New Perspectives on Romance Linguistics: Morphology, Syntax, Semantics, and Pragmatics : Selected Papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005

New Perspectives on Romance Linguistics VolUME I: Morphology, Syntax, Semantics, and Pragmatics

AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE General Editor E.F.K. KOERNER (Zentrum für Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin) Series IV – CURRENT ISSUES IN LINGUISTIC THEORY Advisory Editorial Board Lyle Campbell (Salt Lake City); Sheila Embleton (Toronto) Brian D. Joseph (Columbus, Ohio); John E. Joseph (Edinburgh) Manfred Krifka (Berlin); E. Wyn Roberts (Vancouver, B.C.) Joseph C. Salmons (Madison, Wis.); Hans-Jürgen Sasse (Köln)

Volume 275

Chiyo Nishida and Jean-Pierre Y. Montreuil (eds.) New Perspectives on Romance Linguistics Volume I: Morphology, Syntax, Semantics, and Pragmatics Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005

New Perspectives on Romance Linguistics VolUME I: Morphology, Syntax, Semantics, and Pragmatics Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005

Edited by

Chiyo Nishida Jean-Pierre Y. Montreuil University of Texas, Austin

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA

4-

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   New perspectives on romance linguistics. Vol.I: Morphology, Syntax, Semantics, and Pragmatics. Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005 / edited by Chiyo Nishida and Jean-Pierre Y. Montreuil.    p.   cm. -- (Amsterdam studies in the theory and history of linguistic science. Series IV, Current issues in linguistic theory, ISSN 0304-0763 ; v. 275)  Includes bibliographical references and index. 1. Romance languages--Congresses PC11 .L53  2006 440--dc22 2006045892 ISBN 90 272 4789 7 (Hb; alk. paper) © 2006 – 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

CONTENTS

Acknowledgements

vii

Introduction

ix

A Polarity-Sensitive Disjunction: Spanish ni ... ni Raúl Aranovich

1

Taking a Closer Look at Romance VN Compounds Reineke Bok-Bennema & Brigitte Kampers-Manhe

13

Beyond Descriptivism: Exaptation and Linguistic Explanation Denis Bouchard

27

Do Subjects Have a Place in Spanish? José Camacho

51

On the Conceptual Role of Number Viviane Déprez

67

The Diachronic Development of a French Indefinite Pronoun: Comparing Chacun to Aucun Amanda Edmonds

83

A Syntactic Analysis of Italian De-verbal Nouns Franca Ferrari-Bridgers

97

V-N Compounds In Italian: A Case of Agreement in Word Formation Martina Gračanin-Yuksek

113

A Reinterpretation of Quirky Subjects and Related Phenomena in Spanish Rodrigo Gutiérrez-Bravo

127

Cognitive Constraints on Assertion Scope: The Case of Spoken French Parce que Knud Lambrecht, Julia Bordeaux & Robert Reichle

143

vi

CONTENTS

Avant que- or Avant de-Clauses: When Presupposition Gives Way to an ‘Assertive Construction’ Anne Le Draoulec

155

Null Directional Prepositions in Romanian and Spanish Jonathan E. MacDonald

169

A Unified Account for the Additive and the Scalar Uses of Italian Neppure Alda Mari & Lucia M. Tovena

187

Default Morphology in Second Language Spanish: Missing Inflection or Underspecified Inflection? Corrine McCarthy

201

Early Object Omission in Child French and English Ana T. Pérez-Leroux, Mihaela Pirvulescu & Yves Roberge

213

Agreement Paradigms Across Moods and Tenses Mihaela Pirvulescu

229

Italian Volerci: Lexical Verb or Functional Head? Cinzia Russi

247

Restructuring of Reverse Psychological Predicate Almeida Jaqcueline Toribio & Carlos Nye

263

Subject Index

279

ACKNOWLEDGEMENTS

The papers selected for inclusion in this volume originated as presentations at the Thirty Fifth Annual Meeting of the Linguistics Symposium on Romance Languages (LSRL 35), with parasession on Current Approaches in Historical Linguistics and Dialectology, which took place February 24-27, 2005, at the University of Texas, Austin, Texas. The editors gratefully acknowledge the scholars listed below for their generous assistance in selecting the papers to be presented at LSRL 35 and for reviewing those submitted for inclusion in this volume: Jean-Marc Authier, Dalila Ayoun, Brigitte Bauer, Paola Benincà, David Birdsong, Antonietta Bisetto, Carl Blyth, Ignacio Bosque, Joyce Bruhn de Garavito, Luigi Burzio, Greg Carlson, Ileana Comorovski, Viviane Déprez, Anna-Maria Di Sciullo, Jon Franco, Grant Goodall, Jacqueline Guéron, Javier Gutierrez-Rexach, Paul Hirschbühler, Mary Kato, Paula Kempchinsky, Ekkehard König, Knud Lambrecht, Geraldine Legendre, Rochelle Lieber, Andrew Lynch, France Martineau, Pascual Masullo, Silvina Montrul, Francisco Ordóñez, María Luisa Rivero, Yves Roberge, Sergio Scalise, Carmen Silva-Corvalán, Dominique Sportiche, Christina Tortora, Enric Vallduvi, Dieter Wanner, Steve Wechsler, Karen Zagona, Raffaella Zanuttini Additional thanks to Knud Lambrecht for his Outreach Lecture, to Sergio Acosta, Carl Blyth, Pascal Denis, Laurie Modrey, Robert Reichle, Ellenor Shoemaker, Alexandra Teodorescu, Alfonso Abad-Mancheño, Steve Byrd, Tim Ditoro, and Chris Bryce for their help in the organizational and editorial assistance, and to Anke de Looper for her assistance in editing this volume. Finally, neither the LSRL conference nor this volume would have been possible without the financial support of the College of Liberal Arts, the Department of French and Italian, and the Department of Spanish and Portuguese. Chiyo Nishida and Jean-Pierre Montreuil May 2006

INTRODUCTION

CHIYO NISHIDA

This volume includes papers selected from both the general session and the parasession on historical linguistics and dialectology. It covers topics in morphology, syntax, semantics and pragmatics. The studies represented here reflect the remarkably diverse perspectives one can take in researching Romance languages. Below we have divided the papers according to traditional fields; however, the scope of investigation in many papers is not strictly confined to one particular module. Some are also concerned with the relation among components; others deal with their topics in the context of language acquisition and bilingualism as well as from historical perspectives. Despite variations in approach, each paper sheds new light on the problem it tackles and uncovers something amazing about human language in general. Topics in Morphology. BOK-BENNEMA & KAMPERS-MANHE propose an analysis of Romance VN compounds as nominalizations of vPs with a pro subject, which later moves to the specifier position of the nominalizer. Applying this analysis to analogous English synthetic compounds, the authors argue for an incorporation strategy inherent to the nature of the English nominalizer. Through comparative analysis of VN compounds in Romance languages and English, they show that morphology is highly parallel to syntax, but still has its own ontological restrictions. FERRARI-BRIDGERS proposes a syntactic analysis of Italian deverbal nouns that are formed by merging a nominal feature [n] with either a verbal stem, an AspectP or a VoiceP. By examining the semantic and morphological properties of de-verbal nouns, the author identifies a large array of functional projections involved in the derivation of these nouns. She concludes that semantic and morphological variations across de-verbal noun types can be reduced to syntactically defined structural differences. GRAČANIN-YUKSEK examines the case of ‘gender mismatch’ observed in Italian between the morphological shape of the V-N compound and its syntactic agreement, as in dei poggia-teste “some headrests”. The

x

INTRODUCTION

author argues that the structure assigned to the compound and the featuresharing operation is responsible for this gender mismatch and shows that the model of morphology developed in this paper correctly derives the plural forms of both invariant and inflected V-N compounds in Italian. Throughout this study, she postulates that morphology and syntax are the same with respect to the processes they employ in deriving legitimate structures. MCCARTHY contends that second language morphological errors are constrained by features available in the UG and that the ‘underspecification theory’ explains why certain features are available and others are not. According to this theory, only underspecified morphemes may act as defaults. Using the data collected on verbal morphology (person/number and tense) in a spontaneous production of L2 Spanish the author shows that the majority of errors committed by L2 learners are due to underspecification and not to feature clash. This paper adds a new perspective to the on-going discussion on the second language acquisition of morphosyntactic features. PIRVULESCU investigates why the agreement paradigm in the subjunctive and imperative moods seems to show the same agreement affixes as those of the paradigm in the indicative present across Romance languages. Adopting the framework of distributed morphology, the author argues that such a similarity is a consequence of the syntactic status of TENSE in the subjunctive and imperative moods as well as in the indicative present; in all three paradigms TENSE is underspecified. Pirvulescu observes that this generalization seems to be supported from examples in unrelated languages outside Romance languages. Topics in Syntax. BOUCHARD compares two views on what is unique to human language—the generative mode proposed by Hauser, Chomsky & Fitch (2002) and the ‘exaptive’ model proposed by the author himself. Observing that humans apply recursion to signs, Bouchard hypothesizes that the distinctive property of human language is not recursion, as claimed by the proponents of the generative model, but is rather the signs that link two kinds of substance, perceptual and conceptual. He offers several arguments in support of his model. First, this model provides a basis to explain the presence of recursion in human language. Second, it explains why several novel traits that seem to be unconnected in a generative model appeared simultaneously in human language. Finally, the model leads to insightful reanalyses of five well-studied syntactic phenomena. Introducing a new set of data, CAMACHO maintains that preverbal subjects in Spanish can occupy a number of positions, some left-peripheral (lexical subjects) and some within IP (pro, expletives) and that the L or non-Lrelated nature of subject positions is derivationally determined by verb

INTRODUCTION

xi

movement. To account for the distribution of preverbal subjects, he proposes two principles, agreement-active (a given functional projection may be inert for agreement until activated by movement of a head to it or filling of its specifier) and the Phase Consistency Condition (all categories within a phase are consistent with respect to being selected or non-selected). GUTIÉRREZ-BRAVO argues that experiencer dative arguments that surface in the preverbal position in Spanish cannot be analyzed as an underlying or a quirky subject, as has been previously proposed. Instead he postulates the notion of the ‘Pole’ of the clause defined as the XP that occupies the highest inflectional specifier and is occupied by the argument that ranks highest in the ‘Thematic Hierarchy’. The author claims that his analysis accounts for the fact that dative arguments surface in the preverbal position of some Spanish verbs, without compromising the evidence that the nominative argument of these verbs is the true grammatical subject. MACDONALD proposes that non-reflexive, non-argumental dative clitics of Spanish and reflexive non-argumental clitics of Romanian are introduced as the complement of a null directional-like preposition that merges as a complement of the verb. He contends that such an analysis can account for the cluster of properties that constructions containing these two clitics share: 1) the inability to produce a telic interpretation of a predicate; 2) the lack of on/with entailment; 3) the ability to express temporary relations; 4) the ability to prevent idiomatic interpretations; and 5) the resistance to adjectival secondary predication. PÉREZ-LEROUX, PIRVULESCU & ROBERGE present a comparative study of object omission in child French and English. The experimental cross-linguistic comparison of children’s performance shows that while null objects decrease with age for English children, they do not for French children for both the individuated and the non-individuated object. Based on these findings, the authors postulate that while the object omissions for English children might be a sporadic, performance-driven fall back to a default null object, those for French children are due to the overgeneralizaion of a null object as a referentially bound anaphora as well as one with independent reference. TORIBIO & NYE examine the alterations manifested in the grammars of Spanish heritage speakers with respect to their production and interpretation of constructions with the so-called ‘reverse’ psychological predicates, whose experience argument is mapped into the dative. From the experiments conducted with heritage speakers, the authors show that such speakers use two strategies for this class of predicates: 1) mapping the animate argument to the structural subject position and the inanimate argument to the structural object position; 2) placing the constituents in the SVO order.

xii

INTRODUCTION

EDMONDS gives a detailed characterization of the diachronic development of the French indefinite pronoun chacun “each” using an extensive corpus that dates from the beginning of the 12th century through the middle of the 20th century and compares it to an existing analysis of another indefinite pronoun aucun “none, not any”. After concluding that a unified account for both quantifiers is not possible, she proposes that chacun assumes one underlying structure during Old French and Middle French that develops into two separate structures after the 16th century. RUSSI examines the Italian verb volerci “take/be necessary”, which consists of the verb volere “want” and a grammaticalized clitic ci, in an attempt to determine if volerci can be considered a functional element, i.e., a modal head. Based on a detailed comparison of volerci with the Middle and Old Italian bisogna “must/be necessary”, she concludes that volerci is still a lexical verb; ci incorporation may have altered the thematic grid of volere, but the resulting verb volerci still projects arguments. Topics in Semantics. ARANOVICH proposes a formal semantic analysis of phrases coordinated by the Spanish particle ni (ni XPs). It has been observed that ni XPs must be preceded by a negation when appearing after the verb. Postulating that the ‘inclusion lattice’ proposed for polarity-sensitive items can also be associated with conjunctive coordinations with the conjunctions at the top and disjunctions at the bottom, the author argues that ni XPs are negative polarity disjunctions. He claims that due to the fact that they are the bottom members of an inclusion lattice, disjunctions need to appear in the scope of negation; they can satisfy both their semantic (widening) and pragmatic (strengthening) requirements only in such environments. DEPREZ investigates the role of number in French, English, and Haitian Creole (HC) and argues that languages can differ in the way they individuate kind terms. The author proposes that individuation is done through number in languages like French and English, whereas individuation takes a route that she terms as ‘spatial individuation’ in languages like HC. In these languages, number is used when an entity must be re-identified in a changing context. Deprez’s proposal has empirical consequences for NP syntax and the licensing of bare argument nominals (BAN) in two classes of languages. MARI & TOVENA argue that the Italian neppure contributes in two ways to the interpretation of the clause hosting it. Neppure adds negative information (additive) and poses a strong constraint on scales (scalar). The authors then propose a unified account of these two uses in which they make two claims. First, in both uses, neppure focuses on an element that is understood as a member of a class/series and adds negative information. Second, in both interpretations, neppure sets a temporary closure in the

INTRODUCTION

xiii

speaker’s state of knowledge. The authors conclude that the difference between additive and scalar neppure lies in the way the class/series is built. Topics in Pragmatics. LAMBRECHT, BORDEAUX & REICHLE present a corpus-based analysis of the conversational use of the French conjunction parce que “because”. They propose that in p parce que q, parce que q is either the focus of a larger proposition in which p is presupposed or it expresses an independent assertion due to the effect of a universal cognitive constraint against focus projection from adjunct position. This study contributes to the growing body of research on the difficulties associated with the traditional distinctions between subordination and coordination. LE DRAOULEC contends that temporal conjunction avant que or avant de “before” not only introduces a presuppositional clause, but also has the potential of introducing a non-presuppositional clause, as in the similar case of the well-known ‘quand (when)-inverse’. Using written data, the author shows that, as with quand inverse, a necessary condition is that the avant queor avant de-clause is postposed; i.e., the respective position of the main and subordinate clauses matches the temporal order in which the eventualities take place. As a final note, we hope that this volume will be viewed as a significant contribution to the field of Romance Linguistics and will inspire Romance researchers to pursue these promising lines of investigation.

.

A POLARITY-SENSITIVE DISJUNCTION SPANISH NI ... NI

RAÚL ARANOVICH University of California Davis 0.

Introduction Spanish has a coordinating particle, ni, which is subject to a polarity requirement. As argued in Bosque (1980, 1994) and Herburger (1999a), phrases coordinated with ni (ni-XPs, from now on) must appear in the scope of a negative operator. For instance, ni-XPs must be preceded by negation if they are in postverbal position (1).1 (1)

Las cartas *(no) llegarán ni por avión ni por barco The letters not will.arrive ni by plane ni by ship. “The letters will neither arrive by plane nor by ship.”

In this paper I argue that ni-XPs are subject to a polarity requirement because they are disjunctions, which can only satisfy a requirement of pragmatic strengthening in the scope of negation. My argument assumes the widening/strengthening (W/S) account of polarity sensitivity developed in Heim (1984), Kadmon and Landman (1993), Israel (1996), and Lahiri (1998), among others. The W/S account states that polarity-sensitive items are subject to a couple of seemingly incompatible constraints. From a semantic point of view, NPIs have a wider denotation than a set of alternatives. From a pragmatic point of view, however, NPIs must strengthen the statement they occur in. These two requirements can only be satisfied in contexts like the scope of negation, which reverse the normal direction of entailments. A key hypothesis in my analysis is that ni-XPs are disjunctions, a claim that I defend 1

The following abbreviation is used in the glosses: RECP (reciprocal). Since negative expressions like nada, can be translated as either “nothing” or “anything”, it will be glossed as “n-thing” (similarly for nadie “anybody, nobody” or ninguno “any N, no N”. Occurrences of ni are glossed as ni.

2

RAÚL ARANOVICH

based on syntactic and semantic properties of ni-XPs. Disjunctive coordinations, I suggest, are the weakest members of a set of expressions that includes conjunctive coordinations and the independent coordinated phrases (forming an inclusion lattice, in terms of Krifka 1992). Their disjunctive nature makes Ni-XPs semantically wide (being the bottom element of a polarity lattice). In accordance with the W/S account, then, ni-XPs are limited to appearing in the scope of negation because only there can they satisfy a pragmatic strengthening requirement. The paper is structured as follows: Section 2 shows that ni-XPs are in fact polarity-sensitive items, and discusses other details of the grammar of niXPs against the background of negative concord. Section 3 gives evidence for the hypothesis that coordinated constructions with the particle ni are disjunctions, and not conjunctions. In Section 4 I summarize the widening/strengthening account of polarity sensitivity, and I apply this analysis to explain the distribution of ni-XPs. Conclusions are presented in Section 6. 1.

Negative concord and ni-XPs Spanish is a 'negative concord' language. When expressions like nada “nothing”, nadie “no one”, and ningún N “no N” (often referred to in the literature as n-words) precede the verb, they count as an expression of negation (2). But when they follow the verb, they must occur with the negative adverb no “not” or with another n-word in preverbal position. When this occurs, as in (4), the clause is interpreted as having a single instance of sentential negation, and not as a statement with multiple negations. (2)

Nadie ha leído este poema N-body has read this poem “Nobody has read this poem.”

(3)

López *(no) sabe nada López (not) knows n-thing “López (doesn't) know(s) anything.”

(4)

Nadie ha leído ninguno de estos poemas N-body has read n-one of these poems “Nobody has read any of these poems.”

Bosque (1980) and Laka (1990) make the influential suggestion that Spanish n-words are ‘negative polarity items’, i.e. expressions whose

SPANISH NI ... NI

3

distribution is conditioned by the polarity of their context of occurrence.2 They reach this conclusion on account of the limited distribution of n-words. Across languages, it is not uncommon to find lexical items that must be licensed by negation or other ‘affective’ expressions (the term that Klima 1964 uses to describe the phenomenon). An English NPI like anything, for example, is excluded from simple statements (5), but it occurs in the scope of negation (6), in the restriction of universal quantifiers (7), and in a variety of other environments. (5)

*Sandy ate anything.

(6)

Sandy did not eat anything.

(7)

Every student who had ever read anything on phrenology attended the lectures. (Ladusaw 1979:149)

According to Bosque (1980) and Laka (1990), the limited distribution of n-words in Spanish shows that they are NPIs as well. Following the same line of reasoning, ni-XPs are NPIs too. As shown before, in example (1), niXPs must be preceded by a negative expression if they are in postverbal position. This negative expression can also be an n-word. (8)

Ninguna carta llegará ni por avión ni por barco n-one letter will.arrive ni by plane ni by ship “No letters will arrive by plane or by ship.”

In preverbal position, ni-XPs count as the sole expression of negation (9). I follow the analysis in Giannakidou (1998, 2000), according to which preverbal n-words are licensed by a negative adverb that deletes when preceded by an nword in some negative concord languages. Ni-XPs pattern with other n-words in this respect as well.

2

This is by no means the only position found in the literature about n-words in Spanish, or in the Romance languages in general. Vallduví (1994) and Aranovich (1996) question claims about the parallel distribution of n-words and English NPI's, while Herburger 1999b, following work by Zanuttini (1987, 1991), argues that n-words are ambiguous between a universal and an existential interpretation.

4 (9)

RAÚL ARANOVICH

Ni Fulano ni Mengano (*no) salieron a caminar Ni Fulano ni Mengano (not) went.out to walk “Neither Fulano nor Mengano went on a walk.”

An intriguing fact about ni-XPs, mentioned in Bosque (1994), is that the doubling of the conjunction is mandatory in preverbal position (10), but optional in postverbal position. He argues that the initial ni is necessary to identify a ni-XP as expressing negation. (10) *(Ni) Fulano ni Mengano salieron a caminar. Ni Fulano ni Mengano went.out to walk “Neither Fulano nor Mengano went on a walk.” There is another function of ni, not as a coordinating particle but as a focus particle. These phrases with focus ni are minimizers, which can sometimes be translated as “not even NP”. (11) Ni Gómez asistió a la reunión Ni Gómez attended to the meeting “Not even Gómez attended the meeting.” Like ni-XPs, focus ni is a kind of NPI. In preverbal position a ni-minimizer expresses negation by itself, without any additional negative expressions. In postverbal position it requires a preverbal negative expression, and when the two co-occur they do not give rise to multiple negative readings in the clause. The relationship between the use of ni as a conjunction, and its use as a focus particle, raises interesting questions which I am unable to explore in detail due to space limitations. In this paper, then, I will exclude consideration of focus ni. 2.

The nature of ni as a coordinating particle A fundamental hypothesis in my analysis of the distribution of ni-XPs is that the coordinating particle ni is a disjunction, and not a conjunction.3 In this section I develop three arguments in favor of this hypothesis. They are based on evidence provided by the doubling of coordinating particles in Spanish, on the ability of coordinations to occur as subjects of reciprocal

3

De Swart (2001) reaches a similar conclusion regarding the disjunctive sense of the French counterpart of Spanish ni. Her arguments are based on interpretations of the relative scope of the negative operator with respect to the coordinating particle ni.

SPANISH NI ... NI

5

predicates, and on the possibility of assigning a temporal interpretation to some coordinated structures. The first piece of evidence supporting the hypothesis that ni-XPs are disjunctive in meaning is that, like the disjunction o “or” in (12), ni can occur multiple times in a coordinated XP, as in examples (1) and (9). (12) Las cartas llegarán o por avión o por barco The letters will.arrive or by plane or by ship. “The letters will either arrive by plane or by ship.” Doubling of conjunctions is possible in some languages, as the Polish example below (from Borsley 2005) shows. French conjunctions can also be doubled (Piot 2000), and so can Latin and Modern Greek ones (Johannessen 1998). In Spanish, however, the conjunction y “and” cannot be doubled, as shown in (14). (13) i Jan i Maria and Jan and Maria “both Jan and Maria.” (14) Las cartas llegarán (*y) por avión y por barco The letters will.arrive and by plane and by ship. “The letters will arrive by plane and by ship.” The multiple occurrence of a coordinating particle in Spanish is a characteristic feature of disjunctions, and not of conjunctions. The fact that ni can be doubled, then, is evidence that ni-XPs are disjunctive.4 A second piece of evidence for the disjunction hypothesis is found in the interaction between ni-XPs and reciprocal expressions. A reciprocal pronoun can have an XP coordinated by y 'and' as its antecedent (15), but not one coordinated by o “or” (16). 4

Bosque (1994) assigns the structure of a Conjunction Phrase (CoP) to ni-XPs. In this analysis, the coordinating particle ni is the head of a phrase that has one of the coordinated terms (the first one) as its specifier, and the other term as its complement. As noticed in Borsley (2005), however, the doubling of a coordinating particle, as in examples like (1) or (12), has been long recognized as a problem for the CoP analysis. Johannessen (1998) suggests that the initial particles in cases of doubling should be treated as adverbs, similar to English either or both, thus salvaging the status of the second particle as the head of a construction. The issue of finding the right syntactic representation for coordinated phrases, with or without a double coordinating particle, is beyond the scope of this paper, and does not bear on the argument that doubling is a characteristic feature of conjunctions in Spanish.

6

RAÚL ARANOVICH

(15) Fulano y Mengano se encontraron en la esquina Fulano and Mengano RECP met in the corner “Fulano and Mengano met each other at the corner.” (16) *Fulano o Mengano se encontraron en la esquina Fulano or Mengano RECP met in the corner ("Fulano or Mengano met each other at the corner.") The explanation for the contrast between (15) and (16) is simple: the reciprocal predicate requires a subject that is semantically plural, and this interpretation can only be assigned to conjunctive coordinations, not to disjunctive coordinations. Ni behaves like a disjunction with respect to reciprocals as well, as in (17). If ni-XPs were conjunctions, sentence (17) should be grammatical as the negation of (15).5 (17) *Ni Fulano ni Mengano se encontraron en la esquina Ni Fulano ni Mengano RECP met in the corner (“Neither Fulano nor Mengano met each other at the corner.”) A third piece of evidence for the hypothesis that ni-XPs are disjunctions is the absence of a temporal reading for this kind of coordination. Conjunctions have a temporal interpretation, in which the first conjunct precedes the second one in time. Thus, a sentence like (18) means that the action of opening the door precedes the action of leaving. Reversing the order of the terms results in a nonsensical statement. (18) Fulano abrió la puerta y salió Fulano opened the door and got.out “Fulano opened the door and got out..” The terms of a disjunction are not subject to the same temporal constraints as a conjunction. Thus, there is no common sense incongruence on a sentence like (19). Phrases coordinated with ni pattern with disjunctions in this respect as well, as shown in (20).

5

Moreover, reciprocal predicates cannot take arguments that receive a distributive interpretation. Ni-XPs and disjunctions share this semantic property, while conjunctions do not (I am indebted to a reviewer for bringing up this point).

SPANISH NI ... NI

7

(19) Fulano salió o abrió la puerta Fulano got.out or opened the door “Fulano either got out or opened the door.” (20) Fulano ni salió ni abrió la puerta Fulano ni got.out ni opened the door “Fulano neither got out nor opened the door.” Having shown that ni-XPs are disjunctions, I will argue that their limited distribution is the result of a potential conflict between their semantics and a pragmatic requirement that statements containing ni-XPs be more informative than contextually salient alternatives. In negative contexts the pragmatic requirement can be satisfied. 3.

The W/S account of negative polarity The semantic theory of polarity sensitivity I assume in this work has its roots in Ladusaw's (1979) semantic account of the distribution of NPIs like English anything. Ladusaw argues that the operators that licenses the NPI (also referred to as the ‘trigger’) are downward-entailing (DE), defined below in (21). Negation is downward-entailing, as the formula in (22) shows. (21) Downward-entailing operator: An operator Op is downward-entailing iff p ⊃ q = Op(q) ⊃ Op(p). (22) (P → Q) ≡ (¬Q → ¬P) Ladusaw's semantic approach to negative polarity has inspired a fruitful line of research, which also extends to negation in the Romance languages and other negative concord languages. Many have noticed that the class of operators that license n-words in negative concord languages tend to be a sub-set of the downward-entailing ones (Ladusaw 1992, De Swart 2001). According to Giannakidou (1998), the triggers that license NPIs in negative concord languages like Spanish and Greek (on which most of her research is focused) share the semantic property of being ‘nonveridical’, a semantic concept introduced in Zwarts' (1995) analysis of Dutch and English polarity, and defined in (23). Given the scope of this paper, however, the distinction between DE and nonveridical operators will be glossed over. (23) Nonveridical operator: An operator Op is nonveridical iff Op(p) ⊃ p is not a logically valid statement.

8

RAÚL ARANOVICH

Kadmon and Landman (1993) suggest that NPIs are restricted to appear in the scope of DE operators because the entailment reversal that is verified in these contexts allows the NPIs to satisfy their contradictory semantic and pragmatic requirements. The denotation of English NPIs like lift a finger or anything, they argue, must include even the most marginal cases (what they call ‘semantic widening’), but at the same time the clause with the NPI must provide more information than the corresponding clause without it (what they call ‘pragmatic strengthening’). In upward-entailing contexts (i.e. contexts that preserve entailments), widening clashes with strengthening, since widening the the denotation of a term results in a statement of lesser informative value. It is only in downward-entailing contexts (i.e. contexts satisfying the biconditional in (22)) where by widening the denotation of a term a sentence is strengthened. NPIs are restricted to appearing in the scope of negation (and other downwardentailing operators) because those are the only contexts where their strengthening requirement can be satisfied, given their widening semantic value. Explanations for polarity sensitivity similar to Kadmon and Landman’s are also presented in works dealing with other languages and other polarity items (Heim 1984, Lahiri 1998). An analysis that shares some of the insights of Kadmon and Landman’s approach is developed in Krifka (1992). In his analysis, all NPIs are the minimal element of a ‘polarity lattice’, which can be informally characterized as an ordering among denotations of the same sort as the NPI. The occurrence of an NPI in a statement must make that statement more informative than alternatives resulting from a substitution of the NPI by other members of the lattice. Since the NPI is the bottom element of the lattice, this requirement (analogous to the strengthening requirement of Kadmon and Landman) can only be satisfied in DE contexts. Krifka’s concern is the distinction between any-NPIs and polaritysensitive minimizers like lift a finger. The polarity lattice for minimizer NPIs is ordered according to a pragmatic scale. An NPI like lift a finger, for instance, is the lower limit of a lattice built on the basis of the idiosyncratic scalar property of physical effort or labor. Only in downward-entailing environments can an NPI be more informative than other members of the associated lattice. For the interpretation of NPIs like anything, however, no idiosyncratic pragmatic scale seems to be necessary or available. The polarity lattices associated with non-idiosyncratic NPIs are built on the more general property of ‘inclusion’. In Krifka’s analysis, NPIs like anything are existential quantifiers, denoting the set of properties that some thing has. This set is maximally inclusive. Suppose a domain D where there is something red and something blue. Then blue and red are in the denotation of something (or

SPANISH NI ... NI

9

anything, for that matter). These properties, on the other hand, do not belong to the denotation of everything, since some things are not red (or blue). I will argue that this is the relevant semantic property to understand the polarity requirement on ni-XPs. To find the associated polarity lattice for niXPs, it is necessary to go back to the claim that ni-XPs are disjunctions. I interpret disjunction as the set-theoretic operation of union (∪), and conjunction as intersection (∩). The union of two sets A, B includes each set, and each set includes their intersection, as in Figure 1. Or, in intensional terms, the set of properties that A or B have includes the set of properties that A (or B) has, which in turn includes the set of properties that A and B have. The conclusion is that complex XPs coordinated with disjunctions are also the bottom element of a polarity lattice based on inclusion.6 !$" !

" includes !#"

Figure 1: Inclusion Lattice

By associating ni-XPs with the bottom element of an inclusion lattice, their polarity-sensitive nature can be explained. A sentence with a disjunction will be less informative than all the alternative propositions associated with the other terms in the inclusion lattice, as in (24). That is, a sentence like Max saw Sandy entails (and therefore is more informative than) Max saw Sandy or Pat. (24) P(a) → P(a∨b) I assume that ni-XPs have a strengthening requirement that governs their use in any sentence. That is, the sentence with the ni-XP must be more informative than the corresponding sentence with an alternative member of the lattice in place of the ni-XP, be it one of the coordinated terms or their conjunction. But since disjunctions include the other terms in their associated polarity lattice, the strengthening requirement on a disjunction can only be satisfied in the 6

In this reasoning I am also relying on the well-known equivalence between disjunction and existential quantification. If NPIs like anything, which are at the bottom of an inclusion lattice in Krifka (1992), are indeed existential quantifiers, then disjunctions are also at the bottom of a polarity lattice, since an existentially quantified proposition (∃x)(Px) is equivalent to a disjunction (Pa) ∨ (Pb) ∨ ... ∨ (Pn) of all atomic propositions with property P that are true in the domain Dn.

10

RAÚL ARANOVICH

scope of downward-entailing operators like negation, as shown in (25). In a simple affirmative statement the strengthening requirement of ni-XPs cannot be satisfied. (25)

¬P(a∨b) → ¬P(a)

4.

Conclusions My analysis of the distribution of ni-XPs is based on a simple insight: polarity lattices based on inclusion can be associated with coordinated XPs, so that disjunctions are at the bottom and conjunctions at the top. When this insight is paired up with the W/S account of polarity sensitivity, a prediction is made: there should be a class of negative polarity disjunctions, which can only satisfy their strengthening requirement in downward-entailing environments. In this paper I have argued that this is exactly what ni-XPs are. The polaritysensitive nature of ni-XPs in Spanish, then, offers an argument in favor of the semantic approach to polarity sensitivity developed by Krifka (1992), Kadmon and Landman (1993), and others. References Aranovich, Raúl. 1996. Negation, Polarity, and Indefiniteness: A Comparative Analysis of English and Spanish. Doctoral dissertation, University of California San Diego. Borsley, Robert. 2005. "Against ConjP". Lingua 115, 461-482. Bosque, Ignacio. 1980. Sobre la negación. Madrid: Cátedra. ----------. 1994. “La negación y el principio de las categorías vacías”. Gramática del español, ed. by Violeta Demonte, 167-199. México: El colegio de México. De Swart, Henriëtte. 2001. “Négation et coordination: la conjonction ni”. Adverbial Modification, ed. by Reineke Bok-Bennema, Bob de Jonge, Brigitte Kampers-Manhe, and Arie Molendijk, 109-124. Amsterdam: Rodopi. Giannakidou, Anastasia. 1998. Polarity Sensitivity as (Non)veridical Dependency. Amsterdam: John Benjamins. ----------. 2000. “Negative... Concord?” Natural Language and Linguistic Theory 18.457-523. Heim, Irene. 1984. “A Note on Negative Polarity and Downward Entailingness”. Proceedings of NELS 14, ed. by Charles Jones and Peter Sells, 98-107. Amherst: University of Massachusetts.

SPANISH NI ... NI

11

Herburger, Elena. 1999a. “Lexical Ambiguity is not always Evil: The Example of ni-ni”. Advances in Hispanic Linguistics, ed. by Javier GutiérrezRexach and Fernando Martínez-Gil, 378-393. Sommerville: Cascadilla Press. ----------. 1999b. “On the Interpretation of Spanish N-Words”. Semantic Issues in Romance Syntax, ed. by Esthela Treviño and José Lema, 89-104. Amsterdam: John Benjamins. Israel, Michael. 1996. “Polarity Sensitivity as Lexical Semantics”. Linguistics and Philosophy 19.619-666. Johannessen, Janne Bondi, 1998. Coordination. Oxford: Oxford University Press. Kadmon, Nirit. & Fred Landman. 1993. “Any”. Linguistics and Philosophy 16. 353-422. Klima, Edward. 1964. “Negation in English”. The Structure of Language, ed. by Jerry A. Fodor and Jerrold J. Katz, 246-323. Englewood Cliffs: Prentice-Hall. Krifka, Manfred. 1992. “Some Remarks on Polarity Items”. Semantic Universals and Universal Semantics, ed. by Dietmar Zaefferer, 150-189. Dordrecht: Foris. Ladusaw, William. 1979. Polarity Sensitivity as Inherent Scope Relations, Doctoral dissertation, University of Texas at Austin. ----------. 1992. “Expressing Negation”. Proceedings of the Second Conference on Semantics and Linguistic Theory: Ohio State University Working Papers in Linguistics 40, ed. by Chris Baker and David Dowty. Columbus: The Ohio State University, 237-259. Lahiri, Utpal. 1998. “Focus and Negative Polarity in Hindi”. Natural Language Semantics 6.57-123. Laka, M. Itziar. 1990. Negation in Syntax: On the Nature of Functional Categories and Projections. Doctoral dissertation, MIT. Piot, Mireille, 2000. “Les conjonctions doubles”. Lingvisticae Investigationes 23.45-76. Vallduví, Enric. 1994. “Polarity Items, N-words, and Minimizers in Catalan and Spanish”. Probus 6.263-294. Zanuttini, Raffaella. 1987. “Negation and Negative Concord in Italian and Piedmontese”. Canadian Journal of Italian Studies 10.135-149. ----------. 1991. Syntactic Properties of Sentential Negation. Doctoral dissertation, University of Pennsylvania. Zwarts, Frans. 1995. “Nonveridical Contexts”. Linguistic Analysis 25.286-312.

TAKING A CLOSER LOOK AT ROMANCE VN COMPOUNDS

REINEKE BOK-BENNEMA BRIGITTE KAMPERS-MANHE Groningen University/CLCG

0.

Introduction This paper is the partial result of a substantial project on Romance compounding. Applying the rules and principles of sentential syntax to word syntax, we intend to account for the similarities, but most importantly for the differences, between Romance and West Germanic compounds. In this article we concentrate on Romance compounds of the type given in (1), which are usually called exocentric VN compounds: (1) essuie-glace wipe windshield “windshield wiper” These compounds, illustrated in (2)-(4) for French, Italian and Spanish respectively, are the result of a very productive process. (2) a. protège-slip protect pants “panty liner”, coupe-faim cut hunger “diet biscuit”, sèche-linge dry wash “dryer”, essuie-tout wipe all “paper towel”, cache-boutons hide pimples “concealer stick”, porte-tampons hold tampons “tampon holder”, vide-grenier empty loft “jumble fair” b. porte-parole carry word “spokesman”, casse-pieds break feet “bore”, fouille-merde search shit “journalist”, trouble-fête trouble party “spoilsport”, lèche-botte sucks boot “bootlicker” (3) a. pulisciscarpe clean shoe “shoe polish”, portaspilli hold needles “pincushion”, accendisigari light cigars “cigar lighter”, tagliacarte cut letter “letter opener”, sparaneve shoot snow “snow cannon”, acchiapavirus catch virus “antivirus program”

14

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

b. mangiamaccheroni eat macaroni “good-for-nothing”, mangialeggi eat laws “law eater” (Berlusconi), guardaboschi watch woods “forester”, guardacoste watch coasts “coastguard”, portalettere bring letters “postman” (4) a. buscapersonas look+for persons “pager”, tragaperras swallow coins “gaming machine”, rompecabezas break heads “puzzle”, exprimelimones squeeze lemons “lemon-squeezer”, rascacielos scrape skies “skyscraper”, llenacines fill movie-theaters “popular movie” b. sacamuelas pull molars “(bad) dentist”, chupasangre suck blood “blood-sucker”, limpiabotas clean boots “shoeblack”, portavoz carry voice “spokesman”, guardaespaldas watch backs “bodyguard” These lists contain recent words like cache-boutons in (2a), manggialeggi in (3b) and acchiapavirus in (4a). All these compounds consist of a verb and noun. The relation between the verb and the noun is the same as the relation between a transitive verb and its complement in a syntactic Verbal Phrase: the noun is the theme of the verb. As for the interpretation of the compounds, they all refer either to a device (cf. (2a), (3a) and (4a)) or to a person, as shown in (2b), (3b) and (4b), that acts as the instrument or the agent of the action the verb refers to: for instance, a protège-slip is a device that protects pants, a mangiamaccheroni is a person who eats macaroni. Traditionally Romance VN compounds are considered problematic because they are apparently headless. They have generally been analyzed as nominalizations of Verbal Phrases consisting of a verb and a nominal complement. We will provide a new analysis for these compounds. We will start by considering the status of the verbal projection involved in their formation in Section 2. We will subsequently develop our analysis in Section 3 and compare the structure of Romance compounds to that of their English counterparts like screensaver, that is to say to that of English synthetic compounds (Section 4). We will show that the proposed analysis can account for the differences between the two types of languages and, at the same time, does justice to the properties they share. We finally consider the implications of the proposal for the relation between syntax and morphology in Section 5. 1. 1.1.

VP or vP? Introduction As we have just mentioned, Romance VN compounds have often been analyzed as nominalizations of VP. Di Sciullo and Williams (1987), for instance, provide the word-creating rule in (5):

TAKING A CLOSER LOOK AT ROMANCE COMPOUNDS

(5) N -----> VP

15

Di Sciullo & Williams, 1987: 81

The VP hypothesis can also be found in Rohrer (1977), Zwanenburg (1990), Lieber (1992), Ferrari-Bridgers (2003), a.o. The differences between the authors do not concern the status of VP, but the way the nominalization is achieved. However, within the actual generative framework we are employing here, in which VP is the inner shell of a bigger vP (Chomsky 1995), it seems natural to wonder whether the nominalization involves a VP or a more complex structure, namely that of a vP. Several arguments lead to the conclusion that it is a vP. 1.2.

Arguments in favor of vP The verb occurring in Romance VN compounds is a transitive/causative verb. Under the current assumption that the causative element is hosted in small v, the null hypothesis is that small v is indeed involved. This can also account for the fact that ergative verbs are excluded from these compounds, as shown by the ungrammaticality of (6) in French: (6) *tombe-pierres fall stones *“stonefaller” The second argument comes from the Spanish and Italian words given in (7a) and (7b) respectively: (7) a.

b.

cuentagotas count drops “pipette” muoviculo move bottom “foreman” (pejorative)

The verbal stem vowel of cuenta and muovi has undergone diphthongization, a process that requires that the stem vowel — in addition to being lexically specified as diphthongizing — bears primary stress, as illustrated in (8a) and (8b) for the Spanish case: (8) a.

Esta cosa cuénta gótas “This thing counts drops.”

16

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

b.

Esta cosa sirve para contár gótas “This thing serves to counts drops.”

In both cases the normal stress-assignment rules (see e.g. Harris 1983) have applied. The result is that in (8a), the stem vowel is stressed and therefore diphthongized. No diphthongization occurs in (8b) however because the theme vowel of the infinitive bears stress. To account for the diphthongization in the compounds cuentagotas and muoviculo in (7), we have to assume that, although the stem vowel does not bear primary stress, it must have done so at some point of the derivation and, therefore, that Spell-Out has occurred in an earlier phase. Given the assumption that vP is a cyclic phase, adopting a vP-structure for the compound enables us to account for the facts. The derivation is that of (9): conta adjoins to v, vP is spelt out, the stem vowel of conta gets primary stress, the same way as in (8a), and diphthongizes. (9) [vP conta [VP conta gotas ]] ---> Spell-Out ---> [cuentagotas] After nominalization, the verb, as an element of the edge of vP, remains accessible and can be de-stressed in the next phase, while primary stress is assigned to gótas. The third argument concerns the contrast between Romance and English compounds that is illustrated in (10): the nominal complement screen cannot be pluralized in English, whereas the plural form of the complement can occur in Romance VN compounds: (10) a. b.

salvapantallas “screensaver” (Spanish), porte-jarretelles “suspenders” (French), manggialeggi eat laws (Italian) screensaver/*screenssavers

It has often been proposed in the literature (cf. Roeper & Siegel 1978, Lieber 1992, Kayne 1994, Sportiche 1999, a.o.) that the complement in English compounds of that type is licensed by incorporation. Adopting this idea enables us to account for the fact that it cannot be plural: it is just a stem (cf. Baker, 1988). As for the Romance complements, they have to be licensed in another way. Our claim is that they are licensed by Case-checking. Given the fact that v is the probe for object Case, it must be present in the structure of the compound. Licensing can take place as in (11): V adjoins to v, which activates Case-checking, so that the complement is licensed.

TAKING A CLOSER LOOK AT ROMANCE COMPOUNDS

17

(11) [vP [ essuie+v] [VP essuie glace ]] |____________| CASE-CHECKING

Finally, the vP analysis enables us to account for the semantics of the compounds, as we will show in the next section. 2.

Extension of the analysis VN compounds involve an agent or an instrumental role: an essuieglace is a device that wipes windshields, a sacamuelas is a person who pulls molars. The idea we adopt is that the agent/instrumental role is structurally assigned to the specifier position of small vP. This implies that this position must be filled by an empty DP that can bear this theta role. Two possibilities come to mind: either PRO or pro is present. This nominal element does not refer to an arbitrary person, nor is it co-referential with another DP, contrary to PRO. This leads us to the proposal that the agent theta role is assigned to a pro present is the specifier position of vP. This pro is comparable to the one that occurs in Romance DPs of the type illustrated in (12), according to Sleeman (1996): (12) a.

b c

los pro que están ahí the that are there “the ones over there” Je veux le pro rouge. “I want the red one.” il pro più attento della classe “the most attentive one of the class”

(Spanish)

(French) (Italian)

The structure of the vP is then that of (13): (13) [vP pro [v’ essuie [VP essuie glace ]]] For the next step of the derivation, the nominalization, a nominal head is needed. This head obviously lacks phonetic content, but possesses a phifeature for gender. Following GraČanin-Yuksek (2005), who refers to Riente (2003), we propose that this gender feature is [natural], which is realized as [+masculine] for nouns referring to male humans, and [-masculine] if they refer to female humans. The unmarked [+masculine] is also used for devices, which lack a natural gender. This is illustrated in (14), where the article is masculine in all three Romance compounds referring to

18

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

devices and masculine or feminine depending on the natural gender of the referent of the compound: (14) a. b.

un essuie-glace, el salvapantallas, il tagliacarte un fouille-merde, une fouille-merde

Our claim is that the nominal head under consideration is a noun, and not an affixal head, contrary to Rohrer (1977) and Lieber (1992), among others. As we will show in Section 3, this enables us to account for some crucial differences between Romance VN compounds and English synthetic compounds. After the merging of vP with the nominal head, the structure is as in (15): 1 (15) [NP N0 [vP pro [v’ essuie [VP essuie glace]]]] This structure consists of a nominalized vP. However it is not sufficient with respect to the semantics of the compounds. In fact, the interpretation of (15) will be something like “the wiping of windshields by something (or someone)”, instead of “something that wipes windshields”. Under the same analysis Spanish sacamuelas, for instance, would mean “the pulling of molars by someone (or something)”, instead of “someone who pulls molars”. This means that we have to extend our analysis in such a way that we do obtain the required interpretation. In general terms, the interpretation of the compounds is the one presented in (16). (16) x such that [x [V N]] This is essentially the interpretation of a relative clause. In our opinion this implies that the compounds as a whole have a relative clause-like structure. We can derive such a structure from (15) under the assumption that relativization in general involves the raising of the so-called ‘antecedent’ from an A-position to an A'-position. This view on relativization is the one proposed in Kayne (1994) and earlier in Vergnaud (1974). It is illustrated in (17) for Spanish los que están ahí (example (12a)), using Kayne’s implementation, which differs slightly from that of Vergnaud. 1

An anonymous reviewer remarks that there is no independent evidence from sentential syntax that N0 can take vP as a complement. This is true, and it is one of the reasons why we think the morphological component cannot be equated to the syntactic one (see Section 5).

TAKING A CLOSER LOOK AT ROMANCE COMPOUNDS

19

(17) los [CP pro [C’ que [IP pro están ahí]]] In (17) the pro subject has raised from its A-position within IP to become the A'-specifier of the complementizer que, so that the relative interpretation is obtained. The relevant aspects of the analysis we have in mind for Romance VN compounds, which we present in (18) for essuie-glace, are very similar to those in (17). (18) [NP pro [N’ N0 [vP pro [v’ essuie [VP essuie glace]]]]] As in (17), the pro subject raises here from an A-position to become a specifier. The differences are that the A-position is the vP internal subject position, and that the specified head is the nominalizer. Notice that we have to assume that the specifier-position of the nominalizer is an A'-position. As far as we can see, there is no counterevidence for this.2 Besides providing us with the correct semantic interpretation, this analysis has another advantage. Given the relation between the nominalizer and pro we can assume that the former acts as licenser of the latter under spechead agreement. This entails that the phi-feature [natural] of the nominalizer is sufficient for the licensing under consideration. We will return to this point in Section 5. The structure in (18) represents our final analysis of Romance VN compounds, the most important characteristics of which are: 1) they consist of transitive nominalized vPs with a pro subject and 2) they involve raising of this subject to the A'-specifier of the nominalizer. Before going on with the next section, we will briefly digress on another type of compound in Romance. We refer to ‘exocentric’ PN compounds, which are exemplified in (19) for French, in (20) for Italian and in (21) for Spanish. (19) sans-papiers without papers “illegal immigrant”, sans-abri without roof, “homeless person”, contrepoison against poison “antidote”, après-soleil, after sun “after-sun product”, hors-série outside series “special edition”

2

Cf. Di Sciullo (1996), who proposes that the English affix –er has an A'-specifier. She also explicitly mentions the parallelism between relative clauses and synthetic compounds with –er. It is worthwhile to remark here as well that in the early stages of Generative Grammar the English forms were sometimes derived from relative clauses (cf. Levi, 1978). We will argue below in the text that the English forms have essentially the same structure as Romance VN compounds.

20

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

(20) lungo fiume along river “road along the river”, dopoguerra after war “period after the Second World War”, dopovoto after vote “period after elections”, fuoribordo outside board “boat with outboard motor”, fuoriborsa outside Stock-Exchange “off the floor transaction” (21) sinvergüenza without shame “shameless person”, simpapeles without papers “illegal immigrant”, entrepiernas between legs “crotch”, contratorpedero against torpedo boat “torpedo boat destroyer”, sobrecama over bed “bed spread” Compounds of this type can quite easily be analyzed along the lines we proposed for VN compounds. They involve PPs instead of VPs, but these PPs arguably serve as predicates for a silent subject, with the interpretation “someone” in for instance sans-papiers, “something” in contrepoison, “some period of time” in dopoguerra etc. Under the assumption that this subject is present as a small pro, we have to do with PP small clauses. As in the case of VN compounds, the status of the compound as a whole is nominal and the interpretation is relative clause-like (a sans-papiers is a “person who is without papers”, a contrepoison is “something that acts against poison”, etc.). These facts are accounted for by assuming that: 1) there is a nominalizer present here as well and 2) the pro subject moves to its A' specifier position.3 In (22) we present the structure of sans-papiers by way of illustration of the analysis discussed in the last paragraph. (22) [NP pro [N’ N0 [PP pro [P’ sans papiers]]]] 3.

English synthetic compounds There is a great deal of literature on English synthetic compounds such as screensaver or truck-driver. For an overview of proposals up until 1988 we refer to Spencer (1991); some newer analyses are presented in Lieber (1992), Sportiche (1999) and Di Sciullo (1996, 2003). As far as we know, there have hardly been any attempts to come up with analyses that generalize over both of

3

PN compounds are less frequent than VN compounds. This is partially due to the fact that the class of prepositions is smaller than that of verbs. But another factor is that Romance has a class of PrefixN forms with the same semantic characteristics as PN compounds. Examples of these are French antirides “anti-wrinkle product”, Spanish proguerra “person in favor of war (the war in Iraq)” and Italian transatlantico “ocean steamer”.

TAKING A CLOSER LOOK AT ROMANCE COMPOUNDS

21

these English forms and Romance VN compounds.4 Some authors even explicitly propose quite different structures for them. Thus, for instance, Lieber (1992) presents the structure in (23a) for French essuie-glace and that in (23b) for English thirst-quencher.5 (23) a b

[N [VP essuie glace] 0/ ] [NP [N’ [N thirsti [N quench er] ti ]]]

No doubt these discrepancies in the analyses for both language types stem from the fact that the forms manifest a number of clearly observable differences. In the first place, whereas in English the N precedes the verb, it follows it in Romance. Secondly, the English forms end in the affix –er, while such an affix is absent in Romance. And finally, both the verb and noun are stems in English, and not words as in Romance. It should be noted, however, that the English and Romance forms also have a number of properties in common. In both types of languages the verb is transitive-causative and the noun is its complement. They also share the presence of an agentive or instrumental element in their interpretation.6 Moreover, they have the same overall semantic structure, the one of (16) in Section 2. Our purpose here is to present an analysis of the English forms that does justice to their similarities with the Romance compounds and at the same time accounts for the differences between them. In particular, we want to propose that the analysis we have presented above for Romance holds for English as well, but that, in addition, the English forms involve some extra steps, which follow from the distinct status of the nominalizer. Recall that we claimed that the (empty) nominalizer in Romance is non-affixal. For English, we follow the standard assumption that it is the overt affix –er. We suppose that –er merges with a vP with a pro subject, just as this happens in the case of the Romance nominalizer. The affixal character of –er however requires a further step: the verb must incorporate in it, in order to comply with the Stray Affix Filter (Baker 1988). Apart from the fact that this 4

Kayne (1994:41) briefly discusses the possibility that both English and French forms are partially subject to the same analysis. Varela (1987) presents an analysis for Spanish which greatly resembles the one presented in Lieber (1992) for English. 5 Di Sciullo and Williams (1987:82) claim that their rule for Romance VN compounds (cf. section 1.1 above) is a “word creating rule of the periphery of the grammar”. Such a rule plays no role in their analysis of the English forms. 6 Our reviewer points out that English also has forms such as early riser. We leave these out of consideration here.

22

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

entails that the verb must be a stem, there is another consequence. We know from the work of Baker (Baker 1988) that incorporating a verb in a lexical affix annihilates the verb’s licensing ability with respect to its complement. In the Minimalist framework verbs are no longer uniquely responsible for the licensing of their complements, but Baker’s insight can be easily adapted to this framework under the assumption in (24): (24) Verbal stems cannot activate the Case-checking feature of v It follows that the complement in the English compounds cannot be Caselicensed. This accounts for the fact that, as we have already briefly mentioned in 1.2, English resorts to a different licensing strategy than Romance, which is incorporating the complement, a stem, in the verb. The analysis we now obtain is presented in (25) for screensaver: first screen incorporates in save, then screensave moves to v and from there it incorporates in –er. (25) [N’ screen+sav+er [vP pro [v’ screen+save [VP screen+save screen]]]] The final step of the derivation is the same as in Romance: the pro subject moves and merges as the A'-specifier of the nominalizer, which gives us the result in (26). (26) [NP pro [N’ screen+save+er [vP pro [v’ screen+save [VP screen+save screen]]]]] Notice that the only visible part of the structure in (26) is the incorporation complex. This complex, which we reproduce in a more detailed form in (27), is in fact the same as the one proposed in Lieber (1983), Fabb (1984), Sproat (1985) and others for the compounds as a whole.7 However, in our approach it is just a by-product of the affixal status of the nominalizer and for this reason absent in comparable Romance forms, which apart from this have the same derivation. (27) [N [V [N screen] save] er]

7

We are not implying that the authors mentioned here disregard the semantic properties of the compounds. However, their approaches differ from ours in that they do not implement these structurally.

TAKING A CLOSER LOOK AT ROMANCE COMPOUNDS

23

There is one detail of the analysis above that requires further discussion: the fact that it involves the same pro as the Romance compounds. Such a pro is not otherwise attested in English. French Je veux le (pro) rouge, for example, translates as “I want the red one” and not as “I want the red”. An assumption that accounts for this is that English morphology is normally too poor to license pro. It should be noted however that –er, which we consider to be the licenser of pro in the forms under consideration, bears a number feature. This is evidenced by the opposition –er/–ers, as in the pair screensaverscreensavers. We assume that it is this phi-feature for number that accounts for the possibility to use pro (see also Section 4 below). Corroborating evidence for the pro-licensing capacity of number features comes from Dutch, another Germanic language, where we find pairs like those in (28), in which pro appears with an adjective that manifests morphological alternation for number (rood/rod-e). (28) a

b

Jan heeft een blauw overhemd en Piet een rood (pro) Jan has a blue shirt and Piet a red “Jan has a blue shirt and Piet a red one” Jan heeft twee blauwe overhemden en Piet drie rode (pro) Jan has two blue shirts and Piet three red “Jan has two blue shirts and Piet three red ones”

4.

On the relation between morphology and syntax It is obvious that our analysis of Romance VN compounds and their English counterparts is very syntactic in nature. Nevertheless we would like to maintain that the structures we propose are not derived in syntax. The main reason for this is that there are clear restrictions on the type of the vPs that can be nominalized. For instance, the complement cannot be a clause. Thus, alongside French compte-tours “revolution counter”, a form such as the one in (29) does not occur. (29) *un [compte combien de tours le moteur fait ] a count how+many of revolutions the motor makes Similarly, the complement cannot be a referential DP, as in (30). (30) *un [compte ces tours ] a count these revolutions

24

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

In general terms, the functional categories that make up referential expressions or account for the anchoring of propositions are absent in the compounds under consideration. In our opinion, the lack of such functional categories is a crucial factor that distinguishes morphology from (sentential) syntax and follows from its very nature, i.e. the fact that it is the component of grammar that is concerned with the formation of words (or ‘predicates’ in the Fregean sense), instead of sentences. In other words, we consider morphology an ‘impoverished’ version of syntax. Under this view both components in essence respect the same rules and principles of Universal Grammar. However, the poverty of morphology may sometimes lead to the use of strategies that are absent in syntax. As an example, consider NN compounds such as French homme-grenouille or its English equivalent frogman. In these compounds the head noun is restrictively modified by the other noun. To achieve an equivalent relation in syntax, one has to resort to full-fledged relative clauses, as in un homme qui ressemble à une grenouille or a man who looks like a frog. Morphology, lacking the functional clausal domain, cannot make use of modification by such clauses and must therefore resort to an alternative, which results to be modification by pure nouns. Returning to VN compounds, we assumed above that, whereas the complement noun is either Case-licensed or incorporated, the pro subject is merely licensed by the presence of the phi-feature on the nominalizer ([natural] in Romance and [number] in English). We take this to be another consequence of the poverty of morphology. In syntax the subject would be made visible for theta-role assignment by Case-checking T, and its formal content would be identified by the phi-features of V in T. The absence of T in morphology causes this component to accept pure phi-feature identification as a sufficient condition for theta visibility. One final remark is in order here. We have analyzed the VN compounds as rather complex structures involving complements and specifiers. Thus in fact we claim that the distinction between morphology and syntax does not go as far as to prohibit projections of heads. In doing so, our approach crucially differs from approaches to morphology such as that of Selkirk (1982) and Di Sciullo and Williams (1987), who only allow for head-adjunction structures in this component. Our position has as a consequence that morphological constructs may have the status of XPs. Of course complex words are still words and function as heads in syntax. This entails that the XPs under consideration must be reinterpreted as such before entering syntactic numerations. We call this reinterpretation recycling. The general format of recycling, which involves no change in category, is presented in (31).

TAKING A CLOSER LOOK AT ROMANCE COMPOUNDS

25

(31) Convert XP to X0 In the case of our VN (and PN) compounds, which are morphologically NPs, recycling then entails conversion to N0. 5.

Conclusion In the first part of this article we have argued that the analysis of Romance VN compounds as nominalizations of VPs should be extended in such a way as to involve vPs with a pro subject. We have also argued that movement of this pro subject to the specifier position of the nominalizer provides us with an adequate overall analysis of the compounds. In the second part we have shown that our analysis for Romance can be transposed to analogous English synthetic compounds, but that these, in addition, involve an incorporation strategy that is a consequence of the nature of the English nominalizer. Finally, we have stressed the fact that our proposals entail that morphology is highly parallel to syntax, but still has its own ontological restrictions. References Baker, Marc C. 1988. Incorporation: A theory of grammatical function changing. Chicago: University of Chicago Press. Chomsky, Noam. 1995. The minimalist program. Cambridge, Mass.: MIT Press. ----------. 2001a. Beyond Explanatory Adequacy. Ms., MIT. ----------. 2001b. “Derivation by Phase”. Ken Hale: A Life in Language, ed. by Michael Kenstovicz, 1-52. Cambridge, Mass.: MIT Press. Di Sciullo, Anna-Maria. 1996. “X’ Selection”. Phrase Structure and the Lexicon, ed. by Johan Rooryck and Laurie Zaring, 77-107. Dordrecht, Boston & London: Kluwer Academic Publishers. ----------. 2003. “Morphological Relations in Asymmetry Theory”. Asymmetry in Grammar, Volume 2: Morphology, Phonology, Acquisition, ed. by Anna Maria Di Sciullo, 9-36. Amsterdam: John Benjamins. ----------. & Edwin Williams. 1987. On the Definition of Word. Cambridge, Mass.: MIT Press. Fabb, Nigel. 1984. Syntactic Affixation. Doctoral dissertation, MIT. Ferrari-Bridgers, Franca. 2003. “[V-N] compound nouns formation in Italian”. Abstract Going Romance 17. Gračanin-Yuksek, Martina. 2005. “Italian V-N Compounds and Agreement”. Paper presented at LSRL 35, Austin, TX.

26

REINEKE BOK-BENNEMA & BRIGITTE KAMPERS-MANHE

Harris, James W. 1983. Syllable Structure and Stress in Spanish. A Nonlinear Analysis. Cambridge, Mass.: MIT Press. Kayne, Richard S. 1994. The antisymmetry of syntax. Cambridge, Mass.: MIT Press. Levi, Judith N. 1978. The Syntax and Semantics of Complex Nominals. New York: Academic Press. Lieber, Rochelle. 1983. “Argument Linking and Compounding in English”. Linguistic Inquiry 14.251-286. ----------. 1992. Deconstructing morphology: word formation in syntactic theory. Chicago: University of Chicago Press. Roeper, Thomas & Muffy Siegel. 1978. “A Lexical Transformation for Verbal Compounds”. Linguistic Inquiry 9.199-260. Riente, Lara. 2003. “Ladies first: The Pivotal Role of Gender in the Italian Nominal Inflection System”. McGill Working Papers in Linguistics 17:2.153. Rohrer, Christian. 1977. Die Wordzusammensetzung in Modernen Französisch. Tübingen: TBL Verlag Gunter Narr. Sleeman, Petra. 1996. Licensing Empty Nouns in French. The Hague: HIL Dissertations. Spencer, Andrew. 1991. Morphological Theory. Oxford: Basil Blackwell. Sportiche, Dominique. 1999. “Reconstruction, Constituency and Morphology”. Abstract, Glow. Sproat, Richard. 1985. On Deriving the Lexicon. Doctoral dissertation, MIT. Varela Ortega, Soledad. 1987. “Spanish Endocentric Compounds and the ‘Atom Condition’”. Studies in Romance Linguistics, ed. by Carl Kirschner and Janet Ann DiCesares, 397-441. Amsterdam: John Benjamins. Vergnaud, Jean-Roger. 1974. French Relative Clauses. Doctoral dissertation, MIT. Zwanenburg, Wiecher. 1990. “Argument structure in derivation and compounding”. Recherches de linguistique française et romane d’Utrecht, 37-42. Utrecht University.

BEYOND DESCRIPTIVISM EXAPTATION AND LINGUISTIC EXPLANATION

DENIS BOUCHARD Université du Québec à Montréal

0.

The Uniqueness of Language What exactly distinguishes human language from the systems of communication of other animals? The question is interesting in itself. In addition, the answer we give to this question correlates directly with the way we analyze linguistic data in its fine details. The beliefs we have about what is qualitatively new in human language affect our method of investigation. To illustrate this, I will present two views on what is unique to human language—the view in Hauser, Chomsky & Fitch (2002) (henceforth HCF), and my own view—and I will then compare the kinds of analyses that they lead to. 1.

Hauser, Chomsky & Fitch (2002): Recursion as a mediator is unique HCF assume the standard position that the faculty of language in the broad sense (FLB) is a link between meaning and perceptual form. It comprises three systems as represented schematically in (1): (1) FL Broad: Sensory-Motor=computational system=Conceptual-Intentional The computational system which operates the mediation between the sensory-motor system (SM) and the conceptual-intentional system (CI) has the crucial property of being recursive. FLB is very inclusive and covers all of the capacities that support language independently of whether they are specific to language and uniquely human. Since language is unique to our species, it is possible that some subset of the mechanisms of FLB is both unique to humans, and to language itself. HCF propose the term ‘faculty of language in the narrow sense’ (FLN) to refer to this subset of mechanisms unique to human language.

28

DENIS BOUCHARD

HCF is a vast study which evaluates the possibility that all the components of the systems in FLB may be spandrels, in the sense of Gould & Lewontin (1979), by-products of preexisting constraints rather than an end product of a history of natural selection. If this turns out to be the case, then the traits of FLB did not evolve specifically for human language. HCF argue that there is quite a lot of evidence supporting this hypothesis. They discuss several traits of SM and CI which are present in nonhuman animals. Their presence elsewhere indicates that these traits did not evolve specifically for human language. For instance, perceptual discontinuities like those in SM are found in other species, which means that categorical perception is not unique to language: it appears to have “evolved for general auditory processing, as opposed to specific speech processing” (HCF 2002:1572). Likewise, they discuss several studies which show that the mechanisms of CI are not specific to language, since we find rich conceptual representations in other animals and in other domains of human cognition, such as social hierarchies and abstract concepts like tool, color, geometric relationships, food, number. Even recursion is not unique to language: Fitch et al. (2005) note that spatial reasoning and navigation exhibit limitless embedding of place concepts within place concepts. Also, some animals have an impressive cognitive representation of time and space. Therefore, our capacities to pinpoint locations and mark the passage of time are “not unique, and may not rely on language at all (though language may enhance or modify the capacities).” The conceptual-intentional system of humans shares major similarities with other vertebrates. Surprisingly, however, there is a mismatch between the conceptual capacities of animals and the communicative content of their vocal and visual signals […] For example, although a wide variety of nonhuman primates have access to rich knowledge of who is related to whom, as well as who is dominant and who is subordinate, their vocalizations only coarsely express such complexities. (HCF 2002:1575).

The difference between the communicative content of humans and nonhuman animals cannot simply be due to the absence of recursivity in the capacities of nonhuman animals, since they have rich conceptual capacities which suggest that they have a CI component with elementary parts and formation rules similar to ours. Moreover, some recursive aspects of the human CI appear not to rely on language. Therefore, HCF hypothesize that what is unique to human language is not recursivity per se but rather how recursion is set in FLB: a recursive system links SM and CI, thus allowing humans to express complex and potentially infinite conceptions by

BEYOND DESCRIPTIVISM

29

combining words into sentences. “No species other than humans has a comparable capacity to recombine meaningful units into an unlimited variety of larger structures, each differing systematically in meaning” (HCF 2002:1576). More precisely, HCF assume that recursion developed in organisms with a brain as an optimal solution for problems such as spatial navigation. In evolutionary studies, when a structure that evolved for one function starts serving a quite different function, this is referred to as an exaptation. HCF suggest that recursion was exapted from its navigational function and took on the additional function of linking the SM and CI systems in order to solve the problem of communication. During evolution, the modular and highly domain-specific system of recursion may have become penetrable and domain-general. This opened the way for humans, perhaps uniquely, to apply the power of recursion to other problems (HCF 2002: 1578). Because recursion evolved for reasons other than communication, it is not unique to human language. However, according to HCF, the way recursion is embedded as the mediator between SM and CI is unique. They hypothesize that this property is the only one which is unique to human language, that FLN reduces to this single property. Under this hypothesis, recursion has a rather odd status. The transition between applying recursion to navigation and applying it to communication involves a reification of recursion: it is a modular system which exists independently and humans apply its power to another problem— communication. But what is this problem? From their discussion, it appears that recursively combining meaningful units solves the problem of expressing unlimited conceptions. An infinity of conceptions presupposes recursion in CI, and HCF themselves have presented evidence that recursion occurs in nonlinguistic aspects of the CI system of both humans and other animals. This undermines the need for an additional recursive system: it is possible that all the recursivity observed in language comes from the independently motivated recursive systems in CI. In that case, recursion in the linking between SM and CI is totally redundant and can be dispensed with. HCF do not explore that avenue. That’s a crucial difference between their model and mine. 2.

My view (Exaptive Grammar): Signs are unique I follow the general view of Bouchard (2002) which emphasizes the fact that the initial conditions for language are those of the initial state of a human being. i.e., the substantive properties of the physiology and cognition of humans determine what is a possible language. The faculty of language is conditioned by the fact that “the brain in which the language system is

30

DENIS BOUCHARD

represented also contains a conceptual system with its own properties, and is set in human bodies that have particular sensorimotor systems that determine the kind of form which can participate in language” (Bouchard 2002:2). These initial conditions are properties which are logically prior to linguistic theory, in the sense that they must preexist linguistic theory since its object of study presupposes them. If we restrict our analytical tools to those which derive from these prior properties, then the tools of the theory are not aprioristic. These analytical tools are of no cost because they derive from the initial conditions. Under this view, recursion has no special ontological status, as it appears to have in HCF: instead, as we will see, recursion is a by-product of the linguistic sign, of the linking of elements of two very different kinds of substances—perceptual substance and conceptual substance. So what is unique to human language is not recursion, but the fact that it includes signs. As a starting point, note that there are already countless cases which show that recursion is not always required in linking SM and CI. The meaningful units on which recursion operates are themselves linking rules: in Saussure’s terms, they are signs which link an acoustic image (signifiant) with a meaning (signifié). The linking between a signifiant and a signifié at the level of words does not involve recursion. So the schematic representation in (1) is only part of the model: it corresponds to the complex linking which takes place when two or more signs combine in a complex sign. However, in a simple sign such as a word, no recursion is involved in the linking itself. For instance, the word assassin has a very complex meaning which most likely involves recursion at CI (evoked by the capitalized words in (2)), and also a complex phonological structure which involves recursion at SM. However, the linking between the signifié and the signifiant does not involve recursion. (2)

Conceptual-Intentional:{PERSON, KILL, POLITICAL MOTIVE} ↑ ↓ Sensory-Motor: {asasIn}

As Saussure strongly insisted, the link between a signifié and a signifiant is arbitrary, in the sense that each language partitions the acoustic and conceptual realities in its own way. The source of this arbitrariness and the reason why the meaning-form pairings vary considerably from one language to another, is that there is no ‘reason of nature’ for any particular pairing, because the sounds produced by our phonatory articulators cannot

BEYOND DESCRIPTIVISM

31

have a meaningful, logical or iconic relation with the meanings to be expressed. So arbitrariness and the variation that derives from it are an artefact of the physiological makeup of human beings. The problem with the recursion approach is that it takes for granted that there are meaningful units, and that the uniqueness of human language is the application of recursion to these units to yield discrete infinity. However, if the only distinctive property is that nonhuman animals lack this recursive system, this leaves open the possibility that some animals could develop a large lexicon of signs, since signs do not involve recursion in linking SM and CI. But this prediction is not borne out: nonhuman animals do not even have words/signs. So this recursion is not the distinctive property of human language, and there must be some other constraint which animals have not overcome. We now have the key to what is unique to human language. Humans apply recursion to meaningful units. But you must have these units in the first place. There is no mystery in the fact that other animals, despite their having recursive processes in other domains, did not apply recursion to communication: other animals simply don’t have the proper kind of units from which recursion derives in language, they don't have signs. Animal calls are not signs, they are not a link between a signifiant and a signifié; animal calls are just reflex responses. A reflex response is independent of the will, it is an automatic response to a perceptual stimulation, like blinking your eyes as a protective reaction, or crying out if surprised or frightened. HCF (2002:1576) give a summary of recent studies on animal calls. Key points indicate that animal calls are reflex responses. First, calls are situation-specific: distinctive calls are in immediate response to functionally important contexts, such as detection of predators and the discovery of food. Second, the repertoire of calls is small, “restricted to objects and events experienced in the present, with no evidence of creative production of new sounds for new situations.” Third, “there is no evidence that calling is intentional in the sense of taking into account what other individuals believe or want.” I hypothesize that the only distinctive property of human language is that it has signs, i.e., a signifiant linked to a conceptual signifié instead of a behavioral reflex. The crucial transition is from reflexes which are immediate and limited to specific situations, to signs which have a nonimmediate and potentially unlimited signifié. As HCF remark, most of the words of human language are not associated with specific functions (e.g., warning cries, food announcements) but can be linked to virtually any concept that humans can entertain. Such usages are often

32

DENIS BOUCHARD highly intricate and detached from the here and now” (HCF 1 2002:1576).

