Creolization and Contact (Creole Language Library)

Creolization and Contact

Creole Language Library (CLL) A companion series to the “Journal of Pidgin & Creole Languages”

Editors Jacques Arends

John Victor Singler

Amsterdam

New York

Editorial Advisory Board Mervyn Alleyne

Salikoko Mufwene

Kingston, Jamaica

Chicago

Norbert Boretzky

Pieter Muysken

Bochum

Nijmegen

Lawrence Carrington

Peter Mühlhäusler

Trinidad

Adelaide

Glenn Gilbert

Pieter Seuren

Carbondale, Illinois

Nijmegen

George Huttar

Norval Smith

Dallas

Amsterdam

John Holm Coimbra

Volume 23 Creolization and Contact Edited by Norval Smith and Tonjes Veenstra

Creolization and Contact Edited by

Norval Smith University of Amsterdam

Tonjes Veenstra Free University Berlin

John Benjamins Publishing Company Amsterdam
/
Philadelphia

8

TM

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

Library of Congress Cataloging-in-Publication Data Creolization and Contact / edited by Norval Smith, Tonjes Veenstra. p. cm. (Creole Language Library, issn 0920–9026 ; v. 23) Includes bibliographical references and index. 1. Creole dialects. 2. Languages in contact. I. Smith, Norval. II. Veenstra, Tonjes, 1962- III. Series. PM7831.C743 2001 417’.22--dc21 isbn 90 272 52459 (Eur.) / 1 58811 1113 (US) (Hb; alk. paper)

2001043202

© 2001 – 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

Introduction Norval Smith and Tonjes Veenstra 1. Relexification in creole genesis and its effects on the development of the creole Claire Lefebvre

1

9

2. Voodoo Chile: Differential substrate effects in Saramaccan and Haitian Norval Smith

43

3. Language intertwining: Its depiction in recent literature and its implications for theories of creolisation Anthony P. Grant

81

4. Paralexification in language intertwining Maarten Mous

113

5. Pidginization, creolization and creoloids in Stockholm, Sweden Ulla-Britt Kotsinas

125

6. The origin of creole languages: The perspective of second language learning Pieter C. Muysken 7. Koine formation and creole genesis Jeff Siegel 8. Koineization and creole genesis: Remarks on Jeff Siegel’s contribution Frans Hinskens 9. Convergence and explanations in creole genesis Silvia Kouwenberg 10. Contact-induced language change and Pidgin/Creole genesis Sarah G. Thomason

157 175

199 219 249

vi

Contents

11. Yiddish as a contact language Ellen Prince

263

12. Social stratification and network relations in the formation of Sranan Jacques Arends

291

Index of languages

309

Index of authors

313

Index of subjects

317

Introduction Norval Smith and Tonjes Veenstra

.

Introduction

The idea for the collection of chapters in this book derives originally from a workshop held at the Institute for General Linguistics of the University of Amsterdam in 1995. The Amsterdam Workshop on Language Contact and Creolization was intended to break down some of the artificial barriers that had grown up between researchers in the two fields of Creole Studies and Contact Linguistics. In particular it was the desire of the workshop organizers to invite researchers in various branches of the field of contact linguistics to address the question of what their own particular research agenda had to say about the phenomenon of creolization. Not all the chapters given at that workshop are represented by contributions here, nor were all the articles contained here presented at the workshop. Most workshop articles have since been extensively rewritten, and the original workshop format whereby invited papers were examined in detail by discussants has been abandoned. The theme of this volume is also broader, as is implied by the title Creolization and Contact. Not just contacts between linguistic communities, but contacts between social groups and individuals are considered which may also have had linguistic implications for the formation of the special kind of contact languages represented by creole languages.

. Relexification/Intertwining The first four articles are concerned with the application of the concept of relexification to creole languages. The term intertwined language has been developed to identify those mixed languages that have arisen by processes of relexification. The concept of relexification is not new, deriving as it does from



Norval Smith and Tonjes Veenstra

the work of Sylvain in the 1930s (Sylvain 1936). She claimed that Haitian was relexified Ewe, that is, an Ewe with the Ewe words replaced by French ones. More recently, and more scientifically, the relevance of relexification has been demonstrated for the field of Mixed Languages by Pieter Muysken (1981). The first article here is also concerned with Haitian. Rather than comparing this language to Ewegbe the language principally used in this comparison is the closely related Fongbe. So, Claire Lefebvre discusses the relexification hypothesis of creole genesis. In addition to the initial process of relexification that takes place in language mixing, she adds two other processes: dialect levelling and reanalysis. Dialect levelling accounts for the homogenity of the ensuing creole, while reanalysis applies to lexical entries created through relexification but assigned a phonologically null representation. She presents analyses of determiners, plural markers and aspect markers in Fongbe, French and Haitan. Relexification accounts for the close parallels between the determiner systems of Haitian and Fongbe, the process of dialect levelling is invoked to explain the not-so-close parallel between Haitian and Fongbe, and reanalysis is used to explain why and how the French adverb après was reanalysed as an aspect marker, and ended up having semantic properties similar to the definite future marker in Fongbe. Norval Smith’s contribution is in part a reaction to one aspect of Lefebvre’s article. He examines the Surinam creole languages, where slaves from the Fongbe-speaking area, or more likely the Eastern Gbe coastal area in general, are assumed to have formed the major component of the early slave imports. If it is correct that Fongbe-speaking slaves formed such an important component in both plantation colonies one would expect to find large-scale parallels in the resultant creole languages. There is certainly a shared feature of post-nominal markers present to some degree in the creoles of both colonies, but there is also a significant difference in what is actually marked postnominally. Smith goes on to examine three other facets of grammar, finding one possible parallel feature, but two other aspects where the results are completely different. Anthony Grant in his contribution provides an up-to-date overview of the literature on language intertwining and discusses the implications it has with respect to creole genesis. He identifies a few factors that set intertwining apart from creolization. First, speakers of intertwined languages are in most cases familiar with the languages which go to make up their new language. This is not always the case with speakers of creoles. Second, the results of the two processes differ, since language intertwining involves the perpetuation of most or all of

Introduction

the structure of a previously existing language, and this includes the perpetuation of morphs and their functions. Creolisation involves the creation (albeit often by use of syntagms and zero morphs rather than by inventing new overt morphs) of a new structure. Although language intertwining does not hold the key to the development of creole languages, he argues that if one could separate out the elements which were there in the early stages of creoles from those which are later accretions, it would not be surprising if there was more evidence of language intertwining in the earlier stages of several modern creoles. Maarten Mous argues that it is the process of paralexification that ultimately leads to language intertwining, which he defines as the process by which parallel word forms for one and the same lexical entry come to exist while sharing semantic and morphological characteristics. He illustrates this process with data from Ma’a, and shows that it plays a decisive role in the creation of not only ordinary mixed languages, but also symbiotic mixed languages, secret languages, slang, ritual languages, taboo languages and mother-in-law languages. The extent to which it can replace the process of relexification as one of the forces in creole genesis is unclear according to him. Although it cannot be ruled out, the relative sparsity of morphology in creole languages prevents us from discovering such evidence.

. Second-language learning Both authors treating the relevance of Second-language learning for creolization agree that there appear to be parallels between the two processes. Starting from her long experience with second-language varieties, Kotsinas finds agreement between the strategies used in second-language varieties and those used in the development of pidgins. Starting from the creole end, Muysken finds similar parallels between certain features of creole languages and strategies used in second-language learning. Ulla-Brit Kotsinas investigates second-language varieties in Sweden with respect to their pidgin/creole features, and compares them to Russenorsk and Gastarbeiterdeutsch. She discusses some apparently deviant features in pidginized varieties of Swedish as used by adult first generation immigrants: repetition, circumlocution, analytic decomposition, lexical over-use and semantic overextension. She argues that these features are not primarily caused by interference from the speakers’ mother tongues, but are, instead, the results of compensatory strategies used by the speakers to fulfil their grammatical and pragmatic





Norval Smith and Tonjes Veenstra

needs. Furthermore, most of these strategies seem also to be utilized in pidgins, and just as in pidgins they represent the very first steps in an incipient creolization process in the course of which grammatical markers are developed. The crucial question in Pieter Muysken’s contribution is to what extent the grammatical properties of creole languages can plausibly be atttributed to what we know of the processes of L2 acquisition of their lexifier languages. He takes Negerhollands, the Dutch-related creole language of the Virgin Islands as an example, and compares features of this language with what we know of the acquisition of Dutch as a second language. He argues that the following features of Negerhollands may well be explicable as the possible result of the acquisition of Dutch as a second language: the rigid SVO order, the absence of postpositions, pre-verbal negation, the absence of tense and person marking on verbs, periphrastic possessives, the absence of Dutch er forms in Negerhollands, the use of tense/mood/aspect auxiliaries, the absence of inflection on the adjective, the loss of passive morphology in basilectal Negerhollands, and juxtaposed verb + particle combinations. Among the features that are not as easily explicable in terms of second language acquisition are the pronominal system, the semantic features of the TMA system, serial verb constructions, the prepositional system and number marking, of which some at least appear to reflect West-African patterns.

. Koine-formation and convergence Levelling or convergence is a topic that has been much studied in connection with koine-formation in historical linguistics. The first koine was the Greek koine that arose in the dialectally diverse context of Ancient Greek, as a result of wide-ranging political merger. In this volume Siegel and Hinskens discuss the relevance koine-formation might have for theories of creolization. Kouwenberg, however, is very critical of the uncritical use made by many creolists of notions of convergence. After defining the notion koine (and thereby distinguishing it from other language contact varieties) Jeff Siegel discusses the processes involved in koine formation and shows to what extent they may be relevant to pidgin/creole genesis. The three processes are mixing, levelling, and simplification. He sketches the contours of a sociolinguistic model in which basic similarities, differences of degree, as well as essential differences (for example, the role of nativization) between koineization and creole genesis are insightfully ordered

Introduction

relative to each other. He argues that all three processes play a role at different stages of creole genesis. In addition, he argues that in the levelling process substrate reinforcement plays an important role in deciding which of the features will ultimately survive. Frans Hinskens’s contribution, although originally a reaction to it, is more than a thorough review of the main points of Jeff Siegel’s chapter. He discusses Siegel’s findings regarding the relationship between koineization and creole genesis from three angles, viz. conceptual and methodological aspects, the currently occurring processes of levelling and koineization among (traditional) dialects of languages spoken in the Old World, and insights into the approach to koineization in historical linguistics as well as the study of creole languages, language contact and language acquisition. Silvia Kouwenberg critically discusses the notion of convergence which has rapidly gained an enormous popularity among creolists in the last decade. She shows that there is no coherent theory of convergence in historical linguistics, nor is there a unified class of phenomena which go under its label. She concludes that convergence in historical linguistics really characterizes a result, rather than the process that brings about this result, and, as such, is a descriptive rather than explanatory concept. The reverse seems to be true when this concept is applied to creole issues. She distinguishes two senses of the notion in creole studies. It refers either to superficial similarities between languages in contact, or to the similar results of different possible explanations in creole genesis. With respect to the latter sense, she argues that the present desire among creolists to leave all options open and allow for any and all possible factors to contribute to explanations in creole genesis seriously detracts from its validity, and may even mean the return of the Cafetaria Principle. Only in the case when strong hypotheses are available, has the cumulative evidence of converging explanations the potential to strengthen the case for a particular scenario of creole development.

. Contact linguistics Two authors from the field of contact linguistics have contributions here. Thomason approaches the study of the relationships between contact linguistics and creole linguistics from a general comparative perspective, concluding that one kind of language-contact phenomenon is relevant for the study of creolization. Prince approaches matters from the point of view of the contro-





Norval Smith and Tonjes Veenstra

versial case of Yiddish, concluding that the kind of contact phenomenon found in that language has little to do with creolization, or language mixing, as is often claimed. Sarah Thomason discusses the relationship between types of contact situations and processes of crosslinguistic influence. She distinguishes ‘‘borrowing’’ (that is, the incorporation of features by native speakers into their own language from another language) from ‘‘shift-induced interference’’ (that is, the incorporation of features by non-native speakers into the target language from their native language), and argues that the former does not play an important role in pidgin/creole genesis. It does, however, play a decisive role in language mixture (relexification/language intertwining). With respect to shift-induced interference, she argues that the picture is reversed — that is, unimportant in language mixture, but important in pidgin/creole genesis. Ellen Prince argues that the motivation for the vast majority of contact effects in Yiddish is not an imperfect competence in the Germanic system on the part of the speakers, but a particular semantic or pragmatic intent: to exploit the formal possibilities of Yiddish in order to express in it concepts which were expressed in some different but ‘‘analogous’’ way in the languages present in the contact situation. She rejects out of hand the idea that Yiddish represents some kind of mixture of Germanic, Slavic and Semitic. The idea that Yiddish might be a kind of relexified Sorbian also finds no favour in her eyes. Nor is creolization of relevance.

. The sociolinguistic aspects of contact in creolization In recent years the importance of small-scale factors in some cases of language contact, in particular in the type of language contact involved in creolization, has come to be more and more realized. The size of the European plantation colonies was often (still) very small at the moment of the assumed development of creole languages. In this context it is obviously of great importance to subject the kind of social network that existed on the early plantations to as close an examination as possible. As different languages were spoken by the various participants in such networks, the relevance of this type of examination for a book such as this will be obvious. Jacques Arends’ investigation of sociohistorical factors in creole genesis shows that the stereotypical view of plantations as extremely isolated, strictly bistratal micro-societies to be largely incorrect. He discusses two aspects of the

Introduction

social structure of the plantation system in Surinam: (1) the internal social stratification of the plantation community and (2) the external network relations (i.e. contacts outside the plantation) maintained by the slaves. Although it is not (yet) possible to link the sociohistorical evidence he presents directly to purely linguistic developments (cf. Labov 1994), it seems to be possible to make some general inferences about this relationship. The two aspects may have had linguistic consequences which were more or less opposed: an internal social stratification favoring linguistic differentiation, and an external social network system favoring homogenization (cf. also Lefebvre’s chapter). The first factor provides indirect support for the hypothesis that creole continua may have arisen quite early on. The contribution of the second factor to the levelling of creole varieties spoken on different plantations resulted in a more or less homogeneous creole, rather than a number of different creole ‘‘dialects’’, one for each plantation.

References Labov, William. 1994. Principles of Linguistic Change: Internal factors. Oxford: Blackwell. Muysken, Pieter C. 1981. ‘‘Halfway between Quechua and Spanish: The case for relexification’’. In Historicity and Variation in Creole Studies, A. Highfield and A. Valdman (eds.), 52–78. Ann Arbor: Karoma. Sylvain, Suzanne. 1936. Le créole haïtien: Morphologie et syntaxe. Wetteren: Imprimerie de Meester/Port-au-Prince. [Private publication.]



Chapter 1

Relexification in creole genesis and its effects on the development of the creole* Claire Lefebvre

.

Introduction

There is a large consensus of opinion in the literature to the effect that creole languages are mixed languages in that they derive some of their properties from those of their substratum languages and some of their properties from those of the superstratum language (cf. Alleyne 1981; Holm 1988; etc.). Several scholars, however, have noticed that the type of mix we find in creole languages is not random. For example, Adam (1883: 4–7) states that: J’ose avancer . . . que les soi-disant patois de la Guyane et de la Trinidad constituent des dialectes négro-aryens. J’entends par là que les nègres guinéens, transportés dans ces colonies, ont pris au français ses mots, mais qu’ayant conservé dans la mesure du possible, leur phonétique et leur grammaire maternelles . . . Une telle formation est à coup sur hybride . . . La grammaire n’est autre que la grammaire générale des langues de la Guinée.1

Speaking of Haitian creole, Sylvain (1936:178) observes that: Nous sommes en présence d’un français coulé dans le moule de la syntaxe africaine, ou (. . .) d’une langue éwé à vocabulaire français.2

Similarly, in his extensive study of French-based creoles, Goodman (1964) observes, over and over again, that particular lexical items in the creoles have a phonological representation similar to a French word but that they share properties with corresponding lexical items in the African substratum languages. On the basis of data drawn from Ndyuka, Huttar (1971: 684) also remarks that ‘‘the use of morphemes borrowed by a pidgin or a creole language (. . .) from a European language often diverges from the use of the source morpheme in the source language’’ and often corresponds to the use of the corresponding word in the substratum languages. Voorhoeve (1973)



Claire Lefebvre

makes a similar remark on the basis of Sranan and Saramaccan data. Koopman (1986) compares a number of lexical and syntactic properties in Haitian and in a sample of West African languages (Kru and Kwa languages, as well as one Mande and one Gur language). Her conclusion is twofold (Koopman 1986: 246): First, W. African languages share many properties amongst themselves, and secondly, these properties which include both lexical and syntactic properties tend also to be characteristic of Haitian.

Finally, Keesing (1988:1) writes: I had earlier been struck, when I had learned Solomons Pidgin in the 1960s through the medium of Kwaio, an indigenous language I already spoke fluently, that the learning task mainly required learning Pidgin equivalents of Kwaio morphemes. The syntax of Solomons Pidgin was essentially the same as the syntax of Kwaio, . . . there was a virtual morpheme-by-morpheme correspondance between Kwaio and Pidgin.

These observations suggest that creole languages are not formed by an arbitrary mixture of the properties of the languages present at the time they are being created. The pattern that seems to emerge from the observations reported above is the following: while the phonological forms of the lexical entries of a creole come from superstratum expressions, the semantic and syntactic properties of these lexical entries follow the pattern of the substratum languages. This raises the question of what the process which generates such a division of properties is. On the basis of Haitian data involving functional category lexical entries, I argue that the mental process of relexification accounts for this division of properties in a straightforward way (see also Lefebvre 1986, 1992, 1993a, b, 1998b and references therein; Lefebvre and Lumsden 1989, 1994). By its very nature, however, relexification cannot be the only process involved in creole genesis, even in the case of a radical creole such as Haitian. As is pointed out in Lefebvre and Lumsden (1994), relexification applies in creole genesis when the speakers of the substratum languages are targeting the superstratum language; when these speakers start targeting the relexified lexicons, that is, the early creole, they are no longer using relexification to develop the creole. It has been proposed that, when the speakers of the substratum languages start targeting the relexified lexicons, two other processes play a role in the development of the creole: dialect levelling (cf. Lumsden and Lefebvre 1994) and reanalysis (cf. Lefebvre 1984; Lefebvre and

Relexification in Creole genesis

Lumsden 1994). The data resulting from relexification will be shown to feed the processes of dialect levelling and reanalysis. In this chapter, I illustrate how these three processes apply in creole genesis on the basis of Haitian data involving functional category lexical entries. Before doing so, I provide a definition of these three processes.3 . Relexification The mental process of relexification has been shown to play a role in the genesis of mixed languages (cf. e.g. Muysken 1981). This process has also been argued to play a central role in creole genesis (cf. Lefebvre 1986, 1993a, b; Lefebvre and Lumsden 1989, 1994). According to Lefebvre and Lumsden’s (1994) formal definition of the process, relexification builds a new lexicon in the following way. The lexical entries of the lexicons of the substratum languages are copied, and the phonological representations in these copied lexical entries are replaced with phonological representations derived from the phonetic strings of the superstratum language or by null forms. The second step is referred to as relabelling. The choice of the pertinent phonetic string in the superstratum language to relabel a copied lexical entry is based on their use in specific semantic and pragmatic contexts such that, as is advocated in Muysken (1981), the semantics of the superstratum string must have something in common with the semantics of the substratum lexical entry that is being relabelled. In the literature on creole genesis, it has been pointed out that the makers of a creole do not identify the functional category lexical entries (i.e. determiners, complementizers, tense, mood and aspect markers, etc.) of the superstratum language (cf. Lefebvre 1984; Carden and Stewart 1988; Mufwene 1991; Lefebvre and Lumsden 1994; etc.) because of the limited access that they have to the data (cf. e.g. Thomason and Kaufman 1991). In Lefebvre and Lumsden (1994), it is claimed that the functional category lexical entries copied from the substratum languages are relabelled on the basis of phonetic strings of superstratum lexical categories. It is further claimed that when relexifiers do not find any appropriate phonetic string in the superstratum language, that is, a form which is both semantically and distributionally suitable, the copied functional category lexical entry may be assigned a phonologically null string, such that when this lexical entry is used in an utterance, it is not pronounced. The formal representation of the process of relexification provided in Lefebvre and Lumsden (1994) is illustrated schematically in (1).4





Claire Lefebvre

(1) original lexical entry [phonology]i [semantic feature]k [syntactic feature]n

target language [phonetic string]j used in specific semantic and pragmatic contexts creole [phonology]j ¹ or [Ø] [semantic feature]k [syntactic feature]n

(=(1) in Lefebvre and Lumsden 1994)

Lefebvre and Lumsden’s (1994: 10) proposal makes the following prediction concerning word order in creoles: Since relexification is the first step in second language acquisition, the original aim of the relexifiers is to reproduce the phonetic strings of the superstratum language (. . .) Since the relexifiers intend to reproduce the phonetic strings of the superstratum language, the creole word order for lexical categories will be the word order of lexical categories in the superstratum language (. . .). On the other hand, since the relexifiers do not identify the superstratum functional categories, the word order for creole functional categories will be the same as the word order of the substratum categories that they were relexified from.

The data presented in this chapter will be shown to support the above claims. . Dialect levelling The process of dialect levelling has been observed in situations where dialects or languages are in contact (cf. e.g. Trudgill 1986; Siegel, this vol.). In Lumsden and Lefebvre (1994), it is proposed that this process plays a role in the development of creole languages. Since relexification is a mental process, it is necessarily an individual activity. Typically, situations where creoles are created involve several substratum languages (cf. Whinnom 1971). Thus, although relexification from a single superstratum language provides the early creole community with a common vocabulary (cf. Lumsden and Lefebvre 1994), the relexified lexicons from different substratum languages would not be homogeneous in the early creole. As Lumsden and Lefebvre (1994) proposed, some of these differences might be levelled out by the process of dialect levelling.

Relexification in Creole genesis

. Reanalysis Reanalysis is a process through which a particular phonological form associated with one lexical entry comes to be associated with another lexical entry (Lightfoot 1979). This process, sometimes referred to as grammaticalization (cf. e.g. Sankoff 1990; Hopper and Traugott 1993), has been shown to play a role in cases of regular linguistic change. For example, the preposition of in English has been reanalyzed as a case marker (Chomsky 1981). Likewise, according to Kayne’s (1981) analysis, the French forms à and de have a double status as prepositions and as complementizers. Lefebvre and Lumsden (1994) propose that, when speakers target the speech of the creole community, that is, the early creole, reanalysis plays a role in the further development of the creole. They claim, however, that in the early creole this process applies to a lexical entry that has been created through relexification but assigned a phonologically null representation (cf. (1)). As Lefebvre and Lumsden (1994: 13) put it: In the absence of visible phonological signals for a particular functional category, speakers [of the early creole] use periphrastic constructions to clarify information that is not being signalled. Speakers will then copy the phonological form of the key element of the periphrastic construction into the lexical entry of the previously hidden functional category. In this way, reanalysis provides an explicit signal for a creole lexical entry that was generated by relexification but did not acquire a phonological signal through relabelling . . . When reanalysis assigns this lexical entry a phonological signal, so that it becomes explicit in the creole expression, the creole lexical entry is visibly more like that of the substratum language.

In light of this theoretical background, I will now show how these three processes are implemented in the genesis of Haitian creole. I begin with phase 1 when adult native speakers of different substratum languages target the superstratum language and use the mental process of relexification in order to create a new language. Then I illustrate the processes that apply in the second developmental phase of the creole when the speakers target the relexified lexicons, that is, the early creole.

. Phase : Adult native speakers of substratum languages target the superstratum language This section illustrates how relexification applies to functional category lexical entries during the period when native speakers of substratum languages are targeting the superstratum language. Cases where the copied lexical entry is re-





Claire Lefebvre

labelled on the basis of a superstratum phonetic string and cases where the copied lexical entry is assigned a phonologically null form will be discussed in turn. . Copy and relabel on the basis of superstratum phonetic strings .. The [+definite] determiner This section argues that the lexical entry of the Haitian determiner has been created through the process of relexification. The data and analysis reported on in this section are drawn from Lefebvre (1994a), based on a series of papers on the various facets of this Haitian lexical entry (e.g. Lefebvre 1982, 1992, 1996a, 1998a; Lefebvre and Massam 1988). Haitian creole has a postnominal determiner la (with the phonologically conditioned allomorphs a, an, nan and lan), as illustrated in (2). The presence of this determiner indicates that the information conveyed by the noun phrase is part of the shared knowledge of the participants in the conversation (cf. Fournier 1977; Lefebvre 1982; Lefebvre and Massam 1988). The Haitian determiner is not marked for gender. (2) a.

timounn nan child det ‘the child (in question/that we know of)’ b. liv la book det ‘the book (in question/that we know of)’

Haitian

Haitian

In contrast, the French determiner appears before the noun, as shown in (3), and it is specified for gender and number. Le is masculine singular, la is feminine singular, les is plural, and l’ is a phonologically conditioned allomorph. (3) a.

l’ enfant det child ‘the child’ b. le livre det book ‘the book’ c. la table det table ‘the table’

French

Relexification in Creole genesis

d. les livres/tables det books/tables ‘the books/tables’

In contrast with the Haitian determiner, the French determiner does not necessarily identify old or known information. According to Milner (1978: 23), the definite determiner is either anaphoric, identifying an object that already has been mentioned, or cataphoric. In the latter case, ‘‘l’article annonce une relative ou un génitif sans qu’aucune mention antérieure ne soit requise’’. The Haitian determiner cannot appear with nouns that have a generic or mass interpretation, but the French determiner must appear with such nouns (cf. Milner 1978: 25). These facts are illustrated in (4) and (5), respectively. (4) Pen bòn pou lasante. bread good for health ‘Bread is good for one’s health.’ (= (19) in Lefebvre 1994a)

Haitian

(5) *(Le) pain est bon pour la santé (det) bread is good for det health ‘Bread is good for one’s health.’ (= (24) in Lefebvre 1994a)

French

Furthermore, French has a partitive determiner de+la or du (a contracted form of de + le), which appears with mass nouns, as in (6). (6) Jean a mangé du pain. John aux eat de+le bread ‘John ate bread.’ (=(25) in Lefebvre 1994a)

French

According to Haase (1965), this partitive determiner has been attested in French since the fifteenth century. Milner (1978: 24) points out the exceptional character of French with respect to this partitive determiner and notes that in most languages the determiner does not appear in contexts where the French partitive determiner is manifested. Haitian follows the pattern of the majority of languages, as shown in (7), the Haitian counterpart of the French sentence in (6). (7) Jan manje pen. John eat bread ‘John ate bread.’

Haitian

Finally, in Haitian, the head noun and the determiner may be separated by a relative clause, as in (8).





Claire Lefebvre

(8) Mounn [Ø ki pati] a. man op re-pro leave det ‘The man who left.’ (=(20) in Lefebvre 1994a)

Haitian

By contrast, in French, the head noun and the determiner may not be separated by a relative clause, as shown by the ungrammaticality of the sentence in (9). (9) *Le [qui est parti] homme det [who aux leave] man Lit.: ‘The [who left] man’ (=(26) in Lefebvre 1994a)

French

The determiner in Haitian creole and the determiner in French thus have quite different semantic and syntactic properties, which indicates that the properties of the Haitian creole determiner are not derived from French. Moreover, the French determiner does not appear to have been the source of the phonological representation of the Haitian determiner either. The French determiner is often found as part of Haitian simple nouns, as shown in (10). (10) Haitian nouns larivyè ‘river’ lakay ‘home’ listwa ‘history’ latè ‘world’ zwazo ‘bird’ zonnyon ‘onions’

< < < < < <

Corresponding French DPs la rivière ‘the river’ la case ‘the house’ l’histoire ‘history’ la terre ‘the earth’ les oiseaux ‘the birds’ les oignons ‘the onions’

The French partitive determiner illustrated in (6) above is also found as part of Haitian simple nouns, as shown in (11). (11) Haitian nouns dlo ‘water’ < dife ‘fire’ < diri ‘rice’ <

Corresponding French DPs de l’eau ‘water’ du feu ‘fire’ du riz ‘rice’

The data in (10) and (11) further show that the creators of Haitian did not identify the French determiners as independent morphemes and often analyzed them as part of the phonetic strings of the nouns with which they appear. Further support for this claim comes from the fact that Haitian nouns which contain an agglutinated French determiner may occur with the postnominal Haitian determiner, as shown in (12).

Relexification in Creole genesis

(12) a.

larivyè a river det ‘the river (in question)’ (=(29) in Lefebvre 1994a) b. diri a rice det ‘the type of rice (in question)’

Haitian

Haitian

Consequently, like Sylvain (1936) and Fournier (1977), I conclude that the French determiner could not have been the phonetic source of the Haitian determiner. Where, then, do the properties of the Haitian determiner come from? I argue below that, while the semantic and syntactic properties of the Haitian determiner are derived from the substratum languages’ corresponding lexical entry, its phonological representation is derived from the superstratum lexical item là which occurs at the end of constituents. For example, Fongbe, one of the substratum languages of Haitian (cf. Lefebvre 1993a), has a postnominal determiner ´ɔ (with a phonologically determined allomorph ´ɔn), as illustrated in (13). vˇı ɔ´ child det ‘the child (in question/that we know of)’ b. wémà ɔ´ n book det ‘the book (in question/that we know of)’

(13) a.

