & Cognition. Guéron, J. 2003. Inalienable Possession and the interpretation of determiners. In From NP to DP, Vol. 2: On the expression of possession, M. Coene & Y. D’Hulst (eds), 189–220. Amsterdam: John Benjamins.

Chapter 10. Bare NPs and deficient DPs in Haitian and French  Hazaël-Massieux, G. 1999. Marking specificity in Antillean Creole French. In St Kitts and the Atlantic Creoles [Westminster Creolistic Series 4], P. Baker & A. Bruyn (eds), 271–288. London: University of Westminster Press. Janson, T. 1984. Articles and plural formation in creoles: Change and universals. Lingua 64: 291–323. Joseph, F. 1988. La Détermination Nominale en Créole Haïtien. PhD Dissertation, Université Paris-7. Jun, J.-Y.1999. Recherches sur le nombre en Coréen: Syntaxe, sémantique et morphologie. PhD Dissertation, Université Paris-8. Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge MA: The MIT Press. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The case of Haitian Creole. Cambridge: CUP. Kwon, S.-N. & Zribi-Hertz, A. 2004. Number from a syntactic perspective: Why does plural marking look ‘truer’ in French than in Korean? In Empirical Issues in Formal Syntax and Semantics 5: Selected papers from CSSP 2003, O. Bonami & P. Cabredo Hofherr (eds), http:// www.cssp.cnrs.fr. Longobardi, G. 1999. The structure of DPs: Some principles, parameters and problems. In Handbook of Syntactic Theory, M. Baltin & C. Collins (eds). Cambridge: Blackwell. Longobardi, G. 2001. Formal syntax, diachronic minimalism, and etymology: The history of French chez. Linguistic Inquiry 32(2): 275–302. Lyons, C. 2000. Definiteness. Cambridge: CUP. Meillet, A. & Vendryès, J. 1979. Traité de Grammaire Comparée des Langues Classiques. 5th printing revised by J. Vendryès. Paris: Champion. Milner, J.-C. 1989. Introduction à une science du langage. Paris: Seuil. Muller, C. 1996. La subordination en Français:Le schème corrélatif. Paris: Armand Colin. Müller, A. 2001. Genericity and the denotation of common nouns in Brazilian Portuguese. In The Proceedings of the SULA: The semantics of under-represented languages in the Americas, UMOC 25, A. Weerle & J.-Y. Im (eds), 72–80. Amherst MA: The University of Massachusetts. Pelletier, F. 1979. Non-singular reference: Some preliminaries. In Mass Terms: Some philosophical problems, F. Pelletier (ed.), 1–14. Dordrecht: Reidel. Pompilus, P. 1976. Contribution à l’etude comparée du Créole et du Français – à partir du Créole Haïtien. Morphologie et syntaxe. Port-au-Prince: Editions Caraïbes. Roodenburg, J. 2001. Les ‘noms nus’ en français contemporain: Quelques problèmes. In 6èmes rencontres de l’atelier des doctorants de linguistique, E. Aïm et al. (eds), 125–130. Paris: Université Paris-7. Roodenburg, J. 2003. The interpretations of coordinated bare nouns in French. In Proceedings of ConSole XI, M. van Koppen, J. Sio & M. de Vos (eds), 1–15. Leiden: University of Leiden. Smith, C. 1964. Determiners and relative clauses in a generative grammar of English. Language 40(1): 37–52. Schmitt, C & Munn, A. 2000. Bare nominals, morphosyntax, and the Nominal Mapping Parameter. Ms, Michigan State University. Song, S.-C. 1975. Rare plural marking and ubiquitous plural marker in Korean. In Papers from the Eleventh Regional Meeting of the Chicago Linguistic Society, R. Grossman, L. J. San & T. Vance (eds), 536–46. Chicago IL: Chicago Linguistic Society. Sylvain, S. 1936. Le Créole Haïtien, morphologie et syntaxe. Belgium: Wetteren.

 Anne Zribi-Hertz and Herby Glaude Valdman, A. 1978. Le Créole: Structure, statut et origine. Paris: Klincksieck. Vendler, Z. 1967. Linguistic and Philosophy. Ithaca NY: Cornell University Press. Vergnaud, J.-R. 1985. Dépendances et niveaux de représentation en syntaxe. Amsterdam: John Benjamins. Wagner, R.-L. and Pinchon, J. 1962. Grammaire du Français classique et moderne. Paris: Hachette. Ware, R. 1979. Some bits and pieces. In Mass Terms: Some philosophical problems, F. Pelletier (ed.), 15–29. Dordrecht: Reidel. Zribi-Hertz, A. and Mbolatianavalona, L. 1999. Towards a modular theory of linguistic deficiency: Evidence from Malagasy personal pronouns. Natural Language & Linguistic Theory 17–1: 161–218. Zribi-Hertz, A. 2002. The DP hypothesis and the syntax of identification. Recherches Linguistiques de Vincennes 31: 127–142. Zribi-Hertz, A. To appear. La syntaxe de l’article défini: de la structure à l’interprétation. In Modèles syntaxiques, D. van Raemdonck (ed.), Brussels: Peter Lang.

 

A comparison of romance Creoles

 

Implicit determination and plural1 Viviane Déprez This paper explores and compares the distributions and interpretations of nominal expressions without determiner (bare nouns) in a number of French based creole and in one Portuguese based creole, Cape Verdean Creole. The goals of this exploration are on the one hand to determine whether the range of meanings that bare nouns can have in creole languages parallels the one observed in non-creole languages and on the other hand to examine the role that various morpho-syntactic factors and in particular the morphological expression of number can play in restricting the available interpretations of bare nouns. This paper also evaluates the success of an alternative approach to Chierchia’s Nominal parameter, termed here the Plural parameter in accounting for the observed data. One prediction of this approach that takes nouns to denote Kind terms in all languages is that no language with optional plural should fail to allow for a Kind reading of its bare nouns. This and other predictions are shown to be verified in the sample of languages explored here.

.

Introduction

Simple nominal expressions without determiners such as liv (book) in (1), often termed ‘bare nouns’ (BN) in the semantic literature, are very common in creole languages. (1)

Jan li liv (Haitian Creole) John reads books

Yet, so far, these nominal expressions have received comparatively little attention. In creoles studies, interests have more commonly centered on the emergence of explicit determiners as opposed to their absence so that (with a few exceptions (cf. Déprez . For fruitful comments on various versions of this paper, I’d like to thank R.Kayne, K. Wilkison, V. Dayal and J. Guéron. I would also like to thank D. Adone, M. Bouloux, F. Henri and P. Baker for discussion of the Mauritian Creole data, E.Védrine, Wisly Paul, Francois Canal and Wilson Douce for discussion of the Haitian Creole data and M. Baptista for discussion of the Cape Verdean data. All remaining errors are my own. As this paper was written in 2003, it does not take into account the literature that post-date it.

 Viviane Déprez

1999, 2000, 2001, 2003a, 2004, 2006; Baptista 1999, this volume) the various factors that determine or restrict the distribution and interpretation of BN in creole languages have rarely been investigated. In studies of non-creole languages, in contrast, detailed aspects of the meanings of bare nouns have been extensively studied, particularly in English and in the Romance languages (Carlson 1977, Krifka 1995, Chierchia 1998 among others). The range of meanings that argument BN can have is now rather well known and various formal semantic models that predict their possible and impossible occurrences have been proposed in the literature. The empirical goal of this paper is to compare the properties of BN in some creole languages with that of BN in non-creole languages with two central questions in mind: 1. 2.

Can BN have the same range of meanings in creole as well as non-creole languages? Are choices among possible meanings determined by the same factors?

Methods and paradigms of semantic analysis commonly used for the study of noncreole BN are here applied in the comparative study of French based Creole data from Seychellois (or Seychelles Creole), Mauritian and Haitian Creole and from one Portuguese based Creole, Cape Verdean Creole, which presents a particularly interesting case for our purpose as it retains a comparatively richer nominal morphology. The speech data for this study comes from corpus studies, from the literature or directly from my own field studies with native speaker informants as specified in the relevant sections of the text. The focus of the present comparative study is limited to simple count BN commonly occurring in the argument positions (subject and complements) of regular verbal predicates as in (1). Concretely speaking, this means that, BN in compounds or idiomatic constructions (e.g. to take heed), BN used as proper names (e.g. Father was unhappy), BN in prepositional, predicative and locative constructions (He went home) or in complex nominal constructions such as possessives, relative clauses (i.e. cataphorically specified NPs) and conjoined constructions (e.g. brothers and sisters) are not discussed here. The goal in restricting the focus of our comparative study is to concentrate on the betterstudied cases of BN. While the properties of BN in argument positions are beginning to be well understood, BN in complex constructions manifest properties that are quite distinct from those of regular argument positions and have yet to be investigated. To take a concrete example, in a language like French where BN are usually not allowed as verbal arguments (cf: *Je lis livres “I read books”), idiomatic constructions notwithstanding, BN are quite common as complements of PPs (en danger “in danger”, sans argent “without money”, avec précision “with precision”, un tissu semés d’étoiles “a fabric studded with stars”) and can be used in conjoined constructions (Parents et professeurs sont invités … “Parents and professors are invited…”) in possessive and in pseudo-partitive constructions (des jouets d’enfants “children’s toys”, une poignée de bonbons “a handful of candies”). In such complex constructions BN interpretations that would otherwise not be allowed in regular verbal argument positions can

Chapter 11. Implicit determination and plural 

sometimes be licensed. In English for instance argument count BN never allow for a definite interpretation (I read books ≠ I read the books), but this interpretation is possible with conjoined BN and BN used as proper names (Heycock & Zamparelli 2003). Prior to tackling the properties of BN in this variety of complex constructions, it seemed to us wise for more fruitful comparative results to limit our focus to the cases that are best understood, i.e. the cases of count BN in argument positions. The theoretical goal of this paper is to investigate the role of number and its morpho/syntactic realizations in influencing the range of meanings BN can have in distinct languages. Déprez (1999b, 2000, 2003, 2004, 2006) argues that variability in the morpho-syntactic realization of number is an important factor in determining crosslinguistic variability. This approach is offered as an alternative to Chierchia’s (1998) Nominal Parameter model whose predictions were shown to empirically fail for Haitian Creole and other French based creoles in Déprez (1999, 2001). In English and the Romance languages, number is usually obligatorily expressed in noun phrases and commonly encoded by bound morphemes. In creole languages in contrast, number is often optionally expressed and frequently (yet not always) encoded with a free morpheme. Within the theoretical perspective explored here. I argue that this difference plays a role in restricting the readings that argument bare nouns can have cross-linguistically. This paper provides further testing for this view with a larger set of empirical data. The paper is organized as follows. After reviewing as an empirical background the salient properties that non-creole BN manifest, Section 2 describes the creole data on BN, paying special attention to the morphological expression of number given the hypothesized theoretical importance of this factor. Due to space limitation, the language descriptions are here limited to cases of common count BN occurring in regular verbal argument positions. (See Déprez 2004 for a more extensive discussion). Section 3 summarizes the theoretical approach introduced in Déprez (1999 and following) and outlines its relevance for the present study. Section 4 tests the predictions of this approach against the data offered in Section 2.

. .

Cross-linguistic variations in the meanings of bare nouns Empirical background

A brief review of the properties of argument BN in English and Italian will serve to establish the empirical background of this study. The first part of this section focuses on the possible meanings BN can have in these languages. Creole data are considered in the second part.

..

English bare nouns

The semantic literature on English BN, starting with Carlson (1977), distinguishes three readings for argument BN. These are respectively illustrated here in (2a), (2b) and (2c).

 Viviane Déprez

(2a) presents the so-called existential reading, in which the meaning of a BN resembles that of a typical indefinite noun phrase “some N”, with the proviso that BN always have obligatory narrow scope:2 (2b) presents the generic reading, closest to universally quantified noun phrases like “all N”, but with exceptions allowed. (2c) presents the kind reading that refers as a single whole to the class of entities that forms a kind rather than to its member. This last reading can be paraphrased by expressions such as “the N species”. (2)

a. b. c.

I heard dogs barking in the park yesterday Dogs bark Penguins are disappearing in Antarctica

Existential Generic Kind

As often discussed in the semantic literature, the availability of these different readings is in part conditioned by the temporal/aspectual properties of the sentences BN occur in and in part by the semantic nature of the nominal expressions themselves. Existential readings of BN like (2a) occur in episodic sentences that express statements about particular events and are often temporally or spatially bounded as, for instance, in the reported past event of (2a). In contrast, generic readings of BN as in (2b) occur in characterizing sentences that express a generalization over events, often describing typical or habitual properties of the individuals in the BN. Kind readings, on the other hand, in some sense a subspecies of generic readings, are more typically a property of the type of nominal expressions involved (BN vs. indefinite NPs) and are thus not dependent on the aspectual/temporal properties of the sentences they occur in. Such readings are enforced with a restricted set of predicates – such as be rare/common, be disappearing or invent – that have the particularity of addressing the kind/species as a whole. These predicates can then serve as a test of the capacity for particular types of nominal expressions to support such readings. In English for instance, singular definite noun phrases and BN are nominal expressions that support reference to a kind,

. English BN have obligatory narrow scope with respect to other quantificational expressions or operators. Thus for instance, (i) can only be interpreted informally as in (iii a) but not as in (iii b). Indefinite expressions with overt determiners such as some, a, in contrast, interact in scope with semantic operators, allowing both high and low scope. Thus for instance (ii) can ambiguously be paraphrased as in (iii) a. or (iii) b. (i) (ii) (iii)

John did not see spots on the floor. John did not see a spot on the floor. a. John saw no spots on the floor. b. There was a spot on the floor that John did not see.

That the narrow scope property of BN is maintained in at least some creole languages is shown for HC in Déprez (1999).

Chapter 11. Implicit determination and plural 

but singular indefinite noun phrases are not, as their characteristic infelicity with kind predicates reveals. Thus contrasting with (3a) and (3b), (3c) does not allow reference to the dinosaur kind and only has a reading where a single individual dinosaur is magically disappearing in a Cheshire Cat-like fashion. See Carlson (1977), Chierchia (1998), and in particular (Carlson and Pelletier (1995)) inter alia for an in-depth description of these different readings and their occurrences. (3)

a. The dinosaur is disappearing. b. Dinosaurs are disappearing. c. #A dinosaur is disappearing.3

English BN further manifest an interesting number restriction. To be acceptable arguments, English count BN must occur in a morphological plural form independently of the reading they have. As shown in (4), morphologically singular or more accurately unmarked count BN are quite generally excluded. Unmarked BN must denote a mass as in (4’). They cannot denote singular count entities: (4)

a. b. c.

(4’)

a. b. c.

*I have heard dog barking in the park. *Dog barks. *Dog was common. I have seen water in the basement. Water is good for your health. Water is slowly disappearing from the planet Earth.

Table 1 summarizes the possible readings of English simple argument count BN Kind Generic Existential

..

yes yes yes

Romance bare nouns

In the main Romance languages such as Italian, Spanish and Portuguese (but not French4), the properties of BN present a lot of similarities with those of English but also some interesting differences. Concerning first the number restriction, it is clear that Romance count BN, like the English ones, must bear plural morphology to be acceptable.

. The indefinite nominal expression in this sentence allows for a sub-kind reading whose meaning is akin to the following paraphrase: = One kind of dinosaur (i.e. the brontosaurus kind) is disappearing. This reading is not directly relevant for our present purpose and will thus not be discussed here (see Carlson & Pelletier 1995). . In contrast to the other major Romance languages, French quite generally disallows BN in argument positions. We return to this important point below.

 Viviane Déprez

Singular or unmarked BN generally have a mass denotation, not a singular count denotation. This number restriction, illustrated here with the Italian example in (5), also obtains in Spanish and Portuguese in the most common cases.5 (5)

a.

Ho incontrato *cane/cani I met *dog / dogs

Italian

There is some disagreement in the literature concerning the possible readings of Romance BN (Longobardi 1999). However, what seems to be largely agreed upon is that Romance BN can have an existential and a generic reading comparable to the English ones above and subject to essentially the same temporal/aspectual conditions. That is, Italian BN have an existential reading in episodic sentences that describe a particular event as in (6a) and they can have a generic reading in characterizing sentences. In the latter case, however, the presence of an overt operator such as, for instance, the adverb sempre that overtly signals the generalization over events is highly preferred: (6)

a. b.

Leo questa estate ha corregiato belle ragazze. Leo this summer has courted nice girls. Leo corregia sempre belle ragazze. Leo always courts nice girls.

Existential Generic

Longobardi (1994, 1999), however, argues that Italian BN never have a true kind reading. As shown in (6), Italian BN are indeed infelicitous with characteristic kind predicates. To obtain such kind readings in Italian, the use of an overt definite determiner is required as in (7a’) and (7b’) or else the sentence is unacceptable: (7)

a. a.’ b. b.’

*Leo odia gatti. Leo odia i gatti. Leo hates cats. *Elefanti di colore bianco sono estinti. Gli elefanti di colore bianco sono estint. White colored elephants have become extinct.

Similar observations have been made for Spanish and Portuguese. In this respect, Romance argument BN have a more restricted range of meanings than the English ones. In contrast to English indeed, Romance BN seem to be a type of nominal expression that disallows kind reference, just like indefinite nominals. Romance BN are also

. As noted in the literature (Longobardi 1994), there are exceptions to the plural constraint. In the predicative positions of copula sentences for instance, BN may be singular in Romance for certain classes of nouns like nouns of professions. Predicative BN in Romance and French based creoles are discussed in Déprez (to appear). With Spanish BN, moreover, a mass reading seems also more easily coerced than with English BN. Thus in an imperative sentence like ‘Come manzana’ (Eat apple), a mass reading is rather natural in Spanish but apparently much harder to get in English.

Chapter 11. Implicit determination and plural 

more restricted in their syntactic distribution. They commonly occur in the object positions of most predicates but are generally infelicitous in preverbal subject positions under any reading.6 (8)

a. b.

*cani ama giocare/stato giocandi fuori/sono rari I cani ama giocare/stato giocandi fuori/sono rari Dogs like to play/ are playing outside/ are rare

This characteristic subject/complement asymmetry, much discussed in the literature, is not addressed in this paper (see Longobardi (1994), Déprez (2000) inter alia for discussion).7 As this asymmetry is present in the Romance languages for all possible readings of BN, its source is arguably more syntactic than semantic in nature. In this respect, it remains outside our present concerns. Table 2 summarizes the possible readings of Romance argument count BN: Table 2. Readings of Italian simple argument count BN Kind Generic Existential

No Yes Yes

. As discussed in Longobardi (1994), the felicity of BN in pre-verbal subject positions clearly improves when they are modified: Elefanti*(di colore bianco) hanno creato in passato grande curiosita. White colored elephants raised in the past a lot of curiosity. Zamparelli (1995) argues that modified BN have a different syntactic structure. The presence of a modifier may be sufficient to appropriately license a null determiner. . While Longobardi’s approach capitalizes on Proper Government (ECP) as a condition on the occurrence of nominal expressions with null determiners, Déprez (2000) argues that the external distribution of noun phrases is determined by their internal properties. Null determiners can be licensed internally to nominal projections and yet lack the adequate feature to allow EPP checking (i.e. a categorial D feature in Chomsky 2000). If so, expressions that contain a null D in EPP checking subject positions fail to lead to convergence. This accounts for the observed subject/complement asymmetry. Further, noting that distributional asymmetries with BN are observed primarily in languages with rich verbal agreement – subject BN are quite generally allowed in languages with poor (English) or no agreement (Chinese), albeit sometimes, with interpretational restrictions – Déprez (2000) suggests that such EPP failures may be restricted to rich agreement languages – i.e. to languages in which agreement has a pronominal or D-like character. In addition to accounting for the standard asymmetry, Déprez’s proposal adequately accounts for a number of cases (preverbal focused, coordinated BN) that are problematic for ECP-type analyses.

 Viviane Déprez

.

Bare nouns in creole languages

..

Seychelles Creole

Turning now to creole languages, I begin with Seychelles Creole (SC), a French based creole of the Indian Ocean. A careful study of the corpus of radio recordings collected in Bollée (1977) reveals that BN in this language can have existential (9a), generic (9b,c) as well as kind readings (9d,e), as seen before for English: (9)

a. b. c.

d. e.

I ti ana labitid mãz sotrel ek dimjel sovaz. (Marc 1,6) They were used to eating grasshoppers with wild honey. zot tu zot in bjen dakor pou zot pret zot zuti (VI:46) they all were in agreement to lend them tools Depi sa zur, sat pa ule war lera dã zot lizje. (VI:213) From this day, cats do not want to see rats in their neighborhood. E ziska ozordi torti i ãkor kõtã ziromõ8 (II:82) And up to today turtles like pumpkins ki napa en pli dernje zanimo ki lera (VI:210) there isn’t a more lowly animal than the rat

In addition to these three readings already encountered above, SC BN further allow for a fourth distinct reading, illustrated here in (10), which most resembles that of a definite description. Note that depending on the context, such definite SC BN can be construed as plural or as singular without a change of form: (10)

a.

lapli pe do ẽ das sa ban sat grã zorej …Ler sat i a rekolte, i va rekolte lakok. (VI:186–187) the rain can give a dance to these stupid cats… When the cats will harvest, they will harvest the skin.

. Verbs such as ‘love, like’ are thought to require kind referring expressions in their complement positions by some but not all semanticists. For native speakers of Italian as well as other Romance languages, BN are unacceptable in this position and a definite determiner is required. This strongly suggests that at least in these languages, such verbs do indeed require a kind referring expression as their complement. As my survey of Seychelles Creole is based on a corpus and not on native speakers’ judgments, I have not found other examples typically used in semantic tests of kind reference. My suggestion that SC allows kind readings is thus somewhat tentative, requiring further testing for firmer confirmation. As such data gaps indicate, judgment data are sometimes crucial to complement corpus data. Yet judgment data are often discredited in creole studies as they are often felt to be less objective. It is then worth pointing out that a recent scientific inquiry about judgment data clearly suggests that such feelings are misplaced. In a very careful socio-linguistics study of speaker judgments, Adli (2002) demonstrates that contrary to received opinions, speaker judgments are quite remarkably homogenous, and highly reliable, particularly on infelicitous sentences. His results thus strongly argue for the rehabilitation of judgment data in socio-linguistics and syntactic variation studies.

Chapter 11. Implicit determination and plural 

b. c.

d.

…i..lav sa de lasjet….La lasjet in lave. (I, 122–3) He washes these two plates….Now, the plates are washed. Lerwa i met ẽ solda veje. Ler Sugula i vini pu li vin bejne, i war solda pe veje. (II:5) The king put a soldier in surveillance. When Sugula came to bathe, he saw the soldier keeping watch. Madam Lamur i montre avoka let. Avoka i lir let. (XI:187) Madame Lamour showed the letter to the lawyer. The lawyer read the letter.

Characteristically, examples like (10a–c,) illustrate a definite anaphoric use of BN in episodic sentences. Here these BN refer to unique familiar identified individuals (or sets of individuals) previously introduced in the discourse. They exhibit familiarity and uniqueness, two features that, according to much semantic work, are classic hallmarks of definiteness. Notably in equivalent contexts in both English and Romance, overt definite determiners must be used and BN are disallowed. In this respect, it is clear that in Seychelles creole BN allow for a more extensive range of meanings than the previously discussed languages. Definite readings of BN, however, do not constitute a unique creole feature; they have been observed in other well-known non-creole languages such as Hindi (Dayal 2001), Russian, Japanese and Chinese (Cheng and Sybesma 1999). With respect to number marking, SC nouns and SC BN in particular are always morphologically invariable. This absence of nominal morphology is in line with what is observed in many other creoles. Plural marking in Seychelles creole is optional. That is, there are nominal expressions in SC that are interpreted as plural but remain unmarked. See for instance the definite BN in (10) above. When overtly expressed, plural is signaled by the presence of an independent morpheme, the morpheme ban, which always occurs in a fixed pre-nominal position as in (11):9 (11)

Zot ti kit zot papa Zebede ek ban travajer dã kanot pu suiv Zezi. They left their father Zebede and the workers in the canoe to follow Jesus.

According to Bollée, ban is most commonly used with definite expressions although as seen in (10a–b) above, plural marking is not a necessity even in such cases (Bollé 1977: 37). Given this feature, it should come as no surprise that SC BN exhibit no number restriction comparable to the one observed in English and Romance. SC BN are always morphologically unmarked. Moreover, they are quite generally underspecified for number interpretation, allowing without a change of form for either a semantically singular or plural interpretation, depending on the context. This semantic under-specification for number is observable with all the different readings of BN. In sum, SC BN seem to exhibit the broadest possible range of readings, existential,

. See Déprez (2003b, to appear) for a detailed discussion of the structure DPs in a variety of French based Creoles.

 Viviane Déprez

generic, kind as well as definite, singular or plural and they can occur in all argument positions. To my knowledge, no distributional or interpretative asymmetry has ever been observed in SC: BN seem equally available in subject or complement positions with the same range of readings. Table 3 summarizes the possible readings of SC argument count BN. Table 3. Kind Generic Existential Definite

Yes Yes Yes Yes

.. Mauritian Creole10 Let us turn next to Mauritian Creole (MauC), another French based Creole of the Indian Ocean that is rather similar to Seychelles Creole and often grouped with it under the common denomination of Isle De France Creole. As illustrated in (12), (13) and (14) respectively, Mauritian Creole BN can have existential, generic and kind readings: (12)

a. b.

Zan aste liv. John buys book(s). Mo kamard frances manz kurpa. My French friend eats snails.

Existential

(13)

Lisyen zape, sat miow ek zwaso sante. Dogs bark, cats meow and birds sing.

Generic

(14)

a.

Kind

b.

Lelefan rar dan moris. Elefants are rare in the Mauritius islands. Edison ti invent glob elektrik. Edison invented the electric bulb.

In contrast to SC, however, Mauritian Creole simple common count argument BN do not seem to allow for a regular definite (anaphoric) reading.11 Consider (15a) below,

. I wish to thank Dany Adone for kindly answering my data enquiries. These data have also been checked with her local informants and cross-checked with Philip Baker whom I warmly thank too. They were further checked with two additional native speakers of Mauritian Creole, M. Bouloux and F. Henri. . Here again, it must be kept in mind that our comparative study is limited to syntactically simple common count BN in argument positions. Definite readings of Mauritian (or Morisyen) bare nouns may well occur in a variety of complex nominal constructions (such as locative, nominal possessive, relative clause and conjoined NP constructions) that have special properties, as well as in cases where they are interpreted as proper nouns. My claim here is not that definite readings of BN never occur in Morisyen but that they do not occur for simple bare count nouns in regular argument positions.

Chapter 11. Implicit determination and plural 

which essentially replicates the comparable examples (10) in SC. In their first occurrence as presentational indefinites, the words for man and woman are preceded by the indefinite determiner en. In their second occurrence, as anaphoric definites, both nouns must in our informants’ judgments obligatorily co-occur with the post-nominal definite determiner la. BN would not be acceptable in such contexts. (15)

a.

b.

letan mo rant dan lasam, ti ena en zom ek en fem. Mo apel zom la me fem la ki fin repon. When I entered a room there was a man and a woman in the room. I called the man but it was the woman who answered. Lor latab ti ena enpom ek enpwar. Mari inn pran pom-la e li’nn/li finn manz li On the table there was an apple and a pear. Mary took the apple and ate it.

Such examples directly contrast with the SC examples given in (10) above, which allowed a definite reading for BN in comparable anaphoric contexts. In SC, BN could clearly refer to previously identified people or objects, being, in this respect, semantically definite. But the use of BN in place of the determined nouns in (15) above would be awkward and produce a distinct change in the intended meaning, as it would prevent reference to the previously mentioned individuals. Seychellois and Mauritian Creole thus appear to form a minimal pair with respect to the possibility of allowing anaphoric definite readings for count BN in argument positions,12 while they are in other respects quite similar to one another. As in SC indeed, the distribution of Mauritian BN is uniform, with no distributional asymmetry. Nouns in general and BN in particular are morphologically invariable, lacking both number and gender inflection. BN are also semantically under-specified for number, allowing for both a singular and a plural interpretation in adequate contexts. Finally, as in SC, the expression of plural in Mauritian creole also uses the free pre-nominal morpheme ban: (16)

ban politisyen ban liv la Sa ban lakaz dibwa

politicians the books These wooden houses

Table 4 summarizes the possible meanings of MC simple argument count BN. Table 4. Kind Generic Existential Definite

Yes Yes Yes No

. Philip Baker (pc) has pointed out to me that BN with a definite reading were clearly allowed in earlier forms of Mauritian Creole (cf. Baissac’s stories). He suggested that the possibility of definite BN in Mauritian Creole appears to coincide with the period during which the distinction between sa…la as a demonstrative and la as a definite was not yet established.

 Viviane Déprez

..

Bare nouns in Haitian Creole

Consider next the case of Haitian Creole,13 another French based Creole but from a distinct geographical area, the Caribbean.14 That Haitian Creole BN can have an existential, a generic and a kind reading is shown in (17) below: (17)

a.

b.

Moun koumanse ap pran baton. PROG take sticks People start People started taking sticks Lè chat pa la, rat pran kay la. When cat Neg here rat take house the. When the cat is not there, rats take over the house.

Existential

Generic

If this is correct, it provides interesting support for generalization 2’ in section 4 below, as it appears to confirm on a diachronic level the proposed link between definite BN and the lack of a distinct overt definite marker. . The data on HC has been collected over the years from the literature, from texts on the Internet and from informant sessions with several native speakers. I would like to thank, Wisly Paul, Francois Canal and Wilson Douce who have served as regular informants. I would also like to thank E. Vedrine for kindly answering some of my online questions. Michel DeGraff has raised challenging questions about the generalization expressed here that Haitian Creole does not allow for a regular definite reading of its simple argument count BN. He notes that a definite reading is possible in story contexts, for instance for an animal noun that refers to one of the protagonists of a story (cf. Komper Lapen or Lapen alone). Examples of this kind seem to me to instantiate the use of BN as a proper name or title, as the common capitalization in written texts suggests. Similar examples also exist in French, e.g. the Fable de La Fontaine, (Maitre Corbeau sur un arbre perché.. Goupil par l’odeur alléché), a language known to strongly resist the use of common argument count BN under any readings. Such examples are thus orthogonal to the generalization proposed here, as they present a rather special use of definite bare nouns, i.e. one akin to a proper name, that are outside the scope of the present paper. Other types of examples illustrating definite readings of syntactically complex nominals without an overt definite marker include possessive structures (liv Pol Paul’s book) and cases of noun phrases modified by a relative clause (liv Jan li the book John read). In contrast to syntactically simple bare nouns, these nominal expressions are cataphorically specified by their respective modifiers and in this respect, it is plausible to assume that their definite reading is licensed within the nominal constituent itself. A precise analysis of this type of licensing is beyond the scope of the present paper and is undertaken in further research. According to DeGraff, however, there are other types of examples of the use of definite BN in Haitian Creole. An evaluation of such cases must await the relevant examples, i.e. examples of simple common count BN in argument positions with a definite reading. Should the generalization about Haitian be incorrect, then to my knowledge Martinique Creole instantiates the described properties. I thank Timotée Montoute for his help in providing and collecting the relevant Martinique Creole data. . For a detailed study of the properties of Haitian Creole BN see Déprez (1999b, 2000, 2001, 2004).

Chapter 11. Implicit determination and plural 

c.

d.

Elefan ap vin ra. Elephant PROG come rare Elephants are/the elephant is becoming rare. Edison (te) envante anpoul elektrik. Edison (Past) invent bulb electric Edison invented the light bulb.

Kind

Haitian Creole, however, just like Mauritan Creole, seem to disallow an anaphoric definite reading for syntactically simple count argument BN. The native speakers I consulted quite strongly preferred the use of an overt marker of definiteness such as la, the singular definite marker, or yo, the plural definite marker, to obtain definite readings of nominal expressions in characteristic anaphoric contexts like those of (18): (18)

Selòm fèk achte yon panno, yon ba ak yon makout, men yo gen tan vòlè makout *(la). Selòm just bought a horse cushion, a harness and a makout, but someone has already stolen the makout.

A careful study of texts on the Internet – such as a variety of news articles of Haiti Progrès – reveals that the second occurrence of a noun newly introduced in the discourse with an indefinite determiner is generally followed either by a demonstrative or by a definite determiner.15 A corpus example is provided below. (19)

….yon peyizan ki rele Apredye Tilsen … a peasant who named Apredye Tilsen …a peasant named Apredye Tilsen … …peyizan sa a ki rele Apredye Tilsen alyas Tanpi peasant Dem Def who named Apredye Tilsen alias Tanpi …this peasant named Apredye Tilsen alias Tanpi te fe yon entevansyon sou ed ong…. Past make an intervention on aid ong made an intervention on ong aid… Entevansyon sa a pat pote vis delege Mak Lamou byen… Intervention this the not-past bring vice-delegate Mak Lamou good This intervention did not do vice delegate Mak Lamou any good… Apre entevansyon an vis delege a te poze Tanpi yon kesyon After intervention Def vice-delegate Def past ask Tanpi a question After the intervention, the vice delegate asked Tanpi a question… (Extract of Haiti Progres Volume 15, n02 1997)

. HC nouns in possessive and other complex constructions may well have a definite interpretation without featuring a definite article. However, such complex NP can be considered so to speak ‘constructionally’ definite in similarity with English possessive constructions like John’s book that are clearly definite without the use of an overt definite determiner. That the existence of such constructions does not imply the availability of definite simple count argument BN in a given language is demonstrated by the fact that English disallows such BN.

 Viviane Déprez

In HC, as in SC and MauC, nouns are not morphologically marked for number or gender. Plurality, when overtly expressed, is expressed with the post-nominal morpheme yo and restricted to definite noun phrases. BN allow both for a plural and a singular interpretation, number interpretation being in general determined either by world situational factors as in (20a) or constrained by sentence internal factors as in (20b–c) (dependent singular/plural). In (20a), liv (books) is construed as a plural because buying a multiplicity of books is a rather common event and a speaker would have indicated singularity (yon liv) if that was what was meant. The reverse is true for kay (house), which is most readily interpreted as a singular, as in this case it is the purchase of multiple houses that is sufficiently unexpected to warrant overt marking. In (20b) and (20c) on the other hand, the singular or plural interpretation of kalbas is directly dependent on the singular or plural interpretation of the subject as it is commonly understood that water-carrying only requires one calabash per child. (20)

a. b. c.

Jan achete liv/kay pou Pòl. John bought books /a house for Paul. Ti moun nan pa pote kalbas pou l pran dlo. The child did not bring a calabash to take water. Ti moun yo pa pote kalbas pou yo pran dlo. The children did not bring calabashes to take water

In sum, we have seen that as the data indicate, HC simple argument count BN do not manifest a plural restriction and they allow for existential, generic and kind readings, but not for (anaphoric) definite ones. Table 5 below summarizes the possible readings of HC argument count BN. Table 5. Kind Generic Existential Definite

Yes Yes Yes No

.. Cape Verdean Creole16 Let us now turn to Cape Verdean Creole (CVC), a creole that is somewhat unusual in that, like its superstrate language Portuguese, it can express plural on its nouns with overt bound morphology. Contrary to Portuguese, however, as (21) shows, count BN in Cape Verdean Creole are commonly acceptable as verbal arguments even when they do not bear an overt morphological marking for plural. (21)

Dja’ N faze balei pa’ PERF+I make basket for ‘I made baskets for sale.’

N bende. I sell

(W-ST)

. The data on Cape Verdean Creole is taken from Baptista (p.c, 1999, 2000) and this volume.

Chapter 11. Implicit determination and plural 

Moreover, morphologically unmarked BN do not have to denote a mass to be acceptable. Regular simple argument BN can have a plural count noun interpretation without overt plural marking. In this respect, Cape Verdean Creole BN seem to contrast with English and Romance ones in manifesting no number restriction. However, it must be recognized that plural morphology on nouns is quite restricted in certain varieties of CVC. In basilectal CVC, morphological plural is robustly available only with some classes of nouns, mostly the animate nouns (Baptista 2000) and not with other noun types.17 As a consequence of such restrictions, the distribution of marked and unmarked BN manifests a complexity that requires closer consideration. Let us then consider different cases in turn. Consider first marked BN. As shown in (22), morphologically marked plural CVC BN can have an existential, a generic, as well as a kind reading. (22)

a. b. c.

N odja omis sintadu riba di kaza. I saw men sitting on top of the house. Omis e temozu. Men are stubborn. Pur kausa di kaida demografika, kabuverdianus Because of fall demographic, Cape Verdeans Because of the demographic fall, Cape Verdeans sa ta dizaparese. are disappearing.

Existential Generic

Kind

In addition to these three readings, plural marked CVC BN (Baptista 1997, 2000) can have a fourth reading that is definite. This definite reading is here illustrated in (23a) with omis referring to a group of men already introduced in the discourse. Recall that in comparable episodic sentences, such definite readings are never available for either English or Romance plural BN: (23)

a. b.

N odja omis ta bebe na taberna. I saw the men drinking in the tavern. Raparigas txiga sedu. young women arrive early ‘The young women arrived early.’

. As Baptista (p.c) specifies, her (2000) description appears to be valid only for the basilectal varieties of Cape Verdean Creole, not for the acrolectal ones, where, apparently, the distribution of plural morphology is less restricted, involving a variety of factors not yet well understood (see Baptista this volume for further discussion). As I do not at present have relevant data on the acrolectal dialects, what is said below concerns only the former variety. Should the plural distribution reported here extend to further classes of nouns in other dialects, the analysis suggested below would remain essentially unchanged, provided these other classes of nouns align with the animate ones that are discussed here.

 Viviane Déprez

According to Baptista (2000), the definite and existential readings of plural marked BN occur largely in similar episodic sentence types. This suggests that plural marked BN in CVC allow for both a definite and an existential reading, the distinction being established essentially in the discourse. Newly introduced BN are interpreted as existential, previously introduced ones, i.e. familiar ones as definite. Plural marked BN also seem to have a fully uniform distribution; they can be found in object positions as well as in subject positions with essentially the same range of interpretations. Let us now consider unmarked BN. As a preliminary, it is important to note that in many cases, unmarked BN can be interpreted as semantically singular or plural. In other words, unmarked CVC BN appear to be largely under-specified for number interpretation, allowing both semantically singular or plural interpretations depending on the context. The conditions that determine distinct number interpretations remain at this point to be further studied. As illustrated in (24) (25) and (26) respectively, unmarked CVC BN allow for definite, generic as well as kind readings: Unmarked bare nouns with a definite reading (24) a. N odja omi na taberna. I saw the man in the tavern. b. N ta da un tabor, N ta mara ’l un korda. I will give you a drum and I will attach to it a rope. Kadu bos cega na tera, tokar tabor, When you get to earth, beat the drum, eto N ta larga korda. and I will let go the rope. Unmarked bare nouns with a generic reading (25) a. Omi ten pe di kaxtor men have foot of dog Men are always out b. Na kel tenpu, vida era sabi. in DEM time life was pleasant ‘In those old times, life was pleasant.’

(S-ST)

Unmarked bare nouns with a kind reading (26) a. João toma ritratu di lion na Afrika. João took pictures of the lion in Africa. b. Pur kausa di kaida demografika, kabuverdianu sa ta dizaparese. Because of a demographic fall, the Caperverdean is disappearing.

It is then both surprising and very interesting to observe that existential readings seem to be subject to intriguing restrictions. To better understand these restrictions, a closer look at subclasses of unmarked CVC bare nouns must be taken. Recall that, as noted above, overtly marked plural forms are not equally available for all classes of nouns in CVC basilectal varieties. They are essentially restricted to the class of [+animate,

Chapter 11. Implicit determination and plural 

+human] nouns. It is then interesting to note that for this noun class which manifests a robust singular/plural opposition, existential readings appear to be severely restricted when BN are singular both semantically and morphologically. According to Baptista (this volume), singular [+animate, +human] BN can sometimes have an existential reading in syntactically complex examples such as (27): (27)

Nu tenha xefri ki ta leba libru ku nos.18 we had leader COMP TMA take book with us ‘We had a leader who would take book(s) with us.

(C-ST)

But in common sentences such as (28) where BN are in object position and syntactically simple, definite readings are strongly preferred to existential ones (Baptista 2000). That is, an animated singular BN such as vizinhu is preferably interpreted as the neighbor, i.e. a definite already identified familiar individual, and not as an indefinite existential expression (a neighbor). (28)

a. b.

N odja vizinhu na merkadu. I saw the neighbour in the market. N enkontra amigu. I met the friend.

Even stronger restrictions apply to unmarked singular [+animated, +human] BN in subject positions. That is, in subject positions, existential readings of bare animate singular BN are essentially not found. These BN clearly have a strongly preferred definite reading: (29)

Xefri ka ta zanga ku nos. angry with us leader NEG TMA get ‘The leader does not get angry with us.’

(C-ST)

The interpretation of singular unmarked animated BN appears to and contrast in this respect with that of BN that are either overtly marked for plural or unmarked and simply interpreted as plural. Cases like (22) (23) above, containing a [+animate, +human] plural marked BN are more readily ambiguous between an existential and a definite reading. The same seems also to be true for unmarked BN that have a plural interpretation. These are not subject to asymmetric definite/existential readings in subject and object positions respectively but readily allow both. To sum up, in episodic contexts CVC BN that are singular both semantically and morphologically ([+animate] [+human] unmarked ones in basilectal CVC) favor definite readings over existential ones in subject positions and in some object positions (28). Similar restrictions on existential readings do not seem to obtain for unmarked BN that can be interpreted as plurals or for overtly plural marked BN. . It is notable that in this example, the relevant nominal expression is not strictly speaking ‘bare’ as it is modified by a relative clause. The presence of this modifier may be responsible for the possible existential reading here, which seems otherwise to be dispreferred for simple animated singular BN.

 Viviane Déprez

In sum, argument count BN in CVC can be said to have both a larger as well as a more restricted interpretational range than English or Romance BN. Count BN do not have to bear a plural morpheme to be acceptable, and the absence of plural morphology does not enforce a mass reading of otherwise established count nouns. Moreover, marked and unmarked plural BN allow for a very broad range of readings – they can be definite, existential, generic or kind. However, unmarked BN when they are both semantically and morphologically singular (usually animated ones) are more restricted in their interpretation. In the argument positions of episodic sentences their favored interpretation is definite, and not existential. Table 6 summarizes the possible readings of argument count BN for the basilectal variety of CVC. Table 6.

Kind readings Generic readings Existential readings Definite readings

.

Pl marked BN

unmarked BN

Unmarked animate sing BN

yes yes yes yes

yes yes yes yes

yes yes ??subject/object asymmetry yes

Data summary

Let us summarize the data reviewed so far. We have observed that argument count BN do not have a unique and constant meaning across either non-creole or creole languages. Non-creole languages such as English and Italian have been seen to differ in respectively allowing or disallowing kind readings for their BN. In our limited survey of creole languages, however, we have not encountered a single case of language that disallowed kind readings for its BN. Creole languages may thus systematically allow for kind readings, an interesting generalization to which we return below. We have shown, however, that BN in creole languages can differ in other respect and particularly in allowing or disallowing definite readings. As noted earlier however, this observed distinction with respect to definite readings of BN is not limited to creole languages. There are also noncreole languages such as Russian, Hindi or Chinese that allow for definite readings of their BN and non-creole languages such as English and Romance that disallow them. So perhaps the first point that the facts presented here makes clear is that the range of possible meanings for BN can vary significantly from language to language. There is no predetermined fixed set of meanings that BN are systematically expected to have crosslinguistically. The data further shows that contrary to what is sometimes assumed in the creolistic literature, the meaning of BN is not constant across creole languages. Variability in the meaning of BN is observed in creole and non-creole languages alike. This was illustrated here for instance by Seychellois and Cape Verdean creole, respectively. The former differs from the other French based creoles in allowing definite readings

Chapter 11. Implicit determination and plural 

for BN. The latter appears to favor different readings for its singular and plural marked BN according to their syntactic positions. It may be worth pointing out that this conclusion goes against the idea that creole determiner systems may be different from those of other languages. It was suggested by Bickerton (1981), for instance, that the dichotomy presence/absence of overt determiners in creole languages reflects a distinction between specific and non-specific uses of noun phrases – specific uses requiring overt determiners and non specific uses calling for bare nouns – but this suggestion is not supported by our cross-linguistic data. The use of BN with a definite meaning in both Cape Verdean and Seychelles Creoles clearly refutes this too simple dichotomy as it makes clear that BN can sometimes have an obviously specific and even a definite reading in creoles (see Bruyn 1995 for similar observations about Sranan). The above data also suggest that there is no complete uniformity across the determiner systems of creole languages since there is no complete uniformity in the interpretation of implicit determination, even if there are of course some common grounds. Recall indeed that three French based Creoles, Seychelles Creole, Mauritian Creole and Haitian Creole appear to differ in the use of simple definite count BN. Definite interpretations for simple count BN are present in Seychelles Creole, but absent in Mauritian Creole and Haitian Creole. As these data clearly indicate, there is no prototypical creole determiner system: BN in creole languages manifest both the same range of possible interpretations and the same crosslinguistic variability that was observed with non-creole languages. Turning to somewhat deeper analytical consequences of the data observed, it seems also clear that the variability in the meanings of bare nouns refutes a somewhat simple analysis that would posit the existence of a single null determiner whose semantic interpretation is fixed. Longobardi (1994) had proposed, for instance, that the meaning of this null determiner corresponds by and large to that of an existential or indefinite determiner with plural restrictions. Although it seems clear that an existential interpretation is available for BN in all the languages considered here, a fairly significant generalization to which we return shortly, it is also just as clear that this meaning is not the only ones BN can have. The hypothesis of a single null indefinite determiner is thus simply insufficient to account for the observed data. The data further suggest that a purely discourse-based or pragmatic approach to the meaning of BN in creoles would also fall short of accounting for the observed variability. If BN interpretations arose purely from (con)textual factors, the minimal expectation should be that similar contexts should trigger similar readings across languages. Such a view, however, cannot explain why in comparable anaphoric contexts creole languages differ with respect to the availability of definite readings for BN. Nor could a fully discourse-based approach explain why in CVC, for instance, plural marked animated BN in episodic sentences can have both existential and definite readings, while semantically and morphologically singular ones strongly favor a definite reading, particularly in subject positions. In sum, it seems neither possible to assume that BN can have the same meaning in all languages nor that their interpretation is the mere reflection of some minimal/optimal

 Viviane Déprez

recourse to general discourse strategies. The comparative facts, instead, appear to call for an approach that relates the observed differences in interpretations to learnable distinctions in languages, i.e. a parametric approach. It is important to point out, however, that such a conclusion does not belittle the importance of pragmatic or discourse factors in the determination of possible readings for BN. It just asserts that such factors cannot be the only ones. Granting this conclusion, the central question then becomes what these non-pragmatic learnable distinctions should be. I now turn to a discussion of some concrete proposal. Chierchia (1998), who was the first to offer a parametric approach to the meanings of BN, proposed that the relevant parameter resides in a flexible approach to the basic denotation of nouns. In his view, the basic denotation of nouns can differ crosslinguistically. When coming out of the lexicon of particular languages, nouns can denote either kinds 〈e〉 or predicates 〈e,t〉 or a combination of both in distinct languages, and these choices in turn serve to restrict the possible meanings BN can have. In languages like Italian in which nouns denote predicates, BN are allowed only where a null determiner can be licensed. In languages like Chinese in which nouns denote kinds, BN are allowed without restrictions, but unless a classifier is present they are predicted to have a mass interpretation. Based on a careful study of BN in Haitian Creole, I argued (Déprez 1999) that the predictions of Chierchia’s model are not verified for Haitian Creole data.19 In this work an alternative parametric approach is offered that centers on the distinct morpho-syntactic expressions of plural in languages. Section 3 reviews the main ingredient of this alternative parametric proposal and section 4 tests its predictions against empirical generalizations drawn from the data examined above to see how the model fares in predicting the observed variations.

. A parametric approach to the variable meanings of BN Déprez’ (1999, 2000, 2004) proposal is grounded in a generativist approach to syntax that assumes that nominal constituents consist of a basic lexical category NP dominated by a possibly variable number of distinct functional projections. This structural approach is known in the generative literature as ‘the DP hypothesis’ (Abney 1987 among others). In contrast to Chierchia (1998), Déprez assumes that nouns and their direct NP projections always have the same basic meaning or denotation in all languages: they always denote kind terms, and are thus always expressions of type 〈e〉. Other interpretations are produced by the compositional combination of this basic denotation with a variety of operators introduced by the functional projections of a given language and ultimately by its morphology.

.

See also Déprez (to appear) for further arguments.

Chapter 11. Implicit determination and plural 

For Déprez (1999, 2004) one relevant parameter – called for want of a better name the Plural Parameter – is morpho-syntactic in nature and distinguishes two broad sets of languages, the +Plural Languages (+PL) and the –Plural Languages (–PL), on the basis of whether the structure of their nominal expressions does or does not obligatorily include a functional projection for Number called NumP.20 In +PL languages, a NumP projection is necessarily present.21 That is, NumP always projects even when the noun phrase is ‘singular’, with no apparent overt morphology. The consequence of this proposal is that in +PL languages, apparent bare nouns are in fact always syntactic projections of NumPs as in (30) and never truly bare NP: The structure of bare count nouns in +PL languages (30)

NumP | Num’ Num0

NP | N

Syntactic evidence for this claim discussed in Déprez (to appear) is based on the impossibility of bare nouns to occur in strict predicative positions. If BN could be basic predicates as Chierchia claims, they should be able to occur in copular sentences just like adjectives do. As it turns out, however, such a prediction is empirically incorrect for at least English, German and Romance. In +PL languages the obligatory NumP projection introduces a semantic counter, i.e. a measure function which is the semantic translation of ‘countability’, a notion here assumed to be structural or syntactic in nature, rather than lexical22 since it is brought about by the projection of NumP. In –PL languages, on the other hand, the projection

. It is important to note that despite its name, the proposed parameter does not claim that –PL languages fail to express plurality, only that they do not enforce the expression of number. . Déprez (to appear) suggests that the measure phrases associated with mass nouns are comparable in nature to plural or classifier: they are part of the projection of NumP. If this is correct, then mass nouns too require the projection of NumP. Plural ranges over cardinalities, which is one type of measure, measure phrases and partitives range over less uniformly defined units of measure. While plural forces individualization down to minimal realizations of kinds (atoms), mass nouns tend to individualize maximal ones. . Déprez follows here Bosweld de Smet (1998) in distinguishing count and mass, from countability and uncountability. The former is a lexical distinction that has cognitive but no syntactic consequence. The latter is a syntactic distinction that can supersede the mass count distinction. For a more detailed discussion of this proposal see Déprez (to appear).

 Viviane Déprez

of NumP is optional, and NumP does not have to contain a semantic counter. That is, the NumP projection is possible but not obligatory, with the consequence that BN may either be simple projections of NP or as above, projections of NumPs. The formulation of the Plural Parameter proposed by Déprez (1999, 2001) is given below: (31)

The Plural Parameter NumP must project and contain a semantic counter (+/–)23

Like other parametric distinctions in current day syntactic theory, the proposed Plural Parameter concerns functional structure and capitalizes on an overt morphological difference among languages, namely the presence vs. absence of a ‘rich’ inflectional plural morphology for setting its value. That is, languages that manifest a ‘rich’ plural morphology will have a positive setting for the parameter and languages that manifest a ‘poor’ plural morphology a negative one. However, the parametric choice between +PL and –PL languages, I suggest, is not so much in the ‘richness’ of the morphological marking for number, in the traditional sense, as in its ‘unavoidability’. That is, the parameter separates on the one hand, +PL languages in which, in the most general case, no morphological form of nominal expressions headed by count nouns is understood outside of the number system. Count nouns are either plural or singular and interpreted as such, so that for count nouns the absence of plural morphology generally entails a singular interpretation.24 In -PL languages, on the other hand, one form of count nouns manifests what Corbett (2000) has called ‘general number’ – i.e. a form neutral for number interpretation with the meaning of ‘[one or more] OR [exactly one x]’ – what I have here called a form under-specified for number.25 In such languages, although the presence of a plural morpheme (dependent or not) equally . Déprez (2004) discusses the possibility of a distinct formulation for the plural parameter. As formulated in (31) the parameter contains a conjunction imposing a counter for all languages/nominal expressions that have obligatory NumP. However, the parameter could in principle be subdivided into two separate parts, allowing a language with obligatory NumP but where Num does not have to contain a counter (+Obl NumP; –counter). The expected properties of such a language would be to disallow direct access to kind (obligatory NumP) so that BN would fail to be kind referring expressions as in Italian, but underspecification for number would be allowed. Brazilian Portuguese as described by A. Müller (2001) apparently has these properties, suggesting that dissociated formulation of (31) is pertinent for at least some languages. However, since Brazilian Portuguese is outside the scope of the present paper, I will maintain (31) as is. . Special constructions that allow a mass interpretation of a count noun (cf. There is apple in this salad/There was egg all over the wall) are outside the scope of the present paper. Since such readings are not available in common argument situation (cf. *Book interest students. / *Students read book.), we suspect that they can be constructionally (i.e. syntactically) coerced for count nouns. .