Once triggered, the transition from reflex to sign quickly brings about a system that is as expressive as modern language. As soon as the system goes beyond the immediacy of reflexes, it opens up to expressing the complexity of all that is nonimmediate. The domain of what is nonimmediate comprises all that we can conceive, which appears to be unlimited. This is the source of the communication problem of expressing unlimited conceptions which HCF mention. It is logically possible to obtain limitless expressive power without recursion in the linking of SM and CI. All that is needed is an infinite set of meaningful units, including units that correspond to complex meanings such as those expressed by complex sentences. CI is recursive, so there is a limitless potential of signifiés. On the other hand, our perceptual system sets upper bounds on the distinctions which we can perceive or produce as signifiants. Since these distinctions allow a quite limited number of discrete perceptual elements, recursion operating on these elements is the means to obtain an unlimited number of signifiants linked with the unlimited possible signifiés. If the symbols were holistic vocalizations like primate calls, even a thousand symbols would be impossible to keep distinct in perception and memory (Nowak et al. 1999). Modern language deals with this problem by building words up combinatorially from a repertoire of a few dozen smaller meaningless speech sounds” (Jackendoff 2002: 242).

Any sequence of phonemes making up a word can be augmented by adding a phoneme or syllable to it: so recursion of these elements solves the problem of perceptual distinctivity and allows an unlimited derivation of signifiants. Therefore, the expressive power of words is limitless: any potential conception, however complicated, can be the signifié of a sign, and similarly, any combination of distinguishable percepts can be a signifiant. Indeed, humans do invent words quite freely. However, an infinite lexicon is impossible in practice due to pre-existing constraints, the most obvious one being hat the brain is finite, so that there are limitations on the memory our 1

Fitch et al. (2005) note that “the ability to link novel arbitrary noises to some referent appears to be quite general among vertebrates, present in some form not only in chimpanzees but in parrots, dogs and other species.” Crucially, these are not signs as defined, they do not link a signifiant and a signifié, but rather a sound and a referent: they are calls, i.e., immediate, situation-specific reflexes.

BEYOND DESCRIPTIVISM

33

brain provides to store words/signs.2 The way around the limitation is recursion once again, this time combining words into complex phrases and sentences. For instance, consider the combination of two words as in the expression pink poodle. Each word is a sign, and a sign is necessarily complex: neither the acoustic image nor the meaning is a sign; only the whole resulting from the union of the two elements is a sign (Saussure 1916: 99). Therefore, when the association function Merge operates on actual elements in a particular derivation, it always operates on complex elements which are the union of a signifié and a signifiant. Representations are usually simplified by taking the written form of a word as a symbol for the actual sign. However, this is only a convenient format to help us discuss the forms, and we must bear in mind that the representation of the merger of two words is as in (3) (to which a full representation would add the primitives which combine into the CI and SM of each word). Conceptual-Intentional: (3) Sensory-Motor:

{PINK} === Merge === {POODLE} | | {pink} === Merge === {poodle}

The properties of the signs are already in the system and Merge operates on these. Since signifié and signifiant are irreducibly united, Merge cannot operate on one without operating on the other. In Saussurean terms, Merge creates organized groups of signs which are themselves signs 2

The formation rules involved in CI are most likely not combinations of word-like meanings similar to those found in human languages. First, nonhumans appear to have rich conceptual systems, but do not have language. Second, even in the case of word meanings, Fodor et al. (1980) has shown that they cannot be constructed by combining other word meanings, using the principles that combine words into phrases. As Jackendoff (2002: 335-6) indicates, the fact that lexical meanings cannot be decomposed definitionally suggests that meanings arise from non-definitional composition: the parts of lexical concepts may be units which cannot individually serve as possible word meanings, and the principles by which these sublexical units combine may be different from the principles by which word meanings combine into phrase meanings. See also Bouchard (1995). Jackendoff points out that this situation is parallel with composition in phonology: most speech sounds cannot be words on their own; distinctive features cannot be words nor even sounds on their own. Moreover, concatenation combines speech sounds into syllables, syllables into words, and words into phrases; but distinctive features do not combine into speech sounds by concatenation. A similar parallel can be drawn with particles in physics. There is a first layer of a limited repertoire of substances: oxygen, boron, sulfur, etc., combined by chemical bonds; a next layer of entities can exist in isolation but are not substances: electrons, protons, neutrons, combined by electromagnetic and nuclear forces; a further layer of entities are not substances and cannot exist in isolation: quarks, which are like features of elementary particles.

34

DENIS BOUCHARD

(Saussure 1916: 177). If there is recursion operating on the signifiés, it necessarily operates on the signifiants, and vice versa. The linking between SM and CI transfers the effects of recursion on one level to the other, but recursion never operates on the linking itself. In this case, Type-recursion is involved: embedding an element of type X within other X elements indefinitely. We must distinguish Type-recursion from the Concatenation-recursion of distinguishable percepts mentioned above. As Fitch et al. (2005) indicate, phonology is hierarchical, but appears not to be Type-recursive: Certainly, syllables cannot be embedded within other syllables indefinitely. However, even in this domain of inquiry, there are many open questions. Syllables are not the only relevant components of phonological structure, and other constituents like intonational phrases seem much better candidates for unlimited self-embedding.” (Fitch et al. 2005:201)

Type-recursion derives from memory limitations on the number of signs that can be stored in the lexicon. The listed signs being relatively few, their signifiés tend to correspond to fairly general and usual categories of things, actions, qualities, etc. When someone wants to express a signifié which corresponds to a less general and usual category, distinguishing properties are added to the stored word/sign by combining it with another sign to zoom in on more specific elements of the same kind. For instance, the set of elements denoted by poodle can be further narrowed as in pink poodle or further again as in the phrase pink poodle with a top hat. Type-recursion is the case when one of the restraining elements happens to be of the same type as the restrained element. We are now in a better position to clarify how recursion functions in language. Recursion is not part of the linking in a sign, simple or complex. Moreover, recursion does not come out of the blue. A sign is a link between perceptual substance and conceptual substance. As we saw, these substances restrict the possibilities of language and lead to recursion. Hence, recursion is a by-product of the sign. More precisely, recursion is a by-product of the substantive properties of the physiology and cognition which define the initial state of a human being. The role which I attribute to the sign is quite different from the view espoused in generative grammar. For instance, in his very brief discussion of the sound-meaning pairing of the substantive part of the lexicon, Chomsky (1995: 8) assumes that “these matters [...] appear to be of limited relevance to the computational properties of language.” In contrast,

BEYOND DESCRIPTIVISM

35

I argue that the sign has profound relevance for the computational system, since recursion derives from it. The transition I propose from animal behavior to human language is similar to the scenario presented in Jackendoff (2002: 427): The initial innovation is permitting sounds to be used symbolically in a non-situation-specific fashion. This is admittedly a mysterious advance over primate calls, but vastly less expressive than modern language. From there we are able to see the evolution of the language faculty as the successive addition of more and more “tricks” to the toolkit, in the interest of greater expressiveness and precision in conveying thought.” (Jackendoff 2002: 427)

However, I don’t agree with his assessment that this is a mysterious advance. There are quite plausible intermediate steps in this transition. For instance, a roar repeatedly produced upon the intrusions of a competitor in a territory, or markings of this territory by scents, are still immediate reflexes: the animal establishes its territory here and now. However, the gesture has future consequences, so this is the beginning of non-immediacy, of abstraction towards concepts: it is still a reflex, but with an inferential chain added to it. Another example is when a dangerous animal triggers the reflex response of fleeing and shouting (possibly to startle the adversary and have more time to flee). If this shout starts acting as a trigger on its own and peers respond to it by also fleeing (and possibly making the same shout), there is the beginning of communication which is not immediate, not triggered by the situation of danger, but by a reaction to this situation by a peer. Moreover, in my view, the situation is not that our language faculty provides us with a toolkit for building languages, and languages make choices among the tools; instead, I assume that the toolkit of FLB is made of the logically prior properties of the perceptual and conceptual substances. As for FLN, all it contains is the association which forms signs, the linking between a signifié and a signifiant. 3.

Deriving other novel traits My hypothesis has another important advantage over the proposal in HCF. In addition to recursion in the linking between SM and CI, they point out several other novel traits, constraints in animals which humans had to overcome at some point in order to acquire natural languages. In their model, these traits appear to be unconnected, to have independently emerged from an ancestral node and have converged into human communication. Under their view, each of these specific traits constitutes a separate empirical discovery, a novel development in our recent evolutionary history. This convergence of

36

DENIS BOUCHARD

capacities appears to be fortuitous. Why the correlation should hold remains somewhat mysterious. In contrast, these traits all follow from my hypothesis that the distinctive property of human language is to have signs. Novel trait 1: Words can be linked to virtually any concept. This is the fundamental change according to my hypothesis, the transition from limited situation-specific reflexes to unlimited signifiés. Note that this trait does not follow from HCF’s hypothesis that recursion is the basic distinctive property of human language. Novel trait 2: Human language can convey an unlimited variety of ideas. HCF attribute this trait to recursion in the linking of SM and CI. I argued that there is no recursion in this linking per se, and that, in any event, recursion alone is not sufficient to derive this trait: we need the appropriate units. The signs postulated in my hypothesis are appropriate units, and recursion is obtained derivatively from the interaction of the unlimited signifiés and the perceptual limitations on the signifiants. Novel trait 3: Humans rapidly acquire a large lexicon. The lexicon requires the storage of a large number of different ‘phonological images.’ But there is more than that to it. The lexicon is not a nomenclature of concepts that get labeled: as variation across languages shows, it involves parcelling out the conceptual substance into signifiés which differ from one language to another. Therefore, the elements being stored consist in pairings of a signifiant and a signifié. This irreducible link between signifié and phonological image has a strong mnemonic effect which facilitates the rapid acquisition of a large lexicon. We can see this effect when we try to learn sound bits from lexical items of a language we don’t know: without the leverage from the meaning associated with the sound bits, we are quite poor at it, and quickly limited. This trait—a large lexicon— therefore follows from my hypothesis. But it does not follow from HCF’s hypothesis that recursion is the basic distinctive property of human language. They suggest that this novel trait is an independently evolved mechanism, and they tie it to the human capacity for vocal imitation: Imitation is obviously a necessary component of the human capacity to acquire a shared and arbitrary lexicon, which is itself central

BEYOND DESCRIPTIVISM

37

to the language capacity. Thus, the capacity to imitate was a crucial prerequisite of FLB as a communicative system” (HCF 2002:1574).

Novel trait 4: Humans have an impressive capacity for vocal imitation. This trait also does not follow from recursion. While HCF see this trait as “a novel capacity that evolved in our recent evolutionary history” which helps in the building of the lexicon, I see it the other way around. It is not that imitation helps the lexicon, but instead that the lexicon—more precisely, the sign—is what allows such a qualitative improvement in imitation: the association of a sound image with a concept has a powerful mnemonic effect which makes remembering and repeating sound bits much more productive than imitating the sound images in isolation. In other words, the crucial difference is that humans imitate signs whereas non humans imitate sound images. Novel trait 5: Human language is detached from any specific modality. When human beings lose one modality (e.g., hearing), they make up for this deficit by communicating in a different modality (i.e., signing). This is another trait which does not follow from recursion. Under my view, this trait also illustrates the difference between a reflex and a sign. Reflexes are a response to a perceptual stimulation. The fact that they are ‘hard-wired’ in this way makes them modality-specific. Thus, even in humans, the gesturecall system (laughter, sobbing, facial signaling) is modality-specific. In contrast, the inputs to the language system are signs, which precisely have the property of being arbitrary, without any connection imposed by ‘nature’ between signifiant and signifié. This holds internally to a modality, and there is no reason from nature why it should not also hold across modalities. So again the trait derives from properties of the linguistic sign. Novel trait 6: Human language shows intentionality. Calls are reflexes, automatic responses to perceptual stimulation, so they are not intentional in the sense of taking into account what other individuals believe or want. Moreover, intentionality requires consciousness of self and of others. The reason why nonhuman animals are limited to nonintentional calls has to do with the kind of consciousness that they have. Higher animals have more than perception, they have consciousness. However, although they are conscious, their consciousness is directed outwards to the perceivable world: as Schopenhauer puts it, theirs is a world

38

DENIS BOUCHARD

of perception. This limits them to immediacy in their interaction with the world, hence to calls in their communications. On the other hand, according to Kant, Hegel and many other thinkers, what sets humans apart is that we are self-conscious. Self-consciousness allows us “to distinguish how things seem to me from how they seem to you” (Scruton 2005: 76). This provides the consciousness of self and of others which is a prerequisite for intentionality. It brings about a detachment from immediacy which directly correlates with the nonimmediate and potentially unlimited signifié of a system of signs like human language. The transition from mere consciousness to self-conscious directly correlates with the transition from reflex calls to signs. Novel trait 7: Humans acquire a number system with open-ended recursivity, other animals do not. HCF review some literature on animals and developmental studies of infants. This research shows that there are two core knowledge systems at work for the number concepts which humans and many animals share: a system for representing exact small numbers of objects or events, and a system for representing large approximate numerical magnitudes. After a slow progress from understanding the meaning of one to understanding two, and then three, children take a leap forward: they grasp the idea that the integer list is constructed on the basis of the successor function. By age 5, children can apply this knowledge robustly. Non human animals never undergo this kind of “aha” experience: they “learn the integer list by brute association, mapping each symbol to a discrete quantity” (Hauser & Spelke, in press). HCF underline that there is a convergence of evidence that children possess, but animals lack, the capacity to apply open-ended generative systems in language and in the number system (HCF 2002:1577). Note that this says nothing about where exactly recursion is embedded in the faculty of language. Moreover, Hauser & Spelke remark that “it is not clear whether one mechanism evolved first for language and then for number or whether two distinct mechanisms evolved separately.” In addition, the learning environment of children and the training regime of animals differ in a crucial manner. Animals are taught the meanings of the symbols one by one, whereas children typically first learn an ordered list of symbols: the fact that the list is ordered may play an important role in the transition to acquiring the successor function. Furthermore, recursion in language comes very early on, whereas it is mastered only later in numbers. This connects with Hauser & Spelke’s hypothesis that children construct the natural number concepts by using natural language—number words and verbal counting: so they must

BEYOND DESCRIPTIVISM

39

first acquire a minimum of relevant vocabulary. The hypothesis ties in nicely with the discussion of the limitations of arbitrariness by Saussure (1916: 182). He notes that arbitrariness applied without restrictions would lead to extreme complexity. Language gets around the problem of an infinitely complex list by combining simple signs into a complex sign, so that the complex sign is relatively motivated. Interestingly, one of his examples is French dix-huit “eighteen” whose value is the addition of the values of dix “ten” and huit “eight”. Languages typically express numbers in this way: there is a short list of words corresponding to small numbers, and then the number words are combined in a way that coincides with simple arithmetic functions, and often in a base-ten system. Thus, dix-huit is preceded by dixsept (lit. “ten-seven”) and followed by dix-neuf (lit. “ten-nine”). This strong correlation between simple number theory and how languages express numbers no doubt helps children in learning the number system, including its recursivity. This may help explain some aspects of the learning of the number system, but it leaves open the question why only humans have a number system with open-ended recursivity. The answer may be in the general distinction we have already observed between animals and humans. Animals do not take any distance from immediacy, in contrast with humans. Recursion is an abstract definition of a set, a non immediate notion. So in numbers as in language, animals lack the first crucial step, which is Saussurean arbitrariness in the broad sense of detached from immediacy, as in signs, without any connection imposed by ‘nature’ between signifiant and signifié. 4.

Syntactic analysis and the distinctive property of human language The HCF view and my view share the assumption that recursion is not unique to human language. However, the two perspectives on language differ on how recursion made its way into the faculty of language. HCF follow the method of generative grammar and devise conjectures which rely on formal strategies. Like Chomsky (2001), the general properties of organic systems which they see as part of the initial conditions for language acquisition consist in computational systems. This is what leads them to the hypothesis that recursion is in the mediation between SM and CI, having been somehow transferred from navigation to communication. On the other hand, my exaptive approach asks that we devise conjectures which rely on the logically prior properties of the perceptual and conceptual substances. This leads to the hypothesis that recursion derives entirely from the substantive nature of the elements linked by the signs. These differing points of departure affect the way linguistic analysis is conducted. In the following sections, I briefly

40

DENIS BOUCHARD

compare the way the two approaches tackle five well-studied syntactic phenomena. 4.1.

The Head parameter As discussed in Bouchard (2002: 60-61), when two signs A and B are combined to form a complex sign, the signifié of this complex sign consists in the semantic relation established between the signifié of A and the signifié of B. This complex signifié must have a corresponding signifiant—it must be associated with a perceptual form. In a fundamental semantic relation, there is a primary element P to which a dependent D is added, either as an argument or as a modifier. These semantic relations are always asymmetric: an element is never simultaneously a primary P and an argument or a modifier of its D. The signifiant of the complex sign must indicate which signifié, of A or of B, acts as the primary element, and which one as the dependent. One way to encode this information is to juxtapose the signifiants of A and B: this linear order is then the signifiant of the complex sign. Linear order is also asymmetric: A either precedes or follows B. Semantic combination and temporal linearization are united in a complex sign, and both exhibit an asymmetry: therefore, their asymmetries must be matched. The choice of the temporal order A-B or B-A as the signifiant of the complex sign is arbitrary: this arbitrariness forces the association of the semantic and linear asymmetries to be conventionalized. This derives a head parameter as in (4): (4)

Linearization Parameter: The primary element precedes/follows its dependent.

This account of the Linearization Parameter is in line with my exaptive approach: the notions involved are based on logically prior properties of the conceptual and perceptual substances. Thus, on the conceptual side, the source of the asymmetry between Primary and Dependent presumably lies in our cognitive system. On the perceptual side, the asymmetry in temporal Juxtaposition is due to inherent physiological properties of the human sensorimotor apparatus which are such that two words cannot be produced at the same time in oral languages (Saussure’s principle of linearity). Therefore, crosslinguistic variation in the values of the Linearization Parameter is an instance of Saussurean arbitrariness. It arises because the conditions of the external systems provide more than one potential signifiant for the signifié: whether A precedes or follows B is just as good a signifiant for the signifié which consists in the semantic relation between them. Like all linguistic arbitrariness, the possibility of having different values for the Linearization

BEYOND DESCRIPTIVISM

41

Parameter derives from properties of human beings that are logically prior to language. Exaptive Grammar strives to derive all types of ‘parametric choices’ from deeply motivated properties in this way, i.e., by appealing to legibility conditions of the external systems. In the Minimalist Program however, variation of this kind receives very little attention because it is assumed that “these matters [...] appear to be of limited relevance to the computational properties of language” (Chomsky 1995: 8). Consequently, a different kind of account of variation is sought: “beyond PF options and lexical arbitrariness (which I henceforth ignore), variation is limited to nonsubstantive parts of the lexicon and general properties of lexical items” (Chomsky 1995: 170). These purported nonsubstantive parts are categories and features with no interpretation at the interfaces. Most of syntactic variation is attributed to uninterpretable features of functional categories. For instance, SVO and SOV languages are assumed to differ in nonsubstantive lexical attributes: an SOV language has a functional category with a feature which attracts the Object to the left of the V, whereas an SVO language does not. In this approach, variation of the type expressed by the Linearization Parameter is not linked to logically prior properties, but is stipulated in an ad hoc list of nonsubstantive elements. Taxonomies of this kind are not very helpful in furthering our understanding of the facts. It could be argued that this list of nonsubstantive elements is motivated on a par with the rest of the lexicon. Every language has a lexicon in which substantive elements must be listed because they vary arbitrarily. Putting this other list of nonsubstantive elements in the same part of the grammar could be seen as an interesting generalization since it restricts parametric variation to the lexicon. However, the generalization is only apparent, because the two lists are of a very different nature and must not be confused. As we saw in the discussion of (2), Saussure showed that we must list the meaning-form pairings of the substantive part of the lexicon because the logically prior properties of human physiology and cognition provide no ‘reason of nature’ for any particular pairing. This list is therefore highly motivated. In contrast, there is no explanation of this kind for the presumed uninterpretable features of the nonsubstantive part of the lexicon. This other list does not follow from any logically prior property, nor does it follow from anything internal to the theory: the assumption that a category X has an uninterpretable feature f is an irreducible descriptive statement. This weak type of list tells us that language has certain properties, but it does not help us understand why. Chomsky (1995: 233) recognizes that this kind of formulation “is a restatement of the basic property, not a true explanation.”

42

DENIS BOUCHARD

4.2.

Case marking languages In Case-marking languages, major constituents can appear fairly freely in several different positions. This clashes with the assumption in GG that grammatical functions like subject and object are defined by a universal list of configurations, with corresponding positions. Therefore, corrective solutions must be assumed. For instance, Chomsky (2000) says that Casemarked elements have “a scrambling feature [which] induces pied-piping even after Case assignment, with the pied-piped element ‘attracted’ by a higher probe.” Non-Case-marking languages like English simply do not have such “scrambling” features in their list of features. A list of uninterpretable scrambling features does not help us understand why there is a correlation between overt Case markings and scrambling. In this approach, there just mysteriously happen to be extra mechanisms in Case-marking languages that conspire to give the impression of a freer order. In contrast, Exaptive Grammar brings to the forefront the fact that syntagmatics is the functional use of temporal linearization and that it is crucially linked to the perceptual substance of oral languages. In Bouchard (1996, 2002), I set this in the broad context of the consequences of Saussure’s idea that the linguistic sign takes form according to the substances, in particular when it extends beyond units to cover relations between items. Under this view, linear order and Case marking are two functionally equivalent means among the four provided by the perceptual substance of oral languages to express the combination of signs. (5) Elements of the perceptual substance of oral languages which may combine with a relation of conceptual substance to constitute a form i Juxtaposition: A and B are ordered temporally next to one another, deriving the structural relation of sister and immediately contain. ii Superimposition: B is a modulation superimposed on A, such as intonations to express grammatical functions in tone languages. iii Dependent Marking: the dependent gets a marking, such as Case marking. iv Head Marking: the head is marked, as in predicate marking (polysynthetic languages). These four elements of perceptual substance are equivalent: any one of them is just as good a signifiant for the signifié of a relation. Therefore, covariation between case marking and rigid order does not require costly additions to the theory like a list of scrambling features. On the contrary, the

BEYOND DESCRIPTIVISM

43

fact that languages operate different choices among these means is expected, given the initial conditions arising from the perceptual substance. If there wasn’t this variation, we would have to explain why no language makes use of an option that the perceptual substance provides to the language faculty. 4.3.

Clitics It is a commmon observation that French has clitic pronouns and English does not. (6)

a. a’ b. b'

John sees it/him/her/them. *John it/him/her/them sees. Jean le/la/les voit. *Jean voit le/la/les.

In the generative model, this is accounted for by assuming that French has a list of strong and weak pronouns, and a list of positions in which clitics may appear. Over the years, the list of positions has been expressed by a variety of formal tools: a list of transformational rules which describe where each clitic must appear on the surface (Kayne 1975); a list of items lexically marked as ’simple’ and which must be adjoined to the left of V (Chomsky 1995); a list of items marked by a feature which triggers their adjunction to an I-type functional head and a list ordering these functional heads according to the surface positions of clitics (Kayne 2000); a list of affixal elements which can license an empty pronoun, with a list of hosts corresponding to the surface positions (Borer 1984, Roberge 1990). All these variants in the generative model have in common that they list which elements can be clitics in which languages, and they also list the hosts of the clitics, such as Tensebearing elements. What differs in a non-clitic language is that these lists are empty. Moreover, cliticization is presented as an accidental property, unrelated to any other property of the languages. We would like to go beyond these mere descriptive lists and have an explanation why these particular elements are cliticized and why the Tensebearing element of the clause is the host. The exaptive model of Bouchard (2002) provides such an explanation. Generally, a marking provides information about the interpretation of the element to which it is attached. If we maintain this optimal assumption, it follows that in Jean le voit, the clitic le provides indications to identify an actant of a proposition centered on the semantics of voit and dependent on its Tense; the form of le tells us that it is a direct object, so it relates in a specific manner to the main predicate voit. Since the clitic is affixed to a particular Tense head, this relates it to a

44

DENIS BOUCHARD

particular proposition. This directly accounts for the host, the clauseboundedness and the nature of the relation held by the clitic. The fact that English differs from French with respect to cliticization derives from an independently motivated difference in the way these languages encode semantic Number. Lamarche (1990, 1991) observes that the Det bears the semantic expression of Number in French, whereas semantic Number is borne by the Noun in English. Dwelling on this observation, Bouchard (2002: 34-35) shows that, consequently, Det can appear alone in French, it can be a clitic-marking which identifies an actant, because it bears Number, which provides it with the minimum content required to do so. When a speaker uses such a Det alone, without an N range, this indicates that the referent of the corresponding actant is part of the background shared by the speaker and addressee: the Det functions anaphorically. In English, since it is the Noun which bears Number, the Det does not have the minimum required to function anaphorically. The exaptive model provides an analysis which derives the facts from logically prior elements: the relation of Tense to the proposition and the role of Number in atomisation both derive from properties of the conceptual substance; marking of the Tense bearer by a clitic is one of the means allowed by our physiology—our perceptual substance—to express a relation; the fact that there is a choice to express Number either on Det or N derives from properties of both systems: Why Number is encoded in language at all is forced by CI properties: it comes into grammar as part of a grammaticalization of the logically anterior notions of ‘actant’ and ‘set’. Why Number varies in how it is realized is due to the fact that independent SM conditions force a choice upon languages. French expresses Number on Det, English on N, Walloon as an independent word, because these options correspond to different equally efficient ways to give a perceptual form to Number in the SM system: any of these forms is as good as the other to express the Number of a nominal expression (Bouchard 2002:34).

4.4.

Absence of Determiner There is another difference between French and English which directly correlates with the contrast in cliticization: rather than the Noun being absent, it concerns the absence of Det. English allows the absence of Det quite freely: a common noun can appear without a Det, whether it is plural (7a) or singular (7b). In contrast, such an omission of the Det is generally impossible in French, as can be seen in (8) (for a few special contexts in which this is allowed, see Curat 1999, Bouchard 2002: 273ff).

BEYOND DESCRIPTIVISM

(7)

a. b.

Beavers build dams everywhere. I ate lion yesterday.

(8)

a. b.

*Castors construisent barrages. J’ai mangé du/*Ø lion hier.

45

The situation is the converse of what we observed with clitics in the previous section. A Noun can appear alone in English and be used referentially because it bears Number, which provides it with the minimum content required for atomization, hence for identification of an actant. In French on the other hand, since semantic Number is expressed on the Det, the Noun alone lacks this minimum content for atomization. The same logically prior properties of the conceptual and perceptual substances which explained the cliticization facts also explain the observations about the absence of Det. 4.5.

Adjectives in French and English The distribution of adjectives in nominal expressions is another instance where the generative model and the exaptive model differ in their approaches. While complementless adjectives are almost exclusively prenominal in English (9a), they may appear before or after the noun in French (9b). (9)

a. b.

an expensive Chinese vase un vase chinois cher

A typical generative account of this contrast is found in Cinque (1994) (the same approach is used for adverbs in Cinque 1999). To account for the fact that adjectives appear in a fixed order before the noun in English, he postulates an ordered list of functional categories corresponding to meaning classes of adjectives (poss> cardinal> ordinal> quality> size> shape> color> nation). This ordering is projected in a hierarchical organization of the categories, and an adjective is generated in the Spec of the category corresponding to its semantic class. Underlyingly, French has the same ‘universal’ list of functional categories as in English in the same fixed order before the noun. The distinctive property of French lies in a list of uninterpretable features which trigger movement of constituents to functional categories. For example, the derivation of (9b) starts off with an underlying order similar to that of English, with the adjectives in prenominal position as in (10a). The N vase is attracted by some feature to the left of chinois, deriving (10b). Then vase

DENIS BOUCHARD

46

chinois is attracted by yet another feature to the left of cher, producing the surface order in (10c).3 (10)

a. b. c.

basic order: un cher chinois vase intermediate: un cher vase chinois surface order: un vase chinois cher

==> ==>

The difference between the two languages is that the list of uninterpretable features of French contains these two attracting features, but the English list does not. This just restates the facts. Similarly, to account for pairs as in (11) in which the different orders correspond to different meanings, it is assumed that ancienne is in the Spec of two different functional categories, and accordingly, that there are two different ancienne belonging to different semantic classes of adjectives. (11)

a. b.

église ancienne ancienne église

“church that is old” “former church”

These data immediately bring to mind three questions. (i) Why are bare adjectives obligatorily prenominal in English? (ii) Why can adjectives occur on either side of the noun in French? (iii) Why is there a meaning difference between the two orders, as in (11), and what is it? The answers provided by the generative approach are not very insightful: (i) a list of ordered positions places adjectives in that position; (ii) a list of features triggers this; (iii) ancienne is listed in two classes of adjectives which correlate with these positions. In contrast, by appealing to substantive rather than overly formalist properties, Exaptive Grammar produces informative analyses. First, consider what constitutes the semantic part of the entry of a common Noun. It is well known that it is a network of interacting elements: in the terms of the tradition of Montague Semantics, a common Noun consists in a characteristic function f, a time interval i, a possible world w, and a variable assignment function g. This network of elements determines the set of things that have the property of being a f in w at i, i.e., the extension of the Noun; the variable assignment function g determines the denotation of the expression. I showed in Bouchard (2002) that the semantics of an adjective make it apt to apply to the whole network of the noun, or any of its subparts. For instance, postnominal ancienne in (11a) applies to the whole network, but prenominal 3

For a discussion of other variations on this kind of analysis, see Bouchard (2002).

BEYOND DESCRIPTIVISM

47

ancienne in (11b) only applies to a subpart of the network of église, namely the time interval i, hence the interpretation that only the interval of application of the characteristic function is ancient, not the building denoted. This correlation between the order of the adjective and the noun, and what the adjective modifies, holds very generally. When the adjective is postnominal, it always modifies the noun as a whole: this follows directly from the setting of the Linearization Parameter in French, which is Head+Dependent. Under an Elsewhere interpretation of this setting, the reverse order—in which the adjective is prenominal—indicates that the adjective is not a “normal” dependent: a prenominal adjective is therefore interpreted not as modifying the whole network of the noun but only a subpart of it. By taking into account the substantive properties of the semantics of common nouns and the substantive properties on which the Linearization Parameter rests, it is possible to go beyond a mere listing of the observations which brought about questions (i)-(iii) above. Thus, in answer to (ii), there are two possible orders in French because the Linearization Parameter does not restrict elements to one order, but rather each order is a signifiant corresponding to a specific signifié. This immediately answers to (iii), and predicts what effects the two orders have on the interpretation of the adjective: we explain that the adjective applies to the whole noun or to a subpart of it, depending on its position. As for the answer to (i), adjectives are almost exclusively prenominal in English due to an independent property of its nominal constructions. As we saw, Number is expressed on N in English, whereas it is on Det in French. Consequently, in a French phrase like balle rouge, the adjective is in the sister position of the noun and it can apply to the whole network of the noun, as required by its semantics. On the other hand, the phrase *ball red is ill-formed in English because red is a sister of the complex N+NUM. However, the semantics of red require that it modify just the N, i.e., a subpart of the N+Number element. To do this, according to the Linearization Parameter, red must appear in a position other than the canonical post-head position of the dependent, i.e., it must appear in prenominal position. In sum, while the generative model leads to analyses which merely state the facts in a convenient arrangement, the exaptive model makes use of logically prior properties of our physiological and conceptual make-up to explain the distribution and interpretation of adjectives in French and English. 5.

Conclusion Human language crucially depends on signs, whereas animal calls are not signs but just reflex responses. The arbitrariness of signs and the variation

48

DENIS BOUCHARD

that derives from it are an artefact of the physiological and cognitive makeup of human beings, which provides no ‘reason of nature’ for any particular pairing of signifié and signifiant. This exaptive model provides a basis to explain the presence of recursion in human language: it derives entirely from the nature of the perceptual and conceptual substances linked by the signs. The model also explains why several novel traits appeared simultaneously in human language, whereas these seem to be unconnected in a generative model. Moreover, the exaptive model leads to a different way of conducting linguistic analysis. The results are promising, since devising conjectures which rely on the logically prior properties of the perceptual and conceptual substances leads to insightful analyses instead of ad hoc lists, as shown by a brief survey of five well-studied syntactic phenomena. References Borer, Hagit. 1984. Parametric Syntax. Dordrecht: Foris. Bouchard, Denis. 1995. The Semantics of Syntax. Chicago: The University of Chicago Press. ----------. 1996. “Sign Languages and Language Universals: The Status of Order and Position in Grammar”. Sign Language Studies 91.101-160. ----------. 2002. Adjectives, Number and Interfaces: Why Languages Vary. Oxford: Elsevier Science. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. ----------. 2000. “Minimalist inquiries: The framework”. Step by step: Essays on minimalist syntax in honor of Howard Lasnik, ed. by Roger Martin, David Michaels & Juan Uriagereka, 89-155. Cambridge, Mass. MIT Press. ----------. 2001. “Beyond Explanatory Adequacy”. MIT Occasional Papers in Linguistics 20.1-28. Cambridge, Mass.: MIT Linguistics Department. Cinque, Guglielmo. 1994. “On the Evidence for Partial N Movement in the Romance DP”. Paths Towards Universal Grammar, ed. by Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi & Raffaella Zanuttini, pp. 85-110. Georgetown: Georgetown University Press. ----------.1999. Adverbs and Functional Heads. Oxford: Oxford University Press. Curat, Hervé. 1999. Les déterminants dans la référence nominale et les conditions de leur absence. Geneva-Paris: Librairie Droz. Fitch, Tecumseh, Marc Hauser, & Noam Chomsky. 2005. “The evolution of the language faculty: Clarifications and implications”. Cognition 97:2.179-210.

BEYOND DESCRIPTIVISM

49

Fodor, Jerry, Edward Walker & Cornelia Parkes. 1980. “Against Definitions”. Cognition 8.263-367. Gould, Stephen Jay & Richard Lewontin. 1979. “The Spandrels of San Marco and the Panglossian Paradigm: A Critique of the Adaptationist Programme”. Proceedings of the Royal Society of London, Series B, 205:1161.581-598. Hauser, Marc, Noam Chomsky & Tecumseh Fitch. 2002. “The language faculty; what is it, who has it, and how did it evolve?” Science 298.15691579. ---------- & Elizabeth Spelke. In press. “Evolutionary and developmental foundations of human knowledge: A case study of mathematics”. The Cognitive Neurosciences III, ed. by Micahel Gazzanig. Cambridge, Mass.: MIT Press. Jackendoff, Ray. 2002. Foundations of Language. Oxford: Oxford University Press. Kayne, Richard. 1975. French Syntax: The Transformational Cycle. Cambridge, Mass.: MIT Press. ----------. 2000. Parameters and Universals. Oxford: Oxford University Press. Lamarche, Jacques. 1990. Tête-à-tête et autres relations: La position et l'interprétation des ADs. M.A. thesis, Université du Québec à Montréal. ----------. 1991. “Problems for N0-movement to Num-P”. Probus 3.215-236. Nowak, Martin, David Krakauer & Andreas Dress. 1999. “An Error Limit for the Evolution of Language”. Proceedings of the Royal Society of London B 266.2131–2136. Roberge, Yves. 1990. The Syntactic Recoverability of null arguments. Montreal: McGill-Queen’s University Press. Saussure, Ferdinand de. 1916 [1967]. Cours de linguistique générale, ed. by Tullio de Mauro. Paris: Payot. Scruton, Roger. 2005. “The Unobservable Mind”. Technology Review 108.72-76.

DO SUBJECTS HAVE A PLACE IN SPANISH? JOSÉ CAMACHO Rutgers University

0.

Introduction Overt preverbal subjects in Spanish (henceforth PS) have been claimed to be in an adjoined, peripheral layer of the clause (cf. Contreras 1991, Ordóñez and Treviño 1999) and in an IP-layer (cf. Goodall 1999, Suñer 2003). Part of the motivation for the different analyses stems from mixed A and A-bar properties they display. In this discussion, several independent questions are conflated: 1) Is there a uniquely determined position for PS in Spanish? 2) Is it uniquely associated with certain content (topichood, A/A-bar etc.)? 3) Are overt and null subjects in the same position? In this paper, I address these three questions, arguing that subjects can appear in several functional projections along the extended verbal projection (cf. Grimshaw 2005). In some cases, the properties of those projections will be determined in the course of the derivation: if a verb raises to a certain projection, it will render this projection active for agreement purposes and trigger movement of the subject to its specifier. 1.

The status of subjects in Spanish One prominent line of analysis of the position of PS in Spanish claims that they are, in essence, clitic left-dislocated (CLLD) constituents akin to CLLD subjects in Trentino and Fiorentino, or to direct and indirect objects in Spanish, illustrated in (1)a. In this line of analysis, the structure for (1)b would be (1)c, with the subject adjoined to IP (cf. Contreras 1991) or in a leftperipheral projection (following Rizzi 1997). The subject clitic, is either a null pro, as in(1)c, or the inflection morpheme, as proposed by Ordóñez and Treviño (1999) (henceforth, O&T). (1)

a. A la hermana de Susana, la vi el otro día to the sister of Susana, CL saw the other day “Susana's sister, I saw her the other day.”

52

JOSÉ CAMACHO

b. Susana compró unos regalos Susana bought some gifts “Susana bought some gifts.” c. Susanai [IP proi [I compró [VP unos regalos ]]] 1.1.

PS as CLLD constituents The CLLD analysis of PS rests on several arguments. First, subjects pattern like CLLD objects and indirect objects in VP-ellipsis (cf. O&T): both can appear when the elided VP is in an island, as in (2), (O&P’s (15), underlined text indicates the deleted part). (2)

A ti te admitirán en Harvard pero es probable que a to you CL will admit in Harvard but is probable that to tu amiga no [la admitirán en Harvard] your friend not CL will admit in Harvard “You’ll be admitted to Harvard, but it is likely that your friend won’t.”

Second, Alexiadou and Anagnostopoulou (1998) (henceforth A&A) argue that PS do not involve local, Spec, head agreement with the verb, since adverbs can intervene (cf. (3)a). They also argue that quantificational and indefinite PS have unambiguous scope, as CLLD elements do. Thus, in (3)b, from A&A, the indefinite subject can only have wide scope in Greek, as it does in Spanish. (3)

a. Pedro ayer, inesperadamente, conoció a su profesor Pedro yesterday, unexpectedly met to his professor “Yesterday, after many efforts, Pedro met his professor.” b. Kapios fititis stihiothetise kathe arthro some student filed every article “Some student filed every article.” c. Algún/un estudiante archivó todos los artículos some student filed every the articles “Some student filed every article.”

Finally, A & A argue that if an indefinite PS is in an A-bar position, it cannot reconstruct (like CLLD items), therefore, it should have scope over a modal or negation, assuming that modals do not raise covertly. This is shown in (4), after A&A's examples. In (4)a, the modal can only be deontic, not epistemic, and in (4)b, the only possible interpretation is many > not:

SUBJECTS IN SPANISH

(4)

53

a. Un niño debe presentar el trabajo hasta el final de la semana A boy must turn.in the work until the end of the week “A boy must turn in the work until the end of the week.” b. Muchos hombre no se enamoraron de María Many men not CL fell.in.love of Mary “Many men didn't fall in love with Mary.”

1.2.

PS as A-constituents Goodall (1999) observes that the properties of PS and CLLD phrases do not completely overlap. For example, topics can be bare, but PS cannot. If PS are CLLD phrases (and topics), this is surprising. However, the impossibility of having bare NP subjects must be independent of whether PS are topics, since even topic subjects cannot be bare (cf. (5)). (5)

*Niños, con children, with

Marta, no Marta, not

juegan nunca en el parque play never in the park

Zubizarreta (1993, 1998) points out that CLLD phrases reconstruct, whereas as Suñer (2003) suggests, PS do not, contrary to A&A’s claim. In (6)a (Suñer’s example), the DP (Lea) inside the PS can be coreferential with the clitic, whereas that same coindexing is impossible when the subject is postverbal. If the subject is in an A-position, leaving no trace when moved, it cannot reconstruct in (6)a, so there will no Principle C violation. In (6)b, on the other hand, the DP inside the subject is lower than the coindexed clitic, hence Principle C is violated. (6)

a. [El primer novio de Leai] lai besó en la calle the first boyfriend of Lea CL kissed in the street “Lea's first boyfriend kissed her in the street.” b. *Lai besó en la calle [el primer novio de Leai] CL kissed in the street the first boyfriend of Lea

Along the same lines, a postverbal subject can be coreferential with a DP inside a CLLD constituent (cf. (7)), but if the subject is null (by assumption, preverbal), coindexing with a CLLD item is impossible. If the CLLD phrase reconstructs, to a Clitic Phrase position lower than the preverbal subject but higher than the postverbal one (as seen in b), these contrasts follow.

54 (7)

JOSÉ CAMACHO

a. [La primera página de las novelas de Cortázari] siempre the first page of the novels of Cortazar always la escribía éli en ayunas CL wrote he before.breakfast “The first page of Cortazar's novels, he always wrote before breakfast.” b. [CP ti …[CLP [DP la …de Cortázari] lai [escribía éli …]]] c. *[La primera página de las novelas de Cortázari] proi siempre the first page of the novels of Cortazar always la escribía en ayunas CL wrote before.breakfast “The first page of Cortazar's novels, he always wrote before breakfast.” d. [CP ti …[IP proi [CLP [DP la …de Cortázari] lai [escribía éli …]]]]

These data suggest differences in distribution for PS and CLLD, and different positions for them (A and A-bar respectively). On the other hand, A&A attributed the lack of scope interaction between PS/CLLD items and lower quantifiers (cf. (4) above) to the fact that they can’t reconstruct, and assumed that this is a function of their A-bar status. Notice, however, that in their own account, CLLD items are base-generated in the left-periphery, hence have nowhere to reconstruct to, regardless of what their A/A-bar status is. Thus, it seems that lack of ambiguity for quantifiers cannot be taken as an argument for an A-bar position of PS/CLLD items. One alternative consistent with both Suñer and A&A's conclusions would be to argue that CLLD items are base-generated in a high Clitic Phrase, from which they can move to the left-periphery and to which they reconstruct. Since this position is higher than that of objects, and presumably higher than NegP, we predict the lack of ambiguity between quantified CLLDs and lower quantifiers. Finally, Suñer (2003) shows that lexical subjects and strong pronouns (él, ella “he, she”) can be separated from the verb by a parenthetical, whereas expletive ello “it” in Dominican Spanish cannot (cf. (8)a-b), suggesting that ello is structurally closer to the verb than strong pronouns and lexical NPs, as in (8)c, following Cardenaletti and Starke (1999). (8)

a. Juan/Él, a mi parecer, es muy simpático Juan/He, according to me is very nice “Juan/He, according to me, is very nice.”

SUBJECTS IN SPANISH

55

b. Ello (*a mi parecer) no sería malo estudiar it according to me, not would.be bad to study “It (according to me) wouldn't hurt to study.” c. [YP {Juan/él} [XPparenth. [ZP {ello} V ]]] To summarize existing data, PS pattern with CLLD items in deletion contexts and cannot take narrow scope with respect to other quantifiers, but their behavior diverges with respect to reconstruction. Finally, expletives seem to have a different position than strong pronouns (and lexical PS). 2. 2.1.

New data Ellipsis and modals Recall that one of O&T's arguments for treating PS as CLLD elements stems from their similar behavior as remnants in ellipsis contexts (cf. (2) above). However, with an epistemic modal, CLLD phrases and PS diverge in those contexts: with the PS in (9)a, the missing part’s interpretation cannot include the modal. With the CLLD phrase in (9)b, the epistemic modal can be part of the missing constituent. If the PS is a CLLD item, this is unexpected. (9)

a. Los médicos pueden haber salido, pero Pedro dice the doctors can have left, but Pedro says que las enfermeras no [han salido] that the nurses not have left “Doctors may have left, but Pedro says that nurses haven't (left).” “*Doctors may have left, but Pedro says nurses may not.” b. A los médicos les puede haber dado permiso, pero Pedro to the doctors CL can have given permission, but Pedro dice que a las enfermeras no [pueden haber dado permiso] says that to the nurses not can have given permission “Doctors may have been granted permission, but Pedro says that nurses may not (have been granted permission).”

Assuming the deletion target remains constant (IP) in both cases, these data suggest first, that the epistemic modal verb raises higher with an overt PS than with a null subject and second, that subjects and CLLD items do not occupy the same position (see below for further details). 2.2.

Epistemic modals and PS subjects Preverbal, non-quantified subjects are ungrammatical with a CLLD constituent and an epistemic modal (cf. (10)a), as opposed to postverbal subjects (cf. (10)b) and to PS without the CLLD constituent (cf. (11)(13)a).

JOSÉ CAMACHO

56

This constraint resembles the one observed by A&A (cf. (4) above), but in this case the subject need not be quantified. As seen in (11), without a CLLD item, there is no word order asymmetry. A summary of the data appears in (12). (10)

Background: Full-time doctors have little time to devote to each individual patient, and sometimes, patients suffer on account of this, but in this hospital, all patients seem to be very well tended to… a. #Sí, a los pacientes, los médicos residentes los deben yes, to the patients, the doctors resident CL must atender, por eso están tan bien attend, for that are so good “Yes, patients must be taken care of by resident doctors, that's why they are so well cared for.” b. Sí, a los pacientes los deben atender los médicos yes, to the patients, CL must attend the doctors residentes, por eso están tan bien resident, for that are so good “Yes, patients must be taken care of by resident doctors, that's why they are so well cared for.”

(11)

Background: Full-time doctors have little time to devote to each individual patient, and sometimes, patients suffer on account of this, but in this hospital, all patients seem to be very well tended to … a. Sí, aquí los médicos residentes deben atender a yes, here the doctors resident must attend to los pacientes, por eso están tan bien the patients, for that are so good “Yes, resident doctors must take care of patients here, that's why they are so well cared for.” b. Sí, aquí deben atender a los pacientes los médicos yes, here must attend to the patients the doctors residentes, por eso están tan bien resident, for that are so good “Yes, resident doctors must take care of patients here, that's why they are so well cared for.”

(12)

a.  b. *

CLLD-modal-PS CLLD-PS-modal

SUBJECTS IN SPANISH

c.  d. 

57

PS-modal Modal-PS

These data argue against a unified treatment for preverbal subjects and CLLD elements as occupying the same structural position, on the one hand, and, on the other, as uniformly treating PS as A-bar elements. 2.3.

Subject adjacency effects In addition to the adjacency effects seen in (8)b above, preverbal negative quantifier subjects must be adjacent to the verb (cf. (13)a-b), by contrast to lexical subjects, postverbal quantifiers, and CLLD items (cf. (14)ac)). (13)

(14)

a. Nadie pudo avanzar 3 metros no-one could advance 3 meters “No one could advance 3 meters.” b. *Nadie casi pudo avanzar 3 metros no-one almost could advance 3 meters c. *Ninguno de ellos casi pudo avanzar none of them almost could advance

3 metros 3 meters

a. La tortuga casi pudo avanzar 3 metros the turtle almost could advance 3 meters “The turtle could almost advance 3 meters.” b. Casi no pudo avanzar 3 metros nadie almost not could advance 3 meters no-one “No one could almost advance 3 meters.” c. A ninguno de los culpables, casi lo descubrieron to none of the guilty almost CL discovered “None of the guilty, were almost discovered.”

(14)b does not become grammatical with a restricted quantifier, unlike CLLD phrases (cf. (13)c vs. (14)c, and Cinque 1990 and Goodall 1999), and other adverbs, but not other quantifiers, show the same intervention effects (cf. (15)) vs. (16)). These contrasts suggest that negative quantifier subjects and lexical subjects occupy different positions ((13)b vs. (14)a)), and once again, that PS and CLLD phrases occupy a different position (cf. (13)c) vs. (14)c)).1 Finally, parentheticals do not induce ungrammaticality in the way adverbs do, 1

The CLLD phrase in (13)c is interpreted as focused, showing that CLLD phrases have no unique interpretation.

58

JOSÉ CAMACHO

as seen in the contrast between (13)b above and(17), suggesting that negative quantifiers and Dominican expletives have different structural positions. (15)

a. Juan constantemente viaja a Portugal Juan constantly travels to Portugal “Juan constantly travels to Portugal.” b. *Nadie constantemente viaja a Portugal No-one constantly travels to Portugal

(16)

a. Todos (ellos) casi pudieron avanzar 3 metros all them almost could advance 3 meters “All (of them) could advance 3 meters.” b. Algunos casi pudieron avanzar 3 metros some almost could advance 3 meters “Some could almost advance 3 meters.”

(17)

Ninguno, a mi parecer, pudo avanzar 3 metros no-one, according to me, could advance 3 meters “No one, according to me, could advance 3 meters.”

All of these data suggests different preverbal subject positions depending on subject-type of subject (cf. (8)), verb position (cf. modals in (12)), and on locality constraints (cf. Domincan ello in (8)b, negative quantifiers in (13)). 3.

Extended target positions for subjects The assumptions of the analysis are presented in (18), most notably, 2) a given functional projection may be inert until it is activated by adjoining a head to it or filling its specifier, and 3) certain projections may agree locally only in certain derivational contexts. (18)

3.1.

1. Agree can be local (local agree, in Spec, head configuration) or unbound (unbound agree). 2. A projection becomes active for agreement (agreement-active) by adjoining a head to it or filling its specifier. 3. An agreement-active projection may require local agree.

Adjacency effects The adverb intervention cases described above (cf. (13)-(14), partially repeated below) derive from (18), with one further assumption. The quantifiers involved in (13) are NPIs and as such, they must be licensed by negation in the

SUBJECTS IN SPANISH

59

specifier of a neg-related projection (Haegeman and Zanuttini’s Neg-criterion), which is then not overtly realized as a head (due to some version of the doublyfilled COMP filter). In the case of post-verbal NPIs, they are licensed longdistance by negation. Within the assumptions in (21) the Neg-criterion is a manifestation of agree, movement of preverbal NPIs to the Spec of the relevant negative projection will render this projection agreement-active, and will also require local agree. The contrast in (13) follows from the fact that casi blocks Spec, head (local) agreement with the quantifier, leaving the NPI’s negative feature unchecked, as in (19). Since the feature checked is negative, one expects other non-negative quantifiers not to show the same pattern, as shown in (16) above. (13)

a. Nadie pudo avanzar 3 metros no-one could advance 3 meters “No one could advance 3 meters.” b. *Nadie casi pudo avanzar 3 metros no-one almost could advance 3 meters

(19)

a.

nadieneg

NegP

b.

Neg’ Neg0+ pudo

nadieneg IP

avanzar 3 metros

XP NegP casi

Neg’

Neg0+ pudo

IP

Extending this analysis to the Dominican expletive cases in (8) above, which must be adjacent to the verb, would entail that weak pronouns and expletives are required to undergo local agree with the verb, a configuration that does not obtain when the parenthetical intervenes. However, the adjacency requirement for weak pronouns and expletives may not be syntactic but morpho-phonological: on the one hand, parentheticals do not block negagreement with NPIs (cf. (17) above). On the other, the nature of the parenthetical seems to suggest that it is not selected, but rather adjoined, raising questions as to why it would block a syntactic configuration. On the other hand, if the requirement is morphological (i.e., weak pronouns, as clitics, require a certain type of host), the structural configuration of the parenthetical will no longer be relevant, only the fact that it linearly intervenes between the clitic and its host. Thus, although these data have been argued to be evidence in favor of different positions for the subject, a view I support in this paper, it is very likely that they derive from a different type of constraint.

60

JOSÉ CAMACHO

3.2.

Modals Recall that modals cannot be interpreted in the deleted portion of an ellipsis when there is a preverbal subject, but they can when there is a CLLD phrase (cf. (9) above). I argued that the modal raises higher with a PS than it does with a CLLD phrase. If we assume that the missing material in the second conjunct is interpreted by copying structure from the first conjunct, and that the copied structure is IP, then, the result would be (20), where the copied portion is inside the dash-box. In this structure, the lexical PS must be higher than IP (an observation consistent with O&T’s conclusions), otherwise the deleted material would include the first subject. (20)

a. [ModP los medicos [Mod no pueden [IP [haber salido]] ]] &…[XP las enferemeras ]] b. [CLLDP a los medicos [ModP [IP [les puede]] ]] & …[XP a las enferemeras ]]

Within the assumptions of (18), this means that in (20)a, ModP is agreement-active, since the modal has raised to Mod0 and the PS to Spec, ModP. This triggers local agree with the lexical PS. Why doesn’t the CLLD phrase attract the modal in (9)b? One possible answer is that it is not in ModP, but higher, as depicted in (20)b, hence ModP is not agreement-active. If the subject and the CLLD phrase are in the same projection, there is no obvious explanation for the observed interpretive contrast. The second observation regarding modals was the contrast between ungrammatical CLLD-PS-modal order (cf. (10)a) and the grammatical CLLDmodal-subject order (cf. (10)b). PS-modal and CLLD-PS-non-modal orders are also grammatical (cf.(11)). The analysis just developed would yield the partial derivation in 0 for (10)a: the lexical PS appears in the specifier of ModP, forcing the modal to move to the Mod0. For the CLLD-modal order in (10)b, the CLLD phrase appears in CLLDP, higher than ModP, hence ModP is not rendered agreement-active and the modal does not raise to Mod0. If this is correct, the ungrammaticality of (10)a can be seen in a different light: something blocks the CLLD phrase from merging with the structure in 0. I will propose that in that structure, the CLLD phrase cannot merge with ModP because ModP changes by virtue of becoming agreement active. Specifically, an agreement-active ModP becomes selected (or L-related), and an L-related projection cannot merge with a non-L-related projection within the same strong phase. These two generalizations are formalized in (22)-(23).

SUBJECTS IN SPANISH

(21)

61

Partial merge for (10)a ModP

los médicos residentes

Mod’

los debeni

IP ti

atender …

(22)

An agreement-active projection can become L-related/selected if its head is a selecting/L-relating head.

(23)

Phase Consistency Condition (PCC) a. Projections within a strong phase must be L-consistent. b. Two projections are L-consistent if they are both L-related or both non-L-related.

The PCC predicts that arguments will stack-up in the same strong phase, and context-related projections (focus, topic, etc.) will only be possible at strongphase edges, as suggested by Rizzi (1997) for the left-periphery, and by Ndayiragije (1999) and Belletti (2004) for the VP-phase. (22) allows for a strong phase to be extended: if the strong phase is composed of L-related categories, movement from within it may extend L-relatedness beyond that strong phase. This extension mechanism can be extended to other cases, such as A-scrambling (which extends the binding properties of scrambled elements, cf. McGinnis 1999), or clitic-climbing in restructuring contexts (where the clitic can climb out of the lower phase only if the lower verb incorporates to the higher one, rendering it L-related). These two principles rule out further merger of the CLLD phrase with ModP in (20), as shown in (24). (24)

Impossible merge of CLLD in (10)a CLLDPnon-L

a los pacientes

ModPL los médicosL

Mod’

los debenL

IPL

For the case of the CLLD-modal order in (10)a, no such problem arises, as shown in and (25) below (bold font indicates L-related projections).

JOSÉ CAMACHO

62 (25)

Structure for (10)b CLLDPnon-L

a los pacientes

ModPnon-L IPL los debenL

VPL

When the verb is not modal (cf. (26)), the order CLLD-PS-V is possible. In such cases, the verb only raises to I, the left-peripheral projections are non-L-related, and a CLLD and a PS can merge in the periphery, as in (27). Note that the crucial difference between (24) and (27) is the presence of a dedicated ModP phrase. In both cases the PS merges in the left-periphery, but only in former case does it render the projection agreement-active and attracts the modal to it, blocking further merger of left-peripheral material. (26)

(27)

a. A Marta, los niños la saludan todos los días to Marta, the children CL greet every the days “Marta, children greet her everyday.” b. A los pacientes, los libros, se los traen sus familiares to the pacients, the books CL CL bring their relatives “Patients, books, their relatives bring them.” CLLDPnon-L a Marta

CLLDnon-L

los niños la saludanL

IPL todos los días

The typology of possible and impossible left-peripheral configurations is presented in (28): either a single, left-peripheral, L-related position, derivationally determined by the position of the verb, or a set of possibly recursive, non-L-related positions when no verb movement takes place. (28)

a. V in I, recursive (or non-recursive) left-peripheral positions [CLLDP(-L) DP(-L) [IP(L) PS(L) I ] ] b. *V in left periphery, recursive left-pheripheral positions *[CLLDP(-L) DP(-L) [XP(L) DP(L) [X’ V+X IPL ]]]

SUBJECTS IN SPANISH

63

c. V in left-periphery, non-recursive left-peripheral positions [XP(L) PS(L) [X’ V+X IP]] 3.3.

The role of person (29) is argued by O&T to show that the binding/referential properties of the subject are located in inflection, not in the DP. Descriptively, a 3rd person, plural subject los estudiantes “the students” can appear with 1st, 2nd or 3rd person plural inflection on the verb. In such cases, a clitic in an adjunct clause can only match the verb's inflection, not the DP's person settings (cf. (29)). (29)

Los estudiantes salimos /salisteis /salieron del edificio the students(3pl.) exited(1pl.)/exited(2pl.)/exited(3pl.) of.the building “We/you/they, the students, left from there.”

(30)

a. Los amigosi salimos después de que nosi acusara the friends(3pl.) exited(1pl.) after of that CL(1pl.) accused “We, the friends, left after s/he accused us.” b. *Los amigosi salimos después de que losi acusara the friends(3pl.) exited(1pl.) after of that CL(3pl.) accused

Note, however, that both controlled and non-controlled infinitival verbs that do not show overt person marking, display the same distributional pattern than that observed for (30): in (31), the clitic attached to the infinitival can be 1st or 3rd person, even though the controller los amigos “the friends” is 3rd person. If inflectional morphology always hosts coreferential properties, we would not expect the clitic to be able to be coreferential with los amigos in (31), much less showing variable 1st or 3rd person. The same patterns hold for non-control infinitivals, as seen in (32). (31)

a. Permitió a los amigos [presentar-se al examen] allowed to the friends attend-CL(3p) to.the exam “S/he allowed his/her friends to take the test.” b. Permitió a los amigos [presentar-nos al examen] allowed to the friends attend-CL(1pl.) to.the exam “S/he allowed us, his/her friends to take the test.”

(32)

a. Presentar-se los amigos al examen sería un error attend.CL(3) the friends to.the exam would.be a mistake “For the friends to take the exam would be a mistake.”

64

JOSÉ CAMACHO

b. Presentar-nos los amigos al examen sería un error attend.CL(1pl.) the students to.the exam would.be a mistake “(For us the students) to take the exam would be a mistake.” As an alternative to O&P’s analysis, I will build on Poletto’s (2000) functional structure in (33). She proposes that the subject in Italian crosses a number of functional projections from vP to TopicP, including SpeakerP, HearerP, NumP, NegP and TopP. (33)

[TopPSubj DPi [NegP[NumP ti [Numinfl Vj ] [HearerPti[Hearertj [SpeakerPtj [TP]]]]]]

The proposal for (31) is outlined in (34). In (34)a, the speaker-oriented projection is absent (by economy of structure), and the 3rd person PRO is in IP. Conceptually, the lack of a speaker-oriented projection follows from the fact that the infinitival subject in (31)a does not include the speaker. In (31)b, on the other hand, PRO is in the speaker-oriented projection, which binds the clitic. For finite-verbs (cf. (30)), the structure in (35)a has pro in the specifier of SpeakerP for 1st person coreference, for Peninsular Spanish 2nd person (cf. (35)b), the subject is in HearerP and for 3rd person (cf. (35)c), in IP. (34)

a. Permitió a los amigos [IPPRO(3) presentar-se(3) al examen] b. Permitió a los amigos [SpkrPPRO(1pl.) presentar-nos(1pl.) al examen]

(35)

a. [Los amigos] [SpkrP pro(1pl.) salimos(1pl.)] b. [Los amigos] [HearerP pro(2pl.) salisteis(2pl.)] c. [Los amigos] [IP pro(3pl.) salieron(3pl.)]

4.

Conclusions I have argued for a number of positions for subjects in Spanish, some left-peripheral (lexical subjects), some within IP (pro, expletives). For the former, I have argued that subjects do not occupy the same position as CLLD elements, and that the L or non-L-related nature of subject positions is derivationally determined by verb movement. The principles that derive their distributions include agreement-active, (a given functional projection may be inert for agreement until activated by movement of a head to it or filling of its specifier) and the Phase Consistency Condition (all categories within a phase are consistent with respect to being selected or non-selected).

SUBJECTS IN SPANISH

65

References Alexiadou, Artemis and Elena Anagnostopoulou. 1998. “Parametrizing AGR: Word Order, V-movement and EPP-checking”. Natural Language and Linguistic Theory 16. 491-539. Belletti, Adriana. 2004. Aspects of the Low IP Area. The Structure of IP and CP. The Cartography of Syntactic Structures V. 2, ed. by Luigi Rizzi, 1651. Oxford: Oxford University Press. Cardenaletti, Anna and Michal Starke. 1999. “The Typology of Structural Deficiency”. Clitics in the Languages of Europe (Empirical Approaches to Language Typology), ed. by Henk Van Riemsdijk, 145-233. Berlin: Mouton. Cinque, Guigliemo. 1990. Types of A-bar Dependencies. Cambridge, Mass.: MIT Press. Contreras, Heles. 1991. “On the Position of Subjects”. Perspectives on Phrase Structure: Heads and Licensing. Syntax and Semantics 25, ed. by Ssusan Rothstein, 63-79. San Diego: Academic Press. Goodall, Grant. 1999. “On Preverbal Subjects in Spanish”. Current Issues in Romance Languages, ed. by Teresa Satterfield et al., 95-110. Amsterdam: John Benjamins. Grimshaw, Jane. 2005. “Extended Projection”. Words and Structure, ed. by J. Grimshaw, 1-74. Stanford, Calif.: CSLI publications. Haegeman, Liliane and Raffaela Zanuttini. 1991. “Negative Heads and the Neg Criterion”. The Linguistic Review 8.233-251. McGinnis, Martha. 1999. “A-Scrambling Exists!”. Proceedings of the 23rd Annual Penn Linguistics Colloquium, ed. by Jim Alexander, Na-Rae Han and Michelle Minnick Fox, 283-297. Penn Working Papers in Linguistics. University of Pennsylvania. Ndayiragije, Juvenal. 1999 Checking Economy. Linguistic Inquiry 30.3, 399444. Ordóñez, Francisco and Estela Treviño. 1999. “Left Dislocated Subjects and the Pro-drop Parameter: A Case Study of Spanish”. Lingua 107.39-68. Poletto, Cecilia. 2000. The Higher Functional Field. Evidence from Northern Italian Dialects. Oxford: Oxford University Press. Rizzi, Luigi. 1997. “The Fine Structure of the Left Periphery”. Elements of Grammar, ed. by Liliane Haegeman, 281-337. Dordrecht: Kluwer Academic Publishers. Suñer, Margarita. 2003. “The Lexical Preverbal Subject in a Romance Null Subject Language: Where are Thou?” A Romance Perspective on Language Knowledge and Use. Selected Papers from the 31st Linguistic Symposiumon Romance Languages, ed. by Rafael Núñez-Cedeño et al., 341-358. Amsterdam: John Benjamins.

66

JOSÉ CAMACHO

Zubizarreta, María Luisa. 1993. “The grammatical representation of topic and focus: Implications for the structure of the clause” Cuadernos de Lingüística 2, 181-208. Instituto Universitario Ortega & Gasset, Madrid. ----------. 1998. Prosody, Focus, and Word Order. Cambridge, Mass.: MIT Press.

ON THE CONCEPTUAL ROLE OF NUMBER

VIVIANE DEPREZ Rutgers, ISC

0.

Introduction Bouchard (2002) proposes that the syntactic position of number (#)— on N or D—affects three aspects of nominal syntax: adjective-order, nounomission and Bare Argument Nominals (BAN). This paper shows that number position incorrectly predicts these nominal properties in some languages, i.e. the French-based Creoles, and argues that better results obtain if the conceptual role of number is taken into account. Number interacts with two distinct processes in language: Quantization and Individuation. Quantization, a ‘counting’ operation, is of little concern here. In Individuation, an operation that maps a nominal concept to the individual objects that materialize it, the role of number is to provide a criterion of division mapping given properties onto distinct instances. But Individuation, I argue, can also be achieved independently of number: contextual localization provides an alternative criterion of division. That is, just as objects can be individuated through their properties or their location in space (Xu and Carey 1996), so, I propose, can a kind be mapped onto its realizations through number division or spatial location. Languages opt for different modes of individuation and this choice has empirical consequences on nominal syntax. Number position, I show, matters only for languages with Number Individuation. Accordingly, not all languages that mark number primarily on D have equivalent properties and some are correctly predicted to freely allow BAN. The current paper has three parts. First Bouchard’s proposal, its empirical shortcomings, and the syntax of Haitian Creole (HC) number are discussed. Second, my proposal on the conceptual role of number is outlined. Third, my proposal is shown to predict the syntactic properties Bouchard noted and to remedy the shortcomings. I conclude with additional predictions for the syntax of HC-number.

68

VIVIANE DEPREZ

1. 1.1

Position of # marking Bouchard 2002 Bouchard distinguishes two classes of languages. D-number languages (D-#) mark number primarily on D, have mostly post-nominal adjectives, allow N-omission and exclude BAN as (1a-d) illustrates for French, a characteristic D-# language. N-number languages (N-#) display opposite properties: number is primarily on N, adjectives are largely pre-nominal, Nomission is precluded and BAN are free, as (1a’-d’) illustrates for English. (1)

a. b. c. d.

la/les voiture(s) une voiture rouge Donnes-moi la rouge *Je mange pommes

a’. the car/s b’. a red car c’. Give me the red (*one). d’. I eat apples.

Bouchard accounts for these properties as follows. He argues that intersective adjectives—the largest class—must modify the noun itself. In D-# languages with head-dependent order (French), post-nominal adjectives can modify the noun itself and be intersective predicting that the largest adjective class is post-nominal. In N-# languages (English), post-nominal adjectives modify the cluster [N+#], not the noun itself and can thus not be intersective. To modify the N itself, adjectives must use an Elsewhere-strategy that counters the regular head-dependent order. As a result, intersective adjectives are prenominal. But subsective adjectives that modify less than the N itself are predicted to be pre-nominal in both language types. Contrasts as seen in (2) where French pre-nominal adjectives are only subsective (2a’) while English ones are ambiguous (2a’or b’) are expected. (2)

a. Un vieil ami b. An old friend

a’. long time friend b’. a friend who is old

For Bouchard, number encodes referentiality with consequences on both Nomission and BAN. In D-# languages, nouns can be omitted and referentiality still be achieved through D. In N-# languages, this is not possible, hence Nomission is precluded. The distribution of BAN mirrors N-omission. Nouns in D-# French having no number cannot be referential by themselves, excluding BAN. The reverse is true for N-# English. 1.2

Predictions for Haitian Creole Consider now Bouchard’s predictions for languages like Haitian Creole. As (3) shows, HC nouns are invariable. Number is marked strictly on determiners. Definite DPs encode singular with la (a/an/lan), plural with yo.

69

ON THE CONCEPTURAL ROLE OF NUMBER

Indefinites can encode singular with yon, plural with numerals or quantifiers as shown in (3b). (3)

a. M achte liv la/yo “I bought the book.” b. M achte yon/twa/kek/anpil liv “I bought a/three/some/many book(s).”

HC is thus clearly a D-# language and, confirming Bouchard’s predictions, adjectives are predominantly post-nominal (4a), like French, and when prenominal, only allow a subsective reading (4b). N-omission is allowed (4c). (4)

a. yon liv chè “an expensive book” b. Vje zanmi nou “our long time friends/*our friends that are old” c. Ban mwen wouj la Give me red the “Give me the red one.”

In these respects, HC behaves as expected for a D-# language, but not so with BAN. While in D-# French BAN are excluded, they are, in contrast, freely allowed in HC. (5)

a. Moun koumanse ap pran baton “People started taking sticks.” b. Lè chat pa la, rat pran kay la “When the cat is not there, rats take over the house.” c. Elefan ap vin ra “Elephants are/the elephant is becoming rare.”