Fongbe

Fongbe

This determiner, like the Haitian determiner, indicates that the information conveyed by the noun phrase is part of the shared knowledge of the participants in the conversation (cf. Lefebvre 1992). Thus, like the Haitian determiner (cf. (2)), the Fongbe determiner in (13) is obligatorily anaphoric. The Fongbe determiner cannot appear in noun phrases that have a generic or a mass interpretation, as shown in (14), which parallels the Haitian data in (4). (14) Wɔ` xúxú nyɔ´ n nú lànmε` yén. bread good for health ‘Bread is good for one’s health.’ (= (19) in Lefebvre 1994a)

Fongbe

As is the case in Haitian, Fongbe has no partitive determiner and in Fongbe (cf.(15)), as in Haitian (cf. (7)), bare nouns are interpreted as mass nouns.





Claire Lefebvre

(15) Kɔ` kú ù bléì Koku eat bread ‘Koku ate bread.’

Fongbe

In both languages, a determiner that co-occurs with a mass noun yields the interpretation ‘type of x’, as shown below. (16) diri a mɔ` líkú ɔ´ rice det ‘the type of rice (in question)’

Haitian Fongbe

In Fongbe (cf. (17)), as in Haitian (cf. (8)), the head noun and the determiner may be separated by a relative clause. (17) Súnù [é é yì] ɔ´ man op re-pro leave det ‘The man who left.’

Fongbe

Furthermore, in both Haitian and Fongbe, the determiner is required in the syntactic structure of relative (cf. Koopman 1982; Lefebvre 1986), conditional and factive clauses (cf. Lefebvre and Massam 1988; Kinyalolo 1993; Collins 1994). Finally, in both Haitian and Fongbe, the determiner plays a central role in the structure of the clause (cf. Lefebvre 1992, 1996a, 1998a), and for some speakers, it may appear in constructions involving verb doubling phenomena (cf. Lefebvre and Ritter 1993; Lefebvre 1994b). The above data show that the semantic and syntactic properties of the Haitian creole determiner are strikingly similar to those of the Fongbe determiner. Both languages contrast with French in the same way with respect to this functional category lexical entry. It thus seems that substratum languages of the type of Fongbe were the source of the properties of the Haitian determiner. Following the analysis in Lefebvre and Massam (1988) and Lumsden (1989, 1991) for Haitian, and the analysis in Brousseau and Lumsden (1992) for Fongbe, I assume that the forms la and ´ɔ (and their allomorphs) are the head of the functional category phrase DP. These determiners are specified for the feature [+definite] and they take their complement to the left. What, then, is the source of the phonological form of the Haitian creole determiner? In the literature, it has been proposed that the French form là is a good candidate for the phonetic source of the Haitian determiner la (cf. Faine 1937; Goodman 1964; Fournier 1977; Valdman 1978; etc.). In addition

Relexification in Creole genesis

to its use as a locative adverb meaning ‘there’, the French là may appear following the noun in a noun phrase as an emphatic deictic marker, as in (18), or as a simple interjection after a noun phrase or a clause, as in (19). Note that the use of là shown in (19) has been attested to exist in French since at least the seventeenth century (cf. Flûtre 1970: 141, 146; Hull 1975: 2). (18)

Cet homme -là vient d’arriver. dem man [+deic] come arrive ‘This/that man just arrived.’

French

(19) a.

L’ homme là vient d’arriver. Popular French det man int come arrive ‘The man, uh, just arrived.’ b. Un homme là vient d’arriver. Popular French det man int come arrive ‘A man, uh, just arrived.’ c. Un/l’ homme là, qui vient d’arriver là . . . Popular French det man int who come arrive int ‘A/the man, uh, who just arrived, uh. . .’ (= (30) in Lefebvre 1994a)

These examples show that the French form là has a distribution that is similar to the distribution of the Fongbe determiner and, furthermore, the deictic interpretation of this French form (cf. (18)) overlaps with the interpretation of the Fongbe determiner. Thus, the French form là, pronounced [lɔ] or [lɒ] in popular dialects of French, was probably the source of the phonological form of the Haitian determiner la. As is pointed out in Lefebvre (1994a, 1996a) the syntactic and semantic properties of this French form are not the same as those of the Haitian determiner. The French là can appear with definite or indefinite noun phrases (cf. (19a, b)), whereas the Haitian determiner is incompatible with indefinite noun phrases (cf. (4)). Furthermore, when it is used in the context of a clause (cf. (19c)), là has no semantics in common with the Haitian determiner. Thus, the French form là may have been the source of the phonological representation of the Haitian creole determiner but it did not contribute the syntactic and semantic properties of this functional category lexical entry. The creation of the lexical entry for the Haitian determiner thus proceeded as depicted in (20), which should be read in light of the general schema in (1). The copied lexical entry corresponding to the Fongbe determiner ´ɔ was relabelled on the basis of the phonetic matrix of the French form là, yielding the Haitian determiner la.





Claire Lefebvre

(20) Fongbe lexical entry /ɔ´ / [+definite] complement/head

French [lɔ]/[lɒ] used in specific semantic and pragmatic contexts Haitian Creole lexical entry /la/ [+definite] complement/head

.. The [+deictic] terms The [+deictic] terms of Haitian present yet another case of functional category lexical entries relexified by the creators of Haitian. This section summarizes the findings of an extensive analysis of these terms in Lefebvre (1997, 1999 and 2001). In Haitian, there are two demonstrative terms sa and sila (cf., among others, Goodman 1964: 50; Sylvain 1936). In the grammar of the Haitian speakers who provided me with the data discussed below, while sila is used only to point at objects that are far from the speaker, sa is used as a general deictic term to point at objects that are either close to or far from the speaker. So, while sila is specified as [−proximate], sa is not specified for any value of the feature [" proximate]. In the examples below, the meaning of sila is rendered as ‘that’ and the meaning of sa is rendered as ‘this’/‘that’. (Other interpretive patterns will be discussed at the end of this section.) Furthermore, as is pointed out in Valdman (1978), these Haitian demonstrative terms may be used to point at objects which are either animate and inanimate, as illustrated in (21) below; these terms are thus not specified for the feature [" animate]. These [+deictic] determiners appear postnominally, as shown in (21). (21) a.

bag sa // sila ring [+deic] [−prox] ‘this/that ring’ b. mounn sa // sila man [+deic] [−prox] ‘this/that man’ (=(2) in Lefebvre 1997)

Haitian

Haitian

Furthermore, sa/sila may occur within the same nominal phrase as the possessive marker, the [+definite] determiner, and the plural marker. This is shown in (22).

Relexification in Creole genesis

(22) bag mwen sa // sila a yo Haitian ring poss [+deic] [−prox] det pl ‘these/those rings of mine (in question/that we know of)’ (=(3) in Lefebvre 1997)

Since sa and sila are determiners, they are not identified for the categorial features [+N, −V]. The semantic and syntactic properties of the lexical entries sa and sila are presented in (23). (23) a.

/sa/ ‘this’/‘that’ [+deictic] b. /sila/ ‘that’ [+deictic] [−proximate] (=(8) in Lefebvre 1997)

Haitian Haitian

French has a series of demonstrative terms which can be divided into three major groups on the basis of their categorial status: they are either determinative, pronominal, or adverbial. The forms ce(t), cette, ces constitute the first group of French deictic terms. These forms are nominal determiners and they are mutually exclusive with other determiners. They agree in gender and number with the noun they determine, so they bear gender and number features: ce(t) (m. sing.), cette (f. sing.), ces (pl.). As is shown in (24), these forms occur prenominally. They can be used with animate (24a) or inanimate (24b) objects. Therefore, they are not specified for any value of the feature [" animate]. Furthermore, they can be used to point at an object that is either close to or far from the speaker; thus, they are not specified for any value of the feature [" proximate]. (24) a.

ce 1ces2 this/that 1these/those2 b. cette 1ces 2 this/that 1these/those2

garçon(s)

French

boy(s)’ bague(s)

French

ring(s)’ (=(9) in Lefebvre 1997)

The semantic and syntactic properties of the lexical entries ce(t), cette and ces are presented in (25).





Claire Lefebvre

(25) a.

ce(t): /sə/ ~ /sεt/ ‘this/that’ [+deictic] [−fem] [−pl] b. cette: /sεt/ ‘this/that’ [+deictic] [+fem] [−pl] c. ces: /sεt/ ‘these/those’ [+deictic] [+pl] (=(12) in Lefebvre 1997)

French

French

French

The properties of the French lexical entries in (25) differ from those of the Haitian lexical entries in (23). Thus, it is unlikely that the properties of the Haitian terms could be derived from these French determiners. The second class of French demonstrative terms has two sets of pronominal forms, which are distinguishable on the basis of animacy. The three pronominal forms ça, cela, and ceci make up the first set (cf. (26)). Since these forms may only be used with inanimate objects, they must be specified for the feature [−animate]. Ça and cela are general deictic terms which may be used to point at objects that are either close to or far from the speaker; they are not specified for any value of the feature [" proximate]. In contrast, ceci is used only to point at objects close to the speaker. Consequently, it must be specified for the feature [+proximate]. (26) J’ ai vu ça // cela // ceci. I aux see [+deic] // [+deic] // [+prox] ‘I saw this/that // this/that // this.’ (=(14) in Lefebvre 1997)

French

The semantic and syntactic properties of the lexical entries ça, cela and ceci are summarized in (27). (27) a.

ça: /sa/ ‘this/that’ [+deictic] [+N, −V] [−animate] b. cela: /səla/ ‘this/that’ [+deictic] [+N, −V] [−animate]

French

French

Relexification in Creole genesis

c.

ceci: /səsi/ ‘this’ [+deictic] [+N, −V] [−animate] [+proximate] (=(17) in Lefebvre 1997)

French

The properties of the French forms in (27) are very different from those of the Haitian forms in (23). The French forms are pronominal and are specified as [−animate], whereas the Haitian forms are not pronominal and are not specified for any value of the feature [" animate]. Furthermore, while the paradigm of French pronouns in (27) has a form which lexically encodes the feature [+proximate], the paradigm of Haitian forms in (23) has a form which lexically encodes the feature [−proximate]. Thus, ça, ceci and cela are unlikely to be the source of the semantic and syntactic properties of the Haitian demonstrative terms. The forms celui, ceux and celle(s) constitute the second set of demonstrative pronouns in French. These pronouns are used to point at objects that are either animate or inanimate, and therefore, they are not specified for any value of the feature [" animate]. These forms are specified for gender and number: celui (m. sing.), ceux (m. pl.), celle(s) (f. sing./pl.). These demonstrative pronouns are neutral with respect to the feature [" proximate]. The semantic and syntactic properties of the lexical entries celui, ceux, and celle(s) are shown in (28). (28) a.

celui: /səlɥi/ ‘this/that’ [+deictic] [+N, −V] [−fem] [−pl] b. ceux: /sø/ ‘these/those’ [+deictic] [+N, −V] [−fem] [+pl] c. celle(s): /sεl/ ‘this/that; these/those’ [+deictic] [+N, −V] [+fem] (=(20) in Lefebvre 1997)

French

French

French





Claire Lefebvre

The French forms in (28) share only one property with the Haitian forms in (23): none are specified for a value of the feature [" animate]. Aside from this, the other properties of the French forms in (28) contrast with those of the Haitian forms in (23). Thus, it seems that the semantic and syntactic properties of the Haitian deictic terms sa and sila are not derived from those of the French forms celui, ceux, and celle(s). Finally, there is a third group of deictic terms in French: the adverbials là and ci. These two forms combine with the deictic pronominal forms of the second group, as in (29), or with the deictic determiners occurring in nominal structures, as in (30). (29) a.

J’ ai vu celui- ci. I aux see [+deic] [+prox] ‘I saw this one.’ b. J’ ai vu celui- là. I aux see [+deic] [+deic] ‘I saw this/that one.’ (=(21) in Lefebvre 1997) (30) a. cette bague-ci [+deic] ring [+prox] ‘this ring’ b. cette bague-là [+deic] ring [+deic] ‘this/that ring’ (=(22) in Lefebvre 1997)

French

French

As shown in the above examples, là is a general deictic form, not specified for any value of the feature [" proximate]. In contrast, ci is used to point at objects that are close to the speaker; it is specified for the feature [+proximate]. The properties of the lexical entries là and ci are represented in (31). (31) a.

là: /lɔ/ ~ /lɑ/ ‘there/here’ [+deictic] adv b. ci: /si/ ‘here’ [+deictic] [+proximate] adv

French

French

(=(23) in Lefebvre 1997)

Like the Haitian demonstrative terms, these two French lexical items occur postnominally (cf. (29) and (30)). They differ, however, with respect to categorial features. Furthermore, in French, the positive value of the feature [" proximate]

Relexification in Creole genesis

is lexically encoded, whereas in Haitian the negative value of the feature [" proximate] is lexically encoded. Thus, it appears that these lexical items are not the source of the properties of the Haitian deictic terms in (23). How, then, did Haitian end up with the system described in (23)? The properties of demonstrative terms in the substratum languages provide a clear answer to this question: the properties of the creole lexical entries involving demonstrative terms are derived from those of corresponding lexical items in the substratum languages. For example, in Fongbe, as in Haitian, there are two demonstrative terms: élɔ´ and énε´ (cf. Anonymous 1983). For the Fongbe speakers whose grammar is discussed in this section, while énε´ is used only to point at objects that are far from the speaker, élɔ´ is used as a general deictic term for objects that are either close to or far from the speaker. So, while the first form is specified as [−proximate], the second form is not specified for any value of the feature [" proximate]. In the examples below, the meaning of énε´ is rendered as ‘that’ and the meaning of élɔ´ is rendered as ‘this/that’. (Other interpretive patterns will be discussed at the end of this section.) Furthermore, the Fongbe demonstrative terms may be used with objects which are either animate or inanimate. Thus, they are not specified for the feature [" animate]. As is the case in Haitian (cf. (21)), the Fongbe demonstrative terms occur postnominally, as shown in (32). (32) a.

àlɔ` kε´ élɔ´ // énε´ ring [+deic] // [−prox] ‘this/that // that ring’ b. súnù élɔ´ // énε´ man [+deic] // [−prox] ‘this/that // that man’ (=(24) in Lefebvre 1997)

Fongbe

Fongbe

The [+deictic] determiners may occur within the same nominal phrase as the possessive marker, the [+definite] determiner and the plural marker. This is shown in (33), which parallels the Haitian data in (22). (33) àlɔ` kε´ ce élɔ´ // énε´ ɔ´ lεˆ Fongbe ring poss [+deic] [−prox] det pl ‘these/those rings of mine (in question/that we know of)’ (=(25) in Lefebvre 1997)

Since élɔ´ and énε´ occur as part of the determiner system of nouns, they are not specified for the categorial features [+N, −V]. The semantic and syntactic properties of the lexical entries élɔ´ and énε´ are presented in (34).





Claire Lefebvre

(34) a.

/élɔ´ / ‘this/that’ [+deictic] b. /énε´ / ‘that’ [+deictic] [−proximate]

Fongbe Fongbe (=(30) in Lefebvre 1997)

When we compare the Fongbe and Haitian data, we find a systematic parallel between the properties of the demonstrative lexical entries in the two languages. In both languages, there is a paradigm of demonstrative terms with two forms, both of which are determiners (cf. (22) and (33)). The distribution of the two sets of forms is parallel. In both languages, the terms may be used for either animate or inanimate objects, and in both languages, one term is specified [−proximate] while the other is unspecified for the feature [" proximate]. Sila in Haitian and énε´ in Fongbe are both [−proximate], while sa and élɔ´ are [" proximate] (cf. (23) and (34)). In contrast with French, both languages lack a form that lexically encodes the positive value of the feature [" proximate]. The lexical entries of the demonstrative terms for the two languages are compared in (35). (35)

Haitian a. /sa/ [+deictic] b. /sila/ [+deictic] [−proximate]

Fongbe /élɔ´ / ‘this/that’ [+deictic] /énε´ / ‘that’ [+deictic] [−proximate] (=(31) in Lefebvre 1997)

As can be seen in (35), the semantic and syntactic properties of the Haitian and Fongbe deictic terms are parallel; the Haitian lexical entries are different from the Fongbe ones only in their phonological representations. This is exactly what is expected under the relexification hypothesis. How did the creators of Haitian establish a phonological form for the copied lexical entries? In Lefebvre (1997) I argue extensively that the phonological form of the copied lexical entries was established on the basis of the French form ça, pronounced [sa], yielding sa in Haitian creole and cela, pronounced [səla] or [slɑ], yielding sila in Haitian creole. Alternatively, celui-là might also have yielded sila in Haitian creole. The creation of the lexical entries for Haitian demonstrative terms thus proceeded as depicted in (36), which should be read in light of the general schema in (1). The copied lexical entry corresponding to the general deictic term élɔ´ was relabelled on the basis of the phonetic matrix of the French

Relexification in Creole genesis

general deictic pronoun ça [sa], yielding the Haitian general deictic term sa. Similarly, the copied lexical entry corresponding to the Fongbe [−proximate] pronoun énε´ was relabelled on the basis of the phonetic forms cela [səla] or celui-là [sɥila], yielding the Haitian [−proximate] demonstrative pronoun sila. (36) a.

Fongbe lexical entry /élɔ´ / [+deictic]

French [sa] used in specific semantic and pragmatic contexts Haitian Creole lexical entry /sa/ [+deictic]

b. Fongbe lexical entry /énε´ / [+deictic] [−proximate]

French [səla]/[sɥila] used in specific semantic and pragmatic contexts Haitian creole lexical entry /sila/ [+deictic] [−proximate] (=(32) in Lefebvre 1997)

I would like to end this section by making more precise the interpretive patterns associated with the two Haitian and Fongbe deictic terms. The interpretive pattern discussed in the examples above and summarized in (35) is taken from Lefebvre (1997). The validity of the interpretation of Haitian sila in (35) has been challenged by DeGraff (1999) who claims that, like sa, sila is a general deictic term. In Lefebvre (1999), I show that the interpretive pattern I report in Lefebvre (1997) is also found in the literature (e.g. Sylvain 1936; Étienne 1974). Furthermore, I show that the pattern described in (35) (from Lefebvre 1997) is one of three interpretive patterns. Prior to Lefebvre (1997), two interpretive patterns had been reported for sa and sila: one where there is no distal distinction between the two terms (e.g., Férère 1974: 103; Joseph 1988: 112; Valdman 1978: 194; Valdman et al., 1981; see also DeGraff 1999), and one where there is a distal distinction between them. In his comparative grammar of French based creoles, Goodman (1964: 51) writes: ‘‘In Haiti [. . .], these two meanings [‘this’ and ‘that’] are distinguished formally, sa ‘this’ and sila ‘that’.’’ (See also Hall 1953: 29).





Claire Lefebvre

Tinelli (1970: 28) points out two interpretive patterns: ‘‘sa is sometimes replaced by sila, with no syntactic change [. . .], and no clear modification of meaning, except [my emphasis] for some speakers who distinguish the remote deictic sila from the proximate sa [. . .].’’ Lefebvre (1997) reports a third pattern where sa is a general deictic term and sila is [−proximate]. These three patterns are schematically represented below. interpretive pattern a interpretive pattern b interpretive pattern c

sa [" proximate] [+ proximate] [" proximate]

sila [" proximate] [−proximate] [−proximate]

Note that the third pattern appears to be a compromise between the first two: sa is [" proximate], as in the first pattern, and sila is [−proximate], as in the second one. In Lefebvre (in press) I show that the three interpretive patterns found for Haitian sa and sila are also found for Fongbe élɔ´ and énε´, respectively, thus providing support to the relexification account of the Haitian deictic terms. .. Conclusion The derivation of the Haitian [+definite] determiner and [+deictic] determiners discussed in this section illustrates how relexification proceeds in the case of functional categories when the speakers of the substratum languages are targeting the superstratum language. The data further show that relexifiers use lexical items of the superstratum language in order to relabel copied lexical entries. Other cases of functional items have been argued to have a similar derivation. A detailed discussion of these cases can be found in Lefebvre (1993b, 1994a, 1996b, 1998b). . Copy and relabel by a phonologically null form The following facts involving the realization of genitive Case illustrate the situation where a copied lexical entry has been assigned a phonologically null form. The data and analysis are drawn from Lefebvre and Lumsden (1992). As is extensively argued in Brousseau and Lumsden (1992), Fongbe has a postnominal genitive Case marker tɔ´n, shown in (37). (37) távò Kɔ` kú tɔ´ n ɔ´ table Koku gen det ‘Koku’s table’

Fongbe (= (9) in Lefebvre and Lumsden 1992)

Relexification in Creole genesis

In French noun phrases, Case is realized as de or à (cf. Kayne 1981). The realization of Case precedes the noun, as shown in (38). (38) la table de/à Jean det table case Jean ‘The table of John’

French (= (10) in Lefebvre and Lumsden 1992)

Finally, central and southern Haitian manifest no overt Case marker in the context of a nominal argument, as shown in (39). (39) tab Jan Ø an table Jan gen det ‘John’s table’

Haitian (= (11) in Lefebvre and Lumsden 1992)

Lefebvre and Lumsden (1992) account for the apparent difference between Haitian and both of its contributing languages as follows. They hypothesize that the lexical entry associated with Fongbe tɔ´n has been copied but not relabelled. Since Case is required by universal grammar (cf. Chomsky 1981; Travis and Lamontagne 1992), Lefebvre and Lumsden (1992) assume that there is indeed a functional category realizing Case in the Haitian noun phrase in (39) and that this category is phonologically null. On this account, then, speakers of a language like Fongbe who were relexifying their lexicon on the basis of data from French would have relabelled the copied lexical entry of their genitive Case marker with a phonologically null string. Crucially, the Haitian phrase headed by the null form in (39) is argued to have the properties of a Genitive Phrase, as in (37), rather than those of an Objective one, as in (38) (see Lumsden 1996). . Conclusion The data discussed in this section show that the creators of Haitian did not identify as such the functional categories of the superstratum language. Instead, the creators of Haitian used the properties of their own functional category lexical entries to create the creole lexical entries. This was achieved by the process of relexification whereby the functional category lexical entries were copied, and relabelled either on the basis of superstratum phonetic strings or by a phonologically null form.5





Claire Lefebvre

. Phase : The creators of the creole target the relexified lexicons The second developmental phase of the creole starts when the creators of the creole target the relexified lexicons, that is, the early creole. This situation gives rise to two other processes: dialect levelling (cf. Lumsden and Lefebvre 1994) and reanalysis (cf. Lefebvre and Lumsden 1994). These processes will be discussed on the basis of data involving Haitian functional category lexical entries. . Dialect levelling The process of dialect levelling is illustrated below on the basis of data involving the Haitian plural marker. The data are drawn from Lefebvre (1994a). As shown in Lefebvre (1994a), the Haitian plural marker yo shares a number of properties with the Fongbe plural marker lεˆ . For one thing, both plural markers occur postnominally, as shown in (40). (40) krab yo àsɔ´ n lεˆ crab pl ‘The crabs’ *‘(Some) crabs’ (=(31) in Lefebvre 1994a)

Haitian Fongbe

A noun followed by the plural marker alone, is always interpreted as [+definite]. As shown in (40), it cannot be interpreted as [−definite]. Indefinite plural is not signalled, as shown in (41). (41) M’ achte krab. N’ xɔ àsɔ´ n. I buy crab ‘I bought (some) crabs.’

Haitian Fongbe (=(32) in Lefebvre 1994a)

In both Haitian and Fongbe, the plural marker may occur within the same noun phrase as the determiner. In this case, at S-structure, the plural marker follows the determiner. (42) krab la yo/*yo la àsɔ´ n ɔ´ lεˆ /*lεˆ õ crab det pl ‘the crabs (in question)’

Haitian Fongbe (=(33) in Lefebvre 1994a)

Relexification in Creole genesis

As shown in (43), in these languages, the singular is not indicated by a specific marker. (43) krab la Ø àsɔ´ n ɔ´ Ø crab det num ‘the crab (in question)’ (=(34) in Lefebvre 1994a)

Haitian Fongbe

The properties of the Haitian plural marker discussed so far are similar to those of the corresponding marker in Fongbe. These properties contrast systematically with the expression of number in French. First of all, in French, the feature [±plural] is an obligatory feature of the prenominal determiners and is expressed as part of the determiner. While le and la are the singular forms of the definite determiner, les is the plural form of the determiner. (44) a.

Jean a mangé la pomme. John aux eat the apple ‘John ate the apple.’ b. Jean a mangé les pommes. ‘John ate the apples.’

French

French (=(36) in Lefebvre 1994a)

Second, in contrast with Haitian, in French, the expression of plural is obligatory even when a noun is indefinite. This is shown in (45) where des is the contracted form of de + les (cf. Milner 1978, for a discussion of the French partitive de). (45) Jean a mangé des pommes. John aux eat de+les apple ‘John ate (some of the) apples.’

French (=(37) in Lefebvre 1994a)

Thus, so far there is a systematic parallel between Haitian yo and Fongbe lεˆ , which both contrast with the expression of the category number in French. There is a difference, however, between Haitian and Fongbe, in that the Haitian plural marker is also used as a third person plural personal pronoun, as in (46), while the Fongbe plural marker does not fulfill this function. In Fongbe, the third person plural personal pronoun is rendered by a different lexical item, as shown in (47). (46) a.

krab yo crab pl ‘the crabs’

Haitian





Claire Lefebvre

b. Yo pati. 3pl leave ‘They left.’ (47) a.

àsɔ´ n lεˆ crab pl ‘the crabs’ b. Yˇe yì. 3pl leave ‘They left.’

Haitian

Fongbe

Fongbe

How can the difference between the two languages be accounted for within the framework of the relexification hypothesis? Sylvain (1936) and Goodman (1964) have suggested that the use of yo as a plural marker is in fact an extended usage of the third person plural pronoun. In their view, the phonological representation of yo is derived from the strong form of the third person personal pronoun eux. The pronoun eux in French occurs as an emphatic form before a clitic, as in (48). (48) Eux, ils mangent du riz chaque jour. them, they eat part rice every day ‘Them, they eat rice every day.’

French (=(40) in Lefebvre 1994a)

Furthermore, in colloquial French, the third person pronoun eux may also occur as an emphatic pronoun at the end of a nominal constituent, as shown in (49) (A. M. Brousseau, p.c.). Note the neutralization in gender shown in (49b), also noted by Gougenheim (1973). (49) a.

Les gars, eux, ils . . . French det guys them, they . . . ‘The guys, them, they . . .’ b. Les filles, eux, ils . . . Popular french det girl them they . . . ‘The girls, them, they. . .’ (=(41) in Lefebvre 1994a)

The French pronoun eux thus has a similar surface distribution to the Haitian form yo. It is clear, however, that it shares only a few semantic features with yo since, in French, eux is only pronominal; unlike Haitian yo, it is never used as a plural determiner. These facts suggest that, if the Haitian form yo derives its phonological representation from the phonetic matrix of the French form eux, it does not derive its semantic properties from that form. How did Haitian yo

Relexification in Creole genesis

come to be used both as a third person plural pronoun and as a plural marker, while the corresponding form in French is only used as a pronoun and the corresponding form in Fongbe is only used as a plural marker? Interestingly enough, several West African languages present cases where the third person plural pronoun also serve as a plural marker. For example, in Ewe, woˇ is the form for both the third person plural pronoun and the plural marker (cf. Westerman 1930: 45, 57). According to Goodman (1964: 46, 47), Yoruba, Mandingo, and other languages also show this double function of a single morpheme. The proposal in Ritter (1992) provides a theoretically motivated account of this fact. She argues that, while first and second person pronouns are of the category D(eterminer), third person pronouns are of the category Num(ber). According to her analysis, third person pronouns are derived by raising Num to D, as is schematically represented in (50).

(50) First/second person pronoun

Third person pronoun

DP D

DP NumP Num ti

D [+de²nite]

yoi

On this account then, both the plural marker and the third person plural pronoun are of the category Num. Whereas in Haitian creole, Yoruba, Mandingo, etc., there is only one lexical entry fulfilling these two functions, in Fongbe the category Num is realized by two different morphemes: lεˆ in the context of a noun phrase and yeˇ elsewhere. It thus appears that, in this case, Haitian follows the pattern of substratum languages other than Fongbe. Lumsden and Lefebvre (1994) propose the following account of the historical derivation of Haitian yo within the framework of the relexification hypothesis. They assume that, in languages of the type of Ewe, there is only one lexical entry woˇ whereas, in languages like Fongbe, there are two pertinent lexical entries (i.e. yeˇ and lεˆ ). In their view, the relexification hypothesis predicts that there would be at least two basic dialects in the early creole with respect to these lexical entries. Lumsden and Lefebvre (1994) hypothesize that speakers of a language like Fongbe looked for two morphemes in the superstratum data, one to relexify the plural marker and the other to relexify the third person plural pronoun. They further hypothesize that speakers of a





Claire Lefebvre

language like Ewe looked for only one form, since their own lexicon provided a single lexical entry fulfilling both of these purposes. Finally, Lumsden and Lefebvre propose that the differences between the two types of grammar in the early creole were reconciled through the process of dialect levelling. In this particular case, the early creole dialect that came to dominate was the one that used a single morpheme in both contexts. This analysis accounts for the properties manifested by yo in modern Haitian creole. The above discussion shows how the process of dialect levelling operates on the basis of the early creole lexicons relexified from various substratum languages. The role of this process in creole genesis accounts for the fact that the properties of some specific lexical entries of the creole may depart from those of corresponding lexical entries in the individual substratum languages. . Reanalyis In this section, I illustrate how the process of reanalysis operates in the context of the development of a creole on the basis of data pertaining to the tense, mood and aspect (TMA) system of Haitian. The data are drawn from an extensive analysis of the TMA system of Haitian, as compared to those of its source languages (cf. Lefevre 1996b). The pertinent data involve the Haitian preverbal marker ap. It is a well-known fact that a Haitian sentence containing the preverbal marker ap is ambiguous between an imperfective and a future reading (cf. Damoiseau 1988; Déchaine 1991; Lumsden 1996). This is shown in (51). (51) Mari ap sòti. Mary go-out ‘Mary is going out.’/‘Mary will go out.’