See Déprez (to appear) for a more complete discussion of this distinction.

Chapter 11. Implicit determination and plural 

entails a plural interpretation, its absence on the other hand fails to entail a singular reading for count nouns. The proposed Plural Parameter locates cross-linguistic variation in the lexicon, or more specifically in the inventory of morpho-syntactic features a language may choose to instantiate. It adds, however, a twist to this broadly accepted view of parametric variation, in the sense that morphological ‘richness’ is here tied to a specific interpretation of Num0, with the result that the parametric choice now has clear semantic consequences in addition to the expected syntactic ones. The key idea is that languages that manifest a ‘rich’ (i.e. unavoidable) number morphology project a nominal structure with an obligatory NumP that contains a semantic counter over realizations of a kind. Languages that do not have a ‘rich’ number morphology, in contrast, do not enforce the presence of NumP, nor of a counter, although of course, they may allow them as options in specific cases, i.e. when numerals are overtly encoding such a counter. Within the sample languages considered here, English and the Romance languages are clear instances of +PL languages. In English as well as Romance, an argument count noun is interpreted as singular if it is unmarked. The French based creoles in contrast are clear instances of –PL languages. Unmarked argument count BN can generally be interpreted as either singular or plural.26 The case of Cape Verdean Creole is particularly interesting here because it seems less clear-cut. CVC appears on the one hand to have a class of count nouns (roughly speaking the +animate, +human ones) that functions as in the +PL languages. For these nouns, interpretation for number is a function of morphological marking: marked forms are plural, unmarked forms are singular. CVC has, on the other hand, another class of count nouns (all the others) in which nouns function as in the –PL languages: these nouns allow for both a singular or a plural reading of their unmarked form. What the facts of Cape Verdean Creole then suggest is that the proposed Plural Parameter may apply either language ‘internally’ to particular classes of count nouns, or to a language as a whole to all count nouns without distinctions. That is, in any given language, the obligatory projection of NumP may be specific to a particular class of count nouns with the clear consequence that this class should differ in the behavior of its BN. Or NumP may be required for all count nouns across the whole language. Concretely, I tentatively propose here that Cape Verdean Creole (of the basilectal variety) is overall a –PL language, yet with a class of nouns, the animated ones, that obligatorily projects a NumP. It is in this sense that this class of nouns may be thought to behave like those of the +PL languages. I return below to the consequences this division has on the possible readings of these BN. . Importantly, the claim is not that the meaning of a BN is always ambiguous between singular and plural. In many cases indeed, the context will suffice to clearly set the relevant interpretation. The point is simply that the same nominal form allows for distinct interpretations, albeit, of course, in different contexts. In English argument positions, in contrast, an unmarked noun is never interpreted as a plural and context manipulation never allows such a reading. *I like apple.

 Viviane Déprez

As noted above, the proposal assumes that NumP can have two distinct possible realizations in the two sets of languages distinguished by the parameter: 1.

Num can be a pure instantiation of the ‘Carlsonian realization rule’, as in (32): (32)

Num = λw λ x. λK Rw (x, K)

This realization rule is a relation frοm kind to objects that gives the set of objects x that instantiates a given kind K (where x is a member of the kind K) in a given world w, independent of number.27 In this case, the resulting denotation of a NumP projection is that of a property true of individual(s) that are members of the relevant kind. The property itself is under-specified for number in the sense that it does not specify the number of object(s) that needs to verify it. Note that given the formulation of the Plural parameter in (31) above, this type of NumP is allowed only in –PL languages; it is only in –PL languages indeed that a counter can be missing from the projection of NumP. 2. The second realization of Num takes the form of a Carlsonian realization relation plus a measure function (i.e. a counter) that returns a countable property. I follow here the formulation of Krifka (1995) in assuming a measure function (OU = object Unit), which, when applied to a kind and an object, returns a number, giving us the number of objects x instantiating the kind in a given world w. The denotation of Num for this case is given in (33): (33)

Num = λnλwλx λK[Rw (x,K) & OUw (K)(x) = n]

That is, given a kind K, we obtain sets of objects x with cardinality n that are members (i.e. realizations) of the kind. Importantly, note that the counting function OU introduces a number argument. The presence of this number argument in turn calls for the presence of some element to saturate it. Various morphological elements can saturate the counting function in various ways, but it is crucial that this saturation be accomplished, otherwise the resulting NumP fails to be interpretable. For instance, suppose that NumP contains a numeral term. Then this numeral term can saturate the number argument of the counting function (34) for instance. It gives the relevant interpretation for the numeral three as applied to the kind BEAR. (34)

[NumPThree bears] λwλx [Rw (x,BEAR) & OUw (BEAR) (x) = 3]

In plain language, (34) denotes sets of three individuals x that belong to the kind BEAR in a given world w. The same type of saturation obtains if the numeral term is one or an indefinite determiner similar to English a. A second way of saturating the number argument of (33) is with plural morphology. If NumP contains a morphological plural

.

The resulting set could be a singleton or a multi-membered set.

Chapter 11. Implicit determination and plural 

marker (such as English s), the number argument will be existentially quantified over as in (35). (35) represents sets of bears whose cardinality is superior to 1: (35)

[NumP Bear +s] ∃n λwλx [Rw (x,BEAR) & OUw (BEAR) (x) = n>1]

However, if NumP contains no numeral or morpheme that can adequately saturate the number argument of the counting function and if nothing happens at a higher functional level, such as for instance the introduction of a definite determiner or of a quantifier that could effect the saturating of the counting function, then the number argument will remain unsaturated and the corresponding BN will be un-interpretable: it will contain an unbound variable. On the view proposed here, this is precisely what I propose happens with singular BN in +PL languages like English or Romance. Recall that the proposed parameter (31) states that in +PL languages the projection of NumP is obligatory and must contain a measure function. This implies that in a +PL language a BN calls for saturation. And if there is nothing to saturate the measure function, the BN will be ruled out as uninterpretable. This explains the impossibility of argument singular BN as in *I eat apple. If either plural morphology or a given determiner is present, then saturation of the measure function (i.e. the counter) will be straightforward and a singular NumP licensed. The proposal has thus a number of important consequences. A first important consequence is that it immediately rules out the use of singular BN in +PL languages. In this sense, it accounts for the number restriction on BN discussed above for English and Romance. Let us make the plausible assumption that ‘singular’ forms do not feature an ‘interpretable’ singular zero morpheme in languages like English or Romance but rather the ‘absence’ of a morpheme. On this assumption, the Plural parameter will correctly rule out the use of a singular without a determiner as follows: since there is no semantically interpretable singular morpheme, the counting function that is introduced by the obligatory projection of NumP will fail to be saturated. Consequently, unless a determiner is merged, the number variable is left unbound, and the form is uninterpretable. More generally, the point is that given the Plural parameter, BN in +PL languages are un-interpretable unless some element is there to provide adequate saturation for the number variable of the obligatory NumP. The plural morpheme is the element that can most easily serve this goal; this explains why count BN must be generally plural in languages like English and Romance. I here make the further assumption that determiners always succeed in saturating the measure function but that number morphology may sometimes fail. For English and the Romance languages, I have suggested that ‘plural morphology provides adequate saturation but that ‘singular’ (presumably the absence of a morpheme) does not. Suppose, however, that in some +PL language, even the plural morphology failed to provide adequate saturation for the measure function. That is, if the plural morphology on nouns were in some sense too weak to have any semantic interpretation, this would have as consequence that BN, singular or plural, would be generally ruled out. Nominal expressions would be generally

 Viviane Déprez

un-interpretable in the absence of an appropriate determiner. As a consequence, in such a language, only determiners can provide adequate saturation for the number argument so that their presence is effectively required. The case of French clearly fits the bill quite well. French nominal expressions are either singular or plural and do not generally permit an under-specified interpretation for number. This fact in itself suffices to mark French as a +PL language, thus enforcing the projection of NumP and the presence of a measure function. On the other hand, it is well known that the plural morphology of French nouns remains generally unpronounced, being thus more of an orthographic reality than a true linguistic one. In short, plural morphology on nouns is largely absent from the oral language. French is thus in effect a language that expresses singular/plural distinctions on its determiners and not on its nouns. If French plural marking on nouns is denied any semantic interpretation, the present proposal will derive straightforwardly the fact that BN do not occur in this language. Given that French is a +PL language, if its plural morphology on noun has no semantic interpretation, the NumP measure function will fail to be adequately saturated unless some overt determiner is present to accomplish this task. This in effect, suffices to render argument BN impossible in French and the presence of determiners obligatory. Distinguishing French from the French based Creoles is moreover straightforward in the proposed approach. Recall that French based Creoles differ from their superstrate in not enforcing number marking. Thus they are –PL languages, which in our term means that they are languages with optional NumP and whose NumP, when projected, does not contain a measure function (i.e. a counter). As a consequence, BN in these languages, be they truly bare NP or bare NumP, do not require any saturation and are quite generally allowed. Since they have no obligatory counting function that requires saturation, BN are always interpretable and the presence of a determiner or a plural morpheme is thus not needed for interpretability, although of course, it is allowed. The proposed Plural Parameter as formulated in (31) above has a number of further direct semantic consequences: 1. Only –PL languages allow direct access to the kind denotation of nouns. This result comes about because in these languages, NumP may be missing, and bare nouns may thus be truly bare, i.e. projecting only to NP. Since the basic denotation of nouns is that of a kind, this has as a consequence that kind readings of bare nouns are expected to be possible in all –PL languages and in all contexts compatible with such readings. In +PL languages on the other hand, direct access to kind denotation is impossible given the obligatory projection of NumP. Access to kind readings thus requires the presence of a relevant operator. A concrete consequence of this proposal is that +PL languages may in fact fail to allow a kind reading for BN. Such readings would require the presence of a determiner. In the data reviewed here, we have seen that these predictions are verified. All –PL languages (i.e. here all the creole languages) allow for a kind reading of their BN.

Chapter 11. Implicit determination and plural 

Kind readings on the other hand are missing for Italian and other Romance BN, which are all +PL languages. In these languages, kind readings require the presence of a definite determiner, a requirement that can be seen as a direct consequence of the proposed Plural parameter. 2. Only –PL languages can have BN that are under-specified for number, i.e. compatible with either a plural or a singular interpretation. This result comes about because in –PL languages, NumP can fail to project, and when it does, it does not contain a counter. As a result, there is no counting function that verifies how many instances of objects satisfy the property of being a member of a kind. Again we see that this prediction is verified in our sample, since all the creole languages have BN that are under-specified for number, i.e. compatible with either a singular or a plural interpretation in adequate contexts. 3. In +PL languages, if relevant morphology is missing, BN that contain an unsaturated measure function are excluded. This result comes about because in +PL NumP must project and contain a counter. That is NumP obligatorily introduces a counting function that requires saturation. If saturation is not effected, the resulting expression fails to be interpretable. This prediction again is verified in the languages of our sample. In all +PL languages considered here, singular BN are excluded as singular morphology fails to provide adequate saturation for the measure function. Nominal expressions, if bare, must be plural, or else they must be dominated by overt determiners. Note that this prediction has a particularly interesting consequence for a language like CVC. Recall that as proposed above, CVC is essentially a –PL language except that it has a class of nouns acting like those of +PL language in having an obligatory NumP with a counter. Concerning this class of nouns, the prediction of our approach is that BN should not be possible unless they are marked for plural. Confirming this prediction, recall that existential readings of bare singulars CVC BN were observed to be essentially missing, the reading of a singular [+animate, +human] CVC BN being preferably definite. As it turns out, the fact that singular BN cannot have an existential reading is predicted: these readings are missing because they represent cases in which NumP has projected and contains a counting function that needs saturation, but there is no relevant morphology to perform this job. Unaccounted for at this point, however, is the fact that singular BN can have a definite reading. This reading, I suggest (see Section 4 below) denotes a different structure for a BN, one that in fact contains an additional null D projection. In this respect, the structure of bare singular definites is thus equivalent to that of a full DP with an overt definite determiner. Note that on this view, singular CVC [+animate, +human] BN now behave as expected. Like English or Romance BN, they are excluded unless a determiner is merged to saturate the counting function introduced by their obligatory NumP. Assuming that a null definite D (as opposed to an unrealized singular) can do the saturating job, an assumption that is not so surprising given the familiarity and contextual reference that is at the root of definiteness, the proposed approach turns out to be sufficient to fully account for the

 Viviane Déprez

apparent erratic behavior of the CVC BN. CVC BN behave like those of –PL languages for the most part. The [+animate, + human] nouns behave like those of +PL language in requiring plural or a determiner to be licensed. The difference between CVC and English, however, is that CVC has a null definite determiner while English does not. Since generic and kind readings can generally be built on definite determiners, as the use of an overt definite determiner in French demonstrates, the presence of a null definite determiner can account for these readings. The only one that could not be achieved through the presence of a definite determiner is the existential reading and, as was shown above, this is precisely the one that seems to be mostly lacking for singular [+animate, +human] bare CVC BN. A final prediction of this approach is that a +PL language that has no morphology to provide adequate saturation for the counting function should simply disallow argument BN altogether and always require the presence of some determiner. This prediction is verified with the particular case of French already discussed above.

. Confronting theory and data As the discussion above shows, Déprez’ proposal accounts for a number of important cross-linguistic restrictions on BN. What remains at present to be accomplished is to show how it can account for the different readings BN can have. The goal of this section is to consider the different types of readings in turn and to detail how the proposed model accounts for them.

.

Existential and Generic readings

Quite notable in the range of data considered in Section 2 above is the fact that BN, whenever they are possible and either plural or interpretable as plural, always have at least two common readings, an existential and a generic one. This observation is summed up here as Generalization 1. Generalization 1: Wherever plural BN are possible, they generally have at least an existential and a generic reading.

On the view explored here the possibility of existential and generic readings for BN follows from the projection of NumP. I have proposed that the semantic role of NumP is that of building a property from a kind, namely the property of being an object, member of a kind. Closure of this property under standard Existential closure in episodic sentences will derive straightforwardly the existential reading of BN. For +PL languages, the existential reading of the BN dogs will be as in (36), which in plain language can be roughly translated as: there exist (some) individuals such that they are realizations of (= members of) the kind DOG and they number in (sets of) more than 1.

Chapter 11. Implicit determination and plural 

(36)

∃x ∃n [R (x,DOG) & OU(DOG)(x) = n >1]

The existential reading of BN in a –PL language comes about in essentially the same way, save for the number specification. Thus (37), which in plain language can be translated as: There exist some individual(s) such that it is/they are realization(s) of (= members of) the kind DOG. (37)

∃x [R (x,DOG)]

Similarly, the generic reading can be assumed to derive from the presence of a Generic operator introduced by the verbal tense/aspect morphology of so-called characterizing sentences. The generic reading of the BN dogs in a +PL language would be as in (37), which in plain language would roughly translate as: Generally, elements realizing the kind DOG and numbering in more than one… (bark …). (38)

Generic reading : Generic operator Gen x ∃n [R (x, DOG) & OU (DOG)(x) = n >1]

In a –PL language, this reading would be essentially equivalent, save for the number specification. The denotation is given in (38) and roughly translates as: Generally (an) element(s) realizing the kind DOG…(bark(s)) (39)

Gen x [R (x, DOG)]

Given that the projection of NumP is possible in all languages, Existential closure and binding by a Generic operator will account for the Generalization 1 (G1) deriving, for all BN (in fact bare NumP here), both an existential and a generic reading. Semantically, our proposal is here equivalent to the view that BN can be interpreted as a type of Heimian indefinite variable that can be captured by Existential closure or by other available operators such as the Generic operator, as proposed among others by Diesing (1992) and Krifka (1995). Contrasting with these previous approaches, however, the existential and generic interpretations are here tied to a particular syntactic structure, namely the projection of NumP. That is, BN receiving an existential or generic interpretation are never strictly speaking bare NPs but always bare NumP, a proposal that ultimately accounts for the morphological plural restriction on BN in +PL languages. The proposed Plural Parameter has some important consequences with respect to G1. First, recall that it enforces the projection of NumP in +PL languages. This means that in +PL languages, the existential and generic readings are the only readings “directly” accessible and thus the most basic interpretations for BN. In this respect, +PL languages differ from –PL languages like the French based Creoles in which the most basic reading of BN is the kind reading, i.e. the reading that obtains when NumP fails to project. The predictions of the parameter are thus twofold. First, it predicts that there can be no +PL languages in which BN lack an existential and a generic reading (provided of course that relevant verbal operators are accessible). Second, it predicts that no –PL languages could lack the kind reading of their BN. In contrast, kind readings could well be lacking in +PL languages. As we have seen above, all the clearly –PL

 Viviane Déprez

languages reviewed here (i.e. all the creoles) allow for a kind reading of their BN. In contrast, Italian and other Romance languages, i.e. all +PL languages, illustrate typical cases of missing kind readings for BN.28 To sum up, existential and generic readings in all types of languages involve the projection of NumP. NumP in +PL and –PL languages differ in that the presence of a counter over realizations of a kind is enforced in the former but not in the latter. The presence/absence of the counter in NumP determines in turn whether the number of objects realizing a kind is specified or left unspecified so that BN are compatible in appropriate contexts with either a singular or a plural interpretation.

.

Kind readings

It is a basic tenet of the proposed model that N and its direct projection NP are kind referring expressions of type 〈e〉 in all languages. However, it follows from the proposed Plural Parameter in (31) that this basic reading is directly accessible only in –PL languages. Indeed since in a +PL language, NumP must project, and since the import of NumP is essentially to retrieve a property from a kind, then only the property reading is directly accessible in a +PL language. As a consequence, kind readings will have to be constructed with operators in +PL languages, while they are directly accessible in –PL languages. There are in principle several ways that kind readings might be coerced through an operator. Most +PL languages that have a definite determiner seem to use it to construct kind readings. This is true for instance of all the Romance languages as well as of German. Indeed although German commonly allows generic readings of BN, the presence of a definite determiner is required for kind readings (pc. G. Shoener). However, this is not always true given that in some +PL languages, and English is a case in point, a definite plural determiner does not allow kind reference. To lay out how kind readings can be retrieved in different languages via distinct operators is beyond the scope of this paper. Relevant to this paper is the prediction that kind readings cannot in principle be missing for BN in –PL languages. This follows because in –PL true bare NPs are always an option. This prediction seemed to be verified in the –PL sample of languages studied here, which are all creoles, as well as in well known non-creole

. Following Krifka (1995), we take kind readings of bare nouns in English to be coerced in the presence of a kind predicate by a null operator akin to an intentionalized definite (see Déprez (to appear) for a proposal). This possibility seems to correlate with the presence in English of a null definite in coordinated bare plural as a restricted option not available in Romance (see Heycock & Zamparelli (2003)). This possibly corresponds to a further difference in the way number is computed within +PL languages. A reviewer suggested to us that Longogardi’s proposal that N raises to D0 in English at LF is another possibility to license kind readings of BN in English. This proposal is indeed compatible with the view proposed here. However, given that the movement is assumed to occur at LF, this proposal makes no testable empirical predictions and can thus neither be confirmed, nor entirely dismissed.

Chapter 11. Implicit determination and plural 

–PL languages such as Chinese, Japanese and Korean. Thus, an important point of the present proposal is that it does not cast the overall availability of kind readings in creole languages as a feature that is specific to creole languages in particular but more generally as a feature that is specific to –PL languages. There are clearly –PL languages, like Chinese, that are not creole languages. Whether or not there are in fact creole languages that are +PL languages is at present unknown. But the important point of the present proposal is that it can account for a feature that is clearly prevalent in creole languages without having to conclude that creole languages are in any way special in this regard.

.

Definite readings

As seen above, BN may have definite readings in some but not all the languages considered. The generalization governing this possibility appears to be quite simple. As stated below under Generalization 2, BN seem to allow a definite reading only in languages that lack an overt lexical definite determiner. Generalization 2: Bare nouns allow regular definite readings only in languages that lack lexical definite determiners.

Generalization 2 may seem rather trivial at first, but this apparent simplicity is in fact deceptive. Indeed, it does not appear to be generally true that BN systematically take on the meanings of missing overt determiners in given languages. A particularly telling example of this lack of complementary distribution between overt/covert determiners is found in Cape Verdean Creole. As (40) shows, CVC has both a singular and a plural indefinite determiner un/ uns and yet it clearly allows BN with existential or indefinite readings: (40)

a.

b.

N odja omis sintadu riba di kaza. I saw men sitting top of house ‘I saw men sitting on top of the house.’ N odja uns omi sintadu riba di kaza I saw some men sitting top of house ‘I saw some men sitting on top of the house.’

Similarly, HC and SC feature indefinite determiners and yet BN with indefinite readings are allowed. Simplifying somewhat for the sake of clarity, while there does not seem to be a clear complementary distribution between overt determiners and related interpretations of BN in the indefinite determiner domain, such a complementary distribution does nevertheless seem to hold for definite determiners. Let us have a quick look back at the data to see how strong G2 appears to be. Starting with non-creole languages, clearly English and Italian have overt definite determiners and just as clearly their BN do not allow definite readings. As Dayal (2001) has shown on the other hand, Hindi is a language with no overt definite determiner (although

 Viviane Déprez

it has a demonstrative one) and it clearly allows definite readings of BN. The same is true of Chinese, Korean, Japanese and Russian (Cheng and Sybesma 1999), bringing important confirmation to G2 for non-creole languages. Turning now to creole languages, G2 seems also correct for HC, as argued in Déprez (2001), HC has a definite determiner la and correlatively, it lacks regular definite readings of BN. That the determiner la is strictly speaking just a definite determiner in HC has been contested (Zribi-Hertz this volume), as la is sometimes assumed to have a demonstrative force.29 What seems clear nevertheless is that HC also features an overt demonstrative determiner sa a that is clearly distinct from the simple definite one. Thus, a perhaps better and more cautious formulation of G2 should be: Generalization 2’ Only languages without overt definite determiners distinct from demonstrative ones allow regular definite readings of BN.

That G2’ also obtains for the other creoles considered is perhaps harder to argue convincingly here, given space constraints. Nevertheless, I will try to make a brief case for it, referring the readers to future work for a more careful study of this point. Consider first Seychelles Creole. On the basis of Bollée’s (1977) grammar and of the corpus of texts she provides, Seychelles Creole seems to have a demonstrative determiner namely sa, but no definite one. La with a definite reading does not appear in Bollée’s corpus. All examples of la are adverbial. It seems then that in contrast to the other French based Creoles, at least this dialect of Seychelles Creole fails to have a distinct determiner la. In this respect, it is then quite striking to observe that this French Creole (dialect) also seems to be one that features regular definite readings of BN. Mauritian Creole in contrast seems to form an interesting minimal pair with Seychelles Creole in both regards, although further empirical research is needed to confirm these points. First, la can be used as a regular definite determiner in MauC, and MauC also features the distinct demonstrative sa, sa-la. Second, regular definite readings of BN do not seem to be available, as suggested by examples such as (15) above. The study of a MauC corpus could help confirm these first impressions, but as I have so far not had access to one, this suggestive opposition must remain tentative at present. Should further data confirm this description, it is clear that the minimal pair formed by Seychelles vs. Mauritian Creole would provide rather strong support for G2’.

. As discussed in Déprez (2003b, 2006), the HC definite la is clearly not directly comparable to the definite determiner of a language like French, since for instance, it fails to license kind readings. This, however, does not suffice to qualify it as a demonstrative, contra Zribi Hertz (this volume). Note first that the lack of kind reading is shared by the English plural definite, clearly not a demonstrative determiner under all semantic accounts. Strong evidence also opposes HC la to even the weakest kind of demonstrative like French ‘ce’. The two may share an anaphoric value but they radically differ in their ability to license ostentative oppositions, a distinct hallmark of demonstratives. A full discussion of this issue is taken up in work in progress.

Chapter 11. Implicit determination and plural 

Let us finally consider Cape Verdean Creole. With respect to G2’, CVC appears to be rather similar to SC: at first view, it seems to have a demonstrative determiner kel/kes, but seems to lack a distinct definite one. However, Baptista (1997) describes kel as: ‘the definite article kel (sg)/kes (plur) which is also used as a demonstrative, as in languages such as Japanese’ (Baptista 1997: 11) and mentions that kel/kes has weakened anaphoric uses akin to that of a definite determiner. In the thesis, however, kel is also described as a demonstrative and translated by the English demonstrative: (41)

Kel omi e grandon That man is very tall

(42)

Singular Plural Es …(li) es…(li) This these Kel …(li) or (la) kes …(li) or (la) This or that these or those

“Es and kes would be best translated by English this and that, respectively” (Baptista 1997: 50). So the question of whether kes/kel is a definite or a demonstrative determiner seems rather unclear and in need of further clarification. There are nevertheless a few empirical arguments that, in my opinion, strongly favor considering kel/kes as a demonstrative. First, kel can be used in ostentative opposition and reinforced by elements such as li and la, expressing respectively a proximal or distal relation to the speaker. Reinforcers of this type cross-linguistically tend to combine with demonstrative determiners, not with definite ones. Thus, the fact that kel can be so reinforced argues in favor of its demonstrative nature. Second, in her grammar of CVC, Maria Dulce Almada de Oliveira (1961) defends the view that CVC has no definite determiner, not that it lacks a demonstrative one: (43)

“O artigo definido pode dizer-se que não existe no crioulo de Cabo Verde.” Cabo Verde P89 (de Oliveira 1961: 89)

Third, as shown by Inverno & de Sousa’s (2001) SPCL presentation, common uses of the Portuguese definite article in generic and in possessive constructions only have a null correspondent in CVC; Kel cannot be used in these cases. These facts all suggest that kel is not a definite determiner, but a demonstrative, which like most demonstratives, can have anaphoric uses. If this conclusion is correct, then CVC squarely falls within G2’, as it clearly features definite readings of BN.

.

A tentative account for Generalization 2’

The proposal reviewed in Section 2 does not directly address the question of definite readings of BN. However, within its framework, a rather simple account for G2’ can be given. If we make the plausible assumption that all languages have a definite determiner in the sense that all languages contain in their lexical inventory some functional

 Viviane Déprez

head whose interpretation is that of a definite determiner, it is then sufficient to assume that in some languages this determiner is overt, while in others, it is covert. The fact that the languages in our sample appear to have both an overt and a covert option to express the definite article will then simply follow from very general notions of economy in the lexicon, namely a simple general tendency to avoid lexical redundancy. To my knowledge, there are no languages that have several distinct overt definite determiners with exactly the same meanings. If so, there is no more reason to assume that a language could feature both an overt and a covert definite determiner with the same meaning. Assuming that a given language can have a null definite determiner, then the structure of BN with such a reading is straightforward: it is exactly identical to the structure of nominal expressions with an overt definite determiner in other languages, i.e. a DP. Following much recent work in the generative literature that argues that definiteness is a structural notion, I suggest that there is a particular projection for the definite determiner D0 (see also Lyons 2000) and that the structure of BN with a null definite determiner is thus a full DP as in (37) with a null definite D0: (44)

DefP | NumP | NP

To summarize, this section has explored the validity of G2’ in various creole and non-creole languages. Although further empirical research would be useful to establish our observations more firmly, there seems to be reasonable ground to believe that regular definite interpretations of BN are in complementary distribution with regular overt definite determiners. The assumption made here, namely that BN with definite interpretations are in fact DPs that have a null definite determiner, suggests that this complementary distribution is simply a more specific example of the very general tendency in languages to avoid strict synonymy, and that null determiners can sometimes, but not always play the same role as overt ones in distinct languages.

Conclusion This paper has reviewed a number of empirical facts concerning the interpretation of BN in a variety of creole and non-creole languages in order to draw some generalizations about the former and to test Déprez’ (2001, to appear) proposed model against the cross-linguistic similarities and differences observed. The overall conclusions that emerge from this study are: 1.

Existential and generic readings seem to be universally available in languages that allow plural BN. In the proposed model, this follows from the universal availability of NumP and from standard existential closure and generic binding.

Chapter 11. Implicit determination and plural 

2.

3.

A second conclusion is that kind readings of BN seem to be available in all –PL languages while they may be missing from +PL languages. On the proposed view, this follows from the Plural Parameter because NumP projections can be missing only in –PL languages, thus allowing the kind reading of a truly bare NP as the most basic one. In contrast in +PL languages a NumP projection must be present enforcing the existential/generic reading as the most basic reading for BN. Kind readings must then be derived either constructionally or through the presence of an overt operator. Regular definite readings of BN seem to be in complementary distribution with overt definite determiners. This follows in our view from the proposal that there is a null definite determiner that can occupy the same position in the nominal functional structure as overt ones. The prediction of this proposal is that BN with null definite determiners are expected to be semantically and syntactically equivalent to nominal expressions with overt definite determiners. This prediction is distinct from the one concerning existential readings of BN, which are not expected to be equivalent to nominal expressions with overt indefinite determiners. This prediction remains to be empirically verified, a task left for future work.

References Abney, S. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Adli, A. 2002. Social and style factors in the choice between word order variants. Paper presented at NWAV 31, Stanford University. Almada, M.D de Oliveira. 1961. Cabo Verde: Contribuição para o estudo do dialecto falado no seu arquipélago. Lisboa: Junta de Investigações do Ultramar, Centro de Estudos Políticos e Sociais. Baptista, M. 1997. The Morpho-syntax of Nominal and Verbal Categories in CapVerdean Creole. PhD Dissertation, Harvard University. Baptista, M. 1999. The role of null morphemes in marking specificity and non specificity of NPs: A comparative analysis. Paper presented at the 9th International Colloquium on Creole Studies, Aix-en-Provence, June 24. Baptista, M. 2000. Interpretational variability of bare NPs in Cape Verdean Creole. Paper presented at Rencontres sur les Créoles. Paris. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Bollée, A. 1977. Le Créole Francais des Seychelles. Tubingen: Max Niemeyer. Bosweld de Smet, L. 1998. On Mass and Plural Quantification. The case of French des/du – NPs. PhD Dissertation, Amsterdam/Groningen. (ILLC dissertation series). Bruyn, A. 1995. Grammaticalization in Creoles: The development of determiners and relative clauses in Sranan. Amsterdam: IFOTT. Carlson, G. 1977. Reference to Kinds in English. PhD Dissertation, University of Massachusetts at Amherst. (Published 1980 by Garland Press, New York). Carlson, G. & Pelletier, J. 1995. The Generic Book. Chicago IL: University of Chicago Press. Cheng, L. & Sybesma, R. 1999. Bare and not so bare nouns and the structure of NP. Linguistic Inquiry 30(4): 509–542.

 Viviane Déprez Chierchia, G. 1998. Reference to kinds across languages. Natural language Semantics 6(4): 339–400. Chomsky, N. 2000. Minimalist inquiries: The framework. In Step by Step: Essays in minimalist syntax in honor of Howard Lasnik, R. Martin, D. Michaels & J. Uriagereka (eds), 89–155. Cambridge MA: The MIT Press. Corbett, G. 2000. Number. Cambridge: CUP. Dayal, V. 2001. Number marking and (in)definiteness in kind terms. Ms. Rutgers University. Déprez, V. 1999. De la nature sémantique des nominaux sans déterminant en créole haitien. Langues 2(4): 289–300. Déprez, V. 2000. Parallel (a)symmetries and the structure of negative expressions. Natural Language and Linguistic Theory 18(2): 253–342. Déprez, V. 2001. On the nature of Haitian Bare NPs. In Current Issues in Linguistics Theory, D. Cresti, C. Tortora & T. Satterfield (eds), 48–61. Amsterdam: John Benjamins. Déprez, V. 2003a. Constraints on the meanings of bare Nouns. In Current Issues in Romance Linguistics, R. Nunez-Cedeno, L. Lopez & R. Cameron (eds), 292–310. Amsterdam: John Benjamins. Déprez, V. 2003b. The architecture of DP in French-based creoles. In Romance Languages and Linguistic Theory 2001, J. Quer, J. Schroten, M. Scorretti, P. Sleeman & E. Verheugd (eds), 215–235. Amsterdam: John Benjamins. Déprez, V. 2004. Morphological number, semantic number and bare nouns. Lingua. Déprez V. 2006. On the Conceptual Role of Number. In New Perspectives on Romance Languages, Nishida Chiyo & J. P. Montreuil (eds), 61–85. John Benjamins. Déprez V. 2007. Probing the structuring role of grammaticalization: Nominal Constituents in French-Lexifier Creole. Journal of Pidgin and Creole Languages. 22.2 Benjamins. Diesing, M. 1992. Indefinites. Cambridge MA: The MIT Press. Heycock, C. & Zamparelli, R. 2003. Coordinated bare definites. Linguistic Inquiry 34(3). Inverno, L. & de Sousa, D. 2001. O artigo definido em dois contextos especificos no portugues do Brasil e no Crioulo de Cabo Verde. Talk presented at ACBLPE. Coimbra, Portugual July 28/29. Krifka, M. 1995. Common nouns: A contrastive analysis of Chinese and English. In The Generic Book. G. Carlson & F. Pelletier (eds), 398–411. Chicago IL: University of Chicago Press. Longobardi, G. 1994. Reference and proper names: A theory of N-movement in syntax and Logical form. Linguistic Inquiry 25(4): 609–665. Longobardi, G. 1999. How comparative is semantics? Ms. University of Trieste. Lyons, C. 2000. Definiteness. Cambridge: CUP. Müller, A. 2001. Genericity and the denotation of common nouns in Brazilian Portuguese. Paper presented at SULA, April 20–22, 2001. Ms, University of Massachusetts at Amherst. Zribi-Hertz, A. and Glaude, H. “Bare NPs and deficient DPs in Haitian and French: From morphosyntax to referent construal.” This volume.

 

English-lexified Creoles

 

Bare nouns and articles in Sranan Adrienne Bruyn The paper investigates the distribution of determiners in Sranan, which differs both from that in the lexifier, English, and the Gbe substrate. From an early stage onwards, the article wan functions as a marker of indefinite singularity with nonspecific as well as specific indefinites; it may, however, also be absent. Similarly, the definite articles (n)a (SG) and den (PL), while exhibiting further grammaticalization over time, do not obligatorily appear with all definite nominals even in the modern language, and definite, pragmatically important arguments can appear as bare nouns. On the whole, bare nouns are open to several interpretations with respect to definiteness, specificity or referentiality, and number delimitation, precluding a zero article to assume a unique value. Even though specificity, or pragmatic referentiality, plays a role in the distribution of the articles, number and definiteness are the primary distinctions. The distribution of articles in Sranan does not conform to the proposal by Bickerton (e.g. 1981) and Givón (e.g. 1984b), nor can it be fully accounted for by assuming that articles express individuation while bare nouns typically refer to mass-like collectives, as suggested by Mufwene (e.g. 1986).

.

Introduction

It has often been noted that the use of articles in creole languages may differ substantially from that in the lexifier language.* Especially the absence of any determiner in environments where languages such as English would typically require one, has attracted attention. While some researchers regard the occurrence of such bare nouns as typical for pidgins, others acknowledge their presence in creole languages and try to discover the rules governing the distribution of so-called zero articles, or bare nouns. Thus, Bickerton (e.g. 1981) has argued that bare nouns in creole languages typically have non-specific reference, or that, if the referent has a specific identity, that identity

* This article reports on research supported by the Netherlands Organization for Scientific Research (NWO). Thanks are due to the editors for inviting me to contribute to this volume and for their useful comments on an earlier version of the article.

 Adrienne Bruyn

is not relevant. While speaking of referentiality rather than of specificity, Givón (e.g. 1984b) holds a similar view to that of Bickerton’s. Another approach has been taken by Mufwene (e.g. 1986), who considers bare nouns in Gullah and Jamaican Creole to be non-individuated, i.e. referring to a whole rather than to individual members of a set. In this paper, I will investigate the distribution of determiners in Sranan and discuss the factors determining their use and their absence, assessing the appropriateness of the various proposals. Sranan developed as a creole language in the latter part of the 17th century on the plantations of Suriname (‘Dutch Guyana’), which was an English colony from 1651 until 1667 before it came under Dutch rule in 1668. While it also contains lexemes derived from Dutch, Portuguese, and a number of African languages, the primary lexifier language is English. Among the English functional morphemes that were not retained in Sranan are, however, the articles the and a(n). Instead, Sranan has a singular definite article (n)a, a plural definite article den, and the form wan occurring with indefinite singular nominals — for example, a pikin ‘the child’, den pikin ‘the children’, wan pikin ‘a child’. There is no plural indefinite article, and, as plurality is not marked on the noun itself, plural indefinites appear as unmodified bare nouns. However, nouns that are not preceded by any determiner or other specifying element are open to various other readings as well. In fact, as will be argued below, bare nouns can receive such a wide range of interpretations that they cannot be considered to constitute a unified category with its own value along any dimension. While linguistic and extra-linguistic context may bring about a particular reading, by themselves bare nouns are neutral with respect to definiteness, specificity or referentiality, and number delimitation. Being morphosyntactically unmarked with regard to number and definiteness, they are open to every possible reading in principle. The fact that the possible readings include one as definite and individuated is one of several aspects in which Sranan does not conform to the proposals put forward by Bickerton, Givón and Mufwene to account for the distribution of bare nouns and articles in creole languages. These proposals, providing the background for the present investigation of article usage in Sranan, are introduced in Section 2. The wide range of potential readings of bare nouns in Sranan must be viewed in relation to the fact that the use of neither the indefinite singular article wan nor of the definite articles (n)a (SG) and den (PL) is generalized to the same extent as article usage in, for example, English. In Section 3.1, the distribution of wan is described, and it will be shown that wan has not grammaticalized towards a more generally used indefinite article over time. Its function is then discussed further in Section 3.2, where I will argue that wan is primarily a non-obligatory marker of singularity and indefiniteness. While this may perhaps sound rather trivial, it tends to be neglected in those approaches that assume that forms such as wan express specificity, referentiality, or individuation, and that wan contrasts with zero in these respects. However, even though individuation as well as referentiality or specificity play a greater role in Sranan than they do in English regarding the presence or absence of an article, the primary distinctions conveyed by the Sranan articles are number and definiteness.

Chapter 12. Bare nouns and articles in Sranan 

The definite articles and demonstratives and their diachronic development are dealt with in Section 4. Section 5.1 provides an overview of the contexts in which bare nouns may occur, while the topic of Section 5.2 is the variety of possible interpretations of bare nouns, and the implications for their analysis. A brief comparison of bare nouns in Sranan with those in the lexifier and substrate languages is presented in Section 5.3. Section 6 concludes the paper. The study is based on a variety of texts as well as secondary sources of various periods. As the discussion involves proposals by other scholars concerning a system alleged to be typical for creoles, it is relevant to include the earlier stages of the language, which are less affected by any developments that may have taken place over time. The earliest substantial Sranan source is van Dyk (ca1765), a language manual written for merchants, plantation owners and managers, artisans, and others who had business in Suriname. It includes some dialogues, and a playlet which gives an impression of the daily life on a plantation. The van Dyk text appears to be relatively spontaneous and close to vernacular speech, in particular in comparison with Weygandt (1798), which is partly an adaptation of van Dyk’s book.1 The two other 18th-century sources used for the present study are by Schumann, a Moravian brother. Schumann (1781) is a translation from German into Sranan of a Gospel harmony, a compilation of the four Gospel books of the New Testament. One of the purposes of this text was to be read during services. Schumann (1783) is a Sranan-German dictionary, containing hundreds of example sentences. The detailed information, both linguistic and other, provided in the dictionary, indicates that Schumann must have known the language quite well, even if he was not a native speaker. The larger part of the 19th-century text sources comprises writings by Johannes King. King, of maroon descent, joined the Moravian missionaries after having had visions in which he was ordered to turn to Christianity, and taught himself to read and write in Sranan. He started to preach the Gospel among maroon tribes as well as among the creoles in the Para district, and became known as ‘the prophet of the bushland’. He left a huge collection of diaries, reports, and letters, some of which have been published, including what is known as the Diaries of Johannes King, published in de Ziel (1973). This was written for church people as well as his own tribesmen. The other 19th-century sources used here are Kraag (1894–96), a text by this preacher and teacher in which he reports on his stay among the Kwinti in Copenkrisi; a text by Albitrouw describing a prophetic movement and its effects on the Saramaccan community (Albitrouw 1894–1915); and the dictionary by Focke (1855). For the 20th century, two main texts were investigated. One is the story collection contained in de Drie (1985). Most of the stories, compiled by the anthropologist Trudi Guda, are transcribed from broadcast recordings; others were recorded during

. See Bruyn (1995) for more information on the various sources and discussion of their interpretation.

 Adrienne Bruyn

fieldwork. Guda describes de Drie as a gifted story teller, who used a balanced mother tongue Sranan, combined with the stylistic techniques of the oral tradition. The stories have a variety of topics, including the time of slavery, and traditional Anansi stories. The other 20th-century text source is a selection from the Leysipisi (van der Hilst & Roest 1988), a Bible text translation passages of which are intended to be read in Mass on particular days of the catholic religious year. Like Schumann’s Gospel text, the Leysipisi is an original translation rather than an adapted version of earlier translations. Both texts can be assumed to reflect the language use of their respective periods, even though the style may differ from that of vernacular speech. For the quantitative investigations to be reported on below, text fragments were selected to form a core corpus consisting of 20,000 words for each period. For the 18th-century, it involves 10,000 words from Schumann (1781), and 5,000 each from van Dyk (ca1765) and Weygandt (1798); for the 19th-century, 10,000 words from King (1891–94), 5,000 each from Albitrouw (1894–1915) and Kraag (1894–96); and for the 20th century, 10,000 words each from de Drie (1985) and van der Hilst & Roest (1988).2 Before turning to the use of the Sranan determiners, I will now briefly present the proposals introduced by Bickerton, Givón and Mufwene regarding the distribution of bare nouns in creole languages.

.

Bare nouns and articles in creole languages

The fact that in many creole languages nouns may occur without any determiner has given rise to the suggestion that the distinctions that play a role in the determiner systems of creole languages are different from the distinctions in languages such as English or French. According to Bickerton (1981, 1984) the primary opposition is between

. The following references indicate the sources: vDyk = van Dyk (ca1765), SchGos = Schumann (1781), SchDict = Schumann (1783), Wey = Weygandt (1798), Focke = Focke (1855), King = King (1891–94), Albi = Albitrouw (1894–1915), dDrie = de Drie (1985), Leysi = van der Hilst & Roest (1988). The spelling of the Sranan examples follows that of the sources. The glosses and translations of example sentences are based on the context of the cited sentences. With regard to bare nouns this implies that some translations provide an interpretation that is more specified than the bare noun by itself. Abbreviations used: (NON)SPEC = (non)specific, (NON)REF = (non)referential, (NON)INDIV = (non)individuated, (IN)DEF = (in)definite, DEM = demonstrative, SG = singular, PL = plural, 1/2/3 = 1st/2nd/3rd person, PST = past tense, CNT = continuative aspect, FUT = future, NEG = negation, LOC = locative, Ø = absence of determiner (indicated with nouns under discussion only). For reasons of economy and space, the article wan is glossed as INDEF rather than INDEF.SG, and na as DEF rather than DEF.SG in most cases.

Chapter 12. Bare nouns and articles in Sranan 

specific (SPEC) and non-specific (NONSPEC) in the typical creole language. This would contrast with, for example, English, where the distribution of articles is determined by definiteness, i.e. whether the speaker assumes that the hearer can assign unique reference. The English indefinite (INDEF) article a appears regardless of whether an NP has SPEC reference, as a book in (1a), or NONSPEC, as in (b), the difference between the two readings becoming apparent when the book is pronominalized. (1)

a. b.

She was looking for a book (but she couldn’t find it). She was looking for a book (but she couldn’t find one / any).

Bickerton argues that the article system that emerges during creolization typically distinguishes between three types of NP: the definite (DEF) article is used with NPs that have SPEC reference and can be assumed to be known to the listener, the INDE article with NPs that have SPEC reference and are unknown to the listener, and no article appears with NONSPEC NPs, i.e. generic NPs, NPs in the scope of negation, and NPs with referents whose exact identity is either unknown to the speaker or irrelevant to the point at issue (Bickerton 1981: 23). It is not difficult to find instances of bare nouns in Sranan that can be accounted for along these lines. For example, the noun ijskasi ‘fridge’ in example (2a) clearly refers to fridges as a generic type rather than to one or more particular members of that type. (2)

a.

b.

ijskasi no ben de a ten dati [C20-dDrie29] ø fridge NEG PST be DEF time that ‘(a/the) fridge(s) didn’t exist at the time’ Ne a seri a pernasi, a kon bay pernasi then 3SG sell DEF plantation 3SG come buy ø plantation a Para [C20-dDrie50] at Para ‘then he sold the plantation, and bought (a) plantation(s) in the Para district’

With respect to the bare noun pernasi ‘plantation’ in (2b), one may conclude that in spite of the fact that it has a specific identity it may appear without article, because, apparently, its identity is irrelevant to the point at issue.3 In particular with isolated sentences, it can be difficult to establish this type of non-specificity on the basis of anything other than the presence or absence of an article. The notion of pragmatic referentiality as introduced by Givón provides a useful perspective, however. While Givón (1981, 1984a,b, 2001) uses the term ‘referentiality’ rather than ‘specificity’, his view of the distinctions expressed by articles in creole languages is actually the same as Bickerton’s. As his treatment of the supposedly typical creole system is embedded in a broader discussion of the expression of definiteness and referentiality, including the

. The referent of the noun pernasi in (2b) can be assumed to have a specific identity since it occurs in a past tense context; see below.

 Adrienne Bruyn

implications of certain contexts with regard to referentiality, I will take his framework as a starting point, and follow his terminology. Building on Bickerton’s proposals regarding Hawaiian Creole English and other creoles, and having found supporting evidence in some stories in Krio (English creole of Sierra Leone), Givón concludes that, in creole languages, bare nouns are nominal arguments that are not pragmatically referential. Pragmatic referentiality has to be understood as ‘meant by the speaker to have important reference in the universe of discourse’. Often this corresponds to semantic referentiality, i.e. supposed existence in the universe of discourse. DEF NPs are always referential (REF) in the semantic sense. INDEF NPs are always semantically REF if they occur in a factive environment, i.e. in the modalities of presupposition, as in (3a), and of realis-assertion, for example past tense, as in (3b). (3a) and (b) both imply that Mary saw a particular eagle, whose existence the speaker is committed to assert. The opposite is true with generic expressions, which can only be non-referential (NONREF). Another type of NP that is generally NONREF are INDEF NPs in the scope of negation. (3)

a. b. c. d.

complement of factive V: Mary regretted that she saw an eagle realis-assertion, PAST: Mary saw an eagle generic predicate: Mary is a good bird-watcher non-factive, NEG: Mary didn’t see an eagle

INDEF-REF INDEF-REF

NONREF

NONREF

In non-factive environments other than negation, on the other hand, INDEF NPs can be either REF or NONREF. This is the case, for example, when a NP occurs within the scope of future tense, or as the object of a non-implicative verb such as ‘to look for’. Thus, in example (1) above, a book in (a), which is subsequently referred to by it, is INDEF-REF, whereas a book in (b), which may be pronominalized by one or any, is NONREF. Whereas English a(n) is opaque with regard to referentiality, there are languages where INDEF-REF and NONREF NPs are coded distinctively. Givón states that in a variety of languages, including Mandarin Chinese, Turkish, spoken Hebrew, and creoles, the numeral ‘one’ marks INDEF-REF arguments, while NONREF arguments are commonly left unmarked. Although the referentiality of NPs coded in this way can often be conceived as semantic referentiality, i.e. existing, it is the pragmatic referentiality that eventually determines how a NP is coded in cases where the two types of referentiality do not overlap. Thus, a NP may be semantically NONREF and nevertheless contain ‘one’ if it is of great thematic importance, which is normally reflected by frequent recurrence in the subsequent discourse (cf. the notion of ‘cataphoric topicality’, Givón 2001: 455). Conversely, a semantically REF NP may occur as a bare noun if it is unimportant and nonrecurrent. Givón’s pragmatic referentiality thus corresponds to Bickerton’s specificity, but it is more explicit with regard to possible implications of the contextual modality,

Chapter 12. Bare nouns and articles in Sranan 

and with regard to the question of how an argument’s importance for the discourse may be measured. Another approach is taken by Mufwene (1981, 1984, 1986), who argues that bare nouns are non-individuated, i.e. referring to an undelimited, mass-like collective rather than to individual members of a set. The notion of individuation is comparable to the count–mass distinction, the difference being that count versus mass can be conceived as a lexical feature whereas individuation does not pertain to a certain noun but to the use of a noun in a certain context. While individuation is universally relevant semantically, languages may differ with respect to its morpho-syntactic expression. In English, a few nouns, such as cake, can be either [+count] or [−count], and can occur in an individuated sense — e.g. How many cakes? — or in a non-individuated one — e.g. How much cake? According to Mufwene, in certain languages, including the basilectal varieties of Gullah and Jamaican, any noun can be used in either sense. Non-individuated (NONINDIV) nouns appear as bare nouns, whereas articles and plural marking appear if a noun is meant as individuated (INDIV). In so far as we do not have access to the intentions of the speaker, it is not always easy to establish whether a NP is INDIV or not on criteria other than the presence of an article. There are however many cases of bare nouns in Sranan that can evidently be accounted for by presuming that they are NONINDIV indeed. In (2a) above, for example, ijskasi ‘fridge’ clearly does not refer to one or more individual fridges. Below it will be shown that, even though many bare nouns in Sranan are pragmatically NONREF, or NONINDIV, or both, neither referentiality nor individuation can provide a full account of article usage in Sranan. Before focusing on bare nouns in Section 5, I will first explore the occurrence of overt articles in Sranan versus that of bare nouns, and discuss to what extent this accords with the proposals just sketched.

. .

The use of the article wan The distribution of the article wan at various stages of Sranan

The English article a(n) did not survive in Sranan. Instead, some, but not all, INDEF-SG nouns are accompanied by wan, which is derived from English one and is also the numeral ‘one’ in Sranan as well. Cross-linguistically it is of course quite common to find the lowest numeral marking INDEF-SG nouns (e.g. Lyons 1999). Givón (1981) argues that this is so because ‘one’ is uniquely fit to perform the complex task of introducing a new argument into discourse, involving its presentation as existing in the universe of discourse, i.e. referential, and at the same time making it possible for the hearer to identify it as belonging to a certain type by its connotational or generic properties. As a quantifier, ‘one’ implies referentiality, but not definiteness. Being the lowest numeral, ‘one’ can be used to refer to ‘one of many’, ‘one of the group’, or ‘one of the type’ at the same time. Although the functional extension of ‘one’ towards INDEF article

 Adrienne Bruyn

is a universal one, languages may differ in the extent to which this development has proceeded. Givón (1981) postulates a developmental continuum, the starting point of which is a situation in which no systematic distinction is made between INDEF-REF and NONREF NPs. This may be the case in a rudimentary pidgin where article usage is unpredictable, as in Hawaiian Pidgin English (cf. Bickerton 1981). In the first stage of the development, manifested by the typical creole language according to Givón and Bickerton, ‘one’ marks pragmatically REF NPs only. While Spanish, for example, appears to occupy a later, intermediate position on the continuum by using ‘one’ with some but not all NONREF arguments, English a(n) has reached an advanced stage: a phonologically reduced form of the numeral has extended in scope to occur with by and large all INDEF-SG NPs, regardless of whether they are REF or not. In the following, I will discuss to what extent the distribution of wan in Sranan conforms to the developmental stage considered typical for creole languages, where the INDEF article is assumed to appear only with NPs whose referents are unknown to the listener and pragmatically REF. I will show that the use of Sranan wan does not consistently correspond to such a stage, notwithstanding the fact that there are many instances of wan that can be interpreted as being in accordance with it. In Sranan, new and thematically important participants are typically introduced by using wan. In the following examples the referents presented by means of a wan-NP are central to the story and are subsequently referred to by DEF nominals, in the form of the anaphoric pronoun hem in (4a), and involving the DEF article a in (b). (4)

a.

b.