Existential Generic Kind

The distribution of HC BAN largely parallels that of an N-# language (Déprez 2005) raising a serious empirical challenge for Bouchard’s approach to number. Recall that for Bouchard, number is essential for referentiality, and it is this property that crucially licenses BAN. Since HC nouns do not carry number, they should fail to be referential arguments, contrary to facts.1 Should N-number hence be considered irrelevant for BAN licensing? I think not. That 1

The DP final position of HC determiners could not alone explain this distinction. First, HC determiners are not all DP-final. Some are DP-initial (yon, kek) with no discernable

70

VIVIANE DEPREZ

N-number matters for BAN is not a new idea (cf. Delfitto & Shroeten 1991, Longobardi 1994, Chierchia 1998) and convincing diachronic and crosslinguistic evidence strongly supports it. It is when N-number ceased to be pronounced that BAN became disallowed in French.2 And, while Walloon, like French both lacks N-number and BAN, other main Romance languages, both have N-number and BAN, a fairly clean-cut typological distinction. Nnumber is also demonstrably relevant for BAN even in N-# languages. Singular BAN are not possible in English, a fact naturally related to the lack of ‘overt’ singular morphology and also found in N-# Romance languages. In short, something does seem fundamentally right in the idea that N-number licenses BAN. If so, the interesting question is how French-like languages can be distinguished from HC-like languages while maintaining the idea that Nnumber matters for BAN. The suggestion advanced here is that number crucially plays a distinct conceptual role in the two language types. To articulate this view, a more in-depth study of HC number is needed, to which we now turn. 1.3

Number in HC Consider first this simple fact: in French and English #-marking is necessary to express plurality. Without a plural morpheme, count N cannot refer to more than one entity. Not so in HC. As (6) shows, HC count N can be plural even with no #-marking. Furthermore, #-marking is absent when a plural numeral co-occurs with a noun (6). Thus, #-marking appears optional in HC. (6)

Jan te achte (de) chwal “John bought (two) horse*(s).”

More specifically, HC count BAN are plural or singular without overt morphological differences, depending on contexts of use or lexical meanings as in (7). HC N manifests what Corbett (2000) refers to as ‘general number’. (7)

Jan achte liv/kay pou Pòl (Joseph 1989: 104) ”John bought books/a house for Paul.” That both options are possible is confirmed by HC pronominal reference. As (8) shows, HC BAN can be referred to with either the singular pronoun li, or the plural yo.

consequences. Second, if relevant at all, the position of determiners should also matter for adjective ordering or N-omission, not just for BAN, but this is plainly not the case. 2 These facts and correlations are presented in work in progress.

ON THE CONCEPTURAL ROLE OF NUMBER

(8)

71

Zwazo fè nich li/yo nan prentan “Bird(s) build its/their nest in the spring.”

Importantly, the fact that HC allows general number does not entail that HC N are default mass or lack a mass/count distinction.3 HC mass N, like English ones, are incompatible with plural, numerals and quantifiers like kek (some) restricted to count N. (9)

a. *dlo yo *“the waters” b. *twa /kek dlo *“three/some water”

HC mass N, contrasting with count N, only license singular, never plural, pronominal reference. (10)

Paske lò (se bagay ki) ra, li/*yo chè “Because gold is rare, it is expensive.”

Finally, HC mass N are incompatible with reciprocal predicates. As these facts suggest, minimal parts are accessible to syntax for HC count N, not mass N. (11)

a. Neg, se yon rayi lot “As for people, one hates the other.” b. *Diri, se yon sanble lot *“As for rice, one resembles the other.”

In contrast to French, HC lacks number agreement between N and adjectives or between subjects and verbs: (12)

Yon tifi/ Anpil tifi ap rele ou “One girl is/ Many girls are calling you.”

(13)

Mes anciens amis[me zansjE Zami] *est/sont gentils “My old friends *is/are nice.”

But HC nonetheless manifests an interesting #-agreement effect within nominals. In many languages, NP with plural numerals can be definite. For 3

See Déprez (2004) contra the predictions of Chierchia (1998) or Borer (2004).

72

VIVIANE DEPREZ

such NP, HC manifests a clear choice in how definiteness is marked: yo is required, la ungrammatical. (14)

Kat liv yo/*la four book the (PL)/*(Sing)

Given that four in (14) already signals plurality, la could have logically sufficed to mark definiteness. As its ungrammaticality demonstrates, agreement between a definite D and its NP is required in HC. Further effects of HC number are observed in contexts involving definites. Although indefinite BAN may remain unmarked for number, HC anaphoric definites must specify number. (15)

a. Gen twa liv ak yon magazin. Pran liv yo/*la. “There are three books and one magazine. Take the books.” b. Gwen liv ak magazin sou tab sa. Pran *liv /liv yo. “There are books and magazines on the table. Take the books.”

That is, number marking although optional for HC indefinites appears obligatory for HC definites. HC can thus be said to manifest a definiteness restriction on #-marking. Table 1 summarizes the properties of HC number as compared to those of French and English and shows that the HC number system is quite distinct. The rest of this paper explicates these contrasts.

# marking is optional # is general #Agreement outside DP #Agreement inside DP # Definiteness restriction Mass/count

HC yes yes no yes yes yes

FR no no yes yes no yes

Eng no no yes yes no yes

Table 1: Comparative properties of number marking

2. 2.1

Analysis Central ideas The key idea explored here is that languages can differ in the way they individuate kind terms. In some languages (i.e. French and English), individuation is done through number. In other languages (i.e. HC), individuation takes a distinct route that I term Spatial Individuation. As a consequence of these differing individuations, the role of number is distinct in

ON THE CONCEPTURAL ROLE OF NUMBER

73

the two sets of languages. This, I claim, has empirical consequences for NP syntax and the licensing of BAN. The proposed analysis draws on the following background hypotheses. I assume with Carlson (1977) that N denote names of kinds and with Geach (1962) that their central property is to provide a ‘criterion of identity’, i.e. a standard of sameness by which X can be judged to be the same as Y. Since discourse is most commonly about objects, Carlson proposed that a kind be ‘realized’, i.e. mapped to the set of possible objects that are its members. Borer (2004) argues that number plays a crucial role in this operation. Taking kinds to be undivided masses that must be broken into parts to map onto count nouns, she holds number to be the linguistic factor that operates this division. Reformulating this idea,4 I take number to provide a ‘criterion of individuation’ mapping a kind onto its realizations. The idea explored here is that number is not the only ‘criterion of individuation’ languages rely on: spatialization, i.e. location in discourse space/time, offers another viable option. Count N, I propose, are the syntactic association of a ‘criterion of identity’ (a standard of sameness, e.g. a set of features/properties) with a ‘criterion of individuation’ that serves to make minimal parts accessible to syntax. This ‘criterion of individuation’ is achieved either though morphological number or through contextual spatialization, an operation pairing individuals with unique location indices. These dual options are in my view the linguistic reflex of cognitive abilities at work in object recognition. In the object recognition literature, Xu and Carey (1996) drew attention to the difference between individuating objects on the basis of spatiotemporal information, ‘the WHERE path’, and individuating objects on the basis of property information (e.g. features) ‘the WHAT path’. Kadly and Leslie (2003) proposed the model below, which they describe as follows: Low-level vision provides the brain with information about edges, surfaces, reflectance and retinotopic coordinates. This information is further processed, and representations of features (shape, color, size, etc.) and locations in an allocentric (object-centered) reference frame are established. This information can be used merely to set up an object representation (OR) in working memory (WM) (individuation) or alternatively, it can be bound to the newly created representation and be available later to re-identify the object (identification). The two processes are independent, as are the two possible routes via features vs. locations. (Kadly & Leslie 2003).

4

In contrast to Borer (2004) and Chierchia (1998), Déprez (2004) does not assume that nouns are default mass. I take both mass and count to be distinct realizations of kind, which themselves are linguistically if not cognitively neutral with respect to this distinction.

74

VIVIANE DEPREZ

Object representation in WM

Features

Location

Individuation: signal establishes OR Identification: information is transferred into an already established OR

Low-level vision

Figure 1: Kadly and Leslie (2003) model of object recognition processing

Building on this view, I propose that there is a linguistic reflex of the WHAT and WHERE paths of object recognition that provides distinct ‘criteria of individuation’ for kind realization, i.e. distinct routes for mapping kinds onto individual instances. I take the linguistic reflex of the WHAT path to be associated with Number Individuation. The idea is that this type of individuation consists in replicating/copying the characteristic features of a kind several times, thereby creating distinct entities with the same features or properties. On this view, number signals the repeated binding of the feature bundle that constitutes a ‘criterion of sameness’ onto distinct individuals. In this sense, number individuation puts emphasis on property information. I take the linguistic WHERE path to be associated with Spatial Individuation and to merely signal the discourse presence of a new individual through the allocation of a unique space/time index (There is an x). Properties, i.e. the feature bundle that constitutes the relevant ‘criterion of identity’, may or may not be bound to this individual, partially or entirely. The point is that emphasis is here on location, not on properties. Section 3.2 offers an implementation of these ideas and details their linguistic consequences for the distribution of BAN. 2. 2

Implementation I propose that the WHAT and WHERE paths of linguistic individuation act as nominal aspects—i.e. Seins-arten—similar to verbal aspects or Aktions-

ON THE CONCEPTURAL ROLE OF NUMBER

75

arten. That is, individuation operations introduce ‘modes of presentation’ of a nominal concept, i.e. ways of representing nominal concepts for use by the syntax. Space is too limited here to detail my views on nominal aspect, but for concreteness, I assume that they are introduced in syntax by a little ‘n’ node similar to Chomsky’s proposed little ‘v’ node for verbal aspect. Little ‘n’ nodes come in distinct flavors, two of which are investigated here, the ‘WHATn’ flavor and the ‘WHEREn’ flavor. Each introduces a distinct individuation operation whose output is the realization of a kind into a count noun. Beginning with the WHAT path or Number Individuation, following Déprez (2004) I take WHATn to introduce a realization function R (cf. Carlson’s realization rule) augmented by a predicate requiring realizations to be minimal (Min) and a measure function. I follow here Krifka (1995) in assuming a measure function (OU = Object Unit), which, when applied to a kind and objects, returns a number. OU is a cardinality function that gives the number of objects x instantiating the kind in a given world. (16) presents the syntactic structure and semantic denotation of the WHATn node with an English gloss. (16)

a.

nP

n: WHAT n’

n

NP (kind)

b. [|n’WHAT|]: λwλx λn [Rw(x,K) & Min(x,K) & OUw(K)(x) = n] The x that are minimal realizations of a kind and that number in n An important point to note for our purpose here is that WHATn introduces a number variable. As dictated by Full Interpretation (FI), this number variable must be appropriately bound for the structure to be interpretable. Following Déprez (2004), I assume that the binding of this variable requires the presence of a morpheme or determiner with the correct interpretation. If such a morpheme or determiner is not merged in syntax, FI is violated, as the number variable remains unbound. Consequently, opting for the WHATn ‘mode’ of presentation has the result that number must necessarily be specified. It is in this sense that WHATn is understood as Number Individuation. To put it differently, WHATn introduces a call for a number value without which the resulting nominal expression cannot be interpreted. Without number specification, the nP in (16)—our proposed syntactic representation for count N—is uninterpretable. Turning now to WHEREn, the proposal is that it introduces the realization rule R, the Min predicate but instead of a measure

76

VIVIANE DEPREZ

function, it uses a localization function whose effect is to pair instances of a kind with a contextually determined spatiotemporal index: (17)

n’ WHERE c : λw λx [Rw(x,K’) & Min(x, K’)& Lc(x)] The set of x that are minimal realizations of a kind and have an individuating location. L is a function that associates x with a contextually specified location.

The important point for our purpose here is that WHEREn does not introduce a number variable. As a consequence, there is no requirement for number specification. Spatial Individuation does not require a morpheme to supply an appropriate binder for a variable. Such nP are thus directly interpretable. As Spatial Individuation relies on contextual index, the result can be singular or plural, depending on the state of the context: a noun individuated by Spatial Individuation remains underspecified for number. Finally, although strictly speaking, the WHEREnP of (17) and the WHATnP of (16) differ in their understanding and effects, they nevertheless share some properties. Both associate a ‘criterion of identification’ with a ‘criterion of individuation’ and both introduce a minimality predicate. This is sufficient to meet our definition of count N. Thus, both individuation types clearly produce count N, albeit by different routes. 2.3

Applications and extensions Consider now the consequences of the two modes of individuation for nominal syntax. Nominals with WHAT individuation require number specification to saturate the introduced number variable. This number specification can be provided in different ways, each with their own consequences. I examine here English and French count N showing that for these languages, the position of number actually matters. Consider English whose plural is an affix on N. I take the plural morpheme to head a NumP projection that provides both existential closure of the number variable and a requirement that it be distinct from 1. (18a) shows the structure of the bare plural ‘books’ and (18b) its semantic denotation. (18)

a. [NumP [Num s [nP … n:what [NP book]]]] b. λx ∃n[ R(x, book) & Min(book,x) & OU(book)(x)= n ≠1]

ON THE CONCEPTURAL ROLE OF NUMBER

77

When the x variable gets bound by an existential or generic operator,5 (19) correctly delivers the existential or generic readings of English BAN. Note here the crucial role of plural. Were this morpheme absent, the number variable would remain unbound and a violation of FI would ensue. This, I suggest, is exactly what happens with English singular BAN (*John reads book). As English has no singular morpheme, the number variable introduced by WHATn remains unbound. Such a count N is uninterpretable unless something is introduced in the derivation to provide saturation for the number variable. This role, I suggest is taken up by the indefinite determiner ‘a’ which provides the appropriate saturation by simultaneously giving the value 1 for the number (n) variable and existentially quantifying over the x variable. See (19): (19)

∃x [ R(BOOK,x) & Min(BOOK,x) & OU (BOOK) (x) = 1]

The generalization obtained about English is that the number variable can be saturated at the NumP level. For plurals, saturation is supplied by the presence of a morpheme in Numo to which the n+N head subsequently moves. For singulars, saturation is supplied by the determiner ‘a’. If the determiner is absent, a bare singular nP is uninterpretable and excluded by FI, as it contains an unbound variable. This analysis naturally explains the fact that English BAN must be plural. As for adjective modification, if adjectives never attach lower than Num0, Bouchard’s analysis can be recaptured straightforwardly. Since number is saturated and expressed at this level, English intersective adjectives have to resort to the Elsewhere strategy and be pre-nominal. Similarly, if what was previously assumed to be N-omission is now NumPomission, then in English the entire number chain is omitted, leaving only a numberless D to recapture the omitted nominal expression, a possibility excluded by FI. Turning to French, recall that there, plural morphology is traditionally regarded as too ‘weak’ to have any effect. My analysis allows a precise characterization of this ‘weakness’. Suppose the French plural simply lacks a semantic translation. Accordingly, it is unable to provide saturation for the number variable introduced by WHATn individuation. Thus French ‘morphologically plural’ BAN violate FI. Furthermore, since French lacks a singular morpheme, singular BAN are excluded for the same reasons. In sum, a simple and immediate consequence of the proposed analysis is that FrenchBAN are generally excluded by FI. Some determiner is always required to 5

For the purpose of this paper, which theory of bare noun phrases is ultimately chosen, i.e. the ambiguity approach or the neo-Carlsonnian kind approach, is irrelevant.

78

VIVIANE DEPREZ

saturate the number variable introduced by WHAT individuation deriving the often noted observation that French is a ‘determiner-obsessed’ language. Taking determiners to be merged into (or moved to) D, number saturation is determined at the D-level in French. On this view, Bouchard’s proposed analysis of adjective positioning and N-omission can be recast straightforwardly, taking Num, not N, as the relevant basis. At NumP in French, the number variable has not yet been saturated. Number has not yet been determined, with the consequences predicted by Bouchard following directly. Intersective adjectives can right adjoin to Num(P) and still only modify the N itself, number being yet undetermined. Similarly, NumPomission or lack of pronunciation does not matter since what ultimately determines number interpretation is overtly expressed at the top of the number chain, i.e. in D. Let us now turn to the WHERE path of individuation and discuss the results on HC. The central relevant distinction for our purpose here is that Spatial Individuation introduces no number variable that requires saturation. Individuation is established on the basis of contextual cues that furnish the basis for a localization index. As a result, a nominal expression like liv (“book(s)”) gets directly interpreted without the need of a number morpheme: (20)

a. [NumP [Num s [nP … n:where [NP liv]]]] b.[| nWHERE|]: λx [R(x,LIV’) & Min(x, LIV’)& L(x,i)]

Its interpretation, as singular or plural is contextually determined in ways that are yet to be fully understood. The point here is that BAN are correctly expected to be freely available in HC because FI is never violated; there is no number variable in need of saturation. Note furthermore that since number remains underspecified, there is no reason to believe that it ever would interfere with adjectival modification of the whole N. If so, post-nominal intersective adjectives are predicted to be possible. As for noun omission, expectedly, relevant cases call for a definite determiner in HC. The analysis Bouchard proposed for French extends to HC. The central consequence of my analysis is that BAN are now freely allowed in HC, a welcomed consequence that does not interfere with the importance N-number has for BAN elsewhere. Additionally, the fact that HC BAN are underspecified for number also naturally follows. There are yet further welcomed consequences I now briefly turn to. Since for WHERE individuation the ‘criterion of individuation’ is contextual, it follows that under context change, individuation is made anew. This is indeed what is observed in HC. (21a) features twice the same BAN, with each instance referring to distinct entities. As Spatial Individuation is made anew for each case, the individuated houses end up being distinct.

ON THE CONCEPTURAL ROLE OF NUMBER

(21)

79

a. Jan te achte kay. Mary te achte kay. “John bought a house. Mary bought a house.” b. Jan te achte kay ye. Kay la pat che. “John bought a house yesterday. The house was not expensive.”

To assert that the house under discussion is the same as the one previously mentioned, the use of a definite determiner la is required. Now recall that as noted in Section 3.3, definite noun phrases are always marked for number in HC. Although it is beyond the scope of this paper to offer a precise account of this definiteness restriction on #-marking, I briefly sketch here the original avenue that my proposed analysis allows. Recall that for Kadly and Leslie (2003), the WHERE path of object individuation entails the creation of an object representation (OR) but it does not guarantee the binding of characterizing features to a given OR. As a result, a spatially individuated OR could be specified for just ‘shape’ or ‘color’, or leave most of its features underspecified. Kadly and Leslie (2003) specifically distinguish ‘individuation’ (recognition) from ‘identification’ (tracking) with only the latter requiring that a given OR be associated with features uniquely characterizing it. For them, ‘individuation’ creates an OR while ‘identification’ represents the ability to assert that a given OR is the same as itself in a potentially changing environment. Correspondingly, I proposed that Spatial Individuation delivers a linguistic individual with a specific localizing index. It could be, however, that the particular features that characterize this individual are left partly underspecified. Suppose then that number serves as a linguistic pointer that signals the binding of characterizing features to a given individual. As in object recognition, linguistic feature binding through number may occur either for individuation or identification purpose, i.e. discourse tracking. This suggests that in languages with Number Individuation, the binding of features to given individuals is obtained at the individuation stage. In such languages then individuation and identification are simultaneous. In languages with Spatial Individuation, however, as context change induces individuation anew, commitment to properties, i.e. binding to features, occurs to allow re-identification of a same individual. For these languages, number signals re-identification not individuation. In short, property commitment or feature binding may arise at distinct stages: either when an entity is individuated (Number Individuation), or when an entity must be re-identified in a changing context. In this second situation, it is hardly surprising that number should be associated with definiteness or perhaps more generally with discourse-old entities. This, I suggest, is the source of the above noted ‘definiteness restriction’ on HC number. In languages like HC, number is used

80

VIVIANE DEPREZ

for re-identification, not for individuation, i.e. for tracking a same linguistic individual in changing contexts. It signals identity to a previously individuated entity. This view sheds light on both the definiteness restriction on number marking in HC and on its position. Since definiteness is marked with overt D in HC, number is marked there too. The syntactic position of HC number simply follows from the choice of Spatial Individuation. 3.

Conclusion Based on a comparative study of French, English and HC, this paper explores a novel approach to the role of number in language. Number has been argued to provide a ‘criterion of individuation’ that serves to map a kind to the individuals that realize it; it is not the only such criterion however. Spatial Individuation provides another option. Consequences in the use of these two different modes of individuation include distinctions in the licensing of BAN and more generally distinctions in why, how and when number marking arises. In French and English, number triggers individuation. Thus all NP must specify number and BAN are allowed only if a morpheme appropriately binds the number variable introduced by Number Individuation. Spatial Individuation imposes no such binding requirement on HC NP so BAN are freely available. In HC, number encodes re-identification in context-changing situations and need be specified only on discourse-old entities, i.e. on definites. I take these two modes of individuation to be the linguistic reflex of distinct cognitive routes in object recognition, the WHAT and the WHERE paths. If so, this paper suggests that number is the linguistic reflex of options instantiated elsewhere in human cognition and not directly linked to counting. References Borer, Hagit. 2004. Structuring Sense. In Name Only. Vol I. Oxford: Oxford University Press. Bouchard, Denis. 2002. Adjectives, Number and Interfaces: WhyLanguages Vary. Oxford: Elsevier Science. Carlson, Greg. 1977. “Reference to kinds in English”. Doctoral dissertation, University of Massachusetts-Amherst. Chierchia, Gennaro. 1998. “Reference to kinds across languages”. Natural Language Semantics 6 : 4. 339-400. Corbett, Greville. 2000. Number. Cambridge: Cambridge University Press. Delfitto, Denis. & J. Shroeten. 1991. “Bare plurals and the number affix in DP”. Probus 3:2. 15-185. Déprez, Viviane. 2005. “Morphological Number, Semantic Number and Bare Nouns”. Lingua. 115. 857-883

ON THE CONCEPTURAL ROLE OF NUMBER

81

Geach, Peter. 1962. Reference and generality. Ithaca, NY: Cornell University Press. Joseph, Frantz.L. 1989. “La determination nominale en créole haitien”. Thèse de doctorat 3ème cycle. Université de Paris VII. Kádly, Zsuza & Alan M. Leslie. 2003. “Identification of objects in 9-monthold infants: integrating ‘what’ and ‘where’ information”. Developmental Science 6:3. 360-373 Krifka, Manfred. 1995. “Common nouns: a contrastive analysis of Chinese and English”. The Generic Book. ed. by G. Carlson and F. Pelletier. Chicago: Chicago University Press. Longobardi, Guiseppe. 1994. “ Reference and proper nouns”. Linguistic Inquiry 25. 609-666. Xu, Fei., & Susan Carey. 1996. “Infants’ metaphysics: the case of numerical Identity”. Cognitive Psychology, 30. 111–153.

THE DIACHRONIC DEVELOPMENT OF A FRENCH INDEFINITE PRONOUN COMPARING CHACUN TO AUCUN

‫٭‬

AMANDA C. EDMONDS Indiana University

0.

Introduction In recent years, diachronic syntacticians have begun to consider single lexical items in great detail: Déprez and Martineau (2004) for aucun “none, not any”, Martineau and Déprez (2004) for rien “nothing”, Eguren and Sánchez (2004) for autre/otro “other”, and Longobardi (2001) for chez. In addition to providing insight into the development of single words, these studies have also revealed diachronic information on the general organization of the French DP. I contribute to this growing literature with a diachronic profile of the French universal quantifier chacun “each”. The study that I present draws on a new corpus of more than 4000 examples dating from the beginning of the 12th through the middle of the 20th century. These data allow for a detailed characterization of the diachronic development of chacun not available in previous literature (Section 1). In Section 2, the development of chacun is compared to Déprez and Martineau’s (hereafter D & M) diachronic discussion of the quantifier aucun. Although distributional similarities suggest that a unified analysis of these two lexical items might be possible, I demonstrate that the similarities are superficial and that there exists as yet no compelling evidence in favor of unification. Following this discussion, I advocate an alternative diachronic syntactic analysis of chacun (Section 3). 1.

Chacun: The 12th-20th centuries To construct the database for this project, three electronic collections of French texts — the Amsterdam Korpus, the ARTFL corpus, and the LFA ‫٭‬

I am grateful to Barbara Vance for the discussions, suggestions, and guidance that she generously offered and to Dorian Roehrs for his helpful comments on an earlier draft.

84

AMANDA C. EDMONDS

corpus — were searched for examples of cha(s)cun(e).1 Because few texts from the 14th and 15th centuries are found in these collections, two additional texts were included.2 Given the large number of tokens found for the eight centuries under study, not all instances could be examined. Instead, only tokens from the first quarter of each century (12th-20th) were included. Exceptions were made in two cases: (a) the periods of 1100-1150, 1151-1200, and 1500-1599 were examined, due to a paucity of tokens and (b) an additional sampling period was added (1326-1400) both because of low token counts and because the transition from Old French (OF) to Middle French (MidF) is generally dated around 1325. No tokens were excluded unless questions about appropriate coding could not be resolved. Because chronological date was the sole criterion in selecting tokens, the database is comprised of examples from several geographic regions and includes both lyric and prose forms. The tokens in the final database—half of which are lyric and half of which were taken from prose—were drawn from approximately 300 different texts. The use of chronological date as the sole criterion also resulted in many more masculine examples (3502) than feminine examples (655). 1.1.

Chacun: OF and MidF During OF and MidF, the constructions available to chacun show greater variety than is found today: Whereas Modern French possesses two lexemes to express the distributive notion of ‘each’ (chacun is the pronoun and chaque is the modifier), in OF and MidF, chacun fulfilled both roles. This multiplicity of function can be seen in (1). In (1a-b), chacun is shown as a pronoun and (1c) illustrates the use of chacun as a noun modifier. Note that modifier chacun can be employed simultaneously with other adjectives, as in (1c), but that these adjectives follow the universal quantifier. (1) a. Chascun aime mex soi qu’autrui: “Each person loves himself better than he loves others” 1180, Béroul. Roman de Tristan: LFA b. …chascune de pitié commence a larmoier… “Each girl begins to shed tears out of pity” 1350, Anonymous, Beaudouin de Sebourc: LFA

1

Searches included variant spellings such as ca(s)cun(e), che(s)cun(e), and so forth. Electronic versions of Joinville’s Vie de Saint Louis (1306) and de la Sale’s Jehan de Saintré (1456) were provided by Barbara Vance. 2

85

DIACHRONIC DEVELOPMENT OF CHACUN

c. chascuns bons hom e chascune bone femme qui sent l amor de nostre seignor “each good man and each good woman who feels the love of our lord” 1200-1275, sully2: Amsterdam Korpus Even though chacun fulfilled two different roles, the modifier examples were always less numerous: Noun modifier examples only account for between 10% and 40% of tokens. Proportionally, modifier examples are most numerous between 1300 and 1325, which is followed by a steady decrease in the MidF texts. The changes in amount of modifier examples cannot be clearly imputed to the existence of another modifier form: A search of all OF and MidF texts consulted yielded only four examples of the modern modifier chaque. The distribution of modifier and pronominal uses of chacun is presented in Table 1. Modifier Date OF 1100-1150 1151-1199 1200-1225 1300-1325 MidF 1326-1399 1456

n

%

n

Pronominal %

12/90 96/593 231/869 114/310

13.3% 16.2% 26.6% 36.8%

78/90 497/593 638/869 195/310

86.7% 83.8% 73.4% 62.9%

62/424 6/62

14.6% 9.7%

362/424 56/62

85.4% 90.3%

Table 1: Distribution of chacun between 1100 and 1456

In addition to its use as a pronoun and noun modifier, chacun can be preceded by the indefinite article un within the same DP: un(e) chacun(e) “one each”.3 Buridant (2000) termed this construction the ‘emphatic construction’, potentially because un supports the distributive reading of chacun, a possibility suggested by Roehrs (2004) for the similar German construction ein jeder “one each”. Examples of the emphatic construction are given in (2a-c). (2) a.

3

…mes uns chascuns le despira… “but each and every one4 will scorn him” 1300-1325, contre: Amsterdam Korpus

The two constituents of this emphatic construction always show full agreement for Case, which suggests that the construction is not a compound, as each member of a compound generally receives Case in its original position. Thanks to Dorian Roehrs for this insight. 4 All instances of the emphatic constructions will be translated as “each and every” for the sake of consistency.

86

AMANDA C. EDMONDS

b. il fut despolliez d une chascune “he was deprived of each and every one” OF, bern2: Amsterdam Korpus c. A un chascun couvent cent livres “To each and every convent, one hundred pounds 1378, Anonymous, Miracle de sainte Bautheuch, LFA In my database, there are 22 examples of this construction, including noun modifier (2c) and pronominal examples (2a-b) and examples in both genders (masculine: 2a and 2c; feminine: 2b). Although only three of the 22 examples are feminine, this number must be understood with respect to the overall relatively small number of feminine examples examined (145). Thus, collapsing across gender, the emphatic construction only appears infrequently during the OF and MidF periods, as it accounts for less than 1% of tokens examined (22/2348). 1.2.

Chacun: The 16th-18th centuries Although chacun of the 16th-18th centuries continues to be employed in the same constructions identified in 1.1, the distributional tendencies have changed and continue to change over the course of this time period. Modifier and pronominal examples are attested, although the latter are clearly dominant as the former never account for as much as 10% of all tokens. The distribution of chacun for this time period is given in Table 2. Modifier Date 1500-1599 1600-1625 1700-1725

n 5/63 24/400 1/400

% 7.9% 6% .3%

Pronominal n % 57/63 90.5% 375/400 93.8% 399/400 99.7%

Table 2: Distribution of chacun in the 16th-18th centuries

The original emphatic construction described in 1.1 continues to be attested during this time period and, in the 16th century, a second emphatic construction makes its first appearance in my database. This emphatic construction combines two universal quantifiers whose meanings are generally thought to be in conflict: tout chacun “all each”. An example of the new emphatic construction is given in (3).

DIACHRONIC DEVELOPMENT OF CHACUN

87

(3) …que la vie de tout chacun puisse estre à tout moment ravie… “that the life of each and every one can be taken away at any moment” 1579; Robert Garnier. Antigone: ARTFL Both of the emphatic constructions appear only in the masculine beginning in the 16th century. Because of the relatively small number of feminine examples included in my database, I verified this observation with a search of all texts of this time period in the ARTFL collection: No examples of the emphatic construction in the feminine were found, with only two exceptions.5 The asymmetry between OF and MidF (feminine examples of the emphatic construction are attested, see example 2b) and the 16th-18th centuries (feminine examples are no longer attested, even when examples outside my database are considered) will be important in my analysis. Finally, whereas the emphatic construction accounted for less than 1% of all examples during OF and MidF, in the three centuries that follow, these constructions are more frequent and account for a little over 6% of all examples examined (54/862). 1.3.

Chacun: Modern French (19th-20th centuries) Modern French witnesses the specialization of chacun as a pronoun, and chaque — which begins to come into its own only in the 16th century (Pope 1934) — is established as the modifier counterpart. Thus, no examples of modifier chacun are found after 1800. However, even as its syntactic functions are restricted, the variety of emphatic constructions available to chacun increases: Between 1800 and 1925, un chacun, tout chacun, and the newcomer tout un chacun “all one each” (an apparent hybrid still attested today) are found. As was the case in the previous time period, these constructions are only attested in the masculine. However, unlike the distribution noted for the 16th-18th centuries, in Modern French these constructions are less common than previously: In my database, they account for only .25% of the tokens examined (2/800), although a search of all of the texts dated in this time period on ARTFL revealed examples of each type of construction. An example of the newest addition to the collection of emphatic constructions is given in (4). (4)

5

…révélé à tout un chacun par l’écartement d’une croisée… “revealed to each and every one by the gap of a window” 1900, Courteline. Gendarme. Art. 330. Boubouroche: ARTFL

One from Rabelais (1534), who is known for his archaisms and creativity with language, the second from Charron (1601). Finally, Palsgrave’s grammar (1530) mentions the possibility of une chacune.

88

AMANDA C. EDMONDS

1.4.

Summary During OF and MidF, chacun was both a pronoun and a noun modifier, with modifier examples accounting for approximately 24% of tokens during OF. In the centuries that followed, examples of modifier chacun decreased: In MidF, only 14% of tokens are modifier examples, and this percentage drops to 3.5% during the 16th-18th centuries. In addition to distributional shifts, changes in the emphatic construction (in terms of agreement possibilities) as well as in the variety of these constructions available are apparent. More specifically, although OF and MidF un chacun is attested in the feminine, this possibility is gradually lost, leaving only the masculine version. Finally, data from the centuries following the end of MidF reveal the introduction of two additional emphatic constructions, both of which include two universal quantifiers. These tendencies are summarized in Table 3.

Time period OF

Modifier n %

Pronominal n %

453/1862

24.3%

1408/1862

75.6%

MidF

68/486

14%

410/486

86%

16th-18th centuries Modern French

30/862

3.5%

830/862

96.3%

0/800

0%

800/800

100%

Emphatic Description Un chacun (Pro and Mod), full agreement Un chacun (Pro and Mod), full agreement Un chacun (Pro), invariable Tout chacun (Pro), invariable Un chacun (Pro), invariable Tout chacun (Pro), invariable Tout un chacun (Pro), invariable

Table 3: Diachronic distribution of chacun

2.

Aucun: A microparametric variation analysis In their 2004 study, D & M offer a microparametric variation analysis of the diachronic development of aucun.6 Although perhaps not evident given their modern roles, parallels in both the past functions and forms of chacun and aucun hint at the possibility of a unified diachronic analysis of these two lexical items. First, both served simultaneously as pronoun and modifier during OF and MidF, only to later specialize. Second, as noted by D & M, these two 6

‘Microparametric variation’ is based on the idea of parameters: Whereas changes in major parameters will affect far-reaching and often superficially dissimilar aspects of a grammar, changes in the setting of small parameters may only affect single or a small group of lexical items. Major parameters are thought to be resistant to change, whereas minor parameters are susceptible to it. Microparametric variation generally refers to changes in small parameters.

DIACHRONIC DEVELOPMENT OF CHACUN

89

modern quantifiers (universal and negative, respectively) are superficially similar in terms of morphosyntax: The indefinite article un can be located at the end of both lexical items (but see Grévisse 1993 for differences in their histories). Given these similarities, this section will review the history of aucun and the analysis proposed by D & M with a view to determining whether or not a similar analysis can be fruitfully extended to the chacun data. 2.1.

Aucun: OF and MidF In their discussion of the diachronic development of aucun, D & M (2004) cite distributional and semantic changes in favor of their analysis. These authors find that aucun was both a pronoun and a noun modifier, and that it occurred in the plural and the singular during OF and MidF. They also note that aucun could appear preceded by the definite article (les aucuns “the any”), which resembles the emphatic construction un chacun in its composition (although not in its meaning). Finally, the authors indicate that modifier aucun can appear either pre- or postnominally. In terms of semantics, aucun of OF and MidF is a positive quantifier with a positive meaning, even in negative contexts (i.e., it takes scope over negation). Given these distributional and semantic clues, the authors propose that aucun is found high in the DP (either in D0 or Spec) during OF and MidF: [DP aucun [NP ∅ ]]. 2.2.

Aucun: The 16th-18th centuries The three centuries following the end of MidF witness important changes in the distribution and semantics of aucun. With regard to its distribution, aucun is used predominantly as a singular modifier, can still be preceded by the definite article, and its use as a postnominal modifier increases (postnominally, aucun is indisputably in a modifier position). Turning to the semantics, the authors find that aucun is used more in polarity and negative contexts, to the detriment of positive ones: The meaning of aucun becomes indeterminate and context-dependent (i.e., negation takes scope over aucun). D & M (2004) explain these changes by lowering aucun to an adjectival position within the NP. This lowering correlates with the increase in modifier examples, the increase in postnominal modification, and provides an explanation for the indeterminate interpretation of aucun. More specifically, they propose that a null element that must be licensed for interpretation is located in the upper layers of the DP: [DP ∅ [NP aucun ]]. 2.3.

Aucun: Modern French D & M (2004) note that plural and pronominal uses almost entirely disappear, leaving only singular determiner examples during the 19th and 20th

90

AMANDA C. EDMONDS

centuries. The semantics of aucun have also changed, and this lexical item is now a negative quantifier. (Examples of aucun as a polarity item are almost entirely restricted to uses with sans “without.”) The authors contend that these changes are best accounted for syntactically if aucun has raised from a modifier position to the head of DP: [DP aucun [NP ∅ ]]. Although the authors acknowledge that this analysis leaves questions unanswered (e.g., What roles do number and negation play in this change?), the proposed analysis is generally supported by diachronic evidence, both distributional and semantic. In what follows, I question whether the diachronic history of chacun supports a similar microparametric variation analysis. 2.4.

Does chacun lower? As outlined previously, D & M’s (2004) analysis entails first a lowering of aucun (from high in the DP to an adjectival position within the NP) followed by a raising of aucun (from the adjectival position to D0). To determine whether evidence exists for a similar change in the history of chacun, I will consider the first half of this analysis: the lowering. I chose to examine the possibility of lowering because of chacun’s status as a universal quantifier: Following both literature on universal quantifiers (Giusti 1991; Zamparelli 2000) and observations of my data, I assume that chacun is located high in the DP. Thus, if chacun shows a development similar to the one proposed for aucun, evidence of lowering in its history will be found. On the basis of semantic and distributional evidence, D & M (2004) argue that aucun lowered to an adjectival position during the 16th-18th centuries. In terms of semantics, the authors interpreted the use of aucun in polarity and negative contexts as indicative of the existence of a null element in D0, forcing aucun to be low in the NP. Additionally, three pieces of distributional evidence were cited in support of a lowered, adjectival aucun: (a) modifier uses increased, (b) postnominal modifier examples increased, and (c) the co-occurrence of the definite article and aucun was possible. (However, note that postnominal modification as well as aucun’s co-occurrence with the definite article were both attested during OF and MidF as well.) These same potential sources of evidence—semantic and distributional—will be explored with regard to the diachronic data for chacun. As concerns the potential diachronic changes in the semantics of chacun, I have not yet undertaken a formal examination of all of the tokens in my database. Although I am therefore not able to comment on diachronic semantic changes, an examination of Modern French chacun is revealing. Modern French appears to have two different chacuns at its disposal: invariable chacun, which refers to all people without distinction, only occurs in

DIACHRONIC DEVELOPMENT OF CHACUN

91

the masculine, and is perhaps more accurately rendered in English as “everyone,” and pronominal referential chacun, which can refer to individuals and can be used in the masculine or in the feminine, as dictated by the referent. This distinction will be referred to in the final analysis. In assessing the possibility of the lowering of chacun, distributional evidence is telling and I start by considering the data from OF and MidF. Beginning with the proportion of modifier examples, the highest level was reached at the beginning of the 14th century, when modifier examples accounted for nearly 40% of tokens. Thus, if an increase in modifier examples is a correlate to the lowering of a lexical item to an adjectival position, then it would follow that chacun may have lowered around 1300. However, it must be pointed out that the value of this piece of evidence on its own is questionable. First, although the percentage of modifier examples is highest at this period, modifier chacun fails to account for the majority of tokens, which is perhaps unexpected if chacun is in an adjectival position. Second, I have purposely used the general term noun modifier to refer to examples in which chacun precedes a NP within the same DP, without advocating whether the universal quantifier is in an adjectival position (Spec, NP) or in a determiner or quantifier position (higher in the DP). The vast majority of modifier chacun examples are ambiguous between chacun as a determiner or quantifier and chacun as an adjective (e.g., chacun jour “each day”).7 As such, even the noun modifier examples do not provide clear evidence of the lowering of chacun to an adjectival position. D & M’s (2004) second distributional indicator of lowering can be disregarded with respect to chacun, as this quantifier never occurs as a postnominal modifier in my database. The final piece of evidence to consider is the co-occurrence of chacun and the indefinite article. Just as the linear ordering of les aucuns led D & M to suggest that les occupied D0, forcing aucuns to be placed low in the NP, the superficial ordering of un chacun seems to necessitate a similar analysis. Although this construction is consistent with an analysis in which chacun has lowered to an adjectival position of a NP that is either null or lexically filled (in either case, phi features would be specified in N0), other analyses are possible (see 3.1). Thus, evidence for lowering of chacun during OF and MidF is comprised of an elevated proportion of modifier examples (which cannot be verified as adjectival) and the existence of emphatic constructions that appear infrequently during this time period (accounting for less than 1% of tokens). As such, I do not believe that the evidence supports an analysis that places chacun low during OF and MidF. 7

Chacun can license an argument NP. According to Eguren and Sánchez (2004), this licensing capability implies that chacun was, or at least could be, a determiner.

92

AMANDA C. EDMONDS

During the 16th-18th centuries, the distributional data seem to lead to conflicting conclusions: The decrease in modifier examples argues against the lowering of chacun, whereas the increase in emphatic constructions seems to argue for it. I propose that the increase in emphatic constructions spuriously implies the lowering of chacun, and that this apparent contradiction can be resolved with reference to the emphatic constructions’ agreement patterns. As noted in 1.4, all emphatic constructions from the 16th century through Modern French are masculine, whereas examples in both genders are found in OF and MidF. If the increase in emphatic constructions during the 16th-18th centuries is an indication that chacun is in an adjectival position, the lack of feminine emphatic construction examples is unexpected given the agreement relation that should obtain: Chacun in an adjectival position within a NP — be it null or lexical — should agree with the phi features specified in N0 and, logically, both genders should be possible. Thus, the agreement facts provide indirect evidence that chacun is not in an adjectival position, and I find no proof of lowering to the NP during the 12th-20th centuries. In the following Section, I propose an alternative account of the diachronic data for chacun. 3.

An alternative account for chacun In 2.4, I maintained that chacun and aucun follow different diachronic paths insofar as no compelling evidence for the lowering of chacun to the NP was found. Although the two lexical items follow separate diachronic developments, the syntactic analysis of chacun that I propose in this section entails a microparametric variation analysis. For this syntactic analysis, I assume an agreement relationship that follows from Panagiotidis (2003a, 2003b), who proposed that pronouns are not intransitive. As a result, pronouns select for a null NP and, importantly, the pronoun shows agreement with the phi features specified on the null NP. 3.1.

Analysis: OF and MidF As argued in 2.4, I do not find proof that chacun is located within the NP during OF and MidF. However, the existence of the emphatic construction — in both genders and as both noun modifier and pronoun — implies that chacun is not located as high as D0. This suggests that the universal quantifier is found in an intermediate position in the DP, as illustrated in (5). At present, I have simply labeled this position Functional Projection (FP), although it may be possible to conceive of this FP in terms of the Word Marker Projection of Bernstein (1993).

DIACHRONIC DEVELOPMENT OF CHACUN

(5)

93

DP D1

Spec D | (uni)

FP Spec F | chacuni

F1 NP | N1 | ei jouri

Chacun is located in the same intermediate position as a noun modifier (N0 is lexically filled), as a pronoun (N0 is null), and when part of the emphatic construction (the indefinite article is in D0). Because this analysis proposes that a NP — either null or lexically filled — is always projected, it follows that noun modifier, pronominal, and emphatic construction examples will be attested in both genders. The data support this expectation. 3.2.

Analysis: 16th century to Modern French I propose that the single syntactic structure in (5) evolved into two distinct structures around the 16th century. This evolution from a single structure into two different structures results in the differentiation of invariable uses of chacun (including the emphatic constructions) from pronominal referential uses. Building on Panagiotidis’ notion of agreement, I suggest that modern invariable chacun continues to be generated in the intermediate position of OF and MidF. This modern construction differs from (5) in two important ways: First, the FP no longer selects a NP, allowing only invariable uses and, second, a higher layer QP is projected, providing an explanation for the linearization tout (un) chacun. The structure is given in (6). (6) [QP (tout) [DP (un) [FP chacun ]]] In my account, pronominal referential chacun is distinct from (6) insofar as a NP continues to be selected (explaining the possibility of chacun and chacune) and, like other strong quantifiers, chacun is now found in Q0. (This proposal suggests a difference in the quantificational properties of

94

AMANDA C. EDMONDS

invariable and referential chacun, a possibility that I have not yet examined). This structure is presented in (7). (7) [QP chacun [DP […[NP ]]]] 4.

Conclusions To begin this chapter, I detailed the diachronic development of the French indefinite pronoun chacun. I then examined an existing diachronic analysis for a different French lexical item and determined that it could not be justifiably extended to the data for chacun. Finally, I presented a diachronic syntactic analysis of chacun that assumed one underlying structure during OF and MidF that developed into two separate structures after the 16th century. Despite the advantages of this analysis, certain tendencies — such as the increase in emphatic constructions during the 16th-18th centuries — are left unaccounted for. Additionally, this analysis makes predictions that need to be tested. For instance, it is assumed that all examples of chacun during OF and MidF selected a NP. This assumption implies that invariable uses of chacun, as defined syntactically in 3.2, were not attested before the 16th century. Finally, although both chacun and aucun are quantifiers in Modern French, the presence versus the absence of negation clearly distinguishes these two lexical items; thus, this comparative study points to the possible crucial role of negation in diachronic changes. In addition to these tendencies and predictions deserving of further attention, this project points to several areas that might be explored and subsequently brought to bear both on the development of chacun and on diachronic changes in the French DP. Specifically, this chapter would be complemented by similar studies on the developments of other universal quantifiers in the history of French (e.g., tout “all”) as well as on the development of the modern adjective chaque “each”. References Bernstein, Judy. 1993. “The syntactic role of word markers in null nominal constructions”. Probus 5.5-38. Buridant, Claude. 2000. Grammaire nouvelle de l’ancien français. Paris: SEDES/HER. Déprez, Viviane, & France Martineau. 2004. “Micro-parametric variation and negative concord”. Contemporary approaches to Romance linguistics, ed. by Julie Auger, J. Clancy Clements, & Barbara Vance, 139-158. Amsterdam: John Benjamins. Eguren, Luis, & Cristina Sánchez. 2004. “Contrast and addition in Romance: A case study in microvariation”. Contemporary approaches to Romance

DIACHRONIC DEVELOPMENT OF CHACUN

95

linguistics, ed. by Julie Auger, J. Clancy Clements, & Barbara Vance, 159176. Amsterdam: John Benjamins. Grévisse, Maurice. 1993. André Goosse (Ed.) Le Bon Usage (13th ed.). Paris: Duculot. Giusti, Guiliana. 1991. “The categorical status of quantified nominals”. Linguistische Berichte 138.438-454. Longobardi, Guiseppe. 2001. “Formal syntax, diachronic minimalism, and etymology: The history of French chez”. Linguistic Inquiry 32.275-301. Martineau, France, & Viviane Déprez. 2004. “Pas Aucun/Pas Rien en français classique: Variation dialectale et historique”. Langue française 143.33-47. Palsgrave, John. 1530. Lesclaircissement de la langue francoyse. Menston: Facsimile (Reprinted by Scolar Press, 1969) Panagiotidis, Phoevos. 2003a. “Empty nouns”. Natural Language & Linguistic Theory 21.381-432. ----------. 2003b. “One, empty nouns, and (theta)-assignment”. Linguistic Inquiry 34.281-292. Pope, M. K. 1934. From Latin to Modern French with especial consideration of Anglo-Norman. Manchester: Manchester University Press. Roehrs, Dorian. 2004. “German ein jeder as a ‘late’ compound”. Ms, Indiana University. Zamparelli, Roberto. 2000. Layers in the determiner phrase. New York: Garland Publishing. Data Sources Amsterdam Literary Corpus, established by Anthonij Dees and associates, Free University Amsterdam, and made available to the University of Stuttgart by Piet van Reenen. Tagged version made available by Achim Stein, University of Stuttgart. The Project for American and French Research on the Treasury of the French Language (ARTFL), the University of Chicago http://humanities.uchicago.edu/orgs/ARTFL/. Laboratoire de Français Ancien (LFA), the University of Ottawa. http://www.uottawa.ca/academic/arts/lfa/.

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

FRANCA FERRARI-BRIDGERS Princeton University, Department of French and Italian

0.

Theoretical Framework and Proposal In the spirit of recent minimalist works on word formation processes (see Josefsson 1997, Marantz 1997, Lacarme 2001, Julien 2002, among others) I assume that Merge and Move are the primary operations involved in the derivation of Italian de-verbal nouns. Specifically, I argue that de-verbal noun formation is a syntactic process that consists of the merger of a nominal feature [n] with either a verbal stem, an AspectP or a VoiceP. As illustrated in (1), [n] is morphologically represented either by a derivational morpheme or by a null nominal head [henceforth NØ]. 1 (1) [nP [n (NØ/derivational morpheme)[XP (v-stem/AspP/VoiceP)]]] Because more than one aspectual projection is involved in the derivation of deverbal nouns, I adopt Cinque’s IP-representation (1999:106) as adequate underlying structure. However, given the incompatibility of the feature Tense with de-verbal noun formations (see Alexiadou (2001) and Ferrari-Bridgers (2005a & b), the projections relevant for my analysis are those located below Cinque’s TP-anterior and printed in boldface in (2).

1

It goes beyond the scope of this paper to show how Move is implemented in the derivation of de-verbal nouns. For a step-by-step derivation of Italian de-verbal nouns see Ferrari-Bridgers (2005b), where, following Kayne (1994, 2005), Koopman and Szabolcsi (2000) and Roeper (1999) it is assumed that Move is leftward, phrasal and not necessarily feature-driven. Following Kayne’s constraints on movement (2005:331), in fact, movement can be “blind” to “categorial (and other) features” and “what is moved where is entirely determined by what is merged (in a given derivation) and in what order.” In other words, movement can also be “closeness driven.”

98

FRANCA FERRARI-BRIDGERS

(2) Mood speech act >…XP…> aspect repetitive (I)> aspect frequentative (I)>… XP… >T-anterior > XP > aspect continuative> aspect perfect (?)> XP >aspect durative> aspect generic/progressive> XP > aspect completive (I)> (well) Voice P> (fast/early) aspect celerative (II)> aspect repetitive (II)> aspect frequentative (II)> aspect completive (II). In table 1, I list the most productive Italian de-verbal nouns with their proposed place of syntactic formation. Place of formation Aspect Perfect (?) Aspect Continuative Aspect Generic Voice P Aspect Repetitive (II) Below VP (verbal stem)

De-verbal noun type Feminine Past Participle Nominalizations Infinitive Nominalizations VN compound and nouns in –tore/mento/zione/tura Masculine Past Participle Nominalizations Nouns in –ìo Zero Derived nominals Table 1: De-verbal Nouns

As the analysis of each de-verbal noun type will show in section (2), the main goal of this paper is to demonstrate that semantic and morphological differences across the large variety of Italian de-verbal nouns are reducible to syntactic differences, i.e., the meaning of each de-verbal noun and the morphological properties of its verbal base determine the nature of XP 2.

De-verbal nouns formed on Aspect Generic In table (2) I provide a complete sample of Aspect Generic derived nominals divided according to their verbal theme, i.e., verbal stem+thematic vowel. In Italian there are three thematic vowels, e.g., [-a,-e,-i], each of them representing one of the three verbal conjugations, whose infinitive endings are respectively -are, -ere, -ire. Verbal theme VN compound -tore(trice) -tura -mento -zione

2

Trit-a-(re) Vend-e-(re) Serv-i-(re) trit-a-carne vend-i-fumo apr-i-bottiglia “meatgrinder” “smoke seller” “bottle opener” mangi-a-tore “eater” vend-i-tore “seller” serv-i-tore “servant” spacc-a-tura “split” fend-i-tura (cleft) vest-i-tura “dressing” pag-a-mento sparg-i-mento invest-i-mento “payment” “spreading” “investment” vener-a-zione“worship” perd-i-zione “ruin” eboll-i-zione “boiling” 2 Table 2: De-verbal nouns formed on Aspect Generic

As the anonymous reviewer indicates derived nouns such as scrittore/scrivere (“writer/to write”), produttore/produrre (“manufacturer/to produce”), etc. lack GA morphology. These

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

99 99

All the nouns showed in table (2) have an identical verbal base that is characterized by a verbal stem followed by the vocalic segments [-a,-i,-i]. In section (2.1), I argue that [-a,-i,-i] are aspect generic morphemes and that the generic nature of VPs lacking tense and agreement features justifies the merger of [n] in the IP. 2.1.

Aspect (habitual) Generic I will first analyze VN-compound nouns and de-verbal nouns in -tore. The analysis of de-verbal nouns in –mento/-zione/-tura will be given in section (2.2). I argue that the hypothesis that [-a,-i,-i] are aspect generic morphemes is motivated semantically, syntactically and morphologically. The proposed initial underlying structure for this de-verbal noun type is given in (3). (3) [nP [n [Asp GenericP [XP… [VP]]]]]3 Semantically, as demonstrated in Ferrari-Bridgers (2005b), the assumption that [-a,-i,-i] are aspect generic morphemes accounts (i) for the fact that in VN compounds4 and in nouns in -tore the verb has a habitual generic and not an episodic reading and (ii) for the bare object condition of the complement of these two de-verbal nouns. With regards to the generic reading, I argue that VN compounds and nouns in -tore are types of generic expressions showing many similarities with sentences marked by the presence of a (habitual) generic aspect. According to Dahl (1985: 95-97), (habitual) generic aspect is used in generic sentences to describe “the typical or characteristic properties of a kind” and according to Cinque (1999:99) “a Generic Aspect sentence seems to refer to some inherent characteristic (of an Object) that may not yet have had realization.”

nouns are considered exceptions, because they are derived from a close set of verbs with irregular paradigms. 3 The structure in (3) represents the initial underlying structure of a deverbal noun; its final output is obtained through movement as showed below for the VN compound apri scatole (“can opener”). Merger of GA with VP: [GAP [GA–i [VP [XP scatole] [VP [V apr-]]]]] Move VP to spec of GAP: [GAP [VP apr-]i [GA-i][VP [XP scatole][ti]]] Merger of NØ: [NP [NØ [GAP [VP apr-]i[GA i][VP [XP scatole][ti]]]]] (Ferrari-Bridgers 2005:97) 4

Prati (1931) and Tollemache (1945) observed the habitual reading of VN compound nouns in Italian.

100

FRANCA FERRARI-BRIDGERS

Carlson et al.’s (1995:7) genericity test, i.e., the paraphrases of VN compound nouns and nouns in -tore into generic expressions with characterizing constructions such as is used to and the adverb usually, reveals the generic nature of these nouns. As the examples in (4) and (5) suggest, VN compound nouns and nouns in –tore describe the typical properties of a kind (X). (4) a. apriscatole open cans “can-opener” b. X usualmente apre scatole X “usually opens cans” c. X si usa per aprire scatole X “is used to open cans” (5) a. bevitore (di birra) drinker (of beer) “beer-drinker” b. X usualmente beve birra X “usually drinks beer “ c. X é solito bere birra X “is used to drink beer” (Ferrari-Bridgers 2005a: 98) In addition, as Carlson et al. (1995:7) suggest for English, agentive nouns such as ‘pipe smoker’ have a generic meaning. These nouns describe an agent actually or potentially performing a habitual action. Normally, if they are used as predicates in a sentence, the sentence acquires a generic meaning. Similarly, many Italian VN compound nouns and nouns in –tore correspond to English agentive nouns e.g., fumatore di pipa “pipe smoker” and portalettere “mail carrier”. With regards to the bare condition of the object, the examples in (6) and (7) show that the object of VN compound nouns and nouns in –tore must be bare. For nouns in –tore, however, the contrast occurs in an out-of-the blue context and with an object that is not further modified.

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

101 101

(6) apri (*le/delle) scatole5 “open (*the/some)cans” (7)

il venditore di (*dei) giornali the seller of (*the) newspapers “the newspaper seller”

According to Dahl (1985, 1975: 108) verbs in (habitual) generic aspect can select for generic noun phrases, i.e., a NP that quantifies “over possible objects rather than over actual ones.” According to Chierchia (1998), in Italian bare nouns can be generic NPs, and for Longobardi (2001: 353), “bare nouns are generic objects only when the predicate selecting them has a habitual and not episodic reading.” For Longobardi, the habitual aspect feature is one of the operators providing the characterizing environment necessary for the licensing of bare nouns. It follows that the presence of [-a,-i,-i] as expression of habitual aspect becomes necessary for the licensing of bare generic objects in compound nouns and derived nouns. Morphologically, following Ferrari-Bridgers (2005b), I claim that habitual generic aspect morphology has double derivational and inflectional nature. Such a nature justifies the syntactic merger of nominal feature [n] with Aspect GenericP. According to Bergstand (1994:329-330), in the Eskimo language, Aleut, habitual generic aspect morphology appears in the general, i.e., a finite form of the verb that is tenseless and moodless. Bergstand claims that “being a zero tense, and a zero mood, the general may also be nominal depending on the syntactic construction.” Similarly, it is possible to assume that, in Italian, the presence of (habitual) generic aspect morphology indicates a zero tense and zero mood verbal form, whose derivation depends on the syntactic environment. This predicts that a verb marked by a habitual generic aspect can acquire (i) a nominal reading, if it merges with a nominal element, or (ii) can have a verbal reading, if it merges with other inflectional verbal features. This prediction is borne out. The derivational nature of the habitual generic aspect is confirmed by the fact verbal bases marked for the generic aspect morpheme serve as bases for all the derived nominals given in table (1); whereas, the inflectional nature 5

The anonymous reviewer notes that in French there are nominal compounds with an internal determiner, e.g., coupe-(la)-soif, trompe-la-mort. According to Kampers-Manhe (PC), this type of compounds is unproductive in French. Like Italian compounds, productive French VN compound nouns do not have an internal determiner, e.g., *ouvre la boite “bottle opener”.

102

FRANCA FERRARI-BRIDGERS

of habitual generic aspect is confirmed by the formal similarity of aspect generic forms with the 2nd person imperatives [trita, vendi, apri], which, according to Zanuttini (1997:114), are “bare verbal forms with zero tense and mood.” This suggests that a verb in a Aspect Generic form can acquire a verbal reading, for instance an imperative one, if it is in a verbal environment, i.e., if the habitual generic aspect form can be embedded under a null imperative mood head (Kayne PC). The imperative force is structurally determined by the position the bare verbal form occupies into CP. But it can also acquire a nominal reading if it merges with a nominal head. 2.2.

De-verbal nouns in mento/zione/tura and nominalized infinitives Syntactically, I argue that de-verbal nouns in -mento/-zione/-tura are built also on Aspect GenericP. However, to account for their semantic properties, I propose that their verbal structure contains an additional Aspect Completive Projection as represented in (8). (8) [nP [n [Asp Generic P [Asp Completive [VP]]]]] In order to fully understand the complexity of the structure in (8), it is necessary to set a morphological and semantic comparison with the derivation of nominalized infinitives. I assume that infinitive formation and its nominalization occurs at a higher aspectual projection than Aspect GenericP. My assumption is based on semantic and morphological differences between infinitive nominalizations and nouns in -mento/-zione/-tura. I argue that the derivation of nominalized infinitives is due to the merger of a null nominal head, i.e., [NØ] with Aspect ContinuativeP, as represented in (9). (9) [nP [NØ [Asp ContinuativeP [VP]]]]. Morphologically, the verbal base of nominalized infinitives differs from the verbal bases of nouns in -mento/-zione/-tura. Nominalized infinitives, in fact, are built on a verbal theme, e.g., verb stem + thematic vowel [-a,-e,-i], and not on an Aspect Generic base, as showed in table (2) above. Semantically, the choice of Aspect Continuative for nominalized infinitives is motivated by the following observations. Firstly, according to Gaeta (2002:121) and to Bartsch (1986:17-8) nominalized infinitives refer to temporally unbound situations representing a process/event without focusing on its completion. Nominalized forms of the infinitive denote a process without representing the act of that process as necessarily completed as demonstrated by Gaeta’s (2002:122-123) examples in (10).

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

103 103

(10) a. il gocciolare del lavandino dura già da due ore “the dripping of the sink has been going on for the past two hours” b. *il gocciolare del lavandino è teminato “the dripping of the sink has ended” c. lo sciogliere del ghiaccio da parte del Fohn dura già da molti giorni “the melting of the ice by the Fohn has been going for several days” d. *lo sciogliere del ghiaccio da parte del fohn è stato completato “the melting of the ice by the Fohn has been completed” The examples show that nominalized infinitives are compatible with predicates and temporal expressions that do not focus on the final completion of the action (10a-10c), but they are incompatible with predicate or temporal expression indicating the completion/end of the action/event (10b-10d). Syntactically, looking at the array of AspPs in Cinque’s IP-CP, it is interesting to note that the semantic values associated with nominalized infinitives correspond to the semantic values of Aspect Continuative that according to Cinque (1999:95) refers to actions extending their course over a period of time. Interestingly, in the IP Aspect Continuative is syntactically higher than Aspect Generic. This implies that infinitives, whose bases are verbal themes, are formed and nominalized in a different position than Aspect Generic and it correctly predicts that verbal themes cannot be the bases for the derived nominals of table (2). 2.2.1. Nouns in -mento /-zione/-tura. According to Gaeta (2002:123), contrary to nominalized infinitives, de-verbal in –mento/-zione/-tura semantically represent a situation arrived to its final point. As the contrastive examples between de-verbal in -mento and nominalized infinitive in (11) show, nouns in -mento/-zione/-tura impose a closure effect to the action/event expressed by the predicate. (11) a. l’affondamento della nave si compì in mezz’ora “the sinking of the ship was achieved in half of an hour” b. *l’affondare della nave si compi in mezz’ora *“the sinking of the ship was achieved in half of an hour” c. l’insegnamento del latino é stato completato “the teaching of Latin has been completed” d. *l’insegnare del latino è stato completato “the teaching of Latin has been completed” (Gaeta 2002:123) I attribute the semantic difference between nominalized infinitives and de-verbal in -mento/-zione/-tura to their different aspectual structures. I assume

104

FRANCA FERRARI-BRIDGERS

that the ‘closure effect’ expressed by these nouns is due to presence of an Aspect Completive feature in their underlying structure. Looking at Cinque’s arrays of aspectual projections in (2), syntactically, Aspect CompletiveP is located below Aspect GenericP. One can infer that the closure effect is related to the movement of VP through Aspect CompletiveP before landing to Aspect GenericP to complete the morphological derivation. 6 The presence of Aspect Generic morphology on these nouns, however, is still motivated by the following observations. Firstly, semantically some nouns in mento, zione and tura do not show any ‘closure effect’, but simply are agentive/instrumental nouns like VN compound nouns and nouns in –tore, e.g., the agentive noun accompagnamento “accompaniment” or the instrumental noun nutrimento “nourishment” with the meaning of “the person/thing that V” (Gaeta 2002:218). Secondly, Aspect Generic is semantically neutral with respect to a closure effect, therefore, its presence does not deny the existence of an additional aspectual projection in the underlying structure of these nouns. 2.3.

De-verbal nouns formed on VoiceP De-verbal nouns formed on VoiceP correspond to feminine and masculine past participle nominalizations illustrated respectively in table (3). As the data show, the morphological structure of past participle nominalization consists of a verbal stem followed by past participial morphology [-at,-ut,-it]. Infinitives Feminine Masculine

-ARE -ERE -IRE nuot-at-a bev-ut-a cuc-it-a single act of swimming single act of drinking single act of sewing spacc-at-o dov-ut-o what is due cuc-it-o what is broken what is sewn (ext. vertical section) (ext. the art of sewing) Table 3: Feminine and Masculine past participle nominalizations

Semantically, morphologically and syntactically nouns built on a past participle base can be divided in two sub-classes on the base of their gender. I argue that both sub-classes contain a VoiceP. However, in order to account for the semantic differences between the two noun types, I propose that masculine past participles are formed by the direct merger of a nominal feature [n] with VoiceP, whereas for feminine past participles the merger of [n] occurs in an higher section of the IP as represented in (12).

6

The aspect feature Completive is not overtly marked in these nouns, but its presence is assumed necessary to understand their meaning and the selection of their specific n-heads.

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

105 105

(12) a. [nP [n [VoiceP [VP]]]] b. [nP [n [Aspect XP [VoiceP [VP]]]]] 2.3.1. Feminine past participle nominalizations. Semantically, according to Mayo et al. (1995:912), these nouns cannot be interpreted as types of actions, but rather as individual instantiations of events and according to Gaeta (2002:154), their meaning is “the single act of V”, as indicated by the contrast in (13). (13) a. Il nuoto rilassa i muscoli “Swimming relaxes the muscles.” b. *La nuotata rilassa i muscoli (Gaeta 2002:154) “The (single) act of swimming relaxes the muscles.“ According to Gaeta (2002:156), semantically, the feminine nominalization of past participles has the effect of bounding predicates imposing a closure on the action/event, even onto predicates that are not necessarily unbound such as ‘nuotare’. Such a closure effect, however, differ from the closure effect seen for nouns in -mento/-zione/-tura. In (14), the comparison between de-verbal nouns in –tura and feminine past participle nominalizations reveals their semantic differences. (14) a. La sgrassatura della pentole é durata tre ore *La sgrassata delle pentole é durata tre ore “The removing of the grease from the pans lasted three hours.” b. Ho dato una bella sgrassata alla pentole *Ho dato una bella sgrassatura alle pentole “I gave a nice (single) act of removing of the grease from the pans.“ The examples in (14) show that, although both types of de-verbal impose a closure to the predicate, de-verbals in –tura mark a process that takes place in a X period of time, whereas feminine past participles indicate a one-time event that is not compatible with the idea of a duration. Syntactically, such a semantic difference suggests that both nominals are generated under different projections. As seen above, de-verbal nouns in – tura have a complex aspectual structure comprising Aspect CompletiveP and Aspect GenericP; whereas following Cinque (1999), Italian past participles are generated under VoiceP. Nevertheless, a comparison between feminine and masculine past participle nominalizations reveals that VoiceP alone is not sufficiently representative for the derivation of these feminine nouns.

106

FRANCA FERRARI-BRIDGERS

2.3.2. Masculine Past Participle Nominalizations. The existence of epicene masculine nouns such as il pulito, il cucito with corresponding feminine nouns such as pulita, cucita, is relevant for understanding the different semantics of the two past participle nominalizations. Whereas feminine nouns indicate a process, there is no aspectual meaning for masculine nouns. Masculine past participles are passivization of an event, i.e., they can be paraphrased as “what/who is/or has been V-ed”, as indicated in the semantic contrast between these pairs of nouns, e.g., il pulito “what is cleaned” vs. una pulita “a single act of cleaning”. Syntactically, following Cinque’s (1999: 101-103) all past participles of active and passive verbs are initially generated under VoiceP and then in case of active participle, the verb raises to PerfectP to realize its perfect feature. The adverb indicating the different positions for the different past participle types is bene “well”. Bene is assumed to be found in the spec of Voice P. It follows that the active past participle must be found on the left of bene, whereas bene is found at the left of a passive participle as the examples in (15) indicate. (15) a. Hanno accolto bene il suo spettacolo solo loro Have-aux received well his show only they “Only they have received well his show.” (Cinque 1999:101-103) b. *Hanno bene accolto il suo spettacolo solo loro Have-aux well received his show only they 7 c. Questo genere di spettacoli é sempre stato bene accolto da tutti “This kind of shows are always been received well from everybody.” In addition, Cinque (1999:46) suggests that the active past participle needs to move to the left of the head hosting ‘tutto’(all), syntactically related to the Aspect Completive (I) P, but it must be found to the right of the adverb rapidamente “rapidl”) located under Aspect Celerative (I)P, as indicated by the following examples. (16) a. Hai trovato tutto? (Cinque 1999:100-101) “Have you found all?” b. Gianni ha di nuovo rapidamente cambiato opinione (Cinque 1999:93) “Gianni has again rapidly changed opinion.” 7

The anonymous reviewer points out that, contrary to Italian, in French (15a) is out and (15b) is perfect. According to Cinque (1999:ft.71, p.211), the differences between French and Italian are due to the fact that in French active past participle does not move in the overt syntax; whereas it does it in Italian.

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

107 107

From the above examples, one can conclude that active and passive past participle do not have the same landing site and that active past participles are located in between Aspect Celerative(I) and Aspect Completive (I). Similarly to the derivation of verbal past participles, two IP’s sites are also necessary to explain the semantic differences between masculine and feminine past participle nominalizations. The purely passive nature of masculine past participles and the lack of aspectual content lead to the conclusion that these nouns are formed via merger of the n-feature with VoiceP. Contrarily, feminine past participles, though generated under VoiceP, need to move to a higher AspP to add their aspectual meaning. Under which specific aspectual projection feminine past participle nouns are found is not clear at the moment, but, like active past participles, their landing site cannot be above Aspect Celerative (I), as demonstrated by the following examples where the adjective rapida “rapid/quick” is used instead of the adverb rapidamente “rapidly/quickly”. (17) a. Dai una rapida stirata ai pantalon8, 9 “Give a rapid/quick ironing to the pants.” b. *dai una stirata rapida ai pantaloni. c. Fammi fare una rapida scappata al supermercato. Let me do a rapid/quick run to.the supermarket. “Let me go quickly to the supermarket.” d. *Fammi fara una scappata rapida al supermecato. From the above data one can conclude that feminine past participle nominalizations occur above VoiceP and possibly under an aspectual projection found below Aspect Celerative (I). 2.4.

De-verbal in [ì] I assume that De-verbal nouns in [-ì-], e.g., ronz-ì-o “continuous buzzing” are formed below VoiceP, more precisely under Aspect 8

For Cinzia Russi (PC) the sentences in 17a/b are both fine, especially considering rapido synonymous of veloce (“fast”). According to Cinque (1999:93/103), in Italian the adverb rapidamente can appear in two different positions, one higher associated with Aspect Celerative (I) that quantifies over an event and one lower associated with Aspect Celerative (II) that apparently quantifies over a process. In my Italian, rapida quantifying only over an event; whereas veloce “fast” quantifies over a process, e.g., dai una stirata veloce ai pantaloni. But, interestingly, in Italian, feminine past participle nominalizations cannot be found on the right of veloce, e.g., *dai una veloce stirata ai pantaloni’. 9 As the anonymous reviewer indicates, contrary to Italian, in French (17a-c) are ungrammatical, whereas (17b-d) are grammatical. Similarly to what said in ft.8, I attribute these differences to the lack of overt movement for French past participles.