Haitian

In Lefebvre (1996b), it is argued that there are two lexical entries with the phonological representation ap in the Haitian lexicon: one encoding imperfective and one encoding future. The first argument supporting this claim comes from data drawn from a subset of speakers who allow two aps within a single clause, as shown in (52). (52) M’ ap ap sòti. I def-fut imp go-out ‘I will be going out.’

Haitian (=(20) in Lefebvre 1996b)

Relexification in Creole genesis

A second argument is that speakers who do not accept two co-occurring aps still have the pertinent interpretations. For example, for this second group of speakers, a sentence containing ap such as M ap sòti may be assigned three interpretations: (a) ‘I am going out’, where ap is assigned an imperfective reading; (b) ‘I will go out’ where ap is assigned a future interpretation; or (c) ‘I will be going out’. This latter interpretation shows that the second group of speakers (those who do not pronounce two aps in a row6) still have the interpretation corresponding to (52), which does contain two aps. Thus, the two groups of speakers present similar interpretive data regardless of whether they allow the co-occurrence of two aps at surface structure. These sets of facts argue that there are two lexical entries for ap in the Haitian lexicon. In Lefebvre (1996b), it is argued that the lexical entry of the imperfective preverbal marker ap has been created by relexification. On this analysis, the phonological form of this preverbal marker was established by relabelling on the basis of the French preposition après, which occurs in the French periphrastic progressive, as in Marie est après manger ‘Mary is eating’. The lexical entry of the future marker ap is also argued to have been created through relexification but, in this case, it acquired its phonological representation through the process of reanalysis. In the remainder of this section, I summarize the arguments supporting this claim. I begin by showing that the properties of the definite future marker ap in Haitian systematically parallel those of the definite future marker ná in Fongbe. First, the definite future markers ap and ná both situate, with respect to the moment of speech, an event that is expected to definitely take place in the near future. The event may coincide with the point of reference, yielding a definite future interpretation with verbs of all aspectual classes, as shown in (53)–(55). (53) Dynamic verbs Mari ap prepare pat. Mari ná a wɔˇ . Mary def-fut prepare dough ‘Mary will prepare dough.’ (=(119) in Lefebvre 1996b) (54) Resultative verbs Mari ap wè Jan. Mari ná mɔ Jan. Mary def-fut see John ‘Mary will see John.’ (=(120) in Lefebvre 1996b)

Haitian Fongbe

Haitian Fongbe





Claire Lefebvre

(55) Stative verbs Mari ap kònnèn Jan. Mari ná tu n Jan. Mary def-fut know John ‘Mary will know John.’

Haitian Fongbe (=(121) in Lefebvre 1996b)

Or the event may precede the point of reference, yielding a future perfect interpretation, as in (56). (56) Mari ap deja prepare pat. Mari ná ko a wɔˇ . Mary def-fut already prepare dough ‘Mary will have prepared dough.’

Haitian Fongbe (=(122) in Lefebvre 1996b)

Interestingly enough, in both languages, a clause containing both the anteriority marker and the definite future marker is interpreted as conditional; this holds for all three aspectual classes of verbs. An example containing a dynamic verb is given in (57). (57) Mari te ap prepare pat. Mari ko ná a wɔˇ . Mary ant def-fut prepare dough ‘Mary would prepare dough.’ ‘Mary would have prepared dough.’

Haitian Fongbe

(=(123) in Lefebvre 1996b)

These facts show that the definite future markers in Haitian and Fongbe have similar morphological and semantic properties (cf. Lefebvre 1996b, for a more detailed discussion). According to the relexification hypothesis, the properties of ap parallel those of ná because they have been copied from the original lexical entry. How did the form ap become the phonological representation of the copied lexical entry? Suppose that ap comes from the French adverb après ‘after’ occurring in sentence-initial position (e.g. Après avoir mangé, nous partirons. ‘After eating, we will leave’). But ap could not have become the phonological representation of the copied lexical entry through the process of relabelling. Although Fongbe ná shares some elements of meaning with the French preposition of posteriority après, the two lexical items do not occur in the same surface position. Whereas ná occurs between the subject and the verb, après occurs in sentenceinitial position. In Lefebvre (1996b), it is argued that ap has become the phonological representation of the copied lexical entry through reanalysis of

Relexification in Creole genesis

the Haitian adverb of posteriority apre (< French après ‘after’), which occurs in sentence-initial position as shown in (58). (58) Apre yo fin wè-l yo rakònte . . . After they finish see-him they told ‘After they had seen him, they told . . .’ (from Hall 1953: 221)

Haitian

The following scenario is hypothesized within the framework, outlined in the introduction, defining how reanalysis applies in creole genesis. The creators of Haitian did not find any phonetic string in the superstratum language to relabel the copied lexical entry corresponding to ná. In phase 1, that is during the period where they were relexifying their own lexicons, the relexifiers thus assigned a null phonological form to the copied lexical entry (cf. (1)). They used the sentence-initial preposition of posteriority apre ‘after’ to clarify the information which was not phonologically signalled. In phase 2, that is when the agents of Haitian started targeting the relexified lexicons, the form apre eventually made its way to the position between the subject and the verb. The creators of the creole then copied the phonological form of this adverb of posteriority onto the lexical entry of the previously phonologically null functional category, the copied lexical entry corresponding to ná in Fongbe. In this way, reanalysis provided an explicit signal for a creole lexical entry that was generated by relexification but was not assigned a phonological signal in relabelling. When this lexical entry was assigned a phonological form, it had the properties of the corresponding Fongbe lexical entry.

. Conclusion On the basis of Haitian functional category lexical entries, I have illustrated how relexification, dialect levelling and reanalysis apply in the genesis and the development of creole languages. It has been shown that the data produced by the mental process of relexification feed the two other processes involved in the further development of the creole: dialect levelling and reanalysis. The hypothesis that relexification plays a major role in creole genesis (cf. Lefebvre and Lumsden 1994) provides a direct account of the division of properties observed in the creole lexical entries. Creole lexical entries are predicted to have the same semantic and syntactic properties as the correponding lexical entries in the substratum languages, but with phonological representations taken from phonetic strings of the superstratum language. The relexification hypothesis





Claire Lefebvre

thus explains why creoles reflect the properties of both their superstratum and their substratum source languages in the way they do (cf. Adam, 1883; Sylvain 1936; Goodman 1964; Koopman 1986; Huttar 1971; Keesing 1988, etc.) even after several years of independent evolution. (For further discussion of the issues discussed in this chapter see Lefebvre 1998b.)

Notes * This chapter was written as part of the UQAM Haitian project financed by the SSHRC and FCAR. I would like to thank Anne-Marie Brousseau, Kinyalolo Kasangati, Paul Law, John Lumsden, Elizabeth Ritter, Pierrette Thibault and Raffaella Zanuttini for their very useful comments on the work which led up to this chapter. I would also like to thank the various speakers of Haitian and Fongbe who provided me with the data presented here. Finally, many thanks to Andrée Bélanger for her help in formating the manuscript. . Quote from Adam: ‘‘I go so far as to claim . . . that the so-called patois of Guyana and Trinidad constitute Negro-Aryan dialects. By that I mean that the Guinean Negroes who were transported to the colonies adopted the words of French but, as much as possible, kept the phonetics and grammar of their mother tongues . . . Such a formation is clearly hybrid . . . The grammar is no different from the general grammar of the languages of Guinea.’’ . Quote from Sylvain: ‘‘We are in the presence of a French that has been cast in the mould of African syntax or . . . of an Ewe language with a French vocabulary.’’ . Of course, I assume that, as is the case in other languages, innovation also plays a role in creole genesis. This topic is discussed in Lefebvre (1998b). . The formal representation of the process of relexification in (1) is slightly different from that in Muysken (1981). The motivation for these differences will be discussed elsewhere. . Lefebvre and Lumsden (1992; 1994) also propose that a functional category lexical entry that has been assigned a phonologically null form may be signalled by an overt form in the specifier position of the projection headed by the null form. Due to space limitations, this possibility will not be discussed here. . For an account of this fact, see Lefebvre (1996b).

References Adam, Lucien. 1883. Les Idiomes négro-aryen et maléo-aryen. Paris. Alleyne, Mervyn C. 1981. Comparative Afro-American: An historical-comparative study of English-based Afro-American dialects of the New World. Ann Arbor, MI: Karoma. Anonymous. 1983. Éléments de recherche sur la langue Fon. Cotonou.

Relexification in Creole genesis

Brousseau, Anne-Marie and John S. Lumsden. 1992. ‘‘Nominal structure in Fongbe’’. In Topics on the Syntax and Semantics of Fongbe, C. Lefebvre (ed.), 5–26. (Special issue of the Journal of West African Languages, 22.1) Carden, Guy and William A. Stewart. 1988. ‘‘Binding theory, bioprogram, and creolization: Evidence from Haitian Creole’’. Journal of Pidgin and Creole Languages 3.1: 1–68. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris. Collins, Chris. 1994. ‘‘The Factive Construction in Kwa’’. Travaux de recherche sur le créole haïtien 23: 31–65. UQAM. Damoiseau, Robert. 1988. ‘‘Éléments pour une classification des verbaux en créole haïtien’’. Études créoles 11.1: 41–64. Déchaine, Rose-Marie. 1991. ‘‘Bare sentences’’. In Proceedings of SALT I, S. Moore and A. Wyner (eds.), 31–50. (Cornell University Working Papers in Linguistics, 10.) DeGraff, Michel. 1999. ‘‘Empirical quicksand: Probing two recent articles on Haitian Creole’’. Journal of Pidgin and Creole Languages 14: 359–377. Étienne, G. 1974. Le créole du nord d’Haïti: Étude des niveaux de structure. Ph.D. dissertation, Université des sciences humaines de Strasbourg. Faine, Jules. 1937. Philologie créole: Études historiques et étymologiques sur la langue créole d’Haïti. Port-au-Prince: Imprimerie de l’État. Férère, Gérard A. 1974. Haitian creole sound-system, form-classes, texts. Ph.D. dissertation, University of Pennsylvania. Flûtre, Louis-Ferdinand. 1970. Le moyen picard, d’après les textes littéraires du temps (1560–1660): Textes, lexique, grammaire. Paris: Presses du Palais-Royal. Fournier, Robert. 1977. N ap fè yun ti-kose su la (La grammaire de la particule la en créole haïtien). Master’s thesis, Université du Québec à Montréal. Goodman, Morris F. 1964. A Comparative Study of Creole French Dialects. The Hague: Mouton. Gougenheim, Georges. 1973. Grammaire de la langue française au seizième siècle. Paris: A. et J. Picard. Haase, A. 1965. Syntaxe française du XVIIe siècle. Paris: Delagrave, 5e édition. Hall, R. A. 1953. Haitian Creole: Grammar, Texts and Vocabulary. The American Anthropological Association. Memoir 74, Menasha Georges Banta Publishing Company. Holm, John. 1988. Pidgins and Creoles. (Cambridge Language Surveys, 1). Cambridge: Cambridge University Press. Hopper, P. J. and E. C. Traugott. 1993. Grammaticalization. (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. Hull, A. 1975. On the Origin and Chronology of the French-Based Creoles. Paper presented at the International Conference on Pidgins and Creoles, University of Hawaii. Huttar, Georges L. 1971. ‘‘Sources of creole semantic structures’’. Language 51.3: 684–95. Joseph, F. 1988. La détermination nominale en créole haïtien. Ph.D. dissertation, Université de Paris VII. Kayne, Richard S. 1981. ‘‘On certain differences between French and English’’. Linguistic Inquiry 12.3: 349–371. Keesing, Roger M. 1988. Melanesian Pidgin and the Oceanic Substrate. Stanford, CA: Stanford University Press.





Claire Lefebvre

Kinyalolo, Kasangati K. W. 1993. ‘‘Conditional and Factives in Fɔn’’. In Université du Québec à Montréal. Research report 1992–1993, vol. iv: Études syntaxiques, C. Lefebvre and J. Lumsden (eds.), 161–173. Koopman, Hilda. 1982. ‘‘Les constructions relatives’’. In Syntaxe de l’haïtien, C. Lefebvre, H. Magloire-Holly and N. Piou (eds.), 167–203. Ann Arbor, MI: Karoma. Koopman, Hilda. 1986. ‘‘The genesis of Haitian: Implications of a comparison of some features of the syntax of Haitian, French and West African Languages’’. In Substrata Versus Universals in Creole Genesis, P. Muysken and N. Smith (eds.), 231–258. Amsterdam and Philadelphia: John Benjamins. Lefebvre, Claire. 1982. ‘‘L’expansion d’une catégorie grammaticale: le déterminant la’’. In Syntaxe de l’haïtien, C. Lefebvre, H. Magloire-Holly and N. Piou (eds.), 21–63. Ann Arbor, MI: Karoma. Lefebvre, Claire. 1984. ‘‘Grammaire en contact: définition et perspectives de recherche’’. Revue québécoise de linguistique 14.1: 11–49. Lefebvre, Claire. 1986. ‘‘Relexification in Creole genesis revisited: The case of Haitian Creole’’. In Substrata Versus Universals in Creole Genesis, P. Muysken and N. Smith (eds.), 279–301. Amsterdam and Philadelphia: John Benjamins. Lefebvre, Claire. 1992. ‘‘AGR in languages without person and number agreement: The case of the clausal determiner in Haitian and Fon’’. In Functional Categories, C. Lefebvre, J. S. Lumsden and L. Travis (eds.), 137–156. (Special issue of The Canadian Journal of Linguistics 37.2). Lefebvre, Claire. 1993a. ‘‘The role of relexification and syntactic reanalysis in Haitian Creole: Methodological aspects of a research program’’. In Africanisms in AfroAmerican Language Varieties, S. S. Mufwene (ed.), 254–279. Athens, GA: The University of Georgia Press. Lefebvre, Claire. 1993b. ‘‘The role of relexification in Creole genesis: The case of functional categories’’. Second International Symposium on Bilingualism: Grammatical aspects of bilingualism. Hamburg. Lefebvre, Claire. 1994a. ‘‘Functional categories and relexification’’. Université du Québec à Montréal. Paper read at the MIT Symposium on the Role of Relexification in Creole Genesis: The Case of Haitian Creole. Research report ed. by C. Lefebvre and J. Lumsden. Montreal. Lefebvre, Claire. 1994b. ‘‘On spelling out E’’. Travaux de recherche sur le créole haïtien 23: 1–30. Université du Québec à Montréal. Lefebvre, Claire. 1996a. ‘‘The functional category ‘‘agreement’’ and Creole Genesis’’. In Creole Languages and Language Acquisition, H. Wekker (ed.), 153–195. The Hague: Mouton. Lefebvre, Claire. 1996b. ‘‘The tense, mood and aspect system of Haitian Creole and the problem of transmission of grammar in Creole genesis’’. Journal of Pidgin and Creole Languages 11.2: 231–313. Lefebvre, Claire. 1997. ‘‘Relexification in Creole genesis: The case of demonstrative terms in Haitian Creole’’. Journal of Pidgin and Creole Languages 12.2: 181–201. Lefebvre, Claire. 1998a. ‘‘Multifunctionality and variation among grammars: The case of the determiner in Haitian and in Fongbe’’. Journal of Pidgin and Creole Languages 13.1: 93–150.

Relexification in Creole genesis

Lefebvre, Claire. 1998b. Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole (Cambridge Studies in Linguistics, 88). Cambridge: Cambridge University Press. Lefebvre, Claire. 1999. ‘‘On the empirical reliability of some Haitian data’’. Journal of Pidgin and Creole Languages 14.2: 1–6. Lefebvre, Claire. 2001. ‘‘The interplay of relexification and levelling in Creole genesis and development’’. Linguistics 39.2: 371–408. Lefebvre, Claire and John S. Lumsden. 1989. ‘‘Les langues créoles et la théorie linguistique’’. In La créolisation, C. Lefebvre and J. S. Lumsden (eds.), 249–272. (Special issue of The Canadian Journal of Linguistics 34.3.) Lefebvre, Claire and John S. Lumsden. 1992. ‘‘On word order in relexification’’. Travaux de recherche sur le créole haïtien 10: 1–22. Université du Québec à Montréal. Lefebvre, Claire and John S. Lumsden. 1994. ‘‘Relexification in Creole genesis’’. Université du Québec à Montréal. Paper read at the MIT Symposium on the Role of Relexification in Creole Genesis: The Case of Haitian Creole. Research report, C. Lefebvre and J. Lumsden (eds.). Montreal. Lefebvre, Claire and Diane Massam. 1988. ‘‘Haitian Creole syntax: A case for DET as head’’. Journal of Pidgin and Creole Languages 3.2: 213–243. Lefebvre, Claire and Elizabeth Ritter. 1993. ‘‘Two types of predicate doubling adverbs in Haitian Creole’’. In Focus and Grammatical Relations in Creole Languages, F. Byrne and D. Winford (eds.), 65–91. Amsterdam and Philadelphia: John Benjamins. Lightfoot, D. 1979. Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lumsden, John S. 1989. ‘‘On the distribution of determiners in Haitian Creole’’. Revue québécoise de linguistique 18.2: 64–93. Lumsden, John S. 1991. ‘‘La distribution des modificateurs dans le syntagme nominal en haïtien’’. In La créolisation: Théorie et applications, A. Kihm (ed.), 47–63. (Special issue of Recherches Linguistiques de Vincennes, 20.) Lumsden, John S. 1996. ‘‘On the acquisition of nominal structures in the genesis of Haitian Creole’’. In Creole Languages and Language Acquisition, H. Wekker, (ed.), 184–205. Berlin and New York: Mouton de Gruyter. Lumsden, John S. and Claire Lefebvre. 1994. ‘‘The genesis of Haitian Creole’’. Paper read at the MIT Symposium on the Role of Relexification in Creole Genesis: The Case of Haitian Creole. Research report, C. Lefebvre and J. Lumsden (eds.). Montreal. Milner, J.-C. 1978. De la syntaxe à l’interprétation. Paris: Éditions du Seuil. Mufwene, Salikoko S. 1991. ‘‘Pidgins, Creoles, typology and markedness’’. In Development and Structure of Creole Languages, F. Byrne and T. Huebner, (eds.), 123–144. Amsterdam and Philadelphia: John Benjamins. Muysken, Pieter Cornelis. 1981. ‘‘Half-way between Quechua and Spanish: The case for relexification’’. In Historicity and Variation in Creole Studies, A. R. Highfield and A. Valdman, (eds.), 52–79. Ann Arbor, MI: Karoma. Ritter, Elizabeth. 1992. ‘‘Cross-linguistic evidence for number phrase’’. In Functional Categories, C. Lefebvre, J. S. Lumsden and L. Travis (eds), 197–219. (Special issue of The Canadian Journal of Linguistics 37.2.) Sankoff, G. 1990. ‘‘The grammaticalization of tense and aspect in Tok Pisin and Sranan’’. Language Variation and Change 2: 295–312.





Claire Lefebvre

Sylvain, Suzanne. 1936. Le créole haïtien: Morphologie et syntaxe. Wetteren: Imprimerie de Meester; Port-au-Prince: Chez l’auteur. Thomason, Sarah Grey and Terrence Kaufman. 1991. Language Contact, Creolization and Genetic Linguistics. Los Angeles: University of California Press. Tinelli, Henri. 1970. Generative phonology of Haitian creole. Ph.D. dissertation, University of Michigan. Travis, Lisa and Greg Lamontagne. 1992. ‘‘The case filter and licensing of empty K’’. In Functional Categories, C. Lefebvre, J. S. Lumsden and L. Travis (eds.), 157–174. (Special issue of The Canadian Journal of Linguistics 37.2.) Trudgill, P. 1986. Dialects in Contact. Oxford: Basil Blackwell. Valdman, Albert. 1978. Le créole: Statut et origine. Paris: Klincksieck. Valdman, Albert et al. 1981. Haitian Creole–English–French Dictionary. Bloomington, IN: Indiana University Creole Institute. Voorhoeve, Jan. 1973. ‘‘Historical and linguistic evidence in favor of the relexification theory in the formation of creoles’’. Language and Society 2: 133–145. Westerman, D. 1930. A Study of the Ewe Language. London: Oxford University Press. Whinnom, K. 1971. ‘‘Linguistic hybridization and the ‘special case’ of Pidgins and Creoles’’. In Pidginization and Creolization of Languages, D. Hymes (ed.), 91–115. Cambridge: Cambridge University Press.

Chapter 2

Voodoo Chile* Differential substrate effects in Saramaccan and Haitian Norval Smith

.

Introduction

. General assumptions The title of this chapter ‘‘Voodoo Chile’’ is not frivolously chosen. The creoles to be discussed here — Haitian and Saramaccan — represent the languages of two of the communities in which, as is claimed by various anthropologists and creolists, African slaves from the Slave Coast played a major developmental role. On the Slave Coast — extending roughly from South-East Ghana, through Togo and Benin to South-West Nigeria — languages of the Gbe subfamily of the Kwa1 languages are spoken up to about 100–150 kilometres inland. In particular speakers of the various Fon dialects of present-day southern Benin (the former Dahomey) are supposedly responsible for much of the cultural, religious, and linguistic heritage of both Surinam and Haiti, among other places. I will argue that this is basically correct, if speakers of the very similar so-called Phla-Phela2 dialects (Capo 1994) of the immediate coastal region of Benin are also taken into account. When I wish explicitly to refer to both Fon and Phla-Phela groups together I will use the term Eastern Gbe. The term ‘‘Voodoo’’ represented more accurately in French orthography as Vaudoun (vodu˜) refers to the supreme god in the Eastern Gbe religion. In Haiti the role of Vaudoun religion is well-known, although the reality is a lot different from the popular conception as portrayed in horror stories and films. In Surinam, the syncretic Winti religion comprises several panthea one of which is based on the Eastern Gbe pantheon (cf. Wooding (1972)).3



Norval Smith

. Structure of this chapter The main aim of this chapter is to pose the question why in the case of the two creoles at issue, if the main substrate language is the same, the end-results were not basically identical. I abstract here however from questions of wordorder, where it is clear the respective superstrate languages French and English are responsible for certain differences, like that found in the order of adjectives and nouns in Saramaccan and Haitian, and further all English-lexifier and French-lexifier creoles. In studying the two languages I adopt an eclectic standpoint in preference to either an extreme universalist position, or an extreme relexificationalist position. I review the evidence for substrate influence from Gbe, in particular why this influence should be regarded as more important. In connection with the controversial question of the development of Saramaccan, I first study various questions concerning the development of its sister creole Sranan, as these are intimately bound up with the development of Saramaccan itself. I then turn more directly to the development of Saramaccan, and aspects that appear largely to be shared or partly shared among the Surinam creoles, like the TMA-system. My reason for looking primarily at Saramaccan is that this creole has been exposed to fewer extraneous influences than Sranan. I then address my attention to Haiti. I examine a phenomenon addressed in Lefebvre (1996) as displaying Fon substrate influence in Haitian, that of the Postnominal Determiners. I then attempt to show that a substrate explanation along the same lines is also amenable to some facts of Saramaccan (Definite) Determiners. Three cases where Haitian and Saramaccan do not agree, or at least only partially, and where Saramaccan specifically clearly reflects Fon substrate features, will then be examined. One conclusion I draw is that my eclectic stance is justified. Turning to the question of why both mechanisms seem to operate in the genesis of Saramaccan, I suggest that Saramaccan (and by extension the other Surinam creoles, although there are certainly important differences) involved at least two significant substrate languages, Eastern Gbe and Kikongo, but that Eastern Gbe was the most important. Why do Saramaccan and Haitian differ in the extent to which they reflect Eastern Gbe substrate patterns? This question I leave basically unanswered, apart from suggesting that there might have been some difference in the degree of substrate influence. Much more research is clearly required. What

Differential substrate effects in Saramaccan and Haitian

has been achieved is due to the not inconsiderable amount of research that has already been carried out. This has at least enabled researchers to begin asking the relevant questions. This whole question of the possible Gbe substrate influences on the Surinam creoles is now the object of study of several research projects involving the Universities of Amsterdam, Leiden, and Frankfurt and the Ohio State University.

. Creole Genesis I accept the basic validity of both the universalist and substrate approaches in providing partial but not exclusive accounts of creole genesis. In other words I accept both a version of the Language Bioprogram Hypothesis pioneered by Bickerton (1974, 1981, 1988), and a version of the revamped Relexification Hypothesis pioneered by Muysken (1976, 1980, 1981) and further developed by Lefebvre (1986). Which process comes to play the major role in any given case depends, I believe, on the linguistic and sociolinguistic circumstances pertaining in the creolizing situation. In order to avoid invidious and irrelevant debate on whether particular newly created languages fall under the definition of creole or not, I will, whenever necessary, refer to all newly created linguistic systems under the term Younger Languages. This term will include the classic creoles, as well as mixed languages like Michif and Media Lengua. The important thing, as far as I am concerned, is that the new linguistic system represent a radical departure from any preceding linguistic system, which cannot be explained by the workings of the normal processes of linguistic change. In addition, the new language is assumed to be created in a relatively short time — Bickerton’s one generation seems to fit those recent cases where we know anything at all about the time scale involved. I would extend this also to those cases where adults can be assumed to be primarily responsible for the creation of the new language. . The Language Bioprogram I assume that one prototypical breeding ground for creoles is a multilingual community, with a single socially dominant or primary language (the superstrate language), and a number of socially dependent or secondary languages (substrate languages). Prime examples of such communities would





Norval Smith

be the slave plantation colonies, as they formerly existed, in particular on the tropical litorals of the Atlantic region. I follow Bickerton in assuming that early on in the histories of many such colonies, the creole languages that developed were created by children who were attempting to learn a pidgin based on the superstrate language as if it were a full language. Such pidgins are assumed by Bickerton to have little or no grammar independent of the various grammars of the native languages of the speakers, so that a new grammar has to be constituted from those children’s innate knowledge of language universals, while on the other hand the pidgin lexicon will be utilized to supply the vocabulary. New words will also be created as necessary, principly by compounding. Bickerton’s theory assumes then that creole languages developed in one generation. This is a direct consequence of assuming creole genesis to be an L1-learning phenomenon. During the initial phase of development, termed by Bickerton the establishment phase, when the plantations were on a very small scale, and only had a handful of slaves, these slaves would presumably have had no difficulty in picking up the European language (Bickerton 1989). The next phase of development of most plantation colonies was what Bickerton calls the growth phase when the population balance shifted within a very short space of time in favour of the slaves, as the plantations were expanded. Once we get to this stage, it has to be assumed that most slaves would no longer have easy access to the superstrate language. It can also be assumed that there would now be a growing social need for a slave-community-wide language (Smith 1987: 14ff.). The slaves would speak various different African languages natively and might well retain these under the right conditions for several generations, but as time went on a new common ethnic sense would grow, gradually overlaying, then replacing, their original tribal affiliations, aided by intermarriage, general contact among the slaves, and the disturbed environment of the plantations. The period required for the emergence of this new (pseudo-)ethnic feeling can be assumed to have been fairly short under the prevailing circumstances. Again one generation would seem to be a reasonable estimate of the necessary time. However, the colonial language or some version of it, still has to be presumed to have been the target language for communication, otherwise there would of course be no reason for the present-day creoles ever to have developed. Note that there is no barrier to a duality of ethnicity. Examples of this are legion: Scots Gaelic ethnicity within a larger Scottish ethnicity; Defaka ethnicity within a larger Nkoroo ethnicity in southern Nigeria (Jenewari 1983); Frisian ethnicity within a Dutch ethnicity; etc.

Differential substrate effects in Saramaccan and Haitian

We can summarize the assumptions made in Bickerton’s Language Bioprogram Hypothesis as follows: (1) a. Creoles develop in multilingual communities. b. One superstrate language forms the target for inter-ethnic communication. c. No common substrate language is used by the whole subject community. d. There is a need for a common language among the substrate language speakers. e. Substrate speakers have in general limited access to the superstrate language. f. The languages that emerge in such situations are creole languages. g. These languages are created in a relatively short time.