Lukku, wan uman ben de dapeh, dissi habi wan jeje look INDEF woman PST be there who have INDEF spirit vo siki sinsi tin na aiti jari; hem hele skin bendi, of illness since ten at eight year 3SG whole body bend a no kann hoppo hem fesi. […] [C18-SchGos176] 3SG NEG can raise 3SG face ‘Look, a woman was there, who had a spirit of infirmity for eighteen years; her whole body was bowed, she couldn’t raise her face. […]’ Wan wedwe ben e libi na a foto disi tu. INDEF widow PST CNT live at DEF town this too Ala dey a wedwe […] [C20-Leysi976] all day DEF widow ‘A widow was living in this town too. Every day the widow […]’

The wan-NPs in these examples are REF in the pragmatic as well as in the semantic sense, and are thus in accordance with the typical creole system proposed by Givón and Bickerton. At the same time however, there are cases of wan occurring with semantically NONREF arguments, such as generic predicates connoting a type rather than denoting a particular individual, as in (5a) and (c), and NPs in the scope of negation, as in (5b) and (c).

Chapter 12. Bare nouns and articles in Sranan 

(5)

a.

b.

c.

Morgudan ben de wan sukrupernasi. Morgudan PST be INDEF sugar-plantation ‘Morgudan was a sugar plantation.’ ma no wan dey yu no gi mi but NEG-one day 2SG NEG give 1SG wan pikin bokoboko INDEF little billy.goat ‘but never did you give me a little billy goat’ dan a no kan de wan disipri fu mi then 3SG NEG can be INDEF disciple of 1SG ‘then he cannot be a disciple of mine.’

[C20-dDrie33]

[C20-Leysi953]

[C20-Leysi945]

The use of wan in such cases does not correspond to the first stage of the developmental continuum. As the examples in (5) are from late-20th-century Sranan texts however, the occurrence of wan-NPs in these contexts in general could be interpreted as an indication that wan conformed to the alleged creole system in earlier stages of the language but has developed towards a less restricted INDEF article by the late-20th century. However, similar cases of NONREF NPs involving wan occur in the 18th-century Sranan sources. Example (6a) involves a generic predicate, (6b) negation, and (6c) both a generic predicate and negation: (6)

a.

b.

c.

da wan krommatie negere it.be INDEF Koromanti negro ‘it (he) is a Koromanti’ tog ju no ben gi mi wan tem wan skaap but 2SG NEG PST give 1SG one-time INDEF sheep ‘but you did not once give me a sheep’ a no kann komm wan Discipel va mi 3SG NEG can become INDEF disciple of 1SG ‘he cannot become a disciple of mine’

[C18-vDyk68–9]

[C18-SchGos185]

[C18-SchGos182]

Cases like these suggest that perhaps the distinction between REF and NONREF indefinites was and is not coded in Sranan as consistently as in the alleged creole system. If wan had developed in the direction of a less restricted INDEF article over time, this would have to be reflected in an increase in its overall frequency as well as in certain distributional shifts. Taking Givón’s proposal as a starting point, it would be expected that in the course of time wan will occur relatively more often with NPs in contexts that entail semantic non-referentiality as well as with NPs that are pragmatically unimportant. In the case of English a(n), an overall increase in frequency as well as a relative increase with NPs with a low discourse referentiality are indeed observed by Hopper & Martin (1987). This form began as a numeral in Old English, then became a pragmatically strong presentative, and eventually an unrestricted INDEF article. There is no evidence, however, that a development in the same direction has taken place in Sranan.

 Adrienne Bruyn

Although there are more environments that can be taken to imply semantic nonreferentiality of INDEF NPs, the most important and the most clearly identified are those where the NP is a generic predicate, or an element in the scope of negation, or in that of leki ‘like’ (cf. e.g. Wright & Givón 1987). Taking the use of wan in these three NONREF contexts as one of the gauges for establishing the status of wan as a marker of referentiality, it appears that its referential value has not weakened over time. There is no overall increase in the relative frequency of this type of NONREF wan-NP in the Sranan texts portions of 20,000 words: it starts with 20% in the 18th century (57 out of 292 cases of wan-NP), then raises to 32% in the 19th century (74 of 228), then drops back again to 19% in the 20th century (58 of 299).4 As can be seen in Table 1 below, the absence of a unidirectional diachronic development with respect to semantically NONREF wan-NPs is paralleled by an absence of development of NPs that have low pragmatic referentiality. Sometimes wan marks a NP that is semantically REF but can be regarded as being thematically unimportant, to the extent that no further reference is made to it in the passage following the first mention, as in cases such as the following: (7)

a.

b.

Kaba a go nanga dem na sei, na tappo vo and 3SG go with 3PL at side at top of wan heh bergi, va begi. [C18-SchGos150] INDEF high mountain for pray ‘And he went with them apart by themselves, up onto a high mountain, to pray.’ Dan den nengre e bron wan korfayapatu then DEF.PL negro CNT burn INDEF charcoal-pot heri neti. [C20-dDrie50] whole night ‘Then the slaves burnt a charcoal stove the whole night.’

In order to measure pragmatic saliency, it can be operationalized by looking at the subsequent mention of a referent.5 Contrary to what could be expected if wan had become a more general INDEF marker in the course of time, there is no increase in the

. See Bruyn (1995) for details on the categorization of wan-NPs, and for an explanation for the minor variations in the frequencies in terms of text types. . The issue here is whether the referent is subsequently mentioned or not, rather than the number of subsequent mentionings (cf. Givón 1984b, 2001, Wright & Givón 1987, Hopper & Martin 1987). A NP was regarded as non-recurrent if no coreferential argument occurs within the first 20 clauses after its appearance (cf. Givón 1984b). In the Sranan texts most coreferential NPs appear within the next two clauses, and the picture would be no different if a limit of 10, or 30, clauses were applied. Subsequent mention was interpreted quite broadly; see Bruyn (1995) for details.

Chapter 12. Bare nouns and articles in Sranan 

Table 1. The frequency of wan-NPs in Sranan Semantically nonref 20,000-words text portions

wan-NP TOTAL

%

18th century 19th century 20th century

292 228 299

20% 32% 19%

TOTAL

819

23%

N

Non-recurrent %

N

57 74 58

43% 46% 36%

125 105 109

189

41%

339

frequency of non-recurrent wan-NPs over the three centuries: from 43% in the 18thcentury text portions, to 46% in the 19th-century, and 36% in the 20th-century ones. The quantitative study of the distribution of wan in the three periods as summarized in Table 1 leads to the conclusion that wan did not become a more grammaticalized INDEF article over the course of the 18th, 19th, and 20th century. There is no extension in use of wan in general, nor has there been a consistent rise of semantically NONREF wan-NPs or of wan-NPs that are non-recurrent and in that sense pragmatically unimportant.6 Although it is theoretically possible that, apart from stabilizing or levelling of the earliest creole varieties, there have been changes between the formative period for Sranan and the latter part of the 18th century, this is unlikely in view of the fact that no change is attested in the sources ranging from circa 1765 to 1988.7 Any substantial development in the value of wan that might have been going on before the latter part of the 18th century would have to have suddenly stopped. Therefore, with respect to wan, the article system as it emerges from the sources investigated may be regarded as not very different from the original Sranan system.

. With respect to English an, Hopper & Martin (1987) observe an increasing possibility for an-NPs to occur as subject/topic, reflecting the declining role of an as a presentative marker as part of the development towards general INDEF article. In the case of Sranan this is not a viable diagnostic, however, since the correlation between subjects and topics appears to be weaker anyway. Whereas, in English, non-topical subjects tend not to appear in subject position but rather as in Once there was a …, or are avoided by passivization, this does not apply similarly to Sranan. It is not unusual for new arguments to be introduced in the clause-initial subject position — see for example (4a) and (b) above. . In the Herlein (1718) fragment, wan appears only once, namely in wantem ‘one/some time, once’, which might be regarded as a more or less fixed phrase. At the same time, the text, being quite short and partly consisting of isolated phrases, contains very few NPs with INDEF-SG reference that could have been marked with wan. Together with the fact that wan is optional in later stages of the language, this makes it impossible to determine whether the use of wan in Herlein’s language variety differs from that in later varieties.

 Adrienne Bruyn

.

The use of the article wan versus its absence

While it is clear that wan has not developed towards a less restricted INDEF article in the course of time and that its value with regard to referentiality remained stable, it remains to be established what function the article wan fulfils and what determines its use versus its absence. It appears not to belong properly in the first stage of the developmental continuum sketched above, and thus not to correspond exactly to Givón’s and Bickerton’s typical creole system. For this to be the case, there would have to be lower figures for semantically NONREF as well as for non-recurrent wan-NPs. Of the 189 semantically NONREF wan-NPs, 32% are recurrent in the sense outlined above and can thus be regarded as pragmatically salient, which could account for the use of wan in these cases. That leaves 68%, 129 of the total of 189 NONREF cases, not predicted by Givón and Bickerton. The rather high frequency of non-recurrent wan-NPs in general, with a total average of 41% (see Table 1 above), appears to be even more problematic for the proposal that wan marks pragmatically REF NPs only. Even if one might be able to explain away some of the cases that I categorized as pragmatically NONREF, there are still quite a robust number of wan-NPs that are not in accordance with the typical creole system. In this section I will show that, although referentiality plays a role and can account for many occurrences of wan, the proposal of Givón and Bickerton is inadequate in so far as it fails to acknowledge its function of marking singularity and indefiniteness. And while Mufwene’s suggestion that individuation is the distinctive factor gives due attention to number marking in relation with individuation, he does not take account of definiteness distinctions either. In order to determine what is expressed by wan in Sranan it is worth looking somewhat more closely at the NPs containing wan where this would not be predicted on the basis of the typical creole system as proposed by Givón and Bickerton. The category of non-recurrent wan-NPs consists of three types, distributed more or less evenly: INDEF NPs in contexts that imply semantic non-referentiality, those that are semantically REF since they appear in the context of past or present tense, and those that appear in non-factive contexts and are undetermined with regard to semantic referentiality. Semantically NONREF wan-NPs were illustrated with Examples (5) and (6) above, which involve non-recurrent referents. It may be possible to provide an account for the use of wan in particular cases. Although wan-NPs in the scope of negation are not included in the calculations when it is evident that wan serves to intensify the negation, emphasis may sometimes be taken to account for the presence of wan. With generic predicates, one might try to explain the occurrence of wan in certain cases by pointing out that the parsing would become problematic without wan, considering that (n)a or da can be a DEF article (see e.g. da direkture in (9a) below) as well as a copula (e.g. (8a)), or a sentence-introducing copular element or cleft marker (e.g. (8b); cf. Arends 1995b: 25). Because nominal predicates may occur without any copula, da could be interpreted as a DEF article rather than a copula in sentences such as (8a) if wan were omitted.

Chapter 12. Bare nouns and articles in Sranan 

In (8b), the absence of wan could theoretically give rise to a reading as a phrase, ‘the Koromanti negro’, instead of a copular sentence (cf. Arends 1995b: 25). (8)

a.

b.

hem da wan Samariasomma 3SG be INDEF Samaria-person ‘he was a Samaritan’ da wan krommatie negere it.be INDEF Koromanti negro ‘it’s a Koromanti’

[C18-SchGos190]

[C18-vDyk68] =(6a)

However, apart from the general question of to what extent potential parsing problems can be regarded as explanatory, they cannot fully account for the use of wan in sentences like (8a) and (b) in any case, considering that there are also cases such as those in (9). Although here da and na should not be interpreted as DEF articles either, ron negere and fransman appear as bare nouns, without wan to ensure the intended interpretation: (9)

a.

b.

bekassi da direkture a za kili mi lange gon because DEF manager 3SG FUT kill 1SG with g un liki mi da ron negere [C18-vDyk89] like 1SG be ø run negro ‘because the manager would kill me with his gun as if I were a runaway slave’ bika na fransman [C20-dDrie29] because it.be ø Frenchman ‘because it’s a Frenchman’

Comparing (8a) and (9a) however, the difference regarding the use of wan may be explained by assuming that while both NPs are semantically NONREF as well as nonrecurrent, wan Samariasomma ‘a Samaritan’ in (8a) occurs in a factive context and is somehow more referential than ron negere ‘runaway negro’ in (9a), which is in the scope of an irrealis modality. In this way the use of wan in the former and of a bare noun in the latter may be viewed as compatible with the proposal of Givón, who considers pragmatic referentiality to be a scalar rather than an absolute or binary property. Pragmatic importance may account for the use of wan in (8b) versus its absence in (9b). In view of the respective contexts, it is not unreasonable to assume that in (8b) the fact that the person being talked about is a Koromanti is more important than the fact that the person at issue in (9b) is a Frenchman. The implication of the remark in (8b) that the man in question is a Koromanti is that he is likely to commit suicide when his honour is injured — which, later in the play, indeed happens. That the person referred to in (9b) is a Frenchman is relevant to what is being said, but may be viewed as less important, and, in any case, has already been mentioned before. An alternative account would be to assume, along the lines of Mufwene (e.g. 1986), that wan krommatie negere ‘a Koromanti negro’ is meant as INDIV, while fransman ‘Frenchman’ is not. In this case, this is difficult to establish on the basis of criteria other than the presence of wan. There are cases, however, where it seems rather obvious that

 Adrienne Bruyn

individuation is involved. Consider the following example, also involving a wan-NP which is semantically NONREF and non-recurrent as well: (10)

effi une ha bribi leki wan Mostertsiri, if 2PL have belief like INDEF mustard-seed kaba une takki na dissi boom […] [C18-SchGos189] and 2PL say at this tree ‘If you had faith as a (grain of) mustard seed, and you would say to this tree […]’

The NP wan Mostertsiri ‘a mustard seed’ in (10) is a clear case of an INDIV NP: it does not refer to ‘mustard seed’ as a mass noun or to an undelimited amount of mustard seeds, but to a grain of mustard seed. In general, there are quite a few cases where the use of wan with a semantically NONREF and non-recurrent NP can be understood by assuming that what matters is the fact that the NP refers to an individual entity, rather than whether it is REF or not. The two other categories of non-recurrent wan-NPs occurring in the Sranan texts are NPs that are semantically REF by virtue of the fact that they occur in factive contexts, i.e. in the scope of past or present tense, and NPs in non-factive contexts, which are opaque with respect to semantic referentiality. According to Givón (1981, 1984a, 2001), wan would disambiguate in non-factive contexts by marking REF arguments. Non-factive or irrealis contexts include imperatives and questions; future or habitual tense; the scope of non-implicative verbs such as ‘to look for’ or ‘to want’; the scope of probabilistic verbs or adverbs such as ‘can’, ‘must’, or ‘maybe’; and hypothetical conditionals. The examples in (11) involve a question, a future, and a hypothetical context respectively. In these cases, wan may indeed be considered to convey pragmatic referentiality, albeit of a special kind. (11)

a.

b.

c.

Joe mekie wan oloysie saka toe na da broekoe? 2SG make INDEF watch pocket too at DEF trousers ‘Have you also made a watch pocket on the trousers?’ Èn na tori sa skrifi go moro fara and DEF story FUT write go more far na ini wan tra boekoe. at in INDEF other book ‘And the story will be written further in another book.’ effi hem kau effi hasi fadomm if/when 3SG cow or horse fall na inni wan pütt at in INDEF pit ‘when his cow or horse falls into a pit’

[C18-Wey111]

[C19-King125]

[C18-SchGos179]

Although the wan-NPs in these examples are non-recurrent, those in (11a) and (b) can be regarded as thematically important. At the same time, the use of wan may be taken to suggest some kind of existence in the mind of the speaker. With regard to wan oloysie

Chapter 12. Bare nouns and articles in Sranan 

saka ‘a watch pocket’ in (11a) the utterance is not merely an informative question, but conveys that the speaker expects the addressee, a tailor, to have made a watch pocket on the trousers. In (11b) it may be inferred that the book mentioned does exist, that it has already been written or at least is in the course of being written. (11c) is an example of a case where a wan-NP does not appear to have much pragmatic importance. The effect of the use of wan in such cases is rather that an ‘as-if-it-were’ factive situation is created, that a hypothetical situation is made more concrete in the universe of discourse. This may be subsumed under pragmatic referentiality in a broad sense: it does not pertain to the particular referent at issue but rather to the entire situation being described. In the following examples, the wan-NPs are implicitly in past tense, therefore in factive environments. Although they are non-recurrent and the identities of their referents as such unimportant, they present arguments that are relevant to the setting of the respective stories. (12)

a.

b.

c.

Kaba teh a passa wan woijewoje plesi, place and when 3SG pass INDEF market a miti tin melaatse mannsomma [C18-SchGos190] 3SG meet ten leprous man-person ‘And when he came past a market place, he met ten (male) lepers’ Wan neti den go kari basya Ananka [C20-dDrie40] INDEF night 3PL go call basya Ananka ‘One/some night / on a night, they went to call basya (overseer) Ananka’ den doro na wan pranasi [C20-Leysi913] 3PL arrive at INDEF plantation ‘they arrived at a plantation (settlement)’

The question arises, however, whether it is some kind of pragmatic referentiality that determines the use of wan in these examples, or rather INDEF-SG individuation. The fact that factive contexts are involved not only implies that the NPs themselves are semantically REF but also that there is no need to create an as-if-it-were-factive situation. Thematic salience of the referents at issue cannot account for these occurrences of wan either. When Givón and Bickerton focus on referentiality as the decisive factor with respect to the use of wan, what tends to be neglected is the fact that wan expresses singularity and indefiniteness. In his discussion of the use of wan versus zero in a Krio story, Givón (1984b: 133) accounts for the unexpected use of wan in wan oda vileyj ‘another village’ by pointing out that in this case there is no alternative since oda vileyj alone would necessarily be plural and mean ‘other villages’. Bare nouns, however, are not necessarily plural by themselves, either in Sranan or in Krio, and considering them as necessarily plural conflicts with Givón’s own contention that the use of wan rather than zero is determined primarily by referentiality. Although there is a suggestion that the use of wan in this particular case is somehow connected to oda ‘other’, it is difficult to see how this account can be reconciled with the idea that wan and other articles only mark pragmatically REF NPs while NONREF NPs appear as bare nouns and therefore cannot be

 Adrienne Bruyn

marked for number nor definiteness.8 However, Givón’s account of the Krio case wan oda vileyj ‘another village’ is, while problematic for his own proposal, illuminating in that it indicates that there may be NONREF NPs that need to be marked for singularity, or for singular indefiniteness. There appears to be no reason why it would be generally impossible for NONREF arguments to be construed as individual entities and to be marked for number accordingly (cf. Givón 1984a: 389). If the issue is approached from the perspective of zero-marked nouns it must be concluded that the system proposed by Givón cannot be fully adequate for any language which lacks number marking on the noun. I would like to argue that in Sranan, in any case, the marking of singular indefiniteness is the core function of wan, and that referentiality is not the primary distinction determining the use of wan versus zero even though wan occurs more often with pragmatically salient participants. Certainly, in cases where this is not predicted on the basis of the system alleged to be typical for creoles, the occurrence of wan in Sranan becomes intelligible when it is assumed that the function of wan is to make singularity and indefiniteness explicit, regardless of referentiality. The need to signal INDEF-SG individuation may vary. While the speaker’s intention constitutes the starting point, in some contexts an explicit marking is more appropriate whereas in other contexts a bare noun suffices. Although this will correspond to pragmatic referentiality more often than not, this is not necessarily the case. In the examples in (12) above, wan cannot be omitted without consequences. The omission of wan from wan neti in (12b) would result in such an interpretation as ‘at night’, if it can be interpreted at all. This indicates that wan is needed here to convey the meaning ‘one/some night’, ‘on a night’, even though the specific identity of that night is incidental to the subsequent discourse. In the cases of (12a) and (c) the omission of wan would also result in sentences that may receive different readings than those intended. More specifically, the function of wan in these cases appears to be to rule out a DEF interpretation of the bare noun. The possibility that bare nouns may have DEF reference is not considered by Givón and Bickerton, nor by Mufwene. In Sranan, however, arguments playing a central role in the story or otherwise familiar to the hearer can appear as bare nouns. In Examples (13a) and (b), we have bare nouns whose referents are individual entities that are DEF as well as thematically important:9 (13)

a.

gran mastra ben doore na pranasie great master PST arrive at ø plantation ‘the owner has arrived at the plantation’

[C18-vDyk94]

. In Sranan in any case, nouns with singular reference are occasionally preceded by an adjective without there being an article; see for example (24b) below. It may be remarked here that in the Krio stories discussed in Givón (1984b), many of the NPs that are semantically NONREF as well as pragmatically unimportant have either an INDEF-PL or a mass reading, which renders the use of wan inappropriate anyway. . Notice that na in (13a) must be the preposition rather than the DEF article, since in van Dyk (ca1765), as in other early sources, the DEF-SG article always has the form da.

Chapter 12. Bare nouns and articles in Sranan 

b.

watra e kon nini pernasi water CNT come at-in ø plantation ‘(the) water was flooding into the plantation’

[C20-dDrie40]

In Section 5.2 the implications of such cases for the categorization of bare nouns will be discussed further. What is important at this point is that the occurrence of DEF bare nouns bears on the distribution of wan. When we compare (13a) and (b) with (14), the role of wan in the latter becomes evident: it indicates that a new argument is involved — in other words, it disambiguates with regard to indefiniteness. (14)

den doro na wan pranasi 3PL arrive at INDEF plantation ‘they arrived at a plantation (settlement)’

[C20-Leysi913] =(12c)

In other cases, it is singularity that matters, or singularity and indefiniteness. Compare (15a) and (b), both also in factive past tense contexts: (15)

a.

b.

wan mati fu mi di de na waka INDEF friend of 1SG that be at walk kon sribi na mi. [C20-Leysi918] come sleep at 1SG ‘a friend of mine who was travelling around came to sleep at my place.’ Bikasi bakra ben kon na mi, because Ø white(person) PST come at 1SG mi taki gi den […] [C19-King83] 1SG say to 3PL ‘Because (some) whites came to me, I said to them […]’

Without wan, mati fu mi in (15a) would be open to various interpretations, including INDEF-PL, as with the bare noun bakra in (15b), where the plural pronoun den in the next clause indicates that bakra refers to more than one white person. Since wan mati fu mi and bakra are both thematically salient, the presence of wan does not appear to correspond to a higher degree of referentiality in this case.10 That individuation rather than referentiality should be regarded as accounting for the use of wan in certain cases is also evident from the following examples of NPs occurring in non-factive contexts — imperative in (16), non-implicative in (17): (16)

a.

b.

Joe buy gi den bassia wan zopi 2SG boy give DEF.PL basya INDEF rum ‘You boy, give the basyas (overseers) a (glass of) rum’ Bassia gi zomma zopie basya give person Ø rum ‘Basya, give the people rum’

[C18-vDyk68]

[C18-vDyk108]

. Notice that the use of wan is not implied by the presence of a possessive fu-clause: fu-clauses can follow a bare noun, as in mati fu yu ‘(a) friend(s) of yours’ (cf. (25) below).

 Adrienne Bruyn

(17)

a. b.

Dan mi go suku wan wroko dape. then 1SG go look.for INDEF work there Dan te mi suku wroko, dan […] [C20-dDrie176] then when 1SG look.for ø work then ‘Then I’ll go look for a job there. And while I’m looking for work […]’

While the contexts are similarly opaque with regard to semantic referentiality, the occurrence of wan in (16a) and (17a) does not seem to make the arguments in any sense more referential than the bare nouns in the (b)-sentences. Cases such as these suggest that in some cases it is impossible to relate the occurrence of wan to anything other than the intentions of the speaker with regard to individuation. Were wan omitted from the NPs in the (a)-sentences, the resulting bare nouns would be open to a mass or INDEF-PL reading, and perhaps — though less likely in (17b) in any case — to a  reading. Apparently the respective speakers intended INDEF-SG reference and chose to express this explicitly. Conversely, with the (b)-sentences it must be assumed that no need was felt to mark for number or definiteness. The conclusion must be that Sranan wan is primarily an INDEF-SG article occurring optionally with unknown referents rather than a marker of unknown referents that have pragmatic importance. It thus appears to classify as what Lyons (1999) calls a ‘quasi-indefinite’ article: being segmentally identical to the lowest numeral, it expresses singularity directly while indefiniteness is signalled indirectly. However, in contrast to articles encoding by and large all INDEF-SG nominals — such as English a — Sranan wan is not always obligatory. In other words, to the extent that it does not automatically mark every NP with INDEF-SG reference, it is not fully grammaticalized as an article. A major factor influencing the use of wan is pragmatic referentiality, or specificity in the sense of having a specific identity which is relevant to the discourse. With pragmatically salient referents, the urge to mark explicitly an NP as INDEF-SG will generally be greater (cf. Givón 1984b: 124, Wright & Givón 1987: 29), while, conversely, with referents whose identity is unimportant to the discourse the status regarding number and definiteness may more often remain unspecified. While the focus in this section has been on instances of wan that are not predicted on the basis of the typical creole system as proposed by Givón and Bickerton, there are many cases of wan-NPs that do accord with that system as well (see Example (4) in the previous section, and (34b) below), and it is evident that the use of wan can contribute to a coherent discourse structure. However, since cases that do not correspond to the system based on referentiality, as well as those that do, can be accounted for by regarding wan as an optional INDEF-SG marker, it may be argued that the correlation between the occurrence of wan on the one hand and referentiality and discourse coherence on the other in fact derives from the function of wan as a marker of singular indefiniteness. In this respect then, Sranan does not directly conform to Givón’s (e.g. 1984b) and Bickerton’s (e.g. 1981) proposal for a typical creole system with referentiality or specificity as the primary distinction.

Chapter 12. Bare nouns and articles in Sranan 

To a certain extent, the case of Sranan wan corroborates Mufwene’s proposal that intended individuation determines the use of articles. Although referentiality and individuation tend to go hand in hand in general, singular individuation can sometimes account for the use of wan where it would not be predicted on the basis of referentiality. The fact that Mufwene takes no account of definiteness distinctions, however, weakens his approach. Another problem is posed by the assumption that all nouns occurring without article or number marking are NONINDIV. The fact that bare nouns in Sranan are a less unified category than suggested by both Mufwene and Givón and Bickerton will be discussed further in Section 5.2. First, however, I will describe the situation with respect to the Sranan definite articles and demonstratives.

.

Definite articles and demonstratives

Nominals whose identity is assumed to be known to the hearer sometimes appear as bare nouns in Sranan, an issue to which I will return in Section 5. The present section is concerned with the DEF articles and demonstratives as they occur in early and presentday Sranan. In modern Sranan, the DEF articles are singular na or a, and plural den. Occupying a pre-nominal position, they co-occur with the demonstratives disi ‘this, these’ or dati ‘that, those’, which normally follow the noun when used attributively. (18)

a.

a

pikin child ‘the child’ a pikin disi / dati DEF.SG child this / that ‘this / that child’ DEF.SG

c.

b. den pikin DEF.PL child ‘the children’ d. den pikin disi / dati DEF.PL child this / that ‘these / those children’

Occasionally, disi or the short form di(s) occurs before the noun; in such cases, there is no article. The pre-nominal use of di(si) reflects on older pattern and is largely restricted to more or less lexicalized phrases such as di neti ‘this night, tonight’. Although some present-day speakers, probably under the influence of Dutch, use dati as well as disi before the noun more generally, this is not the general pattern (cf. e.g. Voorhoeve 1953: 58, n104, Wilner 1994). In earlier stages of the language, the singular form is not (n)a but da, which can be assumed to derive from English that in attributive position, as in that man. The early plural form is den or dem, from English them. Both these earlier forms, da and den, have a stronger deictic potential then their modern counterparts. In the 18th-century sources, da and den often function as DEF articles, i.e. they mark a referent that is previously mentioned — cf. wan pikin in (19a) — or can otherwise be assumed to be familiar to the hearer, for example due to situational factors, as in (19b).

 Adrienne Bruyn

Notice that timmere loos ‘carpentry shed’, of which the identity can be assumed to be known as well, appears as a bare noun here. (19)

a.

b.

a kali wan pikin komm na hem, a putti na dem middri, 3SG put at 3PL middle 3SG call INDEF child come at 3SG a brassa, bossi da pikin [C18-SchGos156] 3SG embrace kiss DEF.SG child ‘he called a child towards him, he placed him among them, he embraced and kissed the child’ Cezar joe bin tey da boote bon Caesar 2SG PST tie DEF.SG boat well potti den louw na timmere loos. [C18-vDyk110] put DEF.PL oar at Ø carpentry shed ‘Caesar, did you tie up the boat well, and did you put the oars in the carpentry shed?’

At the same time, 18th-century da and den must sometimes be regarded as demonstratives rather than DEF articles. This is apparent from Schumann’s Sranan dictionary (1783), and from the usage in the text sources. With temporal phrases like da de ‘(on) that day’ or da tem ‘(at) that time’, for example, da evidently has a deictic value. In the following examples, a contrastive demonstrative reading is implied by the context as well as Weygandt’s and van Dyk’s Dutch translations; in (20c) from Schumann’s Gospel text (1781), dem is underlined, apparently to indicate emphatic stress and demonstrative meaning. (20)

a.

Da

stoeloe sa hey toemoesie sontem foe joe, chair FUT high (too)much perhaps for 2SG Diesja sa moro betree [C18-Wey122] this-here FUT more better ‘That chair may be too high for you, this one (here) will be better’ [Kupido.] Mastra mi zi hoe fa tra dirkture libi Kupido master 1SG see how other manager live lange da worke kaba negere no worke haffe zo menni with DEF.SG work and negro NEG work half so much pranasie fo dem krien alle tem. plantation of 3PL clean all time [Eigenaar.] Takki mi hoe fa den dirkture de doe DEM.PL manager CNT do owner tell 1SG how fo holli pranasie krien alle tem. [C18-vDyk87] for hold plantation clean all time ‘[Kupido:] Master, I see how (the) other managers deal with the work, and the slaves need not work half as much (and) their plantations are always clean (of weeds). [Owner:] Tell me how those managers manage to keep their plantations clean all the time.’ DEM.SG

b.

Chapter 12. Bare nouns and articles in Sranan 

c.

Une membre, dem Galileasomma ben de morro ougri DEM.PL Galilee-people PST be more evil 2PL think pikkadosomma, leki dem tarra Galileasomma, sin-people like DEF.PL other Galilee-people bikasi dem miti so ougri? [C18-SchGos175] because 3PL meet so evil ‘You think that those/these Galileans were more evil sinners than the other Galileans, because they suffered such evil things?’

Alongside da and den, the demonstrative disi ‘this, these’ is used attributively in the 18th century, both before and after the noun. If it occurs post-nominally — as it normally does nowadays — either da or den precedes the noun. The demonstrative dati ‘that, those’ on the other hand, while frequent as a pronoun, is not used as a nounmodifying demonstrative until the very end of the 18th century; from then onwards, its position is post-nominal when used attributively. That da and den sometimes have to be regarded as demonstratives in the 18th century is consistent with the absence of dati as an attributive demonstrative in the same period. At the same time, the fact that da and den may have more deictic value than mere DEF articles corresponds to the fact that DEF NPs may appear as bare nouns, which renders the use of da and den more marked. Over time, da/(n)a and den have lost some of their deictic force and grammaticalized towards DEF articles. For example, the 20th-century counterpart to the 18th-century phrase dem Galileasomma ‘those/these Galileans’ in (20c), is den Galileyaman disi [DEF.PL Galilean this] ‘these Galileans’ (Leysipisi 1988: 739). Similarly, temporal phrases like ‘(on) that day’ or ‘(at) that time’ are typically expressed nowadays as a dey dati [DEF day that] and a ten dati [DEF time that], contrasting with the 18th-century phrases involving da ‘that’. The bleaching of the demonstrative value of da/(n)a and den is most clearly evident in the increase in the frequencies of da/(n)a and den on the one hand, and of disi and dati as attributive demonstratives on the other. Table 2 summarizes the total frequencies of da/(n)a and den, of da/(n)a and den taken together without and with post-nominal demonstrative, and of the total number of occurrences of da/(n)a and den together. It shows a clear increase in the use of the DEF articles since the 18th century. While part of the increase is due to those instances of da/(n)a and den that occur together with disi or dati (‘+DEM’ in the table), the frequency of articles not co-occurring with a post-nominal demonstrative (‘plain’) has risen as well.11

. For the frequency counts of da/na and den only cases involving a full noun or, occasionally, a proper name are included. For details on possible ambiguities and their categorization, more detailed figures, and a possible account for the relatively high figures for den in the 19th century in terms of the topics of the texts involved, see Bruyn (1995).

 Adrienne Bruyn

Table 2. The frequency of the Sranan definite determiners da or (n)a, and den da/(n)a & den 20,000-words text portions

da/(n)a

den

plain

+

da/(n)a & den 

18th century 19th century 20th century

411 519 668

209 408 321

609 884 861

11 43 128

620 927 989

The rise in the overall frequencies of da/(n)a and den is indicative of their further grammaticalization as DEF articles. While the items become more frequent, their deictic force diminishes, implying that they become even more generally used. In the case of the singular article there has been a phonological development as well, from da to na and a. With da/(n)a and den losing the deictic potential they still had in the 18th century, the demonstrative function has been taken over by disi and dati in the course of time. However, (n)a and den are not fully grammaticalized as articles to the extent that even in the present-day language, DEF NPs may occur without any determiner, as will be discussed further in the next sections.

.

Bare nouns in Sranan

.

The distribution of bare nouns

Bare nouns, i.e. nouns not preceded by an article, a demonstrative, a possessor, or a quantifier, appear in a variety of contexts in Sranan. To begin with, in the absence of a specialized INDEF-PL article and disregarding the optional use of a quantifier, INDEF plurals cannot but appear as bare nouns. As nouns are not marked for plural, they are realized as unmodified bare nouns. INDEF nouns with a mass or NONINDIV reading always appear as bare nouns as well. (21)

a.

b.

Ma disten, dri boto anga pikin gwe! [C20-dDrie35] but this-time three boat with ø child leave ‘But this time, three boats with children had left!’ Den […] kanti watra na en eri skin tapoe. [C19-Albi31] pour ø water at 3SG whole body top 3PL ‘They […] poured water over his whole body.’

Although not necessarily so, NPs with generic reference often appear as bare nouns. Thus, bare nouns frequently occur in odo’s, proverbs, which as a whole have generic reference. (22)

a. Ø

Makkà soétoe fríman, thorn shoot ø freedman

Chapter 12. Bare nouns and articles in Sranan 

b.

a poéloe hem na sláfoe. [C19-Focke75] at Ø slave 3SG remove 3SG ‘When a/the thorn has pricked a/the freedman, he takes it out on a/the slave.’ Pikién áksi fálla bígi bol. [C19-Focke3] ø small ax fell ø big tree ‘(A) small ax(es) may fell (a) big tree(s).’

However, as we have seen above and will be discussed further below, the use of bare nouns alternates with that of overt articles, not only in the case of generics, but also with INDEF-SG nominals — compare the bare noun pernasi in (23a) with wan pranasi in (b); and with nominals with DEF reference — compare pernasi in (c) with a pernasi in (a): (23)

a.

b.

c.

Ne a seri a pernasi, a kon bay pernasi then 3SG sell DEF plantation 3SG come buy ø plantation a Para [C20-dDrie50] at Para ‘then he sold the plantation, and bought (a) plantation(s) in the Para district’ =(2b) den doro na wan pranasi [C20-Leysi913] 3PL arrive at INDEF plantation ‘they arrived at a plantation (settlement)’ =(14c) watra e kon nini pernasi [C20-dDrie40] water CNT come at-in Ø plantation ‘(the) water was flooding into the plantation’ =(13b)

As shown also by (22b), determinerless nouns may be preceded by an adjective. Although this often involves INDEF plurals, mass nouns, or generics, occasionally adjectives appear in other cases, including nominals with INDEF-SG reference, as in (24b). (24)

a.

biggi sari kissi dem big sorrow get 3PL ‘Big sorrow overcame them’ A meki bigi kawpen. 3SG make ø big cow-pen ‘He made a large cow pen.’

[C18-SchGos160]

Ø

b.

[C20-dDrie91]

While nouns preceded by a possessor — as in Glen oso ‘Glen’s house(s)’, or en oso ‘his/ her house(s)’ — are not considered bare nouns here, it is possible for a bare noun to be followed by a possessive phrase. In some cases the reference can be interpreted as INDEF, as in (25a); in other cases the NP must be taken to have DEF reference, as in (25b). (25)

a.

no NEG

aksi ask

Ø

mati fu yu, friend of 2SG

Ø

brada fu yu, brother of 2SG

Ø

famiri fu yu family of 2sg

 Adrienne Bruyn

b.

efu gudu birtisma. [C20-Leysi941] or ø rich neighbour ‘Don’t invite friends of yours, brothers of yours, family of yours, or rich neighbours.’ Kaba jusnu tatta vo da pikin bari, and soon Ø father of DEF child cry hem hai ronn watra [C18-SchGos153] 3SG eye run water ‘And immediately the father of the child began to cry, his tears came running down’

From the examples of bare nouns given so far it can be seen that they can appear in any syntactic position in Sranan. Here are some more examples of bare nouns not evidently being plural or mass nouns, occurring in the position of predicate, object, prepositional complement, and subject, respectively. (26)

a.

b.

c.

d.

Anake ben de tjariman disi lési, en nanga Asenpe. [C19-Albi28] this time 3SG with Asenpe Anake PST be Ø bearer ‘Anake was (a/the) bearer this time, he and Asenpe. na Krementi, wi tan pikinso, meki den seni boskopoe at Krementi 1PL stay a.little let 3PL send Ø message gi a granman taki wi de krosibé kaba. [C19-King82] that 1PL be close already to DEF chief ‘at Krementi, we stayed for a little while, so they could send a/the message to the chief, that we were nearby already.’ Tu sma ben go na tapu go na tenpri fu begi [C20-Leysi979] two person PST go at top go at Ø temple for pray ‘Two people went up to a/the temple to pray’ watra tranga tumussi [C18-SchDict88] Ø water strong (too)much ‘the current is very strong’

There are no categorical differences between the various stages of Sranan regarding the distribution of bare nouns. However, in the preceding sections it was shown that while the use of the INDEF-SG article wan did not gain in frequency over time, the DEF articles (n)a en den did. This leads to the presumption that the number of bare nouns with definite reference has decreased over time. Voorhoeve & Kramp (1982; cf. Kramp 1983) note that bare nouns appear in Schumann’s dictionary (1783) in cases where present-day Sranan would normally have an article, often DEF, or a possessive pronoun. This would particularly apply in the following contexts: after the preposition na ‘at, to, etc.’ in locative constructions, and after nanga ‘with’; with nouns denoting a characteristic or a defect of a person; with inanimate nouns as the subject of a descriptive sentence; and with inanimate nouns in object position. Although these contexts provide no insight into more general principles

Chapter 12. Bare nouns and articles in Sranan 

determining the use of articles, they capture part of the occurrences of bare nouns in other 18th-century sources as well as in Schumann’s dictionary. The contrast between the 18th and the 20th century is less sharp than suggested by Voorhoeve & Kramp, however. On the one hand, explicit articles sometimes appear in these contexts in the 18th-century sources, including Schumann’s dictionary, and on the other hand, bare nouns do occur in the same contexts in the 20th century. However, a rough count of bare nouns in 5,000-word fragments from van Dyk (ca1765), Schumann (1781), King (1891–94), Albitrouw (1894), de Drie (1985) and the Leysipisi (1988) indicates that the use of bare nouns has indeed diminished somewhat since the 18th century. The frequency of full nouns that are not accompanied by an article, a pre-nominal demonstrative, a pre-nominal possessor, or a quantifier, is approximately 530 out of 10,000 words in the two text fragments representing the 18th century, 450 in the 19th-century text portions, and 400 in the 20th-century ones. Even when the figures for this category must remain somewhat imprecise due to difficulties in determining what should be regarded as a bare noun, it is clear that there is a tendency to fewer such cases. Given that the use of the INDEF-SG article wan has not become more frequent, this indicates that the use of DEF articles has indeed encroached on that of bare nouns to some extent. Even in the modern language however, not every DEF NP contains a DEF article. Thus, bare nouns may have a DEF reading as well as INDEF or generic, rendering it difficult to conceive of bare nouns as a distinctive category.

.

The interpretation of bare nouns

Writing of 20th-century Sranan, Voorhoeve (1953: 9, 56ff) states that while the use of an article renders the noun more concrete, the absence of articles has a generalizing effect. According to him, bare nouns refer to a type rather than a specific referent, and are neutral with regard to number: by themselves, they express singularity nor plurality. Donicie (1954: 28–9), however, not only remarks that the DEF article (n)a may be omitted with nouns with an abstract meaning, but also with nouns referring to unique entities, means of transport, or names of intellectual professions, and with the names of animals in stories. While this may seem a somewhat arbitrary list, Donicie touches upon an important point, namely that some bare nouns are DEF by implication: in some cases, linguistic context and extra-linguistic factors may lead to a DEF interpretation of a noun which is not marked as such. This will form part of the argument to be presented in this section, that the absence of any determiner in Sranan cannot be analysed as a zero form that has its own place in the paradigm of determiners. There are cases where a zero form can be considered to be morpho-syntactically meaningful. Bybee (1994), for example, argues that in some languages the lack of a tense or aspect marker in the verb phrase may convey a particular meaning, depending on the obligatoriness of the overt forms. When a certain marker becomes generally used and increases in frequency, such as progressive be + -ing with dynamic verbs in English, its absence comes to be interpreted as expressing its counterpart, or the

 Adrienne Bruyn

default interpretation in a particular domain — in this case habitual. When a certain overt form that is normally used to express a specific meaning is absent, hearers tend to infer that the meaning expressed by this particular form is not intended, but rather some other meaning. Through the conventionalization of implicature the zero form then becomes associated with its own semantic content when the overt form increases in frequency and becomes obligatory. I will argue, however, that the zero with bare nouns in Sranan does not express one distinctive value. Even though the term ‘zero article’ is used here sometimes for reasons of convenience, this is not meant to imply that zero is a phonologically empty morpho-syntactic form on a par with the overt articles. In the proposal of Givón and Bickerton for a typical creole system, a three-way division is postulated between DEF NPs, INDEF-REF NPs, and NPs that are pragmatically NONREF or NONSPEC, marked by respectively a DEF article, an INDEF-REF article such as wan, and zero. Zero is thus taken to be in complementary distribution to the overt DEF and INDEF articles, and it is assigned its own value: all bare nouns are pragmatically NONREF, or NONSPEC in Bickerton’s terminology, and typically refer to a type. It is not difficult to find Sranan cases compatible with this view. Thus, the bare nouns in the following examples are semantically REF but their particular identity is irrelevant. (27)

a.

b.

c.

Bekassi mi habi kroete lange hem [C18-vDyk98] because 1SG have Ø quarrel with 3SG ‘Because I have (a) quarrel(s) with him’ Dem putti trappo gi hem [C18-SchGos169] Ø trap for 3SG 3PL put ‘They set (a) trap(s) for him’ den prey banya heri wiki [C20-dDrie48] 3PL play Ø banya whole week ‘they played (a) banya(s) for a whole week’ (banya: traditional play with song and dance)

That the bare nouns in these examples refer to types rather than to entities whose individual identity is at issue can be illustrated by rendering them as ‘quarrel-having’, ‘trap-setting’, and ‘banya-playing’. With semantically NONREF arguments, including generic predicates, nouns referring to a type, and NPs in the scope of negation, we also find bare nouns as well: (28)

a.

b.

b.

mi granpapa, di ben de granman daten [C19-King86] DEM-time 1SG grandfather that PST be Ø chief ‘my grandfather, who was chief at that time’ a snorku leki tigri [C20-dDrie31] like Ø tiger 3SG snore ’he snored like (a) tiger(s)’ a no sa gi mi ougri nem [C18-SchGos157] 3SG NEG FUT give 1SG Ø bad name ‘he will not give me (a) bad reputation’

Chapter 12. Bare nouns and articles in Sranan 

However, as discussed in Section 3, it is not exceptional for wan to appear in the same environments, even when the referent at issue does not recur in the subsequent discourse and can therefore be considered pragmatically unimportant. Taken across the three periods, the quantitative analysis of the use of wan showed that 23% (189) of the total of 819 wan-NPs appear in semantically NONREF contexts; 68% of these are non-recurrent. Furthermore, regardless of semantic referentiality, 41% (339 of 819) of the wan-NPs do not recur after the first mention. Even allowing some leeway for a broader interpretation of pragmatic referentiality than as operationalized here, it is evident that wan is used rather frequently in contexts where a bare noun would have been more appropriate according to the alleged creole system. The primary reason for the use of wan with pragmatically unimportant referents appears to be the need to explicitly express singularity, or indefiniteness, or both. Conversely, INDEF-REF nominals occasionally appear without wan, as in (29). The recurrence of the same referent with the DEF-SG article da in the following sentence implies that the referent of the bare noun soema (nowadays spelled <suma>) implicitly has INDEF-SG reference, and that it is thematically important. (29)

den seni soema go loekoe, efi den de kon. 3PL send Ø person go look, if 3PL CNT come Da soema go èn […] [C19-Albi29] DEF.SG person go and ‘they sent (a) person / someone to go and see whether they were coming. The person went, and […]’

Because on the one hand wan is not exceptional where zero would be appropriate following Givón and Bickerton, and on the other hand zero may occur where wan would be expected, zero cannot be regarded as being in opposition with wan with pragmatic referentiality as the distinctive feature — a conclusion already reached in the sections devoted to wan. Looking at the proposal of Givón and Bickerton from the perspective of zero, it appears that not only does the postulation of referentiality as the primary distinction sometimes lead to the wrong predictions for the use of zero and wan, but that the assumption that bare nouns are a unified category alongside REF-INDEf NPs and DEF NPs cannot be valid in the case of Sranan. One complication arises from the lack of plural marking, the other from the occurrence of bare nouns with DEF reference. The fact that Sranan has no plural marking on the noun nor a separate INDEF-PL article poses a structural problem for the alleged creole system. While wan could be added in cases involving singular reference, like (29), it is not possible to express the distinction between thematically important bare nouns such as those in (30) which are implicitly plural — witness the subsequent use of the plural pronoun den in these cases — and bare nouns with a lower pragmatic referentiality. (30)

a.

Bikasi because

Ø

bakra ben kon na mi, white(person) PST come at 1SG

 Adrienne Bruyn

b.

mi taki gi den […] [C19-King83] 1SG say to 3PL ‘Because (some) whites came to me, I said to them […]’ =(15b) banditi lontu a man, Ø bandit surround DEF man dan den puru ala sani na en skin [C20-Leysi909] then 3PL pull all thing at 3SG body ‘(some) bandits surrounded the man, then they stripped him of everything’

Although the quantifiers wantu [one-two] ‘a couple of, a few, some’ and son ‘some, (a) certain’ can be used with nouns with INDEF-PL reference, they are too restricted in frequency and meaning as to consider either of them as an article. Thus, the fact that INDEF-PL nominals appear as bare nouns unmarked for plural regardless of pragmatic referentiality is problematic for the typical creole system, where zero is assumed to correspond categorically to pragmatic non-referentiality. A similar problem arises with respect to the proposal that all bare nouns are NONINDIV. Mufwene (1981, 1984, 1986; cf. e.g. Lyons 1999) points to the similarities between plural count nouns and mass nouns with respect to number delimitation. Both refer to an amount specified only vaguely, and in many languages, including English, the INDEF article for both plural and mass nouns is zero. Accordingly, INDEF-PL NPs in creoles without morphological plural marking are not regarded as a separate category but are subsumed under NONINDIV by Mufwene. Individuation is construed as a scalar phenomenon, going from INDIV-SG through INDIV-PL to NONINDIV, and languages vary with regard to what points on the scale are critical points with respect to morpho-syntactic marking. Thus, while INDEF-PL nouns in English are not preceded by an article but are marked for plural, all bare nouns in Jamaican and Gullah are regarded as undelimited with respect to number, regardless of whether they may receive an INDEF-PL reading. On this view, definiteness cannot be expressed independently of individuation with NPs with plural reference. The options are either an INDEF-NONINDIV bare noun, or a DEF-PL-INDIv NP, while DEF-PL-NONINDIV is excluded as well as INDEF-PL-INDIV. Or, to put it another way, it must be assumed not only that -PL NPs do not exist as such, but also that DEF-PL NPs are necessarily INDIV, and, conversely, PL-INDIV NPs necessarily DEF. This implication is touched upon but not discussed by Mufwene (1986: 40, 48), who is concerned with number delimitation as such rather than with its interrelatedness with the marking of definiteness. However, while it is true for Sranan as well that INDEF-PL cannot be considered to be a separate category to the extent that there is no way explicitly to mark an NP as such, it is questionable whether an INDIV reading of bare nouns such as bakra ‘white(s)’ or banditi ‘bandit(s)’ in (30) is really excluded. In view of the anaphoric reference by means of the plural pronoun den they are certainly less mass-like or generic than other bare nouns, and also arguably less so than buk ‘book(s)’ in the Jamaican example provided by Mufwene which comes closest to the Sranan cases in (30): (31)

Buk de aal uova di tiebl ina im afis. book be.LOC all over the table in him office ‘Books were all over the table in his office.’

Jamaican (Mufwene 1981: 229)

Chapter 12. Bare nouns and articles in Sranan 

While the absence of plural marking with INDEF nominals thus renders problematic the interpretation of bare nouns as a unified category in opposition to either INDIV or pragmatically REF NPs, the issue is complicated even further by the fact that bare nouns in Sranan sometimes have a DEF-INDIV reading. Before turning to such cases, let me point out that, while it is easy to find Sranan examples that are consistent with the creole system proposed by Givón and Bickerton, the same is true of Mufwene’s proposal: many instances of bare nouns can be regarded as being NONINDIV. Below are some examples of bare nouns that can be construed as mass-like, in the (a)-sentences, contrasted with INDIV wan-NPs in the (b)-sentences. (32)

a.

b.

(33)

a.

b.

(34)

a.

b.

tiede ary fadom trange strong today Ø rain fall ‘today rain fell heavily’, ‘today it rained heavily’ wan bigi alen bigin fadon INDEF big rain begin fall ‘a heavy shower started to fall’ Den gi en njanjan 3PL give 3SG Ø food/meal ‘They gave him food’ effi ju setti wan janjam if/when 2sg set indeF meal/food ‘when you organize a meal’

[C18-vDyk71]

[C19-King89]

[C19-Albi31]

[C18-SchGos180]

a de doe ogri na kondre [C19-King127] 3SG CNT do Ø evil at country ‘he is doing harm to the country’ wanten den sabi: da jorka si wan ogri. [C19-Albi29] one-time 3PL know DEF ghost see INDEF evil ‘right away they knew: the ghost had seen an evil thing / something evil’

In such cases, it is not unreasonable to assume that the contrast between the bare nouns and the wan-NPs is indeed one of individuation. On the other hand, it cannot be excluded that the bare nouns in the (a)-sentences may have other interpretations — for example, ‘a meal’ in (33a) — than those suggested by my translations, which are meant to illustrate the notion of non-individuation rather than to provide all possible readings. In general, the absence of independent criteria makes it difficult to assess the validity of Mufwene’s proposal. For example, generics are said to appear typically as bare nouns, yet one type of NP that seems to be in conflict with this proposal is that of generic predicates, which often contain wan in Sranan (see Section 3). However, Mufwene (1986: 55, n13) notes that in Gullah and Jamaican generic predicates sometimes contain an INDEFSG marker, and accounts for this by arguing that this distinguishes INDIV predicates from NONINDIV ones. This indicates not only that wan-NPs as generic predicates in Sranan cannot be taken as evidence against the assumption that individuation determines the presence or absence of articles, but that in general it is difficult to provide evidence either for or against the prediction that NONINDIV nouns are not preceded by an article.