108

FRANCA FERRARI-BRIDGERS

Repetitive/Frequentative (II)P. Syntactically, the stressed vowel [-ì] is an aspectual head, as indicated morphologically by the fact that in the plural the aspectual marker [-ì-] does not fuse with the plural morpheme [-i], e.g. (sg.) ronz-ì-o > (pl) ronz-ì-i, and semantically by the contrastive examples with deverbal nouns built on the same verbal base, e.g., gorgolio “gurgling”– gorgoglìo “constant gurgling”, scricchiolo “creaking”–scricchiolìo “constant creaking”. Because aspect projections select a VP and not a simple verbal stem, I derive that the verbal base of these derived nominals must be a VP. Given the aspectual nature of the morpheme [ì], I propose that the derivation of de-verbal in [ìo] consists in the merger of a nominal feature [n], expressed by the gender masculine marker ‘o’, with Aspect Repetitive/Frequentative (II)P as illustrated in (18).10 (18) [nP[n [Aspect Repetitive/Frequentative (II)P [-ì [VP]]]]] The repetitive aspectual nature of the de-verbal [ìo] is also suggested by Castelli’s (1988:336) and Gaeta’s (2002:114-116) observation that the marker [ìo] is incompatible with expressions of time that do not mark duration, as indicated by the following contrastive examples. (19) a. Il dondolìo della culla é durato a lungo “The rocking of the cradle lasted a long time.” b. *Il dondolìo della culla é stato improvviso “The rocking of the cradle was sudden.” (Gaeta 2002:114) In (19b), the impossibility of combining the semantics of the predicate (essere improvviso “to be sudden”) with the semantics of de-verbal nouns in [ìo] is taken as an evidence that de-verbal nouns reveal information about how an action designated by the verb is performed. The above contrast is also 10

As the anonymous reviewer points out in Italian there exists also the ‘repetitive’ prefix ‘ri’. With the exception of few lexicalized forms, e.g., ri-sciaquìo (“re-swish”), the repetitive morphemes ri and ì do not co-occur because they are semantically incompatible. According to Di Sciullo (2005:85), in fact, ri may only modify events that have an end point, i.e., accomplishments and achievements; whereas as seen above the morpheme ì modifies a process indicating an unbound activity. Syntactically, in Cinque’s IP (1999:92), there are two Repetitive AspectP, i.e., Repetitive (I) located above T(anterior) and used to quantify over events and Repetitive (II) located below TP and used to quantify over processes. Given the different natures of the two repetitive heads, it is possible to assume that ri is associated with Repetitive Aspect I; whereas ì is associated with Repetitive Aspect II. Because events can reoccur, ri can be stacked; but stacking of ì is not possible given that process modification cannot be recursive.

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

109 109

interesting because it reveals the special semantic values of the aspectual morpheme [-ì-]. Specifically, it indicates an action that is performed continuously and progressively over an unbound period of time. The attention is focalized on the process rather than its end and the action is defined as open and unbound. This explains the incompatibility noticed by Gaeta (2002:115) of de-verbal nouns in [ío] with adverbials of time of the type in X time and for X time that indicate that an action has been (or will be) concluded or accomplished in a specific time frame as in the following examples. (20) a. *Il tintinnìo delle campane si completò in due ore “The tinkling of the bells was completed in two hours.” b. Il tintinnìo delle campane durò tutta la mattina “The tinkling of the bells lasted all morning.” Looking at Cinque’s IP/CP representation and it is plausible to conclude that [ì] is an aspectual marker that semantically corresponds to the Aspect Repetitive (II). This aspectual feature is characterized by the repetitive adverb again and it is located below TP anterior. 2.5.

De-verbal nouns built on a verbal stem In Italian not all types of de-verbal nouns are formed on an aspectual base. Several verbs form the corresponding de-verbal nouns by merging the verbal stem with a nominal feature [n] that is morphologically expressed either by a masculine or by a feminine gender marker, e.g., urlare “to scream” > deverbal: urlo “the scream”, danzare “to dance” > de-verbal: danza “the dance”. In (21) I propose the underlying structure for these nouns. (21) [np [n [√url + (v)]]] The assumption that these nouns are built on a lexical stem rather than on a VP is semantically, morphologically and syntactically motivated. Semantically, these nouns refer solely to the action/event/state implicit in the meaning of the verbal stem. The merger of the masculine or feminine nominal feature simply transforms a verbal base into a noun without adding any aspectual meaning regarding how the action expressed by the verb is performed. Morphologically, the absence of aspectual morphology on the base and the non-aspectual nature of the feature gender suggests that these de-verbal nouns are built on a lexical stem, (e.g., √L+(v)), rather than on VP or on a AspectP. Also, syntactically, these nouns do not necessarily surface with an

110

FRANCA FERRARI-BRIDGERS

obligatory object, which confirms the hypothesis that their verbal base is a stem and not a VP. 3.

Conclusion In this paper, I have argued that in Italian de-verbal noun formation is a syntactic process due to the merger of a nominal head with a specific set of IP’s functional projections. Looking at the semantic and morphological properties of derived nominals, I identified a large array of functional projections involved in the derivation of each de-verbal noun types. The analysis shows that semantic and morphological variations across de-verbal noun types can be reduced to syntactically structural differences. References Alexiadou, Artemidis. 2001. Functional Structure in Nominals. Amsterdam & Philadelphia: John Benjamins. Bartsch, Renate. 1986. “On the Aspectual Properties of Dutch and German Nominalization”. Temporal Structure in Sentence and Discourse, ed. by Vincenzo Lo Cascio, Co Vet, 1-28. Dordrecht: Foris. Bergsland, Klaus. 1994. “Aleut Tenses and Aspects”, Tense, Aspect and Action, ed. by Carl Bache, Hans Basboll & Cari Erik Lindber, 323- 367. Berlin: Mouton de Gruyter. Carlson, Gregory, N. & Pelletier, Frances Jeffrey. 1995. The Generic Book. Chicago & London: Chicago University Press Castelli, Margherita. 1988. “La nominalizzazione”. Grande grammatica italiana di consultazione, ed. by Lorenzo Renzi, 333-356, Bologna: Il Mulino. Chierchia, Gennaro. 1998. “Plurality of Mass Nouns and the Notion of "Semantic Parameter"”. Events and Grammar, ed. by Susan Rothstein, 53103. Dordrecht: Kluwer. Cinque, Guglielmo. 1999. Adverbial and Functional Heads. Oxford: Oxford University Press Dahl, Östen. 1975. “On Generics”. Formal Semantics of Natural Language, ed. by Edward L. Keenan. Cambridge, UK: Cambridge University Press. Dahl, Östen. 1985. Tense and Aspect Systems. Oxford: Basil Blackwell. Di Sciullo, Annamaria, 2005. Asymmetry in Morphology, Cambridge, Mass.: MIT press Ferrari-Bridgers, Franca. 2005a. “Italian [VN] compound nouns: A case for a syntactic approach to word formation”. Romance Languages and Linguistic Theory 2003: Selected papers from 'Going Romance 2003', Nijmegen, 20-22 November, ed. by Twan Greerts, Ivo van Ginneken & Haike Jacobs, 63-79. Amsterdam: John Benjamins.

A SYNTACTIC ANALYSIS OF ITALIAN DE-VERBAL NOUNS

111 111

Ferrari-Bridgers, Franca. 2005b. A Syntactic Analysis of the Nominal Systems of Italian and Luganda. How Nouns Can Be Formed in the Syntax. Doctorate dissertation, New York University. Gaeta, Livio. 2002. Quando i Verbi Compaiono come Nomi. Pavia: Francoangeli. Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Kayne, Richard. 2005. Movement and Silence. Oxford: Oxford University Press Koopman, H., Szabolcsi A.. 2000. Verbal Complexes. Cambridge, Mass.: MIT Press. Josefsson, Gunlong. 1998. Minimal Words in a Minimal Syntax. Amsterdam: John Benjamins. Julien, Marit. 2000. Syntactic Heads and Word Formation: a Study of Verbal Inflection. PhD dissertation, Universitet i Tromsø. Lacarme, Jaqueline., 2002, “Gender Polarity Theoretical Aspect of Somali Nominal Morphology”. Many Morphologies ed. by Paul Boucher, 109-141. Sommerville: Cascadilla Press. Longobardi, Giuseppe. 2001. “How Comparative is Semantics? A Unified Parametric Theory of Bare Nouns and Proper Names”. Natural Language Semantics 9:4.335-369. Marantz, Alec. 1997. “No Escape from Syntax”. U.Penn Working Papers in Linguistics volume 4.2., ed. by Alexis Dimitriadis, Laura Siegel, Clarissa Surek-Clark, and Alexandra Williams, 201-224. Philadelphia: University of Pennsylvania. Mayo, Bruce, Marie-Theres Schepping, Christopher Schwarze, Angela Zaffanella. 1995. “Semantics in the derivational morphology of Italian: implication for the structure of the Lexicon”. Linguistics 33:883-938. Prati, Antonio. 1931. “Composti Imperativali Quali Casati e Soprannomi” Revue de Linguistique Romane 7: 250-264. Roeper, Thomas. 1999. “Leftward Movement in Morphology”. MIT Working papers in Linguistics 34.35-66, ed, by Karlos Arregi, Benjamin Bruening, Cornelia Krause & Vivian Lin. Cambridge, Mass.: MIT. Tollemache, Federico. 1945. Le Parole composte nella lingua Italiana. Roma: Rores. Zanuttini, Raffaella. 1997. Negation and Clausal Structure. New York: Oxford University Press.

V-N COMPOUNDS IN ITALIAN A CASE OF AGREEMENT IN WORD FORMATION

*

MARTINA GRAČANIN-YUKSEK Massachusetts Institute of Technology

0.

Introduction Compounds of the type verb-noun (V-N) in Italian have been studied for their argument structure (Di Sciullo & Ralli 1994), exocentricity (Bisetto 1999) and the prosodic status of their component parts (Nespor & Vogel 1986; Pepperkamp 1997). In this paper I explore Italian V-N compounds in the context of the gender and number morphology they bear, with an emphasis on gender marking in the plural. V-N compounds can denote instruments (‘instrument compounds’), shown in (2), or people (‘agent compounds’), illustrated in (3). Instrument compounds are masculine regardless of the gender of the noun contained in the compound.1 This is visible in their masculine agreement with associated determiners and adjectives: the noun mano (“hand”) is feminine in isolation, as shown in (1), but the compound asciugamano (“towel”) is masculine. (1) una mano a-F.SG hand “a hand” (2) a. un piccolo asciuga-mano a-M.SG small-M.SG dry-hand “a small towel” *

I am very grateful to Alec Marantz, David Pesetsky and Donca Steriade for their invaluable comments and discussion. This paper was presented at LSRL 35 at the University of Texas at Austin (February 2005). I would like to thank the audience for their useful questions and comments. All inadequacies are my own. 1 Some compounds, like lavapiatti/lavastoviglie (“dishwasher”) and portaaerei (“aircraft carrier”) are feminine, but they are very few in number and can be regarded as exceptions.

114

MARTINA GRAČANIN-YUKSEK

b. *una a-F.SG

piccola asciuga-mano small-F.SG dry-hand

Agent compounds, on the other hand, show natural gender agreement: their gender depends on the gender of their referent. (3) un /una a-M.SG /a-F.SG “busybody”

ficca-naso thrust-nose

According to prescriptive grammars of Italian, V-N compounds do not inflect in the plural, i.e. they are invariant. Native speakers, however, seem to be quite willing to violate this rule. If plural is overtly marked on the compound, it is reflected in the change of the final vowel of its nominal part, as in (4). (4) a. un asciuga-mano a-M.SG dry-hand “a towel” b. degli asciuga-mani some-M.PL dry-hands “some towels” The surprising fact is that when a compound contains a feminine noun of the first declension (casa, –e, “house”), the plural morphology which appears on the noun is feminine in form (–e), despite the masculine agreement of the compound with adjectives and determiners. For example, the feminine noun testa (“head”) belongs to the first declension and forms plural with –e: teste “heads”. In isolation, it requires a feminine determiner, as in (5). The compound poggiatesta (“headrest”), in the plural also acquires the ending –e, but it agrees with a masculine determiner. This is shown in (6). (5) a. una /*un testa a-F.SG / a-M.SG head “a head” b. delle /*dei some-F.PL some-M.PL “some heads” (6) a. un /*una a- M.SG / a- F.SG “a headrest”

teste/ *testi heads

poggia-testa rest-head

115

V-N COMPOUNDS IN ITALIAN

b. dei /*delle some- M.PL / some- F.PL “some headrests”

poggia-teste/*poggia-testi rest-heads

The morphology that the masculine compound poggiatesta (“headrest”) shows in the plural manifests the feminine gender of the embedded noun (testa, “head”). Striking is the fact that the ending –e is otherwise reserved exclusively for the plural of feminine nouns of class –a (is not found on masculine nouns of any nominal class). Table 1 shows Italian noun endings in different classes in both numbers. The plural ending of masculine nouns of all classes is –i.

Masculine Gloss Feminine Gloss

Class 1 (-a) Class 2 (-o) Class 3 (-e) Stem Sg. Pl. Stem Sg. Pl. Stem Problem -a -i Libr -o -i fulmin problem(s) book(s) lightning(s) Test -a -e Man -o -i vit head(s) hand(s) screw(s) Table 1: Italian nominal classes – Singular & Plural

Sg. -e

Pl. -i

-e

-i

I propose that the phenomenon of ‘gender mismatch’ between the morphological shape of the compound and its syntactic agreement, illustrated in (6b), is the result of agreement between different pieces of the complex structure of V-N compounds. The view of agreement I adopt relies on the notion of ‘feature sharing’ (Pollard & Sag 1994; Frampton & Guttman 2000; Pesetsky & Torrego 2004). It is the very same mechanism that lies at the heart of the convergence of a syntactic derivation. Here is how the process of agreement as feature sharing is defined in Pesetsky & Torrego (2004): (7)

(i) An unvalued feature F (‘a probe’) on a head H at a syntactic location α (Fα) scans its c-command domain for another instance of F (‘a goal’) at a location β (Fβ), with which to agree. (ii) Replace Fα with Fβ, so that the same feature is present at both locations.

What is meant by ‘an unvalued feature’ is a feature that is not specified in the lexicon. The crucial aspect of this definition of agreement is the lack of the reference to a value. Agreement is not seen as a process that happens iff a value can be assigned to a previously unvalued feature. Instead, agreement is a mechanism that applies blindly, as soon as two matching features find themselves in the appropriate syntactic configuration. Suppose that α is an

116

MARTINA GRAČANIN-YUKSEK

unvalued number feature (NumF). If β is also a NumF, and is accessible to α, α can agree with β even though β might also be unvalued. If either α or β later agrees with a valued NumF γ, the value of γ spreads to both α and β by virtue of the previously established agreement relation between the two. In the next section I illustrate how agreement works in the formation of simple nouns. 1.

Agreement in the formation of simple nouns It has been proposed in the literature that in the lexicon, a noun stem is associated with a matrix of features such as category, gender and class (Lieber 1992; Müller 2002). Suppose that a nominal feature matrix also contains the feature ‘number’. Category, class and gender features are valued, while number is unvalued.2 Here are some examples of Italian nominal stems, as they are listed in the lexicon: (8)

a. cas[num: ] [gend: f] [class: 1] [cat: N] “house”

b. libr[num: ] [gend: m] [class: 2] [cat: N] “book”

c. croc[num: ] [gend: f] [class: 3] [cat: N] “cross”

Chomsky (1995, 2000, 2001) claims that unvalued features are not usable at the interfaces. Therefore, all unvalued NumFs must agree with a valued instance of the matching feature for the derivation to converge. I suggest that the source of the number information for nominal structures in Italian is a separate lexical item, ‘inflection’ (Infl0), which is merged with the nominal stem. Infl0 is phonologically null. It comes from the lexicon either as singular or as plural. Under agreement with Infl0, the unvalued NumF on a nominal element acquires value. I propose that the final vowel on a noun, which I will be referring to as the ‘theme vowel’, is a spell-out of the values of features that make up the feature matrix associated with the noun. For concreteness, I assume that category, class and gender feature are also present on Infl0 head, but are unvalued and acquire values through agreement 2

A possible argument for the presence of NumF on nominal stems comes from the nouns that have only singular/plural forms (singularia/pluralia tantum). Grammatical number of such nouns does not depend on their syntactic environment, but is presumably specified in the lexicon. Positing the same feature on all nouns makes singularia/pluralia tantum nouns formally identical to regular nouns.

117

V-N COMPOUNDS IN ITALIAN

with the stem. This mechanism is illustrated in Figure 1.3 a.

N2

b. Infl0

N1 N ThV cas[num: ] [gen: f ] [class: 1] [cat: N]

∅ [num: sg.] [gen: ] [class: ] [cat: ]

N2 N1

⇒

N ThV cas-a [num: sg] [gen: f ] [class: 1] [cat: N]

Infl0

∅ [num: sg.] [gen: f ] [class: 1] [cat: N]

Figure 1: Agreement between the stem and Infl0 in number, gender, class and category features

Once the nominal stem, cas- is valued for number, all its features are valued and its theme vowel can be spelled-out with singular morphology: cas-a (“house”). A note is needed here about the conditions under which agreement happens. The standard conditions on agreement in syntax are the following: -

The unvalued feature that triggers agreement (‘the probe’) is always located on a head (not on a phrase). The probe must c-command the goal.

In Figure 1, the nominal stem cas- (“house”) does not c-command Infl0, under the standard definition of c-command: (9) α c-commands β iff β is not contained in α and every γ that dominates α also dominates β. (Chomsky 1986) However, the projection of N (N1) does c-command Infl0. Chomsky (1995) states that following the merger of two elements, α and β, the label of the newly formed node is either α or β. In the domain of morphology, the elements merged together are heads. It can be maintained then that in word-formation, all the elements created through the merger of two morphemes are also heads, and can as such act as probes. Thus, N1 can probe Infl0, since it c-commands it. According to this proposal, the theme vowel of a simple noun is the expression of the features present on the nominal stem. It is not a phonological expression of the inflection itself. On the other hand, Infl0, following agreement, possesses all the features that N1 possesses, and crucially, N1’s 3

The direction of the arrow is from the probe to the goal; ThV stands for ‘theme vowel’.

118

MARTINA GRAČANIN-YUKSEK

category feature. This enables us to say that in Italian noun-formation, Infl0 always projects, passing the category information of its sister up the tree. I now proceed to show how the agreement mechanism derives the gender mismatch in the V-N compounds, which we saw in (6b). 2.

Agreement in V-N compounds First, we have to establish the structure of V-N compounds in Italian. These compounds are exocentric. Their gender specification (masculine in instrument compounds (ex. 2), natural gender in agent ones (ex. 3) is not attributable to either V or N. Neither are the agentive/instrumental semantics. Agent nouns in Italian are normally derived by the pair of agentive suffixes: – tore and –trice. Both suffixes are lexically valued for gender: –tore is masculine and –trice is feminine, and they both attach to verbal stems. (10) a. allenare b. allena-tore train-INF. train-AGENT.M.SG “to train/to coach” “trainer/coach”

c. allena-trice train-AGENT.F.SG “trainer/coach”

I propose that the structure of V-N compounds contains a similar functional head, which is phonologically null (Kiparsky 1982). This head provides the agentive/instrumental semantics of the compound and turns its complement into an agent/instrument noun. I will call this nominalizing head n. Like –tore and –trice, n is lexically specified for gender. However, the value of its gender feature seems to be natural. This explains why agent compounds reflect the natural gender of their referents, and we will see that it also explains the fact that instrument compounds are masculine. Other instances of morphemes marked for natural gender in Italian include nominal stems like ragazz- (ragazzo/a = “boy/girl”), and bambin- (bambino/a = “male/female infant”). Riente (2003) proposes that in Italian, the gender value of some nominal stems is not fixed in the lexicon. These are subject to a redundancy rule, which marks them ‘feminine’ iff their referent is a female. I adopt this idea, but propose that it is morphemes with natural gender specification that undergo the rule, and not those whose gender feature is unvalued (for the latter, I propose that they acquire the value via agreement). The redundancy rule is given in (11): (11)

[nat]GEN → [f] / ___ , ♀

If the referent of a noun with the natural gender value is not a female, it is

119

V-N COMPOUNDS IN ITALIAN

assigned masculine by default.4 Infl[n] n[V] V V

poggia rest

N N testhead [num: ] [gen: f ] [class: 1] [cat: N ]

ThV -a

n

Infl0

∅

∅

[num: ] [num: sg/pl] [gen: nat] [gen: ] [class: ] [class: ] [cat: N] [cat: ]

Figure 2: Structure of the V-N compound poggiatesta (“head rest”)

5

Given everything presented so far, the structure of a V-N compound must look something like Figure 2.6 The structure in Figure 2, together with the feature sharing operation of agreement is responsible for the gender mismatches observed at the beginning of the paper. In the next section, I show how gender mismatches are derived. 2.1.

Deriving the gender mismatch Here I present step by step derivation of the plural form of the compound poggiatesta (“head rest”), which for most speakers I consulted surfaces as poggiateste (“head rests”). In the first step, the stem is merged with the verb, but no agreement takes place between them. The noun is understood as an internal argument of the verb. It is expected that it is the θ-role assigner that projects, rather than the 4

Additional evidence for n and its gender specification is provided by diminutive formation: instrument V-N compounds in Italian always form diminutives by a masculine suffix –ino, regardless of the gender of the embedded noun. For reasons of space I omit these data here. 5 Infl[n] labels the node projected by Infl0 after its merger with n. Throughout the paper, subscripted brackets in diagrams are used to indicate the head which most recently merged with the projecting head. 6 The verbal element in Italian V-N compounds is a non-agreeing verbal form. Different authors assign it different morphological make-up: Tollemache (1945) considers it to be the third person singular indicative. Bisetto (1999) and Guevara & Scalise (2004) propose that what appears to be the ‘verbal’ part of the compound is actually its agentive nominalization, with invisible (either null or dropped) nominalizing suffix –tore. What is important for this analysis is that no agreement takes place between the verbal and nominal parts of a compound. I assume this to be the consequence of the verbal element lacking the relevant features.

120

MARTINA GRAČANIN-YUKSEK

assignee. The head which projects is thus probably the verb.7 Furthermore, if n, which merges next, is indeed like –tore/–trice, we expect it to be subcategorized for a verbal element. Next, the null nominalizer is merged, as shown in Figure 3. Agreement is triggered by all unvalued features. The unvalued class feature on n is valued by the class feature of N, the closest (and the only) head with the relevant feature in the c-command domain of n. The unvalued NumF on n enters agreement with the matching feature on N, but since both are unvalued, neither acquires a value. However, a link is established between them (marked in the diagram as Ag, for ‘Agreed’), so that once one of them is valued, the other will receive the same value. Since the NumF on the nominal stem has not yet acquired a value, the shape of the theme vowel is still not determined. No agreement in gender feature is happening in the structure, since both gender features have a lexically specified value: the noun is feminine, and n possesses natural gender. At this point, the structure looks like the one in Figure 3b. a.

n[V]

b.

V

V

poggia rest

n[V] V

N N ThV testhead [num: ] [gen: f ] [class: 1] [cat: N]

n

∅ [num: ] [gen: nat] [class: ] [cat: N]

V

⇒

poggia rest

N N ThV testhead [num: Ag] [gen: f] [class: 1] [cat: N]

n

∅ [num: Ag] [gen: nat] [class: 1 ] [cat: N]

Figure 3: Derivation of poggiateste (“head rests”): steps 1 and 2

Finally, the inflection node is merged, with the valued NumF. We are interested in the case where the value is plural, since only then do we obtain a gender mismatch. The unvalued features on Infl0 (gender, class and category) trigger agreement with the valued features on n[V]. The unvalued NumF on n[V] also triggers agreement with the valued NumF of Infl0 (recall that in this model n[V] has the status of a head, therefore, a probe).

7

This is not to say that the resulting constituent is a verb per se, but only that it is ‘V-headed’.

121

V-N COMPOUNDS IN ITALIAN Infl[n] n[V]

V V

poggia rest

N N testhead [num: ] [gen: f ] [class: 1] [cat: N ]

n

Infl0

∅

∅

ThV -e

[num: pl] [gen: nat] [class: 1] [cat: N]

[num: pl] [gen: nat] [class: 1] [cat: N]

Figure 4: Derivation of poggiateste (“head rests”): step 3

Once the value for the NumF on n[V] is set to plural, the same value appears on the nominal stem, by virtue of the feature sharing mechanism of agreement. All the features on nominal stem are now valued. The exponent of the obtained feature matrix on N is the vowel –e, since the noun is feminine and belongs to class 1. The structure obtained is the one in Figure 4. Merger of Infl0 triggers the application of the redundancy rule in (11). Given that the referent of the compound is not a female, the structure is assigned masculine gender, probably by default.8 As a result of the masculine gender assignment, the projected node of Infl0 now contains features [m] and [pl]. In Italian, determiners agree with their complement nouns in number and in gender. So, the determiner that accompanies the noun poggiateste (‘head rests’) must be i (‘the-M.PL’). Thus, we obtain the gender mismatch: a noun with feminine morphology and masculine syntactic features: (12)

i poggia-teste the-M.PL rest-heads “the head rests”

Italian V-N compounds that show gender mismatch include compounds such as puliscitestina/-e (“CD player head cleaner/s”), apriporta/-e (“remote door opener(s)”), segnavia/-e (“signpost/s”), copritastiera/-e (“keyboard cover/s”). The analysis outlined here extends to all compounds that have a distinct morphological form in the plural. The theme vowels vary with the nominal 8

It remains an open question whether all instances of [nat] in the structure are replaced by [m], or just the one on the most recently merged element, i.e. the one on Infl0.

122

MARTINA GRAČANIN-YUKSEK

class and gender, since they are the spell-out of the particular feature combination present on N. With the exception of feminine nouns of the first declension or class 1 (–a), the plural marker for all other nouns in Italian is –i, as shown in Table 1. Thus we obtain the following:9 (13) a. la the-F.SG “the hand” c. il the-M.SG “the port”

mano hand-CL2 porto port-CL2

b. gli asciuga-mani the-M.PL dry-hands “the towels” d. i passa-porti the-M.PL pass-ports “the passports”

This analysis indicates that agreement is operative not only in sentence-syntax, but also in structures below the level of the word. It gives support to agreement seen as feature sharing, triggered by unvalued features, regardless of whether or not they are assigned a value as a result of the agreement operation. The analysis as it stands predicts that in all V-N compounds the NumF on the stem is determined by the number value present on the Infl0 which attaches outside n, i.e. that no V-N compounds are invariant. This is of course not right. In the following section I examine invariant compounds. 2.2.

Deriving invariant V-N compounds The difference between the compounds that have a distinct plural form and the invariant ones can be attributed to whether the number of the noun embedded in the compound is interpreted or not. In the compounds that inflect for plural, the interpretation of the compound does not depend on the plurality of the embedded noun. It does not matter whether a towel is used to dry one or more than one hand, or whether one or more heads rest on a headrest. When the meaning of the compound does not co-vary with the value of the NumF on N, this feature can be valued late in the derivation, as we saw above. Invariant compounds, on the other hand, are transparent enough with respect to the semantics of the embedded noun for the speakers to know whether a singular or a plural noun is appropriate. Thus, there is a difference between portaspazzolino and portaspazzolini (carry toothbrush-SG/PL = “toothbrush holder”): the former holds only one toothbrush, while the latter holds more than one. A compound like cavalcavia (ride street = “overpass”) has to contain a singular noun via (“street”), presumably because an overpass is conceived of as crossing only one street. Here, the denotation of the compound varies (or would vary) with the toggling of the value of the NumF 9

CL1/2/3 indicates the nominal class the noun belongs to, as shown in Table 1.

123

V-N COMPOUNDS IN ITALIAN

on N. The number value of N seems to be built into the meaning of the compound.10 In this model, the NumF on nouns is unvalued in the lexicon. The source of its value is Infl0. If for some compounds N must be either singular or plural in order for the compound to be assigned meaning, and this particular value must be available regardless of the compound’s syntactic environment. I propose that in these cases an additional Infl0 node attaches directly to the noun at the very beginning of the derivation. This step is shown in Figure 5. a.

b.

Infl[N] Infl0

N N cartpaper [num: ] [gen: f] [class: 1] cat: N]

Infl[N] Infl0

N

ThV

N ThV cart -e papers [num: pl] [gen: f] [class: 1] [cat: N]

∅ [num: pl] [gen: ] [class: ] [cat: ]

⇒

∅ [num: pl] [gen: f] [class: 1] [cat: N]

Figure 5: Derivation of fermacarte (stop papers = “paperweight”): step 1

a.

b.

n[V] V

n[V]

V

V

Infl[N] N

n

Infl0

⇒ N ThV ferma cart-e ∅ stop paper [num: pl] [num: pl] [gen: f] [gen: f] [class: 1] [class: 1] [cat: N] [cat: N]

V

Infl[N] N

∅ [num: ] [gen: nat] [class: ] [cat: N]

n Infl0

N ThV ferma cart-e ∅ ∅ stop papers [num: pl] [num: pl] [num: pl] [gen: f] [gen: f] [gen: nat] [class: 1] [class: 1] [class: 1] [cat: N] [cat: N] [cat: N]

Figure 6: Derivation of fermacarte (stop paper “paperweight”): step 2 10

Not all compounds that are invariant rely for their meaning on the number of N. Thus apribottiglie (open bottles = “bottle opener”) is invariant, even though the plural of bottiglie (“bottles”) does not seem to contribute anything to the semantics of the compound. A plausible explanation for why in such cases the number of the embedded noun must be fixed might involve the speakers’ knowledge of the world, i.e. their knowledge of the use of the object denoted by the compound (whether it is used to perform an action on one or more objects).

124

MARTINA GRAČANIN-YUKSEK

All unvalued features trigger agreement and acquire values. Once N possesses values for all the features in its feature matrix, the theme vowel is spelled out as feminine plural of class 1: –e.11 The noun is merged with V, and then with n. Unvalued features on n trigger agreement with the valued features on the closest head in the c-command domain of n with the relevant features. Since V does not have the relevant features, n agrees with the next closest head: Infl[N]. Finally, the ‘outer’ Infl0 is merged. The unvalued features on Infl0 agree with the n[V]. Since n[V] has no unvalued features left in its feature matrix, it triggers no agreement with Infl0. Even though Infl0 acquires gender, class and category features from n[V], no feature values are transferred from Infl0 to n[V] or to any other node in the structure. Thus, the value of the NumF on the outer Infl0 only affects the choice of the determiner. In Figure 7 below, the outer Infl0 bears singular features. This completes the derivation: all unvalued features have acquired values. Natural gender is replaced by masculine by default gender assignment. The outer inflection is marked for masculine singular, and this is the information used for determiner and modifier agreement. Infl[n] n[V] V

V

Infl[N] N

N ThV ferma cart-e stop papers [num: pl] [gen: f] [class: 1] [cat: N]

n

Infl0

∅

∅

Infl0

∅ [num: pl] [gen: f] [class: 1] [cat: N]

[num: pl] [gen: nat] [class: 1] [cat: N]

[num: sg] [gen: nat] [class: 1] [cat: N]

Figure 7: Derivation of fermacarte (stop papers = “paperweight”): step 3

3.

Conclusion The model of morphology developed in this paper correctly derives the plural forms of both invariant and inflected V-N compounds in Italian (including those with the gender mismatch). Moreover, it does so by relying on 11

For this to be possible, some notion of cyclic spell-out of morphological structures is necessary. It seems plausible that spell-out is triggered by the merger of the Infl0 node.

V-N COMPOUNDS IN ITALIAN

125

the operation of agreement, which is well attested in linguistic research. If this analysis is correct, the difference between morphology and syntax lies in the kind of features present on the components that are merged together and in the status of the intermediate nodes built in the course of a derivation (probes vs. non-probes). On the other hand, the two modules are the same with respect to the processes they employ in deriving legitimate structures. References Bisetto, Antonietta. 1999. “Note sui composti VN dell’italiano”. Fonologia e morfologia dell'italiano e dei dialetti d'Italia. Atti del XXXI Congresso della Società di Linguistica Italiana - Padova, Settembre 1997, ed. by Paola Benincà, Alberto Mioni & Laura Vanelli, 25-27. Chomsky, Noam. 1986. Barriers. Cambridge, Mass.: MIT Press. ---------. 1995. The minimalist program. Cambridge, Mass.: MIT Press. ---------. 2000. “Minimalist inquiries: The framework”. Step by step: Essays in Minimalist Syntax in Honor of Howard Lasnik, ed. by Robert Martin, David Michaels and Juan Uriagereka, 89-155. Cambridge, Mass.: MIT Press. ---------. 2001. “Derivation by Phase”. Ken Hale: a life in language, ed. by Michael Kenstowicz, 1-52. Cambridge, Mass.: MIT Press. Di Sciullo, Anna-Maria & Angela Ralli. 1994. “Argument structure and inflection in compounds: some differences between English, Italian and Modern Greek”. Proceedings of the Workshop on Compound Nouns: Multilingual aspects of nominal compositio, ed. by Pierrette Bouillon & Dominique Estival, 61-76. Geneva: ISSCO. Frampton, John & Sam Guttman. 2000. “Agreement is Feature Sharing”. Ms., MIT. Guevara, Emiliano & Sergio Scalise. 2004. “V-compounding in Dutch and Italian.”. Cuadernos de Lingüística del Instituto Universitario Ortega y Gasset – Madrid – XI, 1-29. Kiparsky, Paul. 1982. “From Cyclic Phonology to Lexical Phonology”. The structure of phonological representation, ed. by Harry Van der Hulst & Norval Smith, 131–175. Dordrecht: Foris. Lieber, Rochelle. 1992. Deconstructing Morphology. Chicago: The University of Chicago Press. Müller, Gereon. 2002. “On Decomposing Inflection Class Features: Syncretism in Russian Noun Inflection.” To appear in Explorations in Nominal Inflection, ed. by Lutz Gunkel, Gereon Müller & Gisela Zifonun. Berlin: Mouton de Gruyter. Nespor, Marina & Irene Vogel. 1986. Prosodic Phonology. Dordrecht: Foris. Pepperkamp, Sharon. 1997. Prosodic Words. Den Haag: Holland Academic

126

MARTINA GRAČANIN-YUKSEK

Graphics (HIL dissertations). Pesetsky, David & Esther Torrego. 2004. “Tense, Case and the nature of syntactic categories”. The Syntax of Time, ed. by Jacqueline Guéron & Jacqueline Lecarme, 495-537. Cambridge, Mass.: MIT Press. Pollard, Carl & Ivan Sag. 1994. Head-driven Phrase Structure Grammar. Chicago: University of Chicago Press. Riente, Lara. 2003. “Ladies First: The Pivotal Role of Gender in the Italian Nominal Inflection System”. McGill Working Papers in Linguistics 17:2.153. Tollemache, Federico. 1945. Le Parole Composte nella Lingua Italiana. Rome: Edizione Rores di Nicola Ruffino.

A REINTERPRETATION OF QUIRKY SUBJECTS AND RELATED PHENOMENA IN SPANISH*

RODRIGO GUTIÉRREZ-BRAVO CIESAS-Mexico City

0.

Introduction Icelandic is a nominative-accusative SVO language that is well-known to have quirky subjects (Zaenen et al. 1985, Sigurðsson 2004, and many others). Quirky subjects are oblique arguments that otherwise behave like surface subjects in every relevant respect. Nominative and quirky subjects are exemplified in (1). (1) a. Við hjálpuðum stelpunum we-NOM helped girls.the-DAT “We helped the girls.” b. Henni líkuðu hestarnir her-DAT liked.3PL horses.the-NOM “She liked the horses.”

(Sigurðsson 2004)

It is well-known that Spanish has data that apparently replicate the quirky subject construction. The clearest cases are psych verbs where the dative argument, and not the nominative argument, occupies the preverbal position. (2) A ella le gustaron los caballos to her CL like-3PL the horses “She liked the horses.” *

Many thanks to Ana Aguilar, Judith Aissen, Francisco Arellanes, Donka Farkas, Joan Maling, James McCloskey, John Moore, Enrique Palancar, Anne Sturgeon, Esthela Treviño, María Eugenia Vázquez Laslop, an anonymous reviewer, and the audiences at LSRL 35 and at the Typology Seminar of the National Institute of Anthropology and History (Mexico) for their help and comments on different parts of this analysis. All errors that remain are my own.

128

RODRIGO GUTIÉRREZ-BRAVO

There are two ways in which the ‘subjecthood’ of the dative in (2) has been understood. Working in Relational Grammar, González (1988) proposes that the dative XP in (2) is an Inversion Nominal (I-Nominal). I-Nominals are underlying subjects that surface as indirect objects. Hence they are expected to show some (albeit not all) of the syntactic properties of subjects. The second kind of analysis takes the preverbal dative to be a quirky (and hence surface) subject. Masullo (1993) first proposed that the dative in (2) can be analyzed as a quirky subject, though different from those of Icelandic. The quirky subject analysis has been recently extended to other constructions in Spanish, such as impersonal constructions in Fernández-Soriano (1999) and to a further class of psych verbs in Rivero (2004), both to be discussed in what follows.1 In this paper I provide evidence that the dative in (2) is neither an underlying nor a quirky subject. Instead I argue that the ‘subject’ properties they display are unrelated to subjecthood. Rather, I propose that they result from a word order effect whereby in Spanish the preverbal position is occupied by the argument whose semantic role ranks highest in the Thematic Hierarchy. 1. 1.1.

Spanish preverbal obliques are not oblique subjects González (1988) González (1988) claims that the preverbal datives of verbs like gustar “to like”, show two properties characteristic of subjects in Spanish. First, preverbal datives can bind the anaphor sí mismo “himself”, but ordinary IOs cannot. (3) a. Al príncipei le gustó Soraya para sí mismoi to-the prince CL liked Soraya for himself “The Prince liked Soraya (to keep her) for himself.” b. *Marta le habló al psiquiatrai de sí mismoi. Marta CL spoke to-the psychiatrist about himself However, the status of (3a) is not entirely clear. Three speakers of Mexican Spanish were presented with this example and they all rejected it. My sense is that it is perhaps not ungrammatical, but it is clearly very strongly deviant. More importantly, (3a) and (3b) are not equivalent. In (3a) the anaphor 1

These two ways of analyzing oblique subjects are not mutually exclusive. For instance, Moore and Perlmutter (2000) show that Russian has both I-Nominals and true dative (quirky) subjects. However, I do not make use of this distinction in my analysis since I provide evidence against both the analysis of Spanish preverbal datives as I-Nominals in González (1988) and the analyses that propose that these datives are true quirky (surface) subjects.

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

129

is embedded in a PP adjunct, whereas in (3b) it is in a PP that is an argument of the verb. As noted in Masullo (1993), when the anaphor corresponds to an argument, the preverbal dative is equally unable to bind it, as in (4). In this respect, Spanish is unlike Icelandic, where oblique subjects can indeed bind anaphora. (4) a. * A to b. * A to

Adrianai le gusta sí mismai Adriana CL likes herself Marcosi le preocupa sí mismoi Marcos CL worries himself

(Masullo 1993: 310)

This result is confirmed by a text count carried out as part of my research. In all the 20th century oral Spanish texts in Davies (2004), a total of 347 occurrences of the various inflected forms of 3rd person sí mismo were found.2 Out of these, 305 were bound by the nominative subject (or its corresponding argument in nominalizations) but none were bound by a preverbal dative.3 The second property mentioned by González is the no… sino construction. In this construction the null subject of the second predicate is typically coreferential with the subject of the first predicate, as in (5). González claims that the preverbal datives of psych verbs share this property with subjects, whereas ordinary IOs do not. The data in (6) is from González (1988). (5) Las águilasi no corren, sino que proi vuelan the eagles not run instead that they.fly “Eagles don’t run: they fly.” (6) a. A Juani no le gustan las rubias, sino que proi prefiere to Juan not CL like the blondes rather that he.prefers las morenas the brunettes “Juan doesn’t like blondes: he prefers brunettes.”

2

This anaphor inflects for number and gender in Spanish. Of the remaining forms, 30 were logophors, two were bound by the DO, one by an ordinary IO, and in 9 cases it was not possible to determine what the binding element was. 3

130

RODRIGO GUTIÉRREZ-BRAVO

b. *La CIA no le pidió explicaciones a Nixoni, sino que the CIA not CL asked explanations to Nixon rather that proi obedeció obeyed-3SG (“The CIA didn’t ask Nixon for any explanations; he just obeyed.”) However, it seems to me that this is just a pragmatic effect. An ordinary IO can indeed be coreferential with the pro of the second predicate as long as the IO is fronted to a topic position, as in (7).4 Hence it is reasonable to conclude that the possibility to be coreferential with the second predicate is dependent on topicality and not on subjecthood. (7) ?A Luisi no le dieron acceso al laboratorio, sino que to Luis not CL they.gave access to.the lab rather that proi entró sin permiso entered without permission “Luis was not granted access to the lab: he went in without permission.” 1.2.

Fernández-Soriano (1999) Fernández-Soriano (1999: henceforth F-S) analyzes two kinds of active intransitive constructions as quirky subject constructions. The first kind are intransitive stative predicates and meteorological verbs that can appear with a preposed locative XP. Some of these predicates (such as faltar “to lack”) instead take a dative XP as their argument. The second class corresponds to intransitive eventive predicates (like pasar “to happen”) that alternatively take a locative or a dative XP as an argument. This XP typically also appears in the preverbal position. These classes are exemplified in (8) and (9), respectively. (8)

En Madrid nieva in Madrid it.snows “It snows in Madrid.”

(9) a. Aquí pasa algo here happens something “Something’s going on here.”

4

(Fernández-Soriano 1999)

(Fernández-Soriano 1999)

Most speakers consulted considered (7) less than perfect, but still clearly grammatical.

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

131

b. A Juan le pasa algo to Juan CL happens something “Something’s going on with Juan.” F-S shows that these locative and dative XPs are arguments of these predicates: they are different from adjuncts with respect to extraction from coordinate clauses, interrogative inversion, and the position where they are base-generated. These conclusions will not be contested here. However, I will show that this does not entail that these oblique arguments are quirky subjects. I take (9b) to be a kind of experiencer-verb-theme construction akin with (2), and so I suggest it be analyzed as other psych verb constructions addressed later in Section 2.3. Here I concentrate on evidence that locatives in impersonal constructions are not quirky subjects. The relevant evidence is as follows. First, there is evidence that the locatives in (8) and (9a) do not occupy the preverbal position in the unmarked case (contra F-S). F-S claims that the word order in (8) and (9a) is felicitous in an out-of-the-blue context, a standard diagnostic for unmarked word order. However, it seems that this results from the fact that these data correspond to the habitual present, which is easily amenable to an interpretation where the referent of the locative (a highly definite and individuated NP) has already been introduced in the discourse or is readily identifiable by both the speaker and hearer. Instead, consider a context where speaker A walks into a room and sees speaker B watching the news on television. He then utters the question in (10a). In this context, the answer where the locative PP occupies the preverbal position is robustly infelicitous for all speakers of Mexican Spanish consulted. This points to the conclusion that the locative XP in (8) and (9a) surfaces in the preverbal position not because it is a subject, but rather because it is a fronted sentence topic. (10) a. Qué pasa? “What’s happening?” b. Está nevando en Barcelona it.is snowing in Barcelona “It’s snowing in Barcelona.” c . #En Barcelona está nevando in Barcelona it.is snowing Secondly, consider those impersonal verbs that allow for either a locative or a dative argument. F-S claims that, since these dative/locative XPs

132

RODRIGO GUTIÉRREZ-BRAVO

are quirky subjects, they cannot co-occur in the preverbal position (where Case is assigned), unless the locative is an adjunct as in (11). (11)

[En Barcelona [ nos pasó lo peor]] in Barcelona to.us happened the worst “When we were in Barcelona, the worst happened to us.”

However, impersonal constructions do allow for simultaneous co-occurrence of the dative and locative XPs, as in (12), which casts doubts on their “subject” status. Instead, this again indicates that the presence of the locative XP in [Spec, T] in (8) and (9a) is not related to its status as an argument, but rather to information structure considerations (i.e. topicalization; cf. Section 3). (12)

En esta colonia nos faltan policías in this neighborhood to.us lack policemen “We do not have enough policemen in this neighborhood.” (*When we are in this neighborhood, we don’t have enough policemen)

Finally, a third argument is found in long wh-extraction. A clear distinction between subjects (nominative and quirky) and non-subject fronted topics is that topics block long wh-extraction (Zaenen et al. 1985, Masullo 1993; the data in (13) is from Goodall 2001). As shown in (14), the preverbal locative XPs of impersonal constructions behave like topics in this respect, not like subjects. (13) a. A quién crees [que Juan le dio el premio]? to whom you.think that Juan CL gave the prize “Who do you think that Juan gave the prize to?” b.*A quién crees [que el premio se lo dieron]? to whom you.think that the prize CL CL they.gave (14) a. *Quéi dices [que what you-say that b.??Cuántos díasi how-many days semana pasada]? week past

en Barcelona pasó ti la semana pasada]? in Barcelona happened the week past dices [que en Barcelona nevó ti la you-say that in Barcelona snowed the

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

133

1.3.

Rivero (2004) Icelandic quirky subjects show the peculiar property that they block person agreement between the verb and the nominative argument (Sigurðsson 2004). Accordingly, 1st or 2nd person agreement in (15a, b) is impossible. Instead, the verb form of the quirky subject construction requires default 3rd person agreement and so nominative objects can only be 3rd person, as in (15c). (15) a.* Ég veit að honum líkum I know that him-DAT like-1PL b.* Ég veit að honum líkid I know that him-DAT like-2PL c. Ég veit að honum líka I know that him-DAT like-3PL “I know that he likes them.”

við we-NOM þið you-PL-NOM þeir they-NOM

Rivero claims that there is a class of psych verbs in Spanish (antojar “to fancy”, olvidar “to forget”, and ocurrir “to come up with an idea”) that replicates the Icelandic pattern. She provides examples like (16) as evidence that, just as in Icelandic, in the corresponding Spanish examples agreement is not possible with a 1st or 2nd person nominative argument. (16) a. A Ana siempre se le antojan los mismos to Ana always CL.3.REFL CL.DAT fancy-3PL the same chicos guys “Ana always takes a fancy to the same guys.” b.* A Ana siempre os le antojáis vosotros to Ana always CL.2PL CL.DAT fancy-2PL you-NOM.PL c.* A Ana siempre nos le antojamos nosotros to Ana always CL.1PL CL.DAT fancy-1PL we-NOM However, Rivero’s observation is actually incorrect. It seems that there is more than one factor responsible for the ungrammaticality of (16b-c), but examples with 2nd person singular subjects show that it does not have to do with subjecthood. Subject to some obvious pragmatic restrictions (the verbs olvidar “forget” and ocurrir “to come up with an idea” in Spanish are odd with animate or human subjects), these verbs can show second person agreement, as in (17).

134

RODRIGO GUTIÉRREZ-BRAVO

(17) Tú te me antojas para el papel de príncipe you-NOM CL.2SG CL.1SG fancy-2SG for the role of prince “I fancy you for the role of prince” (i.e. Hamlet in a play). It seems that examples like (16b-c) are ungrammatical not because of the ‘subjecthood’ of the dative argument, but rather because they show a sequence of clitics that is disallowed in Spanish. Rivero herself hints at this when she links the behavior of (16b-c) to the Person-Case Constraint of Bonet (1994:36). (18)

Person-Case Constraint (PCC): If DAT then ACC-3rd

The PCC states that in the presence of a dative clitic, the other clitic must be a 3rd person accusative clitic. We can modify its definition according to some more current assumptions, namely; (a) the clitic se has no person specification (Heap 1998), and; (b) the [dative] feature is exclusive to 3rd person clitics (Grimshaw 2001), i.e. 1st/2nd person clitics never bear it. Now we can redefine (18) as a purely morphological filter on possible clitic sequences as in (19), a notational variant of the PCC. This filter disallows the cooccurrence of clitics specified for [person] with dative clitics, irrespective of their relative order. (19) SPANISH CLITIC-SEQUENCE FILTER * [person] ; [dative] Examples (16a) and (17) comply with this filter, but the clitic sequences in (16b-c) do not. The sequence os le in (16b), for instance, has a 2nd person clitic and a dative clitic, in violation of (19). In contrast, the clitic sequence in (16a) has a dative clitic, but the clitic that precedes it (se) has no person specification, so (19) is satisfied. In turn, in the clitic sequence in (17) both clitics have a [person] specification, but neither of them bears a [dative] feature and so again (19) is satisfied. Summing up, the ungrammaticality of Spanish (16b-c) is unrelated to subjecthood: it is the result of a purely morphological constraint on clitic sequences in Spanish. The dative XP does not block agreement with the nominative argument, which can show full agreement with the verb as long as the resulting clitic sequence complies with (19). As such, there is no relation between the Spanish data in (16) and the Icelandic data in (15).

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

135

1.4.

Masullo (1993) Masullo (1993) is the first and most comprehensive proposal that the preverbal datives in (20a) are akin to the Icelandic preverbal dative in (20b). (20) a. A ella le gustaron los caballos to her CL like-3PL the horses “She liked the horses.” b. Henni líkuðu hestarnir her-DAT liked-3PL horses.the-NOM “She liked the horses.”

(Sigurðsson 2004)

Masullo (1993) shows that the Spanish preverbal IOs in constructions like (20a) are not left-dislocated topics in an A-bar position. These results seem well motivated and will not be contested here. However, it does not seem that from this fact alone we can conclude that (20a) is a quirky subject construction. Specifically, Sigurðsson (2004) identifies three properties that crucially distinguish quirky subjects from the non-subject preverbal obliques of German (see also Zaenen et al. 1985). Spanish has none of these properties and so it is not like Icelandic, but like German, which does not have quirky subjects. The first property is that quirky subjects block person agreement between the verb and the nominative argument. We have seen that preverbal datives in Spanish do not have this property. The other two properties (also identified as crucial diagnostics for quirky-subjecthood in Zaenen et.al. 1985) are in fact also shown to be absent in Spanish by Masullo. First, Icelandic quirky subjects participate in Conjunction Reduction, as in (21). This kind of construction, where a nominative subject is coreferential with a deleted dative in the second conjunct, is not possible in Spanish. (21) Ég hafði mikið að gera og ___ var samt ekki hjálpað I-NOM had much to do and e-DAT was still not helped “I had much to do and was nonetheless not helped.” (22) *Ana ama a los perros y ___ gustan los caballos Ana loves ACC the dogs and like-3PL the horses (Ana loves dogs and likes horses). It might be argued that (22) is ungrammatical not because of the impossibility of correferentiality under conjunction reduction, but rather

136

RODRIGO GUTIÉRREZ-BRAVO

because preverbal datives in Spanish always require a dative clitic (deleted under reduction in (22)) to double them. This is illustrated in (23). (23) ??A ellos les mandé sus papeles y ___ di sus credenciales. to them CL I.sent their documents and gave their IDs This may ultimately show that, because of this independent variable, the conjunction reduction test cannot be applied in Spanish. However, it should be noted that while (22) and (23) are both impossible, there is still a fairly clear contrast between them: (23) is quite bad, but (22) is almost unparsable. The third defining property of quirky-subjecthood, (originally observed in Zaenen et.al. 1985) is that quirks can be controllees in infinitival clauses. 5 (24) Hún vonast til [ að PRO leiðast ekki bókin]. she hopes that for PRO-DAT bore not the.book-NOM “She hopes not to find the book boring.” (Sigurðsson 2004) But again, Masullo (1993) notes that Spanish preverbal obliques do not have this property, as shown in (25). Now, the nominative argument of the infinitival psych verb can be a controllee, as in (26). This is evidence that this argument is the true grammatical subject of psych constructions. (25)*Carlosi hizo todo lo posible [ para eci gustar-lei las Carlos did all that possible for to.like-CL.DAT the matemáticas] mathematics (Carlos did everything possible to like maths) (26) Carlosi hizo todo lo posible [ para eci gustar-lej a Maríaj] Carlos did all that possible for to.like-CL.DAT to María “Carlos did everything possible (in order) for Mary to like him.”

5

Observe that the possibility of an oblique to control a PRO is not a valid test for subjecthood in Spanish. This is because all kinds of non-subjects can control into infinitival clauses in this language, as in (i), where a post-verbal accusative direct object controls a nominative PRO. (i) Nosotros contratamos a Carlosi [ sin PROi we hired ACC Carlos without “We hired Carlos without him having graduated.”

titularse]. to.graduate-CL

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

137

The evidence thus shows that Spanish (like German) does not have any of the three crucial properties that define quirky subjects. However, it is also fairly clear from Masullo (1993) that the preverbal experiencers of psych verbs are not just any fronted indirect object. The two clearest pieces of evidence are their unmarked word order and their relation to long wh-extraction. As shown in (20a), these datives occupy the preverbal subject position even though Spanish is otherwise an SVO language. More importantly, as noted in Masullo (1993), just like preverbal transitive subjects, preverbal dative experiencers do not block long wh-extraction: (27) Cuándoi dijo Marta [que a Marcos se le ocurrió esa idea ti]? when said Marta that to Marcos CL CL occurred that idea “When did Marta say that Marcos came up with that idea?” In this respect, Spanish preverbal datives are like Icelandic quirks (Zaenen et.al. 1985). In the following section I sketch out a proposal that accounts for both these facts. Instead of using the notion of quirky subject, the alternative analysis suggests that these facts are unrelated to subjecthood altogether. 2.

An alternative analysis My claim is that the notion of ‘subject’, nominative or quirky, is not useful for understanding the word order and long wh-extraction facts just discussed. Instead I propose that the relevant element is the XP that occupies the specifier of the highest inflectional projection, irrespective of its grammatical relation. To disassociate this specifier from any specific grammatical relation I refer to it as the Pole of the clause (see Gutiérrez-Bravo 2005 for a full definition). (28)

TP XP

T’

T

vP

Pole compró bought

el periódico the newspaper

Building on Masullo (1993) and on the Generalized TP analysis of Zubizarreta (1998), I assume that preverbal subjects, dative experiencers, and

138

RODRIGO GUTIÉRREZ-BRAVO

preverbal topics all have [Spec, T] as their final landing site. Hence all these different preverbal XPs correspond to the Pole in the examples below.6 (29) a. [TP Pedro compró [vP el periódico]] Pedro bought the newspaper “Pedro bought the newspaper.” b. [TP A ella le gustaron [vP los caballos]] to her CL like-3PL the horses “She liked the horses.” c. [TP [ El periódico ]TOPIC lo compró [vP Pedro]] the newspaper CL bought Pedro “The newspaper, Pedro bought it.” The Pole is a terminological notion that is entirely independent of the subject (nominative or quirky): it has no relation to argumenthood either, since in Spanish any argument or adjunct can function as the Pole of the clause. The Pole is simply a place-filler for the specifier of the highest inflectional projection. It can be thought of as the element that satisfies an EPP requirement like (30), from Gutiérrez-Bravo (2005). (30) EPP Clauses must have Poles The notion of the Pole is a purely structural notion, akin to the notion of onset in syllable structure. Since it corresponds solely to a specific structural position, it can indeed be the case that it is occupied by the subject argument of the verb, as in (29a). Presumably, this is (almost) the only option in languages like English where nominative Case must be checked in [Spec, T]. However, as illustrated in (29b-c), I assume that this is not the case in Spanish, where essentially any fronted constituent may land in the Pole position. In these cases, the notion of the Pole allows us to describe the highest inflectional 6

Strictly speaking, my proposal is that the Pole is a relational notion: if [Spec, T] is the highest inflectional specifier, then whatever XP occupies this position is the Pole, but when there is a further inflectional phrase above TP, then the specifier of this phrase is the Pole (see GutiérrezBravo 2005 for detailed discussion). This issue is tangential for our current discussion and so Pole is equated with [Spec, T] in what follows. However, it should be noted that CP is not an inflectional projection and so [Spec, C] can never be the Pole of the clause. This is one property of the Pole that makes it different from Rizzi’s (1994) notion of Root Specifier. Furthermore, in my proposal any clause with an inflectional layer must have a Pole, irrespective of whether it is a root or subordinate clause. In this respect, the notion of the Pole is again different from the notion of Root Specifier.

139

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

specifier without making reference to the subject grammatical relation. In what follows, I make use of the distinction between subject (understood as a grammatical relation) and Pole (a purely structural position) to account for the apparent quirky-subject phenomena in Spanish in a way that makes no reference to subjecthood. Specifically, I show that the dative arguments of psych verbs happen to surface in the Pole position because unmarked word order in this language is regulated by the Thematic Hierarchy and not by the subject grammatical relation. Hence it is unsurprising that these datives lack the characteristics of true subjects, quirky or nominative. I propose that, in the unmarked case, the Pole in Spanish is the argument of a two-argument verb that ranks highest in the Thematic Hierarchy (Gutiérrez-Bravo 2005): (see also Contreras 1976 and Masullo 1993; cf. Belleti and Rizzi 1988 for Italian). (31) AGENT > EXPERIENCER > THEME > LOCATION The notion of the Pole and the Thematic Hierarchy together derive the observed word order facts without making reference to subjecthood. The theta roles of a transitive verb are agent and theme. Hence the SVO order results from the generalization that the highest argument in the Hierarchy occupies the Pole position in the unmarked case. This same analysis derives the IO-V-S order of the psych clauses discussed so far. The arguments of these psych verbs are a nominative subject which is a theme and a dative experiencer. In this case the IO has the theta role that ranks highest in the Hierarchy. Consequently, it is not the subject, but rather the IO that surfaces in the highest inflectional specifier, as in (32b). (32) a. [TP Pedro compró [vP el periódico]]. (i.e. 29a) b. [TP A ella le gustaron [vP los caballos]]. (i.e. 29b)

SVO IO V S

A different class of Spanish psych verbs behaves in a similar way. The class including verbs like amar “to love” and odiar “to hate”, is different from gustar in that its arguments are a nominative experiencer and an accusative theme. The grammatical relations associated with each of these theta roles are different, but the theta roles are the same. We thus expect the experiencertheme order to be unmarked, which is indeed what is observed. (33)

Juan ama a María Juan loves ACC María “Juan loves María.”

S <EXP>

V

O



140

RODRIGO GUTIÉRREZ-BRAVO

There is yet another class of psych verbs that supports this analysis. Verbs like molestar “bother”, show an alternation where the experiencer is realized as either an accusative DO or a dative IO (Treviño 1992). This alternation in turn correlates to the difference in unmarked word order illustrated below. (34) a. Los niños molestan a Juan the children bother ACC Juan “The children bother Juan.” b. A Juan le molestan los niños to Juan CL bother the children “(The) children bother/irritate Juan.”

S VO

IO V S

As shown in Gutiérrez-Bravo (2005), the word order alternation in (34) depends on how “agent-like” the nominative subject is. The semantic distinctions at play here are best understood with the Proto-Role analysis in Dowty (1991). The subject in (34b) has only two of Dowty’s five proto-agent entailments; causing an event or change of state in another participant and existing independently of the event named by the verb. It is thus more like a theme than an agent: hence the experiencer-theme unmarked word order. In contrast, the subject in (34a) has these entailments plus the volitionality entailment. Apparently, with these three entailments it is more like an agent than a theme and so agent-theme is the unmarked word order. Evidence that (34a) has the volitionality entailment, whereas (34b) does not, can be found with two diagnostics developed in Ackerman and Moore (2001). First, Ackerman and Moore note that when a verb has the volitionality entailment its subject is compatible with adverbials that entail volitionality, but subjects of predicates without this entailment are not. This is what is observed when we compare the examples in (35). (35) a. Los niños molestan a Juan a propósito the children bother ACC Juan to purpose “The children bother Juan on purpose.” b. ??A Juan le molestan los niños a propósito to Juan CL bother the children to purpose Secondly, subjects can control into purpose clauses if their predicates have the volitionality entailment, but not otherwise. Again this is what is observed in Spanish. As shown in (36b), the nominative argument of the IO-VS construction cannot control into the purpose clause.

A REINTERPRETATION OF SPANISH QUIRKY SUBJECTS

141

(36) a. Los niños molestan a Juan [para hacerlo llorar] the children bother ACC Juan for to-make-him cry “The children bother Juan in order to make him cry.” b.* A Juan le molestan los niños [para hacerlo llorar] to Juan CL bother the children for to-make-him cry (The children irritate Juan in order to make him cry.) Thus, subjecthood is not relevant for regulating unmarked word order in Spanish. Instead, with the notion of the Pole the generalization can be established that (in the unmarked case) the Pole in Spanish is the argument of a two-argument verb that ranks highest in the Thematic Hierarchy. This in turn allows us to establish a descriptive generalization of the long wh-extraction facts in (13a) and (27). Concretely, we can now state the generalization that unmarked Poles do not block long wh-extraction, whereas XPs that land in the Pole position as a result of topicalization do. Again, no reference to subjecthood need be made. 3.

Conclusions In this paper I have argued against an analysis of Spanish preverbal obliques as quirky subjects. While Spanish does have non-topical preverbal datives, they do not have the defining properties of quirky subjects. I have proposed that the notion of subject is not helpful for understanding the word order phenomena observed in these cases. Instead I have proposed the notion of the Pole of the clause, which is the XP that occupies the highest inflectional specifier. I have shown that unmarked word order observes the generalization that the Pole is the argument that ranks highest in the Thematic Hierarchy. This analysis accounts for the fact that dative arguments surface in the preverbal position of some Spanish verbs, without compromising the evidence that the nominative argument of these verbs is the true grammatical subject. References Ackerman, Farrell, and John Moore. 2001. Proto-properties and grammatical encoding. Stanford: CSLI. Belletti, Adriana, and Luigi Rizzi. 1988. “Psych verbs and Theta Theory”. Natural Language and Linguistic Theory 6.291-352. Bonet, Eulàlia. 1994. “The Person-Case Constraint: A morphological approach”. MIT Working Papers in Linguistics 22, ed. by Heidi Harley & Colin Phillips, 33-52. Cambridge, Mass. Contreras, Heles. 1976. A theory of word order with special reference to Spanish. Amsterdam: North Holland.

142

RODRIGO GUTIÉRREZ-BRAVO

Davies, Mark. 2004. Corpus del Español. Online, Brigham Young University. Dowty, David. 1991. “Thematic Proto-Roles and Argument Selection.” Language 67.547-619. Fernández-Soriano, Olga. 1999. “Two types of impersonal sentences in Spanish: Locative and dative subjects”. Syntax 2.101-140. González, Nora. 1988. Object and Raising in Spanish. New York: Garland. Goodall, Grant. 2001. “The EPP in Spanish”. Objects and other Subjects, ed. by William Davies and Stanley Dubinsky, 193-224. Dordrecht: Kluwer. Grimshaw, Jane. 2001. “Optimal Clitic Positions and the Lexicon in Romance Clitic Systems”. Optimality Theoretic Syntax, ed. by Geraldine Legendre, Jane Grimshaw and Sten Vikner, 205-240. Cambridge, Mass,: MIT Press. Gutiérrez-Bravo, Rodrigo. 2005. Structural markedness and syntactic structure. London and New York: Routledge. Heap, David. 1998. “Optimalizing Iberian clitic sequences”. Theoretical Advances in Romance Linguistics, ed. by José Lema and Esthela Treviño, 227-248. Amsterdam: John Benjamins. Masullo, Pascual J. 1993. “Two types of quirky subjects: Spanish versus Icelandic”. Proceedings of Nels 23.303-317. Moore, John and David Perlmutter. 2000. “What does it take to be a dative subject?” Natural Language and Linguistic Theory 18.373-416. Rivero, María Luisa. 2004. “Spanish quirky subjects: person restrictions and the Person-Case constraint”. Linguistic Inquiry 35.494-502. Rizzi, Luigi. 1994. "Early null subjects and root null subjects”. Language Acquisition Studies in Generative Grammar, ed. by Teun Hoekstra and Bonnie Schwartz, 151-176. Amsterdam & Philadelphia: John Benjamins. Sigurðsson, Ármann Halldór. 2004. “Icelandic Non-Nominative Subjects”. Non-Nominative Subjects ed. by Peri Bhaskarao and Karumuri Venkata Subarao, Vol. 2.137-159. Amsterdam: John Benjamins. Treviño, Esthela. 1992. “Subjects in Spanish Causative Constructions”. Romance Languages and Modern Linguistic Theory, ed. by Paul Hirschbüler and Konrad Koener, 309-324. Amsterdam: John Benjamins. Zaenen, A., J. Maling, & H. Thráisson. 1985. “Case and grammatical functions”. Natural Language and Linguistic Theory 3.441-483. Zubizarreta, María Luisa. 1998. Prosody, focus and word order. Cambridge, Mass.: MIT Press.

COGNITIVE CONSTRAINTS ON ASSERTION SCOPE THE CASE OF SPOKEN FRENCH PARCE QUE

KNUD LAMBRECHT, JULIA BORDEAUX & ROBERT REICHLE The University of Texas at Austin 0.

Introduction There is a growing body of research on the difficulties associated with the traditional distinction between subordination and coordination. Studies on English (e.g. Green 1976, Declerck 1997), German (König & van der Auwera 1988, Wegener 2001), French (Ducrot et al. 1975, Moeschler 1986, Debaisieux 1994, Le Draoulec 1997), Japanese (Shimojo 1999) and no doubt other languages have shown that various adverbial subordinating conjunctions (corresponding to when, before, although, because, etc.) can also function to introduce independent assertions, i.e. main clauses. In the particular domain of causal conjunctions, there is a striking cross-linguistic trend for the subordinating conjunction ‘because’ to supersede the corresponding coordinating conjunction: English because, French parce que, German weil, Russian potomu chto have replaced, or are in the process of replacing, for, car, denn, ibo, respectively. This trend entails both greater discourse distribution and increased functional load for because-type conjunctions. This study presents a corpus-based analysis of the use of French parce que in spontaneous spoken discourse, using data from the Minnesota corpus.1 1. 1.1.

Approaches to the study of causal conjunctions Previous studies Theoretical attempts at explaining the nature and use of causal conjunctions can be roughly divided into three types, which overlap in various respects. Previous studies have mainly been concerned with logico-semantic issues and with issues of conversational interaction. The current study approaches the subject within a theory of information structure (IS).

1 Available

from Betsy J. Kerr, University of Minnesota ([email protected]), by request.

144

LAMBRECHT, BORDEAUX & REICHLE

Semantically oriented approaches have focused on methods of argumentation in conversation, in particular on the different semantic functions of parce que (cause vs. explanation, etc.) and on the issue of negation scope (Ducrot et al 1975, Wegener 2001). Such studies have adopted a standard notation to refer to the propositions linked by a causal conjunction: the symbol p stands for the proposition that expresses the effect of a causal relationship, and the symbol q stands for the proposition that explains the cause of said effect. We will make use of the p parce que q notation throughout. The conversational approach to the study of parce que is concerned with examining the effect of parce que on the development of discourse. Dataoriented discourse studies like Debaisieux (1994) or Moeschler (1986) have dealt mainly with the observation that, in spoken French, the most common use of parce que is metalinguistic (‘énonciatif’ in Moeschler’s terminology). This use is seen in instances where, instead of explaining why the situation denoted by p occurred, q explains why p was uttered. For example in the attested utterance alors qu'est-ce que tu recommandes comme films? parce que moi, il faut que j'y aille (“so what movies do you recommend? because I have to go”) the parce que clause provides the reason for why the speaker uttered p. The present study will not deal with this metalinguistic use, except to state that it can be easily accommodated within the proposed analysis. Debaisieux and Moeschler have both observed the frequent occurrence of a PAUSE between p and parce que q and have emphasized its importance for the interpretation of the relationship between the two propositions. This pause phenomenon has important implications for the IS approach followed in the present study. Building on those previous works, we will propose an explanation for the grammar of parce que in terms of the informational structuring of the p and q propositions into topical vs. focal portions. 1.2.

An Information Structure approach to parce que Our IS framework distinguishes three analytical parameters, corresponding to different kinds of assumptions (or presuppositions) of a speaker regarding the mental state of a hearer at the time of an utterance. These assumptions concern: (i) the hearer’s knowledge or belief state (ii) her temporary state of consciousness (iii) her state of interest in a given topic The first parameter concerns the question of whether an entity or proposition is known or not yet known to the addressee, or, using Prince’s (1992) felicitous terminology, whether it is ‘hearer-new’ or ‘hearer-old’. The second parameter has to do with the mental activation states of discourse referents or, in Prince’s terms, whether a referent is ‘discourse-new’ or ‘discourse-old’. This parameter

COGNITIVE CONSTRAINTS ON ASSERTION SCOPE

145

is relatively unimportant for the present study and will not be discussed. The third parameter has to do with the issue of topicality or relevance. The first and the third of these parameters invoke the theoretical distinction between the pragmatic PROPERTIES of entities or propositions in a discourse (PRESUPPOSITION vs. ASSERTION) and the pragmatic RELATIONS between propositions and their elements (TOPIC vs. FOCUS). We use the term PRAGMATIC PRESUPPOSITION to refer to the set of entities or propositions evoked in a sentence which the speaker assumes the hearer already knows, or is ready to take for granted at the time of the utterance (‘the old information’). The term PRAGMATIC ASSERTION refers to the proposition expressed by a sentence that the hearer is expected to know or take for granted as a result of hearing the sentence uttered (‘the new information’). The FOCUS of a proposition is that element of information whereby the assertion differs from the presupposition. The TOPIC of a proposition is the given entity or situation relative to which a predication is to be assessed as relevant information. We will argue that the topic-focus and presupposition-assertion distinctions are crucial for explaining the behavior of parce que and we will show how the different IS parameters are formally coded in the sentence. 1.3.

Cognitive constraints on focus marking An important aspect of a theory of IS is the need to capture possible constraints on the respective scope of the presupposed and the asserted portions of a proposition. More specifically, such a theory should state any limitations on the amount of information that is pragmatically assertable within a single propositional unit. One significant principle of IS, much discussed in earlier literature, is the phenomenon of FOCUS PROJECTION (Höhle 1982, Selkirk 1984, and others). This principle states that an argument can project its focus value to its predicate, but not the other way around. In languages in which focus is marked prosodically, this is tantamount to saying that an unaccented predicate can be focal while an unaccented argument cannot. For example in (1) the focus scope of the accented object complement can be limited to that complement, as in (1a), or it can project from the complement to the verb itself, as in (1b): (1) a. Q: b. Q:

What did you see yesterday? A: What did you do yesterday? A:

I saw FOC[ELEKTRA]. I FOC[saw ELEKTRA].

Crucially, the different focus readings are not formally distinguished. The projection principle illustrated in (1) normally does not apply when the complement is an ADJUNCT. This is shown in the formal contrast between (2a) and (2b):

146

LAMBRECHT, BORDEAUX & REICHLE

(2) a. Q: A: b. Q: A:

What happened to the book? It FOC[fell on the wet FLOOR]. Why are you limping? I FOC[FELL], FOC[on the wet FLOOR].