Arends (1994) criticized Bickerton’s position on the grounds that at an early period in the history of this kind of colony there were not enough children to make the scenario envisaged by him possible. He saw this as support for his theory of a slow process of nativization of creole languages. Aside from the fact that it is not clear how many children would be ‘‘enough’’ to introduce a creole language, Arends (1995) has now produced new figures showing that there were more children in early Surinam than he had first thought. We discuss this further below, in connection with the development of Sranan. Another caveat has to be mentioned with respect to Bickerton’s original assumption, based on his work on Hawaiian Pidgin and Creole English (Bickerton and Odo 1976; Bickerton 1981), that the pre-creole pidgin was necessarily rudimentary in type. Work on the various English-lexifier creoles in the Atlantic area reveals a number of parallels in the function word lexicon (Cf. Hancock 1987; Smith 2001b). This suggests that the nature of the pidgin underlying these various forms of creole English was not as primitive as that assumed for Hawaii by Bickerton. We seem to have indications of a stable pidgin which was spread across the Caribbean in the space of a short period along with the English colonizers (Smith 2001b). . Relexification In Muysken (1976, 1980, 1981) a classic and undisputed case of relexification is illustrated — that of Media Lengua. This is a language, or possibly, set of languages, spoken in Ecuador in the towns of Salcedo and Saraguro and possibly others, and deriving from a mixture of Quechua and Spanish. It is





Norval Smith

spoken in the Andes region by small groups who are socially intermediate between the Spanish-speaking population of the towns, and the Quechuaspeaking population of the mountains. They are intermediate in various ways — socially, as we have just noted, and linguistically, but also geographically, as they live on the margin of certain towns, at a higher altitude than the Spanish speakers, but lower down than the Quechua-speakers. What has happened in this classic case of relexification is simple. Two languages — Spanish and Quechua — are in contact. The speakers of the dependent language, Quechua, have in general replaced the phonological forms of their own lexical items with corresponding items from the dominant language, Spanish. The agglutinative morphology and syntax (and phonotactics) of Quechua is however retained virtually complete (Muysken 1980). (2) Si masiado llubi-kpi no anda-sha-chu. (ML-Saraguro) if too.much rain-sub neg go-1fu-neg - yalli tamia-kpi mana ri-sha-chu (Quechua) ‘If it rains too much, I won’t go.’ (‘Si llueve demás, no voy a vir’) (Spanish)

In Salcedo, according to Muysken, Media Lengua derives from the speech of young bilingual adults, and dates from about 1920. In this town the Media Lengua speakers are peasants, weavers, and construction workers. Lefevbre (1986) adopts the concept of Relexification as a hypothesis on creole genesis in connection with a long-running project (1985–1994) carried out by a research group at the Université de Québec à Montréal (UQAM) on the genesis of Haitian. Most of their work assumed relexification from Fon by French. Crucially, however, Lefebvre — in common with Bickerton — regards creolization as essentially involving more than two languages. The speakers of the various languages produce different results from their various processes of relexification, following on which a process of reconciliation will take place (cf. Lefebvre, this vol.). In this scenario too, the relexification is carried out by adult speakers. . Eclectic view As we said above we are convinced of the essential correctness of both views on creolization. What we do not believe is that the two hypotheses operate precisely as described by their respective proponents. Even to state the problem, however, as being just one of which hypothesis is operational under

Differential substrate effects in Saramaccan and Haitian

which set of circumstances is, I think, to be guilty of a gross oversimplification of the complexity of the situation. I in turn do not pretend to offer more than the beginnings of a solution to this thorny problem. In one concrete sense the version of the universalist view being propagated here is a compromise between the Bickertonian and ‘‘Arendsian’’ viewpoints. With respect to Arends however, we prefer to reinterpret the particular evidence he adduces for slow development in the light of other facts. We would suggest that certain aspects of creoles seen by Bickerton as being intrinsically associated with creolization are not in fact necessarily aspects of the creolization process at all. Inasmuch as Arends has produced evidence for the slow development of some of these same aspects of creole languages, we suspect that they will turn out not to be necessary aspects of the creolization process, but rather instantiations of the normal historical processes of change. What of the demarcation between the mechanisms of the Language Bioprogram Hypothesis (LBH) and Relexification as two means of creation of Younger languages? In Smith (1987), we called upon both mechanisms to explain the development of Saramaccan. We proceeded from the assumption that the combination of a single substrate language (like Quechua) with a single superstrate language (like Spanish) would result in a mixed language (like Media Lengua).4 We also proceeded from the assumption that the combination of several substrate languages (Hawaiian, Korean, Japanese and various Philippine languages) with a single superstrate language (English) would result in a creole (Hawaiian Creole English). In the case of Saramaccan we explained the development in two steps: a first phase (external to Surinam) involving the LBH, with several substrate languages and one superstrate language (English) resulting in what we called Proto-Sranan; and a second phase (internal to Surinam) involving Relexification, with two languages Proto-Sranan and a putative N.E. Brazilian Portuguese-lexifier Creole (see for this last aspect also Smith (1999)). The existence of a Portuguese creole in N.E. Brazil is disputed (do Couto 1999; Smith 2001a), but whether such a language was in fact involved, or just the Portuguese superstrate of the Jewish planters, does not significantly affect the point at issue, which is that two significant languages were involved. While retaining my two-stage hypothesis on the origin of Saramaccan, I would now however accept that the LBH probably did in fact operate in Surinam to produce ProtoSranan (on this see Smith (1999, 2001a)). The Media Lengua situation cannot in addition be completely equated with the Saramaccan situation, assuming Relexification played a role in





Norval Smith

this, and neither can the end-results. Media Lengua is virtually completely relexified: (3) Lexicon: Spanish Grammar: Quechua

and Saramaccan is not: (4) Lexicon: Proto-Sranan + Portuguese Grammar: Proto-Sranan (?)

Complete Relexification would have meant that the lexicon would have been basically Portuguese. In Smith (1987) I explained the incomplete relexification to Portuguese displayed in Saramaccan in terms of the fact that initially at least the Portuguese-owned plantations were only a minority as compared to their English-owned counterparts, and that therefore there would only be a minority of plantations which had the Portuguese Creole as their language, and a majority speaking Proto-Sranan. While I would now explain things in slightly different terms, this is not relevant for this article (see Smith 1999). Other writers seem to share the spirit of my eclectic approach (cf. McWhorter 1999).

. Surinam and Haiti . Surinam Before addressing the question of the formation of Saramaccan more closely, it is necessary to spend some time considering various aspects of the genesis of Sranan, as the history of Saramaccan is inextricably bound up with this. .. Sranan Abstracting here as much as possible away from the question of how the various linguistic strands that came together to form Sranan were exactly interwoven, I would like to address three questions. The first is that of the number of children present in early Surinam, and whether this was sufficient to allow the operation of the Language Bioprogram. The second is that of the transmission of what I have referred to as Proto-Sranan over the gap caused by the transference of power from the English to Dutch, and the departure of most of the English with their slaves in the mid-1670s. How come an English-lexifier creole survived at all? The third question concerns the source of the major

Differential substrate effects in Saramaccan and Haitian

substrate influences, in other words the languages spoken by the early slaves. ... Children in Surinam. I claimed in Smith (1987) that the LBH supplied us with a scenario for the development of Sranan, the main English-lexifier creole language of Surinam. I now however in addition accept that the particular operation of the Bioprogram that led directly to the formation of Sranan ought to be localized in Surinam. I do not agree with those (in particular Arends (1989)) who see the ‘‘creolization’’ of creole languages as a process extending over a number of generations. As I noted above, one of the difficulties raised by Arends for the LBH concerned the number of children present in the early years of plantation colonies. Arends considered on the basis of work on the demography of Surinam (Arends 1994) that the number would be insufficient. However, the question of just how many children would be required, a completely unknown factor in any case, has become somewhat academic as new evidence has emerged which has caused Arends to significantly revise his picture of the number of children in early Surinam. Arends (1995) now estimates that for the period 1680–1729 children formed 15.2 per cent of the slave imports. In Table 11 in the same article Arends quotes the result of a poll tax count giving no less than 27.4 per cent of the 4237 slaves present in Surinam as under 12 in 1684 (Van der Linde 1966). Although there is some reason to doubt the complete accuracy of these population figures,5 this is an important piece of information, being the earliest census-type information to show any indication of age. Shipping figures for the seventeenth century also give proportions of 32.2 per cent (1684) and 9.1 per cent (1685) of under-14s for two individual shipments of slaves (Arends 1995). Offical figures for the beginning of the eighteenth century are also available (for the years 1703, 1704, 1706, 1710, 1711 and 1712 (van den Berg 2000a)). These reveal an average of more than 14 per cent children under 12. The received doctrine on second language-learning is of course that children below the critical age of puberty exposed to second languages learn them natively — i.e. as if they were first languages. If there was any accuracy in the figure of 27.4 per cent of slaves being under 12 in 1684, it is clear that we do not have much basis for claiming that at some earlier period there were too few children for the LBH to operate. The question that we must answer is rather, when could it have operated? A recent study (van den Berg 2000b) has revealed the existence of early Sranan sentential material as early as 1707. In general we can see that the new sentence material available for the first half of the eighteenth century





Norval Smith

reveals a language that can be recognized as being Sranan as we know it. Bickerton (1989) advanced the more sophisticated theory that the need for creole languages only became relevant when the ratio of slaves to Europeans became such as to disfavour access to the colonial language. This was not in general the case until some time after the foundation of plantation colonies, as in the initial settlement phase the plantations tended to be small, affording plenty of scope for interaction between the Europeans and the relatively small number of slaves present at the time. The fact that there was scope for interaction is of course no guarantee that there was much. In addition, in the seventeenth century we frequently have present numbers of European indentured labourers, or transported convicts, who were treated little better than slaves, and presumably had more interaction with them, although once again evidence of this would be welcome. To judge from the available figures the establishment phase lasted in Surinam for most of the 1650s. The first population estimate for Surinam indicating a significant preponderance of black over white inhabitants is from 1661 with 2000 slaves as against 1000 whites. In addition we must consider that the population ratio on the plantations would tend to be more extreme as compared with that in the English capital Torarica. The population figures seem to have remained more or less constant during the 1660s — at least as far as the totals are concerned, at the individual level this was a period of great flux. According to Wekker (1991) the average number of slaves per plantation at this period was about ten. While this might not seem to necessarily involve a significant lessening of access to Standard English, we must not neglect the additional factor of an emergent ethnic feeling among the slave community. What can we say about the ethnic mix during this period? Warren (1667) writes this concerning the slaves: They are there a mixture of several nations, which are always clashing with one another, so that no conspiracy can be hatching, but it is presently detected by some party amongst themselves disaffected to the plot, because their enemy have a share in it.

This suggests that, at any rate, there was apparently no single common African language, but certainly some more than primitive means of communication, at least if we can take talk of plotting and detecting plots seriously. We take this as an indication that an English-based pidgin or creole may well already have existed. A relevant point (Adrienne Bruyn, p.c.) is that at the same time as the first children were developing the creole language, they also were either learning

Differential substrate effects in Saramaccan and Haitian

African languages, or already spoke them. To the extent that any one group was dominant in early Surinam this would make for a degree of influence on the creole (adstratal or substratal). The reason why a creole developed and not a mixed language is presumably that Eastern Gbe was not the sole language of the slaves (cf. Smith 1987). That this seems to have been the case we can see below from the figures on slave imports in (7). It should be noted that this represents the last period in which an Englishlexifier creole could have formed — in terms of the LBH. In the 1670s or later any creole that developed would presumably have been a Portuguese-lexifier (c.1675–c.1700)6 or Dutch-lexifier one (after c.1700). A similar point is made in McWhorter (1995) although given a different interpretation. It might be remarked that the date of creolization could be advanced to as late as 1675. This would seem unlikely however in view of the relatively small number of Portuguese or Dutch elements visible in eighteenth-century Sranan (cf. Schumann 1783; Smith 1999). The study of pidgins reveals that in the lack of a population using it as their primary language, lexical stability does not really exist. So, if Proto-Sranan were to have been a pidgin till after 1700, or even till about 1750 as Arends (1989) suggests, one wonders at the lack of influence of Dutch and Portuguese in Schumann’s dictionary of 1783. One example of a pidgin for which we have reasonable lexical information over a longer period of time is Chinook Jargon which changed the nature of its vocabulary quite significantly in half a century, from Chinook as the major component to English (Grant 1994). In the same way, any English pidgin would presumably have rapidly relexified to Portuguese or especially Dutch after English had been removed as a model from the colony. Note that this relexification would then become visible in basic vocabulary. The basic vocabulary of Sranan remains fairly solidly English. The total number of English morphemes in Sranan is roughly 700. The number of words in Sranan is of course far greater with words from Fon, Kikoongo, Portuguese, and latterly a large number from Dutch. In addition, a great deal of use has been made of compounding. However, the number of 700 probably gives us a fair idea of the size of the lexicon of the original pidgin. The fact that English influence was removed more or less immediately following creolization enables us to ‘‘see’’ the pidgin better than in most cases of creolization, I would claim. ... The gap. The colony of Surinam had been settled in 1651. The colony was assigned to the Netherlands as a result of the treaty of Breda, thus coming





Norval Smith

Table 1. The times of departure of the English and their slaves from Surinam (from Arends 1995) Departures 1668 1671 1675 1680

English 67 c.250 250 c.50

Slaves 7

412 c.250 981 c.50

Destination Barbados Jamaica Jamaica ?

under Dutch ownership. The de facto Dutch occupation started in 1668, resulting in the departure of nearly all the English and their pre-1667 slaves during the 1670s. By 1680 then, English superstrate linguistic influence can be regarded as negligible. The major part of the English left in 1671 and 1675, and of their slaves between 1668 and 1675. As has often been pointed out the problem of the transmission of (Proto-) Sranan is compounded by the short time when the ‘‘old’’ pre-1667 slaves of the English overlapped with later slaves (whoever they were owned by). Under the terms of the treaty only the ‘‘old’’ slaves could be removed. Between 1668 and 1675 no less than c.1693 of these ‘‘old’’ slaves departed with their owners. Arends refers to the brief period between 1668 and 1680 when ‘‘old’’ and ‘‘new’’ slaves overlapped. In fact the period of effective overlap is not really even twelve years but basically only seven years (as noted in Bickerton 1994), as there were only fifty ‘‘old’’ slaves who departed later. Arends suggests that some ‘‘old’’ slaves remained with the thirty-nine English who Rens (1953) states remained behind after 1680. If Rens’ total of 2000 slaves present in Surinam before it became Dutch in 1667 is anything like accurate, then there must have been very few ‘‘old’’ slaves left at all, however. The number who left with the English is estimated at c.1693, and we must allow too for a number of deaths and escapes during the ‘‘overlap’’ period. There are a number of ways in which linguistic continuity between the pre-gap and post-gap situations would be strengthened. There was, for instance, apparently a category of ‘‘old’’ slaves that were confiscated by the Dutch. The property of absentee owners, and colonists who had refused to swear allegiance to the Dutch was forfeit. As is mentioned in note 7, the English apparently removed 168 slaves illegally in 1668. Rens (1982) suspects that these were slaves that were forfeit property. They were ordered returned to Surinam. If this return in fact took place, we have at least 10 per cent ‘‘old’’ slaves as a gap-bridging factor to be taken into consideration. A number of other slaves would also have belonged to this forfeit category. This is an

Differential substrate effects in Saramaccan and Haitian

additional factor that as far as I know has not so far been considered in connection with the cross-gap transmission of Proto-Sranan. Allowance must be made for example for a number of slaves purchased by English plantation-owners after 1667, but before the largest party of ‘‘old’’ slaves were removed in 1675. These slaves could not be removed under the terms of the treaty. Such slaves would be confronted especially intensely with the existing means of communication. It is obviously of considerable importance to get some idea of the total number of slaves imported between 1668 and 1675. This is the overlap cohort — the ‘‘new’’ slaves imported while there were still a considerable number of ‘‘old’’ slaves present. Rens suggests a growth in the slave population of 1000 between 1667 and 1673. As c.660 slaves left during this period, we can assume that in excess of 1660 slaves arrived during this vital period (or in excess of 1500 if the illegally removed slaves were returned). A more reasonable figure in the light of the very large mortality typical of Surinam (Van Stipriaan 1993) would be 2000. This group would be large enough in itself, I consider, to guarantee the transmission of Proto-Sranan from the ‘‘old’’ to the ‘‘new’’ slaves, inasmuch we may assume that among the ‘‘new’’ slaves were a significant number of children who could as it were quickly acquire Proto-Sranan as a second first language. A question that we have to ask is whether there was sufficient access for the newcomers to what we have called Proto-Sranan here to maintain it as a language, or whether, as Bickerton (1994) suggests, repidginization would have taken place. We consider that the two factors of a sufficient number of children among the slave-imports, and the presence of an unknown number of forfeited ‘‘old’’ slaves might have been sufficient to avoid repidginization and subsequent recreolization. ... The source of the slaves. As far as the source regions in Africa are concerned the available information specific to Surinam between 1658 and 1675 has not yet been published (Arends (1995), quoting (Postma 1990)). For this period we have to make do with data for the whole Dutch Atlantic trade. For slaves who we know were imported to Surinam (Postma 1990), we give the half-decade totals for the years up to 1710 (see Table 2). Note that the Saramaccan tribe was founded in 1690, and significant accretions continued up till about 1710 (Price 1983). The reason for giving half-decade figures is that this smooths out the idiosyncracies that appear if one uses yearly figures, but preserves interesting trends that would get submerged if the figures are





Norval Smith

examined per decade. How should we interpret these figures? At present, for the period up till 1674, we only have figures derived from the general Dutch slave trade. However these indicate that probably the most significant early input derived from the Slave Coast. And obviously the earliest slaves would have the greatest influence on linguistic habits. The percentages between brackets represent what we get if the slaves of unknown origin are ignored. It must be pointed out here that the proportions of slaves imported into Surinam were later frequently different from those pertaining in the Dutch Atlantic slave trade as a whole (Arends 1992). However, some confidence as to the accuracy of the general picture suggested by these figures can maybe be gleaned from Warren’s statement that the slaves (i.e. in the English period) were mostly from Guinea (Warren 1667). For the late 1670s — a period when nearly as many old slaves left Surinam as new slaves arrived — we can say that most of the rather small number of new slaves would be speakers of Bantu languages. The first half of the 1680s was once again dominated by Slave Coast imports, and the late 1680s by Bantu slaves yet again, with however still a sizeable Slave Coast contingent. Note, however, that we will argue below that Saramaccan, or more correctly, its preescape instantiation — Djutongo — was creolized around 1680, around ten years preceding the effective foundation of the Saramaccan tribe. This would effectively preclude large-scale Bantu influence in Saramaccan, as most of the slaves speaking Bantu languages would be the ‘‘newest kids on the block’’.

Table 2. Slave imports to Surinam

1658–1674 (general Dutch trade) 1675–1679 1680–1684 1685–1689 1690–1694 1695–1699 1700–1704 1705–1709 1710–1714

Gold Coast

Slave Coast

Loango

Other

Unknown

10.4% (18.3%)

28.0% (40.8%)

13.9% (24.6%)

(W) 4.3% (7.6%) (B) 4.9% (8.7%)

43.4%

150 175

1,289 47.8% 1,721 24.3% 950 45.8% 2,197 41.7% 2,397 68.6% 3,250 72.4% 1,360 65.9%

710 61.2% 679 25.0% 3,882 54.9% 615 29.6% 2,384 45.2% 568 16.3% 579 12.9% 705 34.1%

5.6% 2.5%

657 14.6%

W=Windward Coast; B=Bight of Biafra

844 (B) 11.9%

450 38.8% 582 21.6% 450 6.4% 511 24.6% 692 13.1% 528 15.1%

Differential substrate effects in Saramaccan and Haitian

The slave imports during the twenty-odd formative years of the Saramaccan tribe were again dominated by those from the Slave Coast. This section of the table is shaded. Important seventeenth-century slave-trading centres were Ardra, Jacquin and Whydah. Others were Offra, Popo and Appa. Slaves coming from the Slave Coast were identified as Ardras, Foins, and Abos among others. We attempt to identify these names and others frequently mentioned in the literature in Table 3, making use of Capo’s assignment of Gbe dialects to larger groups and his description of where there they are spoken (Capo 1991). The first two columns give some early spellings of the names of the places concerned. The third column contains various historical terms used for identifying slaves from that area, and in brackets the names used for Slave Coast slaves that were shipped through that particular place according to Hartsinck (1770). The fourth column gives the modern name, and the fifth column the local Gbe name. The sixth column is an attempt to identify the Gbe dialect(s) spoken in the neighbourhood of the place. The last column gives Capo’s assignment of these dialects to the two Eastern Gbe groups — Fon and Phla-Phela. Conceivably, the fairly large lexical contribution of the Bantu languages (cf. Daeleman 1972, Huttar 1986) to Saramaccan and Ndyuka (directly Table 3. Gazeteer of Slave Coast ports and trading centres Place

Other variants

Slaves

Modern placename

Gbe name

Gbe Dialect

Dialect group

Ardra7

Ardres

Ardra

Allada

Alada

Arɔhún [aɔʁu˜] Ayizɔ

Fon Phla-Phela

(Foin)

Godomey (?)8

Kpasε

Fon Phla-Phela Fon

Jacquin Whydah

Fida, Juda

(Foin, Juda, Ardra)

Ouidah

Glexwé

Phelá [Xwelá] Kpasε

Popo

Papa

Popo

Grand Popo

Xwla

Phla [Xwla]

Phla-Phela

Popo

Anécho

Anéxɔ

form. Phlaa

Phla-Phela (now Gen)

Cotonou (?)9

Kútɔnu

Urban Fon

Fon

Abomey

Agbómε Agbómε

Fon

Porto-Novo

Xɔgbónu Gun [ogu˜] Alada10

Fon Phla-Phela

Little Popo Appa

Aqua

Abomey Porto Novo (from 1752) a

Gen since 18th c.

Abo, Foin





Norval Smith

descended from Plantation Sranan) is to be explained by the proportionately and absolutely high importation of slaves from Loango during different short periods in the late seventeenth century. It would have been easier for these slaves to influence less central aspects of language, like the cultural lexicon. Bickerton (1994) would emphasize the role of Bantu slaves in the formation of the Saramaccan language over that of Gbe-speaking slaves. It appears however from Price (1983) that the Eastern Gbe component was culturally especially important among the early Saramaccans. So the Matjáu and the Abaísa clans, the earliest escapees, are famous for their competitiveness in the singing of Papá songs — songs in a Gbe-derived ritual language referred to as Papá. Indeed the Abaísa themselves were known as Papa-Negroes during the eighteenth century, and Papa is yet another form of the name Popo (see the above table). Note that the singers of these songs refer to them as Aladá (see above), which apart from being an important slave-trading centre was also the capital of the Kingdom of Alada (Ardra). .. Saramaccan and the Portuguese Jews We have suggested that the creolization of Sranan can be located in the 1660s. This is of considerable importance for the history of Saramaccan. The work of Price, in particular Price (1983), makes clear that the elder clan of the Saramaccan tribe, the Matjáu, dates from a mass escape of slaves in 1690 from a plantation belonging to Imanuel Machado, a Portuguese Jew. In this case the clan name is derived from the family name of the plantation owner. It is clear from Price’s work that the Saramaccans derive in large part from Jewish plantations located on the middle reaches of the Suriname River. The Jews that owned many of the plantations in this area had been first admitted to Surinam in 1665. I hope to have established in Smith (1987) that the term Djutongo (‘‘Jewlanguage’’) is not restricted to just post-marronnage Saramaccan itself, but was also used for the plantation language as it developed further during the eighteenth century. In other words, the term is sometimes used with wide scope for both Saramaccan and the language of those who stayed behind, and sometimes with narrow scope to refer purely to the latter. The Saramaccans are described as speaking Djutongo (e.g. by Schumann 1783), at the same time as he provides some Djutongo forms in his Sranan dictionary which are not identical to the corresponding forms in his Saramaccan wordlist. He defines Djutongo as ‘‘the Negro language mixed with Portuguese [my emphasis]’’. Wullschlägel states clearly that Djutongo ‘‘was spoken on the numerous plantations belonging to Jewish owners and has now nearly disappeared from the

Differential substrate effects in Saramaccan and Haitian

colony, along with the prosperity of those who introduced it’’ (Wullschlägel 1856). For this reason we assume that Djutongo was also the language used on the largely Jewish-owned Middle Suriname River plantations prior to the slave escapes, as well of course as thereafter. (5)

Pre-1690 Plantation Djutongo / \ Saramaccan Djutongo Later Plantation Djutongo

Bickerton suggests that Saramaccan creolized in the jungle (e.g. Bickerton 1981). In the 1690s to the early 1710s the Matjáu, Abaísa and Nasí clans of the Saramaccan tribe lived close together, in an area adjoining the plantation area on the Suriname River. If Saramaccan was created in the jungle it would presumably have to have been at this juncture. For in the immediately following period the Matjáu moved far to the south. However, it is not clear to me why the slaves on the Middle Suriname River area dominated by Jewish plantations should wait until they escaped from their plantations to develop a common means of communication, as is suggested by the work of Bickerton. And as we have seen what little evidence there is about Djutongo suggests that it was spoken by the Saramaccan tribe, and the slaves on the Portuguese Jewish plantations. Our first real linguistic knowledge of Saramaccan of course dates from the period of Herrnhutter’s missionary work, which began in 1765 (resulting in e.g. Schumann 1778). Assuming the creolization of Sranan in the 1660s, when could Saramaccan/ Djutongo have come into existence? The relevant parameters are, I think, the coming of the Jews to Surinam in 1665, and the formation of the first clan of the Saramaccan tribe in 1690. Also relevant is the large shared English element in Saramaccan and Sranan which is virtually all of direct common origin (see Smith 1987), and the wide-ranging grammatical similarities which are appearing as extensive syntactic work on the Surinam creoles increases (Byrne 1987; Bruyn 1995; Huttar and Huttar 1994; Veenstra 1996). There seem to be only two possibilities: (6) Serial origin: Parallel origin:10

Sranan creolized, then Djutongo (as a development from it, as some kind of mixed language) Sranan and Djutongo developed separately (but on a basically common English pidgin basis)

Within the Serial scenario the answer to our question above is fairly easy — Djutongo could only have been formed sometime during the 1670s or 1680s





Norval Smith

— say c.1680. This was a period in which Portuguese Jews possibly formed the majority of the white population of Surinam. Within the Parallel scenario the answer is more difficult. If our conclusions regarding Sranan are correct then, in order to allow parallel development, the date for creolization would have to be c.1670. Against the Parallel scenario are the common elements in grammar. If we have to assume that an English pidgin formed the basis of both Sranan and Djutongo, then Sranan could be satisfactorily explained. Saramaccan would then be derived from the same English pidgin in combination with a (partly) Portuguese stratum. However, the nature of the relations between Sranan and Saramaccan, with the Saramaccan English-derived vocabulary being a near subset of the Sranan English-derived vocabulary, makes this second hypothesis unlikely. .. Saramaccan and Sranan — evidence from language What is, I consider, beyond doubt is that the considerable number (c.350) of lexical items of Portuguese origin in the Saramaccan vocabulary is to be ascribed to the presence of the Jewish population on the plantations, in whatever capacity. A point that must be emphasized here is that Saramaccan is not a Portuguese-lexifier creole, as some (for instance Perl (1995: 247)) have claimed. It is quite clear that the major lexifier-language is English. This becomes even more obvious when the proportions of the sources of function-words are examined. ... Basic vocabulary and function-words. Using a 200-word Swadesh list of basic vocabulary the direct sources of the lexical items, in percentages, are set out in Table 4. Note that the lexicostatistic figures in this table must be interpreted with due care. They let us see only to what extent various languages had a basic influence on a particular creole language. They are certainly not amenable to any standard interpretation in terms of glottochronology. In terms of that (fairly crude) linguistic tool, for Table 4. The lexical sources of the 200-word basic vocabulary list for three Surinam creoles Sranan Ndyuka Saramaccan

English

Portuguese

Dutch

African

77.14% 76.47% 49.96%

3.70% 5.04% 34.88%

17.58% 15.97% 10.45%

1.59% 2.52% 4.74%

Source: Smith (1987)

Differential substrate effects in Saramaccan and Haitian

instance, 75 per cent shared basic vocabulary would imply a separation of Sranan and English about 1000 years ago. In all such calculations however, loaned elements must be disregarded. For creole languages the problem of identifying loans is invidious. Which elements should we regard as being loaned in Sranan and which as being not loaned? Depending on the point of view of the linguist any of the categories of words in the above table might be regarded as loaned. In any case, glottochronology was developed for situations where languages develop by the ‘‘normal’’ gradual means of linguistic change, not by the cataclymic process of creolization. The dominance of the English element in Saramaccan is clear. An even clearer qualitative difference emerges if we look at the percentages pertaining in a list of all function words (see Table 5). The percentage of function words of Portuguese origin is only a quarter of the percentage of English words, about the same as the proportion of Dutch function words. Table 5. The lexical sources of function-words in Sranan and Saramaccan

Sranan Saramaccan

English

Portuguese

Dutch

African

Unknown

71.96% 54.62%

1.35% 13.87%

14.19% 12.61%

2.03% 5.88%

10.47% 13.03%

Source: Smith (1987)

Now that it is clear beyond any doubt that English is the dominant superstrate language in all the Surinam creoles, let us turn to some important agreements and differences between Sranan (and Ndyuka) on the one hand, and Saramaccan on the other. As I stated above, an examination of the shapes of the English items in these creoles demonstrates that nearly all the English items in Saramaccan are also present in Sranan in a form that is consistent with a shared origin (Smith 1987). Only a few items clearly go back to different proto-forms. The English lexicon in Saramaccan is basically a virtual subset of that in Sranan. This applies also with respect to function words, and their associated syntax. Inasmuch as there are parallels here it is clear that not just the lexicons of Saramaccan and Sranan (and Ndyuka) show parallels, but also the morphosyntax (Hancock 1987). As an illustration of this we will examine their TMA systems. At first sight the system, while displaying parallels as to the possible combinations and resultant meanings of the markers, exhibits partial agreement and partial disagreement in the markers utilized.