 Adrienne Bruyn

There is one type of bare noun, however, which poses a crucial problem both for the assumption that bare nouns are always NONINDIV and for the system proposed by Givón and Bickerton. These are bare nouns that have DEF reference and refer to individual entities — compare Examples (13a, b), (19b) and (23c) above. When we take the actual occurrences of bare nouns in Sranan as the starting point, it turns out that a considerable part of the bare nouns are of this type. The very nature of bare nouns, namely that they are unmarked with respect to number and definiteness, makes it impossible to categorize them in an exact manner and to provide precise frequency figures for different types. However, a passage from de Drie (1985) may serve to illustrate the fact that bare nouns with DEF and INDIV reference are not exceptional, and how such bare nouns may function in the discourse. The extract in (35) below is preceded by a passage in which a female slave is introduced, who had a child whose father is the plantation owner. She is sold to another plantation, wan tra pernasi, with another owner, wan tra granmasra, and at this other plantation she has two other children, Opoke and Atyopi. This plantation is then mentioned again in line 1 of (35), accompanied by the DEF article a, and specified as the one where Opoke and Atyopi were born. In line 3 it is referred to as a pernasi dati ‘that plantation’, which confirms the individual identity of this particular plantation that can be assumed to be stored firmly as DEF in what Givón (1984a: 401ff; cf. Givón 2001: 462ff) calls the active discourse file. Then, in line 4, the plantation is referred to by means of the bare noun pernasi. It is rather unlikely that the plantation now should have ceased to be conceived as INDIV, or that its thematic importance should be crucially diminished. Rather, a more suitable account of the use of a bare noun here would assume that the topicality of this plantation as well as its identity are established firmly enough to make the use of a DEF article superfluous. In line 8, pernasi again appears as a bare noun. Between line 4 and line 8, the topic of the story has remained the same: it goes on about the plantation, the bad condition of the sluice and the dam, and the water coming through and flooding the plantation. Therefore, in line 8 it is still not necessary to use a DEF article with pernasi. But from line 11 onwards there is an interruption of the main story: basya Oloku is introduced. How he got his nickname, Oluku, related to the Dutch word ongeluk ‘bad luck, accident’ is explained, and then his mother is introduced. By this time, the topicality of the plantation can be regarded as having shifted to a secondary plane, so that when it is mentioned again in line 15, it is preceded by the DEF article a, in order reactivate it in the discourse file as it were. Whether such reactivation is considered necessary by the speaker will of course differ from case to case. So, in the case of granmasra ‘plantation owner’, there is a considerable gap between the establishment as a topic in line 2, and its appearance as a bare noun in line 12, and again in line 19. (35)

1. […] a

pernasi pe Opoko gebore, pe Atyopi gebore, plantation where Opoke born where Atyopi born ‘[…] the plantation where Opoke was born, where Atyope was born,’ DEF

Chapter 12. Bare nouns and articles in Sranan 

2. a

granmasra dati ben de wan wreedaardige granmasra. […] great-master that PST be INDEF cruel great-master ‘that plantation owner was a cruel plantation owner […]’ Ma a pernasi dati tusyuru sroysi e broko. […] and DEF plantation that always ø sluice CNT break ‘and that plantation, its sluice(s) was/were always breaking down. […]’ Pernasi feti fu sungu bika sroysi boro. ø plantation fight for sink because ø sluice have.holes ‘The plantation was about to flood because the sluice(s) were cracked.’ Sroysi e feti fu broko, dan broko. ø sluice CNT fight for break ø dam break ‘The sluice(s) was/were about to break down, the dam(s) was/were broken down.’ Wan neti den go kari basya Ananka. Ananka taki INDEF night 3PL go kari basya Ananka Ananka say ‘One night, they came to call Basya Ananka. Ananka said:’ “ […] mi de na mi sribi, skrifman Kofi kon kari mi Kofi come call 1SG 1SG be at 1SG sleep writer “ […] I was asleep, Kofi the writer came to call me, telling me’ taki dan broko, watra e kon nini pernasi. that ø dam break water CNT come at-in Ø plantation ‘that the dam(s) had burst, (the) water was coming into the plantation.’ San mi mu du! A no mi broko en.” […] what 1SG must do it.be NEG 1SG break 3SG ‘What should I do? It wasn’t me that broke it.” […]’ A mu gwe go wiki skopuman nanga lanteri with lantern 3SG must go.away go wake digger ‘He had to go and wake up the digger(s) with lanterns’ kon tap dan. […] Ma boyti basya Ananka come close ø dam but apart.from basya Ananka ‘(and have them) come and close the dam. […] But besides Basya Ananka’ dan i ben abi wan tra basya then 2SG PST have INDEF other basya ‘then you had another basya,’ den ben kari basya Oloku. […] Dan granmasra then ø great-master 3PL PST call basya Oloku ‘called Basya Oloku. […] Then, the owner’ ben kari basya Oloku, yo ongelukkige vent, bika […] PST call basya Oloku, 2SG(?) unlucky fellow because ‘used to call him Basya Oloku, you poor fellow (ongelukkige vent), because […]’ Ma now Mari Oloku ben habi wan ma, and now Mari Oloku PST have INDEF mama ‘Well now, Mari Oloku had a mother,’ DEF

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

 Adrienne Bruyn

16. den ben kari en Ma Regina, a ben de n’ a pernasi, 3PL PST call 3SG Ma Regina 3SG PST be at DEF plantation ‘they called her Ma Regina, she lived at the plantation,’ 17. wan obiamama. Na en a gronwinti ben kisi. INDEF obya-mama FOC 3SG DEF earth-spirit PST catch ‘a magic-woman. HER the earth-spirit (winti) had caught.’ 18. Dan di a sroysi broko wan siksi, seybi tron, seven time then when DEF sluice break one six ‘Then, when the sluice had broken down six or seven times,’ 19. den no habi wan fasi fu du anga na sroysi, 3SG NEG have INDEF manner for do with DEF sluice ’they didn’t have a way to deal with the sluice,’ 20. dan granmasra taygi basya Ananka, […] [C20-dDrie40] then ø great-master tell basya Ananka ‘then the plantation owner told Basya Ananka, […]’

What the examples of pernasi and granmasra clearly show is that the zero article does not always correspond to non-individuation: these bare nouns are undoubtedly intended to refer to singular individual entities. At the same time, cases such as pernasi and granmasra are evidence also against the assumption that all bare nouns are pragmatically non-referential. It can even be argued that it is exactly their thematic centrality that makes it possible for some participants to appear as bare nouns. The cases of sroysi ‘sluice’ and dan ‘dam’ in (35) differ from pernasi and granmasra in that their first appearances are as bare nouns, in line 3 and 5 respectively. Since both play an important role in the story, this cannot be attributed to pragmatic insignificance. The absence of articles is intelligible in view of the fact that it is not necessary to explicitly express the definiteness of sroysi ‘sluice’ and dan ‘dam’, which can be considered implicit in their being parts of the plantation. Perhaps the zero articles could also be viewed as reflecting the fact that the referents are meant as NONINDIV. It is only later on that it becomes apparent that singular reference is involved: dan is referred to by the singular pronoun en in line 9,12 and by a dan ‘the (SG) dam’ further down in the story. The singularity of sroysi ‘sluice’ is explicitly expressed in line 18. That the DEF-SG article a is used here can be accounted for in the same way as was done before with regard to a pernasi ‘the plantation’: it brings a sroysi ‘the sluice’ back into the foreground after not being mentioned for some time. The difference with the case of pernasi is that the first occurrence of sroysi, and that of dan, is as a bare noun. Whether the zero article . It could be argued that the use of en does not imply INDIV-SG reference, since en can refer to mass nouns as well. Drawing on the correspondence between NONINDIV NPs and mass nouns en would actually have to be seen as the appropriate pronoun to refer anaphorically to NONINDIV NPs. This raises several problems, including how the use of plural den in cases like those in (30) can be accounted for, if at the same time it is assumed that the bare nouns bakra ‘white(s)’ and banditi ‘bandit(s)’ are NONINDIV.

Chapter 12. Bare nouns and articles in Sranan 

in these cases has to do with non-individuation is difficult to establish. What we can say, however, is that in any case they are not necessarily NONINDIV, because bare nouns in general are not necessarily NONINDIV, witness cases such as pernasi and granmasra in (35). Such bare nouns with topical DEF referents are by no means exceptional, neither in de Drie (1985) nor in other Sranan texts of various genres from the various periods. It appears that with DEF NPs, the use of an article is less essential not only if the specific identity of the referent is irrelevant — as is often the case, for example, with prepositional arguments — but also if it is relevant but already available to the hearer. There are several ways in which a referent can be considered to be part of the active discourse file. One way is through explicit establishment in the previous discourse, of which the case of pernasi in (35) is a typical example. First it is introduced with wan (before the cited passage), then it appears with a DEF article and a specifying relative clause, then with dati ‘that’, and after that it can appear as a bare noun since there will be no misunderstanding about which plantation is at issue. To some extent, this phenomenon may be compared with the use of the relative past marker ben in Sranan: once the events being talked about are anchored in time, the verb may subsequently remain unmarked for tense (cf. Winford 2000). In other cases of DEF referents expressed by bare nouns, their identity is not explicitly established in the discourse but rather implied by it. This may proceed through an associative connection such as a part-whole relationship, as is the case with sroysi ‘sluice’ and dan ‘dam’ in (35), where it is automatically assumed that these are located on the plantation under discussion. Similarly, a part-whole relationship with the referent of da negere ‘the negro’ in the same sentence accounts for the identification of hede in (36): (36)

joe zoete da negere retti na ini hede 2SG shoot DEF negro right at in Ø head ‘you’ve shot the negro right in the (his) head’

[C18-vDyk81]

In other cases the interpretation in terms of definiteness and number is implied not by the textual context but by the wider context, including situation, cultural knowledge or expectations, and general knowledge of the world. As there is one foto ‘town (fort)’ in Suriname, called Paramaribo mainly by outsiders, referring to plain foto is clear enough. In the context of missionary activities, there is only one buku ‘book’, namely the Bible. As it is normal to conceive of son ‘sun’ as unique, its identity need not be specified. (37)

a.

b.

c.

den zel jari mi go na fotte 3PL FUT carry 1SG go at Ø town ‘they will bring me to (the) town (Paramaribo)’ bika boekoe fermane wi foe […] because Ø book admonish 1PL for ‘because the Book/Bible admonishes us for […] ‘ So son bigin dongo. so Ø sun begin set ‘So, the sun began to set.’

[C18-vDyk103]

[C19-Albi37]

[C20-Leysi841]

 Adrienne Bruyn

This type of bare noun exhibits a name-like character, as is the case with nouns like granmasra ‘plantation owner’, which have unique reference within a certain setting. They illustrate once again that while in some cases the absence of any article is related to the fact that the referent has no individual identity or that its identity is irrelevant, in other cases it is the very uniqueness of an individual identity within a particular context that makes the use of an article redundant. Bare nouns in Sranan thus turn out to be open to several rather different interpretations, depending on the meaning of the noun as well as the textual and extra-linguistic context. Zero-marked nouns may have generic reference, or mass-like referents, or referents whose identity is irrelevant, but they may also refer to arguments with an individual, pragmatically salient identity, which can be DEF as well as INDEF-PL. The variety of possible readings is the corollary of the relative optionality of the overt articles. For a zero form to acquire its own morpho-syntactic meaning, it has to occur in opposition with overt forms that are more or less obligatory in particular contexts (cf. Bybee 1994). In Sranan, however, bare nouns are not in complementary distribution with the explicit articles with respect to some feature, whether definiteness, referentiality, or number delimitation. Therefore, a specific value cannot be assigned to the zero article. Instead of analysing the zero article as a separate category with a distinctive semantic content, it would seem to be more appropriate to regard bare nouns in Sranan as undetermined with respect to referentiality as well as definiteness and number. While pragmatic referentiality can account for the use of an article versus its absence in many cases, not all insignificant referents appear as bare nouns, and in some cases thematic saliency may even be taken to account for the use of a bare noun. Similarly, the notion of individuation may help to understand the use of articles versus zero to some extent, but it cannot provide an explanation for all cases, if only because the Sranan articles do not only express number but definiteness as well.

.

Lexifier or substrate?

A brief comparison of bare nouns in Sranan with those in the lexifier language and in the main substrate languages indicates that Sranan exhibits some similarities with both in this respect but also important differences. The main difference with English concerns a wider variety of possible interpretations as well as a higher frequency of bare nouns in Sranan. This is due to the absence of number marking in Sranan alongside the fact that neither INDEF nor DEF articles are obligatory in Sranan to the same degree as they are in English. In Sranan, INDEF plurals typically appear as bare nouns, whereas in English they are often accompanied by some. And when they do occur without determiner, they are morphologically marked for plural in any case, and are accordingly identifiable as INDEF-PL, possibly with generic reading. Apart from plurals and mass nouns, English bare nouns are limited to a few idiomatic cases like by train, to school, or in hospital.

Chapter 12. Bare nouns and articles in Sranan 

While the counterparts of such phrases are determinerless in Sranan too, the occurrence of bare nouns is much more widespread in the creole language. Significantly, they are open to a variety of interpretations, including DEF as well as INDEF-SG, readings which are not generally available for English bare nouns. While the context sometimes implies a particular reading — for example DEF-SG when the referent’s identity is previously established or implied — by itself the Sranan bare noun is neutral with regard to definiteness and number, something which has no counterpart in English either. As for the substrate, among the languages that were spoken by West Africans in Suriname in the early colonial period and therefore may have influenced the developing creole language, the most important ones can be assumed to have been Fon and Ewe, both belonging to the Gbe dialect cluster of Kwa languages, and the Bantu language Kikongo (Smith 1987, Arends 1995a). Influences from these languages have been identified on the levels of phonology, vocabulary, and lexical semantics (e.g. Huttar 1975, Smith 1987) as well as morpho-syntax (e.g. Migge 1998, Smith 2001). The general picture that arises is that whereas various African languages including Kikongo as well as Gbe languages have contributed to the Sranan lexicon, instances of morpho-syntactic influence are rather attributable to Gbe. As it is beyond the scope of this paper to describe the determiner systems of the various languages in detail, I will briefly compare Sranan with the general pattern as it applies to the languages of the Gbe group, drawing on Westermann (1907), Essegbey (1999), Aboh (1999) and Lefebvre & Brousseau (2002), thereby glossing over some of the differences that may hold between them. In the Gbe languages, new arguments are often introduced into the discourse as bare nouns. A post-nominal determiner, such as Ewe a
é or Fon
é, may occur with specific INDEF NPs. Furthermore, DEF markers, such a Ewe (l)a or Fon ɔ́(n), also appearing post-nominally, encode NPs whose identity is not only known to speaker and hearer, but also anaphorically recoverable from the discourse. All other nominals appear as bare nouns, which thus includes not only generics, but also indefinites when not explicitly specific, and definites when not discourse-anaphoric. Plural is marked by a separate morpheme. Although there appears to be some variation across the languages with regard to the interaction of plural marking and definiteness, bare nouns by themselves can be regarded as neutral with respect to number in all varieties. Notwithstanding the fact that there is a certain overlap in the usage of bare nouns in Sranan and in the Gbe languages, there are certain differences in the distribution of explicit determiners versus that of bare nouns, and, concomitantly, differences regarding the possible interpretations of the latter.13 One difference concerns the role . Pace Aboh (2003), who rather focuses on similarities in article usage between Sranan and the Gbe variety Gung(gbe). He argues that the nominal markers, while differing in syntax, fulfil similar functions, and that specificity in the sense of being discourse-anaphoric is a crucial distinction in both languages. Accordingly, Aboh assumes Sranan bare nouns to be similar to those in Gun in that they may, depending on the context, be interpreted as generic, INDEF, or DEF, but are excluded from having discourse-anaphoric reference.

 Adrienne Bruyn

of discourse-anaphoricity. In Sranan, as in English but in contrast to Gbe, nominals whose referents are identifiable on the basis of shared extra-linguistic knowledge are construed as DEF, regardless of whether they are discourse-anaphoric or not (cf. Lyons 1999: 158–60). DEF articles may thus appear with NPs that are not anaphorically recoverable from the discourse in Sranan, whereas a bare noun would be expected in Gbe in such cases. At the same time, discourse-anaphoric DEF NPs may occur without a DEF article in Sranan. Indeed, as argued in the preceding section, with some instances of Sranan bare nouns it is the very fact that their identity is established in the preceding discourse that accounts for their ability to appear as bare nouns. Another difference in the distribution of bare nouns follows from the wider use of Sranan wan in comparison with the markers of specific indefinitess in the Gbe languages. As shown above, Sranan wan rather frequently appears with NPs with a generic reading as well as with other types of NONREF referents, and it certainly is less restricted in function than Fon
é, or Ewe a
é, ‘a certain’, which are limited to cases where the speaker has a particular member of a class in mind (cf. Lefebvre & Brousseau 2002: 40, Essegbey 1999: 43). For example, while a
é occurs in cases such as (38) in Ewe, it would be inappropriate in (39a) and (b) where reference is to any dog rather than a particular one. Similarly, the Fon determiner
é is not used with generic predicates such as (39c). (38)

(39)

Avu áé le afé-á me dog SPEC LOC house-DEF in ‘A certain dog is in the house’ a.

b.

c.

Ewe (Essegbey 1999: 43)

me Ewe Avu le afé-á dog LOC house-DEF in ‘There is a dog in the house’ (Essegbey 1999: 42) uu-a wu avu le mɔ́-á dzí Ewe vehicle-DEF kill dog LOC road-DEF top ‘The vehicle killed a dog on the road’ (Essegbey 1999: 111) Kɔ́kú nyí mέsì àgbè Fon Koku be teacher good ‘Koku is a good teacher’ (Lefebvre & Brousseau 2002: 143)

In Sranan, by contrast, wan, although not necessarily so, can appear in counterparts of the Gbe Examples in (39) (cf. e.g. (5a) and (11a) above). The wider distribution of wan entails that, in Sranan, generics and other NONSPEC NPs are not expressed exclusively by bare nouns, which implies another difference with the Gbe languages with regard to the position of bare nouns vis-à-vis NPs involving overt determiners. The comparisons with English and Gbe must remain cursory here, and it would certainly be worthwhile to look in more detail into, among other things, actual article usage in Gbe texts. Yet, it seems justified to conclude that Sranan has developed its own system with regard to the presence and absence of articles. With the Gbe languages it shares a rather frequent use of bare nouns and the fact that these are neutral with

Chapter 12. Bare nouns and articles in Sranan 

regard to definiteness and number. The Sranan INDEF and DEF determiners have less restricted functions than those in Gbe however, so that Sranan may have a determiner where a bare noun would be appropriate in Gbe. Conversely, due to the optionality of the Sranan determiners — not derived from English articles anyway — we find bare nouns in Sranan where English would require an article. Sranan thus occupies the middle ground, as it were, between Gbe and English, while at the same time exhibiting more alternation between nominals with and without determiner than either Gbe or English.

.

Conclusion

As shown in the preceding sections, the use of overt articles is rather variable in Sranan, to the extent that the INDEF-SG article wan and the DEF articles (n)a (SG) and den (PL) are not always obligatory. If there is no need to explicitly mark a nominal for number and definiteness, an article does not necessarily appear. Concomitantly, bare nouns are open to several interpretations with respect to definiteness, referentiality, and number delimitation, precluding the zero article to assume a unique value with respect to any of these dimensions. Thus, bare nouns are not exclusively NONSPEC, or pragmatically NONREF, as would be the case according to Bickerton’s and Givón’s typical creole system, nor are they always NONINDIV, as proposed by Mufwene. It is certainly true that many instances of bare nouns in Sranan can be understood along the lines of referentiality, and the same applies to individuation. However, by themselves neither of these dimensions can fully account for the absence versus the presence of articles. According to Mufwene’s (e.g. 1986) proposal, the use of articles in creole languages such as Gullah and Jamaican is determined by individuation: while articles would express individuation, bare nouns would typically refer to mass-like collectives. Although referentiality and individuation tend to go hand in hand in general, singular individuation indeed can sometimes be regarded as accounting for the use of wan in Sranan where it would not be predicted on the basis of referentiality. A problem is posed however by the assumption that all nouns occurring without article or number marking are NONINDIV. In Mufwene’s view, INDEF-PL does not constitute a separate category but is subsumed under NONINDIV in those creole languages where nouns with INDEF-PL meaning remain unmarked, as is the case in basilectal Gullah and Jamaican, and also in Sranan. Although many INDEF-PL nouns may indeed be regarded as referring to a collective, it may be doubted whether this is necessarily so, and whether, conversely, all NPs containing an article, in particular in the case of DEF-PL articles, should necessarily be construed as INDIV. By focusing on number delimitation, Mufwene tends to neglect the role of definiteness. However, as the articles mark not only for number but also for definiteness at the same time, the expression of definiteness may sometimes interfere with that of individuation. Furthermore, bare nouns can, in Sranan in any case but also in some other creoles, be DEF and have referents with a

 Adrienne Bruyn

unique individual identity. Thus, notwithstanding the fact that many bare nouns may indeed be considered NONINDIV, this does not apply to all bare nouns. Bare nouns with DEF reference, in particular those that refer to thematically important arguments, also constitute one of the problems with respect to the typical creole system as proposed by Bickerton (e.g. 1981) and Givón (e.g. 1984b). In this system, the distinctive feature determining article usage is specificity (Bickerton) or pragmatic referentiality (Givón). It is assumed that explicit articles, both DEF and INDEF, are used with pragmatically REF arguments only, and that, conversely, the referents of bare nouns do not have a particular identity, or that, if they do have a specific identity, it is pragmatically unimportant. To the extent that article usage in Sranan does not accord with the typical creole system, it might be thought that Sranan is exceptional among the creole languages, and that its deviance from the system as proposed by Bickerton and Givón does not necessarily challenge the general validity of that system. That would be a rather unsatisfactory approach since, in view of its morpho-syntactic properties as well as of the fact that contact with its lexifier language was more or less cut off at an early point in its history, Sranan is generally included among the genuine creole languages. If there exists something like a typical creole determiner system at all, Sranan would be expected largely to conform to such a system. Furthermore, the study of article usage in Sranan suggests that certain complications with respect to the typical creole system as proposed may apply to other creole languages as well. In Sranan, bare nouns are not always pragmatically NONREF, or NONSPEC. Nor is the use of explicit articles restricted to cases where the referent has a specific identity; there are too many instances of wan-NPs that are both semantically NONREF and pragmatically unimportant for this to be the case. The frequency and the distribution of wan in the 18th and in the 20th century are remarkably similar, and the wider use of wan cannot be attributed to a further grammaticalization over the centuries, away from a more typical creole system. A closer inspection of the wan-NPs that are not in accordance with Bickerton’s and Givón’s proposal showed that these can rather be accounted for by regarding wan as a marker of singularity and indefiniteness, albeit one which is not necessarily present with every INDEF-SG nominal. Although pragmatic referentiality certainly influences the use of wan versus zero, it cannot be regarded as the primary distinction. With NONREF arguments too, wan may be used to explicitly express singularity, or indefiniteness, or both. The fact that the use of wan is sometimes brought about by a need to disambiguate for number must be viewed in connection with the absence of plural marking on the noun in Sranan. This aspect of the function of wan, and of similar items in other creole languages without plural morphology, is not sufficiently taken account of in the allegedly typical creole system. In Section 3.2, it was noted how Givón (1984b) accounts for the use of wan ‘one, a’ in a Krio story in a context where this is not expected according to his proposal: it is argued that in this case, wan serves to disambiguate with regard to number. This indicates that pragmatic non-referentiality may sometimes be

Chapter 12. Bare nouns and articles in Sranan 

overruled by the need to explicitly mark a nominal as singular not only in Sranan but in Krio as well. In fact, this may apply to all creole languages that do not have plural morphology, including not only the Suriname creoles and Krio, but also basilectal Jamaican, Haitian, and several other Atlantic creoles. Furthermore, in the absence of a specialized INDEF-PL article as well as of plural morphology, INDEF-PL nominals necessarily appear as unmodified bare nouns in such languages, whether they have pragmatically important referents or not. Thus, pragmatically important plural referents are introduced into the discourse by unmodified bare nouns, thereby posing a problem for the proposal of Bickerton and Givón. It should be noted that the same problem does not occur in Hawaiian Creole English, where plurality may be marked on articleless nouns by the ending -s, of which Bickerton (1981) states that it is used with SPEC nominals only. However, INDEF-SPEC plurals in Sranan and in other creoles where the situation with respect to number marking is similar cannot be accommodated in the typical creole system. Articleless nominals may also have DEF reference in Sranan. Correlating with the further grammaticalization of the DEF articles over the course of time, this type of bare noun is somewhat less frequent in present-day Sranan than it is in the earlier stages, but it is still a common phenomenon even in the modern language. Sometimes the specific identity of a DEF bare noun’s referent is irrelevant to the point at issue. In other cases, however, the identity is relevant but easily accessible to the hearer because it is implied by linguistic or extra-linguistic context. If the identity of a referent is, in a certain context, unique, it may appear without a DEF article. As we have seen, this includes cases of major participants in the current discourse, whose identity has been established previously and is therefore readily available to the hearer. As the absence of any article can be regarded as resulting from the pragmatic saliency of the referent in some cases, whereas in other cases zero corresponds to a lack of pragmatic referentiality, the inevitable conclusion is that there is no zero article with its own unique value with respect to pragmatic referentiality, and that the latter cannot be the primary distinction determining the use of articles in Sranan. In Sranan as well as in other languages, including Hawaiian Creole English and Krio, on which the proposals of Bickerton and Givón are mainly based, many bare nouns can be regarded as referring to a type or as pragmatically unimportant, and by focusing on such cases, one may argue that pragmatic referentiality determines the use of articles. However, the investigation of article use in Sranan has shown that, even though many cases conform to the proposed creole system, there are various kinds of exceptions. Two rather crucial types of counter-evidence are the use of wan with NONSPEC, pragmatically unimportant referents, and bare nouns referring to pragmatically important arguments, either after having been introduced into the discourse, or with an identity otherwise familiar to the hearer already. Both phenomena are widespread in Sranan and cannot be ignored in analysing the distribution of articles. While similar facts have occasionally been reported for other creole languages (cf. e.g. Kouwenberg 1994, Christie 1998, Baptista 2002), they have, generally speaking, received less attention than

 Adrienne Bruyn

the cases that can be interpreted in accord with Bickerton’s and Givón’s system, and they have not really affected the widely-held view that bare nouns in creole languages correspond to non-specificity or non-referentiality. In particular the phenomenon of thematically important arguments that may appear as bare nouns after their identity has been firmly established in the discourse is rarely mentioned, and it is difficult to determine how general this phenomenon is across the creole languages. However, bare nouns with referents that have an individual identity and are thematically important, which are not uncommon in Sranan in any case, do not conform to the proposals put forward with respect to bare nouns in creole languages. Bickerton and Givón assume that zero is a separate category in creole languages, and that its occurrence versus that of explicit articles is primarily determined by referentiality, in contrast to the distinctions determining article usage in English and other European languages that provided the lexical base for creole languages. It appears however that the role played by referentiality should not be overestimated vis-à-vis that of number and definiteness. The article system of Sranan cannot be properly understood if due attention is not paid to the fact that the articles primarily express number and definiteness. While the use of the articles is not obligatory, pragmatic referentiality, as well as individuation, evidently play a role in their distribution. Neither referentiality nor individuation can be regarded as constituting the primary distinction with respect to the article system as a whole however. The difference between Sranan and languages such as English does not lie in the fact that crucially different distinctions determine the occurrence of articles, but in the degree of generalization of article use. Bare nouns occur in English as well, as in by train or on shore. As with many bare nouns in creole languages, the absence of any article may be attributed to the fact that the reference is to a type rather than a particular individual (‘train’), or that the identity of the referent is irrelevant, or NONINDIV (‘shore’). In English and other languages where article usage is fairly extended, such bare nouns are to some extent idiomatic. For example, one Dutch parallel to by train involves a DEF article: met de trein, whereas no article can appear after the Latinate preposition in per trein. The fact that article use in Sranan, and, by extension, in comparable creole languages, is less generalized than in languages such as English entails a more general occurrence of bare nouns, and a larger role to play for referentiality and individuation in the alternation between overt articles and bare nouns. As I hope to have made clear, however, this should not be taken to imply that bare nouns constitute a separate category, nor that number and definiteness are not the primary features expressed by the articles. The forms of the Sranan articles are not reflexes of the English articles, and the Sranan article system is not a continuation of that of English. Rather, Sranan has developed its own system, involving grammaticalized forms of the numeral wan and of the erstwhile potentially demonstrative forms da and den as articles. The fact that these articles are not fully grammaticalized implies that relatively many nouns are not accompanied by any article, and that bare nouns are open to a wider range of readings than English bare nouns. Sranan shares with its major substrate, the Gbe langùages,

Chapter 12. Bare nouns and articles in Sranan 

the common occurrence of bare nouns as well as the fact that these can be regarded as undetermined with respect to number, definiteness and referentiality. To the extent that overt articles in Sranan have a different distribution than the Gbe determiners however, the Sranan system is not a continuation of Gbe either. Thus, Sranan has developed its own system of article usage, in which articles express number and definiteness, and where, as a corollary of the fact that the articles are not obligatory, bare nouns are open to a variety of interpretations.

References Aboh, E. O. 1999. From the Syntax of Gungbe to the Grammar of Gbe. Sierre, Switzerland: Editions à la Carte SA. Aboh, E. O. 2003. D: On the modularity of substrate transfer. Paper presented at Workshop From Alada to Paramaribo 1651 to 1750: What happened to the language?, April 2003, NIAS, Wassenaar. Albitrouw, I. 1894–1915. Tori vo dem bigin vo Anakee en moro fara. Published in Tori foe da bigin foe Anake: Verslag van een Messianistische Beweging [Bronnen voor de studie van Bosneger Samenlevingen 2], M. Sterman (ed.), 28–80. Utrecht: Centrum voor Caraïbische Studiën, 1978. Arends, J. 1995a. Demographic factors in the formation of Sranan. In The Early Stages of Creolization, J. Arends (ed.), 233–285. Amsterdam: John Benjamins. Arends, J. 1995b. Introduction to ‘Part I: The Sranan texts’. In Early Suriname Creole Texts. A collection of 18th-century Sranan and Saramaccan documents, J. Arends and M. Perl (eds), 11–71. Frankfurt: Vervuert. Baptista, M. 2002. The Syntax of Cape Verdean Creole. Amsterdam: John Benjamins. Bickerton, D. 1981. Roots of Language. Ann Arbor MI: Karoma. Bickerton, D. 1984. The language bioprogram hypothesis. The Behavioral and Brain Sciences 7: 173–221. Bruyn, A. 1995. Grammaticalization in Creoles: The development of determiners and relative clauses in Sranan [Studies in Language and Language Use 21]. Amsterdam: IFOTT. Bybee, J. 1994. The grammaticization of zero. Asymmetries in tense and aspect systems. In Perspectives on Grammaticalization, W. Pagliuca (ed.), 235–254. Amsterdam: John Benjamins. Christie, P. 1998. Noun phrases in Dominican Creole. In Studies in Caribbean language II. Papers from the Ninth Biennial Conference of the Society for Caribbean Linguistics, 1992, P. Christie, B. Lalla, V. Pollard & L. Carrington (eds), 265–279. St. Augustine, Trinidad: Society for Caribbean Linguistics. Donicie, A. 1954. De Creolentaal van Suriname. Spraakkunst. Paramaribo: Radhakishun. de Drie, A. 1985. Sye! Arki tori!, compiled by T. Guda. Paramaribo: Afdeling Cultuur Studies van het Ministerie van Onderwijs, Wetenschappen en Cultuur. van Dyk, P. [Ca1765]. Nieuwe en nooit bevoorens geziene onderwyzinge in het Bastert Engels, of Neeger Engels, zoo als het zelve in de Hollandsze Colonien gebruikt word. Amsterdam: Erven de Weduwe Jacobus van Egmont. Essegbey, J. 1999. Inherent Complement Verbs Revisited: Towards an understanding of argument structure in Ewe. Nijmegen: Max-Planck-Institut für Psycholinguistik. Focke, H. C. 1855. Neger-Engelsch Woordenboek. Leiden: P. H. van den Heuvell.

 Adrienne Bruyn Givón, T. 1981. On the development of the numeral ‘one’ as an indefinite marker. In MIT Working Papers in Linguistics 3: Theoretical issues in the grammar of Semitic languages, H. Borer and Y. Aoun (eds), 233–255. Cambridge MA: Massachusetts Institute of Technology. Givón, T. 1984a. Syntax. A functional-typological introduction, Vol. 1. Amsterdam: John Benjamins. Givón, T. 1984b. The pragmatics of referentiality. In Georgetown University Round Table on Languages and Linguistics 1984. Meaning, form, and use in context: Linguistic applications, D. Schiffrin (ed.), 120–138. Washington DC: Georgetown University Press. Givón, T. 2001. Syntax. An introduction, Vol. 1, revised edn. Amsterdam: John Benjamins. Herlein, J. D. 1718. Beschrijvinge van de Volk-plantinge Zuriname. Leeuwarden: Injema. van der Hilst, E. & B. Roest (transl.). 1988. Leysipisi fu den sonde nanga den fesadey. Paramaribo: Lomsukerki. Hopper, P. & Martin, J. 1987. Structuralism and diachrony: The development of the indefinite article in English. In Papers from the 7th International Conference on Historical Linguistics, A. Giacalone Ramat, O. Carruba & G. Bernini (eds), 295–304. Amsterdam: John Benjamins. Huttar, G. L. 1975. Sources of creole semantic structures. Language 51: 684–695. King, J. 1891–94. Diaries as published in Life at Maripaston, H.F. de Ziel (ed.). The Hague: Martinus Nijhoff 1973 [Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde 64]. Kouwenberg, S. 1994. A Grammar of Berbice Dutch Creole. Berlin: Mouton de Gruyter. Kraag, C. 1894–96. Copenkrisi 1894–1896. Published in C. de Beet and M. Sterman, Aantekeningen over de Geschiedenis van de Kwinti en het Dagboek van Kraag (1894–1896) [Bronnen voor de studie van Bosneger Samenlevingen 6], 34–46. Utrecht: Centrum voor Caraïbische Studies, 1980. Kramp, A. A. 1983. Early Creole Lexicography: A study of C.L. Schumann’s manuscript dictionary of Sranan. PhD Dissertation, Leiden University. Lefebvre, C. & Brousseau, A.-M. 2002. A grammar of Fongbe. Berlin: Mouton de Gruyter. Leysipisi 1988 see: van der Hilst & Roest 1988. Lyons, C. 1999. Definiteness. Cambridge: CUP. Migge, B. 1998. Substrate influence in creole formation: The origin of give-type serial verb constructions in the Surinamese Plantation Creole. Journal of Pidgin and Creole Languages 13: 215–265. Mufwene, S. S. 1981. Non-individuation and the count/mass distinction. In Papers from the 17th Regional Meeting of the Chicago Linguistic Society [CLS 17], R. A. Hendrick, C. S. Masek & M. F. Miller (eds), 221–238. Chicago: Chicago Linguistic Society. Mufwene, S. S. 1984. The count/mass distinction and the English lexicon. In Papers from the Parasession on Lexical Semantics, D. Testen, V. Mishra & J. Drogo (eds), 200–221. Chicago: Chicago Linguistic Society. Mufwene, S. S. 1986. Number delimitation in Gullah. American Speech 61: 33–60. Schumann, C. L. (transl.). 1781. Die Geschichte unsers Herrn und Heilandes Jesu Christi, aus den vier Evangelisten zusammengezogen [Moravian Archives Utrecht 12/617], durch S. Lieberkühn; in Neger-Englische Sprache übersezt, zum Gebrauch bey der Neger-Gemeine. Ms. Paramaribo. Schumann, C. L. 1783. Neger-Englisches Wörter-Buch [Moravian Archives Paramaribo 12/648]. Editio tertia. Ms. Paramaribo. Smith, N. S. H. 1987. The Genesis of the Creole Languages of Surinam. PhD Dissertation, University of Amsterdam.

Chapter 12. Bare nouns and articles in Sranan  Smith, N. 2001. Voodoo Chile: Differential substrate effects in Saramaccan and Haitian. In Creolization and Contact, N. Smith & T. Veenstra (eds), 43–80. Amsterdam: John Benjamins. Voorhoeve, J. 1953. Voorstudies tot een Beschrijving van het Sranan Tongo (Negerengels van Suriname). PhD Dissertation, University of Amsterdam. Voorhoeve, J. & Kramp, A. 1982. Syntactic developments in Sranan. Paper presented at the Fourth Biennial Conference of the Society for Caribbean Linguistics, Paramaribo. Westermann, D. 1907. Grammatik der Ewe-Sprache. Berlin: Dietriech Reimer (Ernst Vohsen). Weygandt, G. C. 1798. Gemeenzaame Leerwyze om het Basterd of Neger-Engelsch op een Gemakkelyke Wyze te Leeren Verstaan en Spreeken. Paramaribo: Beeldsnyder. Wilner, J. (ed.). 1994. Wortubuku fu Sranan Tongo, 3rd trial edn. Paramaribo: Summer Institute of Linguistics. Winford, D. 2000. Tense and aspect in Sranan and the creole prototype. In Language Change and Language Contact in Pidgins and Creoles, J. McWhorter (ed.), 383–442. Amsterdam: John Benjamins. Wright, S. & Givón, T. 1987. The pragmatics of indefinite reference: Quantified text-based studies. Studies in Language 11: 1–33. de Ziel, H. F. (ed.). 1973. Life at Maripaston [Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde 64], by Johannes King. The Hague: Martinus Nijhoff.

 

Aspects of the syntax and semantics of bare nouns in Jamaican Creole Michele M. Stewart In this paper I argue that there is no plural marking in Jamaican Creole (JC), since post-nominal dem, traditionally analyzed as a marker of plurality, is instead best considered to be what I call an Inclusiveness marker. As a consequence of the lack of plural marking, all the interpretations of both singular and plural noun phrases in a number-marking language such as English, may be assigned to the JC noun phrase. I argue that this availability of multiple sense possibilities points to the need for functional structure above the NP, which projects in a plural interpretation, when there is a need to individuate. I extend the analysis to definiteness and indefiniteness, which the data show may exist even when they are not phonetically realized. Thus, there is also functional structure above the JC noun phrase which regulates the expression of definiteness even where there is no overt manifestation of this phenomenon. Noun phrases lacking overt functional modifiers, traditionally “bare” noun phrases, then, are not considered to be structurally bare if they express definiteness or indefiniteness.

Introduction In this paper, I argue that there is functional structure above the noun phrase in Jamaican Creole (JC) which regulates the expression of number and definiteness even where there are no overt manifestations of these phenomena.1 Noun phrases lacking

. Unless otherwise attributed, in this paper I am reporting on results of original fieldwork conducted by me between November 2003 and February 2005. This fieldwork also formed the basis of my PhD dissertation, recently approved for award by UWI, Mona. I would like to thank the members of my committee for their contributions to the development of my thinking, my examiners for their very useful comments, and especially my thesis supervisor, Silvia Kouwenberg, who also made valuable contributions to this paper. Finally, my thanks go to the editors, Marlyse Baptista and Jacqueline Guéron whose questions and comments served to greatly improve the paper.

 Michele M. Stewart

overt functional modifiers, traditionally “bare” noun phrases (BNPs), may receive a plural individual interpretation without plural marking via a null functional head. Similarly, BNPs are not considered to be structurally bare if they express definiteness or indefiniteness. In Section 1, I argue that there is no plural marking in JC, showing that the JC post-nominal dem, traditionally analyzed as a marker of plurality,2 is instead best considered to be what I call an Inclusiveness marker. The idea here is that the marking of plurality is not the primary function of this element, but that plurality is subsumed in the concept of inclusiveness. Section 2 explores the semantic consequences of the lack of (overt) number marking, and reviews the demands which the phenomenon makes on the syntax. Without such marking, the noun enters the derivation underspecified for number. As a result, there exists the possibility of assigning to the JC noun phrase all the interpretations of both singular and plural noun phrases (non-individuated as well as individuated) in a number-marking language such as English. I argue that this availability of multiple sense possibilities, points to the need for functional structure above the NP, which projects when there is a need to individuate, as is the case in a plural interpretation. Following Borer (2005), the individuating function is housed in its own universally available functional projection Cl(assifier)P, which houses the plural –s morpheme in English, an overt classifier in a language such as Chinese, and, I claim, a covert classifier in JC. Section 3 extends the analysis to the distinctions of definiteness and indefiniteness, which the data show may exist even when they are not phonetically realized. The definite article di, following traditional analyses, is housed in D, but the indefinite wan is considered to be a numeral in Num(ber)P. JC nouns may surface undetermined (“bare”) from the derivation, but where they have the same properties as determined nouns, they are in fact not structurally bare. Similarly, the bare noun in its indefinite reading is thought to project NumP, and where individuation takes place, this is achieved via ClP. Existential readings require only ClP; only generic readings are considered to be structurally bare.

.

Post-nominal dem as an Inclusiveness Marker

Here, I will argue that JC dem in the sequence di () + N + dem does not have the properties of a plural marker. This will provide background to Section 2 where I will show that plurality is expressed without overt plural marking, via functional structure above the NP.

. Also see the function of dem as a deictic marker in Devonish & Pochard (1988).

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

Traditional analyses have been based on differences between sentences such as (1a) where the JC sequence di + N + dem may not refer to a singular individual, and (1b), where di + N must do so.3 (1)

a.

Di gyal dem gaan a maakit.  girl  gone to market4 ‘The girls have gone to the market.’ *‘The girl has gone to the market.’ b. Di gyal gaan a maakit. *‘The girls have gone to the market.’ ‘The girl has gone to the market.’

Rijkhoff (2004: 36) provides cross-linguistic evidence for number being expressed in a portmanteau element which simultaneously expresses features such as gender, definiteness and/or case. He offers Caribbean Creole languages as cases where the expression of one notion, definiteness (here in the form of di), necessarily involves the expression of another notion (here expressed by dem). In contrast, in English and other inflectional plural-marking languages, there is no constraint on the appearance of a marker of definiteness with N+plural morpheme. Indeed, the constraint that number-marking be restricted to definite NPs is curious. Importantly, plurality can be expressed without the use of dem in referring NPs: (2)

a.

b.

Chii bwai kil dem faada. three boy kill 3 father ‘Three boys killed their father.’ Di chii bwai dem kil dem faada.  three boy  kill 3 father ‘The three boys (together) killed their father.’

The use of dem appears not to be totally arbitrary. Mufwene (1981) links its presence with (scalar) individuation. In this scheme, the most individuated NP is considered

. Note that in a noun phrase restricted by a relative clause, a plural interpretation is possible, as in (i) below. It seems clear, however, that this does not have a referential reading: (i)

Watch out fi di big taim tiif uu kliem se dat dem smaat. watch out for  big time thief who claim say that 3 smart ‘Watch out for the big-time thieves who claim that they are smart.’ [Adapted from Richie Spice’s ‘Earth a Run Red,’ lyrics available at http://www.lyricsandsongs.com/song/499346.html]

. Abbreviations used in this paper are as follows: 1 – 1st person singular; 3pl – 3rd plural; 3s – 3rd singular;  – aspect marker;  – copula;  – definite article;  – demonstrative;  – inclusiveness marker;  – indefinite;  – locative;  – plural.

 Michele M. Stewart

to be marked with the definite (di ‘the’) or indefinite (wan ‘a’). Plural marking with di … dem is considered to be less individuated, and bare nouns, least individuated: Marked. Sg Count – most individuated: (3)

Di mango/wan mango depan a raip nou.  mango/ mango be-on  ripe now ‘The mango/one/a mango is ripening now.’

Marked. Pl Count – less individuated: (4)

Di mango dem depan a raip nou.  mango  be-on  ripe now ‘The mangoes are ripening now.’

Unmarked – non-individuated: (5)

Mango depan a raip nou. mango be-on  ripe now ‘Mangos are ripening now.’

Contra Mufwene, however, I claim that di … dem does not express plurality in the same way that plural marking does in a language such as Engish. Indeed, the reason for Mufwene’s intuition that NPs with di … dem are less individuated, is that this construction marks inclusiveness (), and not plurality. There is an intuition, for example, it is likely in (2b) that the killing event represents one incident, but this is not necessarily the case in (2a). I claim that the naturalness of the use of the plural with the JC definite di is easily explained when it is considered that inclusiveness is a feature of definiteness. Thus for Lyons (1999: 11ff), inclusiveness is relevant to definiteness in the case where the reference of plural and mass nouns is to the totality of the objects or the mass. What this means is that in di bwai dem (‘the boys’), dem indicates that it is the totality of bwai which is under consideration. What is taking place in (2), then, is that post-nominal dem marks a group reading. Its plural construal is a consequence of it being a marker of inclusiveness – a group of boys must necessarily denote (semantic) plurality. Other restrictions have been noted by Patrick (2001: 37). These serve to provide further support that dem cannot be considered to be a marker simply of plurality, since unlike such a marker, it pluralizes count nouns with certain semantic features only. Patrick claims, for example, that di … dem appears more often with nouns denoting humans than animals, and more often with nouns denoting animals than inanimate objects:5 (6)

Di man dem dig di huol an di uman dem plant di kaan.  man  dig  hole and  woman  plant  corn ‘The men dug the holes and the women planted the corn.’ [Sistren 1986: 48, cited in Patrick 2001: 36]

. Patrick notes this as a tendency. I am not sure how systematic the distinction is, since it was not included in my fieldwork. I leave that to further research. It is not crucial to my argument,

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

Patrick (2005: 34) suggests that this may be related to dem’s history of grammaticalization from a third person plural pronoun with primarily human reference. In addition, possibly also since its historical antecedent is a third person pronoun, dem is only available for third person referents, not first person or direct address: (7) *Aal yuu bwai dem all you boy  ‘All you boys’

[Patrick 2005: 33]

Finally, it is well-known that post-nominal dem is also used in constructions such as Mieri dem [Mary ] where it receives the interpretation ‘Mary and her friends’ or, perhaps, ‘Mary and her family,’ depending on the context. Interestingly, it would seem that the composition of the group need not be fixed, but that Mary’s name might simply be a convenient way of identifying some arbitrary group, whose membership may be unknown, but for Mary. I suggest that di is not necessary in such cases, since Proper Names are themselves rigidly referring. Dem could not reasonably be said to be marking the plurality of a proper name.6 To claim ambiguity of dem would be uneconomical. Instead, an analysis in terms of inclusiveness, where reference is to the group of persons in which Mary is included, allows for a single interpretation in all instances. I stress that JC di … dem does entail plurality. As the ungrammaticality of the b. sentences below shows, groups must have plural reference: (8)

a.

Di bwai dem a Jan an Maak.  boy   John and Mark ‘The boys are John and Mark.’ b. *Di bwai a Jan an Maak.

(9)

a.

Di bwai a Jan.  boy  John ‘The boy is John.’ b. *Di bwai dem a Jan.

Semantically, inclusiveness is not closure under summation.7 Instead, inclusiveness involves group formation. A group requires a grouping criterion, a property that objects

but, if true, serves only to provide yet another difference between the JC dem and a traditionally understood plural morpheme. Interestingly, it is well-known that nominal expressions with the plural suffix –men in Chinese must have human reference (see Iljic 2005: 79, for example). This, as well as other properties, have caused –men to be re-analysed as a collective. . This use seems restricted to Proper Names. Di kwiin dem, for example, may not refer to the Queen and her entourage, but must mean ‘the queens’. . Closure under sum formation, or summation, is taken to mean that whenever x and y are members of De (the domain of individuals) then x + y (the sum of x and y) is a member of De (Kratzer 2001: 2).

 Michele M. Stewart

have to satisy in order to belong to the group (Chierchia (1998: 63)). For instance, being associated in some way with Mieri is the property which membership in the group Mieri dem requires. In summary, one would expect that if dem were a plural marker, it would be required to yield a plural reading, and that it would be blind to the various properties of the NP which it modifies. Instead, we have seen that the plural is often unmarked and that dem, where it appears, is constrained to appear in definite NPs only. Also, where dem appears, a single event or group reading seems to be implied. This is unexpected for a true number marker, which typically marks individuation, or the partitioning of the denotation of noun into discrete atomistic entities. That dem does not mark individuation is further supported by the fact that it cannot be used in distributive contexts. Finally, it is well known that dem occurs also as an associative plural with proper names. I conclude that in JC, there is no morphological distinction for number, and discuss in the following section, the implications that this has for the specification of number.

.

The expression of Number in JC

Bach (1989: 78ff)8 suggests that in languages where there is no morphological distinction for number, the noun gathers together everything that in English is described by both the singular (‘horse,’ say) and the plural (‘horses’). Applied to JC, the denotation of the noun haas is the union of the denotations of English ‘horse’ and ‘horses.’ Using Link’s (1983) terminology, we would say that such a noun covers both atomic and non-atomic individuals. Each atom represents an element which resists decomposition. It is into atoms that individuation partitions the denotation of the noun. The JC data show that, as predicted by Bach, the unmarked noun may receive either an individuated or a non-individuated interpretation. (10) may, for example, quite felicitously, be followed discursively by (11a) revealing either a singular or plural individual interpretation, as in (10a). We know that the noun has undergone partitioning, since the anaphoric pronoun bears number – dem makes clear reference to plural individuals, and i to a singular individual. This is partitioning without overt marking. In contrast, with (11b) as a rejoiner, it is clear that (10) has taken on a part-of reading to yield (10b), and has not therefore undergone partitioning. (10)

.

Mi 1 a. b. c.

iit mango yeside. eat mango yesterday ‘I ate mangos/a mango yesterday’ ‘I ate a piece of a mango yesterday.’ ‘I ate mango yesterday.’

Bach’s position is summarized nicely in Portner (2005: 97–98).

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

(11)

a.

b.

Dem/i9 swiit, yu si! 3/3 sweet you see ‘They were/it was so sweet!’ Piis liiv uova iina di pliet. piece leave over in  plate ‘There is a piece remaining in the plate.’

This has consequences, of course, not only for plural interpretations, but for the relevance of the count/ mass distinction in JC, since NPs containing singular or plural count nouns are said to be built out of “atoms” and in contrast, NPs containing mass nouns are non-atomic. While (11a) reveals the possibility of either a singular or a plural (count) reading as in (10a), the mass interpretation (10c) is also possible for the same noun in the same context. I note that a further non-count part-of interpretation (10b) is quite natural here. That both count and mass interpretations are available for the unmarked noun seems naturally to explain the ease with which in JC nouns traditionally considered to be mass, are able to be measured without the help of any overt measure phrase:10 (12)

a.

b.

Milk de aal baut. Milk  all about ‘There’s milk all over the place.’ Sel mi wan milk de. sell 1  milk  ‘Sell me a (bottle/ box/ packet/ can/ glass of) milk.’

If it is the case that the denotation of a noun contains atomic and non-atomic individuals as well as sums of individuals, then what this must mean, is that nouns entering the derivation are underspecified11 for number. The ability to incorporate both singular and plural denotations must, of course be regulated. This is achieved by plural marking in a language such as English. In languages such as Chinese where there is no built-in morphological distinction for

. The singular/plural distinction is also neutralized with what I have labelled the 3rd singular pronoun i. It can therefore actually be interpreted as either ‘It was so sweet’ or ‘They were so sweet.’ The 3rd person plural subject pronoun dem cannot have singular individual reference. . See Stewart (forthcoming) for a more detailed discussion of individuation and ‘mass’ nouns. . I differentiate between unspecified and underspecified. If nouns are unspecified for a feature (Homogeneity, say), it means that this feature is irrelevant, and will at no point be valued, since it is not included in the item’s repertoire of features. In contrast, if a noun is underspecified for a feature (Number, say), it means that this feature must be valued in the course of the derivation, since it contributes either to the syntax or the semantics, or both.

 Michele M. Stewart

number, the ambiguity arising from multiple possibilities is resolved by means of the classifier system, which matches the right kind of classifier (counter), with each type of noun. The classifier for ‘horse’ in Chinese (pi) restricts the denotation of the noun to just the atoms, so that when used, the noun phrase in question must refer to the individuals. This is a way of introducing explicitly the difference between the denotations of singular and plural NPs.12 In JC there is no overt classifier system, yet individuation clearly occurs without such marking. I suggest, therefore, that the projection whose job it is to individuate in JC is the Classifier Phrase (ClP), but that its head is not phonetically realized. Interestingly, individuation in JC is not required with a numeral. Rijkhoff (2004: 38) calls nouns in such languages ‘set nouns,’ since nouns not normally marked for number when in direct construction with a free numeral modifier seem to denote a set of individuals. Rijkhoff includes in languages with set nouns those where plural forms do exist, but need not be used to derive a plural meaning. In such languages, then, a Num+N sequence denotes an X-numbered set. This is different from a plural individual with an X-number of atomic parts which, as we have seen, the plural morpheme in English returns. This explains why it is that in (13) below, the a. interpretation is not possible in JC. Instead, chii man denotes a set of men described as having three members, which participated in a carrying event involving one suitcase.13 Here, there is counting without individuation. (13)

Chii man kyari wan grip. three man carry  suitcase ‘Three men carried a suitcase.’ a. *‘Three men each carried a suitcase.’ b. ‘Three men (together) carried a (one large) suitcase.’