In (2a), the PP on the wet floor is a directional argument of the verb to fall, hence it can project its focus on to this verb. Following Jacobs (1993), Wegener (2001), and other German scholars, we will refer to the projection reading in (2a) as an instance of INTEGRATION. In (2b), on the other hand, the locative complement is not an argument of the verb and integration cannot take place. Since in the given context both the predicate and its adjunct are in focus, each must receive its own focus marking and form its own intonation unit. The integration reading is of course possible in (b) (I FOC[fell on the wet FLOOR]), but the meaning would be markedly different. The situation is slightly more complex in sentences where a predication involves both an object and an adjunct complement, as in (3), a variant of (1): (3) a.

Q: A: b. Q: A:

Where did you see Elektra? I saw Elektra FOC[at the DRAFTHOUSE]. What did you do yesterday? I FOC[saw ELEKTRA], FOC[at the DRAFTHOUSE].

The response in (3a) is acceptable because the adjunct at the Drafthouse does not have to project its focus value on to the predicate, the proposition I saw Elektra being contextually presupposed. In (3b), the focus on the object Elektra is projected onto the verb saw, as in (1b), while the adjunct at the Drafthouse constitutes its own focus domain, as in (2b). The non-integration of the two focus domains is minimally signalled by an obligatory falling intonation contour on Elektra and in all likelihood a short pause. The integrated version of the answer in (3b) (FOC[I saw ELEKTRA at the DRAFTHOUSE]) would be pragmatically ill-formed, since it lacks the required break between the predicate and the adjunct, suggesting a contextually unacceptable integration reading. Integration is impossible because one cannot process a predication and an adjunct to that predication as a single piece of new information. On the basis of the facts illustrated in (2) and (3), we postulate the following universal cognitive constraint on pragmatic assertions, an extension of the principle of focus projection: (4) An adjunct to a predication cannot project its focus on the predicate to which it is adjoined.

COGNITIVE CONSTRAINTS ON ASSERTION SCOPE

147

For sentences involving parce que, this constraint finds a natural explanation: in order to inform an addressee of the cause or consequence of some situation, that situation must be previously represented in the addressee’s mind. Consider the sequence ‘p parce que q’ in (5): (5)

il galérait parce qu’il avait un problème de famille “he was in trouble because he had a family problem”

There are five theoretically possible pragmatic articulations of the proposition expressed in (5), which are summarized in (6).2 (Square brackets [...] indicate asserted propositions, parentheses (...) presupposed propositions; the TOP and FOC labels indicate a topic or focus relation between the labeled element and the global proposition; a comma indicates absence of integration.) (6) a. b. c. d. e.

FOC[il

galérait parce qu’il avait un problème de famille] galérait) FOC[parce qu’il avait un problème de famille] TOP(il galérait) FOC[parce qu’(il avait un problème de famille)] FOC[il galérait], FOC[parce qu’il avait un problème de famille] FOC[il galérait], FOC[parce qu’(il avait un problème de famille)] TOP(il

(6) suggests an important generalisation concerning the use of parce que: the expression parce que q is always focal. This pragmatic property distinguishes parce que most clearly from the causal conjunctions puisque and comme, which always introduce topical propositions.3 Among the pragmatic articulations in (6), the first and simplest one, involving complete integration of the two asserted clauses, is predicted by the cognitive constraint in (4) to be unacceptable. In the second articulation (6b), p and parce que q are integrated, forming a single intonation unit. Proposition p has a topic relation to the global proposition, i.e. the situation described by p is taken to be a topic under discussion. Being topical, p is necessarily also presupposed or hearer-old (one cannot explain the cause of a situation unless the addressee is aware of this situation). The adjunct proposition parce que q is pragmatically asserted, i.e. its function is to inform the addressee both of a hearer-new situation and of the fact that this situation is the cause of p. This asserted proposition has a focus relation to the global proposition, i.e. it provides the element of information which makes it possible for the utterance to constitute a piece of new information. Clear evidence for the presupposed 2

We ignore possible IS variations within p or q. For lack of space, we cannot discuss the issue of the theoretically possible sequence parce que q p (parce qu’il avait un problème de famille il galérait), which does not occur in the corpora we have consulted. 3

148

LAMBRECHT, BORDEAUX & REICHLE

status of p and the focus status of parce que q is provided by the fact that (b) could serve as an answer to the question Why was he in trouble?. The third articulation (6c) is identical to the second, except that q, which is asserted to be the cause of p, is itself presupposed or hearer-old. The pragmatic assertion thus consists here in the establishment of a causal relation between two known propositions (Lambrecht 1994). The representation of the fourth and fifth pragmatic articulations of (5) is self-explanatory. The utterance of (5) consists here in establishing a causal link between two asserted propositions. The difference between (d) and (e) is the same as that between (b) and (c): in the second member of the pair q is presupposed rather than asserted. Notice that, unlike (b) and (c), (d) and (e) could NOT serve as answers to the question Why was he in trouble?, since p is not presupposed. The basic IS difference between (b-c) on the one hand and (d-e) on the other can be informally paraphrased as follows: The reason for his being in trouble was that he had a family problem (integration) vs. He was in trouble, and the reason was that he had a family problem (non-integration). We should point out that the difference between (6b), where q is asserted, and (6c), where it is presupposed is only contextually induced, not grammatically coded. The same is true of the analogous difference between (d) and (e). We can therefore reduce the four pragmatic types in (6) to two formal categories, i.e. (b-c) on the one hand and (d-e) on the other. 2.

Analysis of corpus data Data from the Barnes/Blyth corpus present ample evidence for the existence of the four postulated information-structural articulations of p parce que q in (6), each including various subtypes. As predicted by our cognitive constraint (4), articulation (6a), involving complete focal integration of p and q, does not occur in the corpus. 2.1.

Integration of p and parce que q In sentences of the type illustrated in (6) (b) or (c), p and parce que q are integrated under a single intonation contour. Given the constraint in (4), this integration is only possible if either p or q is topical, hence presupposed. We have found only examples in which p is topical and parce que q focal, not the other way around, i.e. all attested examples are of the type where parce que q is capable of answering a WH-question introduced by pourquoi. Particularly clear examples of this focus articulation are utterances where the propositional content of p has just been uttered in asserted form, as in (7): (7) ah mais c'est dans mon cartable! (c’est dans mon cartable) [parce que je l’ai pris pour le montrer à Bill]

COGNITIVE CONSTRAINTS ON ASSERTION SCOPE

149

“but it’s in my bag! It’s in my bag because I took it to show it to Bill.” In (7), the speaker first asserts the proposition that something is in her bag and then immediately repeats this proposition, thereby marking it as presupposed. Being now presupposed, p can be integrated with q without any break. Our corpus contains a surprisingly large number of such examples. In other instances, the presupposed status of p must be inferred from the context, as in (8) (the speaker is talking about Mormon genealogical beliefs): (8) y a quelque chose que tu ne dois pas avoir dans tes ancêtres, je sais plus quoi (...) (moi je connais ça) [parce que euh mon père s'intéresse à la généalogie] “there’s something you can’t have in your ancestors, I don’t remember what (...) (I know that) because my father is interested in genealogy” The proposition I know that can be taken as presupposed in (8) because the speaker has just demonstrated her knowledge of the issue in question. While in most cases the presupposed status of p is merely contextually given or inferred, there are instances of formal presupposition-marking, as in the following example: (9) (si je suis encore étudiant) c’est [parce que je me, j’ai pas envie de m’installer dans cette vie professionelle] “if I’m still a student it’s because I, I don’t want to settle into this professional life” (9) is an instance of what Lambrecht (2001) calls the ‘If-because-Cleft construction’, a little-known subtype of WH-cleft, whose existence is motivated by the lack of a causal headless relative marker in the language (if = “the reason why”). It is well-known that the subordinate clauses of WH-Clefts are marked as being presupposed. 2.2.

Non-integration of p and parce que q As mentioned earlier, for the non-integration reading of (5) to be acceptable, p and parce que q must be separated by some grammatical signal of discontinuity. We will informally refer to this discontinuity as a BREAK, to allow for both prosodic and syntactic manifestations. We argue that the pragmatic function of this break is to allow the hearer to reinterpret the justuttered asserted proposition p as being now presupposed, thereby preventing the disallowed integration reading in (6a). Ignoring the presuppositional difference between (6) (d) and (e), the non-integration readings of (5) can be

150

LAMBRECHT, BORDEAUX & REICHLE

represented as in (10), where (p’) represents the reinterpretation of the asserted p as presupposed: (10)

FOC[p]

(p’) FOC[parce que q]

The break in (p’) is minimally signalled by an utterance-final falling intonation contour on p, followed in most cases by a pause and/or different types of lexical material. The most common formal realizations of (p’) are schematically represented in (11), in order of increasing lexical explicitness: (11) a. b. c. d. e. f. g.

FOC[p]

(..) FOC[parce que q] (marker) FOC[parce que q] FOC[p] (turn) FOC[parce que q] FOC[p] (c’est) FOC[parce que q] FOC[p] (oui/non) FOC[parce que q] FOC[p] (pourquoi?) FOC[parce que q] FOC[p] (p’) FOC[parce que q] FOC[p]

(p’ = pause) (p’ = discourse marker) (p’ = intervening turn) (p’ = anaphoric c’est) (p’ = ‘yes/no’ with covert p) (p’ = ‘why’ with elliptical p) (p’ = overt repetition of p)

Notice that (11) can be made to capture the above-mentioned metalinguistic usage of parce que by replacing p’ with a metalinguistic comment (p, and the reason I’m saying p is because q). Type (11a), involving the formal manifestation of the break as a phonological PAUSE, is illustrated in (12): (12) [tu devrais de temps en temps y aller] (..) [parce que nous on allait regarder aussi] “you should go there from time to time..because we used to go to look too” The pause between p and parce que q is a conventional signal to the interlocutor to reinterpret the asserted proposition p as presupposed. The corpus contains a large number of instances of this type. Examination of the audio recordings of the data attests each time to the presence of a pause between p and parce que q. Type (11b), involving the use of a DISCOURSE MARKER, is illustrated in (13) (see Debaisieux (1994) for further discussion): (13) [enfin pour moi c’est l’idéal] (hein) [parce que personne à la dans ma maison a de voiture] “for me it’s really ideal, right? because no one in, in my house, has a car” The discourse marker hein (in (13) has the same reprocessing function as the pause in (12).

COGNITIVE CONSTRAINTS ON ASSERTION SCOPE

151

An obvious sign of a break between p and parce que p is the presence of an INTERVENING TURN in the conversation: (14) A: [moi dans ce jeu, j'avais qu'une peur, c'est que la balle me tombe entre les mains et qu'il fallait que je lance..] B: (A ha!) A: [parce que je savais pas lancer la balle] A: “in this game, I had only one fear, that the ball would slip through my hands and that I would have to throw it .. B: A ha! A: because I didn’t know how to throw the ball.” Speaker B’s interjection is an acknowledgement of A’s completed assertion, making parce que q a separate utterance. A very common way of marking p as presupposed is by repeating it in ANAPHORIC form. This is typically done via the anaphoric construction c’est ‘it is’, as illustrated in (15): (15) [c’est cher, quand tu achètes de l’or ou de l’argent en France] mais (c'est) [parce que c’est plus pur] “it’s expensive, when you buy gold or silver in France, but that’s’s because it’s more pure” The pronoun ce in c’est refers anaphorically to the asserted p expressed in the immediately preceding discourse (here the proposition buying gold or silver in France is expensive), thereby making p necessarily presupposed. Another example is (16): (16) [pourquoi est-ce que un Belge se précipite toujours aux fenêtres quand il a euh, quand y a de, de l'orage, quoi ?] (c'est) [parce que euh, il croit qu'on est en train de le photographier] “why does a Belgian man always rush to the windows when, hmm, when he has, hmm, when there’s a thunderstorm, ok? it’s because he thinks someone is taking his picture” The occurrence of the break is strictly speaking unnecessary in (16), since parce que q is an answer to a WH-question, making p an independent speech act. Yet no redundancy is perceived, since c’est does not refer to the entire WH-question but to the presupposed open proposition it contains (a Belgian man always rushes to the windows when there’s a thunderstorm). Another way of covertly repeating p in order to make it presupposed is by using the anaphoric expressions YES or NO. Consider example (17):

152

LAMBRECHT, BORDEAUX & REICHLE

(17) [la deuxième à droite, c’est ça] mm mm (oui) [parce que c’est jamais des blocks] “the second on the right, that’s it. Mm yes because it’s never blocks” The use of oui in (17) evokes the proposition ‘it’s the second on the right’, thereby forcing its reinterpretion as a presupposition. A common anaphoric strategy for turning p into a presupposition is the use of pourquoi, nicely shown in the following example: (18) [les premiers parfumés, c'est les nobles] (pourquoi) [parce qu'ils puaient, les nobles!] Ils se lavaient jamais “the first perfumed people were the noblemen, why, because they stank, the noblemen! they never bathed” In (18), when the speaker asks pourquoi? she is in effect elliptically restating the proposition p (‘the noblemen were the first perfumed people’). Our corpus contains numerous examples of this anaphoric strategy. Finally, our corpus contains examples in which p’ is a more or less complete overt REPETITION of p, as shown earlier in example (7), Since this type involves two more or less identical versions of the same clause, it is a matter of perspective whether the phenomenon should be described here or in the earlier section on integration. While in Section 3.1.1 we were interested in the question of why within the structure [FOC[p] TOP(p’) FOC[parce que q]] p’ and parce que q are integrated, in the present section we are interested in why p and parce que q are non-integrated within the same structure. The data are the same and the reader is asked to refer back to example (7). 3.

A special case There exists an IS articulation that differs slightly from the two types discussed above: that is the PARENTHETICAL use of parce que. Consider (19): (19)

[on paye [parce qu'on est caretaker] 185 par mois] “we pay, because we’re caretakers, 185 a month” The proposition parce que q in (19) is pronounced with a parenthetical low intonation contour, indicating that the proposition is to be interpreted as APPOSITIVE. Appositive constituents can be presupposed or asserted (they permit the insertion of either ‘as you know’ or ‘by the way’). In the utterance context of (19), the appositive proposition was asserted. (19) is thus not a counterexample to our generalization that parce que q is always focal. The parenthetical parce que q in (20) is a kind of afterthought assertion:

COGNITIVE CONSTRAINTS ON ASSERTION SCOPE

153

(20) il a passé la tête à travers la vitre qui était ouverte et (quand le camion a fait un tonneau) .. [parce qu'il a fait un tonneau] .. [tout le poids de la cabine lui est tombé sur le cou et l'a décapité] “he stuck his head through the window that was open and when the truck rolled over, because it rolled over, all the weight of the cab fell on his neck and decapitated him” In (20), the speaker realizes that she incorrectly marked the proposition expressed in the quand-clause as presupposed (Le Draoulec 1997), when in fact it was not previously known to the hearer. The speaker therefore parenthetically adds the proposition in asserted form, in an attempt to repair the mistake. 4.

Conclusion Our corpus-based analysis of the use of parce que has demonstrated that parce que q is either the focus of a larger proposition in which p is presupposed (as in the cases of integration in 3.1), or it expresses an independent assertion (as in 3.2), due to the effect of a universal cognitive constraint against focus projection from adjunct position. In all instances, parce que q has a focus relation to the global proposition in which it occurs. References Debaisieux, Jeanne-Marie. 1994. Le fonctionnement de parce que en français parlé contemporain. Description linguistique et implications didactiques. Thèse nouveau régime, Université de Nancy II. Declerck, Renaat. 1997. When-clauses and temporal structure. London: Routledge. Green, Georgia. 1976. “Main clause phenomena in subordinate clauses”. Language 52.382-397. Groupe λ-1. 1975. “Car, parce que, puisque”. Revue Romane 10:2. 248-280. Höhle, Tilman. 1982. “Explikationen für ‘normale Betonung’ und ‘normale Wortstellung”. Satzglieder in Deutschen, ed. by Werner Abraham, 75154. Tübingen: Gunter Narr Verlag. Jacobs, Joachim. 1993. “Integration”. Wortstellung und Informationsstruktur, ed. by Marga Reis, 63-116. Tübingen: Niemeyer. König, Ekkehard & Johan van der Auwera. 1988. “Clause integration in German and Dutch conditionals, concesive conditionals, and concessives”. Clause combining in grammar and discourse, ed. by John Haiman & Sandra Thompson, 101-134. Amsterdam: John Benjamins.

154

LAMBRECHT, BORDEAUX & REICHLE

Lambrecht, Knud. 1994. Information structure and sentence form. Topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press. ----------. 2001. “A framework for the analysis of cleft constructions”. Linguistics 39:3.463-516. Le Draoulec, Anne. 1997. Etude présuppositionnelle des subordonnées temporelles. Thèse de doctorat, Université de Toulouse Le Mirail. Moeschler, Jacques. 1986. “Connecteurs pragmatiques, lois de discours et stratégies interprétatives : parce que et la justification interprétative”. Cahiers de Linguistique 7:149-167. Université de Genève. Prince, Ellen F. 1992. “The ZPG letter: Subjects, definiteness, and information-status”. Discourse description: Diverse linguistic analyses of a fund-raising text, ed. by William C. Mann & Sandra Thompson, 295326. Amsterdam: John Benjamins. Selkirk, Elisabeth O. 1984. Phonology and syntax: The relation between sound and structure. Cambridge, Mass.: MIT Press. Shimojo, Mitsuaki. 1999. “The interplay between grammar and discourse: The case of the Japanese topic marker wa in subordinate clauses”. Proceedings of the Western Conference on Linguistics, 1998, ed. by Elly van Gelderen & Vida Samiian, 425-439. Fresno, California State University. Wegener, Heide. 2001 “Integration und Nichtintegration von Satzkonjunkten im Deutschen und Französischen”. Sprachkontakt und Sprachvergleich, ed. by Gerda Hassler, 89-105. Münster: Nodues Publikationen.

AVANT QUE- OR AVANT DE-CLAUSES WHEN PRESUPPOSITION GIVES WAY TO AN ‘ASSERTIVE CONSTRUCTION’

*

ANNE LE DRAOULEC ERSS, UMR 5610 du CNRS / Université de Toulouse - Le Mirail

0.

Introduction In this paper, I wish to challenge the way in which certain constructions are regularly considered to be ‘presupposition triggers’. More precisely, my study will focus on the temporal clauses introduced by avant que or avant de in French (the equivalent of before in English). I will presuppose that ‘presupposition’ is a well-known linguistic notion, and I will only sketch the basic outlines of the traditional pragmatic approach. In a very rough summary, pragmatic definitions are formulated in terms of previous knowledge, shared by the speaker/writer and their interlocutors/readers. A well-known test is the ability of the presupposition, as part of the background, to resist the negation of the sentence in which it appears. Among the vast set of presupposition triggers, one of the best-studied cases is that of definite descriptions. Temporal clauses are also frequently mentioned. I give some simple examples (with temporal clauses in either sentence-initial or sentence-final position): (1) Quand elle a publié un roman, Marie a eu de bonnes critiques / Marie a eu de bonnes critiques quand elle a publié un roman “When she published a novel, Mary got good reviews / Mary got good reviews when she published a novel.”

*

I warmly thank Marie-Paule Péry-Woodley for her precious help with editing the English text.

156

ANNE LE DRAOULEC

(2) Après qu’elle a publié un roman, Marie a été très entourée / Marie a été très entourée après qu’elle a publié un roman “After she published a novel, Mary became the centre of attraction / Mary became the centre of attraction after she published a novel.” (3) Avant que Marie publie un roman, personne ne s’intéressait à cette histoire / Personne ne s’intéressait à cette histoire avant que Marie publie un roman “Before Mary published a novel, nobody was interested in this story / Nobody was interested in this story before Mary published a novel.” In a pragmatic view, the realization of the eventuality described in the temporal clause (whatever its position) must already be known to the interlocutor at the time of speech. The possibility of locating this eventuality (i.e. Marie a publié un roman for each of the above examples) on a temporal scale is necessary for the location of the main clause event. It must be specified that if the interlocutor/reader doesn’t know anything about Mary publishing a novel, they are supposed to act as if they knew it, and simply add it to the context as a piece of indirect information. The phenomenon is known under the name of ‘pragmatic accommodation’ of the presupposition. Some exceptions are recognized for the presupposition of temporal clauses. I will skip the case of avant que-clauses in a so-called ‘non-factual’ or ‘counterfactual’ interpretation (cf. Heinämäki 1974). In my view, this case can be reconsidered as coming within a presuppositional analysis (cf. Le Draoulec 2001). A real exception, quite an intriguing one, is the ‘quand inverse’ (“inverse when”) construction, after the phenomenon known in Latin grammars as ‘cum-inversum’. In this construction, the subordinate clause is necessarily postposed. Example (4) gives an illustration: (4) Marie dormait paisiblement, quand (soudain) la sonnerie du téléphone la réveilla “Mary was quietly sleeping, when (suddenly) the phone rang and woke her up.” In (4), there is no presupposition that the phone rang and woke Mary up. Obviously, the content of the temporal clause is not part of shared knowledge. And there is no way of resorting to accommodation: the piece of information cannot be regarded as having been provided indirectly, ‘in passing’. On the contrary, the emphasis is laid on the content of the subordinate clause, with an

AVANT QUE- OR AVANT DE-CLAUSES

157

effect of surprise which is linked to the unexpected interruption of the main clause eventuality in its unfolding. Suspense and surprise may be reinforced by an adverb such as soudain. It is of particular interest that ‘quand inverse’ has been largely recognized, not only as a non-presuppositional construction, but also as an assertive one. The assertive value of the ‘quand inverse’-clause is seen as precisely allowing the emergence of an effect of surprise. And it is also argued that its assertive value enables the clause to have an independent role in the advancement of the narration — hence the denomination of ‘quand narratif’ (“narrative when”) which is sometimes used instead of ‘quand inverse’. I will not dwell on the case of ‘quand inverse’, since it has already been the subject of numerous studies (cf. inter alia Olsson 1971, Declerck 1997, Vogeleer 1998). My aim is to show that the temporal conjunction avant que (or in fact much more often, as we shall see, avant de) also has the potential of introducing a non-presuppositional, assertive clause. As with quand inverse, a necessary condition is that the avant que- or avant de-clause is postposed: i.e., the respective position of main and subordinate clauses matches the temporal order in which the eventualities take place. I will designate the phenomenon I explore as ‘the assertive construction’ and illustrate it with example (5): (5) Le porte-parole a commencé par se réjouir au nom du gouvernement avant de se raviser et signifier s'être exprimé en son nom personnel. Puis, dans la soirée, le gouvernement s'est finalement félicité. (Daily Libération, 11th October 2003) “The spokesman first expressed satisfaction on behalf of the government before he changed his mind and indicated [/ before changing his mind and indicating]1 that he was speaking in a personal capacity. Later on, in the course of the evening, the government finally said they were pleased.” In (5), the subordinate clause avant de se raviser et signifier… does not play an ordinary role of temporal location, and its narrative behaviour appears quite close to that of a ‘quand inverse’ clause. But it does not display the remarkable property of a construction with ‘quand inverse’. There is no effect of surprise,

1

Whereas in French the infinitive construction (with avant de) proves nearly obligatory when there is no change of subject, English equally allows the gerund construction (before changing) or the finite verb construction (before he changed). One or the other will be selected in the glosses, according to what is intuitively felt to be more appropriate — though a difference in meaning may result, which would deserve investigation.

158

ANNE LE DRAOULEC

so it is not surprising that the phenomenon has not received any attention, and has, to my knowledge, never been mentioned in the literature about presupposition. In what follows, I will explore arguments for the hypothesis that in some cases (such as (5)), avant que or avant de can escape from the domain of presupposition, and allow a relation of temporal succession between two asserted clauses. I will also extend the study through an examination of configurations where the assertive construction is accompanied by specific semantic effects. 1.

Absence of presupposition As with ‘quand inverse’, the content of the avant de-clause in (5) does not appear to be part of shared knowledge; quite the opposite, it appears to convey information. However, since there is no surprise effect, it is less obvious why this information could not be retrieved through accommodation. The test of resistance to negation would be of no use here, for reasons linked to the scope of the negation over a subordinate clause. So in order to be sure that there is no presupposition, I will use another test for presupposition (cf. Le Draoulec 2005 & in press), a discourse-level test proposed by Ducrot (1991) as the ‘loi d’enchaînement’ (“continuation law”). Roughly summarized, this law stipulates that no presupposed material in a sentence A may be linked, whether implicitly or explicitly, to a following sentence B. For the case we are interested in (the case of sentences containing avant de-clauses), the continuation law would stipulate that if the content of the avant de-clause is presupposed, it cannot be connected to the next sentence. Only the main clause, or the whole unit constituted by the main clause plus the avant declause, can be the point of departure for such a connection. This law clearly does not apply to example (5). There is a link between the avant de-clause (avant de se raviser et signifier…), and the following sentence (Puis, dans la soirée, le gouvernement s'est finalement félicité): a narrative/contrastive link which is furthermore explicitly marked by both the connector puis and finalement. As I gradually became aware of the phenomenon, I felt that perhaps journalistic style had made the non-presuppositional use of avant de its specialty. But as a matter of fact, it appears in a great variety of sources. In (6) I give an example taken from a detective novel, where the non-applicability of the continuation law is as obvious as in (5):

AVANT QUE- OR AVANT DE-CLAUSES

159

(6) Oleg l’empoigna à bras-le-corps, comme prévu. Et le précipita par-dessus bord, avant de remettre le moteur en route. L’hélice traça aussitôt son sillon […]. (Thierry Jonquet, Ad Vitam Aeternam) “Oleg grabbed him by the body […]. And threw him overboard, before starting the engine again. The propeller immediately marked out its furrow […].” In (6), there is again a clear link between the avant de-clause (avant de remettre le moteur en route) and the succeeding sentence (L’hélice traça aussitôt son sillon). 2.

Assertive value What remains to be examined is whether the subordinate clause is asserted. It must indeed be made clear that non-presuppositional value does not necessarily go hand in hand with assertive value. For instance, the subordinating conjunction parce que (“because”) is not a presupposition trigger; but it doesn’t follow that the non-presupposed clause it introduces is an assertion. For quand inverse, the assertive property of the clause is commonly accepted without any difficulty: the obvious narrative role combined with the ability to generate a surprise effect gives sufficient arguments for the assertion analysis. For the above examples with avant de, things are not so clear, again due to the lack of creating a surprise effect. So what must be explored is the narrative function of the avant de-clauses in these examples. For this I will appeal to the notion of reference time (RT), or more precisely to the difference between main clause RT and RT in an ordinary temporal clause, as it is developed by Hamann (1989). 2.1.

Reference time and progression of the narration Very briefly, according to Hamann, only a main clause can have a ‘current’ RT in the sense of Partee (1984), i.e. an RT with a narrative function. Still according to Hamann, a temporal subordinate clause is also endowed with an RT, but as the clause is presupposed, its RT is only ‘subsidiary’. ‘Subsidiary’ means that it only contributes to the calculation of the main clause current RT; but it never appears itself as a current RT. In other words, the temporal clause does not take part in the advancement of the narration. Now, it is obvious that Hamann’s analysis, which is meant for ordinary temporal clauses, would not apply in the case of examples (5) or (6). The various eventualities described in the succession of clauses constitute links in a

160

ANNE LE DRAOULEC

single chain: in the same way as the other clauses (main clauses), the temporal clauses avant de se raviser et signifier… or avant de remettre le moteur en marche do take part in the updating of the current RT, and thus in the progression of the narration. And this narrative role2 is a strong argument in favour of their assertive nature. In fact, as well as being seen as an indication of the nonpresuppositional nature of our avant de-clauses, the non-applicability of the law preventing continuation on presupposed material may now be linked to the reference time issue: if the subordinate clauses are endowed with a current RT, which is characteristic of an assertion status, it is to be expected that they can be the starting point of a relation with the next sentence. So from now on, we can assume that there is a regular association between the non-presuppositional nature of our avant de-clauses, and their narrative, assertive nature. Example (7) is one more example where the avant de-clause has a clear role in the advancement of the narration, on a par with the surrounding syntactic main clauses: (7) Charlie gigota […]. Il ouvrit un œil, le referma, à plusieurs reprises, avant de se décider à consulter sa montre. Il était plus de midi. Il se leva, sortit de son repaire […] (Thierry Jonquet, Moloch) “Charlie moved about […]. He opened an eye, closed it again, several times, before deciding to look at his watch […].” Example (8) is not directly inserted in the narrative thread: (8) Pendant six heures, le candidat de l’opposition s’est cru élu, avant d’être déclaré battu. (Daily Le Monde, 22nd February 2005) “For six hours, the opposition candidate thought he had been elected, before he was declared beaten.” However, I regard it as an exemplary example insofar as it is the highlighted summary-sentence set apart from a newspaper article, and corresponding perfectly to the split between the two high points of the text. The text first exposes at length the details of the supposed victory of the opposition candidate. One can read, for instance, that (translated from French), “For six hours, glass in hand, he celebrated the end of seventeen years of social-democratic government in Schleswig-Holstein.” There is then a textual

2

Similar to the role which is commonly granted to quand inverse-clauses.

AVANT QUE- OR AVANT DE-CLAUSES

161

shift, with the announcement of the defeat (which will be developed afterwards): “Shortly before midnight, once the ballot papers had been counted twice, it appeared that the CDU-FDP coalition only had 34 seats left.” So it is quite obvious that in (8), which synthetically and accurately renders the movement of the article, the ‘second phase’ avant de-clause (avant d’être déclaré battu) has the same essential narrative role as the ‘first phase’ main clause. 2.2.

Parallelism with temporal connectors A further remark to be made is that in examples (5) to (8) the temporal subordinate conjunction plays the same role as a connector between two 3 asserted clauses. The substitution of avant de with succession connectors such as puis, ensuite, après (“then”, “afterwards”), etc. would not raise any difficulty (one just needs to conjugate the infinitive verb). For instance, the paraphrase of (6) by (9) gives a perfectly acceptable result: (9) Oleg l’empoigna à bras-le-corps. Et le précipita par-dessus bord, puis remit le moteur en route “Oleg grabbed him by the body. And threw him overboard, and then started the engine again.” Avant de remettre le moteur en route in (6) allows a similar interpretation of on-going narration as puis remit le moteur en route in (9). The possibility of substitution cannot be regarded as a proof in itself. Nevertheless the very closeness of interpretation between (6) and (9) is an element which reinforces the validity of the assertion analysis. According to a native English informant, the English gloss in (9) would in fact be more natural than that in (6), with then instead of before. This raises questions as to the assertive behaviour of before-clauses: I won’t venture into a comparative analysis here, but an exploration of this point of contact between syntax and discourse in different languages would be a worthwhile endeavour. 2.3.

Avant de versus avant que in a perspective of speech act autonomy As assertive clauses with their own current RT, our avant de-clauses are very close to independent utterances, i.e., utterances with a speech act

3

I adopt a minimal definition of temporal connectors, whereby they will simply be characterized by their ability to relate two clauses of equal — assertive — standing (cf. Vogeleer 1998: 84).

162

ANNE LE DRAOULEC

autonomy. A factor which argues for this autonomy is the extremely frequent (though not compulsory, cf. example (5)) separation between main and subordinate clauses by a punctuation mark. Nearly all the cases of assertive avant de-clauses I have encountered are preceded by a comma. The separation can even be marked by a full stop (resulting in an inter-sentences subordination). More surprising, as regards the question of autonomy, is the overwhelming majority of avant de- over avant que-clauses with an assertive interpretation. Until now, all my examples contain infinitive constructions. This does not imply that there is no possibility of an assertive construction with avant que. Consider for instance example (10), where the avant queclause clearly participates in the forward movement of the narration: (10) Je me sens rougir jusqu'à la racine des cheveux, avant que l'implacable pharmacienne n’ajoute : - Vous ne la connaissez pas? (Françoise Dorin, Les Vendanges tardives) “I feel myself blushing to the roots of my hair, before the implacable pharmacist goes on: — don’t you know her?” I was not however able to find many such examples with avant que. What is strange about this is that a restricted autonomy is generally attributed to non-finite clauses (cf. for instance Lakoff 1984). It is therefore puzzling that infinitival avant de-clauses should be more easily endowed with a speech act autonomy than finite avant que-clauses. A first element of explanation for this phenomenon is the general disproportion between the use of avant de and avant que, whatever the interpretation. There is no space here to present a quantitative comparison, but avant de is much more frequent than avant que (cf. Le Draoulec 2005). Thus it would be statistically consistent that within the small set of occurrences with an assertive construction, the ones with avant que should be quite rare. This statistical hypothesis requires further evaluation. It would in any case be unlikely to provide a complete explanation. As a matter of fact, the semantic subject continuity implied by the infinitive construction reinforces the ‘discourse topic’ continuity. I hypothesise that such continuity supports the possibility that main and subordinate clauses have equivalent status, in a succession of episodes that make the action progress on a same theme. Of course, all this remains to be examined in greater depth. 3.

Specific semantic effects Up to now, I have emphasized that the assertive construction with avant de (or avant que) is not accompanied by some remarkable effect such as

AVANT QUE- OR AVANT DE-CLAUSES

163

the surprise effect linked to the use of quand inverse. However, I want to qualify my statement, and show that avant de/que also allows the emergence of specific semantic effects which, though not systematically implied, clearly appear in certain types of discourse configurations. Consider example (11): (11) On n'a qu'à regarder pour voir que l'être humain est divisé en deux parties symétriques dans le sens de la longueur. Et, à l'intérieur, de même notre fonction est desservie par des organes gémeaux [...]. Mais avec le cou, […] toute cette architecture aux étages successifs aboutit à l'unité. Avant que la cervelle avec son double lobe, le visage avec ses deux yeux, ses deux oreilles, [...] ne donne à notre dualité le sceau suprême de la composition (Paul Claudel, Commentaires et exégèses) “You only have to look to see that human beings are divided lengthwise into two symmetrical parts. And inside, our internal workings are in the same way served by twin organs […]. But with the neck, […] all this architecture with its successive levels comes into one. Before the brain with its twin lobes, the face with its two eyes, its two ears, […] gives our duality the supreme seal of composition.” This example prompts me to come back to the case of quand inverse, and in particular to the ‘pre-eminent’ or ‘salient’ status generally granted to the situation described in the quand inverse-clause. Assertive avant de/que-clauses can be re-examined in similar terms. In (11), the avant que-clause can be regarded as particularly pre-eminent, as far as this characteristic is explicitly (lexically) stressed by the expression le sceau suprême de la composition. However, it is not necessary that the pre-eminent characteristic of the eventuality should be made explicit. In (10) for instance, we can now also see an emphasis on the content of the subordinate clause. This effect seems to be primarily linked to the assertive construction with avant que: without any knowledge about the context of the example, one understands that what the implacable pharmacienne adds appears as a climax in the emotional ordeal endured by the je. Of course, certain semantico-pragmatic factors contribute to such an interpretation: the intense blushing which precedes, the adjective implacable which qualifies the pharmacist. But as evidence that the preeminence or salience effect really is linked to the construction with avant que, compare (10) with (12):

164 (12)

ANNE LE DRAOULEC

Je me sens rougir jusqu'à la racine des cheveux, puis l'implacable pharmacienne ajoute […] “I feel myself blushing to the roots of my hair, then the implacable pharmacist goes on […]”

In (12), where avant que is replaced by puis, we lose the increase in intensity that characterizes example (10). Further support for this observation is provided by the following example, taken from a short biography of the actress Jeanne Balibar: (13) L’élève exemplaire s’est d’abord grisée de danse […] puis de littérature, de poésie et d’histoire à Normale Sup, Cambridge, Oxford, avant de s’autoriser à réussir le Conservatoire, en 1992. Trois mois plus tard, elle est engagée à la Comédie-Française [...] (Weekly magazine Télérama, 13-19 December 2003). “The model student first immersed herself into dance […] then into literature, poetry and history at Normale Sup, Cambridge, Oxford, before allowing herself to pass the entry examination to the Conservatoire, in 1992. Three months later, she was taken on at the Comédie-Française [...].” The eventuality réussir le Conservatoire (and even more so s’y autoriser, a sign of utmost ease) can be regarded in itself – and through world knowledge – as a height of success in the career of the actress. But above all it is presented as such linguistically: it is because it is introduced by avant de that this eventuality, rather than another (for instance, être engagée à la ComédieFrançaise), acquires its pre-eminent status in (13). Here again, with puis (puis elle s’est autorisée à réussir le Conservatoire, “then she allowed herself to pass…”), the salience effect would disappear, and only the intrinsic importance of the event would remain. There is no space to develop this point any further here, but it would be worthwhile pursuing the analysis, and clarifying the conditions in which the use of avant de/que is associated with a pre-eminence effect. Furthermore, another characteristic potentially associated with avant de/que remains to be explored: the ability to mark a concluding, or final effect – as was the case in example (11), where the assertion of the duality of human beings obviously corresponds to the author’s conclusion.

AVANT QUE- OR AVANT DE-CLAUSES

165

4.

Conclusion First of all, I must acknowledge that I have simplified the facts by acting as if there were a clear distinction between the two types of analysis for avant de/avant que: on the one hand, the classic presuppositional analysis, on the other hand, the assertion analysis which is the object of my investigation. But it is an oversimplification, insofar as there are many examples which constitute indeterminate, under-specified cases, and for which it is very difficult to decide whether they fall within one or the other category of analysis. This point will simply be sketched out here through one single example: (14) John O’Neill, un agent du FBI, a enquêté pendant huit ans sur le réseau terroriste Al-Qaïda, avant d’être poussé à la retraite. (Télérama, 6-12 September 2003 – end of a documentary summary) “John O’Neill, an FBI agent, investigated the Al-Qaeda terrorist network for eight years, before he was forced to retire.” Here, according to my research, no parameter is sufficiently decisive to favour one or the other interpretation (no link with a succeeding sentence, no obvious salience effect). So I have to admit that the hypothesis I put forward is in fact based on ‘well selected’ examples, which has at once the disadvantage of leaving in the shadow a large part of the problem, and the advantage of presenting characteristics strong enough to be exploitable. Now, on a more general level, some very last remarks can be made. How is it that a normally presupposing construction is able to become a nonpresupposing, assertive one? And most of all, since such a phenomenon exists, and is to be taken into account, the question of presupposition becomes even more complex than it already appeared. As a matter of fact, avant de/avant que is not the only case at stake. There is the well-known exception of quand inverse. But there is also the case of the temporal conjunction jusqu’à ce que (“until”), which normally belongs to the set of presupposition triggers, but which in fact is able to introduce an assertive clause (cf. Le Draoulec in press). The exceptions could also be extended to constructions other than temporal clauses. There is for instance the case of cleft constructions, which are traditionally regarded as presupposition triggers. Yet in some constructions with c’est alors que, as in example (15), the clause which follows c’est alors que is emphasized, and can in no way be considered as presupposed. Here again, there is an assertion:

166 (15)

ANNE LE DRAOULEC

La campagne était sereine. C’est alors qu’un coup de tonnerre déchira le silence. “The countryside was serene. It was then that a clap of thunder tore the silence.”

As a result, the possibility of an assertive construction for avant de/avant que must not be regarded as trivial, anecdotal. As the number of ‘exceptions’ becomes greater and greater, one may ask if they really are ‘exceptions’ which can be left aside, treated apart. And so the problem of presupposition, which has already been the subject of numerous studies, should be revisited, by taking into account the sensitivity of so called ‘presupposition triggers’ in relationship to discourse context. References Declerck, Renaat. 1997. When-clauses and temporal Structure. London: Routledge. Ducrot, Oswald. 1991. Dire et ne pas dire. Paris: Hermann. Hamann, Cornelia. 1989. “English temporal Clauses in a Reference Frame Model”. Essays on Tensing in English, II: Time, Text and Modality, ed. by Alfred Schopf, 31-154. Tübingen: Niemeyer. Heinämäki, Orvin. 1974. Semantics of English temporal connectives. Doctoral dissertation, University of Texas at Austin. Lakoff, Robin. 1984. “The Pragmatics of Subordination”. Proceedings of the 10th Annual Meeting of the Berkeley Linguistics Society, 481-492. Le Draoulec, Anne. 2001. “Les différents types d’interprétation des subordonnées introduites par avant que”. Adverbial Modification, ed. by Reineke Bok-Bennema, Bob de Jonge, Brigitte Kampers-Manhe & A. Molendijk, 157-168. Amsterdam & Atlanta: Rodopi. ----------. 2005. “Avant que / de : possibles passages à la connexion temporelle”. Journal of French Language Studies 15. 131-151. ----------. in press. “De la subordination à la connexion temporelle”. Cahiers Chronos. Olsson, Lars. 1971. Etude sur l’emploi des temps dans les propositions introduites par quand et lorsque et les propositions qui les complètent en français contemporain. Uppsala: Acta Universitatis Upsaliensis. Partee, Barbara. 1984. “Nominal and temporal anaphora”. Linguistics and Philosophy 7. 243-286.

AVANT QUE- OR AVANT DE-CLAUSES

167

Vogeleer, Svetlana. 1998. “Quand inverse”. Revue québécoise de linguistique 26:1. 79-101.

NULL DIRECTIONAL PREPOSTIONS IN ROMANIAN AND SPANISH

JONATHAN E. MACDONALD Stony Brook University

0.

Introduction The main goal of this paper is to provide a formal account of nonreflexive non-argumental clitic pronouns of Spanish (1) and reflexive nonargumental clitic pronouns of Romanian (2).1 The clitics are in bold. (1)

(Yo) le lavé el coche (I) to-him washed the car “I washed his car” or ‘I washed the car on/for him.”

(2)

Eu mi-am spălat maşina2 I myself-have washed car-the “I washed my/the car.”

I will focus on five properties associated with these constructions: 1. the inability to produce a telic interpretation of a predicate; 2. the lack of on/with entailment; 3. the ability to express temporary relations; 4. the ability to prevent idiomatic interpretations; and 5. the resistance to adjectival secondary predication. The Spanish reflexive shows all five properties; the Romanian reflexive shows the first four. To account for these properties, I propose that both the Spanish nonreflexive and the Romanian reflexive are introduced as the complement of a null directional-like preposition that merges as a complement of the verb (3). 1

I refer to non-reflexive non-argumental clitic pronouns as ‘non-reflexives’, and to reflexive non-argumental clitic pronouns as ‘reflexives’. 2 Romanian also has clitics in accusative case. The accusative clitics show fundamentally different properties from the dative clitics. Throughout the paper, all of the Romanian clitics are dative, unless otherwise noted.

JONATHAN E. MACDONALD

170

The directional-like nature of the null preposition is motivated by a set of properties shared with overt directional prepositions. (3)

…V’ ri V PP wo P DP Ø DIR 4 Span. non-refl./Rom. refl.

To motivate the structure in (3), it is instructive to first consider the properties of Spanish reflexive non-argumental pronouns. Therefore, in Section 1, I outline a discussion of Spanish reflexives from MacDonald (2004); he shows that reflexives pattern with goal prepositions. In Section 2, I compare Spanish reflexives with Spanish non-reflexives; I show that non-reflexives pattern with directional prepositions. In Section 3, I show that Romanian reflexives pattern with Spanish non-reflexives. In Section 4, I conclude by briefly considering the question of why Spanish reflexives pattern differently from Romanian reflexives and Spanish non-reflexives. 1.

The properties of Spanish reflexives In this section, I discuss properties of Spanish reflexives. The discussion is taken from MacDonald (2004), in which he shows that Spanish reflexives share three properties with goal prepositions: 1. they produce a telic interpretation of the predicate; 2. they express an on/with entailment; and 3. they can prevent idiomatic interpretation. This section provides a starting point for illuminating the properties associated with Spanish non-reflexive constructions. 1.1.

Telic interpretation of the predicate The presence of a Spanish reflexive results in a telic interpretation of a predicate (MacDonald 2004, Nishida 1994, Zagona 1996). As evidence, observe the (in-)compatibility of the durative phrase (e.g. durante una hora “for an hour”) in (4). (4) a.

Lavé el coche durante una hora. washed the car for an hour “I washed the car for an hour.”

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

b.

171

Me lavé el coche # durante una hora myself washed the car # for an hour “I washed my/the car for an hour.”

Durative phrases are compatible with atelic predicates and incompatible with telic predicates (Dowty 1979, Pustejovsky 1991 among others).3 When no reflexive is present (4a), the durative is compatible. When the reflexive pronoun is present (4b), the durative is incompatible. The reflexive produces a telic interpretation. Compare the aspectual effects of the reflexive in (4) to the aspectual effects of the goal preposition in (5). (5) a. b.

John carried a bag for an hour John carried a bag into the bedroom # for an hour.

When no goal preposition is present (5a), the durative is compatible. However, when the goal preposition is present (5b), the durative is no longer compatible. The goal preposition produces a telic interpretation as well. Both goal prepositions and reflexives produce a telic interpretation of a predicate. 1.2.

Expressing an on/with entailment MacDonald (2004) observes that in the presence of a reflexive, there is what he terms an on/with entailment. Consider an utterance in which there is no reflexive (6). (6)

Abroché la camisa buttoned the shirt “I buttoned my/the shirt.”

(6) can be said in a context in which I am wearing the shirt while buttoning it, or in which I am not wearing the shirt while buttoning it; i.e. it could be on a hanger. Contrast (6) with (7) in which there is a reflexive present.

3

This is a bit of a simplification of the generalization. Durative phrases are incompatible with telic predicates on a single event interpretation of the predicate. Often the presence of the durative phrase with a telic predicate results in an iterative interpretation of the predicate. See MacDonald (2006b and references therein) for a more detailed discussion of the aspectual role that durative phrases play.

172 (7)

JONATHAN E. MACDONALD

Me abroché la camisa myself buttoned the shirt “I buttoned my/the shirt.”

(7) can only be said in a context in which I am wearing the shirt at the time of buttoning. (7) cannot be said in a context in which the shirt is on a hanger. In the presence of the reflexive, the shirt is necessarily interpreted as on the denotation of the subject. Consider another example in (8). (8)

Lavé el coche washed the car “I washed my/the car.”

In (8) there is no reflexive and the utterance can be said in a context in which I washed the car myself, or in a context in which I dropped the car off at a carwash for someone else to wash. Contrast (8) with (9) in which the reflexive is present. (9)

Me lavé el coche myself washed the car “I washed my/the car.”

(9) can only be said in a context in which I washed the car myself. It cannot be said in a context in which I dropped the car off at a carwash for someone else to wash. In the presence of the reflexive pronoun, the car is necessarily interpreted as with the subject. In the presence of the reflexive, there is an on/with entailment. MacDonald (2004) analyzes this on/with entailment of the reflexive as the same entailment elicited by a goal preposition. Consider the utterances in (10). (10) a. b.

Frank threw a ball to Bill. John carried a bag into the bedroom.

In (10a) the ball necessarily arrives at Bill. In (10b), the bag necessarily ends up in the bedroom.4 The object must end up on/with the 4

An anonymous reviewer observes that in the progressive aspect or the future tense, this entailment does not necessarily hold: John was carrying a bag into the bedroom (when I saw him). John will carry the bag into the bedroom. The on/with entailment seems to be a potential reading of the goal preposition restricted to certain environments.

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

173

complement of the goal preposition. Goal prepositions and reflexives express an on/with entailment. 1.3.

Preventing idiomatic interpretation Finally, MacDonald (2004) observes that the presence of a reflexive can prevent idiomatic interpretation.5 Consider the Spanish idioms in (11). (11) a.

b.

Jose cortó los lazos con Ana Jose cut the ties with Ana “Jose broke up with Ana.” Juan come la sopa boba Juan eats the soup stupid “Juan is a sponger.”

Observe that when a reflexive pronoun is added to the utterances in (11), their idiomatic interpretation is lost. This is illustrated in (12). (12) a. # Jose se cortó los lazos con Ana. b. # Juan se come la sopa boba. Observe that goal prepositions can prevent idiomatic interpretation as well (13). (13) a. # John kicked the bucket to the barn. b. # John spilled the beans into the sink. Goal prepositions and reflexives can prevent idiomatic interpretation.6 1.4.

The structure of Spanish reflexive pronouns Given the properties in common with goal prepositions, MacDonald (2004) proposes that the Spanish reflexive is introduced as a complement of a null goal-like preposition that merges as the complement of the verb. This is illustrated in (14).

5

De Miguel and Fernández Lagunilla (2000) make a similar observation. Goal prepositions and reflexives do not always prevent idiomatic interpretation: John spilled the beans to the police; sin comer(se)lo ni beber(se)lo “without deserving it”. This ability depends in part on the semantic contribution of the additional element (see Nunberg et al. 1994). MacDonald (2006a) discusses the effect of Spanish reflexives (and non-reflexives) on idiomatic interpretation. See footnote 11. 6

174 (14)

JONATHAN E. MACDONALD …V’ ri V PP ru P DP Ø GOAL 4 Span. refl

MacDonald (2004) accounts for the range of properties associated with the reflexive in the following way. The property of an overt goal preposition that allows it to produce a telic interpretation of a predicate is also present on the null goal-like preposition.7 The on/with entailment results from a property of the null preposition that contributes to a compositional theta-role assigned to the direct object by the verb and prepositional phrase,8 such that the object is interpreted as on/with the denotation of the complement of the goal-like preposition. That is, the object is interpreted as on/with the denotation of the reflexive and given that the reflexive is coindexed with the subject, the object is interpreted as on/with the denotation of the subject. Finally, MacDonald (2004) argues that the ability to prevent idiomatic interpretation results from the low merger of the null preposition. Typically, only elements merged within the verb phrase can affect idiomatic interpretation (Marantz 1984 and Svenonius 2004). Let us now consider the properties of Spanish nonreflexives. 2.

The properties of Spanish non-reflexives In this section, I discuss Spanish non-reflexives. Non-reflexives pattern with directional prepositions in four ways: 1. they do not produce a telic interpretation of the predicate; 2. they do not express an on/with entailment; 3. they can prevent idiomatic interpretation and; 4. they resist adjectival secondary predication. As such, I propose that the non-reflexive is introduced as a complement of a null directional-like preposition that merges as a complement of the verb. I make these properties of the non-reflexive salient by contrasting them with the properties of the reflexive. I argue that the differences between the reflexive and the non-reflexive result from a distinct set of properties associated with the null prepositions introducing them. I supply more evidence

7

Directional prepositions lack this property. See (16) below. Larson (1988) argues for compositional theta-roles assigned by a verb plus prepositional phrase. 8

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

175

for the distinct set of properties by drawing attention to yet another contrast between these constructions in Section 2.2. 2.1.

A comparison of Spanish non-reflexives and Spanish reflexives As we saw in Section 1, Spanish reflexives pattern with goal prepositions in the following ways: 1. they produce a telic interpretation; 2. they express an on/with entailment; and 3. they can prevent idiomatic interpretation. We will see that Spanish non-reflexives pattern with directional prepositions in the following ways: 1. they do not produce a telic interpretation of the predicate; 2. they do not express an on/with entailment; and 3. they can prevent idiomatic interpretation. Consider the utterance in (15). (15) (Le) lavé el coche durante una hora (to-him) washed the car for an hour “I washed his car for an hour.” or “I washed the car for him for an hour.” The non-reflexive does not affect the compatibility of the durative. The non-reflexive does not produce a telic interpretation. Observe that directional prepositions pattern in the same way (16). (16) a. b.

John drove the car (toward the garage) John carried the bag (toward to beach)

for an hour. for an hour.

They do not affect the compatibility of the durative. They do not produce a telic interpretation. Consider the utterances in (17a-b). (17) a.

b.

Le abroché la camisa to-him buttoned the shirt “I buttoned his shirt.” or “I buttoned the shirt on/for him.” Le lavé el coche to-him washed the car “I washed his car.” or “I washed the car on/for him.”

(17a) can be said in a context in which the shirt is on a hanger during the buttoning. In (17b), the individual denoted by the non-reflexive does not have to be present while the car washing takes place. The non-reflexive does not express an on/with entailment. Observe that directional prepositions pattern in the same way (18).

176 (18) a. b.

JONATHAN E. MACDONALD

John drove the car toward the garage. John threw the ball toward Bill.

The car does not necessarily end up at the garage (18a), nor does the ball necessarily end up at Bill (18b). Directional prepositions do not express an on/with entailment. Consider the Spanish idiom in (19a). The non-reflexive prevents idiomatic interpretation of the utterance (19b). (19) a. Juana bebe los vientos por Javier Juana drinks the winds for Javier Idiomatic meaning: “Juana is in love with Javier.” b. # Juana me bebe los vientos por Javier. Intended meaning: “Juana is in love with Javier on me.” Directional prepositions can also prevent idiomatic interpretation (20). (20)

# John spilled the beans toward the police.

Observe another way in which Spanish non-reflexives pattern with overt directional prepositions: they both resist secondary adjectival predication. Consider the utterance in (21). (21)

Yoj lek lavé el coche borrachoj/*k. Ij to-himk washed the car drunkj/*k “I washed his car drunk.” or “I washed the car on/for him drunk.”

Only the subject can be interpreted as drunk in (21); the non-reflexive resists adjectival secondary predication. Observe in (22) that clitics in general do not resist adjectival secondary predication. The denotation of la “her” is interpreted as drunk. (22)

Juan lak besó borrachak Juan herk kissed drunkk “Juan kissed herk drunkk.”

Observe that, like non-reflexives, complements of directional prepositions also resist adjectival secondary predication (23). Only Ralph can be interpreted as drunk. (23)

Ralphj threw the ball toward Frankk drunkj/*k.

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

177

2.2.

The structure of Spanish non-reflexives Spanish non-reflexives pattern with overt directional prepositions in four ways: 1. they do not produce a telic interpretation of the predicate; 2. they do not express an on/with entailment; 3. they can prevent idiomatic interpretation; and 4. they resist adjectival secondary predication. To account for these parallels, I propose the structure in (24) in which the non-reflexive is the complement of a null directional-like preposition that merges as a complement of the verb. (24)

…vP ru (Yo) v’ (“I”) ru v VP ei DP V’ 5 ri el coche V PP “the car” lavar ru “wash” P DP ØDIR 4 le “him”

I assume that the properties of the null preposition account for the first two properties shared in common with directional prepositions, and the low merger of the null preposition accounts for the second two properties. I discuss the low merger first. To affect idiomatic interpretation a constituent must merge within the verb phrase (Marantz 1984 and Svenonius 2004). As a complement of the verb, the null preposition is in a position to prevent idiomatic interpretation. The resistance to adjectival secondary predication can be accounted for by the low merger of the prepositional phrase if we adopt Bowers’s (2000) proposal. Bowers claims that secondary predicates are V’ adjuncts that contain a PRO in their specifier.9 In order to establish a predication relation, a DP must control PRO. Given that the null preposition merges as a complement of 9

As pointed out by an anonymous reviewer, the result of this V’ adjunction approach to secondary predication established with a direct object is that the VP contains a non-branching V’ dominating only the verb. A structure of this sort is becoming more difficult to justify theoretically.

178

JONATHAN E. MACDONALD

the verb, it merges lower than the secondary predicate and, as such, the nonreflexive cannot control PRO. Thus, no predication relation can be established between a non-reflexive and a secondary predicate.10 Recall that MacDonald (2004) analyzes Spanish reflexives as the complement of a null goal-like preposition that merges as the complement of the verb (14). Given that proposal, and given the structure proposed here for non-reflexives (24), it is natural to conclude that any diverging patterns between the two (i.e. the (in-)ability to produce a telic interpretation and express an on/with entailment) result from distinct properties of the null prepositions themselves. In fact, I claim that the source of variation between reflexives and non-reflexives is the distinct properties associated with the null prepositions that introduce them. Observe another property that is a point of divergence between these null prepositions.11 Consider a context in which I work at a car wash and I am assigned cars to wash. My boss, who is concerned with whether or not I washed a particular car assigned to me, asks whether I did so. To tell him that I washed my car, I cannot use the reflexive pronoun. Thus, the utterance in (25) is not a felicitous response to his question. (25)

# Me lavé el coche myself washed the car. “I washed my/the car.”

Consider a similar situation in which my coworker is assigned a particular car that my boss asks about. Let us say that I washed his car. I can utter the sentence in (26) with the non-reflexive pronoun to respond to my boss.12 10

Demonte (1988) has an account of secondary predication that depends on mutual ccommand between the secondary predicate and the DP with which it establishes a predication relation. Given that the non-reflexive is housed in a prepositional phrase on the present account, it cannot c-command out to establish the mutual c-command relation. Thus, under Demonte’s proposal, the resistance to adjectival secondary predication provides evidence for the existence of the null prepositional phrase itself. However, it is not clear that complements of prepositions generally cannot c-command out of a prepositional phrase. Observe that the complement of the preposition can c-command out and control the PRO of an embedded clause: John gave money to Billj PROj to buy milk. 11 MacDonald (2006a) discusses how the distinct properties associated with these null prepositions can account for the variable behavior of reflexives and non-reflexives in the idiom sin comerlo ni beberlo. Reflexives do not prevent idiomatic interpretation: sin comérselo ni bebérselo, while non-reflexives do #sin comérmelo ni bebérmelo. 12 Example (26) is typical of a benefactive or malfactive in Spanish. Consider another: Juan me compró un libro “Juan bought me a book”. Observe that the benefactive/malfactive

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

(26)

179

Le lavé el coche to-him washed the car “I washed his car” or “I washed the car for him.”

In the carwash scenario, the reflexive pronoun cannot express a ‘temporary relation’ established between an object, in this case a car, and the denotation of the reflexive.13 The non-reflexive pronoun can express this temporary relation. This temporary relation is one example of a general difference between these reflexive and non-reflexive constructions with respect to the interpretation of the direct object. In general, the non-reflexive is not as restricted as the reflexive is. The non-reflexive tolerates temporary relations and is not restricted to objects interpreted as on/with. The reflexive does not tolerate temporary relations and is restricted to objects interpreted as on/with. The direct objects in non-reflexive constructions are less restricted in their interpretation than the direct objects in reflexive constructions. More formally, I claim that the range of possible relations between the direct object and the denotation of the (non-)reflexive is regulated via the compositional theta role assigned by the verb plus prepositional phrase. The restrictions on the interpretation of the direct object vary according to the properties of the preposition. 2.3.

Recapping the findings The properties of Spanish non-reflexives that we have uncovered are listed in (27). (27) a. Do not produce telic interpretation b. Do not express on/with entailment c. Can express a temporary relation d. Can prevent idiomatic interpretation e. Resist adjectival secondary predication

→ like directional PP → like directional PP → like directional PP → like directional PP

The common properties between the non-reflexive and the directional preposition motivate the structure in (24) in which the non-reflexive is introduced as a complement of a null directional-like preposition that merges as a complement of the verb. The properties in (27a-c) are argued to result from the nature of the null preposition itself, while the properties in (27d-e) are interpretation is independent of the reflexive/non-reflexive contrast: Juan se compró el libro. “Juan bought himself a book”. The benefactive/malfactive interpretation is independent of the present discussion. 13 MacDonald (2004) refers to this as ‘temporary possession’. I use ‘relation’ to remain neutral.

JONATHAN E. MACDONALD

180

argued to result from the low merger of the null preposition. Consider now the properties of Romanian reflexives. 3.

The properties of Romanian reflexives In this section I show that Romanian reflexives share the first four properties in (27) with Spanish non-reflexives.14 Based on these common properties, I propose that Romanian reflexives, like Spanish non-reflexives, are introduced as the complement of a null durational-like preposition that merges as the complement of the verb. Consider the utterance in (28). Observe that the reflexive does not affect the compatibility of the durative. The reflexive does not produce a telic interpretation.15 (28)

Eu mi-am spălat maşina timp de zece minute I myself-have washed car-the time of 10 minutes “I washed my/the car for 10 minutes.” Consider the utterances in (29).

(29) a.

b.

Eu mi-am pătat cămaşa I myself-have stained shirt-the “I stained my/the shirt.” Eu mi-am spălat maşina I myself-have washed car-the “I washed my/the car.”

(29a) can be said in a context in which the shirt is not on the subject,16 and (29b) in a context in which the subject dropped the car off at the carwash 14

We cannot test whether or not the reflexives resist adjectival secondary predication, for we cannot determine whether predication is established with the subject or with the reflexive itself. 15 Observe that in the following sentence, the durative phrase is incompatible and there is a reflexive pronoun present: Eu mi-am mâncat mărul *timp de zece minute “I ate my/the apple for an hour.”; observe nevertheless that the durative phrase remains incompatible when the reflexive is not present: Eu am mâncat mărul *timp de zece minute. This suggests that the reflexive does not produce the telic interpretation here; the telic interpretation most likely results from the properties of the verb itself. 16 Manoliu-Manea (1996) observes that when the reflexive is accusative, an overt preposition (la) is required, and the shirt is necessarily interpreted as on/with the denotation of the clitic: Eu m-am pătat la cămaşă “I stained my shirt”(The example is my own). Given the case difference and the presence of the overt preposition, I take this to be a distinct construction

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

181

for someone else to wash. The Romanian reflexive does not express an on/with entailment. Consider the utterance in (30). (30)

Eu mi-am spălat maşina. I myself-have washed car-the “I washed my/the car.”

It can be said in a context in which the car is temporarily assigned to the denotation of the subject. The Romanian reflexives can express temporary relations. Consider the Romanian idiom in (31a). Observe that the reflexive prevents idiomatic interpretation (31b).17 (31) a.

Eu am spart ghiaţa I have broken ice-the “I broke the ice.” → Idiomatic meaning: “I started a conversation.” b. # Eu mi-am spart ghiaţa. I myself-have broken ice-the

The reflexive pronoun of Romanian patterns in exactly the same way as the non-reflexive pronoun of Spanish with respect to the properties in (18a-d). I take this result as evidence that these constructions have the same underlying structure. I conclude that both the Spanish non-reflexive and the Romanian reflexive are introduced as the complement of a null directional-like preposition that merges as the complement of the verb. 4.

Conclusion In this paper we have seen that a range of properties can be used as a diagnostic for the underlying structure of Spanish reflexive non-argumental, Spanish non-reflexive non-argumental and Romanian reflexive nonargumental clitic pronouns. These properties are summarized in Table 1.

than the one under consideration. Space limitations prevent me from any further consideration of what an analysis of this accusative construction could purport to the present discussion. 17 Another construction in Romanian in which idiomatic interpretation is lost in the presence of the reflexive is the following: Eu (*mi)-am vazut negru. literally “I have seen black.”, with the idiomatic meaning “I have become angry.” (Laura Spinu p.c.). Arguably because there is a mass term direct object this is a different construction from those under discussion in the text. However, the proposed structure for the Romanian reflexives discussed here offers a way to understand the loss of idiomatic interpretation in this construction as well.

JONATHAN E. MACDONALD

182

produces a telic interpretation

necessary on/with entailment

tolerates temporary relations

can prevent idioms

resists secondary predication

Span. Refl. Goal Prep.

yes yes

yes yes

no no

yes yes

NA yes

Span. Non-refl. Rom. Refl. Dir. Prep.

no no no

no no no

yes yes NA

yes yes yes

yes NA yes

Table 1: Properties of non-argumental clitic pronoun constructions

Observe that all of these pronoun and prepositional phrase constructions pattern the same with respect to the ability to prevent idiomatic interpretation and with respect to the resistance to adjectival secondary predication (where available). These were accounted for by the low merger of the prepositional phrases. The variation among these (null) prepositions, then, results from the properties associated with the prepositions themselves. Two opposing groups emerge when considering the different properties of these prepositions. Spanish reflexives and goal prepositions pattern together on the one hand, and Spanish non-reflexives, Romanian reflexives, and directional prepositions pattern together on the other. A natural question is why these properties are divided up the way they are. With respect to the ability of a preposition to produce a telic interpretation of a predicate, Beck and Snyder (2001) and Snyder (1995) found that cross-linguistically languages vary with respect to the ability of a goal preposition to produce a telic interpretation. Snyder (1995) posits that this ability is related to the presence or absence of a null telic morpheme. A preposition that has this null telic morpheme produces a telic interpretation and a preposition that does not have it, does not produce a telic interpretation. Along these lines, we can say that goal prepositions and the null preposition introducing Spanish reflexives bear this null telic morpheme, while directional prepositions and the null preposition introducing Spanish non-reflexives and Romanian reflexives do not. This accounts for the variation in ability to produce a telic interpretation in the same way that wider cross-linguistic variation of this type can be accounted for. This suggests that the variation in the production of a telic interpretation is not unique to these Romance (non-) reflexive constructions, and therefore not entirely unexpected. It also introduces a big question: What determines whether or not a preposition bears a null telic morpheme or not? A related question more specific to the present

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

183

study is: Why are Spanish reflexives associated with a preposition that bears a null telic morpheme while non-reflexives and Romanian reflexives are not? There may be telling correlations among the properties in Table 1 that can provide an answer to this specific question. The patterns in Table 1 give the impression that there is a correlation between an on/with entailment and the ability to produce a telic interpretation. It is not clear, however, that this correlation exists. MacDonald (2006b) observes in French that the presence of a reflexive pronoun results in an on/with entailment (32a) that otherwise does not exists without the reflexive (32b). (32) a.

b.

Jean s’est boutonné la chemise Jean himself-is buttoned the shirt “Jean buttoned his shirt.” Jean a boutonné la chemise Jean has buttoned the shirt “Jean buttoned his shirt.”

Moreover, he notes that the presence of the reflexive pronoun does not affect the compatibility of the durative phrase (33).18 (33)

Jean s’est/a boutonné la chemise Jean himself-is/has buttoned the shirt “Jean buttoned his shirt for an hour.”

pendant une heure for an hour

These facts suggest that the property responsible for producing a telic interpretation is independent of the property responsible for expressing an on/with entailment. If these properties are indeed independent, then, there is no way to determine if a preposition can produce a telic interpretation of a predicate if it expresses an on/with entailment, or vice-versa. Nevertheless, as an independent property, we can assume that some heads have the property and some do not. This does not tell us why some heads have the property nor why some heads do not, but it provides us with a way to understand the scope of possible (cross-linguistic) variation based on the presence or absence of the properties in Table 1. That is, given the range of properties, we can determine whether or not any is contingent on the presence or absence of another property, or whether each is a property truly independent of all others. In the former case, we could begin to find insight into and perhaps understand more 18

There does seem to be some variation with respect to the compatibility of the durative phrase here.

184

JONATHAN E. MACDONALD

clearly the patterns of these Romance (non-)reflexive constructions. In the latter case, we expect to find a (null) preposition for each of the possible combinations of the presence and/or absence of each property in Table 1. In this latter case, it would not be clear what the insight into the divergent patterns of these distinct Romance (non-)reflexives would be. References Beck, Sigrid and William Snyder. 2001. “Complex Predicates and Goal PP’s: Evidence for a Semantic Parameter”. Proceedings of the 25th Boston University Conference on Language Development, ed. by Anna H.-J. Do, Laura Dominguez & Aimee Johansen, 114-122. Somerville: Cascadilla Press. Bowers, John. 2000. “Predication”. The Handbook of Contemporary Syntactic Theory, ed. by M. Baltin & C. Collins, 299-333. Malden, Mass.:Blackwell. De Miguel, Elena & Marina Fernández Lagunilla. 2000. “El operador aspectual se”. Revista Española de Lingüística 30:1.13-43 Demonte, Violeta. 1988. “Remarks on Secondary Predicates: C-command, Extraction, and Reanalysis”. The Linguistic Review, 6:1-39. Dowty, David. 1979. Word Meaning and Montague Grammar. Dordrecht: D. Reidel. Larson, Richard K. 1988. “On the Double Object Construction”. Linguistic Inquiry 19:3.335-391. MacDonald, Jonathan E. 2004. “Spanish Reflexive Pronouns: a Null Preposition Hypothesis”. Proceedings of WCCFL 23.528-540. ed. by Vineeta Chand, Ann Kelleler, Angelo J. Rodríguez & Benjamin Schmeiser. Somerville: Cascadilla Press. -----------. 2006a. “Non-reflexive, Non-argumental Clitic Pronouns of Spanish”. Proceedings of WECOL 2004, ed. by Asier Alcazar, Roberto Mayoral Hernandez & Michal Temkin Martinez. Fresno: California State University. ----------. 2006b. The Syntax of Inner Aspect. Doctoral dissertation, Stony Brook University, Stony Brook, NY. Manoliu-Manea, Maria. 1996. “Inalienability and topicality in Romanian: Pragma-semantics of syntax”. The Grammar of Inalienability: A typological Perspective on Body Part Terms and the Part-Whole Relation, ed. by H. Chappell & W. McGregor, 711-743. Berlin: Mouton de Gruyter. Marantz, Alec. 1984. On the Nature of Grammatical Relations. Cambridge, Mass.: MIT Press. Nishida, Chiyo. 1994. “The Spanish reflexive clitic se as an aspectual class marker”. Linguistics 32.425-458.

NULL DIRECTIONAL PREPOSITIONS IN ROMANIAN AND SPANISH

185

Nunberg, Geoffrey, Ivan A. Sag and Thomas Wasow. 1994. “Idioms”. Language 70.3.491-538 Pustejovsky, James. 1991. “The Syntax of Event Structure.” Cognition 41.4781. Snyder, William. 1995. “A Neo-Davidsonian Approach to Resultatives, Particles, and Datives.” Proceedings of NELS 25.1, ed. by Jill Beckma, 457-471. Amherst: GLSA. Svenonius, Peter. 2004. “Slavic Prefixes inside and outside the VP”. Nordlyd 32.2, ed. by P. Svenonius, 205-253. Tromsø, Norway: Faculty of Humanities, University of Tromsø. Online at www.ub.uit.no/munin/nordlyd/ Zagona, Karen. 1996. “Compositionality of Aspect: Evidence from Spanish Aspectual Se”. Aspects of Romance Linguistics. ed. by Claudia Parodi, Carlos Quicoli, Mario Saltarelli, and María Luisa Zubizarreta, 475-488. Washington DC: Georgetown University Press.

A UNIFIED ACCOUNT FOR THE ADDITIVE AND THE SCALAR USES OF ITALIAN NEPPURE

ALDA MARI AND LUCIA M. TOVENA CNRS/ENST and Paris VII

0.