Norval Smith

... Tense, Mood and Aspect in the Surinam creoles. The TMA-markers represent a shared system in function, but not in form (see Table 6). The interpretation of these differences led Bickerton (p.c.) to express the seemingly reasonable conclusion that at the time of formation of Saramaccan, (Proto-)Sranan — undoubtedly one component of Saramaccan — was only a pidgin, and had not yet acquired a grammar. It is certain that the TMAsystem is not (completely) shared. Since the TMA-systems of the two creoles have not undergone changes since they were first committed to writing in the eighteenth century, it would seem likely to be the case that they have been different for a long time. Bickerton (1981) regards the TMA-system as one of the basic characteristics of creoles. The question is whether he is right in doing so. Various authors (e.g. Arends 1989) have pointed to the first then known recording of Sranan (Herlein 1718), and drawn attention to the absence of the Durative marker. The only certain TMA-marker occurring in the short text is Anterior ben, with Future go as an additional possibility. It is also possible that the latter is rather to be regarded as an Auxiliary. The recent work of van den Berg (2000b) has added early examples of Irrealis sa (1745) and Durative de (1762). Recent work on the history of Mauritian Creole (Baker 1995), based on a large number of texts, reveals also that the pre-verbal TMA-markers are not evidenced simultaneously. The first to appear in the available texts is the Completive marker fini/n in 1734. Next the Indefinite future (Irrealis) va/a and the Past ete/ti appear in respectively 1777 and 1779. Finally the other three markers, Definite future pur/pu, Immediate completive fek, and Progressive apre/pe appear in respectively 1818, 1818, and 1822. A similar slow development of TMA-markers is reported from Cayenne Creole (cf. Jennings 1993). Early Saint-Christophe Creole apparently did not have any TMA-particles (Jennings 1995b), although it must be admitted that the early textual evidence of this creole does not amount to much. It has been said that one difference between (unextended) pidgins and creoles is that creoles mark Tense, Mood and Aspect by means of pre-verbal markers, while pidgins utilize adverbs, if they mark these categories at all. One Table 6. TMA markers in the Surinam creoles Sranan Ndyuka Saramaccan

Anterior Tense

Future Tense

Irrealis Mood

Durative Aspect

ben be bi

o o o

sa sa sa

e e ta

Differential substrate effects in Saramaccan and Haitian

problem however with the TMA-markers in Surinam (and maybe other creoles too) is that, with the single exception of the Anterior Tense marker, these markers were all originally homophonous with other items, from which they are clearly derived in some sense. This could of course have occurred during creolization — there is unfortunately too little early evidence from Surinam. It does imply, however, that all these elements were present in the pidgin precursor(s), ready to be grammaticalized, with the possible exception of the Anterior markers. The late-eighteenth-century forms of the TMA-markers are set out in Table 7 (Schumann 1778, 1783). Table 7. Eighteenth-century TMA-markers in Sranan and Saramaccan Sranan Saramaccan

Anterior Tense

Future Tense

Irrealis Mood

Durative Aspect

ben bi

go go

sa sa

de tann

go

homophonous with the lexical verb gó ‘go’; The etymological source is English go. sa homophonous with the auxiliary verb sá ‘can’ (Sar.). The etymological source is probably Pidgin English savvy ‘know how to’ rather than English shall. The full verb sábi ‘know’ also has a short form sá in Sranan, Saramaccan and Ndyuka. de homophonous with the locative copula de ‘be’ (also in Saramaccan — dε´), also in the meaning ‘there’; The etymological source is English there. tann homophonous with the copula tán ‘stay’ (also in Sranan). Huttar and Huttar (1994) note that this also marks Continuous aspect in Ndyuka (possibly as an Auxiliary verb?). The etymological source is English stand, as is demonstrated by the nasal element in the eighteenth-century form, not Portuguese estar, as is frequently assumed by creolists who ignore the eighteenth-century form. Note that while the two creoles have made different choices for expression of the Durative Aspect, both of the copulas are present in both languages. Note also that for the irrealis and future the same choices were made, suggesting the same development, if not a common source. The case of the Anterior is puzzling for a number of reasons. Firstly, if the model is English been the Saramaccan form raises questions. While the Sranan form is presumably from general English [b n], the Saramaccan form is





Norval Smith

aberrant in that the final nasal is missing. In modern Saramaccan nasalized vowels are sometimes unstable, but in the eighteenth-century language this is not generally the case. Function words do of course tend to exhibit unusually drastic phonological reductions. In addition however the vowel is different from that in ben, although the development to /i/ of this vowel is certainly not excluded here (see Smith (1987) for examples). A possibility that has been suggested recently (Veenstra 1996) is that bi is of Portuguese origin. Compare also the Gulf of Guinea Portuguese-lexifier creole Fa d’Ambu where the Anterior marker is also bi (Bakker, Post and Van der Voort 1995), etymologically from vir ‘come’. Has the similarity in segmental structure, with a voiced labial stop followed by a front vowel played any role here? This suggests the possibility that a Sranan ben might have been relexified by a Portuguese creole bi. This explanation would not be available for the two Durative aspect markers. as these are both derived from English models. The disparate origin for this marker in an already-existing Proto-Sranan increases if they did not spring forth ready-armed at creolization like the soldiers who grew from the proverbial dragon’s teeth of Greek mythology, but developed slowly later, by various processes of grammaticalization, as seems to have taken place in the above-mentioned exceptionally well-documented case of Mauritian Creole. Both tann and de existed in both Saramaccan and Sranan in the respective meanings ‘stay’ and ‘(be) there’. There are sufficient other grammatical parallels among the Surinam creoles to make clear that even if the system of TMA-markers is partially different, this does not mean that at a period preceding the development of the full TMA-system we had a grammarless pidgin à la Hawaiian Pidgin English (cf. Bickerton and Odo 1976; Bickerton 1981). What I would like to suggest here is that while Bickeron’s three basic markers, indicating Anteriority, Irreality, and Nonpunctuality, may well represent a universally unmarked system of verbal distinctions, it may not necessarily follow that every creole immediately develops such a system. If a pre-creole pidgin has other (adverbial) elements expressing TMA-type concepts, then this may not be necessary. The operation of subsequent processes of grammaticalization may well lead to the achievement of the ‘‘classical creole pattern’’ later on, without meaning that such an optimal system is required by creolization. Note then that this does not imply that I believe in an ‘‘Arendsian’’-style slow creolization, just that it is not necessary for a full language to possess the claimed ‘‘universal’’ set of TMA particles. I abstract here away from the controversies on the universal order and the membership of the set of TMA particles.

Differential substrate effects in Saramaccan and Haitian

. Haiti — History For this account of the historical and demographic background to the Haitian situation I base myself on Singler (1993). There was a slow growth of the French colony of Haiti from say 1659 until 1681 when the population was roughly 4000 French to 2000 Africans. There was a significant shift by the time the population was next measured in 1715, to c.6500 French and 32,000 Africans (mostly slaves). In other words in 35 years the proportion of Africans had risen from 35 per cent to 83 per cent. Of the slaves imported in the perod to 1710, Singler estimates that ‘‘the largest number of African inported to Saint Domingue (= Haiti, N.S.) were ‘Kwa’, especially Ewe-Fon’’. Although there are no exact figures available, Singler regards as probably typical for Haiti and other French colonies the early demographic data recorded on a plantation in Cayenne — Rémire — in 1690. These are: (7) Slave Coast:

Gbe Yoruba Senegambia: Various Gold Coast: ‘‘Cormanti’’ (= Twi) Bight of Biafra: Calbary (= Kalabari) Congo: KiKoongo

49.3% 1.5% 18.5% 4.6% 9.2% 16.9%

The Gbe slaves in Rémire in 1690 can be basically divided into four categories: Foin Arada Juda Popo

= Fon = Alada = Whydah = (Grand and Little) Popo

= Agbóme (Fon) = Arɔhun (Fon) or Ayizɔ (Phla-Phela) = Xwela (Phla-Phela) or Kpasε (Fon) = Xwla (Phla-Phela)

As we saw above the Fon and Phla-Phela dialects are basically very similar linguistically (Capo 1991). From the work of Jennings (1995a) it is clear that virtually all the slaves imported to Cayenne between 1660 and 1673 were Gbe-speaking. It should be noted that the first shipment arrived during the Dutch ownership of Cayenne. With the eighteenth century, reliable figures become available for the sources of the total French slave trade. As Haiti absorbed the largest proportion of this, as Singler points out, it is relevant to examine the proportions from the various regions. In the 1710s the proportion of Gbe-speaking slaves was falling, but these still just represented the largest group (see Table 8).





Norval Smith

Table 8. Slave imports to Haiti (1711–1730)

1711–1720 1721–1730

Senegambia

Windward Coast

Bight of Benin (Slave Coast)

Angola

10,300 22.4% 13,400 16.2%

15,900 34.5% 40,100 48.5%

16,700 36.2% 18,200 22.0%

3,200 6.9% 10,900 13.2%

. Linguistic parallels with Fon . Haitian and Fon Postnominal Determiners I will now examine a case adduced by Lefebvre (this vol.) in support of her Relexification Hypothesis. This is the striking case of the post-nominal definite determiner la in Haitian. In contrast to French, Haitian has post-nominal definite determiners (det). (8) a.

Timoun nan child det ‘The child’ b. liv la book det ‘The book’

vs.

vs.

l’ enfant det child ‘The child’ la livre det book ‘The book’

The Haitian DET does not appear with nouns with a mass/generic interpretation, in French this is compulsory. In Haitian DET also appears in post-clausal position to indicate a shared presupposition among the conversational participants. (9) Mounn nan kraze manchinn nan an man det destroy car det det ‘The man has destroyed the car (as we know).’

det is also required in the structure of Haitian relative clauses, conditional and factive clauses, as well as in some verb-doubling constructions (see Lefebvre this volume for references). In all such constructions DET has no role in French (insofar as the constructions exist in that language). However, Fon turns out to share all the above-mentioned properties with Haitian, e.g. (10) a.

vˇı ɔ´ child det ‘The child’

Differential substrate effects in Saramaccan and Haitian

b. wémà ɔ´ n book det ‘The book’ c. Súnú ɔ´ gbà m´ɔt`ɔ ɔ´ ɔ´ man det destroy car det det ‘The man has destroyed the car (as we know)’

The phonemic material for the Haitian determiner was probably provided by the French post-nominal emphatic deictic marker -là ‘there’ (see for further details Lefebvre, this volume). (11) cet homme-là dem man-there ‘That man’

The stressless French pre-nominal articles were not interpreted by the slaves as meaningful units — sometimes they were incorporated into the lexical entries of the nouns, sometimes not (see Baker 1984). But this incorporation has no semantic import. . Haitian and Saramaccan (and Fon) determiners I will begin my comparison of the strategies adopted by Haitian and Saramaccan with a consideration of the treatment of definite determiners. As we saw above the French articles were not interpreted as Determiners. Definiteness had to be indicated by a consistently salient (i.e. stress-bearing) unit, such as the postposed -là. Similarly, in Proto-Sranan, the equally unstressed English article the was presumably not interpreted as a meaningful unit either, although the incorporation frequent in the French-lexifier and Portugueselexifier creoles does not occur in the English-lexifier creoles. Presumably this is for phonological reasons — in every case, unlike in French, we would get an initial non-salient syllable if English definite articles were incorporated. In English however, although postnominal adverbial modifiers do occur in the DP, their usage is as modifiers further specifying or intensifying prenominal demonstratives, such as this man here, that man there. These also have a very clear local deixis, referring to proximate and distal local deixis respectively. The more frequently-occurring forms this man, that man are firstly semantically weaker, and secondly phonologically less salient. They are phonologically less salient in the sense that, unlike the postnominal modifier in a DP such as that man there, the prenominal modifier in that





Norval Smith

man does not ordinarily bear the primary phrasal stress. They would however still have been preferable to the English definite article forms because of their full (non-schwa) vowels. They are semantically weaker in that while the forms this and that have the primary function of indicating local deixis in English, they are also frequently utilized for other forms of deixis — temporal or discourse — where the choice is not always clearly interpretable in proximate and distal terms. So the postnominal local modifiers are better at expressing the function of local deixis. For these reasons we can begin to see why the prenominal deictic determiners were selected as sources of unmarked definite Determiners in the Surinam creoles. (12) Ndyuka a. a pasi b. a pasi ya c. a pasi de d. a pasi anda

‘the path’ a < da < *that ‘this path’ (lit. ‘the path here’) ‘that path’ (lit. ‘the path there’) ‘yonder path’ (lit. ‘the path yonder’)

In Proto-Sranan there was apparently dialectal diversity — the Proto-Sranan that was a precursor to Saramaccan selected this, in a reduced form di, while that ancestral to Sranan selected that, reduced to da, and in the modern language further reduced to a. Note that while these reductions are exceptional in terms of the normal phonological development of English forms in the Surinam creoles, the shortening found in both these words is fairly typical in function-words. Note that this and that are also found in various functions in the unreduced forms disi and dati in the Surinam creoles, and that in some non-deictic grammatical functions what was disi in the eighteenth century has also become di in the present-day languages (see Table 9). Although Haitian and the Surinam creoles exhibit clear differences in their Determiner systems, I would claim that in both cases there is a greater importance of the Postnominal postion in the DP, as compared to their respective superstrate languages, French and English. I realise that this short review is fairly inadequate, but my main point was to indicate the fruitfulness of the Table 9. Determiners in Fon, Haitian, Saramaccan and Sranan Language

det

Fongbe

Haitian

French

Saramaccan Sranan

English

lexical entry

lexical entry

lexical entry

lexical entry

phonetic string

NP-[ɔ´]D

NP-[la]D

phonetic string (lV˘)NP–la

[(d)a]D NP

ðs NP ðat NP

[di]D NP

Differential substrate effects in Saramaccan and Haitian

parallellism. I am not, in fact, the first to point this parallellism out. Bruyn (1995) refers to the parallels between Sranan and Ewe deictic patterns. What I have done here is indicate possible additional phonological and semantic arguments bearing on the issue. . Saramaccan and Fon (but not Haitian) We shall first list three structural parallels between Saramaccan and Fon, where these two languages appear to share grammatical features that are expressed differently in Haitian. .. Question words In Saramaccan, but not in Sranan and Ndyuka, the normal analytic question words, consisting of a Question element and a Questioned Semantic Unit, are not found in the words for ‘who’ and ‘what’ (Muysken and Smith 1990); see Table 10. Instead we seem to find instead synthetic forms of Gbe provenance; see Table 11. Table 10. Question-words in the Surinam creoles Sranan 1783 Modern

Ndyuka

Saramaccan Structure

who?

(hu-)somma (o-)suma

sama

〈ambε〉

(Wh)-person

what?

(hu-)sanni

san

(on-)san

〈andi 〉

(Wh)-thing

which? (N)

hu-disi

o-di(si) (o-)sortu

on-di

ún-di

Wh-this

where?

hu-ple(si)

(o-)pe

(on-)pe

ún-kamía

(Wh)-place

where? which direction?

(na) hu-sei

(na) (o-) sortu sei

(na-)ún-sε´ na-a-sε´

(loc) Wh-side

(loc) Wh-time

when?

hu-ten o-ten

when? at which hour?

(na) hujuru

what kind of? hu-sorte

on-ten

(na-)ún-té

(na) (o-) sortu yuru

on-yuu

(na-)ún-júu (loc) Wh-hour

(o-)sortu

(on-)soutu

〈andi 〉

(Wh)-sort

why?

fu hu(-sani) (fu) san ede hede/hu-fa

(fu) san ede fu andí édi etc.

(for) Wh-(thing) head/Wh-fashion

how?

hu-fasi, (hu-)fa

o-fasi/fa

fa

(ún-)fá

(Wh)-fashion

how many? how much?

hu-meni

o-meni

on-men

ún-méni

Wh-many

Source: partly after Bruyn (1999); partly L. Adamson, p.c.





Norval Smith

Table 11. The Gbe origins of the Saramaccan words for ‘who’ and ‘what’

who? what?

Saramaccan

Fon

Gen

Ewe

ambε´ andí

mε`(ε´) aní; é-tε´

amε-kε´ nú-kε´

ame-ka nú-ka

The two question words display pre-nasalized stops instead of plain nasals, however the very fact that both forms display the same change rather strengthens than weakens the proposed etymology. Note that the lack of an initial vowel in the Fon word for ‘who’ apparently represents a loss, if we take the Gen and Ewe forms into account. That Saramaccan has retained precisely these particular Gbe forms is apparently not pure chance. They are the only question words whose structure is morphologically opaque. All the other question words appear in both Fon and Saramaccan at least normally as analytic structures. Why do these forms not appear in Haitian? We cannot attach much importance to this point. In various Eastern Gbe dialects an analytic form for the word ‘what’ appears (see (23) above), and Haitian might in any case also have analogically reformed these forms to bring them into conformity with the rest of the question word system. .. Reduplicated adjectives Saramaccan, similarly to the other Surinam creoles, has a system whereby verbs are reduplicated to give a kind of (stative) participial adjective. (13) Participial adjective formation in Saramaccan: pεndε ‘paint’ pεndε-pεndε ‘painted’ booko ‘break’ booko-booko ‘broken’

Fon has precisely the same morphological formation: (14) Participial adjective formation in Fon sú ‘close’ sú-sú ‘closed’ tre ‘tear’ tí-tre ‘torn’

The nature of the reduplication process does differ in details. In Saramaccan — apart from certain tonal effects we will not go into here — we get straight reduplication. In different Gbe dialects and languages the phonological effects vary. In modern Fon we find total reduplication with disyllabic verbs, but with monosyllables certain reductions apply. The vowel of the reduplicative prefix

Differential substrate effects in Saramaccan and Haitian

is consistently i except with rounded stem vowels, where it is u with a high rounded stem vowels, and either i or u with mid rounded stem vowels. This agreement is at first sight perhaps not so striking, in the sense that Fon and the other Gbe languages are not the only West African language with reduplicated deverbal adjectives with a participial function. However, it does seem to be the only language that Saramaccan could reasonably have acquired them from. The only other West African languages which make their presence felt in Saramaccan in more than a couple of lexical items — Kikongo and Twi, lack such morphological structures. Fon and Saramaccan also share the property that such participial adjectives are virtually the only adjectives in the two languages in the strict sense of the word. Virtually all adjectives in European languages are translated by stative verbs in both these languages. There are very few exceptions in either language. Examples of adjectives in Saramaccan are: (15) bunu ‘good’

ogi ‘bad’

These are shown to be adjectives by the fact that the ‘‘verbal’’ copula dε occurs with them. (16) a dε bun]A 3sg BE good ‘He is fine’

a dε ogi]A 3sg BE bad ‘He is bad’

In Fon an example would be: (17) a kpo o kpεví]A dín 2sg be-still BE little int ‘You are still too little’

Participles also occur with the same copulas: (18) Saramaccan a. di wosu dε pεndε-pεndε]A def house BE painted ‘The house is in a painted state’ b. di bata dε lógo-lógo det bottle BE round ‘The bottle is round’

The parallel formation does occur in Eastern Gbe, but in an extremely restricted fashion (Aboh 2001).





Norval Smith

Gun c. ??àvɔ` lɔ´ tò wìwló cloth def BE re-wrinkle ‘The cloth is in a wrinkled state’ d. mìɔ´ ngbá lɔ´ tò cící lamp def BE re-put.out ‘The lamp is in an extinguished state’

Haitian forms participles by conversion (Brousseau, Filipovich and Lefebvre 1986). (19) kwit ‘to cook’ muri ‘to die’ kale ‘to peel’ bukane˜ ‘to smoke’

kwit ‘cooked’ muri ‘dead’ kale ‘peeled’ bukane˜ ‘smoked’

Why reduplication is not used here is not clear. Haitian also lacks the reduplication found in Fon (and Saramaccan) in other cases. For instance verbal nouns are also formed by reduplication in Fon, but by conversion in Haitian. This difference requires further study. .. Contrastive focus The third parallel concerns the marking of contrastive focus in Saramaccan and Fon. Fon marks contrastive focus by fronting, and postposing a morpheme wε to the fronted constituent. Saramaccan makes use of precisely the same strategy (see Smith (1996) for a preliminary study). (20) Contrastive focus in Saramaccan: (Smith 1996) a. mi wε tei εn I foc take it ‘It was me took it’ b. a unu aki wε mi bi ko to you here foc I ant come ‘It was to you here I had come’ c. nɔɔ pindja wε a ta pindja dee sεmbε fu . . . and squeeze foc he prog squeeze the.pl people to . . . ‘And it was squeeze the people he did in order to . . .’

Differential substrate effects in Saramaccan and Haitian

(21) Contrastive focus in Fon: (Lefebvre 1982a, b, c)) a. Masε vì lε wε wá Massè child pl it-is arrive ‘It is the people of massè who have arrived’ b. lɔ´ n wε súnù ɔ´ lɔ´ n jump it-is man det jump ‘It is jump that the man did (not, for example, run away)’

Haitian however makes use of a preposed highlighter se derived from French c’est. (22) Contrastive focus in Haitian (from Lumsden 1990) a. Se Mari Jan te renmen it-is Marie Jean ant like ‘It’s marie who Jean likes’ b. Se ak Mari Jan ap vini It-is with Marie Jean prog come ‘It’s with marie that Jean is coming’ c. Se achte li te achte flè It-is buy 3sg ant buy flower ‘He bought flowers’ (i.e. he didn’t steal them)

Note that the structure assumed for Fon sentence (21a) by Law and Lefebvre (1992) is comparable to that proposed in Smith (1996) for the Saramaccan cases. Abstracting away here from the analysis given by Law and Lefebvre to the cases of Predicate Cleft (i.e. contrastive focus of verbal elements), we have: (23) [CP DP [C¹ [wε` ]C IP ]

where DP is the focussed element in Spec,CP position. Smith does tentatively suggest that there is a deeper constituent-break following wε` in Saramaccan than before it on admittedly unclear evidence from intonation. This is a matter requiring further research, however. The Lexical Learning Theory approach adopted by Bickerton (1989), admitting substrate effects in cases where the actual morpheme is taken from a substrate language, assumes that the syntax of a substrate morpheme is taken over along with the element itself. Let us go along with this position for the sake of the argument. Under this assumption the position of wε` can be explained.





Norval Smith

The Haitian nominal focus construction, with a highlighter se, is claimed by Lumsden (1990) to involve a biclausal structure: (35) [IP pro [I se ] DP [ op IP ]]

where pro is a null pronoun and op(erator) is marked by a relative pronoun under certain circumstances. If Lumsden is correct it is difficult to see any substrate effect when we compare the Haitian strategy to the Fon one. There is of course another parallel which probably should be attributed to a substrate feature. This is the preservation of a copy of a focussed element just in case this is a verb. This is the same in Fon, Saramaccan and Haitian. Whatever the syntactic interpretation of the Haitian element se, it remains the case that the focussed element is preceded by a highlighting element rather than followed by it, as in both Fon and Saramaccan, where the same element wε` is encountered. .. Comparative conclusion In two of our three cases neither Bickerton’s nor Lefebvre’s approaches can explain the totality of the developments involved. In the case of the participial adjectives of the three languages, this is for instance the case. Bickerton cannot have recourse to universals to explain the occurrence of participial adjectives in Saramaccan (and the other Surinam creoles). These are a specifically Surinam creole phenomena, and also occur in the main candidate for a substrate language — Fon (and closely related Gbe lects). If Lefebvre’s Relexification-mechanism was the only mechanism involved in creolization, then why does Haitian not display reduplication in its participial adjectives (and other cases where reduplication is found in Fon)? We note here that reduplicated verbal nouns, as in Fon, are also a feature of all the Surinam creoles (Smith 1990), though their productivity is unclear. It seems that again neither Bickerton nor Lefebvre have an explanation to offer that will account for the Focus phenomena of all three languages. Bickerton, while being able to explain the syntactic parallels between Fon and Saramaccan on the basis of the Lexical Learning Theory, cannot explain why wε` was adopted by the Saramaccan speakers in the first place. Note that this is not a general Gbe feature (the Focus Marker in Ewe is -é, and in Gen -yé), but a specifically Eastern Gbe one. Lefevbre cannot explain why Haitian did not adopt a monoclausal solution parallel to the Fon one, assuming the correctness of Lumsden’s analysis. And a solution involving an extra clause is hardly redolent of relexification.

Differential substrate effects in Saramaccan and Haitian

. Conclusion How do we account for the occurrence of a mixture of Relexification and Universal features in the creole languages of Surinam? Note that the demographic figures indicate neither of the classic scenarios — a single substrate language (36) leading to substrate-derived grammatical structures and superstrate-derived lexicon,11 or several (equally (non-)potent) substrate languages (37) leading to bioprogram-derived grammatical structures and superstratederived lexicon. (36) SUPER + SUB > SUB [SUPER-lex] (37) SUPER + SUB1 + SUB2 + SUB3 > BIO [SUPER-lex]

Instead, we have a mixture of both situations — at least two significant substrate languages, but with one — Eastern Gbe — dominant. In diagrammatic form what we have in Surinam as a starting point is rather: (38) SUPER + SUB1 + SUB2

The result certainly appears to be: (39) BIO + SUB [SUPER-lex]

In other words, a creole showing both substrate features and universalist aspects, with a core vocabulary largely derived from the superstrate language(s). Here we abstract away from the further complication caused by the Portuguese element in Saramaccan. For a more detailed study of this see Smith (1999). We should not be surprised that actual cases of creolization are frequently much more complex than the proponents of unique solutions suggest. However in this case I would submit that both hypotheses have validity. Some aspects of Haitian seem to be interpretable in terms of relexification. Others seem not to be however, like focus. Why is there a difference? It is most unlikely that precisely the same conditions pertained in the formative years of the two creoles. Even if Eastern Gbe speakers were dominant in both colonies at an early period — and the sociohistorical evidence for this is thin for Haiti, the proportions will most likely have been different. This in itself, I would claim, will have caused differences in the eventual creoles. Our brief examination might appear to suggest a greater Eastern Gbe presence among the ancestors of the Saramaccan tribe than on Haiti. The exact extent to which Eastern Gbe features are present in the other Surinam languages is unclear.





Norval Smith

Much work remains to be done. For instance, it appears to be the case that reduplicated participial adjectives are a feature of all the Surinam creoles. On the other hand, the marking of contrastive focus with wε` appears to be absent in Ndyuka (Huttar and Huttar 1994) at least.

Notes * Apologies to the late great Jimi Hendrix. Thanks for assistance and comments to Enoch Aboh, Lilian Adamson, and Margot van den Berg. . Note in connection with the frequent use of the term Kwa by creolists that in the classification of Niger-Congo languages used by Africanists, Kwa no longer includes Yoruba or Igbo. . Note that this term uses the unified orthography of Gbe (Capo 1994). Phela is in fact [Xwela], [Xweda] and corresponds to the European Whydah, Ouidah, Fida, Juda. . In the seventeenth century Doctrina Cristina produced by Spanish Capucin monks (1658) Vodu is used as the translation for ‘God’ (Labouret and Rivet 1929). . As Bickerton (1994) points out language-mixing by relexification can only arise in situations of bilingualism. . Plantation-owners often attempted to reduce their tax-burden by declaring too low a number of adult slaves. . In this period the Portuguese Jews were the largest European group on the Suriname River. . Of these 168 were apparently removed illegally. The King of England ordered these to be returned to Surinam (Rens 1982: note 43). . Note that Ardra, Ardres, Allada and Alada are different forms of the same name; as also Whydah, Fida, Juda, Ouidah, Phelá and [Xwelá]; and Offra, Xwla and Phla. . Jacquin is identified with Godomey by Labouret and Rivet (1929) and Postma (1990). The information in the columns to the right is conditional on this identification. . Aqua/Appa was on the site of Cotonou according to Labouret and Rivet (1929). See also the preceding note. . Note that the present-day Alada dialect is spoken in Porto Novo, and not in Alada. . This scenario has two sub-variants: one where Djutongo creolized on the basis of a PreSranan pidgin and some form of Portuguese; and one where a Pre-Djutongo pidgin first developed from these components, to creolize later. . This has been claimed to be the classic scenario for a Mixed language.

Differential substrate effects in Saramaccan and Haitian

References Aboh, Enoch. 2001. Review of Anne-Marie Brousseau, Réalisations argumentales et classes de verbes en fɔngbè. Paris: SELAF, 1998. [Submitted to Journal of African Languages and Linguistics.] Arends, Jacques. 1989. Syntactic developments in Sranan. D. Litt. dissertation, University of Nijmegen. Arends, Jacques. 1992. ‘‘Historische demografie van Suriname’’. Paper presented at the Creolistiek-colloquium, Universiteit van Amsterdam, March 1992. Arends, Jacques. 1994. ‘‘The African-born slave child and creolization (a postscript to the Bickerton-Singler debate on nativization)’’. Journal of Pidgin and Creole Languages 9: 115–119. Arends, Jacques. 1995. ‘‘Demographic factors in the formation of Sranan’’. In The early stages of creolization, Jacques Arends (ed.), 233–277. Amsterdam and Philadelphia: John Benjamins. Baker, Philip. 1984. ‘‘Agglutinated nominals in Creole French: Their evolutionary significance’’. Te Reo 27: 89–129. Baker, Philip. 1995. ‘‘Motivation in creole genesis’’. In From Contact to Creole and Beyond, P. Baker (ed.), 3–15. (Westminster Creolistics Series, I.) London: University of Westminster Press. Bakker, Peter, Marike Post and Hein van der Voort. 1995. ‘‘TMA particles and auxiliaries’’. In Pidgins and creoles: An introduction, Jacques Arends, Pieter Muysken and Norval Smith (eds.). Amsterdam: John Benjamins Publishing Company, Bickerton, Derek. 1974. ‘‘Creolization, linguistic universals, natural semantax, and the brain’’. University of Hawaii Working Papers in Linguistics 6: 124–141. Bickerton, Derek. 1981. Roots of language. Anne Arbor, Karoma. Bickerton, Derek. 1988. ‘‘Creole languages and the Bioprogram’’. In Linguistics: The Cambridge survey, vol. ii, F. J. Newmeyer (ed.), 268–284. Cambridge: Cambridge University Press. Bickerton, Derek. 1989. ‘‘The Lexical Learning Hypothesis and the pidgin-creole cycle’’. In Wheels within wheels: Papers of the Duisburg Symposium on Pidgin and Creole Languages, Martin Pütz and René Dirven (eds.). Frankfurt am Main: Peter Lang. Bickerton, Derek. 1994. ‘‘The origin of creoles in Surinam’’. MS., Dept of Linguistics, University of Hawaii at Manoa. Bickerton, Derek and Carol Odo. 1976. ‘‘Change and variation in Hawaiian English’’. Vol. i: ‘‘General phonology and pidgin syntax’’. Final Report on NSF Grant No. GS39748, Mimeo, University of Hawaii. Brousseau, Anne-Marie, Sandra Filipovich and Claire Lefebvre. 1986. ‘‘Morphological processes in Haitian creole: The question of substratum and simplification’’. In Projet Fon-Creole Haitien: Etudes syntaxiques, morphologiques et phonologiques (Rapport de recherche pour l’année 1985–1986), C. Lefebvre and J. Kaye (eds.). Montréal: UQAM, 438–470. Bruyn, Adrienne. 1995. Grammaticalization in Creoles: The Development of Determiners and Relative Clauses in Sranan. (Studies in Language and Language Use, 21). Amsterdam: IFOTT.