Rijkhoff ’s (2004) claim is that nouns in such languages are marked with the (semantic) feature [+Shape]. This means that they have a definite outline in the spatial dimension, and so can be identified as discrete entities. With regard to the semantic category [±Homogeneous], set nouns are said to be neutral.14 Being neutral to Homogeneity means that this feature is irrelevant.

. For Rijkhoff, such nouns are classified separately as sort nouns, since the classifier is considered to be sortal. I follow Bach (1989) who considers that classifiers are a means of distinguishing between atomic and non-atomic interpretations in a language where this cannot be achieved through the overt morphology. .

In this context, grip which can be either a small or a large suitcase, is interpreted as large.

. ‘Water’ is an example of a [+Homogeneous] noun. It is cumulative, in that if we add some more water to water in a glass, we still refer to it as ‘water.’ It is also dissective, to an extent, in that after we have drunk some of the water in the glass, the glass will still contain water.

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

The result is that the noun designates a property that is characterized only as having a definite shape: if the set comprises more than one individual, then it can ultimately be divided into as many singleton sets as there are individuals; in the case of a singleton set, the space for which the property holds cannot be further divided (Rijkhoff 2004: 52). It is for this reason that an individuated interpretation of a numeral NP is also possible. In (14), for example, individuation is forced by the use of reduplicative wan-wan. (14)

Chii man kyari wan-wan grip. three man carry one-one suitcase ‘Three men each carried a suitcase.’

Within DP, then, there must reside at least the function to individuate (or not) – determining whether the range is to be constituted by members of the extension, or by parts of its members. Individuation is not the only function involved, however. In the case of a cardinal, for instance, first the noun may be individuated before a number specification is applied to it. Importantly, even where there is no counter morpheme, the syntactic evidence points to the (optional) existence of partitioning, as seen in (10) above. I assume that each of these operations of partitioning and counting is the responsibility of its own individual head, rather than both being housed in a single head. If both functions were assigned to a single head, this would strongly predict that the two operations will be either both active or inactive,15 which is not the case in JC, as we have seen. What this means is that the notion of semantic number is built in stages inside DP, through the combined semantic effect of two independent functional heads, ClP which prepares the noun for counting (or not) by partitioning the members of the set into individuals, and a second, NumP, dominating it, which is (optionally) responsible for the assignment of quantity. The proposal, then, is for the (here, partial) JC nominal structure in (15), with the optional projection of NumP, ClP, and, as we shall see in Section 3 below, DP. (15)

[DP D [NumP Num [ClP Cl [NP N]]]]

Recall my analysis in Section 1 of dem as an instantiation of group-formation. What I now suggest is that ClP houses the individualizing and the group-forming functions, and that only the latter is overt in JC in the form of dem. Thus, dem would be considered to be the overt counterpart of the covert individuating Classifier.16 I consider wan to be a numeral residing in Num and note that this is in line with recent re-analyses of the indefinite as the spell-out of singular Num (see, for example,

. See Heycock/Zamparelli (2005: 227), where this argument is applied in a different context. .

Since dem is post-nominal, this would necessitate movement of NP to [spec, ClP].

 Michele M. Stewart

Munn/Schmitt (1999: 7) for English ‘a,’ and Kouwenberg (this volume) for Berbice Dutch en). A set can have any cardinality – a multiple set with more than one individual, but also a singleton set with only one member (Rijkhoff 2004: 46). Like numerals, therefore, when ClP is not projected in a wan NP, a set reading (in this case, a singleton set) is returned. In contrast, for wan to return an individual, ClP must project. The contrast is made particularly clear in (16) below. (16)

Evri bwai mek wan tiebl. every boy make  table a. ‘Every boy (independently) made a (different) table.’ b. ‘All the boys (jointly) made a table.’

On the distributive reading in (16a), the object ClP will project, individuating its complement tiebl. This allows for the pairing of sub-events (mek) and individuated themes (wan tiebl). I note that each (mek+wan tiebl) consitituent will then in turn require its own agent. This will be achieved by the projection of ClP in the subject, causing the individuation of bwai, and allowing for evri to quantify over those individual boys. The indefinite wan need not be individuated, however. As seen in (16b), when evri is interpreted collectively, a singleton set for the object wan NP is returned. In this case ClP is not projected.17 To sum up, my assumption is that the noun enters the derivation underspecified for number, as the representation of the space for which the property holds. It is considered to be a set, in the Rijkhoff (2004) sense, and the feature [+Shape] allows individuation to take place. Possibilities are as follows: -

-

ClP projects in the case of individuation. NumP projects if the counting function is applied. In such a construction there are two available interpretations, however. If the noun is being described in terms of the number of members which make it up (an X-membered set), then ClP will not project, since individuation has not taken place. In contrast, in the case of singular or plural individual interpretations involving the partitioning of the denotation of the noun, ClP as well as NumP will project. a noun which is neither individuated nor counted,18 is considered to have no functional structure by way of ClP or NumP above it.

Now we will see that definiteness signalled overtly with the determiner di in D may also be expressed in the absence of overt marking. In addition to the possibility of null heads in ClP and NumP, this points to the projection of DP with a null D. . The ability of JC evri to be interpreted collectively is discussed extensively in Stewart (2006), where it is concluded that this interpretation is best not explained as a scopal (specific) reading of an indefinite in interaction with the distributive universal. . An example of a noun which is neither individuated nor counted, is one traditionally classified as ‘mass.’ See Section 3 for some discussion of the count/mass distinction.

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

.

The expression of (In-)definiteness in JC19

What we have seen is that there is no overt marking of individuation in JC, but that it is achieved by a ClP whose head is null. Where individuation does not occur, ClP does not project. Numerals including wan are housed in NumP, and may, but need not select an individuated NP. I now show that the same pattern emerges with regard to the expression of (in-) definiteness, since these properties need not be overtly marked in JC. Considerations center specifically, then, around phenomena associated with BNPs in JC, where we will also see that the ability of the noun to include both singular and plural individuals in its denotation allows for bare singular count NPs, not normally attested in English. As background, for an expression to be definite, the speaker must be in a position to assume that the hearer is able in some way to select the referent. This has been referred to as the Identifiability condition of definiteness.20 The speaker may be able to pick out the referent either through reference at the level of the discourse, or as a result of shared knowedge. The view of definiteness that the entity denoted by the DP must be familiar to both speaker and hearer is known as the condition of familiarity. Importantly, familiarity is subsumed under identifiability – when present, it is familiarity which allows the hearer to identify the referent. According to this characterization, what differentiates a definite from an indefinite NP is only the ability or inability of the hearer to identify the intended referent by some means.21 The concept of individuation as defined in Section 2, is not a characterizing feature of definiteness, since definite expressions can be either individuated or non-individuated. The following are straightforward cases of definite (17) and indefinite NPs (18). (17)

Di man gaan lang taim.  man gone long time ‘The man left a long time ago.’

(18)

Wan man riich aan.  man reach on ‘A man arrived.’

Recall that the JC indefinite article is wan, housed in Num. In contrast, definite di heads DP. In (18), for example, my assumption is that NumP and not DP projects.

. Data in this section was supplemented by contributions from native speakers, including students taking my Semantics course at UWI. .

Much of my characterization of definiteness is inspired by Lyons (1999).

. Lyons (1999) also characterizes definites in terms of inclusiveness and uniqueness. Inclusiveness is discussed at some length in Section 2. In the case of singular nouns, uniqueness indicates that there is just one object satisfying the description.

 Michele M. Stewart

This differs from more traditional accounts where both the indefinite and definite articles are housed in D, and the ‘bareness’ of NPs, characterized by the absence of an overt D, includes consideration of both these articles. Bareness is said to interact with the count/mass distinction. It is well-known, for instance, that if the determiner is not realized in English, the bare noun is normally either plural count or mass, and that a singular count NP with an unrealised determiner is usually not available in similar environments (Krifka et al 1995: 67).22 (19)

a. Dogs are lovely animals. b. Gold is my favourite metal. c. *Dog is a lovely animal.

Discussions in Section 2 would lead us to expect that the JC bare noun would not be restricted in its interpretation as a result of the count/mass distinction and its interaction with the expression of number. This is because the noun enters the derivation underspecified for number, and because, given the structure of the denotations of nouns, both count and mass interpretations are readily available. At least since Longobardi (1994), much attention has also been given in the literature to accounting for differences between languages as to the possibility of the appearance of BNPs in argument vs non-argument (predicative) positions. For example, most nominals are observed to require a determiner in English, as well as in Romance languages, when appearing in argument position. In what follows, I will show that bare singular NPs may appear in JC without expected restrictions either concerning the mass/ count or the argument/predicate distinctions. I assume that where definiteness or indefiniteness are expressed without overt markers of these phenomena, non-overt heads in DP or NumP project. Investigations will center, therefore, around the environments in which these may arise. As predicted, we see below that bare NPs are not restricted to plural interpretation. Singular count nouns in JC can be bare, for example, and may have definite or indefinite interpretations in both subject and object positions. My assumption is that NumP projects for an indefinite reading, and DP for a definite reading. Individuation is achieved separately as required, with the projection of . In (20a) reference is, by shared knowledge, to a particular river in a community, making the definite interpretation quite natural, and resulting in the projection of DP. In contrast, where the speaker may be unsure as to which river has overflowed, the interpretation (20b) could be the indefinite ‘a river.’ In this case NumP and not DP will project.

. An exception mentioned by Krifka et al is the English ‘man.’ Even so, it would appear here to be used as a mass noun or in its natural kind sense, synonymous with ‘mankind.’ In my account, I do not address bare singular count nouns in environments such as complement to Ps as in ‘She just arrived from hospital’ or in conjoined constructions such as ‘Mother and father alike.’

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

(20)

Riba kom dong. river come down a. ‘The river has overflowed its banks.’ b. ‘A river has overflowed its banks.’

In (21) reference is to an activity of turning on the radio which given the context can be identified as the one that is turned on every morning. Both these sentences contain events involving a specific participant which can be identified by the hearer, and the result is a singular referring (definite) interpretation and the projection of DP. (21)

Fos ting a maanin taim im ton aan riedyo. first thing in morning time 3s turn on radio ‘The first thing he does in the morning is turn on the radio.’

Similarly in (22) given the linguistic context of the dropping of coconuts from a particular tree (di chrii) such that there are no more on the tree, a plural definite interpretation for the BNP is forced, and an indefinite or generic interpretation is blocked. (22)

Kokonot drap so tel di chrii no ha no muo. coconut drop so until  tree not have no more ‘The coconuts fell until there were no more on the tree.’

In (23) below, however, the interpretation of the object BNP is less restricted, and interestingly, may not be definite. (23)

Richi tek egzam yeside. Richie take exam yesterday a. ‘Richie took an exam yesterday.’ b. ‘Richie was occupied with exam-taking yesterday.’

The singular indefinite reading is possible, as in (23a), but is not the only one available. There is the intuition that there is an additional reading where there is an abstraction away from any individual exam as in (23b). In this interpretation, the sentence expresses a comment about how Richie occupied himself yesterday – he was involved in exam-taking. Here, the BNP is kind-referring, and not object-referring. The availability of both interpretations is not surprising, given the structure of the denotation of the JC noun, as outlined in Section 2. (23) is reminiscent of Mi iit mango yeside ((10) above) for which a mass interpretation was said to be one of the possible available readings. The characterization of the denotation as a set allows us to speak in a uniform way about both the kind reading of ‘exam’ and the mass reading of ‘mango.’ In the case of a kind interpretation as in (23b),  will not project since there is no individuation, and without a numeral or indefinite interpretation, neither will NumP. In the absence of a definite interpretation, DP does not project, and the result is a structurally bare NP.

 Michele M. Stewart

Importantly, as predicted by my definition of definiteness above, the definite interpretation is only available when contextual information allows the hearer to identify the referent. The speaker understands in (21) that the radio in question is one which is turned on every morning. Without contextual grounding, the hearer is unable in (23) to identify the exam in question, and the indefinite reading results. With more information, as in (24) below, the definite reading does become available. The indefinite interpretation is blocked, as is reference to involvement in general exam-taking, presumably since it is understood that this is not just any exam, but the Grade Six Aptitude Test, , just mentioned. (24)

Jan tuu yong fi , bot Richi tek egzam yeside. John too young for , but Richi take exam yesterday a. #‘John is too young for , but Richie took an exam yesterday.’ b. #‘John is too young for , but Richie was involved in exam-taking yesterday.’ c. ‘John is too young for , but Richie took the exam yesterday.’

Indefinite plural BNPs appear freely in both subject and object positions. In (25a) a generic interpretation would return an anomaly, since the practice is that only young boys wear short pants to school, and the ability to wear long pants marks a coming of age. It would be unexpected, therefore, for boys to generally wear both short and long pants to school. In (25b), the sense of the verb rien doun ensures a plural interpretation for the bare gonshat. (25)

a.

Bwai go skuul iina shaat pans an lang pans. boy go school in short pants and long pants ‘Some boys go to school in short pants while others go in long pants.’ #‘Generally boys go to school in short pants and long pants.’ b. Dem rien doun gonshat pan im. 3 rain down gunshot on 3s ‘They rained down gunshots on him.’ [Patrick 2005: 31 (100) (adapted)]

It is clear that BNPs can appear freely in argument positions in JC with definite and indefinite interpretations, and that they may be either singular or plural. Quantificational readings are also possible for singular and plural BNPs in JC. The BNPs in (26) and (27), for example, appear in existential constructions with temporary (stage-level) predicates. These predicates denote properties or actions which apply at a particular time, but could not be said to be enduring. (26)

Bad man dem taim-de! bad man  time- ‘There were bad guys around in those days.’ [Patrick 2005: 32 (105) (adapted)]

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole 

(27)

Daag iina di yaad. dog in  yard ‘There’s a dog/ there are dogs in the yaad.’

I assume that the existential BNP is a description with a free variable, with no quantificational force of its own. If the BNP has no implicit quantification, then what daag contributes in (27) is only DAAG(x), not [∃x:DAAG(x)]. Structurally,  does not project since definiteness is not relevant, and neither does NumP since this is not an indefinite NP. ClP will be required to individuate the noun. BNPs with permanent (individual-level) predicates denoting enduring properties as in (28) take on a generic reading. (28)

Waya a paizn ting. wire  poison thing ‘Wire is a poisonous thing.’ [Bailey 1966: 27 (adapted)]

A generic BNP makes a generalization about things of the kind that the noun denotes. Semantically genericity is introduced by the generic operator Gen; structurally the NP is truly bare in the absence of definiteness (DP), indefiniteness (NumP) as well as individuation (ClP). I note that although both bare NPs (29) and the definite di (30) may be used to evoke a generic reading, (31) shows that the use of the indefinite does not return a generic interpretation as it does in English. This may provide further evidence for the numeral status of wan in JC, since it would not be expected that a numeral NP would evoke a generic reading. (29)

Manguus chrikifai. mongoose cunning ‘Mongooses are cunning.’

(30)

Di  a. b.

(31)

Wan manguus chrikifai.  mongoose cunning a. ‘A mongoose (that I saw yesterday) is cunning.’ b. *‘Mongooses are cunning.’

manguuss chrikifai. mongoose cunning ‘The mongoose (over there) is cunning.’ ‘Mongooses are cunning.’

Conclusion There is no plural inflectional morphology in JC, and I have shown that the morpheme dem traditionally analyzed as being the post-nominal plural morpheme should be reanalyzed as an inclusiveness marker. What this means is that all the readings associated

 Michele M. Stewart

with singular and plural forms of the nouns in a plural inflectional-marking language such as English are available to the noun in JC without number marking. Following Rijkhoff (2004), such a noun is considered to be a ‘set’ noun. I have argued that where the noun returns an individuated meaning, this is achieved by a null Cl(assifier) head. By the same token, both singular and plural readings are also available for nouns which surface determiner-less in JC, and when individuated, ClP projects. Interestingly, although definite and indefinite articles exist, definite and indefinite readings are available to determiner-less nouns in argument as well as predicate positions, with both singular and plural interpretations. What determines the overt realization of the articles will need to be left to further research.

References Bach, E. 1989. Informal Lectures on Formal Semantics. Albany NY: SUNY. Bailey, B. 1966. Jamaican Creole Syntax. A tranformational approach. London: CUP. Borer, H. 2005. Structuring Sense, Vol. I: In name only. New York NY: OUP. Chierchia, G. 1998. Plurality of mass nouns and the notion of ‘semantic parameter’. In Events and Grammar, S. Rothstein (ed.), 53–103. Dordrecht: Kluwer. Devonish, H. & Pochard J-C. 1988. On the diversification in the function of the deictic markers: A study of a, da and dem in the Jamaican creole noun phrase. English World-Wide 9(2): 213–226. Heycock, C. & Zamparelli, R. 2005. Friends and colleagues: Plurality, coordination, and the structure of DP. Natural Language Semantics 13: 201–270. Iljic, R. 2005. Personal collective in Chinese. Bulletin of SOAS 68(1): 77–102. Kratzer, A. 2001. The event argument. Chapter 4. 1 Apr. 2005 Krifka, M., Pelletier, F., Carlson, G., ter Meulen, A., Chierchia, G. & Link, G. 1995. Genericity: An introduction. In The Generic Book, G. Carlson & F. Pelletier (eds), 1–124. Chicago IL: The University of Chicago Press. Link, G. 1983. The logical analysis of plurals and mass terms: A lattice-theoretical approach. Reprinted in Formal Semantics. The essential readings, P. Portner & B. Partee (eds) 2002, 127–146. Malden MA: Blackwell. Longobardi, G. 1994. Reference and proper names: A theory of N-movement in syntax and Logical Form. Linguistic Inquiry 25(4): 609–665. Lyons, C. 1999 Definiteness. Cambridge: CUP. Mufwene, S. 1981. Non-individuation and the count/mass distinction. In Papers from the Chicago Linguistic Society 17, R. Hendrick, C. S. Masek & M. F. Miller (eds), 221–238. Munn, A. & Schmitt, C. 1999. Bare nouns and the morphosyntax of number. Proceedings of the Linguistic Symposium on Romance Languages. 14 Feb. 2005 Patrick, P. 2005. Jamaican Creole: Morphology and syntax. In B. Kortmann, E. Schneider, C. Upton, R. Mesthrie & K. Burridge (eds) Draft chapter

Chapter 13. Aspects of the syntax and semantics of bare nouns in Jamaican Creole  Patrick, P. 2001. Outline of syntactic structures in Jamaican Creole. In Comparative Creole Syntax, J. Holm and P. Patrick (eds). London: Battlebridge. Portner, P. 2005. What is Meaning? Fundamentals of formal semantics. Malden MA: Blackwell. Rijkhoff, J. 2004. The Noun Phrase. New York NY: OUP. Sistren, with Honor Ford Smith (ed.) 1986. Lionheart Gal: Life Stories of Jamaican Women. London: The Women’s Press, Ltd. Stewart, M. Forthcoming. The expression of number in Jamaican Creole. Stewart, M. 2006. Quantification in Jamaican Creole. The syntax and semantics of evri (‘every’) in interaction with indefinites. PhD University of the West Indies, Mona, Jamaica.

 

African American English

 

NPs in aspectual Be constructions in African American English1 Lisa Green This paper examines indefinite and bare plural NPs in aspectual be constructions in African American English (AAE). Aspectual be constructions are like simple tense generics in that they receive a habitual interpretation. The interpretation of singular indefinite subject NPs in aspectual be constructions depends on the type of predicate that occurs with the subject. On the other hand, bare plural subjects in aspectual be constructions exhibit a similar type of ambiguity associated with bare plurals in generics in English. Bare plurals with individual-level predicates in aspectual be constructions are unambiguous, but bare plurals with stage-level predicates are ambiguous between a habitual and existential habitual reading. It has been noted that NPs in AAE share some patterns with NPs in Creoles, but bare NPs in the two languages are interpreted differently. Analyzing NPs in aspectual be constructions provides an opportunity to learn more about the differences and similarities between the interpretation of NPs in AAE and Creoles.

Introduction This paper considers the interpretation of bare plural and singular indefinite NPs in aspectual be constructions in African American English (AAE).2 Section 1 presents introductory background by highlighting the controversial claim about the Creole origin of AAE and by briefly reviewing the similarities between genitive and associative NPs in AAE and Creoles. Sections 2 and 3 provide general background necessary for characterizing aspectual be constructions in which bare plural and singular indefinite

. I wish to thank Marlyse Baptista and Jacqueline Guéron for their comments. All errors are mine. . See Spears (this volume) for a different treatment and analysis of bare nouns in AfricanAmerican English.

 Lisa Green

NPs that will be discussed in this chapter occur. Section 2 presents a general overview of the syntactic and semantic properties of aspectual be, which marks habitual aspect, not tense. It is distinguished from copula be and auxiliary be, which are inflected for tense. Aspectual be constructions are like simple tense generics in that they receive a habitual interpretation, but they are distinguished from them in other ways. The focus in Section 3 is on a comparison of aspectual be constructions and characterizing sentences, on the one hand, and sentences that express accidental properties, on the other. As explained in this section, aspectual be constructions indicate regular occurrences and express essential properties. Section 4, the major section of the paper, focuses on singular indefinite NPs and bare plurals in aspectual be constructions. One of the major points is that bare plurals in aspectual be constructions reflect the same type of ambiguity associated with bare plurals in generics in English as discussed in Carlson (1977) and Diesing (1992). This is expected, given the extent to which aspectual be constructions are similar to simple tense generics. However, the type of readings associated with bare plurals in aspectual be constructions are slightly different from that of readings of bare plurals in English generics.

.

The Creole issue and a note on previous research on NPs in AAE

The view that AAE originated as a Creole is controversial, and claims about its Creole origins, on the one hand, and English origins, on the other, are well documented in the literature (e.g. Winford 1997, 1998; Rickford 1998; Mufwene 2000; Poplack 2000; Poplack and Tagliamonte 2001). In these debates, features of AAE have been compared to patterns in Creoles as well as to those in other nonstandard English dialects in attempts to show that AAE is closely associated with Creoles according to Creolists or more closely associated with English according to Anglicists. Patterns and properties of verbs and verbal elements are often used in arguments about the origin of AAE; however, properties of NPs have also been discussed in this provenance issue. Mufwene (1998) highlights some similarities and differences between NP patterns in AAE and other varieties of English. He also notes two ways in which NPs in AAE are like NPs in Atlantic Creoles such as Gullah and Jamaican: genitive NPs are not obligatorily marked with -’s (1a), and the associative plural strategy is used, in which the marker nem (1b) forms a part of the plural NP: (1)

a. b.

That’s Earl book. ‘That’s Earl’s book’ Earl nem left already. Literally: Earl and them left already. ‘Earl and the others have left already’

In (1a) the genitive relationship is indicated by order of the possessor followed by the element possessed not by overt -’s marking, and in (1b) the plural NP is formed with a noun and the plural marker nem.

Chapter 14. NPs in aspectual Be constructions in African American English 

On the interpretation of NPs, Mufwene notes that in AAE, bare singular NPs cannot be used to express a singular indefinite NP or bare plural reading. This is in contrast to Creoles such as Jamaican, exemplified in (2) (from Mufwene (1998: 71)): (2)

Daag no nyam dog. ‘A dog does not eat dog (meat)’ ‘Dogs do not eat dog (meat)’

(Also see Schmitt and Munn (1999) for a discussion of bare nouns in Brazilian Portuguese.) The structure of NPs in AAE has received relatively little attention perhaps because they do not appear to be significantly different from NPs in other varieties of English. However, it is clear that there are innovative uses of NPs in registers of AAE used predominantly by those in the adolescent to young adult age group. For instance, singular indefinite NPs can have first person reference as in the following: (3)

They wouldn’t help a sister out. ‘They wouldn’t help me (feminine singular) out’

This register will not be addressed here although there is a great deal to be learned about its NP patterns. It is the case that NPs in AAE are like NPs in other varieties of English in many respects and that bare singular NPs such as that expressed in (2) are not allowed; however, the interpretation of NPs in certain constructions such as aspectual be sentences still need to be considered in determining the range of distribution and readings in AAE and the way in which the interpretations are different from and similar to that of NPs in other varieties of English and in Creoles.

.

An overview of aspectual Be constructions

The major focus of this paper is on NPs in aspectual be constructions, so before moving to a discussion of NPs, I will summarize properties of the marker itself and review general characteristics of aspectual be constructions. This overview is based on Green (2000).

.

Aspectual Be

In AAE, one way of indicating that an eventuality recurs is by using aspectual be, which is also referred to as habitual be, with a predicate, as in (4). (4)

Tamiya be singing in the morning. ‘Tamiya usually sings/is usually singing in the morning’

Here ‘eventuality’ is used as a cover term for states and events along the lines of Bach (1981). Aspectual be is distinguished from copula be and auxiliary be in AAE. In essence, AAE is like other languages, such as Scottish Gaelic (Ramchand 1996) and Irish

 Lisa Green

(Doherty 1996), that distinguish two be’s. One difference is that while copula be and auxiliary be (5a) are inflected for tense and number, aspectual be (5b) is not:3 (5)

a. b.

She IS in the house/running. She be/*IS in the house/running.

I’m in the house/running. I be/*I’m in the house/running.

Secondly, copula be and auxiliary be can occur on the surface if stressed as in (5a); however, they are not obligatory in all environments in which they might be expected to occur. ‘Ø’ indicates the position where copula be or auxiliary be would occur were it present (6a). On the other hand, aspectual be is obligatory in habitual constructions (6b). (6)

a. b.

Sue Ø in the house/running. ‘Sue is in the house/running’ Sue *(be) in the house/running. ‘Sue is usually in the house/running’

A third way in which aspectual be differs from copula be and auxiliary be is that it does not license V’-ellipsis, host negation, or occur in COMP in questions, so it requires do support in these contexts. Do support is illustrated with an example from negation in (7): (7)

She don’t be/*ben’t in the house/running.4 ‘She isn’t usually in the house/running’

On the other hand, aspectual be shares properties with copula be and auxiliary be. Like the copula, it precedes nominal, adjectival, and prepositional predicates, and like auxiliary be, it precedes verbs in the -ing form. One final observation is that aspectual be is like modals in that it occurs in its bare form, but it is clearly distinguished from modals in its distribution. It precedes verbs in the -ing form and not in the bare infinitive or to-infinitive form. Also, in addition, modals are always finite, but aspectual be is certainly nonfinite when it is supported by do. The syntactic properties of aspectual be can be accounted for in an analysis in which the marker is in an intermediate category, Asp(ect)P(hrase), between VP and TP (Green 1993, 1998).5 I follow Green (1993), in which it is argued that do is generated

. It should be noted that, in some instances, aspectual be occurs as the form bes, as in They bes there for her. ‘They are usually there for her’. This form is not limited to 3rd singular contexts, and the -’s here may be argued to be a redundant habitual marker, not a tense and agreement marker. . No distinction in number marking is made here. Do is used as the default form of the verb, so it occurs with singular and plural subjects. . These properties may also be captured in an analysis in which aspectual be is in vP, and whatever semantic or aspectual features are associated with the marker are also in that projection.

Chapter 14. NPs in aspectual Be constructions in African American English 

in T in aspectual be constructions. While copula be and auxiliary be move (from VP) to TP, aspectual be moves no farther than AspP. The marker may not be able to move to TP because it: 1) has no agreement paradigm, 2) has some meaning, or 3) is generated with do, which occupies T. These explanations will not be discussed any further. Do is always [-past] in aspectual be constructions and shares the habitual feature of be, so it is anaphoric.6 This characterization of do is in line with Pollock’s (1989) analysis of do in do support, in which the theta-grid of the main verb is copied onto do. Aspectual be does not assign a theta-role; however, it does have aspectual features that can be associated with or ‘copied’ onto do. In the following section, I will propose that a habitual operator () is associated with aspectual be constructions. This operator ends up in Tense, where do is generated, so [-past] and the habitual feature are both expressed in the same node.

. Recurring eventualities Aspectual be occurs with different types of predicates to indicate that some eventuality occurs habitually, as in the following sentences. (8)

a. b. c. d.

Tamiya be singing in the morning. ‘Tamiya usually sings/is usually singing in the morning’ Tamiya be nervous. ‘Tamiya is nervous from time to time’ Tamiya be on the deck/there. ‘Tamiya is on the deck/there from time to time’ Tamiya be the scorekeeper. ‘Tamiya is the scorekeeper from time to time’

In the sentences in (8a–d), aspectual be indicates that the singing, being nervous, being on the deck/there, and being the scorekeeper recur. In Green (2000), aspectual be constructions are analyzed as tripartite constructions in which a habitual operator () binds variables over eventualities (e). The representation of (8a) is given in (9): (9)

e [(in the morning, e)] [sing (Tamiya, e)]

The representation says that habitually during mornings, Tamiya sings at that time. As shown in the sentences in (8b–d), the occasions during which an eventuality occurs or holds do not have to be stated explicitly. In such cases, the particular occasions during which the eventuality occurs is determined by pragmatics. The sentences in (8) are similar to simple tense generics, such as Tamiya sings, in which a generic operator (Gen) has been argued to bind times.

. Déchaine (1993) uses the term ‘anaphoric’ in noting that “do can be thought of as anaphoric with respect of whatever activity is specified by the main verb in non-declaratives, and by the antecedent predicated in elliptical constructions” (p. 466).

 Lisa Green

Given the analysis, aspectual be is similar to adverbs of quantification such as usually and always and binds a variable. The eventuality variable is argued to be along the lines of the spatiotemporal argument in Kratzer (1995) that is based on the extra event argument in Davidson (1967) and associated with stage-level predicates. Aspectual be introduces such an argument into the representation, so both stage- and individual-level predicates take this argument. Stage-level predicates are not interpreted any differently in these constructions; however, individual-level predicates that take this eventuality argument are interpreted as having stage-level interpretations. For instance, the individual-level predicate know in (10) takes on a stage-level interpretation. This is illustrated with the following sentence, which means that on different occasions, Tamiya shows that she knows how to build computers: (10)

Tamiya be knowing how to build computers. Literally: Tamiya shows that she knows how to build computers by performing some action such as putting the right parts in the right places, telling people how to build them, etc.

To say that in this context ‘know’ has a stage-level reading does not mean that Tamiya’s knowledge of building computers is temporary such that it only comes on occasion. On the contrary, once Tamiya learns how to build computers, she knows how to build them and does not forget intermittently, under normal circumstances, although she may have to consult computer manuals from time to time or call the computer help line. Then it is not that Tamiya knows from time to time how to build computers, but that she does something on different occasions to demonstrate this knowledge. In this way, the predicate know is interpreted as being stage-level; it has a habitual reading. The sentence in (10) shows one way in which aspectual be V-ing differs from the progressive. While the progressive and states are not compatible, aspectual be V-ing and states are compatible. This explains why in AAE (as in other varieties of English) state verbs in the progressive, auxiliary be (which may occur as Ø) + state V-ing, are ungrammatical (e.g. *Tamiya knowing how to build computers), but aspectual be + state V-ing is grammatical (e.g. (10)). Aspectual be constructions are similar to simple tense generics in that like generics they have a habitual reading. Generic and habitual sentences have been argued to be characteristically similar (e.g. Carlson 1989; Gerstner-Link and Krifka 1993). Simple tense generics can have both universal and existential generic readings, as categorized in Lawler (1973). For instance, Bruce repairs Thunderbirds has an existential generic reading, in which Bruce has the expertise to repair Thunderbirds although he may not have had the opportunity to work on that type of car. Under the universal reading, Bruce typically repairs Thunderbirds. The simple tense sentence has both readings, but the aspectual be construction (Bruce be repairing Thunderbirds) only has the universal reading. As noted in Schubert and Pelletier (1989), some simple tense sentences are ambiguous between a generic/habitual and capacity reading. The sentences in (11a, b) can both be used to indicate that the radio generally picks up fifty stations.

Chapter 14. NPs in aspectual Be constructions in African American English 

(11)

a. b.

This radio pick up fifty stations.7 ‘This radio pick up fifty stations’ This radio be picking up fifty stations. ‘This radio typically picks up fifty stations’

The sentence in (11a) is ambiguous between a habitual and capacity reading in that it can mean that the radio usually picks up fifty stations as well as the radio has the capacity to pick up fifty stations but actually never does because the owner has the radio set for her six favorite stations and only listens to them. However, the aspectual be sentence (11b) cannot have the capacity reading; it only has the habitual reading in which the radio is actually tuned in to the fifty stations from time to time. The property of aspectual be of forcing a habitual reading on the predicate is seen clearly in the comparison of simple tense generics and aspectual be constructions, in which case the latter does not have the capacity reading of the former. The claim is that the extra eventuality argument is, in part, responsible for this reading. An alternative account of the unavailability of the capacity reading with aspectual be constructions is a type of modal analysis. As noted in the previous section, aspectual be shares with modals the property of being uninflected. If aspectual be is modal-like in that way, then the question is whether the marker is an epistemic modal, so a sentence such as (11b) would have an epistemic reading that is distinguished from the ability/ capacity reading in (11a). This is an interesting approach to aspectual be constructions that should be considered further. Although the distinction between the kinds of modality can account for the difference between ability and necessity readings, the extent to which we get an account of habituality is not clear. Furthermore, the classification of aspectual be as a modal is not certain, as noted in Section 2.1. As has been illustrated with know, individual-level verbal predicates get a more temporary interpretation in construction with aspectual be. Along similar lines, sentences consisting of non-verbal states in construction with aspectual be also have a type of habitual reading. Consider the aspectual be + non-verbal predicate sequence be brick in the following sentence: (12)

Those buildings be brick. Literally: You will find that when those buildings appear, they are brick.

The sentence means that on occasions when these buildings occur, they will be made of brick material. The argument here is that the eventuality argument associated with aspectual be is responsible for the stage-level reading of predicates in aspectual be constructions.

. In AAE, the distinction between singular and plural marking on verbs is not always made, so verbs used with third person singular subjects may not be inflected for agreement. For instance, in (11a) the verb pick (in pick up) is not marked with 3rd singular -s.

 Lisa Green

Sentences about the location of permanently stable objects also get this type of habitual reading when aspectual be is used. This is illustrated with the following sentence about the usual location of baptismal pools in churches: (13)

Baptismal pools be behind the choir stands. ‘Baptismal pools are usually located behind the choir stands’

Just as a brick building will be brick unless man or nature changes it, a baptismal pool will be permanently located behind the choir stand unless some type of reconstruction takes place. As such, the sentence in (13) does not mean that the locations of the pools change from time to time. The reading is that it is generally the case that when one finds a baptismal pool, it will be located behind the choir stand. The sentences in (12) and (13) are examples of instances in which aspectual be forces a type of habitual reading on objects with permanently stable properties, so the reading here, as in those in which be is in construction with predicates indicating temporary properties or predicates referring to ‘movable’ objects, is one of a plurality of eventualities. NP types play a major role in the way sentences such as those in (12) and (13) are interpreted. It should be obvious that such a sentence becomes anomalous with a specific singular NP, as in (13’): 13’. #That baptismal pool be behind the choir stands.

One way to make sense of the sentence in (13’) is to look at that baptismal pool as being movable or portable such that it can be placed behind the choir stands on different occasions. Otherwise the sentence does not make sense when it refers to a recess that is built in a church and filled with water because it is difficult to get a reading that indicates plural eventualities.8 Such anomaly does not arise with plural subjects. In the case of the bare plural subject NP, as in (13), there is no recurrence of movement of one baptismal pool from one choir stand area to another. Rather the sentence in (13) makes a general statement about where baptismal pools are located, and the plurality of eventualities refers to the different baptismal pools that are located in the specified areas.

. The description here is simplified. There are situations in which it is possible to get a habitual reading when a specific singular ‘permanently located’ NP is in construction with aspectual be and a predicate indicating a location. For example, if someone says that she never sees a baptismal pool in some specified church when she goes there, but the listener knows for a fact that there is such a pool in that church, the listener can respond by saying: That baptismal pool be right behind the choir stand. You just can’t see it when it’s not being used. The meaning is that the baptismal pool is always located there (even on occasions when the visitor goes to the church and does not see it).

Chapter 14. NPs in aspectual Be constructions in African American English 

.

Characterizing properties

In this section, two additional properties of aspectual be constructions are discussed. These constructions express essential properties and regular occurrences, not particular events. In this way, they are like generics, and, as will be shown in Section 4, NPs in aspectual be constructions have a similar interpretation to NPs in generics. Before discussing the interpretation of NPs, however, I will show that aspectual be constructions are characterizing sentences. Krifka, Pelletier, Carlson, ter Meulen, Link, and Chierchia (1995) give a number of tests that can be used to distinguish characterizing and particular sentences. Characterizing sentences indicate regular occurrences of events, and particular sentences refer to a particular instance of an event. In this section, it is shown that aspectual be constructions pass tests for characterizing sentences. Aspectual be constructions have a habitual or characterizing reading (14a), not a particular, non-characterizing reading as in (14b): (14)

a. b.

Bruce be drinking coffee. ‘Bruce usually drinks/is usually drinking coffee on different occasions’ Bruce Ø drinking coffee. ‘Bruce is drinking coffee’

Aspectual be sentences indicate regular occurrences and not particular events, so (14a) refers to the usual occurrence of Bruce’s coffee drinking, while (14b) refers to a particular instance of Bruce’s coffee drinking. Characterizing sentences are also argued to express essential, not accidental, properties, the explanation given for the judgments of the sentences in (15) (from Krifka et al. (1995: 13)). 15)

a. A madrigal is polyphonic. b. ??A madrigal is popular.

Being polyphonic is an essential property of a madrigal; however, being popular is an accidental property. As such, the characterizing sentence in (15a) has a good reading, but that in (15b) does not because being popular is not an essential property of a madrigal. The sentences in (16) are somewhat parallel to those in (15): (16)

a. b.

A mouse be small. ??A mouse be noisy.

On the characterizing reading, the sentence in (16a) is good, but the one in (16b) is not. Here again, the good reading is associated with the sentence in which the predicate expresses an essential property, that of a mouse being small. Being noisy, on the other hand, is an accidental property of a mouse. Aspectual be constructions are similar to simple tense generics in that they are characterizing sentences and express essential properties, but they exhibit some differences when the predicates in those constructions are in -ing forms. As noted by Krifka et al. (1995: 12), one of the properties of characterizing sentences is that they do not retain

 Lisa Green

their generic reading when placed in the progressive. Compare their examples in (17) and (18): (17)

a. b. c. d.

The Italian drinks wine with his dinner. An Italian drinks wine with his dinner. Italians drink wine with their dinner. Luigi drinks wine with his dinner.

(18)

a. b. c. d.

The Italian is drinking wine with his dinner. An Italian is drinking wine with his dinner. Italians are drinking wine with their dinner. Luigi is drinking wine with his dinner.

The generic sentences in (17) lose the characterizing reading as progressives in (18). There is no such caveat in AAE for aspectual be sentences with verbal predicates; they obligatorily occur in the -ing form, as in (19), although they have a different reading than the progressive: (19)

a. b. c. d.

The Italian be drinking wine with his dinner. (cf. 17a, 18a) An Italian be drinking wine with his dinner. (cf. 17b, 18b) Italians be drinking wine with their dinner. (cf. 17c, 18c) Luigi be drinking wine with his dinner. (cf. 17d, 18d)

The sentences in (19) have a habitual reading, differing in interpretation from sentences in (17) only in that they do not have the existential reading such that Luigi will drink wine with his dinner but has not had occasion to do so very often for one reason or another. The difference between the sentences in (18) and (19) is that the latter have a characterizing interpretation, the property the sentences in (17) and (19) have in common. Note that both definite and indefinite NPs can occur as subjects in the generic sentences (e.g. 17a, b). Likewise, both types of NPs can occur as subjects in the aspectual be constructions (e.g. 19a, b). In what follows, the interpretation of NPs in aspectual be sentences will be considered.

.

Bare plural and singular indefinite NPs

Research such as Carlson (1977) and Diesing (1992) presents accounts of different readings of bare plurals in generic sentences. This section compares the interpretation of bare plurals in aspectual be constructions to the interpretation of bare plurals in simple tense generics. In addition, some observations about singular indefinite NPs in aspectual be constructions are made.

. Bare plurals in aspectual Be constructions Carlson (1977) describes two different readings of the bare plural in English in his argument against an account of bare plurals as NPs with a null determiner. The two readings

Chapter 14. NPs in aspectual Be constructions in African American English 

of the bare plural are generic and indefinite plural. These readings have been discussed more recently in Diesing (1992). I will return to Diesing’s analysis in Section 4.3. The sentences from Carlson (1977: 413, 414) are examples of generic sentences that have a universal reading, so a universal quantifier with the force of most would be appropriate in representing the truth conditions: (20)

a. b.

Horses are mammals/creatures/material objects. Horses are smart/larger than mules/good pets.

Carlson uses the sentences in (21) to show that in some instances of generics, the universal quantifier is inappropriate for representing the truth conditions of the bare plural construction. (21)

a. b.

Horses are widespread. Horses are extinct.

The other type of bare plural, which is commonly referred to as the indefinite plural, has an existential reading, in which an existential quantifier captures the truth conditions with the force of some. The examples in (22) are also taken from Carlson (1977: 36). (22)

a. b.

Doctors tried to save the dying boy. Mice will come out of that wall if you pound on it.

It has already been shown that aspectual be constructions can be classified as characterizing sentences, so it is not surprising that bare plurals in construction with aspectual be have the two types of readings noted in Carlson (1977). Aspectual be sentences that are similar to the generics Carlson gives are in (23): (23)

a. b.

Horses be smart/larger than mules/good pets. ‘Horses are usually smart/larger than mules/good pets’ Mice be coming out of that wall. ‘Mice usually come out of that wall’

In Carlson’s unified analysis, both types of bare plurals are treated as denoting names of kinds; a particular property is associated with them. The difference is that the indefinite plural reading arises as a realization of a kind at a time and place. What is interesting here is that while the description captures the distinction between the two types of generics exemplified in (20), (21) and (22), it is not clear that it accurately distinguishes the aspectual be generics in (23a) on the one hand and in (23b) on the other. The property that is used in Carlson’s analysis to distinguish the two is related to that property that is used to unite all aspectual be constructions. That is, Carlson uses the notion of time to capture the difference in interpretation of the generics with the two types of bare plural NPs, but all aspectual be sentences have a reading in which the eventuality is understood as occurring from time to time. It is the case that aspectual be sentences have two types of readings, but it is not clear that Carlson’s description

 Lisa Green

adequately distinguishes them.9 I return to this issue in the final section of the paper in conjunction with Diesing’s discussion of bare plurals.

. Singular Indefinite NPs In this section, I consider singular indefinite NPs in aspectual be constructions and the extent to which they have readings similar to those of bare plural subjects. As shown in the sentences in (19a–c), and repeated here (24a–c), both definite and indefinite NPs can occur in aspectual be constructions: (24)

a. b. c.

The Italian be drinking wine with his dinner. An Italian be drinking wine with his dinner. Italians be drinking wine with their dinner.

All of the sentences in (24) are characterizing sentences, but the readings in (24a, b) are not exactly parallel. While the sentence in (24a) is a characterizing sentence in which drinking wine with dinner is attributed to a particular Italian or to Italians, the sentence in (24b) does not have the same reading. The most felicitous reading of the sentence in (24b) is not one that attributes drinking wine with dinner as a well established property of Italians; instead it reports the regular occurrence of an individual Italian on certain occasions. That is, the indefinite NP gets a taxonomic reading in that it refers to a certain Italian or sub-group of Italians in the sense of the use of taxonomic in Krifka et al. (1995). The sentence is characterizing in that it expresses an event that occurs from time to time, but it does not necessarily express a general property of Italians. (24a) and (24c), which have a definite NP and bare plural, respectively, both have a characterizing reading in which they express a usual eventuality of a group. In this way, these sentences may be used in some of the same contexts. The basic difference between the two types of sentences is in the interpretation of the NPs, not in the habitual reading. The results are somewhat different in aspectual be constructions with non-verbal predicates. Consider the sentences in (25): (25)

a. b.

The madrigal be polyphonic. A madrigal be polyphonic.

. Jacqueline Guéron (personal communication) raises the question about whether sentences such as those in (23) suggest that aspectual be sentences are divided into the categorical and thetic sentence types discussed in Kuroda (1972). According to Kuroda, generic statements are classified as categorical judgments or predications, and specific events are classified as thetic judgments. While I suggest that all aspectual be constructions are a type of generic, it may be possible to separate them into Kuroda’s categorical and thetic judgment groups.

Chapter 14. NPs in aspectual Be constructions in African American English 

As expected, the sentence in (25a) is characterizing and expresses an essential property of madrigals, and (25b) has the same reading. Aspectual be sentences with non-verbal predicates, which express essential properties and in which the subject is an indefinite NP, have a more felicitous reading in which properties of a group are expressed as compared to aspectual be constructions with verbal predicates and in which the subject is an indefinite NP. The sentences in (26a, b) with indefinite subject NPs are fine with the characterizing reading of a group. (26)

a. a’. b. b’.

A Porsche Boxster be small. ??A Porsche Boxster be red. A mouse be small. ??A mouse be black.

A clear distinction between essential (26a, b) and accidental (26a’, b’) properties is made in the paradigm of aspectual be constructions with singular indefinite NPs and non-verbal and verbal predicates (cf. (24b)). So far, we have seen that a range of types of NPs can occur in aspectual be constructions. Sentences such as (24a, c) show that the definite and bare plural NPs are treated similarly in that both can occur in aspectual be + non-verbal characterizing sentences with a reading that refers to a general group. On the other hand, the indefinite subject NP must be treated differently; it cannot occur felicitously as the subject in aspectual be + V-ing constructions with a characterizing reading (24b). As shown in the sentences in (26), indefinite singular NPs can occur in aspectual be characterizing sentences in which an essential property of the NP is expressed by a non-verbal predicate. Given that all aspectual be sentences express regularities, they are characterizing sentences although limitations are placed on the type of readings that are possible with indefinite NP subjects with aspectual be +V-ing. There is no reason to separate characterizing sentences such as those in (24) from the sentences in (26). In a sense, they all express regularities or essential properties, for that matter.

.

Multiple readings of bare plurals

As noted in Section 3, aspectual be sentences are like progressive sentences in that they take the -ing form of the verb, but they report generalizations over occasions of eventualities, not particular events. These sentences are like simple tense characterizing sentences (in English in general) in that they are stative and express regularities. In this way, they are like sentences that contain stative verbs such as know, cost, weigh, love, fear and have. The difference between mainstream English and AAE is that in AAE stative verbs can occur in context with aspectual be and receive an event reading.10 For instance, beasp knowing is the event counterpart to know, beasp weighing is . Certainly there are strategies in English that can be used to give stative verbs less stative readings, as in Those shoes are costing more and more these days.

 Lisa Green

the event counterpart to weigh, beasp costing is the event counterpart to cost, etc., where beasp refers to aspectual be. One question that is raised here is whether restrictions are placed on the type of subject NPs that can occur in aspectual be constructions with stative verbs. Consider bare plurals and lexical stative verbs (27a, b, c) and their aspectual be counterparts (27a’, b’, c’). (27)

a. a’. b. b’. c. c’.

Lions have bushy tails. Lions be having bushy tails. Electricians know how to wire houses. Electricians be knowing how to wire houses. Professional jockeys weigh from about 118–126 pounds. Professional jockeys be weighing from about 118–126 pounds.

Now note singular indefinite NPs and characterizing verbs (28a, b, c) and their aspectual be counterparts (28a’, b’, c’): (28)

a. a’. b. b’. c. c’.

A lion have a bushy tail.11 A lion be having a bushy tail. A electrician know how to wire houses.12 A electrician be knowing how to wire houses. A professional jockey weigh from about 118 to 126 pounds. A professional jockey be weighing from about 118 to 126 pounds.

Given the analysis of aspectual be, generalization must be over eventualities of having a bushy tail (28a’), knowing how to wire houses (28b’), and weighing about 118 to 126 pounds (28c’). However, given what we know about having a bushy tail, knowing how to wire houses, and even the weight of a jockey, these predicates cannot refer to literal different episodes of eventualities. That is, if things are normal, there will not be habitual occurrences of a lion having a bushy tail or of an electrician knowing how to wire houses. While it is possible for a jockey’s weight to change within a small range, it is not likely that there are recurring episodes of weighing from 118 to 126 pounds. Once a lion has a bushy tail, he keeps it and does not take on another kind of tail from time to time. Likewise once an electrician knows how to wire houses, he generally will not have episodes of forgetting and remembering how to do his job. Actually there are not recurring episodes of these eventualities, but there are instances in which an electrician shows that he knows how to wire houses, for example. The sentences are

. In general, have is used as the default form of the verb, so it occurs with singular and plural subjects. . The distinction between the indefinite determiners a, an is neutralized such that a is the default indefinite determiner used in all contexts.

Chapter 14. NPs in aspectual Be constructions in African American English 

actually about properties of individuals, but aspectual be forces them into a type of habitual reading. The sentences in (28a’, b’, c’) only differ from those in (28a, b, c), respectively, in that the former have the type of habitual reading explained in Section 2.2. I will refer to this type of habitual reading as pseudo-habitual, a descriptive term used to distinguish the reading of sentences such as those in (28a’, b’, c’) from the habitual reading in which there is actual recurrence of an eventuality. These sentences (28a’, b’, c’) are identical in meaning to the aspectual be sentences with bare plural subjects in (27a’, b’, c’), respectively. Both groups are types of characterizing sentences, and in neither case is it possible to get a literal habitual reading in which the entity indicated by the subject NP actually undergoes changes such that on some occasions it has the characterizing property expressed by the predicate but then does not have the property on others. In these sentences, the singular indefinite NPs have the reading of bare plurals; they refer to individuals of a class. It seems that the (a’, b’, c’) sentences in (27) and (28) can be characterized, in part, according to the description in Carlson (1977) of stage- and individual-level predicates. In Carlson’s terms, “stages are more closely related to events than objects.” An individual is composed of sets of stages, which are tied “together to make them stages of the same thing” (p. 64). The distinction is that stage-level predicates, which are more temporary, and individual-level predicates, which are more permanent, are predicated of different types of things. In the analysis, the temporary nature of states is due to predication of stages of individuals, while properties are predicated of individuals. The temporary nature of states is due to the period of duration of a state, and the permanence of properties is due to their being predicated of individuals. Once the properties of aspectual be are added to Carlson’s description, it is possible to get an accurate account of the stage- and individual-level predicates in aspectual be constructions. Although there is permanence of properties predicated of individuals, aspectual be forces them to take on the notion of appearing more temporary. Aspectual be sentences can be divided roughly into two groups: the ones that report more temporary properties (e.g. (8)) and those that name more permanent properties with a pseudo-habitual reading (e.g. (13), (28a’, b’, c’)). In Carlson’s terms, those predicates that name more permanent properties select the generic reading of bare plurals, and those that name temporary properties select the indefinite plural reading of bare plurals. Along the same lines of Carlson’s approach, Diesing separates the readings of bare plurals into two classes. She, too, notes that a stage-level predicate can trigger both generic and existential readings. In Carlson’s analysis, the existential quantifier is due to the predication of a stage that is both temporally and spatially bounded. In Diesing’s analysis, the difference in readings with stage- and individual-level predicates is due to the position of the subject NP in the logical representation. In addition to accounting for generic and existential readings, Diesing also discusses the existential generic reading. According to Diesing, the bare plural in the sentence Firemen are available has three readings due to the stage-level predicate. In the first case, the predicate gets an episodic reading and the bare plural an existential reading. Another reading is the generic,

 Lisa Green

in which being available is an attribute of firemen. In the final reading, the bare plural gets an existential generic interpretation and can be paraphrased as ‘Generally, there are fireman available.’ In this case, the generic and existential operators interact. One question here is whether bare plurals in aspectual be constructions are also ambiguous in the way that bare plurals in English generics are. Consider the sentences below in which the predicate in (29a) is individual-level and that in (29b) is stage-level: (29)

a. b.