Introduction Neppure is an additive and scalar particle of Italian that is used to add negative information. Examples (1) and (2) illustrate the two readings in adnominal (a) and adverbial (b) positions. (1)

a. Giovanni non ha mangiato la verdura e neppure la frutta “John did not eat the vegetables or the fruits.” b. Non telefonava e neppure scriveva “S/he did not phone and did not write either.”

(2)

a. Non ha mangiato neppure il dolce “S/he did not eat even the dessert.” b. Non ha neppure sentito parlare del film! “S/he did not even hear about the film!”

As noted by König (1991:63), simple sentences with additive particles (i) presuppose that at least one alternative value satisfies the predicate and (ii) entail the corresponding sentence without the particle. The second half of (1a), for instance, presupposes that there are items other than fruit that John has not eaten—one (the vegetables) is overtly mentioned in the first half, others may have been mentioned in the preceding discourse or be part of the common knowledge. It entails that he has not eaten fruit. Similarly, the second conjoint in (1b) presupposes that John has not accomplished other actions, one of which is phoning, and entails that he has not written. Sentences with scalar particles involve a specific scalar conventional implicature (König, 1991:98). Not only do they presuppose that alternative values under consideration in a context satisfy the predicate, but also that these values are ordered. They also entail the

188

ALDA MARI & LUCIA M. TOVENA

corresponding sentence without the particle. (2a) presupposes that there are other items that John has not eaten. These were less likely to be eaten than the dessert. Similarly, (2b) presupposes that actions other than hearing about the film, e.g. seeing it, did not take place and implicates that hearing was the most likely. It is generally argued (e.g. Kartunnen and Peters 1979; Rooth 1992 and König himself), that the contribution additive and scalar particles make to the semantic content of a sentence is rather limited. They are presuppositional items that affect the felicity conditions of the utterances. In this paper we show that neppure provides a more consistent contribution to the interpretation of the sentence. It adds negative information, thus potentially affecting the truth value of the clause that hosts it, and poses a strong constraint on scales. We propose a unified account for these uses. We show, as tacitly accepted, that in both uses neppure focuses on an element that is understood as a member of a class/series and adds negative information. But we also show that in both interpretations neppure sets a temporary closure in the speaker’s state of knowledge. At a given point in time, it marks that all possible relevant negative facts relative to a class have been checked out. The difference between additive and scalar neppure lies in the way the class/series is built and not in the presence vs. absence of a presupposition concerning an ordering on the class. Our analysis makes the prediction, crucial for the Italian data, that the presence of an overt antecedent preferably correlates with the additive reading, but the absence of antecedent always correlates with the scalar reading. Note that the scalar reading is compatible with the presence of antecedents (3) and that the sentence is infelicitous if perceived lexical or cultural orderings are flouted (4). (3)

Non ha studiato il capitolo e neppure l’ha letto “He didn’t study this chapter and he did not even look at it.”

(4)

#Non ha letto il capitolo e neppure l’ha studiato “He did not look at this chapter and he did not even study it.”

In this paper, following the terminology in Krifka (1998) and Rullmann (2003), we call ‘associate’ the nominal or the verbal expression to which neppure applies. The sentence containing neppure is called ‘host clause’. The entity or event of the same type as the associate and which is overtly present in the context is called ‘antecedent’ and the sentence that hosts it is the ‘antecedent clause’. In the case of additive neppure, the context proposition and the host clause may be part of the same sentence (1a) and (1b).

A UNIFIED ACCOUNT FOR ADDITIVE AND SCALAR NEPPURE

189

Sections 1 and 2 constitute the core of the paper and contain our two main claims. In Section 1 we state the principle of negative information, arguing that neppure adds negative information (1.1) and introduces the presupposition that for other elements in the context we have negative facts of the type asserted of the associate (1.2). Section 2 is devoted to the discussion of the epistemic closure principle and is articulated in four subsections. We first show that, in both its additive and scalar uses, neppure sets a temporary endpoint (2.1). We claim that this endpoint is agent oriented and discuss previous accounts of the notion of endpoint in correlation with the mandatory scalar reading. Secondly, we argue that in both uses, neppure closes the class or the series (2.2) and propose a way to model this situation. In Section 2.3, we show how the set is closed respectively for additive and scalar interpretations. We claim that in the additive case the set is defined extensionally by overt enumeration of the members of the class, whereas with scalar neppure the set is built intensionally by setting a criterion for membership and imposing an order to support the inferences. In the first case all needed information must be available, while in the second, some information must be accommodated. Some mix cases are discussed in Section 2.4. Section 3 briefly concludes this paper. 1.

Adding negative information and triggering negative facts The first hypothesis that we need to investigate is that neppure presupposes a class or a series of negative facts and adds negative information. 1.1.

Neppure contributes negative information Neppure can occur in postverbal position when the verb is negated (5a) and (5c) but is out when the verb is not negated (5b). (5)

a. Non ha fatto i compiti neppure Daniele “Not even Daniele did the homework.” b. *Ha fatto i compiti neppure Daniele “Did the homework neppure Daniele.” c. Non ha fatto neppure i compiti “He did not even do the homework.”

In preverbal position neppure can be the only negative word in the clause: it contributes a sentential negation and the following verb must not be negated (6a), otherwise the sentence is marginal (6b), as it is standardly the case in the negative concord marking system of Italian. For some speakers it can somewhat be rescued via a double negation reading.

190 (6)

ALDA MARI & LUCIA M. TOVENA

a. Neppure Daniele ha fatto i compiti “Not even Daniele did the homework.” b. *?Neppure Daniele non ha fatto i compiti “Not even Daniele did not do the homework.”

When it comes to the interpretation of data such as (5) and (6), the two labels of NPI or NC-word concur. We claim that neppure’s function of adding negative information is naturally lexicalized via an NC-word, because this type of realization warrants that the host clause is always negative, but we agree that the precise nature of the function and the realization can (and should) be proved separately. Clear evidence in favor of the specialization for negative information is constituted by the requirement that a negative fact of a similar nature to that asserted relatively to the associate is presupposed relatively to the antecedent, but no precise constraint on the shape of the antecedent clause comes with it. Support for the analysis in terms of a lexicalization as an NC-word comes from examples (7) through (12). As we can see, neppure can constitute a negative fragment answer (7), something NPI cannot do, and requires clausemate negation when in postverbal position (8), whereas polarity sensitivity is not a clause-bound phenomenon. (7)

Ha mangiato almeno? Neppure “Did s/he eat, at least? Neither that/Not even that.”

(8)

*Non ho detto che ha mangiato neppure la pera “I did not say that he ate neppure the pear.”

Next, neppure is not acceptable under negative predicates as in (9) that are well-known licensors for NPIs, and is out also in other standard licensing contexts such as questions (10), the antecedent of conditionals (11) and troppo (“too”) as in (12). (9)

*Dubito che neppure Luisa scriva “I doubt that even / not even / either Luisa writes.”

(10) *Ha mangiato neppure la pera? “Did s/he eat even /either the pear?” (11) *Se mangia neppure la pera, la situazione è grave “If s/he eats neppure the pear, it is a serious situation.”

A UNIFIED ACCOUNT FOR ADDITIVE AND SCALAR NEPPURE

191

(12) *Sembra troppo stanco per fare neppure i compiti “He seems too tired to do neppure the homework.” Of course, examples (9)–(12) could be invoked to support a characterization as strong NPI. However, besides the ongoing discussion of whether the characterization as NPI or NC is more appropriate than the other, what matters for us, for the purposes of this paper, is that neppure brings with it negative information. In particular, it brings with it the presupposition that there are, in the context, some other entities or events relatively to which a negative fact holds of a nature analogous to that described in the host clause. 1.2.

Presuppositions and scalar items Neppure is a focus-marking particle. Since Jackendoff (1972) and Rooth (1985) it is assumed that focus elements project a class of alternatives. The focus semantic value of a phrase of category S is the set of alternative values to its semantic value, plus its ordinary semantic value. (13) Giovanni non ha invitato neppure Luisa F “Giovanni did not invite even Luisa.” Ordinary semantic value: [[luisa]]° = {luisa} Focus semantic value: [[luisa]]F = {luisa,maria,serena} (14) Giovanni non ha neppure telefonato F a Luisa “Giovanni has not even called Luisa.” Ordinary semantic value: [[telefonato]]° = {phone} Focus semantic value: [[telefonato]]F = {phone,write,e-mail} Following (Rooth 1992 :93), and in agreement with much literature on scalar particles, we assume that the constraint introduced by the focus interpretation is a presupposition. Let Г mark the position associated with focus, for expression φ, in the syntactic tree. (15) Presupposition with focus for scales (constraint on sets): φ ~ Γ presupposes that Γ is a subset of the focus semantic value for φ and contains both the ordinary semantic value of φ and an element distinct from the ordinary semantic value for φ. For neppure α, rule (15) gives: [[neppure α]]F = Γ ⊂ S.

192

ALDA MARI & LUCIA M. TOVENA

We must add that, for ordered sets, G satisfies the scalar implicature: with respect to (13), for instance, if Luisa had a certain level of probability of being invited by Giovanni but turned out not to be, other people who had a lower probability of being invited also were not invited for sure. Let us derive some consequences. First of all, given its additive negative nature, neppure introduces the presupposition that there is (at least) one entity/event in the context relatively to which it holds a negative fact of a nature similar to that asserted of the associate. Neppure satisfies Negative Information Principle: (16) Negative Information Principle: Increase negative information. Second, as other focus semantic particles, neppure affects the felicity conditions of the utterances by imposing constraints on the focused element and on the context (Kartunnen and Peters 1979). Differently form other focus semantic particles, though, it is also able to affect the truth conditions of the host clause, since it can express negation, see (6) above. Third, Rooth’s rule of presupposition associated with focus (15) introduces a constraint on scales, without any specification. It does not constrain the set of alternatives to be ordered, nor does it specify constraints regarding the position of the focused element on the scale. In particular, it is neutral with respect to the necessity of closing the set of alternative values by identifying this position with one extreme of the scale. Neppure adds this information: it sets an endpoint and closes the scale. The next section is devoted to the discussion of these two issues. 2.

Epistemic closure In the rest of the paper we show that given a speaker’s state of knowledge, neppure signals that a negative fact has been asserted of an entire class. In other terms, it signals that the negative information contributed concerns an entire class. In order to show that this is the case, we need to answer the question of how neppure ensures that all the elements of the class have been checked out. We proceed in two steps. First we show that neppure sets an endpoint in 2.1. Second, in 2.2 we show that all the elements in the class, given the speaker’s state of knowledge, satisfy principle (16). Then, we show in Section 2.3 how the closure is differently obtained for additive and scalar neppure. Since the notion of endpoint is more crucial for scalar neppure, we concentrate particularly on this use.

A UNIFIED ACCOUNT FOR ADDITIVE AND SCALAR NEPPURE

193

2.1.

Neppure sets an endpoint We hypothesize that neppure always sets an endpoint, but this seems to be straightforwardly contradicted by the following example: (17) Maria non ha vinto neppure le semifinali di Wimbledon! “Maria did not win even Wimbledon semifinals!” The semifinals are not an endpoint per se in a tournament. However, the information conveyed by the sentence goes beyond the simple pointing which stage of the tournament was not reached and includes that Maria has a level of tennis playing such that she could have won the semifinals, and possibly even the finals. In other terms, the semifinals are the greatest lower bound of success for Maria and thus an end-point on the scale of tennis matches that she is most likely to win. In order to better capture this point let us consider previous accounts of the notion of endpoint and scalar implicatures, which have tried to explain why the scalar reading is mandatory in the absence of antecedents. 2.1.1. Worldly scales and relevance: previous accounts. Schwenter and Vasishth (2000) study two pairs of scalar additive particles in Spanish, i.e. incluso and hasta, and in Hindi, i.e. –bhii and –tak. The elements of each pair often are interchangeable and are said to translate as English even. The difference, according to them, is that incluso and -tak require their host clause to be more informative than another similar proposition already accessible in the context, like even does, whereas hasta and –bhii do not. As revealed by the pointer to even, ‘accessible’ means ‘inferrable’ rather than ‘overtly present’. Next, they set up a correspondence between requiring an accessible alternative and not being an inherently endpoint marking particle vs. not requiring an accessible alternative and being an inherently endpoint marking particle. Hasta is thus predicted to be incompatible with semifinals in the counterpart of example (17) because the associate (semifinals) is not an inherent endpoint. In our view, contextual information is not relevant for licensing weaker scalar particles specialized for unbounded scales, as opposed to context free strong scalar particles specialized for bounded scales. The choice of marking the most informative point can be explained via a Gricean maxim, and the contextual scales speakers reason with may well be truncated, i.e. end with non-inherent endpoints. However, contextual information is crucial for building the set, hence the main difference is between there being an overt antecedent from which to get information directly or having to rely on inferences to reconstruct the set.

194

ALDA MARI & LUCIA M. TOVENA

One should wonder whether informativity would be more appropriate. Kay (1990) had already explored this line of thought. For (17), it predicts that it is more relevant to mention the semifinals since one can predict that she has not won the finals either. One should not conclude form the fact that she has not won the finals, that she has not even won the semifinals, as it was minimally expected from her. Note, however, that this is standard for scalar inferences (see Horn 1972, Fauconnier 1975 and followers). This account raises some problems per se and in relation with neppure. First, it is not clear why one should mention one stronger alternative and not the strongest. Indeed, to assert that Mary won even the semifinals when one knows she won the final is at least infelicitous if not downright unacceptable. Saliency can hardly lead to selecting a different alternative. Moreover, Kay, and Schwenter and Vasishth after him, reason with worldly scales and, as we are suggesting, the endpoints with which scalar particles correlate, are not necessarily given by the external reality. Finally, his account does not provide us with any anchor for explaining what is common between additive and scalar uses of neppure. König (1991) tries to explain the relation between the additive and the scalar uses of focus particles in the light of the theory of greater informativity of Kay and suggests (König 1991:64) that they share the same presupposition: at least one of the alternative values under consideration in a context satisfies the complex predicates. Under the additive interpretation the alternatives are overtly mentioned, under the scalar interpretation, since only the associate of the particle is usually mentioned, it is the most relevant. The alternative values are triggered on the basis of their degree of informativeness, which is lower than that of the associate on the scale of relevance. Since the complex predicate applies to the most relevant alternative, it must also apply to the less relevant, as standard scalar reasoning goes. König’s work marks a step forward in the individuation of the common core of the additive and the scalar uses. However, apart from their sharing the same existential presupposition, it is not clear what else additive and scalar neppure share. In other terms, one should make clear what positive contribution they make to the interpretation. 2.1.2. Neppure introduces the strongest possible argument in the conversation. Our analysis of (17) leads us to make the hypothesis that neppure does not introduce a ‘more informative proposition’ but the strongest possible argument on an agent oriented scale. As we noted, (17) conveys the information that Maria has not fulfilled the minimal requirements for her

A UNIFIED ACCOUNT FOR ADDITIVE AND SCALAR NEPPURE

195

standards. Given Maria’s world, one could conclude that this competition is not worth mentioning at all. The statement is the strongest that can possibly be made from the point of view of the speaker. The semifinals are an endpoint for both the speaker and the hearer in relation with their understanding of Maria’s world. Example (18) can be analyzed similarly. It is generally assumed that the service is a difficult move for anybody but a professional tennis player. For such a player, it is (one among) the minimal requirement(s) for disputing a match. Consequently, if uttered about a professional tennis player the sentence is interpreted in the way just described. If uttered of an amateur tennis player, it is almost uninterpretable. (18) E non sa neppure servire! “And he cannot even serve!” For the professional tennis player, sentence (18) tells that the service is the greatest lower bound of technicality he must ensure. It is not an endpoint per se, however. Again, the ending point is not fix and universal, but has to be considered in relation with the context and speaker’s state of knowledge, and the nature of the entity denoted by the associate. The provisionary conclusion we draw is that scalar neppure marks its associate as a contextual or temporary endpoint in a series or a class. Marking an endpoint means to make sure that the facts have been checked for the entire class. If the endpoint is inherent, this is not problematic. If this is not an inherent endpoint, and since it is always possible to build speaker-oriented situations in which the associate can be seen as an endpoint, one must accept that a temporary endpoint is, in fact, an endpoint. Since this amounts to claiming that neppure signals that a negative fact has been checked out for all the elements of a class or a series, we have now to show that all the elements of the class or the series have to satisfy negative information principle and this for both the scalar and the additive use. 2.2.

Extensional and intensional definition of a set and the notion of closure In the previous section we have identified the notion of closure with that of flexible endpoint on a descriptive basis for scalar neppure. However, it can be extended to the additive use and can be associated with a precise algorithm of identification of the class or a series. In particular, it can be considered as the outcome of a specific procedure for building the class and of

196

ALDA MARI & LUCIA M. TOVENA

checking that the negative fact is satisfied by the elements relevant for the speaker. This procedure is the positive contribution of neppure. Given a context C, a model, in that context, is a triple < E,D,f > where E is a set of entities (restricted under speaker’s state of knowledge and C), D a set of semantic dimensions and f a function that assigns to each entity a value on a semantic dimension. The semantic dimension is identified on the basis of the lexical contribution of the associate and of the antecedents (if present). Their images on this semantic dimension are determined semantically and pragmatically. Let x be the ordinary semantic value of the focused phrase ([[ ]]°= x), E the focus semantic value of the phrase [[neppure ]]F and d = f(x). E is the union of the ordinary semantic value of the phrase and the set A of its alternative values (E = {{x}, A}). This definition of E is meant to provide the maximal extension of the class/series. It follows that the set of alternatives that satisfy negative information (i.e., the set A) is closed. Our claim is that by stating how the set A is construed, we can show how it comes to be closed. The procedure for defining the elements and closing the class/series is the specific contribution of neppure to the meaning of the sentence. This procedure is different for the additive and the scalar uses. In the additive case the set is built by enumeration (extensional criterion). For all the elements that are mentioned, including the associate of neppure, the value of f is the same. In the scalar case, the procedure requires the individuation of a closure point, a criterion for membership, and the order (intensional criterion). (19) [[neppureadditive α]]F = { a1, a2, …, an  f(a1) = f(a2) = …, = f(an) = d } [[neppurescalar α]]F = { a  f(a) < d} Let us develop this in more detail. 2.3. Additive and scalar interpretations: construction and closure of the class Under the additive interpretation, the set is defined extensionally. The entities denoted by the antecedents and the associate of neppure are members of an unordered set. The antecedents are mentioned in the antecedent clause(s). Antecedent clause(s) do not necessarily form a single sentence with the host clause, as in examples (1), but might be found in a larger portion of discourse preceding it. We have also noted that the presence of antecedent(s) is compatible with the scalar use (see (3)). This is a mixed interpretation and we come back to it in Section 2.4.

A UNIFIED ACCOUNT FOR ADDITIVE AND SCALAR NEPPURE

197

For the additive interpretation, without any scalar inference, we assume that all the elements have to be mentioned or be made available by the context. Since they all are assigned the same value as the associate on a semantic dimension, no other criterion of membership is provided. It follows that the closure is also obtained extensionally by mentioning explicitly all the members relatively to which the relevant negative fact that is presupposed can be verified. The speaker assumes then that negative information has been provided for all the relevant members of an unordered set. For additive neppure, crucially, it is known information that there are other entities/events relatively to which a negative fact of the relevant type is verified, namely the entities and events mentioned in the antecedent propositions (or made available by the context). The additive interpretation is epistemically transparent. The scalar interpretation instead requires updating the information state (Zeevat 1992). In the scalar interpretation the associate determines the nature of the dimension. In other terms, by its content, it determines the characteristic of E. The closure is thus intensionally determined. Since the establishment of an order on the semantic dimension turns out to be the procedure to identify elements other than the associate, and since for all and only the elements whose positions on the dimension are entailed by the position of the associate it can be guaranteed that the presupposed negative fact stands verification, the associate also sets a temporary endpoint. Moreover, the closure is temporary, since it is accepted to hold in the particular context and nothing in the nature of the associate necessarily makes it deserve the privileged position of endpoint in another context. In this way, the series is closed under speaker’s state of knowledge. It follows that, contrary to the case of additive interpretation, two pieces of information need to be accommodated. First, since it is the property introduced by the associate that is relevant to retrieve the nature of the scale, its use as a criterion of membership is accommodated. Second, the presupposition that the property predicated of the associate applies to at least another entity is deemed satisfied in the absence of an antecedent. That the series need to be closed (together with adding negative information and setting an endpoint) is part of the specific contribution of the meaning of neppure. Before we straight out the main prediction of our account, let us state the epistemic closure principle: (20) Epistemic closure principle. Convey an assessment in the information state of the speaker’s knowledge based on the associate and the closure of the class.

198

ALDA MARI & LUCIA M. TOVENA

Neppure imposes a condition for the interpretation of the sentence given in (21): (21) [[neppure]] = λxλPneg : ∃a ∈ E [a ≠ x ∧ Pneg(a)] ∧ ∀y ∈ E [y ≠ x → f(y) ≤ d ].Pneg(x) We can now state the prediction concerning the correlation between the absence of antecedents and the mandatory scalar reading as follows: an antecedent, when present, provides an element in the set besides the associate. In the scalar interpretation, the associate provides the criterion for retrieving the series. In both cases, the set is temporarily closed under the speaker’s state of knowledge. 2.4.

More on the correlation The prediction states that there is a correlation between the presence of antecedents and the preference for an additive reading, and the absence of antecedents with the mandatory scalar reading, let us go through the steps of the explanation. The presence of antecedents is also compatible with the scalar reading. There is an important difference with additive reading, though: in this case, the entities denoted by the antecedent(s) on the semantic dimension have to be assigned a value equal to that of the associate, on that semantic dimension. This is not the case for the scalar reading, where the order of the antecedents must respect the order on the scale. Recall (3) and (4), where the scale had to follow lexical ordering. The scale can also follow a cultural ordering: (22) Non ha votato per le regionali e neppure per le nazionali “He has not voted for the regional, and not even for the national election.” (23) (#)Non ha votato per le politiche e neppure per le amministrative “He has not voted for the political election, and not even for the administrative one.” Since the national election is more important than the regional, one could expect that one should at least vote for the former. In other terms, the degree of probability of voting for the national rather than the regional election is higher. This is why (25) is infelicitous. The pragmatic effect obtained from deriving a scalar reading with antecedents is the strengthening of the argument. Not only the criterion is

A UNIFIED ACCOUNT FOR ADDITIVE AND SCALAR NEPPURE

199

provided by the antecedent, not only the order is accommodated, but also it is verified by the overt mention of the other elements of the scale. Note that (23) and (25) are also accepted under an additive interpretation of neppure. Differently from the scalar interpretations (22) and (24), the additive reading is compatible with the possibility that the knowledge of the speaker is not complete, e.g. he is not aware of the possible ordering. The additive and the scalar interpretations seem to allow different continuations: (24) Non ha votato per le politiche e neppure per amministrative. Avra’ forse votato per il referendum? “He did not vote for the regional or for the national elections. Has he maybe voted for the referendum?” (25) Non ha votato neppure per le politiche. (#)Avra’ forse votato per il referendum? “He did not vote for the national or for the regional elections. Has he maybe voted for the referendum?” This observation points to the fact that intensional closure is stronger than the enumeration of the elements that satisfy negative fact. 3.

Conclusion In this paper we have provided a unified account for the additive and scalar uses of Italian neppure. We have defended two claims. First, neppure focuses on an element that is understood as a member of a class or a series, i.e. shares the existential presupposition found in positive additive particles. What is special is that it adds negative information in the sense that it signals that the negative fact asserted of the associate licenses inferences about related negative facts (Negative Information Principle). Second, it signals that the knowledge relative to a class of negative facts can be considered complete with respect to a certain state of speaker’s knowledge (Epistemic Closure). The additive and scalar interpretations result from the different procedure adopted to verify the existential presupposition and to comply with the principles of Negative Information and Epistemic Closure. A major issue still open at present is what, in the nature of neppure, enables it to cover both interpretations and cannot be found, for instance, in the French negative additive counterpart non plus (“neither”), that does not exhibit scalar uses. The answer perhaps lies in the difference of the negative concord rules of the two languages, a question still to be investigated.

200

ALDA MARI & LUCIA M. TOVENA

References Fauconnier, Gilles. 1975. “Pragmatic Scales and Logical Structure”. Linguistic Inquiry 3.353-375. Horn, Laurence. 1972. On the Semantic Property of Logical Operators in English. Doctoral dissertation, University of California Los Angeles. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, Mass.: MIT Press Kartunnen, Lauri & Peters, Stanley. 1979. “Conventional Implicature”. Syntax and Semantics: Presupposition, ed. by Choom-Kyu Oh and David A. Dinneen, vol 11, 1-56. New York: Academic Press. Kay, Paul. 1990. “Even”. Linguistics and philosophy 13.59-111. König, Ekkehard. 1991. The Meaning of Focus Particles. London: Routledge. Krifka, Manfred. 1998. “Additive Particles Under Stress”. Proceedings from Semantics and Linguistic Theory VIII, ed. by Devon Strolovitch and Aaron Lawson, 111-128, Ithaca, NY: CLC Publications. Rooth, Mats. 1985. Association with Focus. Doctoral dissertation, University of Massachusetts at Amherst. ----------. 1992. “A Theory of Focus Interpretation.” Natural Language Semantics 1.75-116. Rullmann, Holtze. 2003. “Additive Particles and Polarity”. Journal of Semantics 20.239-401. Schwenter, Scott and Shravan Vasishth. 2000. “Absolute and relative scalar particles in Spanish and Hindi”. Proceedings of Berkeley Linguistics Society 26. 225-233. Zeevat, Henk. 1992. “Presupposition and Accommodation in Update Semantics”. Journal of Semantics 9.379-412.

DEFAULT MORPHOLOGY IN SECOND LANGUAGE SPANISH MISSING INFLECTION OR UNDERSPECIFIED INFLECTION?

*

CORRINE McCARTHY McGill University 0.

Introduction Recent discussions of second language (L2) morphology have focused on the source of learner errors. On one account, these errors stem from syntactic deficits; these deficits may include impaired or missing functional projections (Hawkins & Chan 1997, Vainikka & Young-Scholten 1994, among others). On another account, these errors stem from problems at the morphology-syntax interface, with the syntax assumed to be intact (Lardiere 1998a,b; Prévost & White 2000). On both accounts, it has been observed that learners often omit phonological markers for tense and agreement, and that learners often employ default morphology in the place of target morphology. However, in the discussion of morphological variability, the variants themselves have been largely ignored. This paper puts aside the issue of the source of these errors, and adopts as a starting point the observation that morphological errors in L2 Spanish are systematic. I propose that morphological errors are constrained by the features that are available in the L2 grammar, and that underspecification theory provides an explanatory account for why certain features are available and others are not. Under the approach advanced here, only underspecified morphemes may act as defaults. Furthermore, I argue that defaults are not necessarily instances of missing inflection, because finite forms may also act as defaults. Data come from person, number, and tense morphology in the spontaneous production of L2 Spanish. One recent finding for L2 French is that nonfinite verbs act as defaults, and may occur in finite (raised) positions; the reverse, however, does not occur: finite verbs do not occur in nonfinite contexts (Prévost & White *

I am indebted to Lydia White and Susana Bejar for their supervision at all stages in the development of this paper, and to Eva Villalba for assistance in data collection. I would also like to thank the audience at LSRL 35 for useful comments, some of which have been implemented in the recoding of data.

202

CORRINE McCARTHY

2000; see also Haznedar and Schwartz 1997, Lardiere 2000). This generalization is captured under the Missing Surface Inflection Hypothesis (MSIH; Prévost & White 2000), and is proposed to follow from the underspecification of finiteness. Under this hypothesis, features and featurechecking mechanisms are unimpaired, which suggests that if learners produce agreement, it will always be accurate. For Prévost and White, this was borne out: the overall error rate for regular and irregular verbs was around 5 percent. However, the low error rate for regular verbs is difficult to interpret, as French is homophonous across all persons in singular forms of present tense. As it turns out, this homophony is important: the most common type of person error found in the data collected here is the substitution of 3rd person for 1st. In Spanish, these forms are, for the most part, distinct; in French, they are usually homophonous. Prévost and White found that French irregular verbs être “be”, avoir “have” and aller “go” did surface with a few agreement errors, mainly involving the overgeneralization of 3rd person singular forms to other persons (Prévost & White 2001:121): (1)

j’a fait la I.have-3SG made the

bagarre avec lui-meme fight with him

In the production of tense morphology, White (in press) and Lardiere (1998a) found that learners of L2 English often omit regular affixal -ed. White shows further that they do not oversupply -ed, lending support to the MSIH. However, it should be noted that the persistent lack of suppliance of -ed is not completely informative: it is not known whether a bare English verb is nonfinite (as the MSIH predicts), or an instance of a finite, present-tense verb, since English nonfinite verbs and present-tense verbs are homophonous, excepting 3rd singular. In sum, ‘default’, under the MSIH, is equivalent to an absence of agreement or tense inflection. The MSIH makes no predictions regarding variability among finite forms, only to say that inflection, when supplied, is accurate.1 The defaults that learners employ are largely systematic, but at present, the only predictable default is a nonfinite one. There are two points at which the methodology of the present study diverges from the previous studies of L2 morphology discussed here. First, the 1

Prévost and White briefly discuss the notion that 3rd person may be underspecified, following a suggestion made by Ferdinand (1996) for L1 acquisition. However, the MSIH does not predict this result. Underspecification of 3rd person would be at odds with their suggestion that inflection, when supplied, is accurate. If Prévost and White predict that inflection in person agreement is accurate (contra the impaired-syntax view), then they cannot also predict an asymmetry across persons.

DEFAULT MORPHOLOGY IN L2 SPANISH

203

spontaneous production studies conducted to date have been case studies of one particular L2 speaker (Lardiere 1998a, 1998b, Franceschina 2001) or have had a very small number of participants (Prévost & White 2000). The present study examines production data from eleven L2 speakers, and can be interpreted as more representative of L2 Spanish speakers in general. Second, this study also avoids a complication of some earlier studies that looked for evidence of knowledge of finiteness, tense, and agreement in languages that either lack unique nonfinite bound morphology (English) or show extensive homophony across persons (French, English). Spanish, on the other hand, has relatively little homophony across persons. In addition, Spanish has morphologically-identifiable nonfinite and present-tense forms, whereas in English nonfinite and present-tense forms are largely homophonous. Spanish therefore allows a more detailed analysis of L2 errors, and can reveal whether defaults are truly nonfinite, as the MSIH predicts, or finite, with nontargetlike inflection for tense and/or agreement. 1. 1.1.

Background: Underspecification Underspecification in Theoretical Morphology Theories of underspecification emphasize economy of representation by excluding redundant information. Arguments in favor of underspecification cite its predictive capacity regarding the shape of inventories, the restriction of combinatorial possibilities available to the grammar, and the neutralization of contrast under certain conditions. I assume that underspecified forms correspond to unmarked forms, following assumptions made in theoretical literature on morphological features (see Noyer 1992, Harley & Ritter 2002, Carstairs-McCarthy 1998, and many others). In this section, I present independent criteria for establishing markedness values for the variables under consideration, and make reference to the theoretical literature that adopts these values as unmarked. For the variable of person, there is some evidence to suggest that 3rd person is universally unmarked. Typologically, unmarked values tolerate more distinctions than marked ones; 3rd person is more likely than 1st or 2nd person to show gender/number distinctions, suggesting that 3rd person is unmarked. Harley and Ritter (2002) adopt a feature-geometric approach to person-number features, and use typological evidence to motivate their geometry. Markedness values are reflected in their geometry, in that unmarked values have fewer nodes. Under their approach, 3rd person and singular are underspecified: 3rd person is realized via the absence of a participant node; singular is realized via the presence of a bare individuation node and absence of a group node.

204

CORRINE McCARTHY

For the variable of tense, structuralist morphology has generally assumed that past is marked relative to present or ‘nonpast’. Jakobson (1957/1971) defines markedness in terms of the generality of the unmarked category; Greenberg (1966:26) states this concept as the “ambiguous nature of the unmarked term”, and applies it as a diagnostic for establishing markedness values. Whereas the meaning of the marked category states the presence of a given feature, the meaning of the unmarked category states nothing about the presence of a given feature. In Spanish and English, the unmarked category of present (nonpast) can surface in past contexts, such as the historical present of narratives. In more recent theoretical literature on tense, Cowper (2004) has proposed a feature-geometric approach to the representation of tense and aspect in the inflectional system of Spanish. In the proposed geometry, the feature [precedence] means that at least one moment associated with the event denoted by the clause precedes the temporal anchor of the clause; if precedence is absent, the unmarked relation of simultaneity is established. Thus past is marked relative to present, as it is associated with the presence of [precedence], whereas the present is not. For verbal morphology relating to finiteness, it is further assumed that nonfinite forms are underspecified for finiteness, as they lack specification for tense. For the variables of person, number, past/nonpast tense, and finiteness, underspecification means that there is an asymmetry built into the feature specifications of morphemes, whereby some morphemes are associated with more features than others. This asymmetry distinguishes the current proposal from full-specification theories by making it possible to predict which morphemes will emerge as defaults.2 1.2.

Deriving Defaults I assume Distributed Morphology, a theory that adopts underspecification and late-insertion of vocabulary items. I further assume privative feature values: features are present or absent, with absent features corresponding to unmarked values. Vocabulary insertion is a competition in which the most highly specified vocabulary item, barring feature clash, is inserted into the fully-specified syntax. Where no features match between the terminal node and feature bundle, an elsewhere form is inserted. Competition for vocabulary insertion proceeds from the most highly specified entry to least specified entry (the elsewhere form). Where there is an equal number of 2

In principle, the adoption of full-specification is not fatal in itself. One may appeal to feature hierarchies in order to capture the facts I will present here, but current L2 theories have not done so. Underspecification allows the facts to be predicted without the need for extraneous mechanisms.

DEFAULT MORPHOLOGY IN L2 SPANISH

205

features specified for two or more forms, the order must be stipulated. (For details on vocabulary insertion, see Halle and Marantz 1993). According to the markedness values established in Section 2.1, [3], [singular], [present], and [nonfinite] are all, by assumption, unavailable as features. In principle, two types of errors may occur in the competition for lexical insertion: feature clash and underspecification. For person agreement, [3] is assumed to be underspecified. Suppose the syntax supplies the [1] feature, but instead -as is produced, a realization of the [2] feature: (2)

yo I

hablas speak-2SG

This is an error of feature clash between the syntax, which supplies [1], and the vocabulary item, which is associated with [2]. Suppose, on the other hand, that the syntax supplies the [1] feature, but instead -a is produced, a realization of the elsewhere condition: (3)

yo I

habla speak-3SG

This is an error of underspecification: where the more highly specified form -o should have won the competition for vocabulary insertion, the underspecified form was chosen instead. This does not result in feature clash, since the elsewhere morpheme represents an absence of features. Extending this pattern from person to number, the substitution of a singular form for a plural one constitutes an underspecification error.3 For tense, the same logic applies: if the syntax supplies the feature [present] in a present-tense context, the insertion of a past morpheme results in feature clash between [past] (or, following Cowper 2004, [precedence]) and [present]. On the other hand, the insertion of present-tense morphology in a past-tense context results in an error of underspecification, as present-tense morphology, by assumption, lacks a specification for tense. It is hypothesized that learners will avoid feature clash in their interlanguage grammar. When errors occur, they are errors of underspecification. I propose the hypothesis in (4): 3

While I focus on the use of 3rd singular as a default, it should be noted that these are not the only kind of underspecification errors that are possible under the hypothesis I advance here. For example, 1st singular may substitute for 1st plural without any feature clash; 3rd plural may substitute for 1st plural without any feature clash. As it turns out, these type of errors do not occur.

206

CORRINE McCARTHY

(4)

Morphological Underspecification Hypothesis: L2 errors are ones of underspecification, not of feature clash.

2.

Methodology The data come from spontaneous production of 11 speakers of Spanish as a second language. All participants were asked to rate their level of proficiency in spoken Spanish. Responses ranged from intermediate to advanced, except for one who reported being near-native. All participants scored either intermediate or advanced on a written proficiency test. Interviews were conducted by a native speaker of Spanish. Participants were told that they should consider the interview a casual conversation, and were encouraged to ask questions of the interviewer if they wanted. Interview lengths ranged from 15 to 35 minutes. Speech was transcribed by a near-native speaker of Spanish. Following the methodology of Lardiere (1998a,b), utterances that were followed by self-correction were excluded; the final, corrected forms were included. Self-repetitions and repetitions of the interviewer were excluded. As Spanish permits null subjects, only those verbs whose (null or overt) subject was judged to be unambiguous were included in the analysis. This was done in order to eliminate any chance of reporting an error when none was produced. Tokens that were produced with indeterminate stress were not counted as errors; for instance, some tokens such as habló “speak-3SG-PAST” bore equal stress on both syllables, and could not be distinguished from hablo “speak1SG”. For tense, only those contexts in which past tense was obligatory, but present tense was supplied, were counted as errors. 3.

Results Error types are presented in Table A. Excluding data from one participant, 92 percent of errors are underspecification errors. One exceptional participant (Sheila) repeatedly employed a first-person past preterite copula/auxiliary, estuve “was”, in 3rd person contexts, constituting an error of feature clash. She does not, however, use first person as a default in other tenses: in simple present, 3rd person surfaces twice in a non-targetlike context. Underspecification Errors Feature Clash Errors Total Errors

Participants 1-10 11 (92%) 1 (8%) 12

Participant 11 3 (12%) 23 (88%) 26

Table A.: Person agreement: Number of tokens by participant and error type

(5,6) below are examples of errors of underspecification in person agreement.

207

DEFAULT MORPHOLOGY IN L2 SPANISH

In (5), the participant asks a question that lacks 2nd person agreement. There was a slight pause between repetitions of the verb, as she was waiting for the interviewer to answer her. The interviewer didn’t understand her the first time, since the verb’s agreement indicated 3rd person. (6) was uttered in a context in which the participant was asked for information about herself, making the intended referent 1st person. (5)

y manejará? (pause) manejará tú? (Samantha, intermed.) and drive-FUT-3SG drive-FUT-3SG you “and will you drive?”

(6)

nació en Boston be.born-PAST-3SG in Boston “I was born in Boston.”

(Beth, intermed.)

For number agreement, results are presented in Table B.4 Underspecification errors account for 92 percent of all number errors produced. Underspecification Errors Feature Clash Errors Total Errors

N 25 (92%) 2 (8%) 27

Table B: Number agreement on verbs: Error type

All underspecification errors for number involved the substitution of a 3rd singular form for a 3rd plural one. An example of an underspecification error is given in (7). (7)

Los italianos puede entender un poco the Italians can-3SG understand a little “The Italians can understand a little (Spanish)”

(Linda, advanced)

The results for the variables of tense (past/nonpast) and finiteness are given in Table C. While present-tense morphology sometimes occurs in obligatory past contexts, the reverse does not occur. Nonfinite morphology occurs a few times in finite contexts, but the reverse does not occur.

4

There were a few tokens of plural agreement with a singular noun gente “people”. These were not coded as errors, since gente sometimes occurs with plural agreement in the speech of native Spanish speakers.

208

CORRINE McCARTHY

Underspecification Errors Feature Clash Errors Total Errors

Tense Finiteness 17 (100%) 4 (100%) 0 0 17 4

Table C: Past/nonpast tense and finiteness on verbs: Error type

Examples of present-tense morphology in obligatory past contexts are given in (8,9). Nonfinite morphology in a finite context is given in (10): (8)

Cuando nos conocimos, yo no hablo (Tom, advanced) when REFL meet-PAST-1PL I NEG speak-PRES-1SG ningún palabra en español NEG word in Spanish “When we met, I didn't speak a word of Spanish.”

(9)

Nací en la ciudad de Nueva York, (Rachel, advanced) be.born-PAST-1SG in the city of New York, pero mis padres se mudan en 1985 but my parents REFL move-PRES-3PL in 1985 “I was born in the city of New York, but my parents moved in 1985.”

(10) yo nunca hacer los I never do-INF the “I never do the dishes.” 4.

platos dishes

(Beth, intermed.)

Discussion The results presented here support the hypothesis that learners avoid feature clash in their productions. The majority of errors in production are errors of underspecification, barring the productions of one exceptional participant for the variable of person. Overall, the results show a clear preference for finite defaults over nonfinite ones, something that is inconsistent with the MSIH (Prévost & White 2000). The MSIH can account for only those errors in verbal morphology that involve the substitution of a nonfinite (infinitive) verb in a finite context, or 4 of 86 errors (5 percent) in verbal morphology. By adopting the underspecification-based model here, 60 of 86 errors (70 percent) are accounted for, or 57 of 60 (95 percent) if we exclude Sheila’s data for the variable of person. These data suggest that morphological errors are better characterized as instances of underspecified morphology than missing morphology.

DEFAULT MORPHOLOGY IN L2 SPANISH

209

One clear counterexample to the predictions made here is Sheila’s repeated use of 1st person preterite estuve “was” (infinitive estar) in 3rd person contexts (11). In (11), Sheila uses the wrong copula; ser would be used in this context, not estar. She has a tendency to use estuve anywhere she means was. (11)

el objetivo estuve ... the objective was-1SG “the objective was...”

(Sheila, intermed.)

Although this is clearly a problem for the MUH, it should be noted that Sheila uses 3rd person as a default in present tense. In addition, these forms are irregular past preterite forms, and Sheila shows no evidence that 1st person generally acts as a default in the past preterite. Current conceptualizations of L2 morphological variability focus on the presence or absence of an overt phonological marker that signals tense or agreement, such as English -ed for past, or -s for 3rd person singular. However, default morphology should not be equated to missing phonological marking. It is true that plural is associated with an overt marker in Spanish present indicative (-n, as in 3rd singular/plural habla/hablan), but Spanish 3rd and 1st person present indicative have an equivalent amount of morphological marking (hablo vs. habla), and past preterite 3rd person has more morphological marking than 1st person in the case of –er/-ir class verbs (nací “I was born” vs. nació “he/she was born”). These facts mean that the substitution of 3rd person for 1st person is not easily explained as the absence of an overt marking that signals 1st person. Finally, the results of this study illustrate the importance of feature inventories in the L2 grammar. L2 morphology should be considered at the level of features, rather than at the level of overt phonological marking. Furthermore, by adopting a feature-based approach to morphology, generalizations about learner errors are predictable across a range of domains. This study also adds to the literature on the role of markedness in second language acquisition, as markedness plays an important role in determining the features that are available in the L2 grammar. 5.

Conclusion The Morphological Underspecification Hypothesis correctly predicts the kinds of errors in L2 production. It predicts that learners will not produce errors that result in feature clash, but that errors of underspecification may occur. Underspecified features are, by assumption, those features that are

210

CORRINE McCARTHY

unmarked. These predictions were supported: errors produced were primarily underspecification errors. This pattern was found for person, number, past/nonpast tense, and finiteness in verbal morphology. The MUH is able to predict the shape of learner defaults under one broad principle: in interlanguage grammar, learners avoid feature clash, in much the same way they do in their native-language grammar. References Carstairs-McCarthy, Andrew. 1998. “How Lexical Semantics Constrains Inflectional Allomorphy”. Yearbook of Morphology 1997, ed. by Geert Booij & Jap van Marle, 1-24. Dordrecht: Kluwer. Cowper, Elizabeth. 2004. “The Geometry of Interpretable Features: INFL in English and Spanish”. Ms., University of Toronto. Ferdinand, Astrid. 1996. The Development of Functional Categories: The Acquisition of the Subject in French. The Hague: Holland Academic Graphics. Franceschina, Florencia. 2001. Morphological or Syntactic Deficits in Nearnative Speakers? An Assessment of Some Current Proposals. Second Language Research 17:3, 213-247. Greenberg, Joseph H. 1966. Language Universals with Special Reference to Feature Hierarchies. The Hague: Mouton. Halle, Morris & Alec Marantz. 1993. “Distributed Morphology and the Pieces of Inflection”. The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, ed. by Ken Hale & Samuel Jay Keyser, 111-176. Cambridge, Mass.: MIT Press. Harley, Heidi & Elizabeth Ritter. 2002. Person and Number in Pronouns: A Feature-Geometric Analysis. Language 78.482-526. Hawkins, Roger & Cecilia Yuet-hung Chan. 1997. “The Partial Availability of Universal Grammar in Second Language Acquisition: The ‘Failed Functional Features Hypothesis’”. Second Language Research 13:3.187226. Haznedar, Belma & Bonnie Schwartz. 1997. “Are There Optional Infinitives in Child L2 Acquisition?” Proceedings of the 21st Annual Boston University Conference on Language Development, ed. by Elizabeth Hughes, Mary Hughes, & Annabel Greenhill, 293-306. Somerville, Mass.: Cascadilla Press, Jakobson, Roman. 1957/1971.“Shifters, Verbal Categories and the Russian Verb”. Selected Writings, Vol. 2: Word and Language, 130-147. The Hague: Mouton. Lardiere, Donna. 1998a. “Case and Tense in the “Fossilized” Steady State”. Second Language Research 14:1.1-26. ----------. 1998b. “Dissociating Syntax from Morphology in a Divergent L2

DEFAULT MORPHOLOGY IN L2 SPANISH

211

End-state Grammar”. Second Language Research 14:4.359-375. ----------. 2000. “Mapping Features to Forms in Second Language Acquisition”. Second Language Acquisition and Linguistic Theory, ed. by John Archibald, 102-129. Oxford: Blackwell. Noyer, Rolf. 1992. Features, Projections and Affixes in Autonomous Morphological Structure. Doctorate dissertation, MIT. Prévost, Philippe & Lydia White. 2000. “Missing Surface Inflection or Impairment in Second Language Acquisition? Evidence from Tense and Agreement”. Second Language Research 16:2.103-133. Vainikka, Anne & Martha Young-Scholten. 1994. “Direct Access to X’theory: Evidence from Korean and Turkish Adults Learning German”. Language Acquisition Studies in Generative Grammar, ed. by Teun Hoekstra & Bonnie D. Schwartz, 265-316. Amsterdam: John Benjamins. White, Lydia. in press. “Some Puzzling Features of L2 Features”. The Role of Features in Second Language Acquisition, ed. by Juana Liceras, Helmut Zobl & Helen Goodluck. Mahwah, NJ: Lawrence Erlbaum.

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH*

ANA T. PÉREZ-LEROUX, MIHAELA PIRVULESCU & YVES ROBERGE University of Toronto 0.

Introduction Two of the main syntactic properties of early child language are: a) Early alignment with the distributional patterns of the target adult grammar as exemplified in (1a) (see e.g., Brown 1973, Pinker 1984); and, b) Constrained patterns of omissions of functional material as shown in (1b) (see Borer & Rohrbacher 2002), and of arguments as in (1c-d) (see Rizzi 2002 and Guasti 2002 for comprehensive reviews).1 (1)

a. b. c. d.

Elle roule pas “It does not roll.” Mary go. Ø want Ø! Mets Ø dedans “Put in.”

(Grégoire 1;11) (Sarah 2;3) (M, 1;11) (Grégoire 1;11)

Argument omission in child language raises important questions about the initial representation of syntactic categories in clausal structure and the relationship between the various stages of development. Our research concentrates on object omission, an area of research that did not receive extensive attention until recently (Jakubowicz et al. 1997, Müller & Hulk 2001, Schaeffer 2000, Wexler et al. 2003). The main question explored in this paper concerns the nature of early object omissions. Our objective is to apply recent approaches to transitivity *

We would like to thank I. Belzil, E. Scott, D. Dascalu, K. Orgill, R. Kadoguchi, G.Tom, S. Burns and C. Lam for assistance in testing and materials development, and T. Gruter, T. Kupisch and one anonymous reviewer for their comments and suggestions. This research is funded in part by a SSHRC grant (410-05-0239). 1 The examples in (1a,b,d) are from the CHILDES database; MacWhinney (2000). (1c) is from personal data.

214

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

and null objects in a Minimalist framework to the study of object omission in child grammar. We are particularly interested in the developmental resolution of the contexts for object omission. To the extent that there is variation in patterns of object omission across language, explanations for such variation need to examine crosslinguistic differences in the licensing and identification mechanisms used. We present a comparative study using an experimental cross-linguistic comparison (French vs. English) of children’s performance with two contexts that vary with respect to the informational demands placed on the object: a) individuated object (what is X doing with Y?): transitive structure with an object singled out; b) non-individuated object (what is X doing?): transitive with no object singled out. 1. On null objects For the purposes of this study we make use of the description of the French and English adult systems found in Cummins & Roberge (2004, 2005), which is based on the important contributions found in Larjavaara (2000) and Lambrecht & Lemoine (1996). According to Cummins & Roberge (2004, 2005), a direct object is always merged to V in VP and this Transitivity Requirement is a property of UG. The object can be null or lexical. There are two syntactic types of null objects in French; see 1.1. In (2), the null object corresponds to an unspecified object. We call this context ‘non-individuated’ and the object is analyzed as a null bare noun, a null cognate object. (2) Non-individuated interpretation: Null Cognate Object De loin, la France continue de séduire Ø, mais, de près, les Français agacent Ø. (Larjavaara 2000: 44) “From afar, France continues to attract, but close up, the French annoy.” In (3), the null object replaces an entity that is specified in the linguistic context.2 We call this context ‘individuated’ and the object is analyzed as a pro, which is normally recovered by an accusative clitic in French. Descriptively, this can be seen as a clitic-drop construction.

2

A reviewer points out that their interpretation of (3) is that the person is not interested in the book because s/he has already read something else, not that specific book. Note that while this interpretation is possible for (3), we are specifically interested in the other possible interpretation (see eg. Larjavaara 2000) according to which it is that specific book that has been read.

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

(3)

215

Individuated interpretation: Clitic-drop A: Tu veux ce livre? “You want this book?” B: Oh ! Mais j’ai déjà lu Ø. *“But I already read.”

Interactions between these two syntactic types and other modules of grammar, especially pragmatics, creates a rich and flexible system of null objects in French; see Cummins & Roberge (2005) for details. 1.1

Variation: On French and English Crosslinguistic variation should be found primarily in the particular identification mechanisms used in recovering the semantics and reference of null objects, and secondarily, in the behavior of specific lexical items. Cummins & Roberge (2005) propose the following typology of null objects in French. Only the null objects associated with N are available in English. Reference Antecedent Contextual recovery Syntax Semantics

Pragmatics

Referential non-referential reference to antecedent no reference to antecedent recovered clitic-drop Deictic contextual contextual by clitic ("delinked") clues clues present absent pro N f features via via deixis lexical semantics of V of clitic antecedent (i.e. Null Cognate Object) I-principle I-principle on I-principle I-principle on extralinguistic (less (more linguistic context stereotype, stereotype, context more less context) context) Table 1: Null Objects in French (Cummins & Roberge 2005)

Inasmuch as we can attribute the possibility in (3) in French to some kind of clitic-drop — quite independently of the exact nature of this drop — it follows that null objects in this context do not exist in English. As can be seen from the translation of (3), this is indeed the case. Therefore, the target grammar for French is complex in that it goes beyond what is provided by UG (N-types of null objects) to include the pro null objects associated with clitics and clitic-drop constructions (i.e. in individuated contexts). In contrast, the target grammar for English in the specific case of null objects does not go beyond the N-types (i.e. non-individuated contexts).

216

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

1.2

Null objects in child language Data leading to the determination of the types of null objects come from the spontaneous speech of adults or, as in Larjavaara (2000), a corpus with very specific properties. This can be problematic for an analysis of null objects in the spontaneous speech of children. Specifically, Pirvulescu & Roberge (2005) question the use of the notion of “obligatory context” as applied to object omission in spontaneous child speech. While this notion is helpful for omissions of clear grammatical elements, the range of object omission constructions found in adult speech sheds doubts on its usefulness. They also argue that adult judgments of the acceptability of children utterances with null objects are quite variable and not empirically reliable. Thus, controlled tasks in an experimental setting are needed. 2.

The experimental framework: the target grammars For individuated contexts, English and French vary in the way the object is realized when it has been previously mentioned and focused in the discourse. In French, both a null pronominal recovered by a clitic and a null pronominal not recovered by a clitic (clitic-drop) are possible.3 English only allows a pronominal in this situation. For non-individuated contexts, English and French are alike in that both a lexical and a null cognate object are possible. Current studies on object omission use variants of the following tasks (Jakubowicz et al. 1996, Schaeffer 1997, Ingham 1993/1994, and subsequent work): (4)

Individuated context (A girl is drawing a flower) English: A: What is the girl doing with the flower? B: She is drawing it. French: A: Qu’est-ce que la fille fait avec la fleur? B : Elle la dessine Ø or Elle dessine Ø

(5) Non-individuated context (A girl is eating a sandwich) English: A: What is she doing? B: She is eating the sandwich. or She is eating Ø.

3

The two possibilities in (4) in French share many properties but Cummins & Roberge (2005) do not claim that they are entirely equivalent. Indeed, they differ stylistically and discursively. Although to our knowledge this has not yet been proven statistically, European varieties of French seem to allow clitic-drop more freely than North American varieties.

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

French:

A: B:

217

Qu’est-ce qu’elle fait? Elle mange le sandwich or Elle mange Ø

2.1.

Hypotheses and predictions Our first hypothesis is that the different mechanisms for object realization in the target grammars will lead to different rates of object omission in individuated contexts for English and French children. Specifically the presence of the clitic system and the existence of clitic drop in adult French will present special difficulties for the French child. The second hypothesis is that English and French speaking children will not diverge with respect to the non-individuated context, since the presence of both the lexical/null objects is regulated by pragmatics and idiosyncratic verb variation. Previous studies concentrated on object omission in the individuated context and indicated a high rate of omissions in languages such as French and Italian. Significant object omission rates were found in experimental setting in French (omissions as high as 60% in a group of French children with MLU between 2.94 and 3.5; Jakubowicz et al. 1996, 1997); Italian (64% omissions in 2-year-olds; Schaeffer 2000) and Catalan (79% omissions in 1 to 2 yearsolds; Wexler et al. 2003). There are no studies comparing object omission in different contexts. 2.2.

Methodology: Elicited production Children were tested individually, using the elicited production paradigm described in (4) and (5). The experimenter asks a question to a puppet introduced as unreliable, and the child is asked to help in correcting the puppet’s factual mistakes. The puppet answers incorrectly, and the target question is repeated for the child. We tested each discourse condition with the same six transitive verbs, selected on the basis of the comparability of both the verbs and the objects across languages and to maximize the phonological salience of the clitic in French if the child produced it. (6) Construire/Build: castle/tower Dessiner/Draw: flower/girl Manger/Eat: cake/sandwich 2.3.

Boire/Drink: juice/milk Frapper/Hit: ball/piñata Couper/Cut: paper/curtain

Story conditions: individuated vs. non-individuated In the individuated condition, the story included an explicit mention of the direct object, to introduce it as a possible discourse antecedent. The prompting question in this condition focused on the VP, and the direct object

218

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

was mentioned as a topical adjunct. The puppet’s factual mistake also involved the direct object, with the wrong verb (i.e., the puppet claims the girl smells the flower instead of drawing it). The non-individuated condition consisted of events and illustrations fully comparable to those in the individuating condition, with two differences. First, there was no direct mention of the objects in the story. Second, the puppet error referred to an entirely different event that does not involve the target object (i.e., phoning his friend rather than eating the bone). Individuated: VP focus with object topic: What did x do with Y? STORY : La petite fille veut dessiner la plus belle chose de la salle. Regarde! Elle prend un crayon et elle dessine la fleur. EXP : Hey Croco. Qu’est-ce que fait la fille ? CROCO : Je sais! La fille sent la fleur. CHILD : ________(NO) CROCO : Non? La petite fille ne sent pas la fleur? EXP : S’il te plaît, dis à Croco ce que la petite fille fait avec la fleur. CHILD :_____________[elle la dessine] ----------STORY: The little girl wants to draw the prettiest thing in the room. Look! She takes a pencil and draws the flower. EXP: Hey, Croco! What's the girl doing? CROCO: I know! The girl is smelling the flower! CHILD:____________(NO) CROCO: No, the little girl isn't smelling the flower? EXP: Please tell Croco what the little girl is doing with the flower. CHILD:____________(she's drawing it) Non-individuated: VP focus with no previous mention of object: What did x do? STORY: Regarde Bertrand. Il a l’air très content. MMMM. EXP : Qu’est-ce que Bertrand fait? CROCO: Je sais, je sais. Il appelle son ami. CHILD : ________(NO) CROCO: NON? Qu’est-ce que Bertrand fait? CHILD: ___[Bertrand mange l’os/ il mange.] …………………….. STORY: Look at Clifford. He seems very happy. MMMM. EXP: What is Clifford doing? CROCO: I know! I know! He's calling his friend. CHILD:________(NO) CROCO: No? He isn't calling his friend? EXP: Please tell Croco what Clifford's doing. CHILD:______[Clifford is eating (his bone).]

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

219

3. 3.1.

Results Development in English Seventeen English-speaking children aged 2;9-4;11 (Mean = 3;8.5, Median = 3;7), were recruited in English speaking daycares in Toronto. A group of 12 adults acted as controls. As shown in Figure 1, overall, the English-speaking children’s performance is already very close to the adult target performance. In non-individuated contexts, the data are almost identical: A high proportion of responses are lexical NPs (approximately 60-70%), and the remainder (25-30%) was primarily null objects. There were few (5% or less) pronominal mistakes made by either group, showing that both children and adults attend to the discourse status of the referent, and avoid replacing the non-individuated objects with pronouns. For the individuated contexts, adults and children produce the same amount of lexical NPs (close to 30%, which is an over-explicit but not ungrammatical response, as shown in (7)). (7) Q: What did she do with the flower? A: She drew the flower. The remainder of adult responses is made up of the expected target pronominal responses. In contrast, children fall short of this mark: they produce close to 20% illicit null objects. This occurs at the expense of the target response of overt pronominals, which makes up approximately half of the answers.

Figure 1: Proportion of direct object types produced in response to individuated and non-individuated contexts by English children and adults

220

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

Developmentally, the older children appear to have less null objects in both contexts. We found a negative correlation between age in months and null objects for both conditions, but this reaches significance only for the individuated object contexts (*individuated object contexts r=-.509, p=.03; n.s. non-individuated objects r=-.125, p=.63). 3.2.

Development in French Twenty-seven French-speaking children aged 2;8 and 4;3 (mean = 3;6.5, median = 3;7) were recruited in French-speaking day cares in Toronto. These daycares were comparable in location, and social status to those of the English daycares. A group of 9 adults served as control group. For the French data, the comparison between children and adults reveals a radical difference; see Figure 2. In the individuated condition, adults produced mostly clitics (61%), some lexical NPs (35%) and virtually no null objects (4%). In the non-individuated condition, they produced 73% lexical NPs 25% null object responses and no pronominal responses, as expected (2%).

Figure 2: Proportion of direct object types produced in response to individuated and non-individuated contexts by French children and adults.

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

221

In contrast, children produced very few clitics in the individuated condition: only 10% clitic responses, but 38% lexical, and 52% null. Like their English-speaking counterparts, the Francophone children produced lexical NP responses in the individuated object condition at rates close to their controls. However, the French children traded-off their target production of clitics for that of null objects. Interestingly, we find the mirror image of the adult data in the non-individuated contexts: children produce more null than lexical objects (about 50% and 40% respectively), and the converse is true of the adult (25% and 70%). Like the English-speaking children, as well as the French controls, these children do not use object clitics in these contexts, demonstrating pragmatic control of pronoun use. The French children were also different from the English children in that the former show no discrete effect of age in development. The correlation between age in months and proportion of null objects is positive in both contexts, but non-significant (individuated, r=.337, p=.08; non-individuated, r=.321, p=.1). We also did not observe an age effect for the clitics. Only 6 children produced clitics, and they were in the middle range of age (3;4-3;7). The most striking fact is that there were no quantitative differences in the production of null objects across contexts for the French children. Approximately half the answers in the non-individuated contexts are null objects, where such an answer is grammatical, and the same proportion is found as well in the individuated context, where it may be considered an illicit response, giving the adult’s overwhelming preference for overt clitics under these experimental conditions. The rates of object omission in nonindividuated contexts for children and adults are significantly different at F1,34=10.107, p=.0031. 3.3.

The children compared To test our hypotheses, we examined the quantitative differences between French and English children in both contexts, as shown in Figures 3 and 4. In the individuated contexts, the children are similar in their production of lexical objects, but they differ sharply in terms of the production of pronouns (sizeable in English) and clitics (minimal in French). Consequently, French children produce more null objects than English children; the difference is shown to be significant by a t-test (mean difference=.321, p=.0019).

222

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

Figure 3: Distribution of responses on individuated contexts for children per language group.

In non-individuated contexts, both groups produce negligible proportions of pronouns/clitics, showing sensitivity to pragmatics. The Frenchspeaking children produce more null objects at the expense of lexically realized objects. This difference is smaller but statistically significant (mean difference=.218, p=.015). This goes against our hypothesized no crosslinguistic difference in non-individuated contexts.

Figure 4: Distribution of responses on non-individuated contexts for children in the two language groups

Recall that we hypothesized language differences in the clitic (individuated) contexts but no language differences where null objects and lexical objects freely alternate (i.e., in non-individuated contexts). We concentrate on the observed differences in the next section.

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

4.

223

Discussion Children in this comparative study behave differently across language groups in both the predicted context (clitics vs. pronouns) as well as in the nonindividuated context where we predicted comparable behavior. The developmental facts in this cross-language comparison are directly relevant to the issue of the grammatical status of object omissions. As seen above, object omissions did not decrease with age for the French children, but they did for the English children. It seems therefore that object omission in English children might be compatible with a sporadic, performance-driven fall back to a default null object (in the sense of Lebeaux 1988 and Roeper 1999). In contrast, object omissions in the French children require an explanation at the level of the grammar. The French study shows a radical discontinuity between children’s data and the adult controls’: Children prefer null objects in the individuated context (where clitics are expected, and uniformly produced by adults) and significantly more null objects than adults in the non-individuated context (where no clitics are involved). We therefore propose that an overgeneralized null object is capable of both serving as a referentially bound anaphora (in individuated condition) as well as capable of producing independent reference (in the non-individuated condition). Another point of interest is the absence of pronominal errors in the nonindividuated conditions in both language groups. This fact undermines pragmatic accounts of pronoun omission in individuated contexts. Pronoun use in this context would be a pragmatic violation, since the referent for it has not been introduced. That these errors do not occur argues for pragmatic differentiation of the two contexts. In sum, crosslinguistic differences suggest that different patterns of object omission in the target language (see Table 1) can have a substantial impact on grammatical development. What makes this developmental difference more striking is that it cannot be predicted directly from the localized behavior: adults perform comparably in this experiment across language groups, producing primarily pronouns when the context is individuated (i.e., as a response to what is she doing with the flower?) and alternating omissions and lexical DPs when it is not (as response to what is she doing?). The striking high rates of null objects in French data make it appear as if these children have not acquired object clitics. But we argue that this is not the case. First, longitudinal studies of spontaneous speech document the emergence of object clitics well before the age of the children in our experiment. Children produce clitics around the age of 2.3; cf. van Kampen 2004, Pirvulescu 2005, Hamann 2003). Second, our data yield higher rates of

224

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

omission than other experimental studies in French. Schmitz et al (2004) tested 17 French children mean age 2;10 produced 23% accusative clitics, using a similar task set up as a fishing game. Jakubowicz et al. (2000) tested 3 and 4 year olds (N=12 in each group) using a picture plus probe question (Que fait Kiki a Nonours? ‘What did Kiki do to Nonours’). The 3 year olds produced approximately 40% clitics and the 4 year olds 78%. The source of these different results is likely methodological, originating from a number of factors: prompt type (the target referent as an explicit theme, rather than as a related adjunct, e.g. do to instead do with), saliency (our stories mentioned the referent verbally several times besides the prompt), animacy (our stories had inanimate objects, which favor omissibility; Serratrice et al. 2004). We can safely assume that our children are capable of producing clitics but are failing to meet the informational demands of the present task, thus falling far from adult standards. Leaving aside the issue of magnitude of omission, two facts remain: 1) French and English children have large differences with identical tasks, and 2) French children overgeneralize null objects across contexts. For the overgeneralization we observe in this data, we can consider the standard explanations. First, the possibility of a morphophonological gap, i.e. avoidance of incorrect forms where a morphophonological paradigm has not been fully acquired. Borer & Rohrbacher (2002) propose an account for the apparent error-free acquisition of inflectional material, where children omit functional elements but rarely err by inserting the wrong morpheme. This pattern of acquisition, they argue, provides support for implicit knowledge of the underlying grammatical structure. Under this view the child would have the equivalent of an adult clitic construction in the individuated context, but the clitic fails to spell-out. This hypothesis would be very appealing given the results from English and French children: omissions in full pronoun languages (English) are less frequent than in the clitic language (French). This line of analysis predicts specific difficulties with clitic constructions, resulting in omissions, but it would not explain why French children overextend omissions to the non-individuated context, where no clitics are involved. This overextension is also difficult to explain within the maturational view of Wexler et al (2003), designed to explain asymmetries in clitic omission in the various clitic languages. We therefore suggest an explanation based on the idea of parametric discontinuity between children and adult grammar, in the sense of Rizzi 2002. Specifically, it is striking that French children omit objects to the same extent in both individuated and non-individuated contexts (see Figure 3 and Figure 4). We tentatively suggest that overgeneration of object omissions in individuated and non-individuated contexts has the same source. Looking back at the null

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

225

object typology in Table 1, we see that for adult French there are two possibilities: a DP level object pro and a null bare noun. One possible interpretation of our results would be that children start with a null bare noun (N) inside the VP for both individuated and non-individuated contexts. When the possibility of grammatical identification becomes established, the domain of application of the null bare noun reduces: it becomes restricted to nonindividuated contexts and pragmatically inferred contexts.4 The French children are not yet able to grammatically interpret variation in null objects across domains (pro and N in Table 1) and use the generalized null bare noun default. English children, on the other hand, only have available a restricted projection on null bare noun (N in Table 1). One potential problem for this account is that French children produce significantly more null objects than English children in non-individuated contexts (Figure 4). This suggests that French children behave differently from English children in this context, which is unexpected under the null bare noun account: this type of null object should be equally attested in both English and French early grammars. This opens the possibility of interpreting the French early grammar as being of a topic-drop type, as proposed by Schmitz et al. (2004). Specifically, in this grammar null object arguments are a minimum NP-pro bound by a discourse-identified zero topic. Alternatively, their null cognate object has fully retained all referential properties, and is capable of functioning both as null anaphora (in individuated contexts) and of independently referring (in non-individuated contexts). Clearly, more research is needed to tease out these two possibilities. 5.

Conclusion We return to the question we asked at the beginning of this article, namely, what is the nature of early object omission? Our answer, from the data gathered at this point, is that cross-linguistically, object omission does not have the same source: children can have residual omission, as in English, or object omission can be an active property of the child grammar, as in French. We have shown that the input that constitutes the target adult grammar interacts in crucial ways with initial UG-based settings to create differential developmental paths. Specifically, we propose that French children, faced with an input containing a variety of null objects, retain the minimal type (null cognate object) and overextend it beyond the distribution found in adults.

4

In an independent study of bare Noun objects in Catalan, Gavarro et al. (2004) found evidence of children overextending generic bare Noun objects to specific (i.e. individuated) referents.

226

PÉREZ-LEROUX, PIRVULESCU & ROBERGE

References Borer, Hagit & Bernhard Rohrbacher. 2002. “Minding the absent: Arguments for the Full Competence Hypothesis”. Language Acquisition 10: 123–175. Brown, Roger. 1973. A first language: the early stages. Cambridge, Mass.: Harvard University Press. Cummins, Sarah & Yves Roberge. 2004. “Null Objects in French and English”. Contemporary Approaches to Romance Linguistics, ed. by Julie Auger, Clancy Clements & Barbara Vance, 121-138. Amsterdam: John Benjamins. -------. 2005. “A Modular Account of Null Objects in French”. Syntax 8: 44-64. Gavarrò, Anna, Ana Teresa Pérez-Leroux & Tom Roeper. 2004. “Definite and Bare Noun Object Contrasts”. Paper delivered at The Romance Turn, Workshop on the Acquisition of Romance Languages. Madrid, Spain. Guasti, Maria Teresa. 2002. Language acquisition: the growth of grammar. Cambridge, Mass.: MIT Press. Hamann, Cornelia. 2003. “Phenomena in French normal and impaired language acquisition and their implications for hypotheses on language development”. Probus 15.91-122. Ingham, Richard. 1993/1994. “Input and learnability: direct object omissibility in English”. Language Acquisition 3: 95-120. Jakubowicz, Celia, Natascha Müller, Ok-Kyung Kang, Beate Riemer & Catherine Rigaut. 1996. “On the acquisition of the pronominal system in French and German”. Proceedings of the 20th BUCLD, ed. by Andy Springfellow, Dalia Cahana-Amitay, Elizabeth Hughes & Andrea Zukowski, 374-385. Somerville, Mass.: Cascadilla Press. ----------, Natascha Müller, Beate Riemer & Catherine Rigaut. 1997. “The case of subject and object omissions in French and German”. Proceedings of the 21st BUCLD, ed. by Elizabeth Hughes, Mary Hughes & Annabel Greenhill, 331-342. Somerville, Mass.: Cascadilla Press. ----------, Celia, Laurice Tuller, & Catherine Rigaut. 2000. “Phonologically weak items in abnormal acquisition in French”. Proceedings of the 24th BUCLD, ed. by S. Catherine Howell, Sarah A. Fish & Thea Keith-Lucas, 450-461. Somerville, Mass.: Cascadilla Press. Van Kampen, Jacqueline van. 2004. “Learnability order in the French pronominal system”. Selected Papers from Going Romance 2002, ed. by Reineke Bok-Bennema, Bart Hollebrandse, Brigitte Kampers-Manhe & Petra Sleeman (eds.), 163-183. Amsterdam: John Benjamins. Lambrecht, Knud & Kevin Lemoine. 1996. “Vers une grammaire des compléments zéro en français parlé”. Absence de marques et

EARLY OBJECT OMISSION IN CHILD FRENCH AND ENGLISH

227

représentation de l'absence 1, ed. by Jean Chuquet & Marc Frid, 279-309. Rennes: Presses universitaires de Rennes. Larjavaara, Meri. 2000. Présence ou absence de l’objet. Limites du possible en français contemporain. Helsinki: Academia Scientiarum Fennica. Lebeaux, David. 1988. Language acquisition and the form of grammar. Doctoral dissertation, University of Massachusetts at Amherst. MacWhinney, Brian. 2000. The CHILDES Project: tools for analysing talk. Mahwah: Lawrence Erlbaum. Müller, Natascha & Aafke Hulk. 2001. “Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient languages”. Bilingualism: Language and cognition 4.1-21. Pinker, Steven. 1984. Language Learnability and Language Development. Cambridge, Mass.: Harvard University Press. Pirvulescu, Mihaela. 2005. “The acquisition of object clitic in French L1: spontaneous vs. elicited production”. Paper delivered at Generative Approaches to Language Acquisition. Siena, Italy. ------- & Yves Roberge. 2005. “Licit and Illicit Null Objects in L1 French”. Theoretical and Experimental Approaches to Romance Linguistics, ed. by Randall Gess & Edward Rubin, 197-212. Amsterdam: John Benjamins. Rizzi, Luigi. 2002. “On the grammatical basis of child language development: A case study”. Ms., University of Siena. Roeper, Thomas. 1999. “Universal bilingualism”. Bilingualism: Language and Cognition 2.169-186. Schaeffer, Jeannette. 1997. Direct Object Scrambling and Clitic Placemen in Dutch and Italian Child Language. Doctoral dissertation, UCLA. -------. 2000. The Acquisition of Direct Object Scrambling and Clitic Placement. Amsterdam: John Benjamins. Schmitz, Katrin, Katja Cantone, Natascha Mueller & Tanja Kupisch. 2004. “Clitic realizations and omissions in early child grammar: a comparison of Italian and French”. Paper delivered at The Romance Turn, Workshop on the Acquisition of Romance Languages. Madrid, Spain. Serratrice, Ludovica, Antonella Sorace & Sandra Paoli. 2004. “Crosslinguistic influence at the syntax–pragmatics interface: Subjects and objects in English–Italian bilingual and monolingual acquisition”. Bilingualism: Language and Cognition 7.183-205. Wexler, Kenneth, Anna Gavarrò & Vicenç Torrens. 2003. “Object clitic omission in child Catalan and child Spanish”. Reports de Recerca, Grup de Gramatica Teorica, Universitat Autonoma de Barcelona.