Norval Smith

Bruyn, Adrienne. 1999. ‘‘Early forms of question words and relativizers in Atlantic English Creoles’’. In St Kitts and the Atlantic Creoles: The texts of Samuel Augustus Mathews in perspective, Philip Baker and Adrienne Bruyn (eds.), 289–314. (Westminster Creolistics Series, 4.) London: University of Westminster Press. Byrne, Frank X. 1987. Grammatical Relations in a Radical Creole: Verb Complementation in Saramaccan. Amsterdam and Philadelphia: John Benjamins. Capo, Hounkpatin C. 1991. A comparative phonology of Gbe. Berlin: Foris. Daeleman, Jan. 1972. ‘‘Kongo elements in Saramacca Tongo’’, Journal of African Languages 11: 1–49. Grant, Anthony. 1994. ‘‘Languages in intimate contact — a study of mixed languages’’. MS, University of Bradford. Hancock, Ian F. 1987. ‘‘A preliminary classification of the anglophone Atlantic creoles with syntactic data from thirty-three representative dialects’’. In Pidgin and creole languages: Essays in memory of John E. Reinecke, Glenn G. Gilbert (ed.), 264–333. Honolulu: Univerity of Hawaii Press. Hartsinck, Jan J. 1770. Beschrijving van Guiana, of de Wilde Kust in Zuid-Amerika, (. . .). Amsterdam: Gerrit Tielenburg. Herlein, J. D. 1718. Beschrijvinge van de volks-plantinge Zuriname. Leeuwarden: Injema. Huttar, George. 1986. ‘‘Kikongo, Saramaccan, and Ndjuka’’. In Language in global perspective: Papers in honor of the 50th anniversary of the Summer Institute of Linguistics 1935– 1985, Benjamin F. Elson (ed.), 563–586. Dallas: Summer Institute of Linguistics. Huttar, George and Mary Huttar. 1994. Ndjuka. London: Routledge. Jennings, William. 1993. La genèse du cayennais: étude de sa démographie et de l’évolution de son système verbal. M. A. thesis, University of Auckland. Jennings, William. 1995a. ‘‘The first generations of a creole society: Cayenne 1660–1700’’. In From contact to creole and beyond, Philip Baker (ed.), 21–40. (Westminster Creolistics Series, 1.) London: University of Westminster Press. Jennings, William. 1995b. ‘‘Saint-Christophe: Site of the first French creole’’. In From contact to creole and beyond, Philip Baker (ed.), 63–80. (Westminster Creolistics Series, 1.) London: University of Westminster Press. Labouret, Henri and Paul Rivet. 1929. Le royaume d’Arda et son evangélisation au XVIIe siècle. (Travaux et mémoires de l’Institut d’Ethnologie, Université de Paris VII). Paris: Institut d’Ethnologie. Law, Paul and Claire Lefebvre. 1992. ‘‘On the relationship between predicate cleft and clausal determiners’’. Travaux de Recherche sur le créole haïtien 9: 35–60. Lefebvre, Claire. 1986. ‘‘Relexification in creole genesis revisited: The case of Haitian Creole’’. In Substrata versus universals in creole genesis, Pieter C. Muysken and Norval S. H. Smith (eds.). Amsterdam and Philadelphia: John Benjamins. Lefebvre, Claire. 1992a. ‘‘Agr in languages without person and number agreement: The case of the clausal determiner in Haitian and Fon’’. Travaux de recherche sur la créole haïtien 8: 1–33. Lefebvre, Claire. 1992b. ‘‘Tryptique sur AgrO en créole haïtien’’. Travaux de recherche sur la créole haïtien 8: 35–47. Lefebvre, Claire. 1992c. ‘‘Double object constructions in Fongbe’’. Travaux de recherche sur la créole haïtien 9: 1–34.

Differential substrate effects in Saramaccan and Haitian

Lumsden, John. 1990. ‘‘The biclausal structure of Haitian clefts’’. Linguistics 28: 741– 760. McWhorter, John H. 1995. ‘‘Sisters under the skin: A case for genetic relationship between the Atlantic English-based creoles’’. JCPL 10: 289–333. McWhorter, John H. 1999. ‘‘A creole by any other name: Streamlining the terminology in creolistics’’. In Spreading the word: Papers on the issue of diffusion of Atlantic creoles, M. Huber and M. Parkvall (eds.), 5–28. London: University of Westminster Press. Muysken, Pieter C. 1976. The importance of being halfway between. MS, University of Amsterdam. Muysken, Pieter C. 1980. ‘‘Sources for the study of Amerindian contact vernaculars in Ecuador’’. Amsterdam Creole Studies 3: 66–82. Muysken, Pieter C. 1981. ‘‘Halfway between Quechua and Spanish: The case for relexification’’. In Historicity and variation in creole studies, Arnold Highfield and Albert Valdman (eds.), 52–78. Ann Arbor: Karoma Publishers. Muysken, Pieter C. and Norval Smith. 1990. ‘‘Question words in pidgin and creole languages’’. In Linguistics 308, P. A. M. Seuren and S. S. Mufwene (eds.), 883–903. (Special number, Issues in Creole Linguistics.) Perl, Matthias. 1995. ‘‘Introduction to Part II’’. In Early Suriname creole texts, J. Arends and M. Perl (eds.), 243–250. (Bibliotheca Ibero-Americana, 49) Frankfurt: Vervuert; Madrid: Iberoamericana. Postma, Johan. 1990. The Dutch in the Atlantic Slave Trade, 1600–1815. Cambridge: Cambridge University Press. Price, Richard. 1983. First Time: The historical vision of an Afro-American people. Baltimore: The Johns Hopkins University Press. Rens, L. L. E. 1953. The Historical and Social Background of Surinam’s Negro-English. Amsterdam: North-Holland. Rizzi, Luigi. 1990. Relativized Minimality (Linguistic Inquiry Monographs, 16). Cambridge. MA: MIT Press. Schumann, C. L. 1778. Saramaccanisch–Deutsches Wörter-Buch. [Published in: Hugo Schuchardt, Die Sprache der Saramakkaneger in Surinam (Verhandelingen der Koninklijke Akademie van Wetenschappen te Amsterdam, afdeeling letterkunde nieuwe reeks XIV, 6). Amsterdam: Johannes Müller, 1914.] Schumann, C. L. 1783. Neger-Englisches Wörter-Buch. [Edited in: André Kramp, Early creole lexicography: A study of C. L. Schumann’s manuscript dictionary of Sranan. Unpublished D. Litt. dissertation, State University of Leiden, 1983.] Singler, John V. 1993. ‘‘African influence upon Afro-American language varieties: A consideration of sociohistorical factors’’. In Africanisms in Afro-American language varieties, Salikoko S. Mufwene (ed.). Athens, GA: University of Georgia Press. Smith, Norval S. H. 1987. The genesis of the creole languages of Surinam. Ph.D. thesis, University of Amsterdam. Smith, Norval S. H. 1990. ‘‘Deverbal nominalization in Sranan: A search for regularities’’. In Unity in diversity: Papers presented to Simon C. Dik on his 50th birthday, Harm Pinkster and Inge Genee (eds.), 265–277. Dordrecht: Foris.





Norval Smith

Smith, Norval S. H. 1996. ‘‘Focus-marking wε` in Saramaccan: Grammaticalization or substrate’’. In Changing meanings, changing functions: Papers relating to grammaticalization in creole languages, Philip Baker and Anand Syea (eds.), 113–128. (Westminster Creolistics Series, 2.) London: University of Westminster Press. Smith, Norval S. H. 1999. ‘‘Pernambuco to Surinam 1654–1665? The Jewish slave controversy’’. In Spreading the word: Papers on the issue of diffusion of Atlantic creoles, M. Huber and M. Parkvall (eds.), 251–298. London, University of Westminster Press. Smith, Norval S. H. 2001a. ‘‘The creole languages of Surinam — past and present’’. In Language atlas of Surinam, J. T. G. Arends and E. Carlin (eds.). Smith, Norval S. H. 2001b. ‘‘Reconstructing Proto-Caribbean Pidgin English’’. Paper presented at the Workshop ‘‘Pidgins: Their nature and significance’’, University of Westminster, April 2001. van den Berg, M. 2000a. ‘‘Extracts from various early Surinam archives’’. MS. van den Berg, M. 2000b. ‘‘ ‘Mi no sal tron tongo’: Early Sranan in court records 1667– 1767’’. M. A. thesis, University of Nijmegen. van den Linde, Jan M. 1966. Surinaamse Zuikerheren en Hun Kerk: Plantagekolonie en handelskerk ten tijde van Johannes Basseliers, predikant en planter in Suriname, 1667– 1689. Wageningen: Veenman. van Stipriaan, A. 1993. Surinaams Contrast: Roofbouw en overleven in een Caraïbische plantagecolonie, 1750–1863. Leiden: KITLV Press. Veenstra, Tonjes. 1996. Serial Verbs in Saramaccan: Predication and creole genesis (HIL Dissertations 17). Den Haag: Holland Academic Graphics. Voorhoeve, Jan and Ursy Lichtveld (eds.). 1975. Creole drum: An anthology of creole literature in Surinam. New Haven: Yale University Press. Warren, George. 1667. An impartial description of Surinam upon the continent of Guinea in America. London: Thomas Osborne. Wekker J. 1991. ‘‘Surinaams plantagewezen: Een kwestie van aantallen’’. OSO, Tijdschrift voor Surinaamse Taalkunde, Letterkunde en Geschiedenis 10: 71–85. Wooding, Chris J. 1972. Winti: Een Afroamerikaanse godsdienst in Suriname. Meppel: Krips Repro.

Chapter 3

Language intertwining Its depiction in recent literature and its implications for theories of creolisation* Anthony P. Grant

.

Introduction

Language intertwining (formerly ‘‘language interweaving’’) is one of the newer phenomena to be accorded independent recognition in the continuing study of processes and results of contact between languages. It is fitting that it should be discussed at a workshop at the University of Amsterdam, where so much work has been conducted on the topic as a theoretical construct, and on languages which exemplify it. The cardinal text for the study of the mechanics and results of language intertwining is the collection of sketches of fifteen such languages in Bakker and Mous (1994). Reference will also be made to Norval Smith’s introduction to the list of pidgins, creoles and mixed languages (Smith 1995: 331–336), in which he provides a useful guide to the differing forms of mixed languages. The idea that a language could consist of an internally consistent combination of large bodies of elements and features from two or more languages was already old in 1876, when J. C. Clough published his essay in book form (Clough 1876). However, his examples include pidgins, such as Chinook Jargon, genuine mixed languages such as Angloromani, and languages with much external borrowing, such as Ottoman Turkish. There has been much blurring of the categories of pidgins and creoles on the one hand and mixed and intertwined languages on the other. Just as it is true that no single sociohistorical mise-en-scène will fully account for the genesis of all creoles, so mixed languages have evolved in many ways, of which language intertwining is only one, albeit a very productive one. This chapter concentrates on processes and results of language intertwining and compares them with those of creoles, seeking to find parallels and divergences in the paths of development.



Anthony P. Grant

. Mixed languages and language intertwining Having used such terms as mixed language and language intertwining fairly freely so far, I will now present usable definitions of them. My definition of the first follows Thomason and Kaufman (1988). Mixed languages are those languages which cannot be shown, by an analysis of their components, to derive straightforwardly from a single genetic parent, and which cannot therefore be classified as belonging to a single language family or subfamily. The classic example of languages which can be classified according to genetic linguistic criteria are the Romance languages. These arose from Latin, which itself was one of several Italic languages which had sprung as a subgroup from Proto-Indo-European. The Italic languages changed over time, and one evolved into Latin, became the lingua franca of a huge empire and gave rise to descendants such as French, Spanish, and Rumanian, while others such as Faliscan had no descendants but died out as their speakers shifted to Latin. This process of development is not the case with creole languages, where the bulk of the vocabulary may be derivable from one language, but where the structure of the lexifier language has not been similarly transmitted. Instead a new structure has arisen, even if this usually consists of structural elements of the lexifier language which have been (re)grammaticalised, as is the case with the Creole French of Mauritius. The structure of Mauritian has very few features which are most readily attributed to typological transfer from the Bantu and Malagasy languages of the slaves who were originally brought to settle Mauritius in the eighteenth century. For the most part the structural typology of Mauritian is not particularly close to Malagasy, Bantu or French. Such languages are not genetically classifiable according to Thomason and Kaufman (1988), and thus count as mixed languages in a broad sense. In one sense, any language can be said to be a mixed language, since all languages have borrowed from others. This usually involves transfer of items from the lexicon of one language into that of another, especially if these items relate to entities which are culturally new to the speech community. But some languages borrow further features: the transfer of phones via loanwords and the development of new phonemic contrasts are assisted by heavy lexical borrowing, as is the transfer and (semi-)productive use of elements of derivational morphology. Borrowing of inflectional morphology occurs more rarely, as does borrowing of core vocabulary, the semantic delimitation or complete replacement of an original lexeme by a borrowed item, while partial or complete replacement of an inflectional morpheme by a borrowed one is rarer still.

Language intertwining

Nevertheless, there is probably nothing — lower numerals, personal pronouns, verbal affixes, items meaning ‘eye’, ‘water’, ‘man’ or whatever — which is completely impervious to borrowing. The etymological structure of the lexicon of a language can tell us nothing certain about its genetic affiliations, although it may impute a great deal about the external history of that language. The proportion of elements from different sources in the inflectional (and to a lesser extent derivational) morphology of the language and in its core lexicon, however the latter is defined, are the deciding factors. The term mixed language has been used excessively in the last century or so. Languages such as Chamorro, Albanian, Armenian, Swahili, Yiddish and of course English have been adduced as mixed languages, although it is clear that in each case the vast bulk of the core lexicon belongs with the grammatical core of the language as the genetic components, while other elements in the lexicon have accumulated over time, entering first as loanwords and then, by their strength in numbers and the prestige of their source languages among speakers of the borrowing language, affecting the phonology and derivational morphology of the language too. The fact that there are more than twice as many words in English which have entered the language from French or Latin than there have been transmitted into Modern English from Old English does not make English a Romance language, any more than the facts that one-third of the Germanic vocabulary in English is without etymology in Indo-European, and that the structure of Germanic has diverged considerably from Indo-European as the result of substratum influences from speakers of unknown languages of northern Continental Europe vitiate the claims of English to be an Indo-European language. European Romani could be added to the list of wrongly-claimed mixed languages. Kalderash, the most widely spoken variety of Romani, contains maybe four times as many morphemes from Rumanian as it does from the genetic Indic component. In addition it has been overlaid with morphemes borrowed from other languages, principally Persian, Armenian, Greek, South Slavic and Hungarian, which together are almost as numerous as the inherited Indic elements. Nevertheless it is still possible to construct sentences containing only Indic morphemes, since these comprise much of the basic vocabulary and most inflectional and productive derivational morphs. Rotuman, spoken in the southwestern Pacific (Biggs 1965, Pawley 1996), comprises no less than three strata of elements from the Oceanic subgroup of Austronesian. As Biggs showed, they can be distinguished from one another on diachronic phonological grounds. One is the Oceanic genetic element,





Anthony P. Grant

another is borrowed from Tongan or an ancestor language of the Tongic subgroup of Polynesian (ancestral to Tongan, Niuean and East ‘Uvean), while a third, which predates the Tongic element, is derived from a language of the Samoic subgroup, possibly Samoan itself, or East Futunan. The Tongic layer of borrowings is almost as numerous as the genetic elements in Rotuman, but this of itself does not make Rotuman a mixed language. The wider affiliations of Yapese of Micronesia were long rather mysterious. It exhibits several large layers of elements from Austronesian languages (Ross (1996a) finds five), including non-Oceanic Palauan and Oceanic Ulithian, as well as loans from Tagalog, German, Spanish, Japanese and English, but the well-hidden genetic layer shows it to be an Oceanic language after all. But sometimes judgments on the genetic affiliation of a language cannot be made on structural criteria. This is the case with many creoles, where the inflectional morphology of the lexifier language has not been transmitted and where grammatical relations are expressed through particles or through the use of fixed word orders. It is also the case with Vietnamese, according to SEALANG-L (Southeast Asian Languages List) exchanges on the Internet in the summer of 1994. The vocabulary of Vietnamese derives from MonKhmer, but there are a number of independently-developed forms, many shared by Vietnamese with its sister-language Muong, a number of early loans from Tai languages, a huge superstrate of forms from Chinese languages over a period of more than a millennium, and finally some borrowings from French and English. However, it lacks the inflectional morphology typical of Mon-Khmer languages. It uses word-order as the means of identifying arguments in sentences, and concatenation rather than infixation in its derivational morphology, and in contrast to Mon-Khmer languages it has lexical tone and mostly monosyllabic words. The lexicon of Vietnamese makes its Mon-Khmer affinities clear, although the typological and grammatical evidence for its affinity that one can find in other Mon-Khmer languages is generally not available. Perhaps a better way of defining a mixed language would be to state that it is one where overt elements — free or bound morphemes — deriving from two or more languages are obligatorily combined at sentence or paragraph level, wherever there is a sentence which contains both an overt noun phrase and an overt verb phrase. I emphasise ‘‘overt’’ because mere typological similarity between two languages is not conclusive proof of either genetic affiliation or close linguistic contact. Or we may say that a mixed language is one where essential grammatical and lexical elements are obligatorily encoded

Language intertwining

by items from two or more languages. This would serve to distinguish mixed languages from instances of extensive code-mixing. If Language X derives its pronouns from Language A and its verb morphology from Language B, then these form part of the grammar of Language X. One is not speaking Language X if one uses pronouns from Language B in an otherwise Language X sentence, since Language B pronouns do not form part of the grammatical structure of Language X. One can find loopholes in most definitions but the key point is that overt elements from two or more languages are foregrounded in the basic structure of a mixed language. There are languages whose genetic affiliations are impossible to untangle because their core structure and core lexicon are so thoroughly mixed that it is impossible to decide which is the genetic element and which are the borrowed items. Several authors, such as Capell (1976) and Wurm (1978), have discussed languages in Melanesia which combine elements of Austronesian (hereafter AN) and non-Austronesian (NAN, formerly ‘‘Papuan’’) origin. Such languages are Magori and Maisin in southeastern Papua New Guinea, and Ayiwo, Nanggu and Santa Cruz on the Reefs and Santa Cruz islands in the northern Solomons. These are internally-consistent languages, whose speakers sometimes know no other. Needless to say, the languages in Papua New Guinea and those in the Solomons derive from amalgamations of features of quite different AN and NAN languages. They probably arose over long periods of time as the result of intimate language contact and widespread bilingualism, first additive and then replacive in nature. They combine AN and NAN elements at the basic and nonbasic lexical level, in addition to elements of unknown origin. The contents of the various grammatical subsystems also derive from different languages, for example pronouns from NAN and nominal morphological marking from AN, with NAN-style noun-classes on nouns of NAN and AN origin, classes created using AN-derived noun-class markers. In such cases it is impossible to tease out the original genetic component. There are also languages in Papua New Guinea, such as Takia on Karkar Island, which retain Austronesian morphemes but whose syntax, and whose grammatical use of these morphemes, is non-Austronesian, as a result of generations of bilingualism in a non-Austronesian language, in this case Waskia (Ross 1996b). Speakers of Takia are no longer bilingual in Waskia, but Takia syntax mirrors Waskia syntax very closely. The language of the Nahalis, a tribe in Orissa, India (Kuiper 1962) shows considerable mixture. Most of the verbal conjugation derives from the Munda language Korku. Other parts of the grammar and a fair proportion of the





Anthony P. Grant

lexicon are of unknown derivation. The remainder of the Nahali material derives from two layers of Munda loans, the first from a variety which has not been more precisely identified and the second from Korku, while there are numerous loans and grammatical affixes from the nearby Dravidian language Kurux, and a top layer of lexical items from Indic Oriya, Bengali and Hindi. Here we see mixture within both structural and lexical realms. Something similar, though less dramatic, is also true of certain creoles, Papiamentu of the Netherlands Antilles derives most of its lexicon from Spanish, with some elements from Arawakan and later accretions from Dutch, French and English, though it also contains a component, comprising maybe a few score elements, deriving from some form of (possibly creolised) Portuguese. This is principally found in the core vocabulary, although there are also some grammatical features, such as most of the personal pronouns and the irrealis marker lo, which is far more readily derived from Portuguese logo than Spanish luego, both meaning ‘soon’. In addition, some items show features of both Portuguese and Spanish, for instance hariña ‘flour’, which resembles both Portuguese farinha and Spanish harina (in which the initial aspirate was formerly pronounced). There are also a few core elements of African origin, most notably the plural marker and third person plural pronoun nan (compare Bini íràn: Norval Smith, personal communication), and a small amount of Kwa and Bantu cultural vocabulary. It is not certain that all the terms attributed to Portuguese derive from Portuguese (Grant 1996 ms), since the phonologies of seventeenth-century Spanish and Portuguese were less dissimilar than they are now. Many words which are now pronounced quite differently in the two languages were then almost or fully homophonous. Possibly Papiamentu comprises a Spanish overlay on an (?Afro-)Portuguese proto-pidgin. . Specifics of language intertwining The term language intertwining, coined by Pieter Muysken and Peter Bakker, describes the process of evolution of a certain type of mixed language. Language intertwining typifies a mixed language which has a grammatical structure derived from one language and a lexicon derived from another. Some simplification may occur in both subsets of the language, as rule reduction in the former and morphemic loss in the latter, but the structural component of an intertwined language is not pidginsed. An important consideration here is that of which levels of a language are important for its genetic classification. English spoken with Welsh intonational

Language intertwining

patterns but with a grammatical structure and lexicon identical to other forms of English is not a Celtic language. It does not matter whether the speaker is a native speaker or a learner of Welsh, or even completely ignorant of the language. Such English, even if it is spoken as the first or only language, is not a mixed language or an example of language intertwining. Nor is English as spoken by a native speaker of French, Igbo or Tuvaluan, who has acquired few or none of the English phones which do not occur in their first language. What we have is simply English with a heavy foreign accent. Since they are mixed languages, intertwined languages cannot always be genetically classified, as their structures and lexica derive from different sources. The donor languages may be related. Stedsk or Town Frisian can be claimed for West Germanic, because it derives from Dutch and West Frisian. Both of these are West Germanic languages, although they belong to different subbranches, so that one can usually etymologise elements and attribute them to Dutch or Frisian. Core components may be unrelated: Javindo, once spoken in Semarang, Java, combines elements of Austronesian Javanese and IndoEuropean Dutch. Intertwined languages can be spoken by people who do not know either of the languages involved, and they are not intelligible to people who know one (or even both) of the languages. Thus most of the speakers of Michif, the Cree-French mixed language of Western Canada, know neither Cree nor French, though they all share English. Cree-French bilinguals cannot understand Michif. An important statement of the characteristics of intertwined languages by one of the major workers in the field is Bakker (1996). Bakker’s approach to the development of these languages is sociologically oriented: he distinguishes between those which are used as secret languages by largely nomadic groups, such as cryptolects involving Romani, and those serving as speech markers of a new emerging ethnic identity among members of social groups which resulted from mixed marriages. In both cases the underlying motivation is the same. The people in question wish to be able to speak differently from the people around them, even if in the case of secret intertwined languages such as Shelta the very existence of the language may be hidden. They may descend from a long societal tradition of ethnically mixed marriages and may not wish to identify with either of the groups from which they are descended (or those groups in question may prevent this identification). They may find that using words in their speech which people in the surrounding community do not understand





Anthony P. Grant

is a good way of keeping secret their business, be it curing or horse-trading, without drawing attention to themselves by appearing to speak a foreign language. In any case, they decide to use a mixed language in which the structure derives from the language which they know best, often the language which they share with the rest of the community and the one which they speak every day, while the vocabulary derives from elsewhere. This can sometimes be from a language which the group developing the intertwined language formerly had as their main language, but which they replaced through language shift, thus losing familiarity with its structure, although they may recall many words and unanalysable phrases. The language providing structural elements is Language A, the structural donor language, the one providing the lexicon Language B or the lexical donor language. A simple dichotomous model for language intertwining can be proposed, as it appears that only two languages are mainly involved when language intertwining takes place, although these languages make different but essential contributions to the new language. One might say as a rule of thumb that if the translation of a sentence such as ‘the man saw the dogs’ requires the use of elements of two languages, especially if there is a structure versus lexicon split, then one has a case of language intertwining. The mechanics of language intertwining, as seen from Bakker’s study, are as follows. The bound grammatical morphemes come from A, free lexical morphemes from B. Free grammatical morphemes come from either A or B, or from A and B together (that is, either a mixture of forms within these categories, or else parallel forms). The grammatical system (phonology, including morphophonology, and morphosyntax) derives from the language which the devisers of the language know best; and which they want their language to sound like. In this formulation the new language, Language X, differs from Language A in regard to its free lexical morphemes and some or all of its free grammatical morphemes. It is essentially wholesale relexification, the substitution of one morph for another, without affecting the semantics of the morph in question. Indeed Maarten Mous referred to this at the 1995 workshop as paralexification Grammatical simplification of the structure of Language A has not taken place. What has been discarded in the formation of Language X is a greater or lesser amount of the lexicon of Language A. It is easier to acquire a new lexicon and to insert that into a previouslyexisting structure than to acquire a new structure and to use that with one’s

Language intertwining

everyday lexicon. People who have not learned second languages by formal means often equate a knowledge of languages with a knowledge of many individual words and are less concerned with how words are put together in the language being learned in order to generate grammatical sentences. They may expect to be understood by stringing the words of the new language together, using the grammar of their first language or else a jumble of words, without productive control of the grammar of the new language. This is the converse of language death, where productive command of the structure of a language may weaken even though much of the lexicon is still retained.