Basketball players be tall. ‘Basketball players are usually tall’ Surgeons be on call. ‘Surgeons are usually on call’

The same type of split in simple tense generics that occurs between bare plurals with individual-level predicates on the one hand and bare plurals with stage-level predicates on the other also occurs with bare plurals in aspectual be constructions. That is, bare plurals with stage-level predicates in aspectual be constructions are ambiguous. There is no existential reading of (29a) just as there is no existential reading of bare plurals in simple tense generics with individual-level predicates. The sentence in (29b) is two ways ambiguous. Informal representations are given in (30): (30)

Surgeons be on call. i. HABe [P (e)] [on call (surgeons, e)]] ii. HABe [P (e)] [(∃x) [surgeons (x) & on call (x, e)]]

The ‘P’ variable in the representations in (i) and (ii) is used in place of a specified time in the restrictive clause during which the eventuality indicated by the predicate (on call) occurs or holds. The first reading (i) given in (30) is the habitual reading meaning that habitually, surgeons are on call. In the second reading (ii), the existential quantifier is under the scope of the habitual operator. It may be referred to as an existential habitual and paraphrased as ‘Habitually, there are surgeons on call.’ In this case, there are always occasions during which surgeons are on call. The sentence does not have the simple generic reading that has been identified for English generics with stage-level predicates, in which being on call is an attribute of surgeons. What has been shown in this section is that bare plurals in aspectual be constructions are also ambiguous. What is revealed here is that the distinction between stage- and individual-level predicates almost seems to be neutralized in aspectual be constructions because both types of predicates get a habitual reading although in the case of individual-level predicates there is no literal reading in which the eventuality expressed by the predicate actually recurs. Nevertheless, the distinction between the two types of predicates is clearly maintained in the constructions with bare plurals and stage- and individual-level predicates. That is, in aspectual be constructions, bare plurals with individual-level predicates are unambiguous. On the other hand, bare plurals in construction with stage-level predicates in aspectual be constructions are ambiguous; they get a habitual and existential habitual reading.

Chapter 14. NPs in aspectual Be constructions in African American English 

Conclusion One of the claims in previous research on NPs in AAE is that, for the most part, they behave like NPs in other varieties of English, except with regard to the two properties they share with NPs in English Creoles. It is the case that AAE differs from Creoles in that bare singular NPs are not permitted, so they cannot be used to express singular indefinite or bare plural readings. NPs in AAE, like those in Gullah and Jamaican, are not overtly marked with genitive -s. Also, AAE and the two Creoles share similar patterns of expressing the associative plural. Studying NPs in aspectual be constructions provides an opportunity to learn more about the interpretation of NPs in AAE and to gather more data for comparison to Creoles. In this paper, it is shown that singular indefinite NPs get different types of readings in aspectual be constructions, depending on the type of predicate. Also, bare plurals with individual-level predicates in aspectual be constructions are unambiguous, but bare plurals with stage-level predicates in aspectual be constructions are ambiguous between a habitual and existential habitual reading.

References Bach, E. 1981. On time, tense, and aspect: An essay in English metaphysics. In Radical Pragmatics, P. Cole (ed.), 63–81. New York NY: Academic Press. Carlson, G. N. 1977. A unified analysis of the English bare plural. Linguistics and Philosophy 1: 413–457. Carlson, G. N. 1989. The semantic composition of English generic sentences. In Properties, Types, and Meaning [Semantic Issues 2], G. Chierchia, B. Partee & R. Turner (eds.), 167–191. Dordrecht: Kluwer. Davidson, D. 1967. The logical form of action sentences. In The Logic of Decision and Action, N. Rescher (ed.), 81–95. Pittsburgh PA: University of Pittsburgh. Déchaine, R.-M. A. 1993. Predicates across categories: Towards a category-neutral syntax. PhD Dissertation, University of Massachusetts, Amherst. Diesing, M. 1992. Indefinites. Cambridge MA: MIT. Doherty, C. 1996. Clausal structure and the modern Irish copula. Natural Language and Linguistic Theory 14: 1–46. Gerstner-Link, C. & Krifka, M. 1993. Genericity. In Encoding of Semantic Aspects of Meaning [An International Handbook of Contemporary Research 1], J. Jacobs, A. von Stechow, W. Sternefeld & T. Vennemann (eds), 966–978. New York NY: De Gruyter. Green, L. 1993. Topics in African American English: The verb system analysis. PhD Dissertation, University of Massachusetts, Amherst. Green. L. 1998. Aspect and predicate phrases in African-American Vernacular English. In African-American English: Structure, history and use, S. S. Mufwene, J. R. Rickford, G. Bailey & J. Baugh (eds), 37–68. New York NY: Routledge. Green, L. 2000. Aspectual be-type constructions in African American English. Natural Language Semantics 8: 1–25. Kratzer, A. 1995. Stage-level and individual-level predicates. In The Generic Book, G. N. Carlson & F. J. Pelletier (eds), 125–175. Chicago IL: The University of Chicago.

 Lisa Green Krifka, M., Pelletier, F. J., Carlson, G., ter Meuler, A., Link, G. & Chierchia, G. 1995. Genericity: An introduction. In The Generic Book, G. N. Carlson & F. J. Pelletier (eds), 1–124. Chicago IL: The University of Chicago. Kuroda, S.-Y. 1972. The categorical and the thetic judgment. Foundations of Language 9: 153– 185. Lawler, J. 1973. Studies in Generics [University of Michigan Papers in Linguistics 1:1]. Ann Arbor MI: University of Michigan. Mufwene, S. S. 1998. The structure of the noun phrase in African-American Vernacular English. In African-American English: Structure, history and use, S. S. Mufwene, J. R. Rickford, G. Bailey & J. Baugh (eds), 69–81. New York NY: Routledge. Mufwene, S. S. 2000. Some sociohistorical inferences about the development of African American English. In The English History of African American English, S. Poplack (ed.), 233–263. Malden MA: Blackwell. Pollock, J.-Y. 1989. Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry 20: 365–424. Poplack, S. 2000. The English History of African American English. Malden MA: Blackwell. Poplack, S. & Tagliamonte, S. 2001. African American English in the Diaspora. Malden MA: Blackwell. Ramchand, G. C. 1996. Two subject positions in Scottish Gaelic: The syntax-semantics Interface. Natural Language Semantics 4: 165–191. Rickford, J. R. 1998. The creole origins of African American vernacular English: Evidence from copula absence. In African-American English: Structure, history and use, S. S. Mufwene, J. R. Rickford, G. Bailey & J. Baugh (eds), 154–200. London: Routledge. Schmitt, C. & Munn, A. 1999. Against the nominal mapping parameter: Bare nouns in Brazilian Portuguese. In Proceedings of the Northeast Linguistic Society, N. Hall, M. Hirotani & P. Tamanji (eds). Amherst MA: Graduate Linguistics Students’ Association, University of Massachusetts. Schubert, L. K. & Pelletier, J. 1989. Generally speaking, or using discourse representation theory to interpret generics. In Properties, Types, and Meaning [Semantic Issues 2], G. Chierchia, B. Partee & R. Turner (eds), 193–268. Dordrecht: Kluwer. Winford, D. 1997. On the origins of African American English—A creolist perspective, Part 1: The sociohistorical background. Diachronica 14: 305–344. Winford, D. 1998. On the origins of African American English—A creolist perspective, Part 2: Linguistic features. Diachronica 15: 99–154.

 

Bare nouns in African American English (AAE) Arthur K. Spears Bare count (and mass) nouns, defined as nouns with no determiner or number suffix, occur in African American English, a partially restructured, noncreole language. It is noteworthy that bare nouns in AAE may take definite and generic interpretations. Bare nouns with definite interpretations were previously believed to occur in creoles outside the Americas only. However, these bare nouns occur in Palenquero, a creole spoken in Columbia. Bare nouns with generic interpretations are common across creoles but, interestingly, do not occur in non-African American dialects of English in the U.S. Bare count nouns with a definite interpretation are limited to subject position in finite sentences. This suggests that the bare noun may in fact contain a null determiner bound by a possibly null sentence topic, as might be the case in some creole languages. However, this analysis does not hold for a bare singular in the scope of negation. In such cases, we could assume that the null determiner is bound by a /existential operator.

.

Introduction

In this chapter1 I will present some remarks on bare nouns in AAE.2 Although I have made a distinction between African American Vernacular English (AAVE), and African American standard3 English (AASE), (e.g. Spears 1988), I will disregard the distinction

.

This chapter is an elaboration of Spears (2004).

. Also see Green (this volume) for a syntactic and semantic study of bare NPs in aspectual be constructions in African American English. . The word standard is usually not capitalized, even when it is used to name a variety of language, because the variety is not the property of a group with an established social identity. Although standards are commonly considered as attaching to regional groups among others, they are not usually associated with specific ethnic (or racial) groups, e.g. African Americans. The term AASE implies that at least in the case of African Americans, there is a standard variety that satisfies the conditions for standardness and also has distinctive grammatical traits associated

 Arthur K. Spears

for the purposes of this chapter, unless some grammatical explanation of the examples is required. Otherwise, I will simply use the term AAE.4 My principal goal is to bring out the differences between, on the one hand, AAE and, on the other, non-African American dialects of English, i.e. other dialects of English (ODE), the latter having had no significant influence from West African or creole languages. Examples of dialects in the latter category would be those of the British Isles and non-African American dialects in the U.S.

. African American English As noted in the introduction, AAE has two subvarieties: AASE and AAVE, a distinction I set aside in this discussion (see above). Holm (2003) has classified AAE as partially restructured. This to say that it has undergone restructuring but significantly less than languages traditionally referred to as creoles (e.g. Jamaican and Guyanese). The term restructuring refers to “all structural modifications that a lexifier language undergoes in the selection and evolution of new linguistic elements, influenced by other, competing languages, in a contact situation”(Neumann-Holzschuh and Schneider, 2000: 6). Simply for convenience,5 I will use the term semi-creole in reference to AAE below and to refer to any language such as AAE, which may be viewed as partially restructured, not restructured as much as most creoles, both in its genesis and its development historically. We have to be careful to distinguish when the partial restructuring may have occurred (see Mufwene 2001). Clearly, it is very difficult to talk about partially restructured languages of the African diaspora in the Americas as opposed to more or “fully” restructured ones. Let it be clear that the term semi-creole here simply refers to the less restructured end of a continuum or cline, which includes some varieties of languages traditionally labeled creoles (e.g. some varieties of Trinidadian and Barbadian Creole English).

with this ethnic group. The conditions for standardness are essentially negative: not having certain grammatical features considered nonstandard, e.g. multiple negatives, ain’t use, and double modals, as in We might could do that. Distinctively African American grammatical features occurring in AASE are mostly undiscussed in grammatical literature, and so fall beneath the radar of those who might want to label them as nonstandard due to their being distinctively African American. . Some speakers who typically use AASE occasionally use a bare noun in a definite sense. Some AASE speakers never do. When any speaker uses bare nouns with a definite sense, s/he can be considered to be speaking AAVE, as, for example, when a speaker uses ain’t or multiple negatives. This observation would, of course, raise a number of questions in the mind of a quantitative sociolinguist, but space does not permit dealing with them here. . It’s less cumbersome than partially restructured language and no less precise than it or other terms used in basically the same sense (e.g. creoloid).

Chapter 15. Bare nouns in African American English (AAE) 

AAE is of interest for creole studies because it has grammatical (not simply lexical) features that can be classified as Africanisms (features whose source can be traced to West African languages). An example is the semi-auxiliary6 come (Spears 1980, 1982, 1990), which is found in the West African language Bambara (and probably others), and in at least the following creoles: Haitian, Jamaican, and Guyanese (Spears, to appear).7 It occurs too in Surinamese Dutch (de Kleine 1999). AAE is of interest also because it has grammatical features that can be classified as creolisms, e.g. associative them,8 the semi-auxiliary come (a creolism and Africanism, as a number of creolisms are), and serial-like verb constructions9 (Spears, To appear).

. Use of the term semi-auxiliary has become established in identifying this item in AAE grammar; it has become its name, so to speak, since the first journal article appeared that treated it (Spears, 1982). Semi-auxiliary was originally used because this come appeared too different from both main verbs and auxiliaries to be called either one. Clearly, how it is labeled depends on the theoretical framework one uses. This said, I will point out that it differs from main verbs in that a. it cannot occur in compound verbal constructions, e.g. be + V+ing, will be V+ ing, have V+ed, AUX V+ing, etc. b. it does not inflect: *coming, comes, etc. (except in some special cases that need not detain us here) c. it takes a present participle or bare verb complement, the former as the progressive auxiliary be (which complement cannot be elided as can that of progressive be) It also has properties that distinguish it from both AUX in non-AAE varieties and a third constituent type that exists in AAE but not in non-AAE that is “verb-like.” There has been almost no syntactic investigation of this third category, discussion of it would take us away from the focus of this chapter. .

Some examples:

AAE: He  coming in here actin a damn fool ‘He had the nerve to come in here acting [like] a damn fool.’ Guyanese: kom shee out diiz jii sii ii peepos, giv os chrii shiits ov peepo, an lef do  hand out these  papers gave us three sheets of paper, and left the ruum. room. ‘She had the nerve to hand out… .’ (Rickford 1987: 616) Jamaican: Di gyal kom kom kaal mi fuul. The girl  come calling me a fool ‘The girl had the nerve to come calling me a fool.’ (Pauline Christie, personal communication)

 Arthur K. Spears

. AAE & U.S. Creoles AAE forms a continuum with creoles indigenous to the U.S. I state this based on personal experience. It may be noted that the well-known AAE scholar William Stewart concurred with this view of the contemporary U.S. [cited personal communication in Holm 1983: 314]. Kautzsch and Schneider (2000) make a corresponding but more refined observation for earlier AAVE in the state of South Carolina. Their take on the matter is that “the varieties of earlier AAVE in South Carolina as recorded in the ex-slave narratives can be accounted for by the concept of ‘differential creolization’, the assumption that the amount of creolization to be observed in the sub-areas of the state correlates with the density of the black population proportion in the period after Emancipation” (Kautzsch and Schneider 2000: 271). They state further that “the grammar of coastal speakers is predominantly creole and that of the inland speakers is predominantly non-creole, but that of the speakers from the intermediate region, still part of the coastal plains, has turned out to be less strongly creole than would have been anticipated” (Kautzsch & Schneider 2000: 271). In other words, there is a comparatively greater thinning out of creole features going from the coastal to the intermediate area than in going from the intermediate to the inland area.10 Their work indicates that the continuum in South Carolina (it still exists) is not new. It is plausible that the continuum in other areas is not recent either, that it is not, e.g. the result of language change in the 20th century or the post-World War II period. There are at least two grammatically distinguishable mainland English-lexifier creole varieties: Gullah and Afro-Seminole (Holm 1989: 479). However, the grammatical distance between them, so to speak, is certainly less than what one finds among most English-lexifier Caribbean creoles.

Haitian: Pa vin di m anyen sou fè manje.   tell 1sg nothing about make cook ‘Darn it! Don’t tell me how to cook. (Elisée St. Preux, personal communication) .

Example: John an them [or ‘em] left this morning. ‘John and his friends/family/gang, etc. left this morning.’ (There’s no one good term to use in translating associative them.)

.

Those having three or more bare verbs in sequence: Hurry come go walk to the store with me. Rush run come go take this to Grandma. (ODE’ s would break up the sequence with one or two ands, if indeed so many bare verbs were used in a simple sentence at all.)

. They duly address the difficulty of making firm claims about what is and is not a creole grammatical feature. The reader is referred to their work for further discussion.

Chapter 15. Bare nouns in African American English (AAE) 

In the creole speech of Florida today, there may also be a creole variety distinguishable from Afro-Seminole and Gullah − indeed a Bahamian creole, given the back and forth of Afro-Seminoles, Gullahs, other African Americans, and Bahamians between the Bahamas and the U.S. mainland. There are pockets of Bahamian descendants in Florida, for example, whose speech falls along a creole/AAVE continuum. This statement is based on my own observations. Research is required to determine whether the creole end of this continuum can usefully be distinguished from Gullah, which has historically influenced Bahamian (Holm 1983) and, perhaps, to a lesser extent been influenced by Bahamian. Of course, Bahamian (i.e. in the Bahamas), Gullah, and Afro-Seminole are quite close; and, Bahamian and Gullah, at least grammatically, share more with one another than they do with other English-lexifier Caribbean creoles. Moreover, Afro-Seminole is an offshoot of Gullah (Holm 1989: 495). Bahamian is in most respects an offshoot of Gullah too. “In many respects the ‘mother country’ of the Bahamas was actually the North American mainland…” (Holm 1989: 489).

.

Source of data

For this chapter, I am using my own grammatical intuitions as a native speaker of AAE, born in the U.S. My family and ancestors, my clan11 − to use the anthropological term, are without exception, for the past one hundred and fifty years or so at least, members of the largest language group in the African American community, that which speaks AAE. In other words, none of my family has Caribbean (and possibly, creole language) connections. No one in my family except me lives in a part of the country where there is a significant Caribbean population, so Caribbean creole (or West African creole, for that matter) language influence is quite out of the question. Overwhelmingly, African American (AA) communities are multi-class, as were and are the communities in which my family lives/d. Note also the high level of residential segregation by race that marks U.S. residential areas. The great majority of AAs live in racially segregated AA communities. The communities in which I and the rest of my family of my age were raised were rigidly segregated – and multi-class. Thus, we spoke, unavoidably, a variety of AAE. In my family’s case, it was a standard AAE, but we were in constant, everyday contact with vernacular (i.e. nonstandard) varieties of AAE (AAVE). We are all capable of speaking a somewhat acrolectal12 AAVE, but do not normally do so owing to social factors. . It is indeed a clan, since I am speaking of a once unilocal extended family that has recently dispersed. The clan, as it happens, is a collection of relatives descended from a nineteenth century ancestor (actually, there are paternal and maternal clans). The presence of Native Americans and European-Americans (whites) in African American family histories is common and is of no consequence for this discussion. .

That which is close to the standard.

 Arthur K. Spears

Thus, the data I present in this chapter contain sentences that I normally would not speak, but am perfectly capable of speaking and that I do use on occasions, e.g. when socializing briefly with AAVE speakers or when in a social situation where it is to my advantage to speak AAVE.13 Furthermore, since I live in Harlem (New York City), I hear daily even the most basilectal varieties of AAVE. Harlem is a kaleidoscope of AA class groups, not to mention other groupings based on identities such as regional ones. New York City’s population density and the use of public transportation allow one to hear many kinds of speech daily.

. AAE bare nouns and creoles AAE is often compared to Atlantic creole languages (e.g. Jamaican, Barbadian, and Trinidadian), primarily because of hypotheses on its origins that relate it to creole languages (see Holm 2000; Holm 2003; Wolfram & Thomas 2002, for discussions) and because AAE has creolisms (see the discussion below). The most interesting feature of AAE bare nouns, that is, those without determiners or a plural suffix, is that A. B.

count nouns may take definite and generic interpretations, which is not the case in ODE. mass nouns may also take definite interpretations, which is not the case in ODE.

The following examples in (1) illustrate A, and those in (2) illustrate B: (1)

a. b.

(2)

(The) dog got fleas. (AAE) (Count, Definite)14 ‘The dog’s got fleas.’ Dog ain’t got no sense. (AAE) (Count, Generic) ‘Dogs don’t have any sense/ A dog doesn’t have any sense.’ (The) butter you bought is good! (AAE) (Mass, Definite)

Of interest in relation to creole languages are two facts involving the differences mentioned above between AAE and ODE: C.

AAE bare nouns (count and mass) may take a definite interpretation. Among Western Hemisphere creoles, this is the case in Palenquero15 (Schwegler and Green, to appear; Schwegler, this volume) as well as Jamaican (personal communication, Silvia Kowenberg) and probably other creoles there. Also, this trait is found in several West African creoles.

. E.g. in conflict-prone ones, where AAVE implies the ability to defend oneself, verbally and physically. .

Optional elements are in parentheses. Nouns of interest are in boldface.

.

Palenquero is a Spanish-lexified creole spoken in Colombia.

Chapter 15. Bare nouns in African American English (AAE) 

D.

AAE bare nouns may take a generic interpretation (as in 1b). This is typically the case in creole languages. (E.g. Haitian Creole: Lè revèt fè dans li, li pa envite poul ‘When cockroaches/a cockroach give/gives a party, they/he don’t/doesn’t invite chickens’.)

Concerning C, note that as recently as a few years ago, it was claimed that definite bare nouns do not exist in Atlantic creoles, although they do in non-Atlantic ones, for example, Tok Pisin and Nubi (Holm 2000: 214). It should also be pointed out that such bare nouns are found in African Portuguese-lexifier creoles in West Africa (see Baptista, Kihm, Alexandre & Hagemeijer, and Kihm in this volume). Schwegler (2002, this volume) and Schwegler and Green (to appear) show that bare nouns in Palenquero may take a definite interpretation depending on pragmatic context. Observe the following example with its possible translations: (3)

Pelo asé ndrumi mucho Dog  sleep much ‘The dog sleeps a lot.’ (Singular, Definite) ‘There’s a dog that sleeps a lot.’ (Singular, Indefinite) ‘Dogs sleep a lot.’ (Generic) ‘The dogs sleep a lot.’ (Plural, Definite) ‘Some dogs sleeps a lot.’ (Plural, Indefinite) (Example from Schwegler 2002; author’s translations)

As Schwegler and Green (to appear) elaborate, …[Palenquero] can and generally does signal number and “±definite” in nouns via un (singular indefinite), Ø (singular definite), un ma (plural indefinite), and ma ‘plural definite’. But the use of these articles is conditioned not by grammatical but discourse pragmatic principles, i.e. the need to disambiguate number and ±definiteness. (Schwegler and Green to appear: 40)

Thus, the Palenquero bare noun may contextually be interpreted as singular (definite or indefinite) or plural (definite or indefinite) as well as generic. This follows from Schwegler’s and Green’s discussion and Schwegler’s observations (this volume). Additionally, as noted above, Jamaican and no doubt other Caribbean creoles have bare nouns with definite (singular and plural) and generic interpretations. Consequently, certain behaviors of AAE bare nouns are like those of creoles (even if not all). I use the term creolism for grammatical features ●

●

that are unlike anything found in a creole’s lexifier language (or ODE, in the case of AAE; in some cases, the features may be found in ODE, but their presence can most likely be attributed to contact with AAE) but that are like features found in some, if not all, Atlantic16 creoles.

. The reference group of creoles in this chapter is Atlantic creoles. It is much less revealing to speak of all the languages that have been traditionally labeled creoles in using the locution “grammatical features found in some creoles, if not all.”

 Arthur K. Spears

Lists of creolisms that accord with this characterization can be found in Schneider 1990; Kautzsch and Schneider 2000; and Spears, in preparation.17 My use of the term creolism does not imply that AAE has its origins in one or more creoles. The creolisms in AAE, I might add, are not enough to reasonably classify AAE as a creole language. This is to say that languages that are commonly classified as creoles have significantly more creolisms (i.e. traits distinguishing them from their European lexifiers but shared with at least some other creoles). Additionally, my use of the term creolism does not imply that all creole languages form a typological class (McWhorter 1998, 2001a, and 2001b; Mufwene 1997, 2001). It simply recognizes that there are features found in a number of Atlantic creoles (not all creoles that have been traditionally labeled as such) that are not found in their European language lexifiers.

. .

The semantics of bare nouns Determiners in AAE and ODE

It is well known that ODE, as well as AAE, have bare singular mass nouns (e.g. butter and salt) and plural nouns with no determiner, interpreted as indefinite (e.g. tables). Note the use of determiners in ODE, as shown in Table 1: Table 1. Use of determiners to express definiteness in ODE − Count Nouns Singular nouns Definite The N

Plural nouns Indefinite a(n) N

Definite the N+PL

Indefinite N+PL

Examples would be, from left to right, the car, a car (an apple), the cars, and cars. Of course, mass nouns normally occur only without a determiner and normally do not

. However, not every single item in those lists is a creolism in this sense: e.g. don, the perfect tense/aspect marker, which occurs in a number of English varieties with no connection to creoles (e.g. Ozark American English). Also, some traits are limited to intentionally (as opposed to extentionally) definable creole subgroups, e.g. the postposed articles of French-lexifier creoles: e.g. Haitian gason-an, ‘boy-’ = ‘the boy’. Note finally that any definition of creolism will ultimately run into problems because this language set is still not strictly defined, grammatically or otherwise, pace McWhorter (1998). Even his later remarks on creole typology fail to mark off the set of languages that have commonly been labeled creoles. Any creolist can come up with a structural definition of creole languages that will define a subset of the languages commonly called creoles. At this point in our understanding, we can only hope to narrow down as much as possible what we are talking about and rely on the forbearance of the reader until our knowledge of contact languages expands significantly.

Chapter 15. Bare nouns in African American English (AAE) 

occur as plural, the notable exception being when one refers to more than one kind, e.g. all the cheeses sold at the supermarket. AAE is different in that it has count nouns that are singular and bare.

.

Plurality

Plurality must be marked in the following AAE sentences, as in ODE, but the definite article is optional under a definite interpretation: (4)

A: You got too many suits! B: Yeah, but (the) suits I got are all good. (Plural, Definite) A: Yeah, but you don’t even have a chance to wear them all. B: Well, (the) suits still good, so I’m gonna (= going to) keep them. (Plural, Definite)

In line with the statement just made, the following example is ungrammatical: (5) *Yeah, but suit I got are all good. (Plural, Definite)

In other words, for a noun to be definite and plural, it must be marked with the plural. Since AAE bare count nouns, as in some creoles, can take a definite interpretation, even though this is impossible in other English varieties, it is reasonable tentatively to classify this feature as a creolism. In any case, the occurrence of definite bare count nouns in Atlantic creoles and in an Atlantic semi-creole forces us to reformulate our views of the grammars of Atlantic creoles and semi-creoles.

. Optionality, definiteness, and number The examples thus far underlie the following claims concerning AAE (see A–D above): E.

Optionality of the determiner – the definite article is optional with bare nouns taking a definite interpretation; see the examples in (1) and (2). F. Number – bare nouns may take a singular definite interpretation (see (1a) and below), while determinerless plural nouns may be indefinite, as in ODE, or definite, unlike ODE; see examples in (4). Bare nouns cannot be interpreted as definite and plural; see example (5). G. Definiteness – in AAE, unlike ODE, the definite article is not required to express definiteness; see E and F. (This is another way of stating A and B.)

.

Specificity

Going further, I can point out the behavior of AAE bare nouns with respect to specificity. By specificity is meant denotation by the speaker of an entity or a group that is uniquely identifiable and distinguished from all other entities inside and outside of the

 Arthur K. Spears

class of entities to which it belongs.18 Example (1a), repeated here as (1a)’, which is also SPECIFIC, shows that a bare noun may be specific. (1)

a.’

(The) dog got fleas. (Count, Definite, Specific)

Plural nouns, with or without the definite article, may be specific, as shown in the examples in (4) above. Suits is specific in those examples. Also note that in (4), even if suits were not modified, it could still be interpreted as specific and definite. Thus, H.

.

Specificity – in AAE, unlike ODE, bare nouns may be specific (see the examples referenced in the immediately preceding discussion).

Tense

None of these interpretations are affected by tense. This is shown in the examples in (6) and (7). In (6), the bare nouns are definite and specific. In (7), it is generic. (6)

a. b. c. d. e.

(The) Norwegian gonna [‘is going to’] marry Betty. (Definite, Specific) (The) dog barkin like he crazy. (Definite, Specific) Cop (‘policeman’) was telling them what to do. (Definite, SPecific) Cop told them what to do. (Definite, Specific) (The) man always act crazy. (Definite, Specific)

(7)

(A) little dog fight more than a big one. (Generic)

Thus, we have I: I.

.

Tense – The interpretation of bare nouns is not affected by tense.

Genericity

The examples in (8) illustrate J: (8)

a. b. c. d.

White man don’t want you to have nothing. (Generic)19 Pretty woman ain’t nothing but trouble. (Generic) Black woman got it hard. (The black woman has a hard time in life.) (Generic) Cow eat grass. (Generic) (This sentence shows 3sg verbal -s absence. See also example in (1b).

. This definition, which does not make critical use of the notion of presupposed existence, handles cases of specificity in embedded clauses of “world-changing” verbs (such as dream), e.g. Last night, I dreamt that I married a unicorn. In this example, the unicorn is not presupposed by the speaker ever to have existed. . In practice, bare noun generics are often used to talk about racial and gender issues, but they are certainly not limited to such discussions.

Chapter 15. Bare nouns in African American English (AAE) 

J.

.

Genericity - Bare nouns may be specific, but they may also be generic, i.e. quantifying over all denotated entities (or more accurately in the case of mass nouns, all instances of a substance), i.e. the extension of the noun. (The same is true with definite article generics such as the tiger, e.g. The tiger is a carnivore, which also takes nongeneric interpretations. An example of the latter would be where the tiger designates a type, as in The tiger is found in India. Obviously, not all tigers are found in India.) Determinerless plural nouns, in AAE as in ODE, may also be interpreted as generic, e.g. Tigers are carnivores.20

Summary – interpretation of bare nouns

Table 2 shows the possible interpretations of count nouns in AAE, indicating among other things that all five forms of the count noun may contextually express specificity and genericity. Table 2. Forms and possible interpretations of count nouns in AAE Singular count nouns

Definite Indef. Specific Generic

Plural count nouns

The N

A(N) N

N (bare noun)

+ − + +

− + + +

+ − + +

The N+PL + − + +

N+PL + + + +

A summary of the same information for mass nouns is presented in Table 3. Table 3. Forms and possible interpretations of mass nouns in AAE Singular mass nouns

Definite Indef. Specific Generic

The N

A(N) N

+ − + −

− + − −

Plural mass nouns

N (bare noun) +† − +† +

The N+PL

N+PL

+†† − +†† −

+ +††† + −

†e.g.

Last night we had cheese and gumbo, and, man, I tell you, cheese was good! Gumbo didn’t taste like nothing. (Specific and Definite) ††Meaning ‘types’ only, e.g. The cheeses I got are more like brie and camembert. †††Meaning ‘types’ only, e.g. At the gourmet store, cheeses are a better buy than wines. . In standard ODE, singular nouns with the definite article (e.g. the tiger; the tiger is a carnivore) having a generic interpretation are rarer than determinerless plural-marked generics (e.g. tigers; tigers are carnivores). The former are somewhat bookish, for one thing, and have a more restricted use. Singular nouns with the indefinite article can also be used as a type of generic, e.g. A dog is a canine just like a wolf.

 Arthur K. Spears

The comparison of AAE with ODE can be made explicit by pointing out that, as implied by the preceding discussion, ODE do not have bare nouns. Thus, Tables 3 and 4, showing the facts for ODE, would have no bare nouns column.

.

Bare nouns and sentential position

Bare nouns whose formal and semantic behavior does not conform to that of ODE, discussed above, occur freely in tensed clause subject position. They do not occur in other argument positions (Examples 9a–c). Thus, the bare noun phenomena under discussion occur freely in subject position in conjoined clauses (Examples 9d, e), and tensed complements (9g, h) but not in untensed complements, e.g. want, in (9f), which is highly questionable. The nouns in question in the following examples are to be interpreted as definite. (9)

a. b. c. d. e. f. g. h.

*Yeah, I saw man you know. *Give it to woman upstairs. *I saw him with car I want. Dog is dead, but man got out alive. Man left, and woman did too. ??I want dog next door to stop messin around in my yard. I know man ain’t got no sense. I told him man ain’t got no sense.

Curiously, the sentences in (10) are acceptable, and they both have objects of negated verbs. (10)

a. b.

I didn’t see woman nowhere. I ain’t found bucket nowhere. (Both ain’t and didn’t can be used for negating the Simple Past in AAVE.)

The behavior may be relatable to the fact that crosslinguistically, noun complements of negated verbs are sometimes treated differently (morphosyntactically) from those of their non-negated verb counterparts. For examples, in the following French example, the object noun in the affirmative sentence has the indefinite article une, while that in the negative has the preposition de: (11)

a. b.

J’ai vu une femme dans la salle. ‘I saw a woman in the room.’ Je n’ai pas vu de femme dans la salle. ‘I didn’t see a woman in the room.’

An interesting generalization emerges from the data in (9) and (10). It seems that a bare singular noun is definite only in the subject position of a finite sentence. This could suggest that the bare noun may in fact contain a null determiner bound by a possibly null sentence topic, as might be the case in some creole languages. It is interesting that this solution does not hold for a bare singular in the scope of negation.

Chapter 15. Bare nouns in African American English (AAE) 

Then, we could assume that the null determiner is bound by a NEG/existential operator in such cases.21

.

Conclusion

The preceding observations raise a number of points. One is the possibility that there are dialects of AAE unknown to me in which bare nouns occur in a wider range of sentential contexts, conceivably some for which the proviso concerning sentential context is unnecessary. This possibility should be pursued in further research. Given the partially restructured nature of AAE, we could expect its grammatical subsystems to be different in distinct ways from both ODE and creoles, and this is true. This point about the partially restructured nature of AAE is all the more important in view of the fact that there are other varieties of European languages in the Americas (as opposed to creoles) showing the phenomenon of bare nouns taking definite interpretations and probably taking an array of interpretations similar to those treated above (see Green, 1997 on nonstandard Dominican Spanish and Melo 1997 on Brazilian nonstandard varieties). This observation strongly suggests that creolists should make the study of partially restructured languages, “semi-creoles,” or whatever one wishes to call them, a central feature of the creole studies agenda. Indeed, this group of languages, including AAE and other African-language-influenced varieties of European languages (again, as opposed to “creoles”) including Cuban and Dominican vernacular Spanish, vernacular Brazilian Portuguese, and Surinamese Dutch22 (de Kleine 1999) should be studied with reference to each other, to creoles, and other languages falling within the purview of language contact studies.

References Green, K. R. 1997. Non-standard Dominican Spanish: Evidence of partial restructuring. PhD Dissertation, The City University of New York. Holm, J. 1983. On the relationship of Gullah and Bahamian. American Speech 58(4): 303–318. Holm, J. 1989. Pidgins and Creoles, Vol. 2: Reference survey. Cambridge: CUP. Holm, J. 2000. An Introduction to Pidgins and Creoles. Cambridge: CUP. Holm, J. 2003. Languages in Contact: The partial restructuring of vernaculars. Cambridge: CUP. Kautzsch, A. & Schneider, E. W. 2000. Differential creolization: Some evidence from Earlier African American Vernacular English in South Carolina. In Degrees of Restructuring in Creole Languages, I. Neumann-Holzschuh & E. W. Schneider (eds), 247–274. Amsterdam: John Benjamins. .

I am indebted to Jacqueline Guéron for clarifying these possibilities.

. Afrikaans could certainly be included here, even though for sociopolitical reasons it is not considered a variety of Dutch and it has some non-African substrate influences.

 Arthur K. Spears de Kleine, C. M. 1999. A Morphosyntactic Analysis of Surinamese Dutch as Spoken by the Creole Population of Paramaribo, Suriname. PhD Dissertation. The City University of New York. McWhorter, J. H. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74: 788–818. McWhorter, J. H. 2001a. Defining creole as a synchronic term. In Degrees of Restructuring in Creole Languages, I. Neumann-Holzschuh and E. W. Schneider (eds), 85–123. Amsterdam: John Benjamins. McWhorter, J. H. 2001b. The world’s simplest grammars are Creole grammars. Linguistic Typology 5(3): 125–156. Melo, H. R. de. 1997. The Genesis and Development of Brazilian Vernacular Portuguese. PhD Dissertation, The City University of New York. Mufwene, S. S. 1997. Jargons, pidgins, creoles, and koines: What are they? In The Structure and Status of Pidgins and Creoles, A. K. Spears & D. Winford (eds), 35–70. Amsterdam: John Benjamins. Mufwene, S. S. 2001. The Ecology of Language Evolution. Cambridge: CUP. Neumann-Holzschuh, I. & Schneider, E. W. 2000. Introduction: Degrees of restructuring in creole languages? In Degrees of Restructuring in Creole Languages, I. Neumann-Holzschuh & E. W. Schneider (eds), 1–18. Amsterdam: John Benjamins. Rickford, J. R. 1987. Dimensions of a Creole Continuum: History, texts and linguistic analysis of Guyanese Creole. Stanford CA: Stanford University Press. Schneider, E. W. 1990. The cline of creoleness in English-oriented creoles and semi-creoles in the Caribbean. English World-Wide 11: 79–113. Schwegler, A. 2002. Reconsidering the evidence: Bare nouns in Palenquero and what they really mean. Paper presented at the Society for Pidgin and Creole Linguistics meeting, San Francisco, CA. Schwegler, A. & Green, K. To appear. Palenquero (Creole Spanish). In Comparative Creole Syntax, J. Holm & P. Patrick (eds). London: Battlebridge. Spears, A. K. 1980. The other come in Black English. Working Papers in Socio-Linguistics, No. 77. Southwest Education Development Laboratory and National Institute of Education. Spears, A. K. 1982. The Black English semi-auxiliary come. Language 58(4): 850–72. Spears, A. K. 1988. Black American English. In Anthropology for the Nineties, J. B. Cole (ed.), 96–113. New York NY: The Free Press. Spears, A. K. 1990. The grammaticalization of disapproval in Black American English. In CUNYForum: Papers in Linguistics [No. 15.1 & 2], R. Rieber (ed.), 30–44. New York NY: Linguistics Department. The City University of New York. Spears, A. K. 2004. Los sustantivos sin determinantes en el palenquero y en el ingles afroestadounidense. In Actas del III Encuentro de ACBLPE [Asociación de Criollos de Base Léxica Portuguesa y Española], M. Fernández, M. Fernádez-Ferreiro & N. Vázquez Veiga (eds), 227–235. Frankfurt: Verveurt. Spears, A. K. In preparation. Disapproval marking in African American English and creoles. Spears, A. K. To appear. African American English and creoles. In Pidgin/Creole Handbook, S. Kouwenberg & J. V. Singler (eds). Malden MA: Blackwell. Wolfram, W. & Thomas, E. R. 2002. The Development of African American English. Malden MA: Blackwell.

 

Dutch-lexified Creoles

 

Bare nouns in Berbice Dutch Creole Silvia Kouwenberg Berbice Dutch determinerless nouns occur frequently in factive contexts, as specific reference nouns – an unexpected finding, as specific reference is usually associated with definiteness, and Berbice Dutch does not lack a definite article. I argue that specific reference as established in discourse is marked by an overt definite article, but that specific reference which is established outside the discourse is marked by a covert definite article. Roberts’s (2003) distinction between strong and weak familiarity is able to distinguish between these types of specific reference. Determinerless nouns occur also as plural and mass indefinites on first mention. Such bare nouns can be overtly pluralized and are in complementary distribution with overt indefinites, pointing to the presence of a covert indefinite article. This, I argue, heads NumP, like its overt counterpart. Only non-referential determinerless nours, which appear as predicates, are structurally bare.

Introduction Berbice Dutch Creole (BD), a plantation creole of mixed Dutch and Eastern Ijo lexicon, is the former vernacular of the Berbice River and Canje River area (Guyana, South America), which was a Dutch colony from the early seventeenth to the early nineteenth century.1 This article considers the distribution and interpretation of bare nouns in BD, where “bare nouns” is taken to refer to determinerless NPs; such NPs may contain adjectival modifiers and plural markers.2 In the following, we will first briefly consider the BD determiner system and the extent to which the use of determiners in BD follows that . I carried out fieldwork with the last speakers of BD over three separate periods during 1986–1990. I am grateful to the Netherlands Foundation for Tropical Research (WOTRO), which funded my research, and to the Berbice Dutch speakers who shared their language and their life-stories with me. I wish to thank Michèle Stewart, Enoch Aboh, and the editors for insightful comments on some of the ideas presented here. . Proper names will not be considered here. Nor will we consider the use of bare animal names to denote individual characters in the context of fables (e.g. tigri mєtє hondo, eni bєt-tє mati [tiger with dog, 3p bet-PF each_other] ‘Tiger and Dog, they made a bet with each other’).

 Silvia Kouwenberg

described by Bickerton (1981) as the typical creole article system (Section 1). Although that model can be applied in broad terms, once a body of data is considered in more detail, the predictions that follow from it break down. We then turn to a more detailed consideration of the determiner-taking preferences of BD nouns (Section 2). There are some contexts where bare nouns are expected. These include cases where the reference of a noun is inherently plural (collective) or to an entity of indefinite size (mass). Another such context is that where a noun is used with generic reference, or where it is a predicate. Notions such as “nonspecific” and “nonindividuated” can be called upon to account for these bare noun uses. But there are also some unexpected contexts which facilitate the use of bare nouns (Section 3). One is where the noun has a unique referent. “Uniqueness” is not a unified phenomenon: we will see that it can come about in different ways. Another is where the bare noun signals the existence of a referent whose exact identity is irrelevant or cannot be identified.3 Can or should there be a single account for all these cases? This issue is tied up with the question whether bare nouns have an extended projection headed by a null determiner (Section 4). Evidence to be taken into account includes distributional and functional properties of bare nouns, as well as theory-internal considerations. Where bare nouns are seen to have properties which are typically associated with the presence of a determiner, we have cause to assume that the noun is in the scope of a null determiner. In that case, its “bareness” is only a surface property. We will see, however, that some cases are better accounted for by postulating a null Numeral heading an extended projection of NP, while nonargumental bare nouns can be considered truly bare. This paper is based on fieldwork material. My fieldwork did not consider the kinds of questions which constitute its subject matter. The data represent mainly naturalistic production, with the limitations inherent in such data. This means that not all relevant issues can be satisfatorily resolved.

.

The BD determiner system

The BD determiner system consists of the prenominal markers di [definite] and en [indefinite], the latter homophonous with and presumably derived from the numeral

. Before we proceed, it is useful to acknowledge that nouns do not necessarily inherently belong to any one of the types distinguished here. Some nouns can be used both countably and uncountably, for instance. Similarly, uniquely referring nouns are that by virtue of a cultural or pragmatic context. In other words, one and the same noun can potentially occur in different usages, although there may well be preferred or default interpretations. In the following, where I use terms such as “mass noun”, “count noun”, “uniquely referring noun” etc., the terms refer to usage, not to inherent properties of the nouns in question.

Chapter 16. Bare nouns in Berbice Dutch Creole 

en ‘one’. Thus: di wari ‘the house’, en wari ‘a house / one house’.4 Let us first consider what is expected of the article system of creole languages, and then turn to its BD illustration.

.

The creole prototype

The most complete statement is that of Bickerton (1981: 22–24), who describes the article system of Hawaii Creole English (HCE) as follows: (1)

The Hawaii Creole English article system: a. the definite article is used for all and only specific reference NPs that can be assumed known to the listener; b. the indefinite article is used for all and only specific-reference NPs that can be assumed unknown to the listener; c. all other NPs have no article, including generic NPs, NPs within the scope of negation (i.e. clearly nonspecific NPs), and cases where, while a specific referent may exist, the exact identity of that referent is either unknown to the speaker or irrelevant to the point at issue.

Bickerton (1981) claims that the grammars of plantation creoles–which includes both HCE and BD–reflect the properties of universal grammar.5 It is legitimate, therefore, to take his characterization of the HCE system as constituting a claim regarding the properties of the article systems of plantation creoles generally.6 Bickerton predicts the occurrence of overt articles based on a combination of factors including referential properties (specific / nonspecific), familiarity (which is usually reflective of discourse status, i.e. old / new) and syntactic contexts (presence of negation).

.

BD and the creole prototype

Because of its reliance on discourse factors, the validity of (1) can be ascertained only by considering the occurrence of nominals in discourse, as for instance the occurrences of

. Demonstratives are postnominal, and require the presence of the prenominal definite article: di wari di / dida [the house this / that] ‘this / that house’. Possessed nouns are assumed to acquire definiteness by virtue of their possessor. We will not consider any of these here. . In Bickerton’s view of creole genesis, children born in the plantation environment were exposed to a rudimentary jargon from which no grammar could be inferred; as a result, creole languages are thought to display the default settings of the innate language acquisition device. . The generalizations in (1) presumably hold of the countable usage of nouns. It appears that Bickerton did not take non-count nouns into consideration.

 Silvia Kouwenberg

the noun kui in (2). The extension7 of the noun kui is the set of cows. In the case of its first occurrence, kui1, which occurs in the context of a potential event, there is no member of that set which is picked out by it: it is nonspecific, in Bickerton’s category (c). At this point, any arbitrary cow can represent the set. As predicted by (1c), it occurs without a determiner. Kui2 on the other hand, is used in a factive context where the activity of purchasing (kopu) involves the selection of a member of the set. En kui2 has specific reference, therefore. The indefinite article signals the fact that some member is selected–as expected from a noun which is in the class of Bickerton’s specific-reference NPs that can be assumed unknown to the listener (1b). We can say that the reference of kui has now shrunk to this single member. The function of the definite article is to refer to whatever set is now the established reference of the noun. Hence, di kui3 represents reference to the cow which has been identified as the one purchased. Not only is the reference specific, it now also represents known information: Bickerton’s category of specific reference NPs that can be assumed known to the listener. Lyons (1998: 4) considers this use of the definite article to be anaphoric, since the familiarity of the referent is established by the preceding discourse. In contrast, kui4, like kui1, is used with arbitrary reference, in the context of a statement about a hypothetical state of affairs. Finally, di kui5 again refers to the now well-known specific cow which was purchased.8 (2)

andaka iši wa habu boki … other_day 1p  have money … iši bi iši ma kopu kui1 an 1p say 1p IRR buy cow and iši kopu-tε en kui2 mεtε o. 1p buy-  cow with 3s εkε pama bifi iši mu kopu di kui3 1s tell say 1p must buy  cow in kes aši titi εkε doto-tε o habu kui4 fi paši o, in case if time 1s die- 3s have cow for care 3s bikas o kan furkopu di kui5 because 3s can sell  cow ‘The other day we had some money (…) We said we would buy a cow1 and we bought a cow2 with it. I told (him) that we should buy the cow3, just in case, when I die he will have a cow4 to secure him (financially), because he can sell the cow5’

. The term extension is used in the usual way, as referring to the set of individuals in the universe of discourse which are denoted by the word or phrase in question. . BD examples are provided in full morphemic form, without the reductions which usually occur in natural speech. Abbreviations in the glosses are:  = copula or focus introducer,  = definite article,  = focus marker,  = habitual aspect,  = indefinite article,  = imperfective aspect,  = irrealis mood,  = locative,  = negator,  = perfective aspect,  = plural,  = resultative auxiliary. Pronouns are distinguished as s = singular, p = plural, poss = possessive. Unadapted loans from Guyanese Creole are in italics.

Chapter 16. Bare nouns in Berbice Dutch Creole 

Thus, we can say that the referential properties of the noun kui in (2) are differentiated by the choice of determiner: none for nonspecific reference, indefinite en for a situation where reference is to a specific but as yet unknown member of the set which is the extension of the noun, definite di for contexts where reference is to a known member of the set. A similar situation seems to obtain for plural NPs, as shown here for the noun namblu ‘horse’. It occurs as a bare noun on first mention, where the presence of horses in a world different from ours is asserted. It thus establishes the reference to a subset (horses in the world of the water spirits) of the set (horses) which is the extension of the noun namblu. Having established the subset which constitutes the domain of reference in discourse, this subset becomes a known entity, as signalled by the presence of the definite article on second mention. Note that the noun now appears with plural marking; we shall see below that plural marking is permitted with bare nouns too:9 (3)

namblu alma das jєnda danga … titi єnє koro-a, horse all  be_there … time rain fall- єnє titi dєnє ju das horo di namblu-apu das bre mєhєhєhє rain time then 2s  hear  horse-  bray meheheheh ‘Even horses are there (= in the world of the water spirits) (…) When rain is falling, in the rainy season, you can hear the horses braying meheheheh’

Bickerton’s (1c) predicts that a noun with nonspecific reference occurring in the scope of negation occurs without a determiner. This is true of kєnє and wari in (4–5), but we will see below that it is in fact the generic reference of the nouns rather than the negative context which is responsible for the absence of an overt determiner. (4)

o no-ko mja kєnє fi boro danga ka 3s not- make person for pass there  ‘He (= God) did not intend for people to pass there’ [‘there’ refers to a place of great danger]

(5)

ju fєndє wari kandi-kandi ka 2s find house side-side  ‘You won’t find houses near to each other’

. Plural marking is present in BD where a count noun with plural reference is used, except in the presence of a plural quantifier, where plural marking is rather unusual, though not impossible or unacceptable. Another favorable context for non-marking of plurality is where reference is to body parts which naturally come in pairs (eyes, arms, etc.). Other Caribbean creoles restrict plural marking to definite nouns; this is not the case in , where an otherwise bare noun can appear with the plural suffix, as we shall see in subsequent examples. There is also no animacy effect, i.e. animate nouns such as namblu ‘horse’ do not differ in respect of plural marking from inanimate nouns such as wari ‘house’.

 Silvia Kouwenberg

The examples in (2–5) conform to the creole prototype as characterized by Bickerton (1981). In the following, we shall see that not all noun occurrences follow these expectations. Before we proceed with such cases, let us briefly consider what the preceding examples may tell us about the existence of a null determiner. The contrast between en kui (2), a specific-reference singular noun, and namblu (3), a specific-reference plural noun, suggests that en and Ø are possibly in complementary distribution. This provides a first indication that there may be merit in postulating a null indefinite article which subcategorizes for plural NPs and which denotes specific reference with first-mention nominals. However, so far, we have seen four bare nouns, kui ‘cow’, namblu ‘horse’, k єn є ‘person’ and wari ‘house’. Of these, only namblu is used in a context where it has plural specific reference; the bare nouns kui, k єn є, wari are used in contexts where their reference is nonspecific, the first denoting any arbitrary member of the set which constitutes the extension of the noun, the latter two generically denoting the entire set. Were we to assume that the referential status of the nominal is a property of the determiner, it would seem that more than one null determiner needs to be distinguished. Alternatively, these data perhaps suggest that not every bare noun occurrence involves a null indefinite determiner. We shall return to these issues in Section 4. Not all nominals are as well behaved as those in (2–5). A few examples will suffice to introduce the problem here. In (6), the first mention of tau ‘snake’ takes the form of a bare noun. At this point, before considering the continuation of the text, we would have to assume that reference is made to a plural subset of the set which is the extension of the noun, similar to the first mention of namblu ‘horse’ in (3). The factive environment in which it occurs here does not allow for an arbitrary or a generic interpretation. But, as seen from its second mention, di tau, reference is to a subset containing a single individual. Thus, an indefinite determiner would have been expected in its first occurrence: en tau. The difference between prototypical cases of bare nouns, such as those in the preceding, and non-prototypical, unexpected cases such as this one will form the subject of the following sections. We will return to this particular example in 3.2. (6)

tau biti-tє o ben ši fišpen. di tau biti-tє ši fingri snake bite- 3s inside 3s.poss fish pen.  snake bite- 3s.poss finger an o doto-tє and 3s die- ‘A snake bit him inside his fish pen. The snake bit his finger and he died’

The following examples show the occurrence of indefinite NPs rather than bare nouns in the scope of negation. BD standard negation is clause-final. Constituent negation is achieved by scope over indefinite quantifiers, in contrast with most other creole languages, which employ preverbal standard negation and a prenominal constituent negator. As a result, indefinite NPs frequently occur in negative contexts in BD. The earlier examples of bare nouns in negative contexts in (4–5) involved generic interpretations;

Chapter 16. Bare nouns in Berbice Dutch Creole 

this, rather than the negative context, accounts for the bare nouns. I will not consider negative contexts separately below. For further discussion of BD negatives, Kouwenberg (1994a,b) can be consulted. (7)

gaavna

danši mєtє en kєnє

mєrє ka

ju nimi-tє bat di

jєrma

Governor dance with  person more  2s know- but  woman ‘The Governor dances with nobody else you know, but this woman.’ (8)

en gutu mjє ju ka  thing do 2s  ‘Nothing harms you.’10

A final point to be made here pertains to plural marking, which, according to Bickerton (1981), is restricted to definite nouns in HCE. This is true also of many creole languages of the Caribbean (see, for instance, Dijkhoff 1983). In (3), we noticed that plural marking appeared on the definite noun, but not on the bare noun, as expected. However, the BD plural suffix in fact appears quite frequently with bare nouns, as shown here, where both bare plural nouns have generic reference. We will return to plurality in the discussion of generic nouns below. (9)

.

muni-apu, da boko-apu kanє black-person-,  Amerindian-  ‘Black people, not Amerindians’ [i.e. It is Afro-Guyanese, not Amerindians, who would do this sort of thing]

Nonspecific reference nouns

In the following, we will consider the different nonspecific reference contexts distinguished in the introduction, i.e. collective and mass nouns, generic reference nouns, and nonargumental / predicative usage.