AGREEMENT PARADIGMS ACROSS MOODS AND TENSES∗

MIHAELA PIRVULESCU University of Toronto

0.

Introduction Within recent developments of the Minimalist Program (Chomsky 2000), an important research agenda investigates how much information is exchanged at the syntax-morphology interface and the role of each component. The simplest assumption within the Minimalist Program is a model where, in the default case, there is a transparent interface between morphology and syntax, as in Distributed Morphology (Halle and Marantz 1993, Bobaljik 2000, Embick and Noyer to appear): morphology is a direct interpretation of syntactic structure. More specific morphological operations are departures from the default assumption and are therefore highly constrained and language specific. In such a model, it becomes crucial to investigate cross-linguistic morphological generalizations/constraints because they should either follow from the syntactic structure or be proven not to exist. In this article, I will investigate the agreement paradigms of the subjunctive and the imperative across Romance languages. I will depart from the observation that these paradigms seem to always be identical to an indicative agreement paradigm. The generalization I propose is the following: the realization of agreement in subjunctive and imperative verbs is a consequence of the syntactic status of Tense in these two moods. The generalization put forward here has a clear limitation: it is based on evidence from Romance languages. However, examples from unrelated languages outside the Romance area, seem to support the generalization. I will depart from an observation made by Carstairs (1987:80): (…) certain morphosyntactic property contrasts, such as ones involving Aspect and Tense, seem to define partitions of verbal paradigms which ∗

I would like to thank Carmen Dobrovie-Sorin,Virginia Hill, Yves Roberge and one anonymous reviewer for their comments and suggestions.

230

AGREEMENT PARADIGMS

are morphologically in some sense more fundamental than other property contrasts such as those of Person and Number.

My aim is to begin where the quote above from Carstairs (1987) has left off. Within Distributed Morphology, I propose a precise characterization of what it means for one morphosyntactic feature (such as tense features) to be more fundamental than another (such as agreement features). I will also examine some of the consequences of this asymmetry. Specifically, I will address the following questions: a) In what sense can a tense feature be said to define verbal agreement paradigms? b) What happens if a morphosyntactic contrast, such as Tense, is not active at the morphological level? The set of fundamental assumptions of Distributed Morphology of interest here are the following (see, among others, Embick and Noyer, to appear): a) syntax is the only generative component of the grammar; b) morphology interprets syntactic structures, rather than feeding them; c) in the default case, morphological structure is simply syntactic structure; d) additional PF processes are minor operations that manipulate nodes generated by syntax in a sharply constrained fashion (i.e. insertion or copying of nodes/features, impoverishment, rebracketing and movement); e) PF operations are constrained, among other things, by being language specific; i.e. each rule is triggered by a language-specific requirement that must be learned by speakers of that language. The article is organized as follows: Section 2 presents the relevant data from Romance languages and proposes the core idea of this article: agreement realizations in subjunctive and imperative always have only the most widely distributed agreement affixes – the elsewhere affixes, which are not marked for tense. In Sections 2.2 and 2.3 this is illustrated with a morphological analysis of Romanian verbal agreement affixes. Section 3 discusses the syntactic representation of a verb in subjunctive and imperative mood respectively. In Section 4, the implications of the proposed generalization are discussed. 1.

Morphological patterns The observation at the core of this article is that when looking at agreement verbal paradigms across Romance languages, one cannot help but notice close resemblances between certain paradigms. I will focus here on the similarity that is robustly present across Romance languages: the one between the indicative (present or imperfect), the subjunctive present and the imperative.

MIHAELA PIRVULESCU

231

Data from Romance languages1 Looking across Romance languages in general, a striking similarity is noticed: subjunctive paradigms are identical, in most cases, to the indicative paradigms (present or imperfect), as Table 1 illustrates: 1.1.

Romanian

French

Spanish

Subj.~Ind.

Subj.~Ind.

Subj.~Ind.

Portuguese

Latin

Italian

Subj.~ 2 Subj.~ Ind. Subj. Imperfect Imperfect 1sg -ø -ø -ø -ø -m -o -ø 2sg -i -ø -s -s -s -i -ø 3sg -ø -ø -ø -ø -t -ø -ø 1pl -m -ons -mos -mos -mus -mo -o 2pl -ţi -ez -is -is -tis -te -te 3pl -ø -ø -n -m -nt -no -o Table 1: Agreement paradigms for the indicative (present or imperfect) and the present 3 subjunctive in Romance languages

Table 1 presents agreement affixes for subjunctive and indicative (present or imperfect) across the Romance languages. In those languages (except Italian) the agreement affixes are identical in subjunctive and indicative (Subjunctive ~ Indicative). In Italian, the forms of the agreement in subjunctive are a subset of those in indicative present (two separate columns). In these languages, the same resemblance can be noticed with respect to the paradigms of the imperative and the present of indicative. Romanian

Italian

2 sg.

vorbeste-Ø ! “speak” ~ind. 3sg

2 pl

vorbi-ţi ! ~ind. 2pl

parla-Ø! “speak” ~ind. 2sg or 4 3sg parla-te! ~ ind. 2pl

1 pl.

-

1

Portugu ese fala-Ø! “speak” ~ind. 3sg fala-i ! specific ending -

Spanish

French

Latin

habla-Ø! “speak” ~ ind. 3sg

parle-Ø! “speak” ~ ind. 3sg or 2sg parl-ez! ~ ind. 2pl

amā-Ø! “love” specific ending amā-te! specific ending

habla-d! specific ending -

parlia-mo! parl-ons! ~ ind. 1pl ~ ind. 1pl Table 2: Agreement paradigms for the present indicative and imperative in Romance

The affixes presented in Table 1 and Table 2 are based on the segmentation provided by the following authors : Iliescu and Mourin (1991) for Romanian ; Le Goffic (1997) for French ; De Bruyne (1995) for Spanish ; Hills and Ford (1944) for Portuguese ; Bennett (1900) for Latin ; Ragusa (1984) for Italian. 2 Past Perfect and conditional have the same pattern. 3 Table 1 presents only the agreement endings because in some languages there is variation in material other than agreement forms between indicative and subjunctive (for example different theme vowels in Romanian and Spanish). Full forms would have taken too much space. 4 Depending on the verbal class.

232

AGREEMENT PARADIGMS

The examples presented above in Tables 1 and 2 show that in the verbal inflection system, agreement paradigms in subjunctive and imperative are identical to the indicative (present or past).5 What is the status of this identity? Is this accidental identity over surface phonological forms, or does it come from something deeper in the grammar? In this article, I will explore the second possibility. Given the framework of Distributed Morphology adopted in this article, a deeper explanation would have to be placed at the syntactic level: cross-linguistic morphological generalizations should follow from the syntactic structure. I propose that the identity of the agreement paradigms of the subjunctive and the imperative with the indicative (present or past) is the result of the way Tense features are represented at the syntactic level. The reason for choosing this possibility is the breadth of the phenomenon: as the forms in Table 1 and Table 2 show, the same similarity reoccurs across Romance languages. Moreover, if we look outside Romance languages, we find that same phenomenon holds. In Section 4 I will discuss some unrelated languages outside the Romance family. How can we, therefore, formalize this paradigmatic identity between the subjunctive and indicative (present or past), and the imperative and present indicative? In the next section, I will illustrate with an analysis of Romanian verbal morphology. This analysis suggests that the subjunctive and imperative agreement paradigms always only have the most widely distributed agreement affixes – the elsewhere affixes, which are not marked for Tense. I will advance the idea that the agreement affixes in subjunctive and imperative are the result of the verb lacking a certain specification for the Tense node. 1.2.

Romanian verbal agreement morphology This section presents data from the Romanian verbal system. This data is analyzed within the framework of Distributed Morphology and it is used to formalize the observation that subjunctive and imperative agreement paradigms are identical with the present indicative ones. Romanian verbs are classified in five classes determined by the infinitive endings. The infinitive is made of the particle a, and five different thematic vowels which attach to the radical.

5

I will ignore for the moment the differences in imperative forms in some languages but see the discussion in Section 4.

MIHAELA PIRVULESCU

(1)

I -a (a cânta) “to sing”

II -ea (a tãcea) “to keep quiet”

III -e (a merge) “to go”

233

IV V -i -î (a fugi) (a coborî) “to run away” “to descend”

Below I give the inflection of the Romanian verb a cânta, “to sing”: (2) Indic.

a cânta “to sing” Pers. Present Imperfect 1sg cânt cântam 2sg cânţi cântai 3sg cântã cânta 1pl cântãm cântam 2pl cantaţi cântaţi 3pl cântã cântau

Subjunctive6

Pers. 1sg 2sg 3sg 1sg 2sg 3sg

Present sã cânt sã cânţi sã cânte sã cântãm sã cântaţi sã cânte

Simple Past cântai cântaşi cântã cântarãm cântarãţi cântarã

Past Perfect cântasem cântaseşi cântase cântaserãm cântaserãţi cântaserã

IMPERATIVE Pers. Present 2sg

cântã !

2pl

cântaţi !

The verbal structure is the following (Iliescu & Mourin 1991, Chiţoran 1997): (3)

stem + cânt “(we) sing”

theme vowel +tense + -ã -Ø

person and number -m

The tense position is justified by past forms such as the following form from the past perfect : (4)

6

cânt -a-se-m “I had sung” sing-theme vowel-tense-person and number

Past subjunctive does not have agreement affixes; it is an invariable form composed from the subjunctive particle sã, the infinitive form of the auxiliary fi “be” and the past participle of the lexical verb: (i) sã fi cântat SA be sing-past participle

AGREEMENT PARADIGMS

234

Person and number are usually expressed by a single affix as in (5): (5)

noi cânt-ã-m we sing-theme vowel-1pl “we sing”

The agreement affixes and their distribution are the following (I follow the segmentation proposed in Iliescu and Mourin, 1991): (6)

-m -i -şi -Ø

-u -rã -ţi

1sg (imperfect, past perfect) 1pl everywhere 1sg simple past 2sg indicative present; subjunctive; imperfect 2sg simple past; past perfect 1sg indicative present, subjunctive 3sg everywhere 3pl everywhere except imperfect 3pl imperfect plural simple past, past perfect 2pl everywhere

The tense affixes are the following: (7)

-se-a

Past Perfect Imperfect7

I will use the values for person and number from Halle (1997:429): (8)

Person Participant Speech Event Author of Speech Event Number

1 2 + + + ± plural

3 -

Using the features above, the agreement affixes can be analyzed as following:

7

See Chiţoran 1997 for reasons to consider that imperfect has a temporal affix.

MIHAELA PIRVULESCU

(9)

235

Agreement affixes in Romanian -u [-PSE, -A, +pl] / ___ imperfect -şi [+PSE, -A, -pl] / ___ simple past, past perfect -i1 [+PSE, +A, -pl] / ___ simple past -i2 [+PSE, -A, -pl] -rã [+pl] / ____ simple past, past perfect8 -ţi [+PSE, -A, +pl] -m [+PSE, +A] -Ø elsewhere

The list in (9) shows competing Vocabulary Items (Halle and Marantz, 1993); this list is consulted disjunctively from top to bottom, in order to yield the form for any given combination of features. Vocabulary Insertion takes place after the syntax; therefore, Vocabulary Items do not have to be fully specified, since they do not supply features involved in syntactic operations. Accordingly, some affixes in the list in (9) are underspecified for the context of insertion (such as -i2) or for their intrinsic features (such as -m which is not specified for number). Underspecified or unspecified affixes encode syncretisms within paradigms: for example, the underspecified affix -m encodes the fact that this affix appears as an exponent for 1st person plural through the whole verbal paradigm and that it also appears as 1sg in indicative imperfect and past perfect. There are not two homophonous -m affixes, specified for the different contexts of insertion, rather the affix -m is treated as underspecified. The unspecified affix -Ø, serves the same purpose; it appears in 1st person singular and 3rd person singular and plural in indicative present, subjunctive, and imperative 2nd person singular. In addition to insertion of Vocabulary Items, other morphological operations are allowed to apply in Distributed Morphology (Impoverishment, Fission, Fusion, etc.). These operations are triggered by language-specific requirements. Such operations are posited when there are mismatches between the syntactic representation and the forms obtained. A rule of Impoverishment has to apply in the Romanian 1st person singular in indicative present in order to obtain the correct forms and to express the neutralization within the paradigm:

8

The affix -rã is considered in traditional grammars and in some formal analyses as being the exponent for the feature [plural] in simple past and past perfect (Irimia, 1994; Ştefãnescu 1997). I adopt here this idea, however this might not be the whole story since the 2nd person plural has -rã as well as -ţi, which in the list in (9) is also marked [+plural].

AGREEMENT PARADIGMS

236 (10)

Impoverishment rule [+PSE, +A]  Ø / [-pl, present indicative]

The operation in (10) prevents the more specific morpheme -m from being inserted in the 1st person singular and expresses the fact that the agreement affix for 1st person singular is syncretic, in the present, with the 3rd person (singular and plural). This rule correctly represents the retreat towards a less marked form, the 3rd person (Bobaljik 2002, Halle 1997). Given the analysis for agreement affixes above, as well as the operations for the Spell-out of syntactic nodes, I will now investigate the way the subjunctive and imperative forms are obtained. 1.3.

Morphological agreement affixes in the subjunctive and imperative How are the affixes distributed in subjunctive and imperative? From Tables 1 and 2 we see that the agreement affixes of subjunctive and imperative in Romanian are identical to the indicative present, as repeated below: (11)

Present indicative 1sg -Ø 2sg -i 3sg -Ø 1pl -m 2pl -ţi 3pl -Ø

Present subjunctive -Ø -i -Ø -m -ţi -Ø

Imperative -Ø

-ţi

First, let us examine the subjunctive paradigm. The majority of subjunctive agreement affixes are obtained by Vocabulary Insertion from the list in (9) as below: (12)

2sg subjunctive: [+PSE, -A, -pl] 3sg subjunctive: elsewhere 1pl subjunctive : [+PSE, +A] 2pl subjunctive: [+PSE, -A, +pl] 3pl subjunctive: elsewhere

 -i2  -Ø  -m  -ţi  -Ø

As the list in (12) shows, these affixes are the ones that do not contain any tense specification in their intrinsic or contextual features. They are in fact exactly the same affixes as those for indicative present. For the 1st person singular subjunctive, an Impoverishment rule needs to apply in order to obtain the correct result:

MIHAELA PIRVULESCU

(13)

1st singular subjunctive: Impoverishment: [+PSE, +A] Vocabulary Insertion: elsewhere

237

 Ø / [-pl, subjunctive]  -Ø

This Impoverishment rule is exactly the same as the one proposed for indicative present in (10), differing in only the contextual information [subjunctive]. Let us assume [subjunctive] to be a Mood feature. The crucial assumption I make is that [subjunctive] is not needed as a feature and therefore I propose to unify the two Impoverishment rules as in (14): (14)

1st person singular indicative present and subjunctive: Impoverishment: [+PSE, +A]  Ø / [-pl, present]

This rule contains no Mood information; as we will see in Section 3.2, the subjunctive syntactic structure justifies the decision to remove Mood from contextual information: in Romanian, the Mood node is not adjacent to the agreement node. Turning now to the imperative, in Romanian it has only two forms, 2nd person singular and plural. The form for 2nd plural is the same as for the other tenses/moods, therefore simply the result of Vocabulary Insertion: (15)

2nd plural imperative: [+PSE, -A, +pl]

 -ţi

However, 2nd person singular has a -Ø morpheme. I will assume that this is the elsewhere morpheme in the list in (9). Therefore, an Impoverishment rule needs to apply in order to obtain the correct form: (16)

2nd singular imperative: Impoverishment: [+PSE, -A] Vocabulary Insertion: elsewhere

 -Ø / [imperative]  -Ø

The Impoverishment rule in (16) above has to apply in order to prevent the affix for 2nd person singular, -i2, from being inserted. 2.

Analysis The observations that emerge from the data presented in the previous section are the following: a) The subjunctive and imperative agreement affixes are elsewhere affixes (with some or no feature specification). These are the affixes that are not restricted by contextual information pertaining to Tense; b) Some tense information is available for operations like Impoverishment, in the

238

AGREEMENT PARADIGMS

subjunctive: in this case, the tense feature is the same as in the indicative mood (‘present’). c) Mood information can appear in the case of the imperative.9 It seems therefore that tense information is not used in moods such as the subjunctive or imperative, unless it is exactly the same information as in another paradigm – the present indicative paradigm in this case. How can we explain this? In the next section, I link the agreement morphology in Moods such as subjunctive and imperative to the syntactic status of Tense. 2.1.

The status of the syntactic category Tense in subjunctive It is generally accepted that some verb forms can have an independent temporal interpretation whereas others have to acquire temporal interpretation from elements outside their clause, for example the difference between the indicative and subjunctive forms. In generative grammar, it is proposed that the difference in the temporal interpretation has to be syntactically represented by a difference in the status of the functional category T. Therefore, there is a difference between an independent Tense category as in the indicative mood and a dependent one as in the subjunctive mood (cf. among others, Progovac 1994 for ‘dependent Tense’; Picallo 1985 for ‘anaphoric Tense’, Tsoulas 1995 for ‘temporal indefiniteness’). The difference could be characterized in terms of features: for example, Motapanyane 1995 proposes specific tense features (present, past, etc.) for the independent Tense category and unspecified features for the anaphoric Tense; Tsoulas 1995 proposes a [-definite] feature in T for subjunctive. Now let us return to the Romanian subjunctive for an illustration. 2.2.

Syntactic Tense in subjunctive Consider the syntactic structure of a subjunctive clause (Motapanyane 1995; Dobrovie-Sorin 1994):

9

I assume here that [imperative] is a Mood feature. It is not crucial on which Functional Projection the [imperative] feature is located, only that it is not a Tense feature.

MIHAELA PIRVULESCU

(17)

239

MP

3

Spec

M’

3

M

AgrP Agr’

3

Agr

TP T’

T Ion sã cumpere John SA buy

3

VP [nişte mere] some apples

Once the syntactic operations have applied, the structure is as follows: (18)

M 3 M Agr 3 T Agr 3

sã

V T cumper-e10 -Ø

-Ø

According to Motapanyane 1995, the verb can only raise as far as the Agr head; Mood is expressed by the independent morpheme sã before the verb. Conditioning for allomorphy is subject to a strict locality condition on the syntactic structure after movement has taken place as in (19) (see Bobaljik 2000, Adger et al. 2003, among others): (19)

Y

X

V

2

2

Y

X

The branching nodes in (19) are labeled just as in syntax, i.e. the feature specifications are visible even after vocabulary items have been added. This permits inward sensitivity of node Y to node X in a cyclic root-out insertion of vocabulary items. In the structure in (19), node X can condition node Y if and only if, they are sisters.

10

A theme vowel appears, -e. There should probably be additional structure, such as a position for the theme vowel (Oltra-Massuet 1999).

240

AGREEMENT PARADIGMS

In (18) the only node that could provide contextual information for Vocabulary Insertion or for other morphological operations, such as Impoverishment, is the Tense node. As we saw in Section 2.3, the Tense feature value “present” does indeed serve as contextual information for Impoverishment in both the indicative present and subjunctive. Consider again the morphological Impoverishment rule proposed for 1st person singular in the present indicative and subjunctive: (20)

1st person singular present indicative and subjunctive: Impoverishment: [+PSE, +A]  Ø / [-pl, present]

The context of application for this rule contains agreement and tense information. If subjunctive Tense is ‘unspecified’ (according to its syntactic and semantic behavior), where does the feature ‘present’ come from? There appears to be a mismatch between the feature content of the Tense node as required by its syntactic behavior and semantic interpretation, and the feature content of the same node as required by morphological operations. I will maintain the idea that subjunctive is essentially a Mood where the verb is unspecified for Tense features. I tentatively propose a Feature Copying rule (Embick and Noyer, to appear). A feature introduced by such a rule is a ‘dissociated feature’ where ‘dissociated’ is “a term which emphasizes that such material is an indirect reflection of certain syntactic morphemes, features or configurations, and not the actual spell-out of these” (Embick and Noyer to appear: 15). This rule would apply for the purpose of Vocabulary Insertion and would copy the feature of the Tense in the main clause, [present] in this case, onto the Tense node in the lower clause. 2.3.

Syntactic Tense in imperative According to Zanuttini (1994, 1997), the essential property of the imperative is the absence of certain functional projections. I will follow her approach, and I will further propose that the morphological realization indicates the Tense projection is one of the missing functional projections (see also Pirvulescu and Roberge 1999). The morphological structure of an imperative after syntactic operations have taken place would be the following (adapted from Zanuttini 1994: 136):

MIHAELA PIRVULESCU

241

FP11

(21)

3

Agr

M

3

3

V cânt-a

M -Ø

FP

Agr -ţi

We saw that in order to obtain the correct form for the imperative 2nd person singular, an Impoverishment rule is needed: (22)

2nd singular imperative: Impoverishment: [+PSE, -A] Vocabulary Insertion: elsewhere

 -Ø / [imperative]  -Ø

This rules shows that an imperative feature is needed (it is called ‘imperative’ for convenience, to emphasize the fact that this rule is only needed in this Mood). In order for the Impoverishment rule to be sensitive to [imperative], this feature has to be on a node adjacent to the Agr node. As [imperative] is not a Tense feature, let’s assume it belongs on the Mood node (some models propose such a feature in CP, cf. Rivero 1994). In this case, because Tense is absent, the Mood node is necessarily adjacent to the Agreement node. 3.

Discussion Let us return to the generalization at the core of this article: that subjunctive (and imperative) agreement forms are identical to the indicative ones and that this identity follows from the status of syntactic Tense. Specifically, Tense is unspecified in subjunctives and absent in imperatives. We saw that in the Romanian subjunctive a morphological operation has to make appeal to a tense feature [present]. I proposed that this feature is not active in the syntax but copied for the purpose of morphological operations. To the extent that operations across languages will make appeal to a tense feature (present or past) in order to spell-out subjunctive agreement morphemes, a Feature Copying rule will need to be posited. It will therefore become more than a PF operation as defined in the introduction, i.e. triggered by a language-specific requirement and as such its status might have to be reconsidered. Under the idea that the Tense node in subjunctive is uniformly adjacent to the Agreement node across languages, it is this node that will condition 11

FP represent another functional projection (it could be a complementizer projection,a polarity one, etc.)

242

AGREEMENT PARADIGMS

allomorphy at the Vocabulary Insertion point.12 What we expect to find then, is that subjunctive agreement paradigms will always be identical with another agreement paradigm within the verbal system. In other words, subjunctive agreement paradigms a) will never show specific tense-conditioned morphological operations and b) will not have a specific tense affix. I would like to stress that, even if the Feature Copying rule I propose is a PF rule in the version of the Distributed Morphology framework I am working with, it is nevertheless a consequence of the status of the syntactic Tense in the subjunctive. The status of the syntactic Tense in the subjunctive as dependent or anaphoric (with the proposed features ‘unspecified’ or ‘-definite’, etc.), is presumably universal. Therefore, because of this fact, there will always be PF effects. However, the PF effects could vary within or across languages, according to the precise feature that will be copied onto the subjunctive Tense node (for example, the difference between subjunctive present and past in some languages). For imperatives, I proposed that there is no Tense projection in the syntax. Therefore, it is expected that a) morphological operations will never make appeal to a tense feature in this mood; b) no tense morpheme should appear in imperatives. However, Mood can condition morphological operations and therefore significant differences with respect to indicative are expected. One indication that this is so comes from the imperative in Spanish, which has a distinct ending in the plural (see Table 2). The generalization I propose could simply have a historical explanation. Therefore, it would be interesting to see whether it holds for unrelated languages outside Romance area. An indication that this is so comes from languages such as German, Russian and Turkish where subjunctive/imperative agreement paradigms seem to show, on the surface, the same identity with the indicative one. An indication that more generally verb forms with anaphoric/dependent Tense have no specific agreement paradigms comes from languages such as Portuguese and Hungarian: in these languages, inflected infinitives have an agreement paradigm identical with future subjunctive and personal possessive affixes from the pronominal paradigm respectively.13 Clearly, more research is needed.

12

Alternatively, we can propose that agreement and tense features are on the same syntactic node and that allomorphy of agreement features is obtained in a different manner than from structural relation among independent heads. 13 For reasons of space, I cannot provide examples here, but see Pirvulescu (2002).

MIHAELA PIRVULESCU

243

4.

Conclusion My aim in this article was to investigate an observation made by Carstairs (1987) regarding the fact that a morphosyntactic contrast such as Tense can be more important than a morphosyntactic contrast such as Agreement. I investigated the relation between the realization of agreement affixes and the realization of Tense across Tenses/Moods within the Romance area and found that there was a correlation between the two. Systematically, moods such as subjunctive and imperative do not show specific tense affixes or specific tense-induced allomorphy on their agreement affixes. Therefore, the paradigms associated with these moods show the same agreement affixes as those of another paradigm that is established on the basis of a Tense specification. I hope to have shown that there is a generalization to be stated and to be explained. The hypothesis I pursued was that the generalization follows from the syntactic structure. References Adger, David, Suzanna Béjar and Daniel Harbour. 2003. “Directionality of allomorphy: a reply to Carstairs-McCarthy”. Transactions of the philological society 101:1. 109-115. Bennett, Ch. 1900 A Latin Grammar. Toronto : G. N. Morgan. Bobaljik, Jonathan. 2000. “The ins and outs of contextual allomorphy”. University of Maryland Working Papers in Linguistics 10, ed. by K. K. Grohmann and C. Srujke. 35-71. Bobaljik, Jonathan. 2002. “Syncretism without paradigms: Remarks on Williams 1981”. Yearbook of Morphology 2001, 35-71. Carstairs, Andrew. 1987. Allomorphy in Inflection. Beckenham: Croom Helm. Chiţoran, Ioana. 1997. The phonology and morphology of Romanian glides and diphtongs: a constraint-based approach. Doctorate dissertation, Cornell University. Chomsky, Noam. 2000 “Minimalist Inquiries: the Framework”. Step by step, ed. by Roger Martin, David Michaels & Juan Uriagereka, 5-67. Cambridge, Mass: MIT Press. De Bruyne, Jacques. 1995. A comprehensive Spanish grammar. Oxford : Blackwell. Dobrovie-Sorin, Carmen. 1994. The syntax of Romanian. Berlin : Mouton de Gruyter. Embick, David and Rolf Noyer. “Distributed morphology and the Syntax/Morphology interface”. To appear in The Oxford Handbook of Linguistic Iinterfaces, ed. by George Ramchand and Charles Reiss. Oxford: Oxford University Press.

244

AGREEMENT PARADIGMS

Halle, Morris. 1997. “Distributed Morphology: impoverishment and fission”. Papers at the interface, MIT Working Papers in Linguistics 30, ed. by B. Bruening, Y. Kang & M. McGinnis, 425-449. Cambridge, Mass. : MIT. ---------- and Alec Marantz. 1993. “Distributed Morphology and the pieces of inflection”. The view from building 20, ed. by Kenneth Hale and Samuel Jay Keyser, 35-67. Cambridge, Mass. : MIT Press. Hills, Elijah. C. and Jeremiah. D. M. Ford. 1944. Portuguese Grammar. Boston: D. C. Heath and Company. Iliescu, Maria and Louis Mourin. 1991. Typologie de la morphologie verbale romane. Innsbruck : Amoe. Irimia Dumitru. 1994. Morfo-sintaxa verbului românesc. Iasi : Ed. Universitãtii. Le Goffic, Pierre. 1997. Les formes conjuguées du verbe français. Paris: Ophrys. Motapanyane, Virginia. 1995. Theoretical Implications of Complementation in Romanian. Unipress: Padova. Oltra-Massuet, Isabel. 1999. “On the notion of theme vowel”. MIT Working Papers in Linguistics. Cambridge, Mass.: MIT Picallo, Carme. 1985. Opaque domains. Doctoral dissertation, CUNY. Pirvulescu, Mihaela and Yves Roberge. 1999. “Objects and the Structure of Imperatives”. Formal Perspectives on Romance Linguistics, ed. by JeanMarc Authier, Barbara Bullock & Lisa Reed, 211-226. Amsterdam : John Benjamins. ----------. 2002. Le concept de paradigme et la morphologie verbale. Doctoral dissertation, University of Toronto. Progovac, Liliana. 1994. Negative and positive polarity: a binding approach. Cambridge UK: Cambridge University press. Ragusa, Olga. 1984. Italian verbs – regular and irregular. New York : S. F. Vanni. Rivero, Maria-Luiza. 1994. “Clause structure and V-movement in the languages of Balkans”. Natural Language and Linguistic Theory 12. 63120. Ştefãnescu, Ioana. 1997. The syntax of agreement in Romanian. Doctoral dissertation, CUNY. Tsoulas, George. 1995. “Indefinite clauses: some notes on the syntax and semantics of subjunctives and infinitives". Proceedings of WCCFL 13, ed. by Raúl Aranovich, William Byrne, Susanne Preuss & Martha Senturia, 515-530. Stanford, Calif.: CSLI Publications.

MIHAELA PIRVULESCU

245

Zanuttini, Raffaella. 1994. “Speculations on negative imperatives”. Rivista di Linguistica 6.1. 67-89. ----------. 1997. Negation and clausal structure, New York : Oxford University Press.

ITALIAN VOLERCI: LEXICAL VERB OR FUNCTIONAL HEAD?

CINZIA RUSSI The University of Texas at Austin

0.

Introduction Italian shows a number of verbs characterized by the presence of the clitic ci, which, however, has completely lost its pronominal value and has become, through a process of grammaticalization, an obligatory element of the verb (a morphological marker). Among the most common verbs affected by this phenomenon of ci incorporation we find essere “be” (> esserci “be, existential”) and avere “have, auxiliary” (> averci “have, possessive”); verbs of perception like sentire “hear” (> sentirci “be able to hear”) and vedere “see” (> vederci “be able to see”), mettere “put” (> metterci “take/be necessary”), and volere “want” (> volerci “take/be necessary”). Thus, as shown in (1)-(4), verb pairs can be found in Italian, which differ in the presence/absence of ci. (1) a. Ci vogliono/*Vogliono nove giorni per farsi installare in casa il telefono dalla Telecom “It takes nine days to have the phone installed in your house by Telecom.” (from CORIS/CORDIS1) b. Vogliono/*Ci vogliono partire entro mezzogiorno “They want to leave by noon.” (2) a. Non ci sento/*sento “I can’t hear.” b. Sento /*Ci sento un rumore “I hear a noise.” 1

CORIS/CODIS is a corpus of written Italian under development at CILTA “Centre for Theoretical and Applied Linguistics” (University of Bologna) that consists of a collection of authentic and commonly occurring texts in electronic format chosen by virtue of their representativeness of modern Italian. Presently, the corpus contains 100 million words. http://www.cilta.unibo.it/Portale/RicercaLinguistica/coris_eng.htmL

248

CINZIA RUSSI

(3) a. Carlo è simpatico/* c’è simpatico “Carlo is nice.” b. Alla festa c’era /* era Carlo “There was Carlo at the party.” (4) a. Dove metti/*ci metti le chiavi? “Where do you put your keys?” b. Quanto ci metti/*metti ad arrivare in ufficio? “How long does it take you to arrive at the office?” The grammaticalization of lexical verbs into (modal) auxiliaries and functional elements in general is a widespread phenomenon (see, among others, Bybee et al. 1994). A typical aspect of the process is a morphosyntactic impoverishment of the verb, which goes hand in hand with modifications in the verb’s argument structure. In two fairly recent articles Benincà & Poletto (1994, 1997; henceforth B&P) offered a detailed analysis of the evolution of the Italian verb bisogna “must/be necessary” (B&P’s gloss) from full transitive (i.e., lexical) verb in Old Italian (OI) to deontic modal in Modern Italian (MI), and reconstruct an analogous process for volere “want” and toccare “touch” of some Italian dialects. B&P show how these verbs developed a number of morphosyntactic restrictions (i.e., loss of forms in the verbal paradigm), accompanied by a significant alteration of the thematic grid of the verb that resulted in loss of argument structure. That is, bisogna underwent a grammaticalization process and changed from lexical verb to functional head. In a first attempt at examining the structural properties of volerci “take/be necessary” I proposed (Russi 2003a) that ci incorporation leads to substantial changes in the argument structure of volere “want”. Based on significant parallels between volerci and Verbs of Inherently Directed Motion (VIDMs, Tortora 1996, 1999), I argued that Tortora’s (1999) ‘implicit argument hypothesis’ can be extended to volerci and proposed to analyze volerci as a two-argument unaccusative. This proposal received strong criticism (by two anonymous reviewers), on the grounds that volerci is more properly analyzed as a modal head similar to Modern Italian bisogna, both in terms of synchronic behavior and diachronic development. My main objective in this paper is to determine whether Italian volerci can indeed be considered a functional element (modal head). More specifically, the paper investigates whether the grammaticalization process reconstructed by B&P for bisogna can be posited for volerci as well. A detailed comparison of volerci and MI bisogna reveals that the two verbs do

ITALIAN VOLERCI

249

not display analogous morphosyntactic restrictions. On the other hand, a striking parallel emerges between volerci and OI bisogna. Based on the morphosyntactic similarities between OI bisogna and volerci, it can then be concluded that volerci is still a lexical verb: ci incorporation may have altered the thematic grid of volere but the resulting verb volerci still projects arguments. The paper is organized as follows: In Section 2 I summarize the morphosyntactic properties of OI and MI bisogna as discussed in B&P, which I then compare to the structural properties of volerci in Section 3. Section 4 summarizes the main points of the paper and briefly addresses the question of whether we can actually claim that volerci (or volere) underwent grammaticalization. Given that the structural properties of volerci appear to have remained unaltered since the oldest attestation I was able to locate (mid 18th century), I propose that we cannot posit a process of grammaticalization that affected volere or volerci. The only element that actually grammaticalized is the clitic ci, which from anaphoric element developed into an (empty?) morphological marker. 1. 1.1.

Italian bisogna: from lexical verb to functional head Bisogna in MI A crucial assumption of B&P’s analysis is that the thematic grid of a verb affects the verb at both the morphological and syntactic level, in the sense that impoverishment or loss of a verb’s θ-grid (grammaticalization) brings about major changes in the verb’s syntax and morphology. Evidence for this a claim is provided by the synchronic and diachronic analysis of Italian bisogna ‘must/is necessary’ and some modals of necessity found in Northern Italian dialects. According to B&P (1997:94-95) bisogna “has the ‘most reduced meaning’ among Italian verbs of necessity … [it] means a pure state of necessity; its meaning does not involve any cause of the necessity itself nor a particular person or object individually concerned with it.” As shown in (5), bisogna can take either a bare infinitive (5a) or a subjunctive clause (5b), which denote what is necessary. (5) a. Bisogna studiare bisogna-P(RESENT) I(NDICATIVE)-3SG study-INF “It is necessary to study” b. Bisogna che tu studi bisogna-PI-3SG that you study-P(RESENT) S(SUBJUNCTIVE)-2SG “It is necessary that you study.”

CINZIA RUSSI

250

Compared to other Italian deontic modals, bisogna is characterized by a number of morphosyntactic restrictions, which are given in (6). (6) Special morphosyntactic properties of MI bisogna (adapted from B&P 1994:36; 1997:98) a. It can only be inflected in the 3rd person singular in tense/mood forms that: (1) are not marked for aspect, (2) are compatible with unspecified time localization, and (3) can express modality. b. It lacks a subject position. c. It is not a raising verb. d. It cannot host clitic pronouns. The person and tense/mood restrictions of MI bisogna are illustrated in (7) and (8) respectively. (7) * Bisogno/ bisogni / bisognamo/ bisognate/ bisognano leggere Bisogna- PI-1SG/ 2SG/ 1PL/ 2PL/ 3PL read-INF quest’articolo this article (8) a. * Bisognando leggere quest’articolo, lo leggeremo bisogna-GER read-INF this article it read-1PL-FUT “Being necessary to read this article, we will read it.” b. * È/ Ha bisognato leggere be-PI-3SG / have-PI-3SG bisogna-P(AST) P(ARTICIPLE)-3SG read-INF quest’ articolo this article “It was necessary to read this article.” c. Bisognerebbe leggere quest’articolo bisogna-P(RESENT) C(ONDITIONAL)-3SG read-INF this article “It would be necessary to read this article.” Evidence that bisogna lacks a subject position comes from grammaticality contrasts among sentences in which the PRO subject of the infinitive is controlled by the subject of the governing sentence. As shown in (9), both argumental subjects (9a) and quasi-argumental subjects (9b) in a matrix clause are able to control the PRO subject of the infinitival clause: (9) a.

Andrò a Roma senza vedere il papa “I will go to Rome without seeing the pope.”

ITALIAN VOLERCI

b.

251

Nevica senza necessariamente fare molto freddo “It snows without being necessarily very cold.” (B&P 1997:96)

Expletive subjects of impersonal verbs, on the other hand, can control PRO only marginally: (10)

?Sembra che si tratti di un delitto senza essere chiaro chi sia il colpevole “It seems to be a murder without being clear who the culprit is.” (B&P 1997:97) In contrast, the subject of bisogna cannot give PRO any content:

(11) a. * Bisogna che lo leggiamo senza esser necessario che lo facciamo subito “It is necessary that we read it without being necessary that we do it at once.” b. * Bisogna leggerlo senza esser necessario farlo subito “It is necessary to read it without being necessary that we do it at once.” (B&P 1997:97) Additional evidence for the claim that bisogna lacks a subject position comes from the following contrast between bisogna and sembrare “seem”, which shows that (unlike sembrare) bisogna is not a raising verb. (12) a. * Mario bisogna leggere “Mario needs to read.” b. * Bisogna partire Mario “Mario needs to leave.” (13) a. Mario sembra leggere “Mario seems to read.” b. Sembra conoscerlo Mario “Mario seems to know him.” (B&P 1997:97-8)

CINZIA RUSSI

252

Finally, the examples in (14), to be contrasted with (15), show that bisogna can host neither clitic pronouns thematically related to it (14a) nor clitics related to an embedded predicate via restructuring (14b). (14) a. * Gli bisogna leggere quest’articolo to.him bisogna-PI-3SG read-INF this article “He needs to read this article.” b. * Lo bisogna leggere it bisogna-PI-3SG read-INF “It is necessary to read it.” (15) a.

b.

Gli è necessario leggere quest’articolo to.him be-PI-3SG necessary read-INF this article “To him it’s necessary to read this article.” Lo sembra leggere con interesse it seem-PI-3SG read-INF with interest “S/he seems to read it with interest.”

Summarizing, based on the aforementioned morphosyntactic restrictions, B&P conclude that MI bisogna lacks in argument structure; i.e., it is characterized by the structural properties typical of a functional head. An anonymous reviewer brought up an interesting point, namely that the properties in (6) are not incompatible with the analysis of bisogna as a regular (i.e. lexical) verb that subcategorizes for a clausal complement but lacks a subject theta role. According to him/her, the inflectional restrictions (6a) are accounted for in terms of 3SG default agreement, due to the fact that the only argument of bisogna is a clause. In contrast, volerci takes a nominative object NP and therefore it agrees with that. The lack of a subject position (6b) is explained by the following: Italian is a pro-drop language, the equivalent of English expletive it is null, and ci (i.e. the equivalent of there) is ruled out with clausal arguments. As for not being a raising verb (6c), this property is verb specific and cannot be related to the absence of a subject theta role (cf., for instance, It was hoped that John would leave vs. *John was hoped to leave). Finally, the impossibility of hosting clitic pronouns (6d) results from the lack of arguments that would give such a pronoun. We could then conclude that the analysis of MI bisogna as a functional head might need to be reconsidered. Although this is certainly an issue worth pursuing, it must be put aside at the moment because it is not within the scope of this paper.

ITALIAN VOLERCI

253

1.2.

Bisogna in OI The morphosyntactic restrictions discussed so far did not characterize OI bisogna. as illustrated in (16), in OI bisogna could assign a thematic role to both an experiencer, which would be realized in the dative, and a theme, which would surface in the nominative (16a) or as the pronoun ne, which is the clitic required for pronominalization of part of a quantified theme (16b), and function as the grammatical subject of the verb (i.e., it triggered verbal agreement). Moreover, OI bisogna could host clitics expressing one of its arguments and could be inflected for tenses/moods that are ungrammatical in MI (16c vs. 8a). (16) a. Mi bisognano fiorini dugento d’ oro 1SG IO bisogna-PI-3PL florins two.hundred of gold “I need two hundred gold florins.” (B&P 1997:109) b. ... coloro che ne bisognano ... those who of-it bisogna-PI-3PL “... those who need it ...” (B&P 1997:112) c. … bisognando- gli una grande quantità di denari bisogna-GER-3SG IO a large amount of money “… him needing a large amount of money…” (B&P 1997:109) Based on the structural differences discussed so far, B&P conclude that bisogna changed from a canonical transitive verb into a pure modal head (functional element). That is, bisogna underwent a grammaticalization process, which considerably affected (in fact, eliminated) its θ-grid. Therefore, it is to be analyzed as a functional element directly inserted into the head of a functional projection corresponding to its semantics, namely the head of RootMood, following Cinque (1995) (see also Cinque 1999). 2.

Volerci as a modal of necessity? Example (4a), repeated here as (17) represents a typical construction with volerci. (17) Ci vogliono/*Vogliono nove giorni per farsi installare in casa il telefono dalla Telecom “It takes nine days to have the phone installed in your house by Telecom.”

254

CINZIA RUSSI

The construction illustrated in (17) comprises the NP nove giorni “nine days” which, based on verb agreement facts, can be identified as the grammatical subject of the verb, plus an infinitival clause introduced by per ‘for’, whose categorial status will not be discussed in this paper. The subject NP represents the entity needed to bring about the event expressed by the infinitival clause. Most commonly the NP is a temporal expression, as in (17) or an inanimate (material or abstract) entity (18a, b) but it can also be animate (18c). (18) a. Credo che per quelli all’ultima fila ci voglia addirittura il binocolo per $2 “I think that for those in the last rows binoculars might be necessary to…” (from LIP3) b. … ma a me colpisce una cosa che eh ci sia voluta una Guerra per parlare del tema della giustizia… “…but one thing strikes me, that it took a war to speak about justice…” (from LIP) c. … per fare grande la Fiorentina ci vuole anche Dunga… “…to make Fiorentina big, it takes also Dunga…” (from LIP) Comparing volerci and MI bisogna we notice that they differ crucially with respect to the possibility of taking an NP subject: volerci allows it whereas bisogna does not (because it lacks a subject position; cf. (6b)). (19) a. …mentre ci vuole moltissimo impegno per cambiare… “…while it takes very much effort to change…” (from CORIS/CODIS) b. * Bisogna moltissimo impegno per cambiare “Very much effort is necessary to change.” In addition, volerci and bisogna differ considerably with regard to the various morphosyntactic restrictions given in (6). More precisely, volerci can occur in tenses and moods that are rejected by bisogna. (20) a. Volendoci del pane, Carlo è andato a comprarlo “Since bread was needed Carlo went to buy it.” 2

$ indicates an unintelligible word. Lessico di frequenza dell’italiano parlato De Mauro et al. (1993). http://languageserver.unigraz.at/badip/badip/home.php 3

ITALIAN VOLERCI

255

b. * Bisognando comprare del pane, Carlo è andato in panetteria “Since it was necessary to buy bread, Carlo went to the bakery.” (21) a. Ci sono voluti quattro anni per creare “Standing Stone” “It took three hours to create Standing Stone.” (from CORIS/CODIS) b. * È/Ha bisognato leggere quest’articolo “It was necessary to read this article.” Furthermore, unlike bisogna, volerci shows no inflectional restrictions in person and number (cf. (7)).4 (22) Ci voglio io/ Ci vuoi tu/ Ci vogliamo noi/ Ci volete voi per... “It takes me/ it takes you SG / it takes us / it takes you PL to…” In this respect, volerci resembles OI bisogna, which could be inflected for both 3SG and 3PL (23a, b), and is also found in 2SG (23c). (23) a. … a me bisogna la vostra fede … to me bisogna-PI-3SG the your PL faith “…to me your faith is necessary…” (Decameron, IV 8)

4

An anonymous reviewer remarked that “[a]n interesting, though obscure, descriptive detail concerns (22); the sentences are not perfectly grammatical; they become perfect with an imperfect or conditional tense/mode.” I have to say that such detail does not characterize my speech. A search of LIP and CORIS/CODIS for volerci + pronoun construction in present indicative, imperfect indicative, passato prossimo and present conditional revealed the following four examples, all from CORIS/CODIS. (i) a. Ci voglio io per ridargli una motivazione (PRESENT INDICATIVE) volerci-PI-1SG I for give.back-INF-3 SG IO a motivation “It takes me to give in back some motivation.” b. Sono robusti, ci vogliono loro (PRESENT INDICATIVE) be-PI-3PL strong volerci-PI-3PL they “They are strong, it takes them.” c. Ci volevi tu, eh, Grassone! (IMPERFECT INDICATIVE) volerci-II-2SG you fat.AUGM “It took you, eh, Fatso.” d. Ci voleva lui, Silvio Berlusconi in persona (IMPERFECT INDICATIVE) volerci-II-3SG he Silvio Berlusconi in person “It took him, Silvio Berlusconi in person.”

CINZIA RUSSI

256

b. … troppe gran forze bisognano … too.many big forces bisogna-PI-3PL “…too many big forces are necessary…” (Decameron, VI Introduction) c. … tu ci bisognavi per dir orazioni … you 1PL IO bisogna-I(MPERFECT) I-2SG for say-INF prayers “…you were necessary to us to say prayers…” (Decameron, VII 3) In fact, as we can notice by comparing (24) and (25) the parallel that obtains between volerci and OI bisogna becomes even more remarkable in view of the following: (i) both verbs can select an experiencer, which is realized in the dative, and a theme, which is realized in the nominative or as ne and functions as grammatical subject, and (ii) both can host clitics thematically related to them. (24) a. Gli ci sono voluti otto mesi per rimettersi 3SG IO volerci-PPERF-3SG eight months for recover-INF dall’ incidente from.the accident “It took him eight months to recover from the accident.” b. Gli ce ne sono voluti otto “It took them eight (months).” (25) … si

diede al bere, e benché non ne give-PAST-3SG at.the drink-INF and although NEG of.it gli bisognasse 3SG IO bisogna-ISUBJ-3SG “… he took to drinking, although he didn’t need it…” (Decameron, VIII 6) RFLX

To conclude, volerci and MI bisogna exhibit no common morphosyntactic properties, whereas the structural similarity between volerci and OI bisogna is quite substantial. 3.

Summary and concluding remarks The fact that no structural parallel obtains between volerci and MI bisogna indicates that volerci cannot be considered a modal head (functional element). Rather, volerci appears to be a verb of the same type as OI bisogna, that is, a regular transitive verb according to B&P’s analysis. The process of ci

ITALIAN VOLERCI

257

incorporation has, to some extent, altered the thematic-grid of volere “want” but the resulting verb volerci can still project arguments.5 My previous analysis of volerci (Russi 2003a), then, may seem to be inaccurate, given that volerci appears to be a simple ‘transitive’ verb. Notice, however, that volerci, exactly like OI bisogna, differs from canonical transitive verbs (e.g., mangiare “eat”) in two quite relevant aspects: (i) auxiliary selection, i.e., canonical transitive verbs take avere “have” (26a), whereas volerci takes essere “be”, which is the auxiliary selected by unaccusatives (26b); and (ii) surface realization and thematic roles of the arguments, i.e., transitive verbs select a nominative subject (agent/experiencer) and an accusative object (patient/theme) (27), while volerci selects a dative object (experiencer) and a nominative /ne subject (theme) (28). (26) a. Ha mangiato un panino have-PI-3SG eat-PP a sandwich “S/he ate a sandwich.” b. Ci sono volute sei uova be- PI-3SG volerci-PP six eggs “It took six eggs.” (27) a. Lui (NOM.) /*Gli (DAT.) ha bevuto un caffè “Carlo drank a cup of coffee.” b. Lui lo(ACC.) ha bevuto “He drank it.” (28) a. * Lui/Gli ci vorranno almeno due ore per prepararsi “It is going to take him at least two hours to get ready.” b. Gli ce ne/*le (ACC.) vorranno almeno due “It will take him at least two.” 5

An exhaustive discussion of the structural differences between volere and volerci cannot be included here due to space constraints. I will just mention that volere typically takes a subject (nominative) NP whose referent is [+ human] and either an (accusative) NP object (ia) or a clausal object (ib, c). (i) a. (Loro) vogliono un gelato “They want an ice-cream.” b. Vogliono uscire “They want to go out.” c. Vogliono che tu esca “They want you to go out.” See §3 above for the selectional properties of volerci.

258

CINZIA RUSSI

These differences suggest that categorizing volerci as a standard transitive verb may be inappropriate; rather, they indicate that it simply is a verb with a thematic structure, and the same is to be concluded for Old Italian bisogna. In any case, however, the suggestion that volerci is better analyzed as a modal head cannot be considered adequate. In view of the impossibility to analyze volerci as a functional element the (crucial) question arises of whether it is actually possible to claim that volerci has undergone a process of grammaticalization. In other words, is there empirical evidence that any of the morphosyntactic changes typical of grammaticalization have affected volerci? A preliminary screening of Italian texts from different periods shows that, unlike bisogna, volerci did not undergo any structural modification.6 As the examples in (29) indicate, the structural properties of volerci have remained unchanged since is the oldest attestation I found (29a). (29) a. Si vede che né epicicli, né eccentrici ci voleno a farli alzare e retrocedere. “It means that it takes neither epicycles nor eccentrics to make them rise and regress.” (La città del sole; Tommaso Campanella 1602, my translation CR) b. Avanti ch’io superi l’avversion per le donne, ci vuol altro “Before I can overcome my adversion for women, it takes more.” (La locandiera, Act I, scene 1; Carlo Goldoni 1752, my translation CR). c. Credimi: ci vuole una stupida bassezza o una sovrumana filosofia per lasciarsi a beneplacito d’un nemico che ha faccia impudente, anima negra, e mano tremante “Believe me: it takes either a foolish lowness or a superhuman philosophy to give oneself to an enemy who has an impudent face, a black soul, and a shaky hand.” (Jacopo Ortis; Ugo Foscolo 1802, my translation CR) d. Ci voleva poco a capire “It took little to understand.” (Cuore, Novembre Lo spazzacamino 1; Edmondo De Amicis 1886, my translation CR) 6

The texts I screened are the following: La vita nuova (Dante Alighieri ca. 1292); Divina Commedia, Inferno (Dante Alighieri ca. 1306-21); Decameron (Giovanni Boccaccio 1349-51); Orlando furioso (Ludovico Ariosto 1532), La città del sole (Tommaso Campanella 1602); La locandiera (Carlo Goldoni 1752); Gli innamorati (Carlo Goldoni 1759); Le ultime lettere di Jacopo Ortis (Ugo Foscolo 1802); Cuore (Edmondo de Amicis 1886); Canne al vento (Grazia Deledda 1913). http://www.crs4.it/HTML/Literature.html

ITALIAN VOLERCI

259

e. Quanto ci vuole? Non so dirtelo quanto ci vuole in bicicletta. Poche ore “How long does it take? I can’t tell you how long it takes by bike. Few hours” (Canne al vento, 3; Grazia Deledda 1913, my translation CR) The point is that the only element that indeed grammaticalized is ci. It is ci that underwent semantic and morphosyntactic alteration, changing from pronoun (anaphorical element) to morphological (lexical?) marker (for more details on the grammaticalization of ci see Sala-Gallini 1996, Russi 2003b). Ci incorporation to volere, then, may have affected the morphosyntax of volere (see fn.4) but a process of grammaticalization that affected volere or volerci per se cannot be posited. In other words, a diachronic trajectory of the type ‘from lexical to functional’, which represents the only possible instantiation of grammaticalization within the generative approach (as in Roberts and Roussou 2003), does not pertain to the emergence of volerci from volere, since both the source and the target are lexical verbs. Rather, two concomitant processes have taken place in this case, namely grammaticalization of ci and lexicalization of volerci, in a scenario which appears to support Lehmann’s (2002: 7) claim that it is not a linguistic unit per se that undergoes grammaticalization or lexicalization: “it is the construction of which the element is a constituent that embarks on either course.” Grammaticalization and lexicalization are thus to be seen as orthogonal processes, which can apply alternatively to complex units. The crucial difference between the two is that “[g]rammaticalization involves an analytic access to a unit, lexicalization involves a holistic access to a unit, a renunciation of its internal analysis” (Lehmann 2002:13). I would like to conclude by drawing attention to an appealing hypothesis about a possible grammaticalization scenario for volerci which has been suggested by an anonymous reviewer and definitely deserves further researching. As we see in (30), OI bisogna had a second possible surface realization of its two arguments, which volerci lacks: the experiencer could be nominative and the theme in objective case (or pronominalized as ne; see also (16b). (30) Quasi tutte le giovani si fanno più belle in viso almost all the young.girls REFLX make more beautiful in face e non bisognano d’ altri ornamenti and NEG bisogna-PI-3PL of other ornaments “Almost all the young women make their faces more beautiful and do not need any other ornaments.” (Ugo Foscolo, Odi IV 342, in B&P 1997:112)

260

CINZIA RUSSI

This second realization resembles that of a regular transitive verb, apart from the fact that nominative case is assigned to a non-agentive role, while in the case of a canonical (prototypical) transitive verb it would be assigned to an agent. Transitive volere could then be viewed as a lexical variant of volerci that selects the second type of case assignment just mentioned (i.e., NOM  EXPER and ACC  THEME; see fn. 5). Also, the two possible argument realizations of OI bisogna may be viewed as two coexisting stages of grammaticalization and the case of volerci can possibly suggest that the second realization (i.e. nominative/experiencer and objective/theme) is older than the realization dative/experiencer and nominative/theme. Within this scenario, volerci can in fact be related to a grammaticalization process of transitive volere through the intervention of the locative clitic. The process of grammaticalization that affects volere/volerci though differs from that of bisogna in a quite important aspect: in the latter, no semantic alteration takes place (i.e., the lexical meaning of bisogna remains unchanged), whereas in the former we witness a semantic shift from volition to necessity. References Benincà, Paola & Cecilia Poletto. 1994. “Bisogna and Its Companions: The Verbs of Necessity”. Paths Towards Universal Grammar. Studies in Honor of Richard S. Kayne ed. by. Guglielmo Cinque et al., 35-57. Washington, D.C.: Georgetown University Press. ----------. 1997. “The Diachronic Development of a Modal Verb of Necessity”. Parameters of Morphosyntactic Change, ed. by Ans van Kemenade & Nigel Vincent, 94-118. Cambridge: Cambridge University Press. Bybee, Joan, Revere Perkins & William Pagliuca. 1994. The Evolution of Grammar. Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press. Cinque, Guglielmo. 1995. “Adverbs and the Universal Hierarchy of Functional Projections”. Glow Newsletter 14-15. ----------. 1999. Adverbs and Functional Heads. A Cross Linguistic Perspective. New York: Oxford University Press. Lehmann, Christian. 2002. “New Reflections on Grammaticalization and Lexicalization”. New Reflections on Grammaticalization ed. by Ilse Wischer and Gabriele Diewald, 1-18. Amsterdam & Philadelphia: John Benjamins. Roberts, Ian & Anna Roussou. 2003. Syntactic Change. A Minimalist Approach to Grammaticalization. Cambridge: Cambridge University Press. Russi, Cinzia. 2003a. “On the Argument Structure of Italian Volerci”. Paper presented at the 33th Linguistic Symposium on Romance Languages, Bloomingtoon, April 2003.

ITALIAN VOLERCI

261

----------. 2003b. The Grammaticalization of Italian Clitics. Doctoral dissertation, University of Washington. Sala-Gallini, Mario. 1996. “Lo statuto del clitico nella dislocazione a destra: pronome vero o marca flessionale?” Archivio glottologico italiano 81.7694. Tortora, Christina. 1996. “Two Types of Unaccusatives: Evidence from a Northern Italian Dialect”. Grammatical Theory and Romance Languages. Selected Papers from the 25th Linguistic Symposium on Romance Languages, ed. by Karen Zagona, 251-262. Amsterdam & Philadelphia: John Benjamins. ----------. 1999. “The Post-verbal Subject Position of Italian Unaccusative Verbs of Inherently Directed Motion”. Semantic Issues in Romance Syntax, ed. by Estelha Treviño & José Lema, 283-98. Amsterdam & Philadelphia: John Benjamins.

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES ALMEIDA JACQUELINE TORIBIO AND CARLOS NYE The Pennsylvania State University

0.

Contact and contracting Spanish Extant research on the linguistic behaviors of Spanish heritage speakers in the U.S. unveils evidence of convergence and indeterminacy in their Spanish language performance. Specifically, their performance exhibits non-target-like lexical semantic properties and selection of discursively-conditioned options (cf. Montrul 2002, 2005, Silva-Corvalán 1994/2000). For example, heritage speakers may demonstrate an insensitivity to mapping irregularities of unaccusative predicates (Zapata et al. 2005) and a loss of the discursive constraints that govern the felicitous usage of null and overt personal pronominal subjects (Toribio et al. 2005). However, it merits pointing out that notwithstanding their target-deviant performance, heritage bilinguals demonstrate speech forms with properties diagnostic of the full variety spoken by monolinguals as regards core or ‘narrow’ syntactic phenomena, among these, the functional projections and features of INFL that control verb raising and license null subjects and object clitics (cf. Sorace 2000, 2004, Tsimpli et al. 2004, de Prada et al. 2005). 1.

Target of investigation: Psychological predicates In the present study, we scrutinize the alterations manifest in the grammars of Spanish heritage speakers by reference to their production and interpretation of constructions with psychological predicates exemplified by gustar. The gustar class of psychological predicates differs from other classes of Experiencer-taking verbs such as gozar de and amar, exemplified in (1). (1) a. El goza de buena salud he gozar-de3SG good health “He enjoys good health.”

264

A. JACQUELINE TORIBIO & CARLOS NYE

b. El ama he amar-3SG “He loves his wife.”

a ACC

su esposa his wife

In the above ‘direct’ psychological predicates, the Experiencer occupies the structural subject position, receiving Nominative case, while the Theme occupies the structural object position, and each demonstrates the respective ensuing properties. This ‘direct’ thematic mapping parallels that of the English verbs enjoy, love, and like. However, the Spanish equivalent of like, gustar, exemplified in (2), demonstrates quite disparate behavior—the Theme appears in Nominative case and controls agreement, and the Experiencer is marked with Dative case; for this reason, gustar is often called a ‘reverse’ psychological predicate or Dative-Experiencer verb, whose mapping mirrors that of to be pleasing to (cf. Martín 1998). (2) a. (A

los estudiantes) les gusta el libro the students CL-DAT-3PL gustar-3SG the book-NOM “The students/They like the book.” b. El libro les gusta (a los estudiantes) the book-NOM CL-DAT-3PL gustar-SG DAT the students “The students/They like the book.” DAT

The full range of properties that typify the gustar class of psychological predicates in the monolingual variety of Spanish are described in (3): (3) a. The Theme or the Experiencer may precede the verb; b. The Theme is marked with structural NOM case and the Experiencer is marked with lexical DAT; c. Agreement on the verb is controlled by the Theme (3SG/PL); d. An obligatory clitic (DAT-3SG/PL) ‘doubles’ the Experiencer. As outlined, there is a dispersion of ‘subject’ properties in reverse psychological predicates. In particular, the properties of NOM case and control of agreement that characterize a canonical subject reside with the Theme. However, the Experiencer can be shown to demonstrate additional behavior of structural subjects. More significantly, as an animate argument that dominates the Theme in the thematic hierarchy of the argument structure (Grimshaw 1990), the Experiencer is understood as the semantic, non-Agentive subject. The properties of the Spanish indirect psychological predicate gustar relative to the English direct psychological predicate like render this class of verbs especially difficult and confusing for English speakers acquiring Spanish

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

265

as a second language (cf. Montrul 1997, 1998, Toth 2003). These reverse psychological predicates are also vulnerable in contact and contracting varieties of Spanish. Silva-Corvalán (1994/2000), in her longitudinal observations of Mexican-Americans in Los Angeles, has remarked on the emergence of non-target agreement and case marking for such constructions. To date, however, only two studies have probed the (in)stability of this construction among heritage speakers of Spanish. Dvorak & Kirschner (1982) elicited Spanish-language translations of English ‘like’ sentences from Puerto Ricans in New York City. Results revealed a strong preference for preverbal Experiencers that regulated verbal agreement, as in (4a). Such forms, the authors suggest, are promoted by the fact that the pronominal 3SG Experiencer subject ‘a él’ is distinguished from a 3SG Nominative subject él solely by the appearance of the Dative marker a. Dvorak (1983) further pursues this line of research, employing a grammaticality judgment task and an aural interpretation task administered to Puerto Ricans and Cuban-Americans. Results from this study led to a re-evaluation of previous conclusions. In particular, Dvorak construes data such as in (4a) as indicating that the Experiencer, in preverbal position, is interpreted as the ‘true’ subject of the verb. In fact, in evaluating the item in (4b), in which the Theme is inanimate and in preverbal position, her participants often rejected it, commenting that “the candy can’t like the kids.” (4) a. A

él le gusta los deportes he CL-DAT-3SG gustar-SG the sports “He likes sports.” b. El dulce les encanta a los niños the sweets CL-DAT-3PL enchant-3SG DAT the children “The children like the candy.” DAT

While these two studies describe heritage speaker grammars that diverge from the target norm, there are several limitations that must be redressed, among these, the reduced number and diversity of speakers represented, the production versus interpretation of target forms, and the inattention to the animacy and number (SG/PL) of the arguments in the test items. In addition, several hypotheses remain unexplored. For example, while Dvorak & Kirschner noted a tendency for the Dative marker ‘a’ to be lost, there are no examples to demonstrate whether the source is a phonological reduction of an already weak unstressed vowel, as suggested, or whether it is the erosion of a syntactic case marker. Likewise, the validity of the account for (4a) could have been confirmed with Dative forms that are morphologically distinct—the prediction is that heritage speakers should produce correct translations for I like sports, A mí me gustan los deportes, in which the

266

A. JACQUELINE TORIBIO & CARLOS NYE

Experiencer, a mí, is more fully distinguished from the nominative counterpart ‘yo’ (cf. *Yo me gusta/gustan los deportes). Furthermore, Dvorak concludes that the patterns do not yet appear to be stable, but fails to identify which of the properties in (3) might be most centrally implicated in this indeterminacy. Finally, while Dvorak has added to the available empirical data addressing psychological predicates in heritage speakers’ Spanish, she fails to examine implications for syntactic knowledge and representations. 2.

The Present Study Consonant with the foregoing discussion, the present study is guided by the following questions: (5) a. Do speakers of contact/contracting varieties of Spanish reveal a tendency towards reduction and resolution of the indirect mapping properties that characterize psychological predicate constructions? If so, how? b. If attested, is this non-target behavior reflective of vulnerability that is constrained to the areas where the core syntax interfaces with the lexical/semantic and the discourse/pragmatic modules? The prediction is that heritage speakers will indeed seek to reconcile the attributes of psychological predicates, and will converge on a construction that will favor transparency in thematic mapping and syntactic licensing. 2.1.

Participants Twenty-four Spanish-English bilingual undergraduate students from The Pennsylvania State University participated in the pilot study. All were enrolled in a Spanish-for-Heritage-Speakers class at the time of the study. The participants ranged in age from 18 to 21 years and all had been exposed to Spanish at an early age. 2.2.

Instruments This study was performed during a fifty-minute class period; however, the average time for completion of the five-page questionnaire was approximately forty minutes. The participants were first asked to provide brief biographical information in regard to their acquisition and use of Spanish and English, and were subsequently asked to complete tasks designed to examine their production and knowledge of psychological predicates. Task A consisted of 24 questions and a prompt for a response. All questions included psychological verbs and were modeled such that participants would have to answer with either the Experiencer or the Theme in

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

267

the sentence-initial position. Examples of the question formats are given in (6). The verbs included in this section were: fascinar, interesar, gustar, disgustar, aburrir and hacer falta. The sentences were varied for number in the Experiencer argument, and number and animacy in the Theme argument. (6) Sample questions a. ¿A quién le encanta la salsa puertorriqueña? (Mi tía) La salsa puertorriqueña … “Who likes Puerto Rican salsa music? (My aunt) Puerto Rican salsa music…” b. ¿A quién le gusta el dulce de leche? (Mis compañeros) A mis compañeros … “Who likes dulce de leche? (My room/workmates) My room/workmates …” Task B was an 80-item judgment task in which participants were asked to render a judgment on the sentences according to three criteria: me suena bien “it sounds fine to me”, me suena más o menos bien “it sounds more or less fine to me”, and no me suena bien “it doesn’t sound good to me”. Sixtyfour experimental sentences were presented, each including the verb gustar. Sixteen filler sentences sampled VS and SV word orders in declaratives and interrogatives. All sentences were randomized. As shown in (7), the 24 grammatical test items were of three types: Experiencer-initial, Theme-initial (animate and inanimate), and (clitic+)Verb-initial. (7) a. A

los atletas les gusta el entrenador the athletes CL-DAT-3PL gustar-3SG the trainer “The athletes like the trainer.” b. Las chicas/Los motores le gustan al joven the girls/ the motors CL-DAT-3SG gustar-3PL DAT.the boy “The boy likes the girls/motors.” c. Le gustan las rosas amarilla a la anciana CL-DAT-3SG gustar-3PL the roses yellow DAT the old.woman “The old woman likes the yellow roses.” DAT

The 40 ungrammatical sentences within each of the three types targeted verb agreement (SG/PL), clitic agreement (SG/PL), expression/omission of the Dative marker, and case marking of the Experiencer argument. Sample items follow:

268

A. JACQUELINE TORIBIO & CARLOS NYE

(8) a. *Al

cantante le gusta las admiradoras singer CL-DAT-3SG gustar-3SG the fans “The singer likes the fans.” b. *A la escritora les gustan las narraciones breves DAT the writer CL-DAT-3PL gustar-3PL the narratives short “The writer likes the short stories.” c. *Las rosas amarillas le gustan la anciana the roses yellow CL-DAT-3SG gustar-3PL the old.woman “The old woman likes the yellow roses.” d. *Yo me gustan los mangos pro-NOM-1SG CL-DAT-1SG gustar-3PL the mangoes “I like mangoes.” DAT.the

3.

Results The results revealed in heritage speakers’ written productions and grammaticality judgments are consonant with the hypotheses here: their contact/contracting grammars demonstrate significant instability with respect to their knowledge of the features associated with gustar and an attendant preference for canonical word order patterns (SVO or Agent/Experiencer V Theme). We consider the results of each task in turn. Non-target forms were manifest in the written productions of all but two of the bilinguals. As shown in (9), variability was noted in the expression of the clitic that doubles the Experiencer. Some participants produced no clitic, others produced only the singular form. Again, an explanation based in phonetic reduction avails itself since some Spanish speakers reduce and elide syllable-final /s/, rendering /les/ as [le]; but, only one of the participants was a speaker of such a dialect. (9) a. Los artistas de rock-en-español disgustan a mi madrina the artists of rock-in-Spanish desgustar-3PL DAT my godmother “My godmother dislikes Spanish rock artists.” b. A mis profesores fascina la lingüística DAT my professors fascinar-3SG linguistics “Linguistics fascinates my professors.” c. El candidato le aburre a mis primos the candidate CL-DAT-3SG aburrir-3SG DAT my cousins “The candidate bores my cousins.” d. La religión le fascina a mis tíos the religión CL-DAT-3SG fascinar-3SG DAT my uncles “Religion fascinates my uncles.”