. Patterns and types of intertwined languages The ideal intertwined language would have 100 per cent structural replication from Language A and 100 per cent lexical replication from Language B, apart maybe from free grammatical morphemes — adverbials, negators, demonstratives, numerals, disjunctive pronominals — of which there would be two equal sets, each from the two parent languages. Such perfect intertwined languages do not appear to exist. At least, the perfect instance of language intertwining has yet to be recorded. We always seem to find some slippage — lexical elements in the form of content words from language A, or else a lexical or occasionally structural component from a third language. What we find are approximations to this ideal, and it is in the ways in which they diverge from Platonic perfection that we can learn much about the development of intertwined languages. Probably the nearest instance to this ‘‘perfection’’ is Media Lengua, spoken in at least two dialects in parts of central and southern Ecuador (Muysken 1997). As its name suggests, it is a ‘‘halfway’’ language. The morphosyntax and phonology are Quechua, whereas the stems derive from Spanish. Spanish stems participate in the Quechua-derived inflectional and derivational morphological processes of the language, and the semantics of these stems are those of the Quechua equivalents which have been relexified by Spanish forms. The morphosyntax is not simplified. This language is not made up ex nihilo every time it is spoken, since it includes Spanish-derived stems, such as nuwabi‘there is not’, which are not now found in Spanish (Spanish now has no hay). This is clearly a stable part of an independent language system. Younger speakers of Media Lengua often know Spanish but not Quechua. The reverse is true of older speakers. Even so they speak the language much the same way as





Anthony P. Grant

middle-aged people, who may know Quechua, Spanish and Media Lengua. Another close approximation to perfect language intertwining is what Norval Smith (Smith 1995: 333) called Symbiotic Mixed Languages (formerly known as Mixed Replacive Languages). These can be distinguished from ordinary mixed languages, or from Lexical Replacive Languages, where the lexicon is the only thing which differentiates the language from another language with which it shares the structure. The latter can serve as the sole language of a community, but Symbiotic Mixed Languages coexist with the language which has provided their structure. This structural donor language is generally the mother tongue of the users of the Symbiotic Mixed Languages. Sometimes, but less often, these users also know the lexical donor language. In the case of secret languages. the lexicon may derive from several sources. In Symbiotic Mixed Languages part of the lexicon derives from one language, while the grammatical structure and the rest of the lexicon come from another. The amount replaced can be considerable, though it is not the whole lexicon. It can include function words such as adverbials and pronouns, free grammatical morphemes and free lexical morphemes (especially the commoner of the latter). Since users of the Symbiotic Mixed Language know the lexicon of the structural donor language, except in those cases where lexenes encode concepts unique to the group (adlexification of a pre-existing vocabulary), these are cases of paralexification. What distinguishes them from the structural donor language is the paralexicon. The majority of these languages are used for cryptolectal purposes among members of in-groups. A simple example is Senkyoshigo, used by American Mormon missionaries in Japan, which uses Japanese content words in an English grammatical framework and with English phonology (Smout 1988). This dichotomy between the sources of the lexicon and structure also seems to be the case with Angloromani, a mixed language spoken by the majority of Romanichals (the self-ascription of British Gypsies) in Britain, the USA and Australia. These people reached Britain in the early sixteenth century at the latest. Some productive use of Romani grammatical structure is found in earlier records of Angloromani, for example in Smart and Crofton (1875), and Romani grammatical structure and lexicon remained in use among Gypsies in Gwynedd, North Wales, into this century (Sampson 1926) (this group came to North Wales from Somerset in the 18th century). Nowadays little or no Romani inflectional morphology remains in Angloromani, except for the retention of feminine and masculine adjectival endings in the variety used in South Wales, probably due to the fact that Welsh makes the same

Language intertwining

distinction. Derivational morphology has fared better: a few common nominalising suffixes remain productive. In modern Angloromani, Romani words, including many free-standing function words, are inserted into an English grammatical framework and are pronounced with English phonology, thus losing features such as phonemic aspiration which were present in Welsh Romani till well into this century. Theories of the genesis of Angloromani, including the possibility that its earlier stages reflect uncompleted language shift (the so-called ‘‘U-turn hypothesis’’) and its relations with Welsh Romani and other varieties, are discussed in Grant (1998). Typologically Angloromani is one of the ‘‘Para-Romani’’ languages, in which Romani-derived lexicon is used with the structure of the languages of the areas where the Roma or Gypsies have settled. Varieties have been recorded from the Iberian Peninsula, Scandinavia and Armenia. They are surveyed in Boretzky and Igla (1994). Speakers of Angloromani are raised speaking English and acquire Angloromani in the course of their late childhood. What sets Angloromani apart from English is its lexicon. The individual user knows between 300 and 800 items of this, of which at least four-fifths derive from Romani. The rest mostly come from Shelta (the cryptolectal language of Irish Travellers), dialectal English, English Cant, and backslang. They are not usually known to nonspeakers of Angloromani. Discourse considerations govern the degree to which Angloromani words are substituted for their English equivalents in the workaday or cryptolectal speech of Romanichals (Hancock 1986). Symbiotic Mixed Languages can possess structural features which their source languages lack. For instance, both Scottish Traveller Cant (a partial derivative of Shelta, with extensive borrowing from Romani and from English Cant) possesses an unusual strategy for forming personal pronouns not found in the lexifier or structural donor languages. These are formed by adding a possessive to a semantically empty noun, for example Scottish Traveller Cant my nawkin ‘I’. A similar structure appears to occur in Shelta and other secret languages. It is important to realise that while intertwined languages may start out with contributions from two languages, they are also as subject to change and innovation as other languages. Different varieties of Para-Romani have adopted as personal pronouns differing cases of Romani pronouns: possessive or ablative in Swedish Para-Romani, comitative in Brazil and in Caló (Spanish Para-Romani), locative for the form in Angloromani, and dative for ParaRomani varieties in Portugal and Andalusia, while Basque Para-Romani chooses either the dative or the nominative.





Anthony P. Grant

There are Symbiotic Mixed Languages involving numerous other languages, including Spanish, Dutch, Farsi and Turkish, which are combined with lexica deriving from various sources, these often including a number of words which have been invented ex nihilo. Indeed such languages can evolve from any language where the bulk of the lexicon is expressed by free-standing morphemes rather than by bound morphemes. The chance of some sort of intertwined language arising from such languages as Inuktitut, Caddo of Oklahoma, or the Salishan languages of the Pacific Northwest, where rich derivational (and inflectional) morphology, often including affixes with specifically lexical meanings, is found, is presumably more limited. In such cases native speakers’ perceptions of morpheme boundaries would come into play. In some cases the special lexicon in this mixed language has no existence outside the language, either because the combination of languages which provide the lexicon is unique to this mixed language, or because of the nature of the elements themselves. For instance, Rotwelsch, used as a secret language in a number of forms largely within the German-speaking area, comprises a lexicon of items from Sinto (German Romani), Hebrew, Latin, newer creations, German words used with different senses, and other sources. This lexicon has no independent existence outside Rotwelsch; this is especially true of forms which are found in Rotwelsch but in no other language. It is not a relic of some lost language once spoken in Germany but now extinct. . Some case histories of language intertwining Bakker and Mous (1994) contains descriptions of fifteen individual mixed languages, or groups of similar languages. The editors recognise that other such languages exist, for example in Melanesia, Australia and Central Asia, but were unable to obtain sufficient documentation of these languages for their book. The two languages which are most problematic for Bakker’s theory of language intertwining are Michif, the Cree-French language of Alberta, Saskatchewan and Montana (Bakker 1997), which combines French noun phrases with Plains Cree verb phrases, and Mednyj Aleut, the Aleut language of Copper Island (Russian Mednyj Ostrov), which combines Russian finite verb morphology with Western Aleut lexicon, non-finite verbal and other sections of morphology (Thomason 1997a). Others fit more neatly into the division between lexicon and structure. In Michif, now spoken by a dwindling number of Metis (people of mixed Amerindian and European descent, who coalesced as an ethnic group in the

Language intertwining

early nineteenth century), both verbal stems and their morphology derive from Plains Cree, while noun stems and most of their adjuncts (though not demonstratives) derive from French, with an increasing number of nouns being replaced by their English equivalents. The split is not perfect, however. A few nouns derive from Cree, the related Algonquian language Ojibwe or the Siouan language Assiniboine. About twenty verbs, relating mostly to practices introduced by Whites, derive from French, as does the copula. Most French verbs are integrated into the Plains Cree conjugation with an adaptor suffix. Cree obviative and locative markers are sometimes found on French nouns. French nouns are classified both as animate or inanimate, as in Cree, and masculine or feminine, as in French. The demonstratives used before or after nouns derive from Cree, and reflect the Cree feature of animacy and the proximal/distal distinction, more overt in Cree than in French, in addition to distinguishing singular and plural nouns, as both Cree and French do. Post-nominal adjectives in Michif derive from French but are invariable; pre-nominal adjectives, also from French, modify for the number and gender of the following noun. The Michif verbal system is a little simplified by comparison with that of Plains Cree, which is itself simpler than those of other types of Cree, but its morphology and semantics are thoroughly Algonquian and more complex than the French verbal system. In Mednyj Aleut the division of labour between the two sets of structural elements is different from that of Michif. Grammatical subsystems have been taken from two languages, as in Michif, but from different parts of the morphosyntax. The Russian influence touches the finite verbal morphology, including the borrowing of personal pronouns and auxiliaries. Some kinship terms and much aculturational vocabulary have also been borrowed from Russian. The rest of the language derives from Western Aleut, and the phonology of Mednyj Aleut is closer to Western Aleut than it is to Russian. Here the verbal system has not been simplified; rather, it has largely been abandoned in favour of a simpler one from another language. The other languages profiled in Bakker and Mous (eds. 1994) fit much more snugly into the division between lexicon and structure. Nurse (1994: 221) comes to the conclusion that Ilwana of northern Kenya is a sister of Swahili heavily influenced by extensive borrowing from the Cushitic language Orma. Simiarly Maltese is shown to be a variety of North African Arabic which has borrowed heavily from Sicilian, standard Italian, and latterly English, and this has involved transfer of derivational morphology and some associated morphophonology in addition to lexicon.





Anthony P. Grant

Kimwani of northern Mozambique appears to be the result of imperfect shift towards Swahili several centuries ago by speakeers of Makonde. Here two Bantu languages are involved, and again lexicon and morphophonology have been influenced, this time by Makonde. Imperfect language shift also characterises Stedsk, in this case the imperfect acquisition by Frisian townspeople of Dutch in the sixteenth century. This gave rise to a speech form which stabilised and then developed norms of its own. The classical pattern of language interwtining, with structure from one language, lexicon from another, and some slippage of lexicon from the structural donor language, works quite well in the case of Petjo (Malay structure with a two-thirds Dutch lexicon and the rest from Malay), Javindo (Javanese structure, Dutch lexicon), and the long-extinct diplomatic language AmarnaAkkadian (East Semitic Akkadian lexicon and West Semitic Canaanite structure). It is true too for Island Carib Men’s language, in which some lexicon and a couple of structures derive from Carib, while other lexicon and structures are from the Arawakan language misleadingly known as Island Carib. The other three examples in the book are more problematic. Callahuaya, a secret language used by a group of male curers in Bolivia, uses Quechua structure and lexicon from many sources, including Quechua and Tacana, but among which the extinct and poorly-documented Puquina language is thought to be inportant. Puquina had five contrasting vowels and phonemic vowel length, as has the lexicon (but not the morphology) of Callahuaya. Quechua has three vowels without contrasting vowel length. The mixed lexicon and the blend of Quechua and non-Quechua features in the phonology distinguish Callahuaya from the more formulaic intertwined languages. The histories of the two other languages illustrate a fact which is too often overlooked in studies of mixed languages. As soon as mixed languages emerge as separate entities, they can evolve, add or subtract rules and lexicon, and can change. Shelta or Gammon, the in-group language of Irish Travellers, is an example. Shelta comprises a Hiberno-English grammatical framework into which are inserted lexicla items of diverse origins. Many of these are phonologically disguised Irish Gaelic words, some are Gaelic words without any disguise, a number derive from English Cant, some more are from Romani, and the origin of many is completely mysterious. There are a number of synonyms for common words in the lexicon (this is also true of Angloromani) in which the words have diffrerent origins, for instance skai/paani ‘‘water’’, where the first and commoner word derives from Irish uisce and the second from Romani

Language intertwining

pani. Hiberno-English shows the impact of Irish Gaelic on its phonology and syntax. Material in Shelta from the 1890s, profiled in Grant (1994), shows structural features — tense-markers, some elements of derivational morphology, word-order characteristics — which are not of Hiberno-English origin. Some of these can be attributed to Irish Gaelic, others cannot. It is also likely that some distinctively Shelta words in use last century have since been lost. It is known that Shelta was once spoken within a predominantly Irish Gaelic grammatical framework, and many speakers of Shelta last century also spoke Irish, in addition to, or instead of, English. Shelta is still spoken to a slight extent in the Highlands of Scotland, where Shelta lexicon is used in a Scottish Gaelic grammatical framework. Unfortunately we have no textual material from ‘Irish Gaelic’ Shelta, only from a form of ‘English’ Shelta which nevertheless shows some relics of Irish grammatical structure, including a few instances of the characteristically Gaelic morphophonemic initial consonant mutation. It is possible that, since much of the lexicon of Shelta derives from non-Irish and non-English sources, the mysterious elements in the lexicon may be the remains of the language underlying Shelta which was then swamped by elements from Irish, first of all, and then from English. That is to say, the persistent and characteristic element of Shelta is (in part) its lexicon, which has been inserted into grammatical frameworks of at least two different languages over time. In modern Shelta with its English-derived structural framework we may be witnessing the later stages of a process of language intertwining which began at a time when English was unknown to Shelta-speakers. Ma’a, or as Mous calls it, Inner Mbugu (Mous 1994), is spoken in northern Tanzania. It remains as a largely Cushitic-origin lexicon available for use by speakers of the Bantu language Outer Mbugu (basically a dialect of Pare). Ma’a words are perfect synonyms of Outer Mbugu words. However, there were previously some independent features of Ma’a structure which are not of Bantu origin and not shared with Outer Mbugu. These, such as the productive causative suffix -ti, were still able to be recorded in the 1960s, but are now obsolete. The problem is that the distinctive lexicon of Ma’a does not derive from one single Cushitic language. In addition to late loans from the Bantu languages Pare and Mbugwe, there are forms from Oromo (Southern Cushitic), Iraqw (Eastern Cushitic), and many from Maasai (Nilo-Saharan), and a number of morphs whose origins are currently unknown. Thomason (1997b: 484) suggested that Ma’a might originate from a macaronic Cushiticlexifier pidgin. This was used by people who were deprived of their cattle and then enslaved by the Maasai.





Anthony P. Grant

The conclusion from thsi is that some of the intertwined languages which have developed were not purposely created, but have evolved from previous states of being through the processes of structuiral impoverishment and the effects of replacive bilingualism. A theory which can deal diachronically with language intertwining is one which is best equipped to explain the process. There is not just one technique of language intertwining. Unfortunately we lack historical records or extensive documentation of earlier stages of these intertwined languages, and must rely on conjectures and the occasional linguistic clue to help us see how the languages became intertwined. These instances show us that language intertwining can occur as the end-product of rule loss and language contact as much as a first step towards the construction of a new language for a new ethnic identity.

. Creoles, language intertwining and mixed languages What does language intertwining have to tell us about the origin of creoles? The answer depends upon one’s assumptions about the origin of morphosyntactic elements in creoles, and upon the languages which one classifies as creoles. The structure/lexicon dichotomy which we see in Media Lengua might please the substratists who wish to see it exemplified in the structure of creoles, but it has long been shown that no single substrate language is generally a perfect fit as the sole source of all typological patterns, let alone the transferred structural morphemes, in any given creole. Language intertwining is not a sufficient or necessary condition to account for such languages. Caribbean creoles may share typological similarities with certain Kwa or other West African languages, as indeed does Kituba (Mufwene 1997), and in both cases the languages may result from the interaction of speakers of West African languages with socially dominant speakers of other languages (say English or French, or in the case of Kituba, Kikongo) which provided the lexical input for the contact language, but they do not usually share the same grammatical morphemes as these languages, which is what true language intertwining would require. Substratists see Kwa grammatical features being reproduced in creoles by morphemes which are usually of European origin. Even those who wish to attribute most creole features to substrate influence cannot say that Guadeloupean, for instance, blends French lexicon and Kwa structure — at best it combines French lexicon and some Kwa typology.

Language intertwining

One can show this by looking at Haitian Creole. For nearly two decades a team at the Université de Québec à Montréal under Professor Claire Lefebvre have examined structural similarities between Haitian Creole and Fon, a Gbe language known to have been the language of a large proportion of the slaves transported to Haiti, especially in the seventeenth and early eighteenth century. The scholars in question are exploring the impact that Fon-speakers equipped with French lexicon might have had on the genesis of Haitian Creole. It sounds like a good scenario for the development of an intertwined language. But many Fon-speakers also settled in French Guiana in the late seventeenth century and there are major differences between Guyanais and Haitian. If Haitian Creole were an instance of language intertwining we would expect to find a language with French lexicon but Fon semantics. Free grammatical morphs would be either from French, or from Fon, or there would be a set from each language. The tense-aspect systems and the means to mark it would be Fon, as would markers for other grammatical subsystems. The syntax of the language would be Fon as to word-order, element order and the structure of clauses. The impact of French lexicon might preserve some French phones not found in Fon, but the phonology of this fictive language would be largely Fon in appearance, even if lexical tone had not developed in the language. In fact, Haitian Creole is rather different from this, as is Guyanais. There are similarities between Haitian and Fon, but Haitian has many major typological similarities with other creoles, both French and non-French lexifier, which were not clearly influenced by Fon. Grammatical morphemes mostly derive from French, as does most of the lexicon. The phonology is as different from Fon as it is from French. Most creoles cannot be easily fitted into Bakker’s framework for intertwined languages, and this is largely because overt structural features, that is, grammatical morpehemes, have not usually been transferred en bloc from a substrate language to the creole. There are some exceptions, as the discussion of the Papiamentu pluraliser above showed, while the noun pluraliser ma- and some personal pronouns in Palenquero Creole Spanish of Colombia derive from Kikongo, but they are relatively rare. Typological features, especially those which depend upon the ordering of elements, do not provide evidence for interlinguistic influence as strong as that of morphs which have actually been transmitted from one language to another, since languages can change typologically as the result of various stimuli. Some creoles look like candidates for language intertwining but are not. Saramaccan may contain many lexical items (Smith 1987b) and some free





Anthony P. Grant

grammatical morphemes of Gbe origin (for instance the words for ‘who’ and ‘what’(Smith 1987a: 147)), in addition to such Gbe (and indeed widespread West African) typological features as serial verb constructions, a large number of ideophones, prenasalised and labiovelar stops (which have spread to lexical items of European origin) and lexical tone. It may thus have a phonology which seems to have more salient characteristics of Kwa languages than of Portuguese or English, the two main sources of core lexicon (over 300 Saramaccan forms come from Portuguese while about 600 derive from English via early Sranan). However, it is not an intertwined language as such, though it is a fascinating example of the results of multi-layered language contact involving several very different languages. For one thing, the grammatical morphemes derive mostly from English by way of Sranan, of which Saramaccan can to some extent be regarded as a historically-invaluable early eighteenth century witness. In any case, the Gbe-derived lexicon in Saramaccan, some 127 items (Smith 1987b), is outnumbered by the 142 terms from Kikongo, which exerted no overt grammatical influence on the languiage. Furthermore, one ‘‘Kwa’’ phonological trait which makes Saramaccan look different from its lexifier, namely the absence of r and its substitution by l, can be shown from early data on the language to be a late eighteenth-century innovation given impetus by deletion of intervocalic l (Smith 1987a: 307ff.). Some creoles certainly exhibit features of true language intertwining. This is the case with Berbice Dutch of Guyana, and Zamboangueño and Cotabateño of Mindanao in the southern Philippines. These case histories will be looked at more closely. . Berbice Dutch Berbice Dutch is currently spoken or remembered by a handful of people of mixed Amerindian, African and European descent, the so-called Bovianders, living near Wiruni Creek on the Berbice River in the interior of Guyana. Kouwenberg (1994) provides a very full description of the language. This is a example of a creole which shows a very high degree of influence from African languages, and from one in particular. The language in question is Eastern Ijo, from the Niger Delta region of eastern Nigeria. All the productive bound morphemes in Berbice Dutch derive from Eastern Ijo. The creole is unusual among Caribbean creoles in having bound morphemes in the first place. The morphemes borrowed include the continuous marker -a, the past marker -tε, the pluraliser -apu and a locative element -ana. It almost seems as

Language intertwining

if rudimentary morphosyntactic structures were adopted from Eastern Ijo in order to form the basis of a grammatical framework. Berbice Dutch phonology is not very close to its main African component, or indeed to Dutch. It lacks lexical tone, and though it distinguishes open and close forms of e it does not do this for o. Schwa is also present. The complex vowel system, contrastive vowel length and consonants such as x (replaced by  or k) are present in Dutch but not in Berbice Dutch. Closed syllables are permitted in Berbice Dutch but not in Eastern Ijo. The TMA system is evidently more similar to those in West African languages than to Dutch, although it deploys rather more preverbal particles and bound morphemes than most Caribbean creoles do. The basic word order of Berbice Dutch is interesting: both Dutch and Eastern Ijo are SOV languages, but Berbice Dutch is resolutely SVO. The evolution of Berbice Dutch is anomalous in respect of the general patterns of genesis among Caribbean Creoles. The vast majority of Africanderived elements in the language derive from Eastern Ijo, and the lexical items which have been borrowed are overwhelmingly part of the basic vocabulary of the language, whereas in other Caribbean creoles they tend to refer to survivals of African folklife and material and spiritual culture. Words from other African languages seem to have entered Berbice Dutch via Guyanese Creole English (‘‘Creolese’’), which is nowadays the main language of Berbice Dutch speakers. The origin of the Berbice colony and its connection with speakers of Eastern Ijo are fairly well established. Ijo-speakers were brought from the Niger Delta area in the seventeenth century by Dutch slave-traders and were put to work on plantations on the Berbice River owned by the Dutch colonist Abraham van Peere (Smith, Robertson & Williamson, 1987). A count of Berbice Dutch morphemes in Kouwenberg (1994) shows that 435 are of Dutch origin, 182 derive from Eastern Ijo and five could be from either language. 222 morphemes derive from Creolese, 164 are Arawak, 29 are from other languages, mostly African ones. The origin of 57 is at present unknown. Most core vocabulary is taken from Dutch (c.65 per cent) and Eastern Ijo (<30 per cent), while the remainder derives from Arawak (a language spoken by many Bovianders’ ancestors) or Creolese. Some semantic fields, such as numerals, are wholly Dutch in origin, but such fields as body parts, verbs of motion, and items of clothing, take elements from both major components. Eastern Ijo items include pronouns, verbs, adjectives and nouns. Dutch has not been available as a lexifier for over a century; new loans come from Creolese.



 Anthony P. Grant

Creolese is probably the most recent element to enter the language, and appears to have come into Berbice Dutch over the past century or so, a process accelerated by language shift away from Berbice Dutch. It has also exerted some little influence on Berbice Dutch structure. The TMA particles das marking continous aspect and its negative equivalent dasn derive from Creolese, and they are used consistently even by fluent speakers of the language. Many speakers also have an imperfective marker yustu from the Creolese, parallel with English used to, while the anterior marker bin and the past continuous marker was are used by speakers more at home in Creolese. Berbice Dutch could be claimed as an instance of language intertwining, since the productive morphology of the language is derived from a language other than that which has provided the bulk of the core lexicon. However, the Eastern Ijo elements in Berbice Dutch account for more than just grammatical features, while neither its phonology nor basic word order are typical of the two major, generally analytic, components. Much convergence of features of Dutch and Eastern Ijo has occurred, and features have developed which are not directly paralleled in either language. The impact of Creolese should not be ignored either. With few exceptions our textual and lexical records of Berbice Dutch only begin in the 1970s, and we can merely guess at what the language was like in the period before Creolese began to have an impact. . Mindanao Creole Spanish: Mixed language, Intertwined language, Creole? The kind of substrate influence present in Berbice Dutch is found even more with two Spanish-lexifier creoles from southwestern Mindanao, in the Philippines. These are Zamboangueño and its mutually intelligible sister-dialect (or descendant) Cotabateño. The first is a first language for more than 200,000 Christians in Zamboanga City and the surrounding area, including parts of the nearby island of Basilan, and a second language to as many more, including Muslims and adherents of traditional religons. The second is spoken by a dwindling population of a few thousand Catholics in Cotabato City. It is clearly a derivative of the former, as can be seen by the large amount of shared vocabulary deriving from the Central Bisayan language Hiligaynon, with which the ancestors of the present Zamboangueño speech community are known to have come into

Language intertwining

contact during the period of the formation of the creole, although this is not the case with the speakers of the creole in Cotabato City. Riego de Dios (1987), herself a native speaker of the language and its chief describer, points out that the Cotabateño speech community sprang from a group of children ransomed by local Spanish Jesuits from local Muslims and converted to Christianity in the 1870s. The children developed a mixed Spanish-Magindanaw (Southern Philippine Austronesian) language. This was modified by contact with speakers of the first creole when the community was evacuated to Zamboanga City in 1900 at the outbreak of a cholera epidemic. Members of this speech community, being Catholics in a sea of Muslims, were directed to marry one another and to give rise to a new community with a distinctive religious and ethnolinguistic identity. For most purposes Zamboangueño and Cotabateño could be treated together as Mindanao Creole Spanish. Apart fom the dictionary and grammatical description of Cotabateño in Riego de Dios (1989), my main sources have been for Zamboangueño, and comprise Frake (1971, 1980), McKaughan (1953) and the Catholic New Testament translation (Rivas, ed., 1982). My data sources for Hiligaynon were Sunio and Zorc (1992) and Zorc (1998). There are other Spanish-lexifier creoles in the Philippines. They are generally known as the Manila Bay Creoles, and are said to result from a hispanicised Malayo-Portuguese brought to Manila in 1665 by Catholic refugees from Ternate in the Moluccas. Ermiteño, once spoken in the old Manila suburb of Ermita, is probably extinct; Caviteño and Ternateño each have a few thousand speakers living just outside Manila, in Cavite and Ternate. These languages have remained in contact with Spanish, and to a greater degree, Tagalog, which is the source of the Philippine influence on them. Ternateño is supposed to be the oldest of the creoles and the progenitor of the others, but it contains a number of phonetic pecularities and some loans from Moluccan Malay, Portuguese and the non-Austronesian language Bahasa Ternate, which are not found in the other Manila Bay creoles, and which suggest a partially separate origin (Grant 1996). Together, the Manila Bay and Mindanao creoles are generally referred to by their speakers as Chabacano/Chavacano, which literally means ‘vulgar’, though the technical term for them is Philippine Creole Spanish. There were speakers of Manila Bay Creoles at the time of the founding of the modern city of Zamboanga in 1719, since the city was founded by soldiers from Manila who came to Mindanao by way of Iloilo on the island of Panay, a Hiligaynon-



 Anthony P. Grant

speaking town. It is not improbable that Mindanao Creole Spanish is at least in part a recreolisation of Caviteño or a similar variety under the influence of Hiligaynon. Mindanao Creole Spanish derives four-fifths of its vocabulary from Spanish, usually with minimal phonological adjustment, but its structure owes little to Spanish. Grammatical relations are thus usually expressed by word order patterns and the use of free grammatical morphemes. The syntax of the language is similar in most respects to that of Central Philippine languages such as the Bisayan languages (including Cebuano, Hiligaynon, Samar-Leyte, and Aklanon), Tausug, Bikol and Tagalog, though certain features are specific to Hiligaynon or Bisayan languages and are not found in Tagalog. The language contains a number of overt grammatical morphemes from Hiligaynon, which has also provided the bulk of the non-Spanish lexicon. These include such free grammatical morphemes as the copular verb ʔamó, originally an expletive in Hiligaynon (with the sense ‘that’s it!’), which has been regrammaticalised. It is especially widely used in the Zamboangueño New Testament translation. Also borrowed are several discourse particles, conjunctions such as kay ‘because’, and both the forms and the syntax of gayót ‘very’ and lang ‘only’, both of which follow the items which they modify. The use of the copular verb seems to be replacing the earlier usage, characteristic also of Central Philippine languages, where a sequence of Predicate-Subject was construed as constituting a copular sentence. Later, and mostly lexical, influence, has come from Tagalog/Pilipino, the national language of the Philippines, and from Cebuano, also a Bisayan language and is thus related more closely to Hiligaynon than Tagalog is, though it and Hiligaynon belong to different subgroups of Bisayan. Cebuano is an important lingua franca in Mindanao, not least in the towns. English has also exerted extensive lexical influence, and the creole was strongly influenced, and to a slight degree decreolised, by contact with Standard Spanish (Lipski 1992), for long a compulsory school subject and the medium for Catholic worship. There are some loans from other regional languages such as Sinama and Yakan, closely related languages indigenous to the area, and from Tausug, an important local lingua franca and another Central Philippine language. Cotabateño also has loans from the Mindanao languages Magindanaw and Tiruray, maybe ancestral languages of children who formed the creole speech community there. Zamboangueño phonology includes features which are characteristic of both Spanish and Hiligaynon, but it is not a composite of the two as such. It can rather be seen as an expansion, by transferral of Spanish and English

Language intertwining 

phones, of a typical Central Philippine phonological system. Words from Philippine languages undergo little phonological modification in the creole. The glottal stop [ʔ] has been taken over from Philippine languages, and as in these languages, it occurs non-phonemically before all word-initial vowels, and intervocalically after vowels and consonants, and word-finally. The same is true of the phoneme usually written ng, / /, found only allophonically in Spanish as a form of n before velars, but which is especially frequent in Zamboangueño in words from Philippine languages. As in Tagalog and Bisayan languages, non-glottalised final vowels in the creole are followed by a slight non-phonemic aspiration. There are some divergences from the Spanish consonantal system. The two rhotics of Spanish are not distinguished from one another in the creole, while Spanish f is realised by less well-educated speakers as p (as happens in some Philippine pronunciations of English or Spanish), though better educated speakers have the fricative. Intervocalic voiced consonants are realised as stops as in Philippine languages, rather than as fricatives as in Spanish. Voiceless and voiced stops can occur word-finally in the creole, when they occur in loans from English or Philippine languages. The affricates and fricatives /tʃ, ʃ, d, v/ are found in Zamboangueño. The first belongs to the Spanish element of the vocabulary, the others have come in recently as a result of lexical borrowing from English, and are confined to English loans. The vowel system is the fivemember Spanish set of ieaou rather than the three-member one, namely iau typical of Philippine languages such as Cebuano and of the pre-Spanish element in other Philippine languages, such as Tagalog and Hiligaynon. However, vowel length is contrastive neither in Spanish nor in the creole, though it is so to an extent in Tagalog. Nouns in Zamboangueño do not have grammatical gender as they would in Spanish. The definite article ʔel is unmodified for gender and number, which is not the case with its Spanish prototype but which is true of Tagalog and Hiligaynon ʔang. Nouns which carry Spanish plural markers are regarded as singular; pluralisation is done with (ʔel) manga, a form of Central Philippine origin which is found in Bisayan and Tagalog among other languages. Proper names when used as subjects are preceded by the personal definite article si, another morpheme from Central Philippine languages. Possession is indicated by the use of de or del if the possessor is definite, which is much the same as in Spanish, whereas in Philippine languages possession in noun phrases is expressed by apposition of the possessed noun before the possessor, with the definite article inserted between the two if the possessor is definite. Post-

 Anthony P. Grant

nominal adjectives in the creole do not need a linking particle as in Philippine languages. Personal pronouns are encoded for three ascending levels of formality within the second person (tu, ʔebós, ʔusté, respectively), as is typical of Philippine languages. Plural personal and possessive pronouns derive from Hiligaynon, thus Spanish ellos ‘they’ has been completely replaced by Hiligaynonderived silá. The Philippine distinction between inclusive and exclusive first person plural pronouns is found in the creole. Possessive pronouns follow their nouns, as in Philippine languages. Those adjectives which precede the noun in Spanish do so in the creole; others including those of Philippine origin follow the noun, as in Spanish and Philippine languages. The Zamboangueño verbal system does not use a complex pattern of prefixes and infixes to indicate aspectual features, as Philippine languages do, but instead employs three preverbal TMA particles, while the pronoun or noun subject of the phrase often follows the verb. The particles ya, ta, ʔay/di/ ʔel encode completive action, continuous or habitual action and irrealis respectively. Of the three irrealis particles the first is the one most widely used, while the second is archaic and the third is most common in rural speech. The aspectual system corresponds roughly to the irrealis/non-irrealis distinction typical of aspectual divisions in the verbal systems of Philippine languages. Some of the particles are found in other Spanish-lexifier creoles. Ya is related to the Spanish adverb meaning ‘already’, while ta (deriving from Spanish está, the third person singular present indicative form of the verb estar ‘to be in a location, to be in a temporary state’) is a feature which occurs in most Spanish-lexifier creoles, though similar features are found in Portugueselexifier creoles of Africa and Asia (Lipski 1986, 1992). Origins of the irrealis markers are uncertain: the last is puzzling; the first two may derive from older Spanish ha de ‘have to’, a usage commoner in Portuguese. Other forms which may be relics of a Portuguese creole past for this language are na, a general locative proposition with no ready Spanish etymology, kiláya ‘how’, which includes Portuguese laia ‘manner, way’, unknown in Spanish, and maskín, a particle with the sense ‘no matter what, even if ’, reinforced by the presence of particles with the shape of masín and the sense of ‘however’ in Bisayan and other languages. The Central Philippine particle man- marks some transitive verbs, including some from Spanish, those from English and the vast majority from Philippine languages. It also makes transitive verbs from adjectives: manblándo ‘to soften’ from blándo ‘soft’.