.

Bare mass and collective nouns

Mass nouns and collective nouns denote a totality rather than a set of individuals, and are therefore considered inherently plural. It follows that nouns used in this manner will not occur with the singular determiner en [indefinite], and that they are not eligible for plural marking. (10) illustrates referential use of a collective noun. On first mention, the noun is bare, on repeated mention, a definite article appears, as expected: (10)

iši brєkє biaka moi, an, o sa

mu-tє nau, bi o ma šel

di

biaka

1p break corn good and 3s  go- now say 3s  shell  corn ‘We were breaking corn nicely and he went, saying he was going to shell the corn’ . Elicited judgements and occasional spontaneous occurrences indicate that it is possible for indefinite NPs in negative contexts to take higher scope, resulting in a referential interpretation (as in: ‘One (specific) thing doesn’t harm you’ for (8)).

 Silvia Kouwenberg

Similarly, in (11), the first occurrence of mingi ‘water’ takes the form of a bare noun. This first use serves to restrict the reference of the noun, namely to that body of water involved in the hypothetical event described here. Its third occurrence takes the form of a definite noun di mingi, signalling that it is this restricted reference which is intended. The intervening, second occurrence of mingi has generic reference; it therefore appears as a bare noun. (11)

eni falu-tє mingi anga eni doto-tє, lombo gutu mingi … 3p fall- water  3p dead-, bad thing water somtiti da di mingi gain-a

some_time be  water vex- ‘If they fall into the water, they are dead. It is a bad thing, water (…) Perhaps it is that the water gets rough’ [i.e. making the river dangerous to travel on]

Many nouns allow usage in both count and non-count readings, but some nouns do so more readily than others. Thus, bita is collective ‘clothes’, and countable ‘garment, cloth’. Here, a speaker appears to make use of both options at different points in a story. Collective use is seen in (12), countable indefinite use (with plural marking) in (13), in identical contexts: (12)

fєndє en gu wari nau, en nangwa tau jєnda mєtє bita, kєnє find  big house now,  long rope be_there with clothes, person waši-tє di bita wash-  clothes ‘Then (she) came upon a large house; a long line with clothes was there; some one had washed the clothes’

(13)

mєtє bita-apu fєndє en nangwa tau jєnda find  long rope be_there with clothes- ‘(He) finds a long line there with clothes’

In short, where nouns are used with mass or collective interpretations, they do not accept plural marking, and normally occur as bare nouns on first mention. Where reference is repeated, the definite article appears.

.

Generic use

In its generic use, the reference of a noun is to the entire class consisting of all the entities satisfying the description inherent in the noun or nominal (for instance, Lyons 1998: 179). This is true of the highlighted nouns in (14–15): (14)

bakofu mєtє kukrit, ju kan jefi eni ka banana with maripa 2s can eat 3p  ‘Bananas and maripa palmnuts, you cannot eat them (under certain circumstances)’

(15)

eni bi opropo oko, ori oko moi ka 3p say pig too, 3s too good  ‘They say pig (meat) too, it too isn’t good (to eat)’

Chapter 16. Bare nouns in Berbice Dutch Creole 

But nouns may have generic reference, yet occur with a definite determiner. Consider the contrast in (16): the two utterances were produced by the same speaker on the same occasion, and in both cases, the extension of the noun is the entire class of women. (16)

a. jєrma doko gau-gau b. di jєrma doko gau-gau woman pull quick-quick  woman pull quick-quick Both: ‘Women paddle with short, quick strokes’

These utterances illustrate the difference between class generics and individual generics (Lyons 1998: 182ff). The bare noun jєrma unambiguously refers to the class of women: it constitutes a class generic. Where, instead, the singular noun di j єrma is used in an individual generic, its referent stands as a typical representative of its class (see, for instance, Hawkins 1978: 215); the class can be said to “inherit” from the individual member which represents it the property that is predicated (here one of characteristic behaviour). The presence or absence of a definite determiner ultimately makes no difference in what is said of the class, but it makes a difference in how that property is attributed to the class. Note, by the way, that while di j єrma can also receive a referential interpretation (‘the woman [that we are talking about] pulls with quick, short strokes’), this would be decidedly odd given the cultural context. In (17–18) we see the use of definite plurals with generic reference. According to the form (definite determiner, plural marking), reference is made to sets of individuals–an interpretation which is possible in principle, but which is not the intended one here. Extending the account offered above for j єrma vs. di j єrma, we can say that the generic class, being constituted by the set of individuals that make up the class, inherits its properties from that set of individuals. (17)

di boko-apu bin bangi di doto kєnє  Amerindian-  fear  dead person ‘Amerindians used to be scared of the dead’

(18)

єkє timi mєtє di

hogo skun-apu ababa-ka 1s able with  high shoe- anymore- ‘I cannot cope with high-heeled shoes any more’

A somewhat different case arises where the generic noun takes the form of a bare plural. I assume that this case, too, means that reference is made to a set of individuals: (19)

eni bi eni das mini nama-apu kєkє hatibeš-apu 3p say 3p  swallow animal- like bush_deer- ‘They say they swallow animals, such as bush deer’

(20)

daliki titi kєnє-apu ka langi titi kєnє-apu wa sєtє kєkє long time person-  stay like just now time person-  ‘People of long ago didn’t live like people of nowadays’

 Silvia Kouwenberg

.

Nonargumental use

I use the term nonargumental here to refer to attributive predicate nominals. Attributive predicate nominals lack an extensional interpretation. Predicate nominals are typically bare. In terms of Bickerton’s (1981) model, this follows from the fact that they are nonreferential. Both singular and plural nouns occur as bare predicate nominals: (21)

en wa nam ani, ori da gu-jє, gu-gu tog one  name Annie 3s be big-, big-big tug boat ‘One was called Annie. It was (a) big one, (a) really big tug boat’

(22)

da moi jagi-hondo BE good hunt-dog ‘(It) was a good hunting dog’

(23)

titi iši wa da toko-apu mu-a skul … time 1p  be child- go- school ‘when we were school-going children …’

This analysis extends to nominals in the complement of the copular verb drai ‘turn, become’ (24) and in other quasi-predicative positions (25–27): (24)

da hiri so єkє drai-tє potє man BE here  1s turn- old man ‘This is where I got old’

(25)

wєl titi o mja-tє eni mja-tє wari kriki-tє toko-apu stєdi-stєdi, 3pl make- house get- child- steady-steady, well time 3s make- gu plєkє nau big place now ‘They had families, had children all the time, well then it became a big place’

(26)

єkє hafi gungu kєkє kali toko 1s have_to creep like small child ‘I had to creep like a small child’ [after I injured myself]

(27)

dєn ju fєndє di plєkє findi nau, kєkє gu birbiši then 2s find  place open now, like big river ‘Then you will find the place opening up, (to become) like a wide river’ [describing the upriver journey on the Berbice River]

On the other hand, the contrast between (28) and (29) (uttered by different speakers) suggests that a predicate nominal can take the form of an indefinite. While this represents a minority pattern, it is clearly in violation of Bickerton’s claim that the indefinite article is used for all and only specific-reference NPs that can be assumed unknown to the listener. (28)

єkє da boko jєrma 1s be Amerindian woman ‘I am an Amerindian woman / I am Amerindian’

Chapter 16. Bare nouns in Berbice Dutch Creole 

(29)

o bi kwiina šiiba bin da en potє boko jєrma 3s say Queen_of Sheba  be  old Amerindian woman ‘It is said that Queen of Sheba was an old Amerindian woman’

This is shown again in (30–31), both uttered by the same speaker at the start of a story. The first utterance contains a predicate nominal in the form of an indefinite, the second a predicate nominal in the form of a bare nominal: (30)

Jˇak wa da en loi toko Jack    lazy child ‘Jack was a lazy child’

(31)

ši papa wa da riki man 3s.poss father   rich man ‘His father was a rich man’

In short, nonargument nouns are most typically bare, but the indefinite article is attested in these contexts, against expectations.

.

Specific reference bare nouns

We now turn to the category of specific reference count nouns whose bare occurrence— according to the predictions of the creole prototype–is not the normal option. Recall that this type of usage of nouns is supposed to be subject to the following pattern: (32)

specific reference unknown / new known / old

en + NPsg

Ø + NPpl

di + NPsg/pl

It follows from this schema that in the majority of contexts, these nouns ought to occur in the scope of an overt determiner (or other kind of quantifier). There are, however, some uses which turn out to be typically determinerless. These include nouns with unique reference and those with existential reference.

.

Uniquely referring nouns

There is a small number of nouns which are name-like in having a unique referent in the universe of the speaker. Their uniqueness is absolute, i.e. their reference is always unambiguous. Such uniquely referring nouns usually occur without a determiner, as in (33): (33)

sono das mja lombo fi єkє sun  make bad for 1s ‘The sun makes (it) hard for me [to work in the field]’

 Silvia Kouwenberg

It stands to reason that such nouns do not accept plural marking (*(di) sono-apu [the sun-PL]).11 Nor do they occur with the indefinite article en, as its use is incompatible with a uniqueness presupposition (*en sono [a sun]).12 Nouns of this kind do exceptionally appear with a definite article, as will be seen below, a possibility which presumably arises because of the uniqueness presupposition of that article. I know of three BD nouns whose denotative sets contain only one member: sono ‘(the) sun’, akalu ‘(the) moon’, grotala ‘(the) Governor’. A few other nouns function dually, occurring most frequently in their name-like use, but appearing also as regular nouns. One of these occurs with the definite article in its name-like use: di lanši [the language] ‘Berbice Dutch’ vs. lanši ‘language’; this may be related to the fact that the Berbice Dutch language became a topic of conversation only very recently, namely when linguists showed an interest in it. Other name-like nouns refer to features of the local geography, and are well-established in BD usage. These allow for both their name-like and other interpretations without a determiner: birbiši ‘(the) Berbice River / river’, kr єkє ‘(the) Wiruni Creek / creek’, stati ‘New Amsterdam / town’. Despite their similarity to proper nouns, these nouns clearly have descriptive content, and can thereby be distinguished from proper nouns, which are assumed to lack descriptive content altogether. Their descriptive content makes them potentially ambiguous. Compare name-like kr єkє in (34) with its countable use in (35): (34)

єkє wa

stup-a so, krєkє ben, ofru hilda-apu 1s  stop- so creek inside over Hilda- ‘I was living over there, in the Wiruni Creek, across from Hilda and her family’

(35)

dida das twa mosli di krєkє-apu wanga di mingi das strom that  put mostly  creek- where  water  flow ‘That one [i.e. a type of fish trap] is usually put (in) the creeks, where the water is running’

A second type of uniquely referring nouns are those with pragmatically unique referents. Hawkins (1991: 408) points out that “the immediate situation of utterance in which speaker and hearer find themselves can constitute a pragmatic set of entities for the purpose of uniqueness: Pass me the bucket will be unambiguous for the hearer if there is just one bucket in his field of vision, irrespective of the evidence of countless buckets elsewhere.” (See Hawkins 1978 for more elaborate discussion.) Thus, within the immediate situation in which a conversation takes place, the reference of certain entities can be as unambiguous as the reference of first and second pronouns. This is . It is true of the universe of the speaker that there is only one sun, though it is not true of the physical world. Thus, the size of sets is culturally determined. . See Hawkins (1991: 439), who argues that the negation of uniqueness is “conversationally implicated” by the indefinite article—although he considers it an existential quantifier, neutral with respect to uniqueness.

Chapter 16. Bare nouns in Berbice Dutch Creole 

true, for instance, of the reference of pata ‘path, road’ in (36). Moreover, despite the existence of more than one shop, more than one church, etc., there is no doubt about the unique reference of the corresponding nouns šap ‘shop’, kerki ‘church’ etc. within the larger situation of the community, as shown for šap ‘shop’ in (37). (36)

єkє kan nel єni gutu fan

hiri-sa tutu pata ben 1s can nail any thing from here-so until path inside ‘I could nail anything from here as far as in the path’ [the speaker is talking about his steady aim with bow and arrow]

(37)

a bi a djas mangi-tє mu-tє šap anga 3s say 3s just run- go- shop  ‘He said he just hurried to the shop’

Also, the discourse can set up a “virtual” situation, as is true of the following cases, both taken from a fable: the reference of tafl ‘table’ in (38)—despite the fact that it has not been introduced previously—is unambiguous by virtue of its location in the house where the main protagonist of the story finds himself. The reference of bili ‘axe’ in (39) is unambiguous as an object which is clearly assumed to be within grabbing distance at this point in the story. (38)

(39)

bringi єkє gutu-apu twa tafl bof bring 1s thing- put table top ‘Bring my things and put them on the table’ bi mama pi єkє bili gau, an Jˇak

mama pi-tє o di bili say mother give 1s axe quick, and Jack mother give- 3s the axe ‘(He) says “mother give me the axe quickly”, and Jack’s mother gave him the axe’

In short, where the pragmatic context—be it that of the larger community, that of the immediate situation, or that of a discourse situation—picks out a specific member of the set, we see bare nominals rather than quantified nouns. These all require English translations containing a definite article. In fact, the uniqueness property–be it absolute or pragmatic–is typically associated with definiteness in the relevant literature. The use of bare nouns here clearly conflicts with that view. It also does not conform to Bickerton’s prediction that specific-reference NPs which can be assumed known to the listener appear with a definite article. As we shall see below, this usage of bare nouns has in common with other bare noun occurrences that the identity of the referent is not at issue. In these cases, it is not at issue because it is self-evident. There are, however, some cases where the definite article appears with uniquely referring nouns. One type of case is where a uniquely referring noun is used in a non-prototypical role. In (40), the uniquely referring noun birbiši ‘Berbice River’ denotes not so much the Berbice River as the community of people living along

 Silvia Kouwenberg

its embankments. Since birbiši normally denotes a location, it is largely restricted to oblique contexts. In contrast, it occurs here as subject: (40)

di birbiši wa lombo  river  bad ‘The Berbice River was bad’ [i.e. bad things were happening in the community]

In (41), kirki ‘church’ does not refer to the church building located in the community, but non-prototypically to the church as institution: (41)

o bєtrє dati o lahan di kirki 3s better that 3s leave  church ‘It would be better if he left the church’ [i.e. if he gave up his position in the church]

The following generic statement presents another case: in contrast with (36), which is factive, describing a real event, (37) is non-factive, stating a generalization about the economic situation. Paradoxically, the definite nominal di šap ‘the shop’ seems to refer to “any” shop rather than to the specific one that would be provided by the pragmatic context. (42)

donkє ju kriki kali boki an ju drai mu di šap anga, don’t care 2s get small money and 2s turn go  shop  tutu ju drai wєrє ju drai mєtє en gutu ka until 2s turn again 2sg turn with one thing  ‘It doesn’t matter if you get some money and you go to the shop, when you come back, you come back with nothing’

The definite article in (40–42) seems to act as an instruction to the hearer to locate the referent, rather than assume its “default” interpretation. Hawkins (1978) argues that definite reference “instructs the hearer to locate the referent in some shared set of objects” (p. 167). That “shared set of objects” may be constituted from shared knowledge about the world, shared knowledge about the specific situation, as well as shared knowledge derived from the preceding discourse. In each case, shared knowledge defines a set of unique objects. The English definite determiner “the” is used for all kinds of uniqueness, whether absolute, situational, or discourse-related. BD di, on the other hand, is clearly not used for all kinds of uniqueness. Nouns with absolute or pragmatically / situationally determined unique reference contrast with nominals that are uniquely referring by virtue of their discourse status. Thus, di kui in (2) refers uniquely to the cow bought by the narrator and known through its earlier introduction in the discourse; its specific reference in the discourse is marked by the definite determiner. On the other hand, BD nouns with absolute or pragmatically / situationally determined unique reference are bare, as seen above. Definite reference used with such nouns carries the conversational implicature that their “normal” uniqueness does not obtain. Instead, the hearer is to locate the referent in some other shared set.

Chapter 16. Bare nouns in Berbice Dutch Creole 

While this accounts for most cases of unique reference nouns that I have come across in the texts, there are some that defy explanation. For instance, it is unclear to me what triggers the use of a definite article with sono in (43) as compared to (33) –both produced by the same speaker in the same conversation: (43)

єkє timi deki di

sono kanє 1s able take  sun  ‘I cannot deal with [the heat of] the sun’

As a final aside, note that inherently and pragmatically / situationally uniquely referring nouns differ also from nominals such as that in (44), where unique reference is a result of the presence of noun modifiers which act as functions restricting the larger set (kui) to a subset with a single member only. A definite article marks the unique reference of the noun and must be present for the utterance to be felicitous: (44)

.

ši mama tiri-є o nau fu furkopu di lašti kui wati eni habu 3s.poss mother send- 3s now to sell  last cow what 3p have ‘His mother then sent him to sell the last cow that they had’

Existential reference

Consider (45–47). In each case, we are dealing with first mention bare nouns. Also in each case, the context is factive and the proposition describes an activity which restricts the reference of the pertinent nouns: that of bili to the specific axe observed by the child, that of bariri to the hawk which is responsible for the disappearance of the speaker’s fowl, that of man to the man that “she had”. On repeated mention, these nouns appear with a definite article, as seen in (45) and (47). It would appear that an indefinite article should have been present on first mention, and that its absence constitutes an oversight, if it weren’t for the fact that my database is replete with examples of this kind. (45)

di the o 3s

toko wengi-a, titi o luru, bili, bili kori-a child walk-, time 3s look, axe, axe work- bi “daki om”, di bili bi “daki, šu, ju say goodday uncle, the axe say goodday, dear, 2s

bai ši selfu by 3s.poss self wengi-a?” walk-

‘The child walks; then when she looks, she sees (an/some) axe! (An/some) axe working by it self. She says “goodday uncle”. The axe says “goodday, dear child, you’re taking a walk?” (46)

bariri ku-arє єkє oboko-apu chicken hawk catch- 1s hen- ‘(A) chicken hawk / (some) chicken hawk is catching my chickens’

(47)

o wa habu man an eni lahan-tє, o lahan-tє di man an o kumu-tє wari 3s  have man and 3p leave-, 3s leave-  man and 3s come- house ‘She had a man / some man, and they separated; she left the man and came home’

 Silvia Kouwenberg

Despite the specific reference of these entities, there is also a certain unidentifiability to them. In (45), the identity of the axe and other magical implements which appear in the fable cannot really be known to the interlocutors. While (46) implies the existence of a hawk specifically responsible for the disappearance of the chickens, the identity of this hawk is not known or relevant as such. In (47), the speaker does not know the identity of the individual referred to and made it quite clear in the conversation that she is not the least bit interested in him. In each of the preceding cases, the bare noun appears to signal the existence of a referent, while at the same time indicating that the identity of this referent is unknown, of no interest, or of no consequence. The factive context ensures that the noun is interpreted as referential, while the use of the bare noun allows the hearer to ignore the exact reference of the noun. These data show that such bare noun usage is encountered in both subject and object positions; there appears to be no asymmetry in this regard. Bruyn (this volume) points to the relevance of Givon’s notion of pragmatic referentiality to cases such as these. Since pragmatic referentiality hinges on importance in discourse, this comes down to saying that the speaker uses the bare noun to signal that a particular noun lacks important reference. Bruyn postulates that “importance” can be measured by discourse continuity: does reference to this entity recur? Based on this measure, she notes that pragmatic referentiality cannot account for a number of Sranan cases. Similarly, there is no doubt about the importance in the discourse of the magical axe in (46), which continues a conversation with the girl, and recurs at several points later in the story. The “lack”of reference of bili ‘axe’ in (46) seems, rather, to result from the magical context, which makes it an entity which we, humans, cannot know. The speaker repeats this device when he introduces an equally magical hoe and cutlass. This is also the situation which obtains in the problematic example (6), repeated here. I noted earlier that the occurrence of tau ‘snake’ as a bare noun was unexpected. We are now in a position to interpret the use of the bare noun. The factive context tells us that tau is referential. But its bare use suggests that this snake is not identified. In this manner, the speaker chooses to make the reference of tau existential (“some snake”) rather than specific (“a snake”). Note that the snake clearly plays a central role in the event described here; it does not suffer from lack of importance. (6’)

tau biti-tє o ben ši fišpen. di tau biti-tє ši fingri snake bite- 3s inside 3s.poss fish pen.  snake bite- 3s.poss finger an o doto-tє and 3s die- ‘A snake / some snake bit him inside his fish pen. The snake bit his finger and he died’

Lyons (1998) argues that the distinction between the English definite and indefinite articles can be captured by the notion of “identifiability”: the indefinite article signals that the entity referred to is not associated with a context–a physical situation or

Chapter 16. Bare nouns in Berbice Dutch Creole 

the previous discourse (p. 13), hence not identifiable at that point. It follows that the introduction of the indefinite reference implies that such identification is now possible. Perhaps it is the case that the BD indefinite determiner is “stronger” than its English counterpart: not only does it signal that identification is possible, but also that it is necessary. From that point of view, its absence in the preceding examples can be understood as a indication that identification is not necessary. If a bare noun can be used in factive contexts to signal existential reference to a single individual, there is no reason why such a bare noun could not also appear in contexts where the reference of the noun is plural. Let us consider again the first occurrence of namblu in (3), repeated here: (3’)

namblu alma das jєnda dangga … titi єnє koro-a, horse all  be_there … time rain fall- єnє titi dєnє ju das horo di namblu-apu das bre mєhєhєhє rain time then 2s  hear  horse-  bray meheheheh ‘Even horses are there [= in the world of the water spirits] (…) When rain is falling, in the rainy season, you can hear the horses braying meheheheh’

Although the first proposition restricts the reference of the noun namblu to a subset, there is a clear sense in which the horses remain an unknown entity: they do not inhabit our world, the speaker has never seen them and she knows of noone who has. The members of that subset cannot be known. Again, the interpretation is existential rather than specific. This is clearly so also for the existentially interpreted nouns in the following: (48)

kuta das jєnda alabot matara bead  be_there all_about Matara ‘There are beads all over the place at Matara’

(49)

o bifi iširi falu-tє eni tibi anga 3s say iron fall- 3p head  ‘He said metal (pieces) fell on their heads’

The existential use of the bare nouns kєnє ‘person’ and gutu ‘thing’ distinguishes itself by sheer frequency of occurrence. As seen here, the interpretation of utterances containing these forms as subjects allows for an arbitrary interpretation of the subject. Typically, whenever I attempted to elicit passive constructions from BD speakers, utterances such as these were offered instead. Kєnє, in this type of use, is not restricted to human reference. Similar use of kєnє is seen in (12). (See Kouwenberg 1994b for discussion.) (50)

kєnє bringi-tє kom pama, bifi dati… person bring- come tell, say that… ‘Someone brought the news that…’ / ‘The news came that…’

(51)

gutu wa gjof-tє єkє toro thing  hit- 1s eye ‘Something had hit my eye’ / ‘My eye got hit (by something)’

 Silvia Kouwenberg

.

Existential bare nouns in collocations

The idiomatic combination deki kєnє [take person] is always used with a bare noun despite the fact that in most cases, kєnє has a specific referent, as it does here: (52)

єkє deki-tє kєnє an єkє trou-tє 2s take- person and 1s marry- ‘I chose a partner / I got engaged and I got married’

Lyons (1998: 51) points out that an expression such as “in bed” expresses a state or direction “rather than a relationship to a particular entity”; ‘bed’ here is not referential. Similarly, the BD expression deki kєnє can be considered to express an event, rather than a relationship to an individual. Collocations of this type use bare nouns in BD, as illustrated for a selection here: (53)

a. b. c. d. e.

kapu tun [cut field] ‘cut a field’ [refers to the practice of shifting cultivation] mja wari [make house] ‘build a home’ deki kєnє [take person] ‘choose a partner, get engaged’ deki kapu [take cut] ‘undergo an operation’ kriki toko [get child] ‘have a child’

Again, these cases seem to involve existentials rather than referential nouns. Thus, despite the fact that the factivity of (53) implies a specific individual denoted by the noun kєnє, the identity of this person is not at issue.

.

Discussion

.

Summary of findings

Let us recall the types of bare nouns that we have encountered in the preceding: (54)

Bare nouns occur as (a) plural indefinites on first mention (b) mass and collective nouns on first mention (c) nouns with generic reference (d) nonargumental occurrences (i.e. predicates and the like) (e) uniquely referring nouns (which very occasionally occur with an indefinite or definite article) (f) nouns with existential reference

The literature leads us to expect the bare occurrence of plural indefinites, mass and collective nouns, nouns with generic reference, and predicate nouns. In the case of plural indefinites, the bareness is contrasted with the singular indefinite, which is marked by an indefinite determiner. The inherent plurality of mass and collective nouns means that these nouns can be expected to behave like plural indefinites in this regard.

Chapter 16. Bare nouns in Berbice Dutch Creole 

Nouns with generic reference and predicate nominals are in the category “nonreferential”. Bickerton’s prototype predicts this category to take the form of bare nouns. In contrast, the literature does not lead us to expect the bare occurrence of uniquely referring nouns and of nouns with existential reference: these are bare nouns in factive contexts which suggest specific reference, and specific reference is expected to be marked by a determiner. Some of these cases can be better understood as instances of Mufwene’s nonindividuated category. This term is applied to the bare use of a noun which refers to its denotative kind as an “ensemble” or part thereof (1986: 39). It occurs where a speaker “does not care to individuate the referents” (1984: 205) or makes no reference to “whether or not the kind he is naming has constituent members” (1984: 203). This seems to apply to the bare existentials, where the identity of the referent whose existence is being asserted is unknown, irrelevant, or of no interest. Mufwene’s nonindividuated category has the further advantage that it can subsume existentials, generics, collectives and mass nouns, and predicate nominals. On the other hand, the uniquely referring bare nouns which we considered in 3.1 are clearly not in the nonindividuated class. Interestingly, in those cases too, the identity of a noun’s referent is not at issue, this time because it is given, either by the cultural context or the pragmatic context. Thus, both existentials and uniquely referring nouns involve cases where the identification of the referent need not be made, but for contrasting reasons. Thus, a connection between existentials and uniquely referring nominals is suggested, rather than between existentials and (nonreferential) generics. A partitioning of bare nouns into different classes has implications for an analysis which assumes covert determiners. We will now turn to the evidence for null-D, where we will see that the classification suggested by applying Mufwene’s notion of individuation is not upheld.

.

Do bare nouns have null determiners?

Determiners can be considered as functions, which pick out a member or members of a set. If it can be shown that bare nouns, like overtly determined nouns, pick out a member or members of a set, then it would be reasonable to assume that bare nouns are determined by a null determiner. Thus, its deictic function provides us with a functional argument for an empty determiner. For the bare noun uses reviewed above, it seems evident that there are several that involve such a deictic function: Plural indefinites and mass and collective nouns are determinerless on first mention, but it is clear from their appearance as definites on subsequent mention that they are no different from indefinite singulars on first mention. The plurality of these nouns – recall that it is assumed that mass and collective nouns are inherent plurals – is simply not compatible with an overt indefinite determiner. Moreover, a distributional argument can be put forward here: the overt

 Silvia Kouwenberg

indefinite determiner is in complementary distribution with its covert counterpart, the former occurring with singular nouns, the latter with plurals. The presence of a null indefinite determiner, which signals the identifiablity of the referents, is also supported theory-internally, by the assumption that DPs, not NPs, are arguments (see, for instance, Cheng & Sybesma 1999). Plural indefinite bare nouns occur freely in subject and object positions. It is, however, not clear that indefinite en should be considered a determiner: taking into account that indefinite en is derived from and homophonous with the numeral en ‘one’, and that it is ambiguous between indefinite and numeral interpretation in actual usage, an obvious alternative to postulating determiner status for en is to assign it numeral status instead—a possibility which has emerged in the theoretical literature more generally. Assuming that the extended structure of NP is at least as in (55), where it is dominated by NumP and DP, two functional positions are available in the left-periphery: (55)

[DP … [NumP … [NP … ] ] ]

NumP is thought to be associated with certain referential features (e.g. Aboh 2002); in other words, reference is not the exclusive prerogative of the DP. The numeral status of en means that nominals modified by it are in fact NumP, with en appearing as the Numeral head. Its empty counterpart is then also an empty Numeral head. As it turns out, a distinction needs to be made between the empty indefinite and a “true” empty determiner, which can be argued to modify uniquely referring nouns, where “uniqueness” is either interpreted absolutely, or with reference to the pragmatic or situational context, but not with reference to the discourse. The unique reference of these nouns is akin to definiteness. Again, we have seen uniquely identifiable bare nouns in argument positions, suggesting their DP status. The null D which heads these DPs signals the unique reference of these nouns, thus fulfilling a typical function of definite determiners, but without the implication – which is associated with the overt definite determiner – that the referent can be identified in the preceding discourse. We noted that existentials are similar to uniquely referring nouns in that their bareness signals to the hearer that it is not necessary or relevant to identify its referent from among some set of potential referents. It may be arguable that a conversational strategy is used here which employs the same null definite D which heads uniquely identifiable bare nouns. Since the situational context is such that the referent of these nouns is clearly not available in the immediate environment, it is clear to the hearer that this type of definite bare noun acts in opposition to a noun modified by an overt or covert indefinite – an opposition which is unavailable for uniquely referring nouns. The conversational use of the definite null D in these cases implies an identifiablity contrast with indefinites: the latter are newly identifiable, as signalled by the presence

Chapter 16. Bare nouns in Berbice Dutch Creole 

of Numeral “one” (whether overtly or covertly), the former are not identifiable, or, at least, the hearer is not to bother with an identification.13 For nouns with generic reference, an argument has been made for N-to-D movement (Longobardi 1994). This means that NP is dominated by an extended projection which contains material in a functional position, thus licensing the use of such bare nouns in all argument positions. Finally, we have seen bare nouns occurring also in nonargumental positions, namely as predicates. As pointed out, the extended projection of NP is assumed to allow it to function as argument. As Cheng & Sybesma put it, “D has the function of mediating between the description (predication) provided by the NP and whatever specific entity in the real world the description is applied to” (1999: 513). Where NP is used as predicate, on the other hand, no such mediation is required. In other words, nonargumental bare nouns are the only truly “bare” cases, involving only NP, without the extended projections of NumP and/or DP.

Conclusion An unexpected finding in the data is that there are contexts where bare nouns have specific interpretations. These most typically include contexts where a noun is uniquely referring, as well as contexts where a specific referent is irrelevant. I have argued that these involve the use of a null definite determiner. We have also seen indefinite bare nouns, where the bare occurrence is in complementary distribution with the presence of the overt indefinite en, which I have claimed is a numeral rather than a determiner; its null counterpart can similarly be considered a numeral. The only truly bare nouns in the data are those which occur as predicates, i.e. occupying a nonargumental position.

References Aboh, E. O. 2002. La morphosyntaxe de la périphérie gauche nominale. Recherches Linguistiques de Vincennes 31: 9–26. Bickerton, D. 1981. Roots of Language. Ann Arbor MI: Karoma. Cheng, L.L-S. and Sybesma, R. 1999. Bare and not-so-bare nouns and the structure of NP. Linguistic Inquiry 30(4): 509–542. . Kihm (this volume) suggests an analysis of somewhat similar cases in terms of incorporation. A crucial difference between the BD existentials and the Kriyol cases is that the latter are restricted to direct object positions. This makes an account in terms of lexical incorporation possible. The BD existentials occur also as subjects, where incorporation into V is not an option.

 Silvia Kouwenberg Dijkhoff, M. 1983. The process of pluralization in Papiamentu. In Studies in Caribbean Language, L. Carrington et al. (eds), 217–229. St. Augustine, Trinidad: SCL. Hawkins, J. A. 1978. Definiteness and Indefiniteness. A study in relevance and grammaticality prediction. London: Croom Helm. Hawkins, J. A. 1991. On (in)definite articles. Journal of Linguistics 27(2): 405–442. Kouwenberg, S. 1994a. Berbice Dutch. In Typological Studies in Negation, P. Kahrel & R. van den Berg (eds), 237–266. Amsterdam: John Benjamins. Kouwenberg, S. 1994b. A Grammar of Berbice Dutch Creole [Mouton Grammar Library 12]. Berlin: Mouton de Gruyter. Longobardi, G. 1994. Reference and proper names. Linguistic Inquiry 25: 609–666. Lyons, C. 1998. Definiteness. Cambridge: CUP. Mufwene, S. S. 1984. The count/mass distinction and the English lexicon. In Papers from the Parasession on Lexical Semantics, D. Testen, V. Mishra & J. Drogo (eds), 200–221. Chicago IL: Chicago Linguistic Society. Mufwene, S. S. 1986. Number delimitation in Gullah. American Speech 61: 33–60.

 

A synthesis and a postface

 

Properties of noun phrases in creole languages A synthetic comparative exposition Marlyse Baptista The primary objective of this chapter is to provide summary tables highlighting the properties of noun phrases including both determined and determinerless nouns in the 14 creoles under study.1 Before each table, I also draw a few interesting generalizations and provide a set of comments underlining the most striking features emerging from the empirical data observable in each table. As I did in Baptista (this volume), I offer working definitions for my use of the concepts of (in)definiteness, (non)specificity and (non)referentiality in the tables below. For all purposes, in this chapter, we define definiteness as being about identifiability (or familiarity to) by the speaker and hearer (Lyons 1999: 4). In contrast, an indefinite NP may denote a particular entity (specific and familiar to the speaker) or may be used to speak of an arbitrary member of the class described by the NP (non-specific) (Lyons 1999: 165). Hence, an indefinite NP may be specific or non-specific. The usefulness of the referentiality concept becomes clear in the light of examples such as (1) and (2) taken from Lyons (1999: 165): (1) (2)

I bought a car Pass me a book

“A book” in (2) is non-specific and describes but does not refer whereas specific “a car” in (1) is a referring expression.2 Lyons (1999) noted that the concepts of specificity and non-specificity should be used not only in relation to indefinites (cf. Comrie 1981 and Croft 1990) but also for definites. Consider the following set of examples from Lyons (1999), featuring an indefinite determiner in (3) and a definite one in (4).

.

I thank one of the external reviewers of our manuscript for this suggestion.

. Such a view of referentiality is not uncontroversial. See an elaborate discussion of this matter in Baptista (this volume).

 Marlyse Baptista

(3)

a.

(4)

b. a. b.

Peter intends to marry a banker – even though he doesn’t get on at all with her. Peter intends to marry a banker – though he hasn’t met her yet. Joan wants to present the prize to the winner – but he doesn’t want to receive it from her. Joan wants to present the prize to the winner – so she’ll have to wait around till the race finishes.

The (a) examples show extensional readings that are specific and referential. “A banker” in (3a) and “the winner” in (4a) refer to specific individuals. In contrast, the (b) examples show intensional readings that are non-specific and non-referential. It is clear from the context that “a banker” in (3b) and “the winner” in (4b) have not yet been identified either by the speaker or the hearer; hence, they do not refer to specific entities. These two sets of examples illustrate Lyons’ observation that both indefinites and definites can be specific or non-specific, a distinction that will be upheld in the tables below. I should add here that just as in Baptista (this volume), I treat specific/referential and non-specific/non-referential on a par. Working definitions of definiteness, specificity and referentiality having been provided, we may now examine the tables below. Table 1 shows that most of the creoles under study have both overt definite and indefinite markers (cf. AAE, Berbice Dutch, Réunionnais, Santome…) irrespective of their lexifiers or substratal sources. Let us note, however, that Palenquero lacks a singular definite marker, Guinea-Bissau Creole lacks a definite determiner altogether and only Haitian lacks an indefinite marker. In some creoles, both the definite and indefinite are prenominal (cf. Berbice Dutch, Cape Verdean Creole, Mindanao Chabacano). In others, the definite determiner may be postnominal and its indefinite counterpart prenominal (cf. Lesser Antillean, Mauritian). A first generalization is that there are no postnominal indefinite determiners in the creoles examined in this volume. A number of creoles have overt indefinite singular determiners but no plural counterparts (like English), as in Berbice Dutch, Jamaican, Lesser Antillean and Papiamentu. However, the reverse does not occur in the creoles under study: there are no creoles with overt indefinite plural determiners and devoid of overt indefinite singular counterparts. In the domain of plurality, some creoles have bound morphemes (cf. AAE, Berbice-Dutch, Cape Verdean Creole) whereas others display unbound morphemes (cf. Mauritian, Lesser Antillean). Réunionnais, Santome and Sranan miss an overt plural marker altogether. In this table, NA is used to indicate that a given determiner or plural marking does not exist in a specific creole.

Chapter 17. Properties of noun phrases in creole languages  Table 1. Table for overt determiners and plural marking3

indefinite marker/ quantifier

definite marker Creole

SG

PN PL

PN

AAE Berbice Dutch Cape Verdean Guine-Bissau Haitian Jamaican Lesser Ant. Mauritian Mindanao C. Palenquero Papiamentu Réunionnais Santome

the di kel5 NA (l)a di la la el NA e le/lo/la se13

yes yes yes NA no yes no no yes NA yes yes no

yes a yes4 en yes un NA un no NA yes7 wan no9 on yes en yes un yes un yes12 un yes in no15 u˜ a

Seychellois16

sa17

yes

Sranan

(n)a

yes

the di…apu kes NA (la)yo6 di…dem sé…la ban…la10 el manga ma e…nan le bane/lé inen…se/ inen14 sa ban/ ban18 den

SG

overt plural

PN PL

PN overt

PN

yes yes yes yes NA yes yes yes yes yes yes yes yes

NA NA uns uns NA NA NA ban manga un ma NA in bane u˜ a dôsu

NA NA yes yes NA NA NA yes yes NA yes yes

yes [–s] yes [–apu] yes [–s] yes [–s] yes [yo] yes [dem]8 yes [sé] yes [ban] yes [manga] yes [ma]11 yes [nan] NA NA

no no no no no no yes yes yes yes no NA NA

yes19

ê/en yes

NA

NA

yes [ban]

yes

yes

wan

NA

NA

NA

NA

yes

. Abbreviations used in Table 1: PN = prenominal, NA = non applicable, SG = singular, PL = plural. .

The definite marker di is prenominal and the plural marker apu postnominal.

. The primary function of kel/kes is that of a demonstrative in Cape Verdean Creole but it can also serve as a definite determiner referring to a specific or non-specific entity. . In most Haitian dialects the la yo combination does not appear but it does occur in at least one Northern dialectal variety. .

Di is prenominal and dem postnominal.

. Stewart does not treat dem in her paper as a plural marker but as an inclusive marker (which involves the concept of plurality). She acknowledges, however, that dem has been traditionally described as marking plurality. .

Sé is prenominal and la postnominal.

.

Ban is prenominal in Mauritian and la postnominal.

.

The plural marker ma in Palenquero may be overt but is not obligatory.

 Marlyse Baptista

Table 2 below shows that creoles fall into three distinct groups in the domain of plural marking and its correlates. In this table, we note that in a first group of creoles, overt plural marking can be correlated to specificity alone, as in Haitian, Lesser Antillean, Papiamentu and Seychellois. In a second group, the use of an overt plural marker is correlated to both specificity and animacy, as in Cape Verdean Creole, Guinea-Bissau Creole and Jamaican. In a third group, plural marking is totally unrelated to specificity or animacy, as in AAE, Berbice Dutch, Mindanao Chabacano and Palenquero. It is, however, never the case in the creoles under study that plural marking may be related to animacy alone and not to specificity whereas the reverse is true, as seen with the first group of creoles just mentioned. In this table, NA is used to identify creoles with no overt plural morpheme. Table 2. Overt plural marking associated with specificity or/and animacy Creole

specificity

animacy

AAE Berbice Dutch Cape Verdean Guinea-Bissau Haitian Jamaican Lesser Antillean Mauritian Mindanao C. Palenquero Papiamentu Réunionnais Santome Seychellois Sranan

no no yes yes yes yes yes no no no yes NA NA/yes20 yes NA21

no no yes yes no yes no no no no no NA yes no NA

.

E is prenominal and nan postnominal.

. The primary function of se in Santome is that of a demonstrative but it has also been described as a definite determiner in the literature on the language (Ferraz 1979; Lucchesi 1993; Schang 2000). . Alexandre and Hagemeijer also report inen can be used on its own as a plural definite marker and is prenominal. Its singular counterpart is null. .

When used together, inen is prenominal and se postnominal.

. I took the data from Seychellois presented in these tables not only from Déprez (this volume) but also from Bollée (1977), Corne (1977) and Michaelis (1993), as these latter descriptions provided the full set of Seychellois determiners. These were lacking from Déprez’ exposition of Seychellois, as she focused on its use of bare nouns.

Chapter 17. Properties of noun phrases in creole languages 

In Table 3 below, the various interpretations of each overt determiner and plural marker are considered. I examine here whether the definite, indefinite and plural markers can receive a specific interpretation, a non-specific interpretation, or both. Table 3. Interpretation of overt determiners and plural marking

definite marker

indefinite marker

overt plural

Creole

specific non-specific specific

non-specific specific non-specific

AAE Berbice D. Cape Verd. Guinea Biss. Haitian Jamaican Lesser Antil. Mauritian Mindanao C. Palenquero Papiamentu Réunionnais Santome Seychellois Sranan

yes22 yes yes NA yes yes yes yes yes yes yes yes yes yes yes

yes yes yes no NA yes yes yes yes yes yes yes yes yes yes

yes yes yes NA no no no no yes yes no no no no no

yes yes yes yes NA yes yes yes yes yes yes yes yes24 yes yes

yes yes yes yes yes yes yes yes yes yes yes NA yes25 yes NA

yes yes no23 no no no no yes yes yes no NA no no NA

. Seychellois sa is ambiguous between a demonstrative and a definite article according to Corne (1977). . Sa or sa ban may be used as definite plural markers. . When used together, sa ban appear adjacent to each other in a prenominal position. . In the case of Santome, plurality is expressed via the determiner inen that preferably modifies specific and animate entities. There is, however, no isolated plural morpheme. . There is no overt plural morpheme in Sranan. . Green (this volume) makes interesting observations regarding the use of both definite and indefinite NPs in aspectual be constructions. Consider the sentences in (i): (i) a. The Italian be drinking wine with his dinner. b. An Italian be drinking wine with his dinner. She notes that the sentences in (i) are characterizing but (a) and (b) are not exactly parallel. While in (ia), drinking wine is attributed to a particular Italian or to Italians, the preferred reading of (ib) reports the occurrence of an individual Italian on certain occasions. Therefore, in aspectual be predicates, the NPs modified by the or a may not yield a similar interpretation to that found in ordinary predicates.

 Marlyse Baptista

The creoles in our sample fall into distinct groups in this domain: in AAE, Berbice Dutch, Mindanao Chabacano and Palenquero, the definite, indefinite and plural markers may receive a specific or non-specific interpretation. For the majority of creoles in our study, the definite marker can only receive a specific reading (cf. Haitian, Jamaican, Lesser Antillean). However, in Berbice Dutch, Cape Verdean Creole and AAE, the definite determiner may be interpreted as specific or non-specific. For creoles endowed with overt indefinite determiners, these can be interpreted as specific or non-specific, the only exception being Guinea-Bissau Creole in which the indefinite determiner is only interpretable as specific. For creoles with overt plural markers, the majority (cf. Cape Verdean Creole, Guinea-Bissau Creole, Haitian, Jamaican, Lesser Antillean, Papiamentu, Seychellois…) only use overt plural marking with specific noun phrases. However, AAE, Palenquero and Berbice Dutch may use overt plural marking with non-specific referrents. In Table 3, “NA” indicates that a given marker does not exist in the language. “No” means that a given interpretation does not obtain for a specific marker. The column entitled ‘overt plural’ indicates whether an overt plural may get a specific or a nonspecific reading or both. In Table 4, the full range of available interpretations for bare nouns in each creole is considered. The creoles under study behave differently in this area as well. Bare nouns may receive a wide range of interpretations (specific/non-specific singular, specific/non-specific plural) in Cape Verdean Creole, Guinea-Bissau Creole, Jamaican, Lesser Antillean, Palenquero, Papiamentu, Secheyllois and Sranan. It is also worth noting that creoles such as Cape Verdean Creole, Guinea-Bissau Creole and Palenquero make us of both overt determiners and null determiners with the same semantic import (i.e. un in Cape Verdean Creole, Guinea-Bissau Creole and Palenquero can also be rendered with Ø. On the other hand, in a few creoles, some overt determiners and null determiners are in complementary distribution. For instance in Santome, the indefinite singular marker ˜ua must be overt and a bare noun can never be interpreted as indefinite singular.

. Non-specific nouns do not carry an overt plural marker in the basilectal varieties of CVC that I studied. . Although Alexandre and Hagemeijer do not make this explicit, I found in their paper instances of u˜ a referring to a specific entity, as illustrated in their example in (47): ˜ floli (47) Ua pega mosu n’ope. Soku mina fada floli… a flower grab boy in-foot / then girl say flower ‘A flower grabbed the boy’s feet. Then the girl said to the flower: …’ . As already stated, there is no isolated plural morpheme in Santome but the plural is expressed on the definite determiner.

Chapter 17. Properties of noun phrases in creole languages 

Table 4. Table for bare nouns interpretive variation specificity interpretation Creole AAE Berbice Dutch Cape Verdean Guinea-Bissau Haitian Jamaican Lesser Antil. Mauritian Mindanao C. Palenquero Papiamentu Réunionnais Santome Seychellois Sranan

non-specific interpretation

plural interpretation

SG

PL

SG

PL

overt

non-overt

yes yes yes yes no yes yes no yes yes yes yes yes yes yes

no26

yes yes yes yes yes yes yes yes yes yes yes yes no yes yes

yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes

yes yes yes yes yes yes yes yes yes yes yes NA NA28 yes NA

yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes

no yes yes no yes yes no27 no yes yes yes yes yes yes

In this table, ‘no’ means that a given interpretation does not obtain for bare nouns in a particular creole. NA indicates that a dedicated overt plural marker does not exist in a given creole. PL indicates that a completely bare noun (with no determiner and unmarked for plural marking) is interpretable as a plural with specific (in first column) or non-specific (in second column) interpretation irrespective of the animacy variable observed in some creoles. The last column (plural interpretation) shows whether or not a creole with an overt plural marker also has the option of using a bare noun interpretable as a plural.

. According to Spears, a completely bare noun cannot be interpreted as plural definite but determinerless nouns with a plural marker can be interpreted as specific, as is the case of many creoles under review. . This seems to be contradicted by some data that Dany Adone (in preparation) has shared with me, in which bare nouns may be interpreted as specific singular or plural. . Santome does not have a dedicated isolated plural marker but does use the definite determiner inen to convey plurality and specificity.

 Marlyse Baptista

Table 5. Table for bare nouns Syntactic distribution of bare nouns Creole

subject

object

AAE Berbice Dutch Cape Verdean Guinea-Bissau Haitian Jamaican Lesser Antillean Mauritian Mindanao C. Papiamentu Palenquero Réunionnais Santome Seychellois Sranan

yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes

no29 yes yes yes yes yes yes yes yes yes yes yes30 yes yes yes

Table 5 below shows that all creoles under study, with the notable exception of AAE, allow bare nouns to appear in argumental subject or object position. In Table 6, I consider whether there is a subject/object interpretive asymmetry. This table shows that at least three creoles in our sample (Cape Verdean Creole, GuineaBissau Creole and Santome) display an asymmetry in which subject bare nouns are preferably interpreted as definite whereas their object counterparts are preferably interpreted as indefinite. It is noteworthy that the reverse does not hold, meaning that we have not found it to be the case that bare nouns in subject positions are preferably interpreted as indefinite and objects as definites. We should also keep in mind that these are tendencies and not fixed rules. In conclusion, each table in this chapter and the sets of comments accompanying each table were meant to reflect the diversity of the determiner systems in the creoles under study. We showed that the paradigms of overt determiners and plural marking differ from one creole to the next. For instance, while some creoles such as Cape Verdean Creole display a full paradigm of overt determiners and express plurality overtly via an overt plural bound morpheme, creoles such as Sranan only display a paradigmatic subset of determiners and do not make use of an overt plural marker.

. According to Spears, completely bare nouns in AAE can only occur in object position of negated verb but not in a prototypical argumental position. However, determinerless bare nouns carrying the plural marker can occur in object position, just as in Standard English. .

Bare nouns in Réunionnais seem to be more frequent in object position.

Chapter 17. Properties of noun phrases in creole languages 

Table 6. Subject/Object asymmetry Creole AAE Berbice D Cape Verdean Guinea-Bissau Haitian Jamaican Lesser Antil. Mauritian Mindanao C. Palenquero Papiamentu Réunionnais Santome Seychellois Sranan

Subject/Object asymmetry

Pref. reading for subject

Pref. reading for object

yes no yes yes no no no yes34 no unsure35 yes36 no (only in distribution) yes37 no no

definite NA definite32 definite NA NA NA definite NA NA NA NA

NA31 NA indefinite mass33 NA NA NA indefinite NA NA NA NA

definite38 NA NA

indefinite39 NA NA

. It is non applicable, as there is no occurrence of bare nouns in ordinary object position in AAE according to Spears. . Bare plurals are interpretable as either definite or indefinite irrespective of their syntactic distribution but for bare singulars, the tendency is to interpret a bare noun as definite in subject position and as indefinite in object position. One should emphasize that this is a tendency for Cape Verdean Creole, not a fixed rule. . Kihm observes that the preferred interpretation of bare nouns in subject position is definite whereas bare nouns in object position are preferably interpreted as mass. .

This is based on data from Dany Adone (in preparation).

. Armin Schwegler could not make any categorical pronouncement on this matter and preferred to reserve his judgment until further research is conducted. . Yes for bare plurals. Bare plurals can only occur in subject position in Papiamentu. If they occur in object position, they must be modified. .

Again, these are just tendencies just as in Cape Verdean Creole and Guinea-Bissau Creole.

.

A definite singular interpretation obtains.

.

An indefinite plural interpretation tends to arise.

 Marlyse Baptista

In some creoles such as Guinea-Bissau Creole, the use of an overt plural marker is correlated with specificity and animacy whereas no correlation may be found in AAE or Mindanao Chabacano. In creoles like Berbice Dutch or Palenquero, the entire paradigm of overt determiners may be interpreted as specific or non-specific, whereas in others such as Haitian, the definite marker can only be interpreted as specific. Bare nouns may receive a wide variety of interpretations in some creoles such as Papiamentu where they may be interpreted as specific singular or plural as well as nonspecific singular or plural. In other creoles such as Haitian, their reading is much more restricted and may be limited to non-specific singular or plural. In all creoles except for AAE, bare nouns may appear in both subject and object argument positions but in some creoles, the interpretation of bare nouns will vary depending on their syntactic distribution. For instance, in Santome, bare nouns in subject position tend to receive a definite singular reading whereas bare nouns in object position tend to be interpreted as indefinite plurals. In other creoles such as Seychellois, no such interpretive subject/ object asymmetry obtains. This synthetic and comparative chapter has highlighted the complexity and diversity of the determiner systems in the creoles under examination. In the next chapter (a Postface), we examine closely the use of bare nouns across these creoles and attempt to describe as precisely as possible the role of Universal Grammar in shaping their properties.

References Adone, D. In preparation. DP functional structure in Morisyen. Bollée. A. 1977. Le créole français des Seychelles: Esquisse d’une grammaire, textes, vocabulaire. Tübingen: Max Niemeyer. Comrie, B. 1981. Language Universals and Linguistic Typology: Syntax and morphology. Oxford: Blackwell. Corne, C. 1977. Seychelles Creole Grammar: Elements for Indian ocean proto-creole reconstruction. Tübingen: TBL Verlag Gunter Narr. Croft, W. 1990. Typology and Universals. Cambridge: CUP. Ferraz, L. 1979. The Creole of São Tomé. Johannesburg: Witwatersrand University Press. Lucchesi, D. 1993. The article systems of Cape-Verde and Sao-Tomé Creole Portuguese: General principles and specific factors. Journal of Pidgin and Creole Languages. 8: 81–108. Lyons, C. 1999. Definiteness. Cambridge: CUP. Michaelis, S. 1993. Temps et aspect en créole Seychellois: Valeurs et interférences. Hamburg: Helmut Buske. Schang, E. 2000. L’émergence des créoles Portugais du golfe de Guinée. Doctoral Dissertation, Université Nancy 2.