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

269

e. A

los estudiantes le aburre la profesora the students CL-DAT-3SG aburrir-3SG the profesor “The professor bores the students.” f. A mis amigas le atraen los cantantes colombianos DAT my friends CL-DAT-3SG atraer-3PL the singers Colombian “Colombian singers attract my female friends.” DAT

Participants also demonstrated indeterminacy in verbal agreement. The majority of cases of non-target agreement comprised an animate Theme, and agreement was controlled by whichever argument was in preverbal position, as in (10a,b). In other non-target productions, as in (10c), the sole animate argument (the Experiencer) took precedence in governing agreement, irrespective of its position. (10) a. A

la niñera le disgusta los niños mimados the sitter CL-DAT-3SG disgustar-3SG the children spoiled “The baby-sitter dislikes spoiled children.” b. A mi hermana le disgusta los hongos DAT my sister CL-DAT-3SG disgustar-3SG the mushrooms “My sister dislikes mushrooms.” c. La música latina le gustan a los jóvenes the music Latin CL-DAT-3SG gustar-3PL DAT the youth “The youth like Latin music.” DAT

Finally, participants were observed to variably realize the dative marker on the Experiencer argument; the sentences that appear in (11), produced by a single speaker, are representative of a broader pattern. Note that while data such as (11a) might be attributed to phonetic reduction of an utterance-initial unstressed segment, and (11b) might likewise be ascribed to the coalescence of contiguous vowels, data such as in (11c) argue against the validity of an account based in pronunciation. Moreover, as a writing task, participants were afforded time to carefully consider and edit their productions. (11) a. El jefe le interesan los trabajadores the boss CL-DAT-3SG interesar-3PL the workers “The workers interest the boss.” b. La religión les fascina mis tíos the religion CL-DAT-3PL fascinar-3SG my uncles “Religion fascinates my uncles.” c. Los cigarros dominicanos le disgustan mi hermana the cigars Dominican CL-DAT-3SG disgustar-3PL my sister “My sister dislikes Dominican cigars.”

270

A. JACQUELINE TORIBIO & CARLOS NYE

In Task B, participants demonstrated differential judgments in their assessment of grammatical vs. ungrammatical items. Note that the percentages in Tables 1 and 2 reflect the acceptability of grammatical gustar sentences, and rejection of ungrammatical gustar sentences, both of which should approximate 100%. However, as shown in Table 1, participants did not categorically accept grammatical structures; rate of acceptance was 78%. Word order was shown to be an important variable: gustar sentences in which the Experiencer occupied the initial position were most often correctly accepted (Type 1), and those in which the (clitic+)Verb complex preceded the full arguments were least often accepted (Type 3). Animacy was also of relevance when one of the two arguments occupied the initial position (Type 1, 2): participants more accurately judged sentences in which the animacy of the Experiencer and Theme was dissimilar. This pattern was especially notable when the Theme was in initial position. Interestingly, when the (clitic+)Verb occupied the initial position, the reverse pattern was observed: participants more accurately judged sentences with animate Themes. This latter pattern may owe to the fact that the verb carries enough morphology (number on the clitic and the verb) to allow for the correct interpretation of the animate arguments. Overall acceptance of grammatical structures: 78.67% Construction type % Accept Example 1: Initial Experiencer Inanimate Theme Animate Theme

94.44 94.79 93.75

A los obreros les gusta el café A los obreros les gusta el jefe

2: Initial Theme Inanimate Theme Animate Theme

81.25 85.42 72.92

El café les gusta a los obreros El jefe les gusta a los obreros

3: Initial (clitic+) Verb Inanimate Theme Animate Theme

76.04 75.00 78.13

Les gusta el café a los obreros Les gusta el jefe a los obreros

Table 1: Acceptance of grammatical structures

But of greater interest is participants’ failure to reject ungrammatical sentences; with the exception of one type (Type 8), ungrammatical sentences were accepted at rates greater than chance. That is, participants did not emphatically reject ungrammatical structures; as shown in Table 2, rate of rejection was only 34.62%. Especially noteworthy is the acceptance of those items in which a sentence-initial Experiencer regulated verb agreement (Type 4), and the high rate of acceptance of those items in which a sentence-initial

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

271

Theme agreed in number with the Verb and additionally agreed with the Clitic (Type 5); this low rate of rejection suggests that for some speakers the clitic may be invariable (le). Note that although no statistical information was calculated for the present study, a replication by de Prada et al. (2005) revealed statistically significant differences between heritage speakers’ responses to grammatical versus ungrammatical gustar constructions. Overall rejection of ungrammatical structures 34.62% Construction type % Reject Example 4: Verb agreement Experiencer initial Animate Theme Theme initial Animate Theme

30.73 21.88 27.08 39.58 47.92

A los obreros les gustan el café A los obreros les gustan el jefe El café les gustan a los obreros El jefe les gustan a los obreros

5: Clitic agreement Experiencer initial Animate Theme Theme initial Animate Theme

29.69 30.21 37.50 29.17 31.25

A los obreros le gusta el café A los obreros le gusta el jefe El café le gusta a los obreros El jefe le gusta a los obreros

6: DAT marker Experiencer initial Animate Theme Theme initial Animate Theme

41.67 39.58 27.08 44.79 43.75

Los obreros les gusta el café Los obreros les gusta el jefe El café les gusta los obreros Los obreros les gusta el jefe

7: Two errors

32.29

A los obreros le gustan el café

8: NOM Experiencer Case marking V-agreement

68.75 67.71 69.44

Yo me gusta el café Yo me gusto el café

Table 2: Rejection of ungrammatical structures

4.

Discussion and conclusion The trends in the data speak to Spanish heritage speakers’ non-target and variable performance with reverse psychological predicates: participants proffered ungrammatical psychological predicate constructions in the written production task and failed to reject ungrammatical psychological predicate constructions in the judgment task. That is, in answer to the first research question (5a), the findings demonstrate speakers’ indeterminate or incomplete knowledge of the properties associated with this class of verbs. Closer scrutiny of the elicited data unveils two particular strategies:

272

A. JACQUELINE TORIBIO & CARLOS NYE

(12) a. Direct mapping for gustar: Map the animate argument to the structural subject position and the inanimate argument to the structural object position. b. Fixing of SVO constituent order: Dedicate the preverbal position to the subject. The outcome of the first strategy is an Experiencer that may lack the Dative marker a and may control agreement—note that participants reject Experiencers without Dative marking at 39.58% when in initial position and at 44.79% when in internal position; in turn, the Theme may behave like a direct object. The productivity of this strategy is further confirmed in the present study by (13), in which the participant marked the Theme argument with a personal a, normally reserved for definite animate direct objects. It would prove instructive to learn whether additional object properties may also reside with the Theme, e.g., Accusative case in pronominalization, as in (14a); a spontaneous utterance from a non-participant heritage speaker (14b) provides anecdotal support for this claim. (13) Test Prompt: ¿A quién le disgustan los niños mimados? La niñera DAT who CL-DAT-3SG disgustar-3PL the children spoiled? The sitter “Who dislikes spoiled children? The sitter…” Participant Response: A la niñera…le disgustan a los niños mimados DAT the sitter CL-DAT-3PL disgustar-3SG ACC the children spoiled “The baby-sitter dislikes spoiled children.” (14) a. Test Prompt: A los atletas les gusta el nuevo entrenador DAT the athletes CL-DAT-3SG gustar-3SG the new trainer “The athletes like the new trainer.” Response: ¿→? Lo gustan. CL-ACC-3SG gustar-3PL “They like him.” b. ¿Million Dollar Baby? Lo va a gustar Million Dollar Baby? CL-ACC-3SG go-INF to gustar-INF “Million Dollar Baby? You’re going to like it.” The outcome of the second strategy (12b) is a preference for preverbal Experiencers, which renders structures that coincide with those of English

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

273

‘like.’ Note that participants prefer constructions with initial Experiencers (94.44% accept) over those with initial Theme (81.25% accept). In fact, in the third activity, in which participants were asked to complete sentences drawing on a selection of direct and indirect psychological predicates, initial Themes were often inserted into predicate adjective constructions, e.g., es aburrida “is boring”, son interesantes “are interesting”. This pattern is consistent with the findings of Zapata et al. (2005), in which preverbal constituents were largely interpreted by heritage speakers as structural subjects, perhaps owing to the contribution of English. (Of interest, in this respect, would be the alterations in gustar constructions that might characterize heritage Spanish in the context of typologically diverse languages.) Taken together, the strategies exhibited by the heritage speakers in this study may be construed as reflecting their efforts towards a more transparent mapping of argument structure for this class of verbs coupled with a reduction in the word order options that the Spanish language affords. Nevertheless, participants’ production of target forms, however variable, and their acceptance of grammatical forms, reveals knowledge of properties diagnostic of a core syntax that approximates the full monolingual variety, e.g., as pertains to knowledge of Agreement and Case licensing. Thus, in response to research question (5b), the areas of vulnerability fall outside of the core syntax and into the interface modules. In conclusion, speakers of contact and contracting varieties of Spanish show evidence of vulnerability in the lexicosemantic module, as corroborated in incipient restructuring of the argument structure of gustar, and in the domain of discourse/pragmatics, as witnessed in the preference for preverbal subjects. 5.

Limitations of the study and directions for future research Several limitations of this exploratory study will be rectified for completion of a larger investigation. With respect to materials, tokens will be increased and controlled across construction types. Person agreement will be further tested with materials expanded to include constructions with first person plural Experiencer and singular theme—e.g., A nosotros nos gusta el dulce de leche “We like dulce de leche”—to determine whether the plural preverbal Experiencer will induce plural agreement on the verb, i.e., (A) nosotros nos gustamos el dulce de leche (on par with the attested form A mis compañeros les gustan el dulce de leche). Default values for verbs will be tested through constructions with plural arguments and singular verb and clitic morphology (cf. A los obreros le gusta los jefes, a pattern which is typical of colloquial Icelandic). And verbs will include direct psychological predicates such as amar “to love” and querer “to want, to love,” which are translation equivalents of gustar. Procedures will be revised to include a translation task

274

A. JACQUELINE TORIBIO & CARLOS NYE

(as in the early studies of Dvorak and Dvorak & Kirschner), a pronominalization task (cf. 13b), and a picture-matching aural interpretation task (as in Montrul’s second language acquisition research). The participant pool will be expanded to include second-language learners, monolingual speakers of diverse dialects of Spanish, and passive bilinguals, each potentially presenting patterns reflecting diverse stages of language development (acquisition, decline, and change). Finally, sociolinguistic variables (e.g., gender, age, language history, language usage) will be incorporated as variables of analysis. These modifications in mind, the prospective contributions of the research are manifold and may offer novel insights into numerous areas of linguistic study. The endeavor can inform our understanding of language change as mediated by language acquisition, since the transmission of a linguistic system with variable forms that are biased towards convergence (e.g., Experiencer-Verb-Theme order) could lead to incomplete replication of the original syntactic system and indirectly to syntactic change (cf., Bullock & Toribio 2004, Meisel 2001). Thus, researchers must examine the linguistic input to which heritage speakers are exposed, given that at least some nontarget forms may be attributable to converged forms. Similarly, since linguistic convergence may be promoted by considerations of processing economy (cf., Muysken 2000), researchers must explore heritage speakers’ patterns of language usage; non-target, though licit, forms may be more prevalent among bilinguals who are regularly called on to access and activate both language systems (Toribio 2004). Moreover, as stated by Bullock (2004), the study of contact and contracting grammars “allows us a window into synchronic variation and change in real time.” Indeed, the tendencies reported here mirror those of normal diachronic developments in the history of English and in the development from Latin to Spanish, where the propensity for modification among these verbs finds precedence. English presented like as an indirect predicate that paralleled modern Spanish gustar; loss of verbal and case morphology led to the modern-day direct predicate (cf. Lightfoot 1991, Allen 1986, Jespersen 1927). And Latin presents the drift in reverse: Latin gustare was a direct predicate with the meaning of “to taste”, “to savor”, or “to sample (food)”, which existed alongside the reverse predicate placere (Lipski 1975). Evidence of differential permeability of the interface modules can inform the modularity of lexico-semantics vs. discourse-pragmatics. Extant findings reported in the literature to date suggest that the discursive/pragmatic interface may be more vulnerable to modification than the lexico-semantic interface (de Prada et al. 2005). Comparison of systemic alterations in psychological predicate constructions in monolingual vs. bilingual productions and interpretations would be informative in this regard, as would an

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

275

examination of reverse psychological predicates in the speech of Spanishlanguage bilinguals in diverse linguistic contexts (e.g., heritage speakers of Spanish in contact with VSO languages such as Welsh, Arabic or Zapotec). Future work addressing the systemic vulnerability across bilingual and monolingual speech communities can also contribute to debates surrounding the locus of language change. For if monolinguals may be found to demonstrate similar patterns in the lexico-semantic domain, then change cannot be unequivocally ascribed to contact (though, as noted, contact may be a catalyst that exaggerates change). Finally, this research program can advance our understanding of language maintenance in the context of language contact. In particular, our findings support the observation that contact forms need not signal a ‘turnover’ from one system to another (cf., Myers Scotton’s 1998); quite the contrary, such forms may be an indication of language vitality (witness the stability of the narrow syntax in this population and the synchronic and diachronic modifications of gustar). To be sure, as aptly stated by Valdés (2000:119), it may be case that “the survival of a language in a minority-majority context requires the acceptance of ‘interference’ phenomena by its speakers.” References Allen, Cynthia L. 1986. “Reconsidering the history of like”. Journal of Linguistics 22.375-409. Bullock, Barbara. 2004. “Investigating phonological convergence: Contexts and consequences”. Paper presented at Tulane University. ---------- & Almeida Jacqueline Toribio. 2004. “Introduction: Convergence as an emergent property in bilingual speech”. Bilingualism: Language and Cognition 7.91-93. Dvorak, Trisha. 1983. “Subject-object reversals in the use of gustar among New York Hispanics”. Spanish in the U.S. Setting: Beyond the Southwest, ed. By Lucía Elías-Olivares, 21-36. Rosslyn: National Clearinghouse for Bilingual Education. ---------- & Carl. Kirschner. 1982. “Mary likes fishes:psychological phenomena in New York Puerto Rican Spanish”. The Bilingual Review/La Revista Bilingüe 9.59-65. Grimshaw, Jane. 1990. Argument Structure. Cambridge, Mass.: MIT Press. Jespersen, Otto. 1927. A Modern English grammar on historical principles. London: Allen and Unwin. Lightfoot, David. 1991. How to set parameters: Arguments from language change. Cambridge, Mass.: MIT Press. Lipski, John. 1975. “A semantic-syntactic shift in Spanish”. Foglia Lingüística 7.149-163.

276

A. JACQUELINE TORIBIO & CARLOS NYE

Martín, Juan. 1998. “The lexico-syntactic interface:Psych verbs and psych nouns”. Hispania 81.619-631. Meisel, Jurgen. 2001. Working Papers in Multilingualism: From bilingual language acquisition to theories of diachronic change B.30. Collaborative Research Center on Multilingualism, University of Hamburg. Montrul, Silvina. 1997. “Spanish gustar psych verbs and the unaccusative se constructions: the case of Dative experiencers in SLA”. Contemporary Perspectives on the Acquisition of Spanish, ed. by William R. Glass & Ana Teresa Perez-Leroux, 189-207. Somerville: Cascadilla Press. ----------. 1998. “The L2 acquisition of dative experiencer subjects”. Second Language Research 14. 27-61. ----------. 1999. “Activating AgrIOP in second language acquisition”. The Development of second language grammars: A generative approach, ed. by Elaine C. Klein and Gita Martohardjono, 81-108. Amsterdam: John Benjamins. ----------. 2002. “Incomplete acquisition and attrition of Spanish tense/aspect distinctions in adult bilinguals”. Bilingualism: Language & Cognition 5. 39-68. ----------. 2004. “Subject and object expression in Spanish heritage speakers: A case of morphosyntactic convergence”. Bilingualism: Language & Cognition 7.125-142. ----------. 2005. “Second language acquisition and first language loss in adult early bilinguals: Exploring some differences and similarities”. Second Language Research 21.1-51. Muysken, Pieter. 2000. Bilingual speech. Cambridge: Cambridge University Press. Myers-Scotton, Carol. 1998. “A Way to Dusty Death: The Matrix Language Turnover Hypothesis”. Endangered languages: Language loss and community response, ed. by Leonore A. Grenoble & Lyndsay J. Whaley, 289-316. Cambridge: Cambridge University Press. de Prada, Ana, Gabriela Zapata, Almeida Jaqueline Toribio, & Eva María Suárez-Budenbender. 2005. “The core versus peripheral grammars of heritage speakers of Spanish”. Paper presented at the 2005 Hispanic Linguistics Symposium. The Pennsylvania State University. Silva-Corvalán, Carmen. 1994/2000. Language Contact and Change. Oxford: Clarendon Press. Sorace, Antonella. 2000. “Differential effects of attrition in the L1 syntax of near-native L2 speakers”. In Proceedings of the 24th Annual Boston University Conference on Language Development (Vol. 2), ed. by Anna H.J. Do, Laura Domínguez, and Aimee Johansen, 719-725. Somerville, Mass.: Cascadilla Press.

RESTRUCTURING OF REVERSE PSYCHOLOGICAL PREDICATES

277

----------. 2004. “Native language attrition and developmental instability at the syntax-discourse interface: data, interpretations and methods”. Bilingualism: Language and Cognition 7: 143-145. ----------. Forthcoming. “Selective optionality in language development”. Biolinguistic and sociolinguistic accounts of syntactic variation, ed. by Leonie Cornips & Karen Corrigan. Amsterdam: John Benjamins. Toribio, Almeida Jacqueline. 2004. “Convergence as optimization in SpanishEnglish bilingual code-switching”. Bilingualism: Language & Cognition, 7.165-173. ----------, Gabriela Zapata, Eva María Suárez-Budenbender & Hillary Barnes. 2005. “Indeterminacy in the interpretive component”. Paper presented at the 5th International Symposium on Bilingualism, Barcelona. Toth, Paul. 2003. “Psych verbs and morphosyntactic development in instructed L2 Spanish”. Linguistic theory and language development in Hispaniclanguages, ed. by S. Montrul & F. Ordóñez, 468-497. Somerville, Mass.: Cascadilla Press. Tsimpli, Ianthi-Maria, Antonella Sorace, Caroline Heycock & Francesca Filiaci. 2004. “First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English”. International Journal of Bilingualism 8.257-277. Valdés, Guadalupe. 2000. “Bilingualism and language use among Mexican Americans”. New immigrants in the United States, ed. by Sandra Lee McKay & Sau-ling Cinthia Wong, 99-136. Cambridge: Cambridge University Press. Whitley, M. Stanley. 1995. “Gustar and other psych verbs: a problem in transitivity”. Hispania 78.573-585. Zapata, Gabriela, Liliana Sánchez & Almeida Jacqueline Toribio. 2005. “Contact and contracting Spanish”. International Journal of Bilingualism 9.377-396.

SUBJECT INDEX

A. A-bar Position 52, 54 Accommodation 156, 158 Accusative (ACC) 257, 260 Additivity 187-200 Adjacency Effects 57-59 Adjectival Secondary Predication 169-186 Adjectives 45-47 Adult input 225 Agreement 88, 92, 113-126, 201-211, 229-245 Gender Agreement 205 Number Agreement 207 Person Agreement 205 Agreement Affixes 229-245 Allomorphy 239, 242, 243 Anaphoric Element 249, 259 Argument Structure 264, 273 Aspect Generic 98 Continuative 102 Repetitive 108 Aspect P 97 Assertion/Assertive/Asserted 155-167 Atelic 171 Aucun 83-95 Autonomy 161, 162 Auxiliaries 248 Avant 155-167 B. Bare Argument Nominals (BAN) 67-81 C. Case 42-43 Case-checking in morphology 16-7, 22 Chacun 83-96

Chaque 84, 85 C-Command 115, 117, 120 Child Language 213, 216 Clitic(s) 43-45, 51-66, 213-227, 247-261 Non-reflexive Clitics 169-186 Reflexive Clitics 169-186 Clitic Sequences 133-134 Clitic Drop 213-227 Clitic Left-Dislocated (CLLD) 51-66 Closure Closure of Sets 187-200 Epistemic Closure 187-200 Compounds Agent 113, 114, 118 Instrument 113, 118 Invariant 122 PN 19-20 Synthetic 13-26 VN (V-N) 13-26, 99, 113-126 Conceptual 27-49 Conjunction 1-11 Conjunction Reduction 135 Connector 158, 161 Continuation (Law) 158, 160 Continuity 162 Convergence 263, 274 Creole Haitian 67-81 Criterion of Division 67 D. Dative 251, 254 Default 201-211 Determiner Absence of Determiner 44-45 Determiner Phrase (DP) 83-95

280

SUBJECT INDEX

De-verbal Nouns 97-111 Diachronic 83-95 Diphthongization 15-16 Direct Objects 213-227 Discontinuity 223, 224 Discourse 155-167 Discourse-new vs. Discourse-old 144 Disjunction 1-11 Doubling Coordinating Particles 4 Downward-entailing 6 E. EPP 138 Evolution 29, 35, 37 Exaptation 27-49 Experiencer 247-261 Explanation 41-42 Expletive Dominican 58 F. Feature(s) 201-211 Feature Category 113-126 Feature Clash 201-211 Feature Class 113-126 Gender Features 113-126 Feature Sharing 113-126 Morphosyntactic Features 230 Number Features 113-126 Uninterpretable Features 41, 45, 46 Unvalued Features 113-126 Valued Features 113-126 Finite Forms 201, 202 Finiteness 201-211 Focus 143-154 Focus Projection 145-146 French Old 83-95 Middle 83-95

Functional Functional Element 224, 247-261 Functional Head 247-261 Functional Projection 253 G. Gender Feminine 113-126 Masculine 113-126 Gender Mismatch 113-126 Natural 17, 24 113-126 Goal 115, 117 Grammaticalization 247-261 H. Head 113-126 Hearer-new vs. Hearer-old 144-145 Heritage Speakers 263-277 I. Icelandic 127-142 Identification (tracking) 79-80 Idiom(atic) Interpretation 169-186 Impersonal Constructions 130 Impoverishment 229-245 Inclusion Lattice 9 Incorporation 16, 21-22 Individuation 67-81 Spatial 67-81 Infinitive(s) 157, 161, 162 Nominalized 102 Infelction (Infl0) 113-126 Information Structure 143-145 Integration vs. Non-integration 143-154 Inverse 155-167 I(nversion)-Nominals 128

SUBJECT INDEX

Italian Old 248 L. Left-periphery 54, 61, 63 Lexicalization 257 Linearization 27-49 M. Markedness 202-203, 208 Mass/count Distinction 71 Merge 33-34 Microparametric Variation 83-95 Missing Surface Inflection Hypothesis (MSIH) 202, 203, 208 Modal(s) 51-66 Deontic 248, 250 Modal Head 248, 253, 256 Modifier 83-95 Mood Imperative 229-245 Subjunctive 229-245 Morphology 13-26, 201-211 Distributed 204 N. Narration/Narrative 155-167 Negative Concord (NC) 2, 7, 189, 191 Negative Polarity Item (NPI) 1-11, 191 Neppure 187-200 Ni 1 Nominative (NOM) 247-261 Nominalization(s) of VP 13-26 Past Participle Nominalizations 104 Nominalizer 113-126 Nonveridical 7 Null Bare Nouns 214, 225 Null Cognate Object 214, 216, 225 Null Objects 213-227

281

Number Conceptual Role of 67 General Number 70-71 Plural Number 70-71 Syntactic Position of 67 N-words 2-3 O. Object Omission 213-227 Oblique Controllees 136 On/With Entailments 169-186 Operators 7-8 P. Paradigms 229-245 Parametric Discontinuity 224 Parce que 143-154 Pause (as a marker of assertion scope) 144, 146, 150 Perceptual 27-49 Person-Case Constraint 134 Phase Consistency Condition (PCC) 61 Polarity 1-11 Polarity Lattice 8-9 Pragmatic Assertion 143-154 Pragmatics 215, 217, 222 Pre-eminence/Pre-eminent 163, 164 Preposition Directional 169-186 Goal 169-186 Presupposition 155-167, 187-200 Pragmatic Presupposition 145 pro 13-26 PRO 251 Probe 113-126 Projection 113-126 Agreement-active 58, 61 Speaker-oriented 63

282

SUBJECT INDEX

Pronoun(s) 213-227 Indefinite 83-95 Reciprocal 5 Proto-role Analysis 140 Psychological Predicates 263-277 R. Reconstruct 51-66 Recursion 27-49 Recycling 24-25 Reference Time (RT) 159-161 S. Scope 52, 54-55 of Assertion 145-146 Scalarity 187-200 Second Language (L2) 201-203 Second Language Acquisition 209 Sign 27-49 Strengthening 8 Stray Affix Filter 21 Substance 27-49 Subject(s) Argumental Subject 250 Bare NP Subjects 53 Grammatical Subject 253, 254 Negative Quantifier Subjects 57 Preverbal Subjects 51, 57 PRO Subject 250 Quasi-argumental Subject 250 Quirky Subjects 127-142 Subject Position 250, 251, 252 Subject Theta-role 252 Topic Subjects 53 Subordination 155-167 Syncretism 235

T. Telic (Interpretations) 169-186 Temporary Relations 169-186 Tense 201-211, 229-245 Theme 247-261 Thematic Hierarchy 128, 139 Thematic Grid 248, 249, 251 Thematic Role 253, 257 Topic 143-154 Topicality 143-154 Topic-drop 225 Topic Islands 132, 137 Transitivity 214 U. Unaccusative 248, 257 Underspecification 201-211 Universal Quantifier 83-95 V. Variability 201, 202, 209 Verb Canonical Transitive 254, 257 Impersonal 251 Lexical 248, 249 Psych 127-142 Raising 250, 251 Transitive 247-261 VoiceP 97, 104 W. Widening 8 Word Order 127-142

CURRENT ISSUES IN LINGUISTIC THEORY

E. F. K. Koerner, Editor

Zentrum für Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin [email protected] Current Issues in Linguistic Theory (CILT) is a theory-oriented series which welcomes contributions from scholars who have significant proposals to make towards the advancement of our understanding of language, its structure, functioning and development. CILT has been established in order to provide a forum for the presentation and discussion of linguistic opinions of scholars who do not necessarily accept the prevailing mode of thought in linguistic science. It offers an outlet for meaningful contributions to the current linguistic debate, and furnishes the diversity of opinion which a healthy discipline must have. A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 276 MONTREUIL, Jean-Pierre Y. (ed.): New Perspectives on Romance Linguistics. Vol. II: Phonetics, Phonology and Dialectology. Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005. x, 209 pp. Expected August 2006 275 NISHIDA, Chiyo and Jean-Pierre Y. MONTREUIL (eds.): New Perspectives on Romance Linguistics. Vol. I: Morphology, Syntax, Semantics, and Pragmatics. Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005. 2006. xiii, 282 pp. 274 GESS, Randall S. and Deborah ARTEAGA (eds.): Historical Romance Linguistics. Retrospective and perspectives. 2006. viii, 393 pp. 273 FILPPULA, Markku, Juhani KLEMOLA, Marjatta PALANDER and Esa PENTTILÄ (eds.): Dialects Across Borders. Selected papers from the 11th International Conference on Methods in Dialectology (Methods XI), Joensuu, August 2002. 2005. xii, 291 pp. 272 GESS, Randall S. and Edward J. RUBIN (eds.): Theoretical and Experimental Approaches to Romance Linguistics. Selected papers from the 34th Linguistic Symposium on Romance Languages (LSRL), Salt Lake City, March 2004. 2005. viii, 367 pp. 271 BRANNER, David Prager (ed.): The Chinese Rime Tables. Linguistic philosophy and historical-comparative phonology. 2006. viii, 358 pp. 270 GEERTS, Twan, Ivo van GINNEKEN and Haike JACOBS (eds.): Romance Languages and Linguistic Theory 2003. Selected papers from ‘Going Romance’ 2003, Nijmegen, 20–22 November. 2005. viii, 369 pp. 269 HARGUS, Sharon and Keren RICE (eds.): Athabaskan Prosody. 2005. xii, 432 pp. 268 CRAVENS, Thomas D. (ed.): Variation and Reconstruction. 2006. viii, 223 pp. 267 ALHAWARY, Mohammad T. and Elabbas BENMAMOUN (eds.): Perspectives on Arabic Linguistics XVII– XVIII. Papers from the seventeenth and eighteenth annual symposia on Arabic linguistics. Volume XVII–XVIII: Alexandria, 2003 and Norman, Oklahoma 2004. 2005. xvi, 315 pp. 266 BOUDELAA, Sami (ed.): Perspectives on Arabic Linguistics XVI. Papers from the sixteenth annual symposium on Arabic linguistics, Cambridge, March 2002. 2006. xii, 181 pp. 265 CORNIPS, Leonie and Karen P. CORRIGAN (eds.): Syntax and Variation. Reconciling the Biological and the Social. 2005. vi, 312 pp. 264 DRESSLER, Wolfgang U., Dieter KASTOVSKY, Oskar E. PFEIFFER and Franz RAINER (eds.): Morphology and its demarcations. Selected papers from the 11th Morphology meeting, Vienna, February 2004. With the assistance of Francesco Gardani and Markus A. Pöchtrager. 2005. xiv, 320 pp. 263 BRANCO, António, Tony McENERY and Ruslan MITKOV (eds.): Anaphora Processing. Linguistic, cognitive and computational modelling. 2005. x, 449 pp. 262 VAJDA, Edward J. (ed.): Languages and Prehistory of Central Siberia. 2004. x, 275 pp. 261 KAY, Christian J. and Jeremy J. SMITH (eds.): Categorization in the History of English. 2004. viii, 268 pp. 260 NICOLOV, Nicolas, Kalina BONTCHEVA, Galia ANGELOVA and Ruslan MITKOV (eds.): Recent Advances in Natural Language Processing III. Selected papers from RANLP 2003. 2004. xii, 402 pp. 259 CARR, Philip, Jacques DURAND and Colin J. EWEN (eds.): Headhood, Elements, Specification and Contrastivity. Phonological papers in honour of John Anderson. 2005. xxviii, 405 pp. 258 AUGER, Julie, J. Clancy CLEMENTS and Barbara VANCE (eds.): Contemporary Approaches to Romance Linguistics. Selected Papers from the 33rd Linguistic Symposium on Romance Languages (LSRL), Bloomington, Indiana, April 2003. With the assistance of Rachel T. Anderson. 2004. viii, 404 pp. 257 FORTESCUE, Michael, Eva Skafte JENSEN, Jens Erik MOGENSEN and Lene SCHØSLER (eds.): Historical Linguistics 2003. Selected papers from the 16th International Conference on Historical Linguistics, Copenhagen, 11–15 August 2003. 2005. x, 312 pp.

256 BOK-BENNEMA, Reineke, Bart HOLLEBRANDSE, Brigitte KAMPERS-MANHE and Petra SLEEMAN (eds.): Romance Languages and Linguistic Theory 2002. Selected papers from ‘Going Romance’, Groningen, 28–30 November 2002. 2004. viii, 273 pp. 255 MEULEN, Alice ter and Werner ABRAHAM (eds.): The Composition of Meaning. From lexeme to discourse. 2004. vi, 232 pp. 254 BALDI, Philip and Pietro U. DINI (eds.): Studies in Baltic and Indo-European Linguistics. In honor of William R. Schmalstieg. 2004. xlvi, 302 pp. 253 CAFFAREL, Alice, J.R. MARTIN and Christian M.I.M. MATTHIESSEN (eds.): Language Typology. A functional perspective. 2004. xiv, 702 pp. 252 KAY, Christian J., Carole HOUGH and Irené WOTHERSPOON (eds.): New Perspectives on English Historical Linguistics. Selected papers from 12 ICEHL, Glasgow, 21–26 August 2002. Volume II: Lexis and Transmission. 2004. xii, 273 pp. 251 KAY, Christian J., Simon HOROBIN and Jeremy J. SMITH (eds.): New Perspectives on English Historical Linguistics. Selected papers from 12 ICEHL, Glasgow, 21–26 August 2002. Volume I: Syntax and Morphology. 2004. x, 264 pp. 250 JENSEN, John T.: Principles of Generative Phonology. An introduction. 2004. xii, 324 pp. 249 BOWERN, Claire and Harold KOCH (eds.): Australian Languages. Classification and the comparative method. 2004. xii, 377 pp. (incl. CD-Rom). 248 WEIGAND, Edda (ed.): Emotion in Dialogic Interaction. Advances in the complex. 2004. xii, 284 pp. 247 PARKINSON, Dilworth B. and Samira FARWANEH (eds.): Perspectives on Arabic Linguistics XV. Papers from the Fifteenth Annual Symposium on Arabic Linguistics, Salt Lake City 2001. 2003. x, 214 pp. 246 HOLISKY, Dee Ann and Kevin TUITE (eds.): Current Trends in Caucasian, East European and Inner Asian Linguistics. Papers in honor of Howard I. Aronson. 2003. xxviii, 426 pp. 245 QUER, Josep, Jan SCHROTEN, Mauro SCORRETTI, Petra SLEEMAN and Els VERHEUGD (eds.): Romance Languages and Linguistic Theory 2001. Selected papers from 'Going Romance', Amsterdam, 6–8 December 2001. 2003. viii, 355 pp. 244 PÉREZ-LEROUX, Ana Teresa and Yves ROBERGE (eds.): Romance Linguistics. Theory and Acquisition. Selected papers from the 32nd Linguistic Symposium on Romance Languages (LSRL), Toronto, April 2002. 2003. viii, 388 pp. 243 CUYCKENS, Hubert, Thomas BERG, René DIRVEN and Klaus-Uwe PANTHER (eds.): Motivation in Language. Studies in honor of Günter Radden. 2003. xxvi, 403 pp. 242 SEUREN, Pieter A.M. and Gerard KEMPEN (eds.): Verb Constructions in German and Dutch. 2003. vi, 316 pp. 241 LECARME, Jacqueline (ed.): Research in Afroasiatic Grammar II. Selected papers from the Fifth Conference on Afroasiatic Languages, Paris, 2000. 2003. viii, 550 pp. 240 JANSE, Mark and Sijmen TOL (eds.): Language Death and Language Maintenance. Theoretical, practical and descriptive approaches. With the assistance of Vincent Hendriks. 2003. xviii, 244 pp. 239 ANDERSEN, Henning (ed.): Language Contacts in Prehistory. Studies in Stratigraphy. Papers from the Workshop on Linguistic Stratigraphy and Prehistory at the Fifteenth International Conference on Historical Linguistics, Melbourne, 17 August 2001. 2003. viii, 292 pp. 238 NÚÑEZ-CEDEÑO, Rafael, Luis LÓPEZ and Richard CAMERON (eds.): A Romance Perspective on Language Knowledge and Use. Selected papers from the 31st Linguistic Symposium on Romance Languages (LSRL), Chicago, 19–22 April 2001. 2003. xvi, 386 pp. 237 BLAKE, Barry J. and Kate BURRIDGE (eds.): Historical Linguistics 2001. Selected papers from the 15th International Conference on Historical Linguistics, Melbourne, 13–17 August 2001. Editorial Assistant: Jo Taylor. 2003. x, 444 pp. 236 SIMON-VANDENBERGEN, Anne-Marie, Miriam TAVERNIERS and Louise J. RAVELLI (eds.): Grammatical Metaphor. Views from systemic functional linguistics. 2003. vi, 453 pp. 235 LINN, Andrew R. and Nicola McLELLAND (eds.): Standardization. Studies from the Germanic languages. 2002. xii, 258 pp. 234 WEIJER, Jeroen van de, Vincent J. van HEUVEN and Harry van der HULST (eds.): The Phonological Spectrum. Volume II: Suprasegmental structure. 2003. x, 264 pp. 233 WEIJER, Jeroen van de, Vincent J. van HEUVEN and Harry van der HULST (eds.): The Phonological Spectrum. Volume I: Segmental structure. 2003. x, 308 pp. 232 BEYSSADE, Claire, Reineke BOK-BENNEMA, Frank DRIJKONINGEN and Paola MONACHESI (eds.): Romance Languages and Linguistic Theory 2000. Selected papers from ‘Going Romance’ 2000, Utrecht, 30 November–2 December. 2002. viii, 354 pp. 231 CRAVENS, Thomas D.: Comparative Historical Dialectology. Italo-Romance clues to Ibero-Romance sound change. 2002. xii, 163 pp.

230 PARKINSON, Dilworth B. and Elabbas BENMAMOUN (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XIII-XIV: Stanford, 1999 and Berkeley, California 2000. 2002. xiv, 250 pp. 229 NEVIN, Bruce E. and Stephen B. JOHNSON (eds.): The Legacy of Zellig Harris. Language and information into the 21st century. Volume 2: Mathematics and computability of language. 2002. xx, 312 pp. 228 NEVIN, Bruce E. (ed.): The Legacy of Zellig Harris. Language and information into the 21st century. Volume 1: Philosophy of science, syntax and semantics. 2002. xxxvi, 323 pp. 227 FAVA, Elisabetta (ed.): Clinical Linguistics. Theory and applications in speech pathology and therapy. 2002. xxiv, 353 pp. 226 LEVIN, Saul: Semitic and Indo-European. Volume II: Comparative morphology, syntax and phonetics. 2002. xviii, 592 pp. 225 SHAHIN, Kimary N.: Postvelar Harmony. 2003. viii, 344 pp. 224 FANEGO, Teresa, Belén MÉNDEZ-NAYA and Elena SEOANE (eds.): Sounds, Words, Texts and Change. Selected papers from 11 ICEHL, Santiago de Compostela, 7–11 September 2000. Volume 2. 2002. x, 310 pp. 223 FANEGO, Teresa, Javier PÉREZ-GUERRA and María José LÓPEZ-COUSO (eds.): English Historical Syntax and Morphology. Selected papers from 11 ICEHL, Santiago de Compostela, 7–11 September 2000. Volume 1. 2002. x, 306 pp. 222 HERSCHENSOHN, Julia, Enrique MALLÉN and Karen ZAGONA (eds.): Features and Interfaces in Romance. Essays in honor of Heles Contreras. 2001. xiv, 302 pp. 221 D’HULST, Yves, Johan ROORYCK and Jan SCHROTEN (eds.): Romance Languages and Linguistic Theory 1999. Selected papers from ‘Going Romance’ 1999, Leiden, 9–11 December 1999. 2001. viii, 406 pp. 220 SATTERFIELD, Teresa, Christina M. TORTORA and Diana CRESTI (eds.): Current Issues in Romance Languages. Selected papers from the 29th Linguistic Symposium on Romance Languages (LSRL), Ann Arbor, 8–11 April 1999. 2002. viii, 412 pp. 219 ANDERSEN, Henning (ed.): Actualization. Linguistic Change in Progress. Papers from a workshop held at the 14th International Conference on Historical Linguistics, Vancouver, B.C., 14 August 1999. 2001. vii, 250 pp. 218 BENDJABALLAH, Sabrina, Wolfgang U. DRESSLER, Oskar E. PFEIFFER and Maria D. VOEIKOVA (eds.): Morphology 2000. Selected papers from the 9th Morphology Meeting, Vienna, 24–28 February 2000. 2002. viii, 317 pp. 217 WILTSHIRE, Caroline R. and Joaquim CAMPS (eds.): Romance Phonology and Variation. Selected papers from the 30th Linguistic Symposium on Romance Languages, Gainesville, Florida, February 2000. 2002. xii, 238 pp. 216 CAMPS, Joaquim and Caroline R. WILTSHIRE (eds.): Romance Syntax, Semantics and L2 Acquisition. Selected papers from the 30th Linguistic Symposium on Romance Languages, Gainesville, Florida, February 2000. 2001. xii, 246 pp. 215 BRINTON, Laurel J. (ed.): Historical Linguistics 1999. Selected papers from the 14th International Conference on Historical Linguistics, Vancouver, 9–13 August 1999. 2001. xii, 398 pp. 214 WEIGAND, Edda and Marcelo DASCAL (eds.): Negotiation and Power in Dialogic Interaction. 2001. viii, 303 pp. 213 SORNICOLA, Rosanna, Erich POPPE and Ariel SHISHA-HALEVY (eds.): Stability, Variation and Change of Word-Order Patterns over Time. With the assistance of Paola Como. 2000. xxxii, 323 pp. 212 REPETTI, Lori (ed.): Phonological Theory and the Dialects of Italy. 2000. x, 301 pp. 211 ELŠÍK, Viktor and Yaron MATRAS (eds.): Grammatical Relations in Romani. The Noun Phrase. with a Foreword by Frans Plank (Universität Konstanz). 2000. x, 244 pp. 210 DWORKIN, Steven N. and Dieter WANNER (eds.): New Approaches to Old Problems. Issues in Romance historical linguistics. 2000. xiv, 235 pp. 209 KING, Ruth: The Lexical Basis of Grammatical Borrowing. A Prince Edward Island French case study. 2000. xvi, 241 pp. 208 ROBINSON, Orrin W.: Whose German? The ach/ich alternation and related phenomena in ‘standard’ and ‘colloquial’. 2001. xii, 178 pp. 207 SANZ, Montserrat: Events and Predication. A new approach to syntactic processing in English and Spanish. 2000. xiv, 219 pp. 206 FAWCETT, Robin P.: A Theory of Syntax for Systemic Functional Linguistics. 2000. xxiv, 360 pp. 205 DIRVEN, René, Roslyn M. FRANK and Cornelia ILIE (eds.): Language and Ideology. Volume 2: descriptive cognitive approaches. 2001. vi, 264 pp. 204 DIRVEN, René, Bruce HAWKINS and Esra SANDIKCIOGLU (eds.): Language and Ideology. Volume 1: theoretical cognitive approaches. 2001. vi, 301 pp. 203 NORRICK, Neal R.: Conversational Narrative. Storytelling in everyday talk. 2000. xiv, 233 pp. 202 LECARME, Jacqueline, Jean LOWENSTAMM and Ur SHLONSKY (eds.): Research in Afroasiatic Grammar. Papers from the Third conference on Afroasiatic Languages, Sophia Antipolis, 1996. 2000. vi, 386 pp.

201 DRESSLER, Wolfgang U., Oskar E. PFEIFFER, Markus A. PÖCHTRAGER and John R. RENNISON (eds.): Morphological Analysis in Comparison. 2000. x, 261 pp. 200 ANTTILA, Raimo: Greek and Indo-European Etymology in Action. Proto-Indo-European *ag�-. 2000. xii, 314 pp. 199 PÜTZ, Martin and Marjolijn H. VERSPOOR (eds.): Explorations in Linguistic Relativity. 2000. xvi, 369 pp. 198 NIEMEIER, Susanne and René DIRVEN (eds.): Evidence for Linguistic Relativity. 2000. xxii, 240 pp. 197 COOPMANS, Peter, Martin EVERAERT and Jane GRIMSHAW (eds.): Lexical Specification and Insertion. 2000. xviii, 476 pp. 196 HANNAHS, S.J. and Mike DAVENPORT (eds.): Issues in Phonological Structure. Papers from an International Workshop. 1999. xii, 268 pp. 195 HERRING, Susan C., Pieter van REENEN and Lene SCHØSLER (eds.): Textual Parameters in Older Languages. 2001. x, 448 pp. 194 COLEMAN, Julie and Christian J. KAY (eds.): Lexicology, Semantics and Lexicography. Selected papers from the Fourth G. L. Brook Symposium, Manchester, August 1998. 2000. xiv, 257 pp. 193 KLAUSENBURGER, Jurgen: Grammaticalization. Studies in Latin and Romance morphosyntax. 2000. xiv, 184 pp. 192 ALEXANDROVA, Galina M. and Olga ARNAUDOVA (eds.): The Minimalist Parameter. Selected papers from the Open Linguistics Forum, Ottawa, 21–23 March 1997. 2001. x, 360 pp. 191 SIHLER, Andrew L.: Language History. An introduction. 2000. xvi, 298 pp. 190 BENMAMOUN, Elabbas (ed.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XII: Urbana-Champaign, Illinois, 1998. 1999. viii, 204 pp. 189 NICOLOV, Nicolas and Ruslan MITKOV (eds.): Recent Advances in Natural Language Processing II. Selected papers from RANLP ’97. 2000. xi, 422 pp. 188 SIMMONS, Richard VanNess: Chinese Dialect Classification. A comparative approach to Harngjou, Old Jintarn, and Common Northern Wu. 1999. xviii, 317 pp. 187 FRANCO, Jon, Alazne LANDA and Juan MARTÍN (eds.): Grammatical Analyses in Basque and Romance Linguistics. Papers in honor of Mario Saltarelli. 1999. viii, 306 pp. 186 MIŠESKA TOMIĆ, Olga and Milorad RADOVANOVIĆ (eds.): History and Perspectives of Language Study. Papers in honor of Ranko Bugarski. 2000. xxii, 314 pp. 185 AUTHIER, Jean-Marc, Barbara E. BULLOCK and Lisa A. REED (eds.): Formal Perspectives on Romance Linguistics. Selected papers from the 28th Linguistic Symposium on Romance Languages (LSRL XXVIII), University Park, 16–19 April 1998. 1999. xii, 334 pp. 184 SAGART, Laurent: The Roots of Old Chinese. 1999. xii, 272 pp. 183 CONTINI-MORAVA, Ellen and Yishai TOBIN (eds.): Between Grammar and Lexicon. 2000. xxxii, 365 pp. 182 KENESEI, István (ed.): Crossing Boundaries. Advances in the theory of Central and Eastern European languages. 1999. viii, 302 pp. 181 MOHAMMAD, Mohammad A.: Word Order, Agreement and Pronominalization in Standard and Palestinian Arabic. 2000. xvi, 197 pp. 180 MEREU, Lunella (ed.): Boundaries of Morphology and Syntax. 1999. viii, 314 pp. 179 RINI, Joel: Exploring the Role of Morphology in the Evolution of Spanish. 1999. xvi, 187 pp. 178 FOOLEN, Ad and Frederike van der LEEK (eds.): Constructions in Cognitive Linguistics. Selected papers from the Fifth International Cognitive Linguistics Conference, Amsterdam, 1997. 2000. xvi, 338 pp. 177 CUYCKENS, Hubert and Britta E. ZAWADA (eds.): Polysemy in Cognitive Linguistics. Selected papers from the International Cognitive Linguistics Conference, Amsterdam, 1997. 2001. xxviii, 296 pp. 176 VAN HOEK, Karen, Andrej A. KIBRIK and Leo NOORDMAN (eds.): Discourse Studies in Cognitive Linguistics. Selected papers from the 5th International Cognitive Linguistics Conference, Amsterdam, July 1997. 1999. vi, 187 pp. 175 GIBBS, JR., Raymond W. and Gerard J. STEEN (eds.): Metaphor in Cognitive Linguistics. Selected papers from the 5th International Cognitive Linguistics Conference, Amsterdam, 1997. 1999. viii, 226 pp. 174 HALL, T. Alan and Ursula KLEINHENZ (eds.): Studies on the Phonological Word. 1999. viii, 298 pp. 173 TREVIÑO, Esthela and José LEMA (eds.): Semantic Issues in Romance Syntax. 1999. viii, 309 pp. 172 DIMITROVA-VULCHANOVA, Mila and Lars HELLAN (eds.): Topics in South Slavic Syntax and Semantics. 1999. xxviii, 263 pp. 171 WEIGAND, Edda (ed.): Contrastive Lexical Semantics. 1998. x, 270 pp. 170 LAMB, Sydney M.: Pathways of the Brain. The neurocognitive basis of language. 1999. xii, 418 pp. 169 GHADESSY, Mohsen (ed.): Text and Context in Functional Linguistics. 1999. xviii, 340 pp. 168 RATCLIFFE, Robert R.: The “Broken” Plural Problem in Arabic and Comparative Semitic. Allomorphy and analogy in non-concatenative morphology. 1998. xii, 261 pp. 167 BENMAMOUN, Elabbas, Mushira EID and Niloofar HAERI (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XI: Atlanta, Georgia, 1997. 1998. viii, 231 pp.

166 LEMMENS, Maarten: Lexical Perspectives on Transitivity and Ergativity. Causative constructions in English. 1998. xii, 268 pp. 165 BUBENÍK, Vít: A Historical Syntax of Late Middle Indo-Aryan (Apabhram�śa). 1998. xxiv, 265 pp. 164 SCHMID, Monika S., Jennifer R. AUSTIN and Dieter STEIN (eds.): Historical Linguistics 1997. Selected papers from the 13th International Conference on Historical Linguistics, Düsseldorf, 10–17 August 1997. 1998. x, 409 pp. 163 LOCKWOOD, David G., Peter H. FRIES and James E. COPELAND (eds.): Functional Approaches to Language, Culture and Cognition. Papers in honor of Sydney M. Lamb. 2000. xxxiv, 656 pp. 162 HOGG, Richard M. and Linda van BERGEN (eds.): Historical Linguistics 1995. Volume 2: Germanic linguistics.. Selected papers from the 12th International Conference on Historical Linguistics, Manchester, August 1995. 1998. x, 365 pp. 161 SMITH, John Charles and Delia BENTLEY (eds.): Historical Linguistics 1995. Volume 1: General issues and non-Germanic Languages.. Selected papers from the 12th International Conference on Historical Linguistics, Manchester, August 1995. 2000. xii, 438 pp. 160 SCHWEGLER, Armin, Bernard TRANEL and Myriam URIBE-ETXEBARRIA (eds.): Romance Linguistics: Theoretical Perspectives. Selected papers from the 27th Linguistic Symposium on Romance Languages (LSRL XXVII), Irvine, 20–22 February, 1997. 1998. vi, 349 pp. + index. 159 JOSEPH, Brian D., Geoffrey C. HORROCKS and Irene PHILIPPAKI-WARBURTON (eds.): Themes in Greek Linguistics II. 1998. x, 335 pp. 158 SÁNCHEZ-MACARRO, Antonia and Ronald CARTER (eds.): Linguistic Choice across Genres. Variation in spoken and written English. 1998. viii, 338 pp. 157 LEMA, José and Esthela TREVIÑO (eds.): Theoretical Analyses on Romance Languages. Selected papers from the 26th Linguistic Symposium on Romance Languages (LSRL XXVI), Mexico City, 28–30 March, 1996. 1998. viii, 380 pp. 156 MATRAS, Yaron, Peter BAKKER and Hristo KYUCHUKOV (eds.): The Typology and Dialectology of Romani. 1997. xxxii, 223 pp. 155 FORGET, Danielle, Paul HIRSCHBÜHLER, France MARTINEAU and María Luisa RIVERO (eds.): Negation and Polarity. Syntax and semantics. Selected papers from the colloquium Negation: Syntax and Semantics. Ottawa, 11–13 May 1995. 1997. viii, 367 pp. 154 SIMON-VANDENBERGEN, Anne-Marie, Kristin DAVIDSE and Dirk NOËL (eds.): Reconnecting Language. Morphology and Syntax in Functional Perspectives. 1997. xiii, 339 pp. 153 EID, Mushira and Robert R. RATCLIFFE (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume X: Salt Lake City, 1996. 1997. vii, 296 pp. 152 HIRAGA, Masako K., Christopher SINHA and Sherman WILCOX (eds.): Cultural, Psychological and Typological Issues in Cognitive Linguistics. Selected papers of the bi-annual ICLA meeting in Albuquerque, July 1995. 1999. viii, 338 pp. 151 LIEBERT, Wolf-Andreas, Gisela REDEKER and Linda WAUGH (eds.): Discourse and Perspective in Cognitive Linguistics. 1997. xiv, 270 pp. 150 VERSPOOR, Marjolijn H., Kee Dong LEE and Eve SWEETSER (eds.): Lexical and Syntactical Constructions and the Construction of Meaning. Proceedings of the Bi-annual ICLA meeting in Albuquerque, July 1995. 1997. xii, 454 pp. 149 HALL, T. Alan: The Phonology of Coronals. 1997. x, 176 pp. 148 WOLF, George and Nigel LOVE (eds.): Linguistics Inside Out. Roy Harris and his critics. 1997. xxviii, 344 pp. 147 HEWSON, John: The Cognitive System of the French Verb. 1997. xii, 187 pp. 146 HINSKENS, Frans, Roeland van HOUT and W. Leo WETZELS (eds.): Variation, Change, and Phonological Theory. 1997. x, 314 pp. 145 HEWSON, John and Vít BUBENÍK: Tense and Aspect in Indo-European Languages. Theory, typology, diachrony. 1997. xii, 403 pp. 144 SINGH, R.K. (ed.): Trubetzkoy's Orphan. Proceedings of the Montréal Roundtable on “Morphonology: contemporary responses” (Montréal, October 1994). With the collaboration of Richard Desrochers. 1996. xiv, 363 pp. 143 ATHANASIADOU, Angeliki and René DIRVEN (eds.): On Conditionals Again. 1997. viii, 418 pp. 142 SALMONS, Joseph C. and Brian D. JOSEPH (eds.): Nostratic. Sifting the Evidence. 1998. vi, 293 pp. 141 EID, Mushira and Dilworth B. PARKINSON (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume IX: Washington D.C., 1995. 1996. xiii, 249 pp. 140 BLACK, James R. and Virginia MOTAPANYANE (eds.): Clitics, Pronouns and Movement. 1997. 375 pp. 139 BLACK, James R. and Virginia MOTAPANYANE (eds.): Microparametric Syntax and Dialect Variation. 1996. xviii, 269 pp. 138 SACKMANN, Robin and Monika BUDDE (eds.): Theoretical Linguistics and Grammatical Description. Papers in honour of Hans-Heinrich Lieb. 1996. x, 375 pp.

137 LIPPI-GREEN, Rosina L. and Joseph C. SALMONS (eds.): Germanic Linguistics. Syntactic and diachronic. 1996. viii, 192 pp. 136 MITKOV, Ruslan and Nicolas NICOLOV (eds.): Recent Advances in Natural Language Processing. Selected Papers from RANLP ’95. 1997. xii, 472 pp. 135 BRITTON, Derek (ed.): English Historical Linguistics 1994. Papers from the 8th International Conference on English Historical Linguistics (8 ICEHL, Edinburgh, 19–23 September 1994). 1996. viii, 403 pp. 134 EID, Mushira (ed.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume VIII: Amherst, Massachusetts 1994. 1996. vii, 261 pp. 133 ZAGONA, Karen (ed.): Grammatical Theory and Romance Languages. Selected papers from the 25th Linguistic Symposium on Romance Languages (LSRL XXV) Seattle, 2–4 March 1995. 1996. vi, 330 pp. 132 HERSCHENSOHN, Julia: Case Suspension and Binary Complement Structure in French. 1996. xi, 200 pp. 131 HUALDE, José Ignacio, Joseba A. LAKARRA and R.L. TRASK (eds.): Towards a History of the Basque Language. 1996. 365 pp. 130 EID, Mushira (ed.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume VII: Austin, Texas 1993. 1995. vii, 192 pp. 129 LEVIN, Saul: Semitic and Indo-European. Volume I: The Principal Etymologies. With observations on AfroAsiatic. 1995. xxii, 514 pp. 128 GUY, Gregory R., Crawford FEAGIN, Deborah SCHIFFRIN and John BAUGH (eds.): Towards a Social Science of Language. Papers in honor of William Labov. Volume 2: Social interaction and discourse structures. 1997. xviii, 358 pp. 127 GUY, Gregory R., Crawford FEAGIN, Deborah SCHIFFRIN and John BAUGH (eds.): Towards a Social Science of Language. Papers in honor of William Labov. Volume 1: Variation and change in language and society. 1996. xviii, 436 pp. 126 MATRAS, Yaron (ed.): Romani in Contact. The history, structure and sociology of a language. 1995. xvii, 208 pp. 125 SINGH, R.K. (ed.): Towards a Critical Sociolinguistics. 1996. xiii, 342 pp. 124 ANDERSEN, Henning (ed.): Historical Linguistics 1993. Selected papers from the 11th International Conference on Historical Linguistics, Los Angeles, 16–20 August 1993. 1995. x, 460 pp. 123 AMASTAE, Jon, Grant GOODALL, M. MONTALBETTI and M. PHINNEY (eds.): Contemporary Research in Romance Linguistics. Papers from the XXII Linguistic Symposium on Romance Languages, El Paso/Juárez, February 22–24, 1992. 1995. viii, 381 pp. 122 SMITH, John Charles and Martin MAIDEN (eds.): Linguistic Theory and the Romance Languages. 1995. xiii, 240 pp. 121 HASAN, Ruqaiya, Carmel CLORAN and David G. BUTT (eds.): Functional Descriptions. Theory in practice. 1996. xxxvi, 381 pp. 120 STONHAM, John T.: Combinatorial Morphology. 1994. xii, 206 pp. 119 LIPPI-GREEN, Rosina L.: Language Ideology and Language Change in Early Modern German. A sociolinguistic study of the consonantal system of Nuremberg. 1994. xiv, 150 pp. 118 HASAN, Ruqaiya and Peter H. FRIES (eds.): On Subject and Theme. A discourse functional perspective. 1995. xii, 414 pp. 117 PHILIPPAKI-WARBURTON, Irene, Katerina NICOLAIDIS and Maria SIFIANOU (eds.): Themes in Greek Linguistics. Papers from the First International Conference on Greek Linguistics, Reading, September 1993. 1994. xviii, 534 pp. 116 MILLER, D. Gary: Ancient Scripts and Phonological Knowledge. 1994. xvi, 139 pp. 115 EID, Mushira, Vicente CANTARINO and Keith WALTERS (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. VolumeVI: Columbus, Ohio 1992. 1994. viii, 238 pp. 114 EGLI, Urs, Peter E. PAUSE, Christoph SCHWARZE, Arnim von STECHOW and Götz WIENOLD (eds.): Lexical Knowledge in the Organization of Language. 1995. xiv, 367 pp. 113 FERNÁNDEZ, Francisco Moreno, Miguel FUSTER and Juan Jose CALVO (eds.): English Historical Linguistics 1992. Papers from the 7th International Conference on English Historical Linguistics, Valencia, 22–26 September 1992. 1994. viii, 388 pp. 112 CULIOLI, Antoine: Cognition and Representation in Linguistic Theory. Texts selected, edited and introduced by Michel Liddle. Translated with the assistance of John T. Stonham. 1995. x, 161 pp. 111 TOBIN, Yishai: Invariance, Markedness and Distinctive Feature Analysis. A contrastive study of sign systems in English and Hebrew. 1994. xxii, 406 pp. 110 SIMONE, Raffaele (ed.): Iconicity in Language. 1995. xii, 315 pp. 109 PAGLIUCA, William (ed.): Perspectives on Grammaticalization. 1994. xx, 306 pp. 108 LIEB, Hans-Heinrich: Linguistic Variables. Towards a unified theory of linguistic variation. 1993. xiv, 261 pp. 107 MARLE, Jaap van (ed.): Historical Linguistics 1991. Papers from the 10th International Conference on Historical Linguistics, Amsterdam, August 12–16, 1991. 1993. xviii, 395 pp. 106 AERTSEN, Henk and Robert J. JEFFERS (eds.): Historical Linguistics 1989. Papers from the 9th International Conference on Historical Linguistics, New Brunswick, 14–18 August 1989. 1993. xviii, 538 pp.

105 HUALDE, José Ignacio and Jon Ortiz de URBINA (eds.): Generative Studies in Basque Linguistics. 1993. vi, 334 pp. 104 KURZOVÁ, Helena: From Indo-European to Latin. The evolution of a morphosyntactic type. 1993. xiv, 259 pp. 103 ASHBY, William J., Marianne MITHUN and Giorgio PERISSINOTTO (eds.): Linguistic Perspectives on Romance Languages. Selected Papers from the XXI Linguistic Symposium on Romance Languages, Santa Barbara, February 21–24, 1991. 1993. xxii, 404 pp. 102 DAVIS, Philip W. (ed.): Alternative Linguistics. Descriptive and theoretical modes. 1996. vii, 325 pp. 101 EID, Mushira and Clive HOLES (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume V: Ann Arbor, Michigan 1991. 1993. viii, 347 pp. 100 MUFWENE, Salikoko S. and Lioba MOSHI (eds.): Topics in African Linguistics. Papers from the XXI Annual Conference on African Linguistics, University of Georgia, April 1990. 1993. x, 304 pp. 99 JENSEN, John T.: English Phonology. 1993. x, 251 pp. 98 EID, Mushira and Gregory K. IVERSON (eds.): Principles and Prediction. The analysis of natural language. Papers in honor of Gerald Sanders. 1993. xix, 382 pp. 97 BROGYANYI, Bela and Reiner LIPP (eds.): Comparative-Historical Linguistics: Indo-European and FinnoUgric. Papers in honor of Oswald Szemerényi III. 1993. xii, 566 pp. 96 LIEB, Hans-Heinrich (ed.): Prospects for a New Structuralism. 1992. vii, 275 pp. 95 MILLER, D. Gary: Complex Verb Formation. 1993. xx, 381 pp. 94 HAGÈGE, Claude: The Language Builder. An essay on the human signature in linguistic morphogenesis. 1993. xii, 283 pp. 93 LIPPI-GREEN, Rosina L. (ed.): Recent Developments in Germanic Linguistics. 1992. xii, 163 pp. 92 POYATOS, Fernando: Paralanguage: A linguistic and interdisciplinary approach to interactive speech and sounds. 1993. xii, 478 pp. 91 HIRSCHBÜHLER, Paul and E.F.K. KOERNER (eds.): Romance Languages and Modern Linguistic Theory. Selected papers from the XX Linguistic Symposium on Romance Languages, University of Ottawa, April 10–14, 1990. 1992. viii, 416 pp. 90 KING, Larry D.: The Semantic Structure of Spanish. Meaning and grammatical form. 1992. xii, 287 pp. 89 BURRIDGE, Kate: Syntactic Change in Germanic. Aspects of language change in Germanic with particular reference to Middle Dutch. 1993. xii, 287 pp. 88 SHIELDS, JR., Kenneth: A History of Indo-European Verb Morphology. 1992. viii, 160 pp. 87 BROGYANYI, Bela and Reiner LIPP (eds.): Historical Philology: Greek, Latin, and Romance. Papers in honor of Oswald Szemerényi II. 1992. xii, 386 pp. 86 KESS, Joseph F.: Psycholinguistics. Psychology, linguistics, and the study of natural language. 1992. xiv, 360 pp. 85 BROSELOW, Ellen, Mushira EID and John McCARTHY (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume IV: Detroit, Michigan 1990. 1992. viii, 282 pp. 84 DAVIS, Garry W. and Gregory K. IVERSON (eds.): Explanation in Historical Linguistics. 1992. xiv, 238 pp. 83 FIFE, James and Erich POPPE (eds.): Studies in Brythonic Word Order. 1991. x, 360 pp. 82 VAN VALIN, JR., Robert D. (ed.): Advances in Role and Reference Grammar. 1992. xii, 569 pp. 81 LEHMANN, Winfred P. and Helen-Jo Jakusz HEWITT (eds.): Language Typology 1988. Typological Models in the Service of Reconstruction. 1991. vi, 182 pp. 80 COMRIE, Bernard and Mushira EID (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume III: Salt Lake City, Utah 1989. 1991. xii, 274 pp. 79 ANTONSEN, Elmer H. and Hans Henrich HOCK (eds.): STAEFCRAEFT: Studies in Germanic Linguistics. Selected papers from the 1st and 2nd Symposium on Germanic Linguistics, University of Chicago, 4 April 1985, and University of Illinois at Urbana-Champaign, 3–4 Oct. 1986. 1991. viii, 217 pp. 78 KAC, Michael B.: Grammars and Grammaticality. 1992. x, 259 pp. 77 BOLTZ, William G. and Michael C. SHAPIRO (eds.): Studies in the Historical Phonology of Asian Languages. 1991. viii, 249 pp. 76 WICKENS, Mark A.: Grammatical Number in English Nouns. An empirical and theoretical account. 1992. xvi, 321 pp. 75 DROSTE, Flip G. and John E. JOSEPH (eds.): Linguistic Theory and Grammatical Description. Nine Current Approaches. 1991. viii, 354 pp. 74 LAEUFER, Christiane and Terrell A. MORGAN (eds.): Theoretical Analyses in Romance Linguistics. Selected papers from the Linguistic Symposium on Romance Languages XIX, Ohio State University, April 21–23, 1989. 1991. viii, 515 pp. 73 STAMENOV, Maxim I. (ed.): Current Advances in Semantic Theory. 1991. xi, 565 pp. 72 EID, Mushira and John McCARTHY (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume II: Salt Lake City, Utah 1988. 1990. xiv, 332 pp. 71 O’GRADY, William: Categories and Case. The sentence structure of Korean. 1991. vii, 294 pp. 70 JENSEN, John T.: Morphology. Word structure in generative grammar. 1990. x, 210 pp.

69 WANNER, Dieter and Douglas A. KIBBEE (eds.): New Analyses in Romance Linguistics. Selected papers from the Linguistic Symposium on Romance Languages XVIII, Urbana-Champaign, April 7–9, 1988. 1991. xviii, 385 pp. 68 BALL, Martin J., James FIFE, Erich POPPE and Jenny ROWLAND (eds.): Celtic Linguistics/ Ieithyddiaeth Geltaidd. Readings in the Brythonic Languages. Festschrift for T. Arwyn Watkins. 1990. xxiv, 470 pp. 67 LEHMANN, Winfred P. (ed.): Language Typology 1987. Systematic Balance in Language. Papers from the Linguistic Typology Symposium, Berkeley, 1–3 Dec 1987. 1990. x, 212 pp. 66 ANDERSEN, Henning and E.F.K. KOERNER (eds.): Historical Linguistics 1987. Papers from the 8th International Conference on Historical Linguistics, Lille, August 30-September 4, 1987. 1990. xii, 577 pp. 65 ADAMSON, Sylvia M., Vivien A. LAW, Nigel VINCENT and Susan WRIGHT (eds.): Papers from the 5th International Conference on English Historical Linguistics. 1990. xxi, 583 pp. 64 BROGYANYI, Bela (ed.): Prehistory, History and Historiography of Language, Speech, and Linguistic Theory. Papers in honor of Oswald Szemerényi I. 1992. x, 414 pp. 63 EID, Mushira (ed.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume I: Salt Lake City, Utah 1987. 1990. xiii, 290 pp. 62 FRAJZYNGIER, Zygmunt (ed.): Current Progress in Chadic Linguistics. Proceedings of the International Symposium on Chadic Linguistics, Boulder, Colorado, 1–2 May 1987. 1989. vi, 312 pp. 61 CORRIGAN, Roberta L., Fred R. ECKMAN and Michael NOONAN (eds.): Linguistic Categorization. Proceedings of an International Symposium in Milwaukee, Wisconsin, April 10–11, 1987. 1989. viii, 348 pp. 60 KIRSCHNER, Carl and Janet Ann DECESARIS (eds.): Studies in Romance Linguistics. Selected Proceedings from the XVII Linguistic Symposium on Romance Languages. 1989. xii, 496 pp. 59 VOORST, Jan van: Event Structure. 1988. x, 181 pp. 58 ARBEITMAN, Yoël L. (ed.): Fucus: A Semitic/Afrasian Gathering in Remembrance of Albert Ehrman. 1988. xvi, 530 pp. 57 BUBENÍK, Vít: Hellenistic and Roman Greece as a Sociolinguistic Area. 1989. xvi, 331 pp. 56 HOCKETT, Charles F.: Refurbishing our Foundations. Elementary linguistics from an advanced point of view. 1987. x, 181 pp. 55 HALL, JR., Robert A.: Linguistics and Pseudo-Linguistics. 1987. vii, 147 pp. 54 WEIDERT, Alfons: Tibeto-Burman Tonology. A comparative analysis. 1987. xvii, 512 pp. 53 SANKOFF, David: Diversity and Diachrony. 1986. xii, 430 pp. 52 FASOLD, Ralph W. and Deborah SCHIFFRIN (eds.): Language Change and Variation. 1989. viii, 450 pp. 51 CHATTERJEE, Ranjit: Aspect and Meaning in Slavic and Indic. With a foreword by Paul Friedrich. 1989. xxiii, 137 pp. 50 RUDZKA-OSTYN, Brygida (ed.): Topics in Cognitive Linguistics. 1988. x, 704 pp. 49 WAUGH, Linda and Stephen RUDY (eds.): New Vistas in Grammar: Invariance and Variation. Proceedings of the Second International Roman Jakobson Conference, New York University, Nov. 5–8, 1985. 1991. x, 540 pp. 48 GIACALONE-RAMAT, Anna, Onofrio CARRUBA and Giuliano BERNINI (eds.): Papers from the 7th International Conference on Historical Linguistics. 1987. xvi, 672 pp. 47 LEHMANN, Winfred P. (ed.): Language Typology 1985. Papers from the Linguistic Typology Symposium, Moscow, 9–13 Dec. 1985. 1986. viii, 200 pp. 46 PRIDEAUX, Gary D. and William J. BAKER: Strategies and Structures. The processing of relative clauses. 1987. ix, 197 pp. 45 KOOPMAN, Willem F., Frederike van der LEEK, Olga FISCHER and Roger EATON (eds.): Explanation and Linguistic Change. 1986. viii, 300 pp. 44 JUNGRAITHMAYR, Herrmann and Walter W. MUELLER (eds.): Proceedings of the Fourth International Hamito-Semitic Congress. 1987. xiv, 609 pp. 43 AKAMATSU, Tsutomu: The Theory of Neutralization and the Archiphoneme in Functional Phonology. 1988. xxi, 533 pp. 42 MAKKAI, Adam and Alan K. MELBY (eds.): Linguistics and Philosophy. Festschrift for Rulon S. Wells. 1985. xviii, 472 pp. 41 EATON, Roger, Olga FISCHER, Willem F. KOOPMAN and Frederike van der LEEK (eds.): Papers from the 4th International Conference on English Historical Linguistics, Amsterdam, April 10–13, 1985. 1985. xvii, 341 pp. 40 FRIES, Peter H. and Nancy M. FRIES (eds.): Toward an Understanding of Language. Charles C. Fries in Perspective. 1985. xvi, 384 pp. 39 BENSON, James D., Michael J. CUMMINGS and William S. GREAVES (eds.): Linguistics in a Systemic Perspective. 1988. x, 452 pp. 38 BROGYANYI, Bela and Thomas KRÖMMELBEIN (eds.): Germanic Dialects. Linguistic and Philological Investigations. 1986. ix, 693 pp. 37 GRIFFEN, Toby D.: Aspects of Dynamic Phonology. 1985. ix, 302 pp.