 

Functional deficiency, ellipsis or innovation in creole languages? A Postface Marlyse Baptista and Jacqueline Guéron In this Postface, one of our primary objectives as editors is to weave together into a coherent whole the insights and analyses of our contributors. We aim to show how the chapters in the volume complement each other, highlighting both similarities and differences in the patterns of the nominal system of various creole languages. A second objective is to try to formulate some hypotheses regarding the use of bare nouns in creole languages and the role of Universal Grammar in shaping the linguistic properties of the creoles under study. To the degree that creole bare nouns have common properties, we try to identify the grammatical options that UG makes available to natural languages in the domain of the Noun Phrase.

. .

Some preliminary conclusions Bare nouns

The contributions in this volume clearly demonstrate that the creoles under study differ from their respective lexifiers in at least one startling way. Their use of bare nouns (or more precisely, bare Noun Phrases) is much more prevalent than in the European languages that contributed to their genesis. The studies by Zribi-Hertz & Glaude for Haitian and French, by Chaudenson for Réunionnais and French, by Green and Spears for African American English and other dialects of English, by Baptista for Cape Verdean Creole and European Portuguese, and by Kester & Schmitt for Papamientu and Dutch/Spanish (they also examine Brazilian Portuguese) make this clear. One of the questions raised in the Introduction to this volume was whether the influence of the substrate languages on the creole could explain the widespread use of bare nouns. The few papers that investigated the issue in this volume did not reach any satisfactory conclusion. Indeed, when compared to their African contributors, the study of Lesser Antillean (Gadelii) and of Sranan (Bruyn) did not demonstrate a clear influence from Loango Bantu languages or from Gbe, respectively, on these creoles.

 Marlyse Baptista and Jacqueline Guéron

Both languages exhibit similarities with the African contributors but important differences as well. Nor can the direct influence of former stages of the lexifier language be attested with any amount of certainty. In the three contributions presented from a historical perspective, the authors reach inconsistent results. Bruyn concludes that the properties of the determiner system in Sranan are very distinct from those of the European and African contact languages and must therefore be judged to be innovative. She crucially states that Sranan exhibits more alternation between nominals with and without determiner than either Gbe or English. Even Chaudenson, who traces some of the uses of bare nouns in Réunionnais creole to 17th century French dialects, in which they were used with agglutinative forms and light verbs, cautiously warns us that some uses of bare nouns “could be interpreted either as residues or as innovations [our emphasis], no one knows for sure.” When comparing Mindanao Chabacano to the Philippine languages that have most influenced it over the past few hundred years (Caviteño, Hiligaynon and Tagalog among others), Grant concludes that NPs in MC are quite distinct from those in the contributing languages. So, although a systematic comparison of all creoles to both their lexifier(s) and their African sources (or Philippine ones in Grant’s case) is desirable, we are still not in a position to assert that the source languages are in any way responsible for the prevalence of bare nouns in these languages. At this point, we should make it clear that we do, naturally, acknowledge the linguistic contributions of the source languages (lexifier and substrates) to the genesis of creole languages (see Baptista 2006, on feature convergence). However, we argue in this Postface that for the creoles under study, a nominal system that is quite distinct from their source languages consistently emerges.

.

On the use of bare nouns

Let us define a bare noun (bare NP) as one which is used without a determiner and possibly without number marking as well. Do these properties suffice to define the creoles under study here as a natural class? The fact that the determiner systems of the creoles examined in this volume differ among themselves warns us to be cautious on this score. No two creoles reported on here have an identical nominal system, even when they have the same lexifier. Some creole languages possess overt definite and indefinite determiners, yet even these do not display the same paradigmatic set. And none of them seem to display the complete paradigmatic set of determiners found in their lexifiers: Mindanao Chabacano (Grant), a Spanish-lexified creole, has un and el, but no feminine or plural determiners from Spanish have passed into productive and systematic use in MC. In Palenquero (Schwegler), one of its Spanish-lexified congeners, not even the definite determiner el was inherited. Haitian (Zribi-Hertz & Glaude, Déprez) is endowed with a definite singular and plural determiner but has no indefinite singular or plural marker.

Chapter 18. Functional deficiency, ellipsis or innovation in creole languages? 

Cape Verdean Creole (Baptista) has a definite/demonstrative singular and plural, an indefinite singular, but no genuine indefinite plural determiner (uns is really a quantifier). Other creoles have no overt definite determiners at all (i.e. Guinea-Bissau Creole (Kihm)). Nor do the creoles in this volume display the same range of interpretations of determinerless bare nouns. Déprez shows that among the French-lexified creoles, Seychelles Creole argument count nouns are interpretable as kind, generic, existential and definite. In contrast, in Haitian Creole and Mauritian Creole, argument count nouns can be interpreted as kind, generic or existential but no definite reading is available. For some creoles, (Cape Verdean Creole, and Guinea-Bissau Creole), the syntactic position of a bare noun affects its interpretation: a bare subject is usually construed as definite, while a bare object is construed as indefinite. Other creoles (Jamaican Creole (Stewart)) seem to be insensitive to this variable.1 Some creoles (Cape Verdean Creole and Papiamentu for instance) are endowed with both a definite determiner and a null type-shifting operator that turns a predicative nominal into a kind term. In these languages, bare singulars may occur in argument position, even though the language possesses plural morphology and lacks a generalized classifier system, in contradiction to the predictions made by Chierchia’s 1998 Nominal Mapping Parameter (see Baptista 2000 on Cape Verdean Creole). The same situation can also be found in Brazilian Portuguese (Kester & Schmitt, this volume; Schmitt & Munn 1999, 2003; Munn & Schmitt 2001). In some creoles, animacy, number and specificity are correlated (Cape Verdean Creole, Guinea-Bissau Creole and Santome), for others only number and specificity pair up (Haitian and Jamaican among others). For a third category of languages, number marking and specificity are independent of each other (Mindanao Chabacano (Grant), Palenquero (Schwegler), Berbice Dutch (Kouwenberg)). Even sister creoles like Cape Verdean Creole and Guinea-Bissau Creole display distinct nominal systems with respect to both overt determiners, as already noted, and bare nouns. To be more precise, in Guinea-Bissau Creole, nouns like kabra ‘goat’ must be marked with plural marking in order to be interpreted as referential, but no such requirement holds for Cape Verdean Creole. This overview of the disparate properties of the elements that make up the nominal domain of the creoles under study shows that the NP systems of creole languages differ both from that of their source languages and that of other creoles in their particular use of bare nouns and determiners. Furthermore, even creoles with the same lexifier, such as Haitian Creole and Seychelles French-lexified creoles, or Cape Verdean Creole and Santome Portuguese-lexified creoles, have quite distinct NP systems.

. According to Stewart, singular count nouns in Jamaican Creole can be bare and may receive definite or indefinite interpretations in both subject and object positions with no marked preference.

 Marlyse Baptista and Jacqueline Guéron

As discussed in Section 1.2 of our Introduction (Chapter 1), the disparities among the creole nominal systems prohibit us from viewing them as a typological class. In that same chapter, we stated that in our view, they do not have a genetic historical link but this does not mean that they do not have common properties that can be unveiled by examining UG mechanisms at work in creole formation.

.

Interpreting bare nouns

Sometimes a language possesses both determined and undetermined nouns. A number of the contributors to this volume attempt to discover the factors which allow us to predict when the determiner or bare noun must or may appear in such languages (Baptista, Bruyn, Déprez, Gadelii, Kouwenberg) and the semantic range of these bare nouns when they do appear. Variables such as animacy, uniqueness of the entity, the nature of Tense (episodic versus generic), information and discourse structure, and pragmatics, among other factors, reflect the strategies used by creole speakers to recover information that is provided by overt morphology, in particular by inflectional affixes and determiners, in other languages. As already mentioned, the creoles under study have in common that their use of bare nouns is more prevalent than that in the source languages that were examined. When we compare the creoles to their lexifiers, we immediately observe that the lexifying European language has a full grammatical paradigm with number affixes and determiner morphemes in DP, where the creole language is lacking in grammatical morphemes: it may have no plural inflection or it lacks a full set of determiners. All of our contributors observe the existence of bare nouns in creole languages whose lexifier has no -or else very restricted- bare nouns. This could lead some scholars to view creoles as simpler languages (i.e. McWhorter 2001) when comparing the creole DP against the paradigmatic template of the lexifier’s DP. However, even studies that note a ‘functional deficiency’ or ‘functional category ellipsis’ (Lumsden 1996) in creole languages also show that the creole nominal domain they examine does not bear much resemblance to the lexifier’s NP domain. It actually looks as if creole languages have an NP domain with its own type of completeness and self-sufficiency. This volume, and this Postface (see Section 2, below) should provide ample evidence that what seems to be a simplification in the creole actually bears witness to a highly complex set of principles. If detailed analyses lead us to doubt that DPs in the creoles under study have much in common with those of the source languages, as discussed in Section 1.1, then it is reasonable to assume that the properties of creoles reported in the various contributions to this volume are UG-based innovative systems, much in the spirit of language creation proposals such as those of Kegl, Senghas & Coppola (1999) and DeGraff (1999). A language which lacks both grammatical determiners and grammatical number marking exhibits a reduced use of functional material which plausibly must be compensated for by “radical functional innovation”. The language has recourse to UG

Chapter 18. Functional deficiency, ellipsis or innovation in creole languages? 

principles to formulate an alternative system to compensate for the lack of overt number inflection or overt determiners. As a historical note, the use of bare nouns is attested in the developmental stages of both creole and non-creole languages. Chaudenson (this volume) presents a compelling case where in 17th century French dialects, the use of bare nouns was much more prevalent than in contemporary French. Bruyn (1995, this volume) also provides evidence that Sranan displayed more bare nouns in its early stages than it does today.

. Bare nouns and functional innovation . A query While older languages such as Latin, Russian, and Hindi lack determiners, they all have number morphology on the Noun. But a number of creoles lack grammatical number affixes. So it seems that a reduced use of functional material leading to radical functional innovation may define creole languages, or at least the creoles analyzed in this volume. Chierchia’s blocking principle, adopted and elaborated upon here by Déprez and by Kester & Schmitt, links functional deficiency and bare NPs. This seems to be a natural hypothesis. The problems which remain, however, are (i) how are bare NPs interpreted, and (ii) does UG provide any specific grammar-internal mechanisms beyond pragmatics to compensate for the lack of number inflection and determiners? As the first problem is treated at length in the contributions to this volume, we would like to say a few words here concerning the second problem.

. On natural grammars Natural grammars contain two types of lexical items, substantive morphemes like nouns and verbs, and grammatical morphemes like conjunctions and articles. Focusing on the Noun Phrase, we assume that the basic nominal expression contains a lexical root, the Noun, plus additional functional material; namely, a number morpheme which heads a NbP located above the Noun, and a Determiner which heads a DP located above the NbP. N denotes a class of object, Nb. individuates the members of the class, and the determiner identifies the individuated element(s) in terms either of its presupposed existence, for definite determiners, or of its asserted existence, for indefinite determiners, with respect to the discourse world of the speaker. Creole grammars appear to be fully as productive as non-creole grammars with respect to substantive nouns, but they tend to lack grammatical morphemes denoting number (or individuation), and determiners that assert or presuppose reference in the discourse world (or identification). Consider the property of individuation. Chierchia (1998) has interestingly noted that languages are in complementary distribution with respect to the possession of plural morphemes or classifiers. English and Romance languages have plural morphemes while languages like Chinese or Korean have no

 Marlyse Baptista and Jacqueline Guéron

plural morphemes and effect individuation through classifiers. Soyoung Roger-Nun (2002) has shown for Korean that such complementarity, while real, does not necessarily characterize a language as a whole. It may characterize specific subsets of Nouns within a language. Thus in Korean and other languages, [+human] Nouns may bear either plural affixes or be preceded by classifiers, while [–human] Nouns must be preceded by a classifier. As mentioned in the Introduction (Chapter 1), Schmitt and Munn (1999), Déprez (1999, this volume), and Kester and Schmitt (this volume), have shown that the predicted complementary distribution of plural morphemes and classifiers does not hold for creole languages, which may lack both nb. morphology and classifiers. Their data are offered as evidence that falsifies Cherchia’s generalization. However, the fact that some creoles lack both number and classifiers may reflect a different phenomenon having to do with the different stages and degrees of grammaticalization of a language. It is possible that Cherchia’s generalization holds for “older” languages with full grammatical paradigms and does not hold for languages in which the nominal system only contains a subset of those grammatical paradigms. If the grammatical paradigms of creole languages do not match those that Chierchia’s typology addresses, then Chierchia’s generalization is not necessarily false, but simply doesn’t apply to these languages at this point in time. The nominal systems of various creoles were reported to be at different stages of grammaticalization by some authors in this volume (Alexandre & Hagemeijer on Santome, Baptista on Cape Verdean Creole, and Bruyn on Sranan). We assume that the reduced use of functional material which creole languages display may be attributed to the historical and sociological conditions of their birth, which most (but not all) creoles have in common. We may speculate – but how can we prove it to be true? – that radical functional innovation may possibly be attributed to the conditions under which the creoles developed, referred to in the Introduction to the volume. The first generation of slaves imported from Africa were presumably obliged to make themselves understood in some version of the colonizer’s language. We assume it is not so difficult to adopt new terms for objects present in the immediate discourse world but it is more difficult to adopt a system and a full set of grammatical morphemes, which are unrelated or irrelevant to daily experience. So the first slaves spoke a language that was lacking in grammatical morphemes, by necessity. The grammatical system of the new language was not easily accessible while that of the mother language did not fit in with the new lexical system. This is simplistic but plausible.2 From a language acquisition and learnability perspective, if one assumes that both children and adults are participants in creole genesis (Baptista 2002), then it would

. See Lefebvre’s relexification hypothesis for an entirely different scenario on the genesis of Haitian (1998).

Chapter 18. Functional deficiency, ellipsis or innovation in creole languages? 

not be hard to accept that adult learners experience difficulty with acquiring inflectional morphology in initial stages of L2 acquisition (DeGraff 1999: 517). However, instead of claiming that they practice, as a result, ‘functional category ellipsis’ (Lumsden 1996), we could reasonably assume that adult learners simply ignore or make abstraction of such morphology. DeGraff (1999: 497) formulates a highly plausible scenario for creole genesis applicable to our particular case of nominal systems. He hypothesizes that properties of the substrates and superstrates could be transferred to the creole via adult learners’ interlanguages. These interlanguages comprise at first a much-reduced range of approximations of target patterns (DeGraff 1999: 497) but the new generation of creolophone children introduce, in turn, UG-guided innovations into the linguistic pool. Hence, the original apparent simplification produced by the adults is ultimately compensated for highly complex principles in the children. At this point, several questions come to mind: How do creole languages make up for their lack (or reduced use) of number inflection and determiners? What is the nature of the radical functional innovation that UG makes available to languages which do not have the full set of grammatical paradigms available in other, older languages?

.

The role of UG

We assume that even if the maximal set of grammatical morphemes is not available, the purposes such morphemes fill, that of individuation and identifiability, must be satisfied by speakers. Every grammar is subject to specific output conditions on the interface level where grammatical form, or syntax, interacts with conceptual structure, or meaning. We assume that individuation and identifiability are grammatical interface concepts associated with NPs which must be expressed in a language. In languages like French or English, these concepts are expressed straightforwardly by number morphemes and determiners included within the maximal projection of DP. We propose that a variety of other means are available to express them when purely grammatical morphemes are unavailable. These means are more complex either in terms of their lexical content or in terms of their structural range. They can involve semanticized morphemes like quantificational operators which mingle grammatical and lexical content in the same lexical item within the DP. Or else they involve nonlocal processes like the binding of an empty element inside the nominal by a sentential operator outside of it. More concretely, individuation which is expressed by the opposition between singular and plural number in European languages may be expressed by a quantificational operator having the form of numerical “one” in the corresponding creole language. The identifiability which is expressed by a definite determiner in the European language may be expressed by a deictic operator which links the nominal to the discourse world in the creole, as in Cape Verdean Creole described by Baptista (1998). In other creoles, such as the French-lexified Haitian or English-lexified Jamaican, the European

 Marlyse Baptista and Jacqueline Guéron

language determiner is replaced by a pronominal element in the creole. Another, nonlocal means of expressing identifiability is to posit a null determiner bound by an overt or covert Topic in the Comp domain (cf. Gadelii, this volume). Identifiability expressed by an indefinite determiner in a European language may be associated with the very same quantifier that expresses individuation. In this case, however, an overt or empty quantifier is necessarily bound by the Tense node in T. When a determiner is null, a tense morpheme carried by either the Complementizer or the Tense nodes is crucial to identify its content. The hypothesis that grammatical morphemes are fewer in number in creole languages makes a number of predictions which are realized in fact by the data reported in this volume. 1.

In the absence or reduced use of grammatical determiners, a language has no “expletive” determiners. The expletive determiner is a grammatical morpheme par excellence as it is motivated by nothing other than syntactic constraints. Thus Baptista shows that while an expletive determiner introduces a Proper Name in EP, this is the case neither in Cape Verdean Creole nor even in Brazilian Portuguese (typically considered a non-creole language).

Without expletive determiners of the Italian type, impersonal sentences can only be introduced by bare NPs. 2.

3.

4.

In creole languages, number has a referential import where this is not necessary in the lexifier, or possibly the substrate, language. Kihm points out for Guinea-Bissau Creole, Baptista for Cape Verdean Creole, Spears for African American English,3 Alexandre & Hagemeijer for Santome, and Kester & Schmitt for Papiamentu, that number alone can identify a referential DP. This particular relation between form and function does not characterize any of the European lexifiers examined. A deictic element may assume the grammatical function in a creole that a determiner has in a lexifier language like French or Portuguese. Baptista shows that the determiner in Cape Verdean Creole hesitates between the construal of a deictic element and that of a grammatical determiner, gradually gaining use for generic or kind statements. This suggests that Cape Verdean Creole may be a creole which is evolving towards a more ordinary status in which the deictic morpheme will lose its referential content and function merely as an iota operator denoting either referential or non-referential objects, like le in French or o in European Portuguese. If number and deictic determiners both have referential content, then it is to be expected that bare NPs must be used for non-referential construals such as generic or kind statements.

. According to Spears, for a bare noun to be definite and plural in African American English, it must be marked with the plural. He calls such a feature a creolism, pointing out that African American English has creole properties, perhaps for historical and sociological reasons similar to those often suggested for the creoles.

Chapter 18. Functional deficiency, ellipsis or innovation in creole languages? 

.

On syntactic “bareness”

We ascribe to the opinions of Alexandre & Hagemeijer, Baptista, Déprez, Gadelii, Kester & Schmitt, Kihm, Kouwenberg, Spears and Zribi-Hertz & Glaude in this volume, according to which no referential NP can be bare, that is, deprived of both Nb and D projections. The theoretical reason is, as proposed by Baptista, following Guéron (2006) (see also Lecarme (1996, 2004)), that at the interface between syntax and interpretation, a referential nominal must be placed in discourse time. The referent of the nominal must be linked to the Tense chain of the sentence (Guéron and Hoekstra 1988) headed by the Comp node associated with the Speech or Reference time and continuing with the Tense node associated with the Event or Assertion time. Definite specific DPs whose discourse existence is presupposed in the sentence, are coindexed with the Comp node. Indefinite DPs, whose existence in the discourse world is asserted rather than presupposed, are coindexed with the Tense Node. Thus, indefinite DPs refer successfully only if the sentence which contains them turns out to be true, whereas definite/specific DPs refer successfully independently of the truth-value of the sentence. Only the D node can be linked to a tense node like T or C. N not being a grammatical category, it cannot be directly linked to a grammatical sentential operator. Nb. is grammatical, but it focuses inwards on N, not outwards on T for it is concerned with the individuation of members of the set N defines, not with the identifiability of the referent of N. Only a morpheme in D, be it a determiner or a deictic element, can provide the link between the element or elements selected from the set which N denotes, and the discourse world and time. We assume that referential Nb. can raise to D, or else stays in the Nb head while an empty D is linked to T or C. Referential number operators may even double as pronouns with which they share number and discourse-based content. This would account for the referentiality of the free morphemes wa and inen in Santome. These morphemes would be generated in NbP and raise to D, gaining referential status, like “one” in “only one is left” or “one ought not to do such things” in English. A bare D may be referential if it is identified by tense just as a bare pro or PRO may be referential if it is suitably identified.

. On the role of discourse In “older” languages, which have had the time to develop a full grammatical paradigm, individuation and identifiability are marked in DP by number affixes and determiner morphemes. In creole languages, weaker in grammatical morphemes, individuation is marked by semantically meaningful items such as quantifiers and deictic operators. Nb. is not only number but it asserts existence. The determiner is not a mere iota operator or an expletive, but is a deictic element.

 Marlyse Baptista and Jacqueline Guéron

When discourse anchoring is not supplied inside the DP by a determiner, it must be supplied by an element outside DP. Reference becomes more diffuse. Null nb. may be sg. or pl. according to contextual necessity, while a null determiner may be bound by a discourse topic, as Gadelii proposes for Lesser Antillean.

. Subject-object asymmetries In both European languages and in creoles, bare NPs are able to appear in the direct object position. This is the sole position for which we would posit the insertion of a syntactically bare NP. In this position, the lexical content of a bare NP can combine with the lexical content of a bare V to denote an activity or a state. Then the NP can acquire an existential value indirectly, via the binding of the T feature of the verb by the Event or Assertion time morpheme in T. A bare object cannot be both incorporated and referential, since a referential determiner is bound by C independently of the content of VP. Incorporated NPs are necessarily existentially quantified, under syntactic Locality.

.

On agreement

Is it not paradoxical that all the European lexifiers considered in this volume have number and gender agreement between subject and verb and/or between the noun and the determiner or adjective, while the creoles on the whole do not? This disparity between the lexifier and the creole follows from our hypothesis. Agreement in most cases involves number agreement. Even if gender is part of the agreement feature complex, gender agreement is usually parasitic on Nb. agreement (cf. Renault 1989). This is not absolute, however: in Arabic, verb and subject agree in gender and person but not in nb. whenever the order of constituents is VS. Agreement is a process that checks the identity of the grammatical features of Noun and Verb, or Noun and Determiner or Adjective. In creole languages, which tend to lack grammatical number, we do not expect to find agreement, while we do expect to find agreement both in the substrate and the lexifier languages. The absence of number and gender agreement that distinguishes creole languages from their source languages is expected, if absence or reduced presence of functional morphemes and radical grammatical innovation characterize a creole language.

. Conclusion .

Creoles and the source languages

Creole languages are natural languages, like their source languages. Yet they differ from their source languages formally and historically, two factors which are plausibly related. The creole languages were formed in circumstances that may not have allowed them to either borrow the grammatical system of the lexifier or maintain that of the

Chapter 18. Functional deficiency, ellipsis or innovation in creole languages? 

substrate languages. The first speakers developed a grammatical system which enabled them to represent individuality and identifiability on the basis of lexical material, or more precisely, material with both lexical and grammatical content. Instead of a nb. morpheme, they used a free morpheme functioning as a quantifier, like strong “one” in English. Instead of a determiner they used a deictic morpheme or even adopted a pronoun for a grammatical function. In the absence of a determiner, they used peripheral properties, such as topics able to bind a null determiner. As all these mechanisms belong to UG, the language developed is a natural language, so there was no reason for it to be rejected or even profoundly modified by subsequent generations of speakers. But as a language formed in less than optimal circumstances, it relies more heavily on the more accessible semantic features of lexical items even to accomplish the grammatical functions of individuation and identifiability. As a result, the creoles do not display a maximal distinction between lexical and grammatical morphemes. Maximal distinction could possibly be achieved over time, as creoles evolve towards greater degrees of grammaticalization. The use of semantically charged morphemes like quantifiers, deictics, or pronouns as grammatical morphemes to express the obligatory semantic functions of individuation and identifiability no doubt characterizes all languages to some extent. But in older languages, the mixed morphemes have been weakened. French ille has been weakened to il. English strong quantificational ‘one’ has been weakened to grammatical ‘a’. Similarly, null morphemes are bound by overt or covert topics in older languages as well as in creoles. English has a null pronominal bound by an overt topical Focus in ‘John Mary likes e’. Number varies in its referential value in non-creole languages. Number in English is more referential than in Romance. This allows the plural morpheme to raise to D in English, so that in “men arrive,” the subject gets asserted reference by binding to T. In French, where number is purely grammatical, *Hommes sont arrivés ‘men arrived’ is unacceptable. In Romance, only an overt determiner can be bound by T. Similarly, we may hypothesize that in languages in which there is no number agreement in VS order, but agreement in SV order, as in Arabic, number has referential properties that require it to agree in nb. with a subject in a topic-like position but not with one in the non-topical post-verbal position. In Romance, where number is not referential but purely grammatical, post-verbal subjects agree, in general, in number with the verb, just like pre-verbal subjects. All creoles are not identical and neither are all older languages. Nor is there an open and shut definition of a creole. Brazilian Portuguese seems to be a semi-creole, for instance.

. Towards verifying the hypothesis We have tentatively suggested that a creole language may be characterized as one that accumulates lack (or reduced use) of grammatical determiners and of grammatical

 Marlyse Baptista and Jacqueline Guéron

number morphemes and as a result, is radically functionally innovative. Instead of number affixes, it uses quantifiers or referential number morphemes to satisfy the grammatical necessity for individuation. Instead of grammatical determiners, it uses deictic markers, null determiners bound by a discourse topic, or even pronouns, to satisfy the requirement for identifiability. In principle this hypothesis can be tested. (i) The data and analyses given here for creoles with European lexifiers would have to be checked against a similar study of creoles with non-European lexifiers, say, creoles with Arabic, Asian, Austronesian or Amerindian lexifiers, among others. This work must be done by researchers with as much competence in these creole languages and their lexifiers as our contributors have shown for creoles with European lexifiers. Until this work is done, we cannot prove that a creole exhibits a reduced use of functional material and radical functional innovation as a compensation in the ways we have suggested here. (ii) The work done here for the nominal systems of creoles with European lexifiers must be duplicated for the temporal and aspectual systems of the same languages. Is it the case that tense in creoles can be expressed by lexical morphemes or by an ambiguously construed null morpheme, just as identifiability is expressed by an ambiguously construed null determiner? Is aspect expressed in a creole by morphemes with lexical content where the European lexifier uses a specialized grammatical morpheme? How different are the lexifier and the creole in this respect? The editors hope to attack this second project, with the aid of the community of creolists, in the near future, thus profiting from and, at the same time, testing the tentative hypotheses formulated in this Postface.

References Baptista, M. 1998. The role of null morphemes in marking specificity and nonspecificity of NPs: A comparative analysis. Paper presented at the Society of Pidgin and Creole Linguistics/ Linguistic Society of America. Held in New York City. January 8–10, 1998. Baptista, M. 2000. Variabilité référentielle des syntagmes nominaux nus en créole cap-verdien. Paper presented at the University of Paris III Linguistics Symposium. April 7th, 2000. Baptista, M. 2002. The Syntax of Cape Verdean Creole: The Sotaventa varieties. Amsterdam: John Benjamins. Baptista, M. 2006. When substrates meet superstrate: The case of Cape Verdean creole. In Cabo Verde: Origens da sua Sociedade e do seu Crioulo, J. Lang, J. Holm, J.-L. Rougé & Maria João Soares (eds), 91–116. Tübingen: Gunter Narr. Bruyn, A. 1995. Grammaticalization in Creoles: The development of determiners and relative clauses in Sranan [Studies in Language and Language Use 21]. Amsterdam: IFOTT. Chierchia, G. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. DeGraff, M. 1999. Language Creation and Language Change: Creolization, diachrony, and development. Cambridge MA: The MIT Press. Déprez, V. 1999. De la nature sémantique des nominaux sans déterminant en créole haitien. Langues 2(4): 289–300.

Chapter 18. Functional deficiency, ellipsis or innovation in creole languages?  Guéron, J. & Hoekstra, T. H. 1988. T-chains and the constituent structure of auxiliaries. In Constituent Structures, A. Cardinaletti, G. Cinque & G. Giusti (eds), 35–99. Venice: Annali di Ca’ Foscari. Guéron, J. 2006. Generic sentences and bare plurals. In Non-definiteness and Plurality [Linguistik Aktuell/Linguistics Today 1995], S. Vogeleer & L. Tasmowski (eds), 219–234. Amsterdam: John Benjamins. Kegl, J., A. Senghas & Coppola, M. 1999. Creation through contact: Sign language emergence and sign language change in Nicaragua. In Language Creation and Language Change: Creolization, diachrony, and development, M. DeGraff, (ed.), 179–237. Cambridge MA: The MIT Press. Lecarme, J. 1996. Tense in the nominal system: The Somali DP. In Studies in Afroasiatic Grammar, J. Lecarme, J. Lowenstamm & U. Shlonsky (eds), 159–178. The Hague: HAG. Lecarme, J. 2004. Tense in Nominals. In The Syntax of Time, J. Guéron & J. Lecarme (eds), 440– 475. Cambridge MA: The MIT Press. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The case of Haitian Creole. Cambridge: CUP. Lumsden, J. 1996. On the acquisition of nominal structures in the genesis of Haitian creole. In Creole Languages and Language Acquisition, H. Wekker (ed.), 185–205. Berlin: Mouton de Gruyter. McWhorter, J. 2001. The World’s Simplest Grammars are Creole Grammars. Linguistic Typology 5: 125–166. Munn, A. & Schmitt, C. 2001. Bare nominals and the morphosyntax of number. Current issues in Romance Linguistics, T. Satterfield, C. Tortora & D. Cresti (eds), 217–231. Amsterdam: John Benjamins. Renault, R. 1989. Genre grammatical et typologie linguistique. Bulletin de la Société Linguistique de Paris LXXXII (1): 69–117. Roger-Nun, S. 2002. Les Expressions Nominales à Classificateurs et les Propositions à Cas Multiples du Coréen. PhD Dissertation. Université Paris III-Sorbonne Nouvelle. Schmitt, C. & Munn, A. 1999. Against the Nominal Mapping Parameter: Bare nouns in Brazilian Portuguese. In Proceedings of NELS 29, P. Tamanji, M. Hirotani & N. Hall (eds), 339–353. University of Delaware. Schmitt, C. & Munn, A. 2003. The syntax and semantics of bare arguments in Brazilian Portuguese. Linguistic Variation Yearbook 2: 185–216.

Subject index

A abstract , , , –, –, –, , , , , , –,  abstractability –, –, ,  acrolectal , –, ,  adjective(s) –, , , , , , –, , , , , , , –, , , , , , ,  adjectivized nouns ,  adverb(s) , –, –, , –, , –, , , , , , , –, , , ,  african American standard English – defined , , , , –, , ,  africanism  agglutination (agglutinated determiner) , ,  agree , , , , , –, , , , , – agreement , , , , , , , , , , , , , , , –, –, , – anaphor –, , , , , –, , , –, , , , , , , , –, –, , –, , , , –, , ,  anaphora , , , , , 

anaphoric nouns , ,  animacy , , , , , , , , , , , , , , , , , , , – animacy hierarchy  animate , –, , , , , , –, , –, ,  anthroponyms  appelatives ,  argument(s) , , , , , , , , , , , , –, , –, , –, , , –, , , , , , –, , , , , , –, , , –, , –, –, –, , –, , , , –, –, –, –, –, –, , , , –, , , –, ,  article(s), definite –, , , , , –, , –, , , , , –, , , –, , –, , , , , , , , , –, , –, , –, –, , , , , –, , , –, –, , –, ,  article(s), indefinite , –, , , –, –, , , , , , , , , , , , , , , , , –, –, 

articulatory  aspect , , –, , , –, , , , , , , , , , , , , , , , ,  aspectual , , , , –, , , –, , ,  see also aspectual be , –, ,  associative construction  associative plural , ,  asymmetry(ies) , , , , , , –, , , –,  Atlas ,  atomic , , , , –, , – attributes , ,  auxiliary be –,  B bare nominal(s) , –, , , , , –, , , , –, , , , , , , –, ,  bare nouns –, –, , –, –, –, , –, , , –, , , , –, –, –, –, –, , –, –, , –, , , , –, –, –, –, , –, , –, , –, , –, –, –, –, , , , , , , , , –,

 Subject index –, , –, –, –, , , , , –, –, –, –, , , , –, , – bare plural , , –, , , , , , –, –, , , , , , , , , , , –, –, ,  bare singular , , , , , , , , , –, –, , –, –, –, , , –, , , , ,  base-generated , ,  basilect  basilectal , , –, , –, , –, , , , , ,  Blocking principle ,  BNP(s) , –, –, , –, –, –, –, , , , , , –, – bridging context  C capacity reading – cardinal –, ,  categorical , , , , , , , , ,  chain –, –, , ,  characterizing sentence ,  checking , , ,  classification , , , –, , ,  classifier , –, , –, , , , , , , , –, , –, , – classifier phrase ,  clitic , –,  collective noun ,  collectives , , , , , 

common names, nouns , , , , , , ,  complementary distribution , , , , –, , , , , –, , – complementizer , , ,  complements , , , , , , –, , ,  conceptual , ,  construal(s) , –, , –, , , –, –, , , –, –, , ,  construction(s) , , , , , , , –, , , , , , , , , , –, , –, , , , , –, , , , , , –, , , , , –, , , ,  contrastive focus , –,  copula be – corps  count (noun) , , , , , , –, , , , –, , –, –, –, , –, , –, , , , , , , –, , , , ,  creole(s) –, –, –, –, , –, , , , –, –, , , , , , , –, , –, , –, –, , –, , –, , –, , –, –, –, , –, , –, –, , , , –, , –, –, , –, –, , –, –, –,

, , –, , –, , , –, , –, –, , –, , –, , –, –, – see also genesis , , , – see also creole languages –, –, –, , –, , , , , , , –, , , , –, –, –, –, –, , –,  -, , –, , –, –, , , –, , – creolism , , –,  creolization , , ,  creoloid  D D1 , – decreolized ,  deficiency , , –, , , , –, , , – definite , ,  definite bare nouns , ,  definite determiner –, , , , –, –, –, , ,  definite expression ,  definiteness –, , –, –, , , , –, , – deictic element , – deixis , , , , –,  demonstrative , , –, –, , , –, –, , , –, , , –, , –, –, , –, , , –, , –, , , , –,  demonstrativity ,  denotation , , –, , , , , , , , , –, –, , 

Subject index  determined noun(s) , ,  determiner(s) –, –, –, –, –, –, –, –, –, –, –, , –, –, , –, –, –, –, –, , –, , –, –, –, –, –, , , , , –, –, , –, –, , , , , –, –, –, –, , –, , , –, –, , –, –, , , –, , , , –, , , –, –, –, –, , , , –, , , , , , , , , –, –, –, –, , –, , –, –, , –, –, – determiner Phrase (DP) ,  determiner system , –, , –, , , , , , , –, , –, , , , , , , , , –, , , , , –,  determinerless noun(s) , –, –, , , , , , , ,  D-feature  diglossia ,  direct verbal complements  discourse , , –, –, –, , , , , , –, –, , –, , , , –, –, –, , –, , , , , , , , –, , , , , , –, , –, , –, , ,

, , –, –, , , , –, –, –, , –, –,  see also discourse anaphora , , , , ,  see also discourse status ,  distributed morphology ,  distribution , , , –, –, , –, –, , , –, –, –, –, –, –, , , , , –, , , , , –, , , , –, , –, –, –, –, , , , , , –, –, –, , , –, , , –, – distributive , ,  D-linking  do support – DP , , –, , –, , –, , –, , –, –, , –, –, , –, , , , –, , –, –, , –, , , , , , –, , –, –, , –, –, –, , , , –, , –, –, – DP hypothesis , , , ,  DP structure –, , , , , , , , , , ,  E E-language  elsewhere , , , ,  EP (European Portuguese) , ,  episodic , , –, , , , –, , –,

, , , , , , –, , ,  epithet ,  erasing rule  event time –, ,  eventuality , –, –, – existential bare nouns  see also existential generic , , , , , , – see also existential habitual , – see also existential quantifier , , –,  see also existential(s), existential reference –, –, –, , , , , , , –, , , , , , –, –, –, –, , –, , , –, –, , , –, –, ,  expletive –, , , , – extension , , , –, , , , , , , –,  extensional , ,  F factive, factivity , –, , , , , , –, – familiarity , , –, , , , , , , , , ,  focus , –, , , , , , , , –, ,  frozen expression(s) , , –, – functional deficieny , , ,  functional head , , , , ,  functional structure , , , , , , , , , , –, 

 Subject index G gender , , , –, , , , , , , , , , , , , , , ,  gender features  general number , ,  generative , , ,  generativist , ,  generic –, , –, –, , , –, –, , –, –, –, , –, –, –, , –, , , , –, , –, , , , , , , –, –, , , , , , –, –, –, , –, –, –, , , –, –, –, , –, –, –, –, , –, –, –, –, , –, , –,  see also generic interpretation , , , , , , –, , –, , , ,  see also generic operator , , , ,  genesis , , , , , , , , –, – grammar(s) , –, , , , , –, , , , –, –, –, –, , , –, , , , , , , , –, –, , , –, ,  grammatical features , , , –, ,  grammatical function , , ,  grammatical systems  grammaticalization , , –, , , 

H habitual be  habitual operator (HAB) ,  head movement , , ,  history of the French Language – I identifiability , –, , , –, , , , , , , –, – see also identifiable , , , , , , , , , , , , – I-language  inalienable possession , , –, , – inanimate , , , , , , ,  incorporation , , , , , –,  indefiniteness , , , , , , , , , , , , , , –, , , , , ,  indefinite , ,  indefinite determiner –, , –, , , , –, , –, , , –, , , , –, , , , , , , , , , , , , , , , , , , , –, , , ,  indefinite expression  index –, , –, –, , ,  individual-level , , , , , , , –, – individuation , , , , , , , , , –, –, –, , , –, –, , , –, – inflection –, –, , , , , –, 

inflectional –, , , , , , , , , –,  informational component , ,  instantiation of kinds ,  intensional , –, ,  interpretation –, , –, –, , –, –, , –, , –, , –, , –, , –, , –, , –, , , , –, –, –, , –, , , –, –, –, , , , , –, , , , –, , , –, –, , , –, , , –, –, –, , –, , , –, , , , , , –, –, –, –, –, , –, –, , –, –, , , , –, –, ,  interpretive –, , , , , , , , –,  iota-operator, , , – K kind interpretation –,  kind terms ,  kind(s) , , –, , –, , –, , –, , –, –, , , , , , , , , –, , –, –, , , , , ,  kind-denoting expression , ,  kind-level predicate  kinship terms , , 

Subject index  L lexical incorporation , –,  lexicalization – lexically-governed position  lexicon –, , , –, , , , , , , , –, , , , , , , , –, , , , ,  lexifier(s) –, , , , , , , , , , , , , , , , –, , , , , –, , , , –, –, , –, , ,  LF –, , , , ,  light verb(s) , , –,  M marked , , , , , , , , , , , –, , , –, , –, –, , , , , , –, , , , , , –, –, , –, , , , , , , , , , , , –, , – see also markedness , , ,  mass interpretation , , , –, –, , , ,  mass noun , , , , , , , ,  mass/count contrast  metatypy ,  Minimalist , ,  morphological , , , , –, , –, , –, , –, –, , , , , , , , , , –, , , –

morphology , , –, , , , –, , , , , –, , –, , , , , , , , , , , , , –, , –, –, , , – movement –, , , , , , , –, –, , , , , ,  mutual intelligibility (Spanish/Palenquero)  N narrative – narrow scope , , , ,  negation –, , , –, , , –, –, –, , , , , , , –,  see also standard negation, constituent negation  new information , , , , –, , ,  node , –, , , –, , , , – nominal expression(s) , , , , , , , , , –, , –, , –, –, –, ,  nominal mapping parameter , , , , ,  non-argument  nonargumental , , , , , ,  non-atomic , , –, – noncreole , , , –, , –, –, , , , , ,  nonindividuated, nonindividuation , , , , , , , , , –, ,  nonspecific , –, , –, , –, –, –, –, , ,

, –, , , –, –, , , , –, , , , –, –, –,  see also non-specificity , , , ,  noun phrase(s) , –, , , , –, –, –, , , , , –, , –, –, –, , , –, , , , , , , , , –, , , –, , –, –, , , –, , –, –, , , , , , , –, , , , ,  noun phrase structure  NP(s) , , –, –, , –, –, , –, , , –, , –, , –, , , , –, , , –, , –, , , –, , , , –, , –, , –, –, , , , –, –, –, –, –, –, , –, , , , , , , , –, , , , –, –, –, –, –, –, –, –, –, , –, –, , , –, , , –, , , –, – NP marking , ,  N-raising , –, , ,  see also N-to-D raising , – null determiner , , , , –, , –, , , –, , , , , – number , , –, –,

 Subject index –, –, , , , –, –, –, , , , , , , , , –, –, –, , –, , , –, –, , , , , , , , , , , , –, –, –, , , , , , , , , –, –, –, –, –, , , –, , , –, –, , –, , , , –, , –, , , –, –, –, –, , , , , –, –, , , , , – see also number inflection , ,  see also number marking , , , , , , , , , , , , , , , , , , , , –, , , – see also number phrase (see NumP) , , , , , , , ,  numeral , , , –, , , –, –, , , , , , , , –, –, , –, –, , , , –, , , , ,  NumP , , , –, , , , –, , , –, –, , –, , –, , –, –, –, –, –, –, , –, , , ,  O object

, , , –, , –, –, , , –, –, , –, –, –, , , , , , –, –,

–, , –, , , , , –, –, , , , , , , –, –, , , –, –, –, , , –, , , , , –, , , , –, –, , ,  see also object position , , , , , , , , –, , –, –, , , , , , –, , , , , , –, , , –, , , –,  old information , ,  omission , , , , , –, ,  operator(s) –, , , , , , –, , , , , , , –, , , , , , , , – P Palenquero dialogue (excerpt)  parameter , , , , , , , , , , , , , –, –, ,  partially restructured –,  particle , –, , , –, ,  partitive , –, , , , ,  patient-trigger ,  phonological , , , –, , , , , , –, –, , ,  phonology , ,  phrasal movement , –, – pidgin –, , , ,  pidgins , ,  plantation creole(s) , , ,  plural , , , –, –,

- , –, –, –, –, –, , –, –, -, , , –, –, , , –, –, –, , –, –, –, –, –, –, –, , –, –, –, , –, , , –, –, , –, –, –, , –, , –, –, , , –, –, –, , –, , –, –, , –, , , , , , , , , , –, , –, –, –, –, –, –, –, –, –, –, , –, –, –, ,  see also plural morphology –, , , , , , , , , –, –,  see also plural parameter –, , , –, –,  see also plurality , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,  pluralization , , –, , –, ,  portmanteau morpheme – possessive , , , , , , , , , –, , , –, , , , –, , , –,  pragmatic referentiality , , –, , –, , –, , –,  pragmatic(s) , , –,

Subject index  , , , , , , , , , , , , , , , , , –, , , , , , , , , , , , –, –, – predicate noun, predicate nominal – prepositional complements , – presupposition , , ,  principle(s) , , –, , , , , , , , , , , –, , , , , , , , , , –, ,  pro  PRO  projection(s) , , , , , , , , –, –, –, , , , , , –, –, , , , , , –, , , , –, , , , –, –, –, , –, –, , , , ,  proper name(s) , , , , , , , , , –, , , , , , –, , –, , , –, ,  see also proper nouns , , , , , ,  pseudo-determiners  Q QP –,  quantification , , , ,  quantificational , , , , , , ,  quantified expression  Quantifier phrase ,  quantifier(s) , , ,

–, , , –, –, , –, , , , , , , , , , , , , , , , , –, –, –, , , –, – R reference , –, –, , , , , , , , , , , , , , , , , , –, , , , , , –, –, , , –, –, , , –, –, –, –, , , , , –, –, , – reference time  referent , , , , , , , , , , , , , , , , , –, , , , , , , , , –, , –, , –, –, , , , , , , , , , , –, –, , , –, , , –, , –,  referential , –, –, –, –, , –, –, , –, , , –, –, –, –, , , , –, , –, , , , , , –, , , , , , –, , , , , , , – referentiality , , , –, , –, –, , , , , –, –, –, –, , –, , –,  see also referentiality,

pragmatic , , –, , –, , –, , –,  see also referentiality, semantic , , , ,  representation , , , , ,, , , , , , –,  restructuring ,  S salience ,  saliency , ,  saturation, arguments as saturated expressions – scope , , , –, , , , , , , , , , , , , , , , , , , , –, , , , , –, –,  semantics , , , , –, –, –, – –, , , , –, ,, , , , , , , , , , , , , ,  semi-auxiliary  semi-creole , , ,  serial-like verb constructions  singularity , , , , , , –, , –, , , ,  slavery ,  slaves , , ,  snowballing  Spec-DP –, –,  specific , , , , , , , –, , , , –, –, –, –, , –, –, –, , , –, , , –, –, –, –, , , , , , , , , , –, –, , ,

 Subject index , , , , , , , , , , , , –, –, –, , , , , , –, , –, –, –, , –, –, , –,  specificity , , , , , , , –, , , –, , , , –, , , –, , –, –, , , , –, –, , , ,  Spec-NumP – speech time ,  stage-level , , , , –, – structure , , , , , –, , , , , , , –, , –, , –, –, , , , –, , , , , –, , , –, , , , –, –, –, , , –, , –, , , , , , , , , –, , , , –, –, , , , , –, , , , , , –, –, –, , , ,  structure of DP , , ,  subject , –, –, , –, , , –, –, –, , , –, , , , , , , , –, , , , , , , , , –, , , , , –, , , , , , , , , , –, , , , , , , , , –, , , , –, , –, , , – see also subject position

, , , , –, , , , –, , , , ,, , , –, , , , – substrate , , , , , , –, , , , , , , –, , , , , – syntactic , , , , , , , , , –, , , , , , , , , , , , , , , , , , , , , , , –, –, , , , , , , , , , , –, , , , –, , –, , , , , –, , –, , , , , , , , –, , , – see also syntactic structure , , , , –, , , , –, , , , , ,  syntax , , , , , , , , , , , –, , , , , , , –, –, , , , , , , –, –, , , , , –, , ,  T T , –, , , –, –, , , , – T , – taxonomic interpretation  taxonomic reading –,  T-chain , – Tense , , , , , –, , –, –, , –, –, , , , , , –,

, –, , , , –, –, –, , , , , , , ,  see also Tense, episodic , ,  see also Tense, generic , –, –, –,  theoretical , , , , , , , , –, , , , , , , , ,  thetic  topic , , , , , , , , –, , , , , , –, , , , , , , , , , , ,  topic time , , ,  topicality , –, , , , , –, –, ,  toponyms , , ,  transnumeral , , ,  type-shifter  type-shifting operation  typology , , , , , , ,  U UG

, , , , , –, ,  underspecification  underspecified , , , , ,  uniqueness , , , , , , , , , , , , , , , , –, ,  see also uniqueness presupposition  see also uniqueness, unique reference , , , , , , , , , , , , , , , , –, , , , , , , –,  universal , , , ,

Subject index  , , ,  universal grammar , , , – unmarked , , , , , , , , , , , , , , , , , –, , , , , , , , , –,  V variability , , , , , – variables , , 

variants , , , , ,  variation , , , , , , , , , , , , , , , , , , , , –, , , , , ,  varieties , , , , , –, , –, –, , , –, , , , –, , , –, –, , 

W Wh-movement  X XP

, , , , –, 

Z zero determiner , , , –, –, 

In the series Creole Language Library the following titles have been published thus far or are scheduled for publication: 32 Huber, Magnus and Viveka Velupillai (eds.): Synchronic and Diachronic Perspectives on Contact Languages. 2007. xii, 370 pp. 31 Baptista, Marlyse and Jacqueline Guéron (eds.): Noun Phrases in Creole Languages. A multi-faceted approach. 2007. ix, 493 pp. 30 Siegel, Jeff, John Lynch and Diana Eades (eds.): Language Description, History and Development. Linguistic indulgence in memory of Terry Crowley. 2007. xv, 514 pp. 29 Deumert, Ana and Stephanie Durrleman (eds.): Structure and Variation in Language Contact. 2006. viii, 376 pp. 28 Clements, J. Clancy, Thomas A. Klingler, Deborah Piston-Hatlen and Kevin J. Rottet (eds.): History, Society and Variation. In honor of Albert Valdman. 2006. vi, 304 pp. 27 Escure, Geneviève and Armin Schwegler (eds.): Creoles, Contact, and Language Change. Linguistic and social implications. 2004. x, 355 pp. 26 Mous, Maarten: The Making of a Mixed Language. The case of Ma’a/Mbugu. 2003. xx, 322 pp. 25 Migge, Bettina: Creole Formation as Language Contact. The case of the Suriname Creoles. 2003. xii, 151 pp. 24 Mühleisen, Susanne: Creole Discourse. Exploring prestige formation and change across Caribbean English-lexicon Creoles. 2002. xiv, 332 pp. 23 Smith, Norval and Tonjes Veenstra (eds.): Creolization and Contact. 2001. vi, 323 pp. 22 Neumann-Holzschuh, Ingrid and Edgar W. Schneider (eds.): Degrees of Restructuring in Creole Languages. 2001. iv, 492 pp. 21 McWhorter, John (ed.): Language Change and Language Contact in Pidgins and Creoles. 2000. viii, 503 pp. 20 Rickford, John R. and Suzanne Romaine (eds.): Creole Genesis, Attitudes and Discourse. Studies celebrating Charlene J. Sato. 1999. viii, 418 pp. 19 Spears, Arthur K. and Donald Winford (eds.): The Structure and Status of Pidgins and Creoles. Including selected papers from meetings of the Society for Pidgin and Creole linguistics. 1997. viii, 461 pp. 18 Escure, Geneviève: Creole and Dialect Continua. Standard acquisition processes in Belize and China (PRC). 1997. x, 307 pp. 17 Thomason, Sarah G. (ed.): Contact Languages. A wider perspective. 1997. xiii, 506 pp. 16 Clements, J. Clancy: The Genesis of a Language. The formation and development of Korlai Portuguese. 1996. xviii, 282 pp. 15 Arends, Jacques, Pieter Muysken and Norval Smith (eds.): Pidgins and Creoles. An introduction. 1994. xv, 412 pp. 14 Kihm, Alain: Kriyol Syntax. The Portuguese-based Creole language of Guinea-Bissau. 1994. xii, 310 pp. 13 Arends, Jacques (ed.): The Early Stages of Creolization. 1996. xvi, 297 pp. 12 Byrne, Francis and Donald Winford (eds.): Focus and Grammatical Relations in Creole Languages. Papers from the University of Chicago Conference on Focus and Grammatical Relations in Creole Languages. 1993. xvi, 329 pp. 11 Byrne, Francis and John Holm (eds.): Atlantic Meets Pacific. A global view of pidginization and creolization. 1992. ix, 465 pp. 10 Winford, Donald: Predication in Caribbean English Creoles. 1993. viii, 419 pp. 9 Byrne, Francis and Thom Huebner (eds.): Development and Structures of Creole Languages. Essays in honor of Derek Bickerton. 1991. x, 222 pp. 8 Bailey, Guy, Natalie Maynor and Patricia Cukor-Avila (eds.): The Emergence of Black English. Text and commentary. 1991. x, 352 pp. 7 Fabian, Johannes (ed.): History from Below. The “Vocabulary of Elisabethville” by André Yav: Text, Translations and Interpretive Essay. With the assistance of Kalundi Mango. With linguistic notes by W. Schicho. 1990. vii, 236 pp. 6 Singler, John Victor (ed.): Pidgin and Creole Tense/Mood/Aspect Systems. 1990. xvi, 240 pp. 5 Jackson, Kenneth David: Sing Without Shame. Oral traditions in Indo-Portuguese Creole verse. 1990. xxiv, 257 pp. 4 Lipski, John M.: The Speech of the Negros Congos in Panama. 1989. vii, 159 pp.

3 2 1

Byrne, Francis: Grammatical Relations in a Radical Creole. Verb Complementation in Saramaccan. With a foreword by Derek Bickerton. 1987. xiv, 293 pp. Sebba, Mark: The Syntax of Serial Verbs. An investigation into serialisation in Sranan and other languages. 1987. xv, 218 pp. Muysken, Pieter and Norval Smith (eds.): Substrata versus Universals in Creole Genesis. Papers from the Amsterdam Creole Workshop, April 1985. 1986. vii, 311 pp.