Language intertwining 

As stated above, most syntactic features of this creole owe more to Philippine languages than to Spanish. This is true of the extensive use of reduplication of verbal and adjectival stems and the preponderance of VSO sentences, where Spanish has SVO sentences. It is also true of the use of differing negators: the use of nway (which also means ‘there is not’) with existential statements and the Tagalog-derived hindí’ with contingent statements reflects the use of walá’ and díli’ in Hiligaynon (but Tagalog would not use walá’ before completive aspects, though Hiligaynon would), while Tagalog, Hiligaynon and Mindanao Creole Spanish, but not Standard Spanish, all have special forms for negating imperatives: respectively huwág, ʔayáw and no. Spanish has no in all cases. The use of kon to mark direct objects can be paralleled from Philippine languages where oblique definite forms of certain articles are so used, or from Spanish ‘personal a’ before animate objects (con in Spanish is ‘with’), but there are differences in usage. Not all features of Central Philippine syntax are reflected in Zamboangueño syntax. The verbal affix infixation of central Philippine languages has no correlate in the creole. The intricate goal and focus-marking system typical of many Austronesian languages has not been transferred, and if it needs to be expressed overtly this involves using circumlocutions more or less identical word for word with Spanish equivalents. The productive derivational morphology of Mindanao Creole Spanish is mostly taken from Spanish and has been analysed and reappropriated from the large number of derivative forms in the vocabulary which come from Spanish. Thus we find the productive use of agentive suffixes from Spanish, such as -dor. There are also some derivational morphemes from Central Philippine languages, such as the marker of ordinal numerals ʔáka-/ʔíka-, and the reciprocal verbal suffix (man-)-han. Lipski (1992) points out that the form of Zamboangueño recalled by his most elderly consultants during his work in 1985, people who could remember how the language was spoken at the end of the nineteenth century, was something much closer to standard Spanish than the language is now. Unfortunately he has not published any samples of this archaic form of speech; McKaughan’s texts are the earliest examples of Mindanao Creole Spanish of any great length, and they show a language which is not too different structurally from less anglicised forms of modern-day Zamboangueño. Lipski does not see Zamboangueño as a form of the Manila Bay Creole Spanish language Caviteño which has been influenced by Hiligaynon, as has been suggested in Smith (1995: 373). Instead he provided a creativist model of

 Anthony P. Grant

the origin of the language, seeing in it a pooling of features, including lexicon, common to several relevant Philippine languages (Central Philippine ones such as Tagalog, Hiligaynon, Cebuano and Tausug, plus others such as Magindanaw, Maranaw and probably Yakan and Sinama). These features included typological characteristics of grammatical structure, often with specific free grammatical morphemes, and a certain amount of shared vocabulary, which included the large number of Spanish loans relating to such matters as food, religion and war, which had passed into the Philippine languages. Now Lipski sees this bundle of features as the basis of a new language which sprang out of a need for a shared means of communication by members of several speech communities who had come into the Zamboanga area — local Yakan and Sinama speakers, Tausug raiders, Spanish soldiers and their Tagalogspeaking colleagues and Hiligaynon wives, who reached Zamboanga from Manila and Cavite via Iloilo — at the time when Fort Pilar was built and maintained there with a Spanish garrison in the period from 1719. As time went on, Lipski, says, this emergent language underwent influences from noncreolised and possibly creolised Spanish, while later incorporating elements from Visayan (especially, he maintains, from Cebuano, which has been prominent in the area for over a century), and after World War II, also from English. In other words, the ingredients of Zamboangueño varied from one century to the next. The problem with this idea is that it does not account for the presence of the bulk of the Spanish-derived vocabulary in Zamboangueño, which is culturally neutral in that it refers to a large number of basic concepts for which there would already be words — often close cognates — in all or most of the participating Philippine languages. A similar problem arises with the presence and nature of the Philippine lexical element. It is true, as Frake (1971) made clear, that many terms of Philippine origin in the creoles relate to local realia, and others represent the disadvantageously-marked feature in adjectives and nouns where polarisation is marked (for instance smallness as opposed to largeness), but this also does not account for all the forms. The forms are generally from Hiligaynon rather than Cebuano. Furthermore the lexical semantics of Zamboangueño are much closer to those of Central Philippine languages than they are to those of Spanish (there is much easier ‘‘intertranslatability’’ between the creole and Tagalog or a Bisayan language than between the creole and Spanish), though the fit is not identical. Mindanao Creole Spanish is not simply a calque upon Hiligaynon. The Philippine-derived forms on the Swadesh 200-word list for Zamboangueño do not seem to be words that are especially widespread as close

Language intertwining 

cognates or identical forms across the relevant grouping of Philippine languages. Nor do they appear to present a coherent semantic grouping, apart from the set of plural pronouns. The rest of the items in question are rather more scattershot and comprise glosses for the following: ‘because, child, dirty, dog, dull, father, float, knee, left side, louse, mother, mountain, not, rain, river, rotten, rub, short, small, squeeze, stab, thin, tree/wood’. To sum up, Mindanao Creole Spanish involves the combination of a number of features of Central Philippine and Spanish origins. The segmental phonology is a compromise between the two groups, the lexicon is mostly taken from Spanish, although with a considerable Philippine element which includes the replication of a number of Philippine semantic categories, as might be expected of the linguistic vehicle of a sub-branch of Philippine Christian culture. The feature of pluralisation is expressed by morphemes of Philippine origin. The syntax is largely Philippine, with a predilection for VSO sentences, but some nominal and adjectival syntax from Spanish remains. Not all the syntactic resources of Philippine languages are available in the creole. Given the nature and distribution of its components, Mindanao Creole Spanish is certainly a mixed language. Insofar as the grammatical subsystem of pluralisation has been taken over from Philippine languages, it can be argued for as an intertwined language on grammatical grounds similar to those for Michif and Mednyj Aleut. As it is a relatively new language, which preserves the lexicon of Spanish though not its structure, and which shows clear influences from substrate Central Philippine languages without being an identical calque on any of them, it is a creole. The TMA particles and the small number of other morphemes which are shared exclusively within this area with the Manila Bay Creoles, and which cannot be derived from Spanish, suggest that Mindanao Creole Spanish may represent the recreolisation of a Manila Bay Creole Spanish, which presumably had some Tagalog elements, under the influence of Hiligaynon (perhaps the first language of the wives of the men who brought the creole to Zamboanga via Iloilo) and with the continuing influence of more standard forms of Spanish. In answer to the title of this subsection, Mindanao Creole Spanish can be classified as all three — Mixed Language, Intertwined Language, Creole.

. Conclusions We know that one single historical or demographic model will not account for

 Anthony P. Grant

the origins of all or even most creoles. The more we learn about the histories of creoles, the more diverse we discover them to be. The backgrounds to Guadeloupean, Louisianais, Mauritian and Tayo, for instance, are all different even though all four languages derive their lexicon from French. The same is true of intertwined languages. Two languages can appear to be very similar or typologically congruent as far as the sources and distribution of the elements go, but their origins can be very different. Some languages evolved into intertwined languages as the result of chronic structural or lexical loss. Others seem to have been created purposely as intertwined languages. A crucial sociolinguistic comsideration should be borne in mind. Many speakers of many intertwined languages are familiar with one or more of the languages which go to make up their new language. For instance, speakers of Angloromani always know English. This is not always the case with speakers of creoles: for example many speakers of Dominican Creole French have no knowledge of Standard French. Nor is it always true for speakers of intertwined languages: few speakers of Zamboangueño now know Spanish or Hiligaynon, though many know English, Tagalog or Cebuano. 250 years ago the situation may have been different. On the other hand, speakers of Berbice Dutch have been without contact with Eastern Ijo and (more) standard Dutch for a very long time, for maybe two centuries or more in the first case and probably 150 years or so in the second. The languages most readily available to them, Creolese and to a lesser extent Arawak, have both been used as sources of new vocabulary. The results of ‘‘pure’’ language intertwining differ from those of creolisation, since language intertwining involves the perpetuation of most or all of the structure of a previously existing language, and this includes the perpetuation of morphs and their functions. Creolisation involves the creation (albeit often by use of syntagms and zero morphs rather than by inventing new overt morphs) of a new structure. The syntactic patterning of some creoles can be as covert as overt. Even in those intertwined languages with mixed grammatical systems — Michif, Mednyj Aleut, to some extent Mindanao Creole Spanish — there is little simplification of borrowed structures. For the rest, it seems that the features of the structural donor language which are most likely to be augmented by features taken from the lexical donor language are, unsurprisingly, elements of derivational morphology and of the segmental phonology, especially the more unmarked phones which share phonological features found also in the structural donor language. However, one can never be quite accurate in predicting exactly which features of the lexical donor language will be taken over.

Language intertwining 

Language intertwining does not hold the key to the development of creole languages, and as a process it is less important than, say, relexification. However, if we know more about the histories of many creoles and expanded pidgins, and if we could separate out the elements which were there in the early stages of the language from those which are later accretions, it would not be surprising if there was more evidence of language intertwining in the earlier stages of several modern creoles. The fact that Berbice Dutch and Mindanao Creole Spanish can be interpreted as possessing sketchy morphosyntactic systems built up out of especially frequent features of the grammar of their substrate languages, and thus combine a superstrate lexicon (with minor substrate loans) and a substrate structure (with minor superstrate contributions) shows the possible extent of processes of language contact. But in order to answer all our questions about the development of subtypes of mixed language it is not enough to know what these or other languages are like now. We need to see what they were like centuries ago, and to explore any ways in which their structures differed from the present ones. And this means that many questions about the development of mixed languages and of creoles will remain unanswered forever.

Note * I would like to thank Peter Bakker and Tonjes Veenstra for inviting me to participate in the 1995 Amsterdam Creole Workshop, the Netherlands Government for financial support for the presentation of my paper, and Joy Granados, Ian Hancock, Frans Hinskens, Maarten Mous, Norval Smith, Sally Thomason, the SEALANG-L Internet Usergroup, and especially R. David Zorc for providing comments and data which have found their way into this chapter. I alone am responsible for any infelicities.

References Bakker, Peter. 1996. ‘‘Language intertwining and convergence: Typological aspects of the genesis of mixed languages’’. Sprachtypologie und Universalienforschung 49: 9–20. Bakker, Peter. 1997. ‘A Language of Our Own’: The Genesis of Michif, the Mixed Cree–French languiage of the Canadian Métis. Oxford: Oxford University Press. Bakker, Peter and Maarten Mous (eds.). 1994. Mixed Languages: Fifteen case studies of language intertwining. Amsterdam: IFOTT. Biggs, Bruce. 1965. ‘‘Direct and indirect inheritance in Rotuman’’. Lingua 14: 383–415.

 Anthony P. Grant

Boretzky, Norbert, and Birgit Igla. 1994. ‘‘Romani mixed dialects’’. In Mixed Languages: Fifteen case studies of language intertwining, Peter Bakker and Maarten Mous (eds.), 35–68. Amsterdam: IFOTT. Capell, Arthur. 1976. ‘‘Austronesian and Papuan ‘mixed’ languages: General remarks’’. In New Guinea Area Languages and Language Study, vol. ii: Austronesian Languages, Stephen A. Wurm (ed.), 527–579. (Pacific Linguistics, C-39.) Clough, James Cresswell. 1876. On the Existence of Mixed Languages, Being an Examination of the Fundamental Axioms of the Foreign School of Modern Philology, More Especially as Applied English. London: Longmans, Green and Co. Frake, Charles O. 1971. ‘‘Lexical origins and semantic structure in Philippine Creole Spanish’’. In Pidginization and Creolization of Language, Dell Hymes (ed.), 223–242. Cambridge: Cambridge University Press. Frake, Charles O. 1980. ‘‘Zamboangueño verbal expressions’’. In Language and Cultural Description: Essays by Charles O. Frake, Anwar S. Dil (ed.), 277–310. Stanford: Stanford University Press. Grant, Anthony P. 1996. ‘‘Spanish, Portuguese and Beyond: A sociohistorical , structural and lexical comparison of Zamboangueño and Papiamentu’’. MS. Grant, Anthony P. 1998. ‘‘Romani words in non-standard British English and the development of Angloromani’’. In The Romani element in non-standard speech, Yaron Matras (ed.). Wiesbaden: Otto Harrassowitz. Hancock, Ian F. 1986. ‘‘The cryptolectal speech of the American roads’’. American Speech 63.3: 206–220. Kouwenberg, Silvia. 1994. A Grammar of Berbice Dutch Creole. Berlin: Mouton de Gruyter. Kuiper, Franciscus B. J. 1962. ‘‘Nahali, a comparative study’’. Mededelingen der Koninklijke Nederlandse Akademie van Wetenschappen, Afdeling Letterkunde, nieuwe reeks 25.5: 239–352. Lipski, John M. 1986. ‘‘The Portuguese element in Philippine Creole Spanish: A critical reassessment’’. Philippine Journal of Linguistics 17: 1–17. Lipski, John M. 1992. ‘‘New thoughts on the origins of Zamboangueño (Philippine Creole Spanish)’’. Language Sciences 14 (3): 197–231. McKaughan, Howard P. 1953. ‘‘Notes on Chabacano grammar’’. Journal of East Asiatic Studies 3: 205–224. McWhorter, John H. 1996. Towards a Theory of Creole Genesis. Berlin: Peter Lang. Mous, Maarten. 1994. ‘‘Ma’a or Mbugu’’. In Mixed Languages: 15 case studies of language intertwining, Peter Bakker and Maarten Mous (eds.), 175–200. Amsterdam: IFOTT. Mufwene, Salikoko S. 1997. ‘‘Kitúba’’. In Contact languages: A wider perspective, Sarah G. Thomason (ed.), 173–208. (Creole Language Library, 17.) Amsterdam and Philadelphia: John Benjamins. Muysken, Pieter. 1997. ‘‘Media Lengua’’. In Contact languages: A wider perspective, Sarah G. Thomason (ed.), 365–426. (Creole Language Library 17.) Amsterdam and Philadelphia: John Benjamins. Nothofer, Bernd (ed.). 1996. Reconstruction, classification, description: Festschrift in honor of Isidore Dyen. Hamburg: Abera. Nurse, Derek. 1994. ‘‘South meets North: Ilwana=Bantu + Cushitic on Kenya’s Tana

Language intertwining

River’’. In Mixed Languages: Fifteen case studies of language intertwining, Peter Bakker and Maarten Mous (eds.), 213–222. Amsterdam: IFOTT. Pawley, Andrew. 1996. ‘‘On the position of Rotuman’’. In Reconstruction, Classification, Description: Festschrift in honor of Isidore Dyen, Bernd Nothofer (ed.), 85–120. Hamburg: Abera. Riego de Dios, Sister Maria Isabelita. 1987. ‘‘Composite dictionary of Philippine Creole Spanish’’. Papers in Philippine Linguistics 7.2: i–xii, 1–210. Manila: Linguistic Society of the Philippines. Rivas, Father Carlos, ed. 1982. El buen noticia na Chabacano. Zamboanga City: Claretian. Ross, Malcolm. 1996a. ‘‘Is Yapese Oceanic?’’ In Reconstruction, Classification, Description: Festschrift in honor of Isidore Dyen, Bernd Nothofer (ed.), 85–120. Hamburg: Abera. Ross, Malcolm. 1996b. ‘‘Contact-induced change and the comparative method: Cases from Papua New Guinea’’. In The Comparative Method Reviewed: Irregularity and regularity in linguistic change, Mark Durie and Malcolm Ross (eds.). Oxford: Oxford University Press. Sampson, John. 1926. The dialect of the Gypsies of Wales. Oxford: Oxford University Press. Smart, Bath Charles, and Henry Thomas Crofton. 1875. The Dialect of the English Gypsies. London: Asher and Son. Smith, Norval. 1987a. The genesis of the Creole languages of Surinam. Doctoral dissertation, University of Amsterdam. Smith, Norval. 1987b. ‘‘Gbe words in the creole languages of Surinam’’. Unpublished manuscript, University of Amsterdam. Smith, Norval. 1995. ‘‘An annotated list of creoles, pidgins and mixed languages’’. In Pidgins and Creoles, Jacques Arends, Pieter Muysken and Norval Smith (eds.), 331– 374. (Creole Language Library, 15.) Amsterdam: John Benjamins. Smith, Norval, Ian Robertson and Kay Williamson. 1987. ‘‘The I jo element in Berbice Dutch’’. Language and Society 16: 49–90. Smout, Kary D. 1980. ‘‘A missionary English from Japan’’. American Speech 63.2: 137–149. Sunio, Delicia, and R. David Zorc. 1992. Hiligaynon Newspaper Reader. Kensington, Maryland: Dunwoody. Thomason, Sarah G. 1997a. ‘‘Mednyj Aleut’’. In Contact languages: A wider perspective, Sarah G. Thomason, (ed.), 449–468. (Creole Language Library, 17.) Amsterdam: John Benjamins. Thomason, Sarah G. 1997b. ‘‘Ma’a (Mbugu)’’. In Contact languages: A wider perspective, Sarah G. Thomason, (ed.), 469–487. (Creole Language Library, 17.) Amsterdam and Philadelphia: John Benjamins. Thomason, Sarah G. (ed.). 1997. Contact Languages: A wider perspective. (Creole Language Library, 17.) Amsterdam and Philadelphia: John Benjamins. Thomason, Sarah G., and Terrence Kaufman. 1988. Language Contact, Creolization and Genetic Linguistics. Berkeley and Los Angeles: University of California Press. Wurm, Stephen A. 1978. ‘‘Reefs-Santa Cruz — Austronesian But . . .’’. In Proceedings of the Second International Conference on Austronesian Linguistics, fascicle 2, Eastern Austronesian, Stephen A. Wurm and Lois Carrington (eds.), 969–1010. (Pacific Linguistics, C-61.) Zorc, R. David. 1998. ‘‘Hiligaynon’’. To appear in Encyclopedia of the World’s Languages. New York: Henry Holt.



Chapter 4

Paralexification in language intertwining Maarten Mous

.

Introduction

One of the important processes in language intertwining is what I call ‘‘paralexification.’’ This is the process by which parallel word forms for one and the same lexical entry exist while sharing meaning and morphological characteristics. Paralexification is not another name for language intertwining. It is more than language intertwining and probably also less than intertwining. It is more in the sense that it occurs in various contexts and to various extents that do not always lead to results that anyone would call a mixed language. It is less because paralexification is not sufficient to account for all changes that a particular (mixed) language has undergone. It may be that certain types of paralexification, e.g. those affecting large parts of the basic vocabulary and mainly by taking foreign elements as parallel word forms, correlate with other language changes, for example simplification. Mixed languages, like pidgins and creoles, are interesting from a diachronic perspective and thus the historical processes are the ultimate object of study. Whether a language is to be called a mixed language or not, is no longer important once we know the history of that language. In that event, its genetic classification is also no longer of interest because classification is not a goal in itself in historical linguistics. Studying mixed languages reveals that it is the process of paralexification that leads to intertwining. I will first illustrate how this process accounts for the mixed nature of Ma’a and then look at paralexification in a wider context.

. Structural properties of paralexification. . What is paralexification? Paralexification is the addition of a word form to a lexical entry. This added



Maarten Mous

form is on a par with the existent word form of the lexical entry in question. That is, two word forms share meaning (1)–(5), metaphorical extensions, and morphological properties such as noun class membership for nouns (6–11) and predicate frame and derivational suffixes (12)–(13), (14)–(15) for verbs. The following parallel entries from Ma’a and Mbugu are complete synonyms. For a more detailed analysis of Ma’a as a paralexicon of Mbugu see Mous (to appear). (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

Ma’a bále ‘to move (house)’ míndá ‘house’ hái ‘four’ kú¹u ‘her/his’ fíli ‘to go out of sight over the hill’ m-hé cl.1 ‘person’ va-hé cl.2 ‘people’ m-harégha cl.3 ‘arm’ mi-harégha cl.4 ‘arms’ vu-basá cl.14 ‘face’ ma-basá cl.6 ‘faces’ hlá ‘to open’ hlá-ti ‘to close’ (caus) ’áku ‘to shine’ ’akú-ti ‘to sprout’ (caus)

Mbugu sama ‘to move (house)’ nyumbá ‘house’ nne ‘four’ -akwé ‘her/his’ gongolóka ‘to go out of sight over the hill’ m-nhtu cl.1 ‘person’ va-tu cl.2 ‘people’ m-kóno cl.3 ‘arm’ mi-kóno cl.4 ‘arms’ vu-shó cl.14 ‘face’ ma-shó cl.6 ‘faces’ jughua ‘to open’ jughu-ja ‘to close’ (caus) anga ‘to shine’ anga-ja ‘to sprout’ (caus)

Paralexification can only be recognised if one considers the full system. If one would consider Ma’a only, without looking at Mbugu, then Ma’a would be a mixed language and the true nature of the system, the parallel lexicon to Mbugu, does not show up. Thus, paralexification presupposes drawing the borders, and that not too narrowly, between what is one language and what is another. That is, we have to see Ma’a and Mbugu as one language sharing one grammar in order to recognise the double, parallel lexicon. The aspect of addition to an existent form is what makes paralexification different from additional or replacive borrowing. In time, after loss of the old form, the outcome of paralexification may be indistinguishable from replacive borrowing. However, when the words replaced show formal morphological properties, such as noun class membership, which are identical to another language for every individual lexeme, that in itself is a sign of prior paralexification. Thus, if at some point in the future the speakers of Ma’a no longer

Paralexification in language intertwining

use Mbugu, if they lose the Mbugu lexical base, the traces of paralexification will still be present in the fact that the noun class membership of the individual Ma’a lexeme will still be identical to the corresponding lexeme in Mbugu’s closest relative Pare. As a case in point, we can look at Michif. Paralexification must have happened in the history of Michif, because only paralexification can account for the animate/inanimate distinction that Michif has even in nouns from French origin, matching the Cree idiosyncrasies. For example, the word for ‘money’, from French, is animate, because the equivalent Cree word is animate (Bakker 1997: 99). At that point in time, those speakers were also fluent in Cree, a language that they lost later. The question arises as to whether relexification of Fongbe with French which resulted in Haitian, as it is discussed by Lefebvre (this volume), is in fact paralexification of French and Fongbe forms plus subsequent loss of the Fongbe forms. Such a scenario cannot be ruled out, but the strongest evidence for it should be morphological properties in common between Haitian words and their Fongbe counterparts, preferably excluding other possible lexifier languages. The lack of morphology in creole languages prevents us from having such evidence. An intermediary stage of paralexification is incompatible with relexification of null elements (Lefebvre, this volume). . Where do we find paralexification? Paralexification occurs in many situations. The process may differ in extent and in many other respect, but essentially the phenomenon remains the same. Smith (1995) gives a taxonomy of mixed languages and related phenomena taking paralexification as the central concept, distinguishing mixed languages, symbiotic mixed languages, secret languages, slang, ritual languages, taboo languages and mother-in-law languages. The differences between these are found in the extent of affected vocabulary, the survival of both codes, and in the criteria for the choice of register. The linguistic phenomenon of paralexification should be studied in all its contexts. Situations in which it occurs are for example second language acquisition, language death situations, taboo languages, secret codes, ritual languages. The structural properties of two parallel forms in one lexeme remain the same. Below I give some examples of these situations. .. Registers  of respect Paralexification can be found in registers of respect or taboo languages. The





Maarten Mous

purpose of these registers is that of avoidance of a personal name and words resembling the name, as a sign of respect. We have to keep in mind that names in some languages are meaningful and derived from other words. In Yemsa (Janjero), an Omotic language spoken in Ethiopia, there is a special respectful vocabulary for communication with the elite classes, the king, his council members, the provincial governors and their clans. It is in fact used in order to show respect in some other situations as well. Apart from the lexicon, pronouns and person marking on the verb are stratified too. In addition, there used to be yet another register for words related to the king and the king’s body, see Yilma (1992). There are some Adamawa languages in Northeastern Nigeria where the name of a respected person who has passed away, and words related to that name are avoided (Kleinewillinghöfer 1995). Another widespread custom is for a woman to avoid the name of her father-in-law or other male in-laws. The best known and documented African case is South-African Bantu hlonipha (Van Rooyen (1968), Kunene (1958), Herbert (1990)). But its occurrence goes beyond the boundaries of Southern Africa as it is also reported for Nyakyusa, a Bantu language of Tanzania. For example, if the name Mwa-mbene (mwa- is a name clitic) is to be avoided, then the word imbene ‘goat’ has to be avoided too and can be replaced by either alya, which is the sound used to chase a goat away, or by indyasi which is derived from the verb -lya ‘eat’ to mean ‘the one that eats a lot’; as a noun, indyasi is placed in the same noun class as the word it replaces. The name Mwa-sekema causes avoidance of isekema (class 9/10) ‘fever’ which is replaced by ihoma (class 9/10) ‘fever’ from Swahili homa, see Kolbusa (1995). Similar cases are reported for Australia, see Haviland (1979). What may not be apparent from this brief description is that these registers of respect are usually highly conventionalised not only with regard to which words should replace the taboo words, but also with regard to which words are taboo. Which words are taken to resemble the taboo name is not just a matter of phonology or morphology, but likewise is conventionalised or lexicalised. The register is shared by a community; it is hardly individual. .. Language loss In situations of language decay the meaning and use of words of the language on the verge of extinction are often adjusted to the newly adopted dominant language resulting in a structure of paired lexical items with the same meaning and the same morphological characteristics. For example, in German in the United States gleichen is most often used in the meaning ‘to like, to find

Paralexification in language intertwining

likeable’ rather than in the older meaning of ‘to resemble, to be like, to equal’, (Salmons 1991: 47). The same observation can be found in the Yiddish of bilingual Americans who use denkn where English think would be used. Thus denkn then contains the meanings of European Yiddish denkn ‘to consider’, maynen ‘to hold an opinion’, gloyben ‘to believe’. Haysn ‘to give a name to’ in European Yiddish, is extended to ‘to summon’ under the influence of English ‘to call’. These examples of loanshifts are from Rayfield, who also mentions that Yiddish words and their English counterparts are often used in the same sentence, for example forsitzer and tsherman, or metsivoh and bargain, and are thus true equivalents (Rayfield 1970: 40, 60–61). .. Second language acquisition The term compound bilingual has been introduced for people who do precisely what I call paralexification: ‘‘For compound bilinguals a verbal label and its translation equivalent have one representation common to both languages’’ (Ervin and Osgood quoted in Hamers and Blanc 1989: 93). .. Transitional stage in replacive borrowing In the case of Bowe (Bantu F.34) in Tanzania, the original Bowe words are in the process of being replaced by Swahili words. But at the moment both exist as pairs. Examples of such pairs are in Table 1. The same is true for many other Bantu languages in Tanzania whose speakers use Swahili more often than their mother tongue. The Swahili words are adapted to Bowe in phonological and morphological properties, that is, the Swahili class 15 prefix for infinitives is replaced by its Bowe counterpart and the v after a nasal is changed into a stop, e.g. chumbi from Swahili chumvi ‘salt’. In all these examples of paralexification we see the same structural properties: An identical meaning for two forms sharing the same morphological characteristics such as noun class membership. This morphological parallelism which I take to be a crucial characteristic of paralexification was reported by Table 1. Pairs of new (Swahili) and old lexemes in Bowe Meaning

New-Bowe

Swahili source

Original Bowe

be angry to breath to mould to bend salt

okasirika opumua ofínyánga opínda chumbi (cl.10)

ku-kasirika ku-pumua ku-finyanga ku-pinda chumvi

ovéna ofwéréra otulatúla uúnánya tónyo (cl. 10)





Maarten Mous

Kunene for hlonipha-nouns: ‘‘It would seem, however, that an attempt is always made to employ a substitute noun (or coin one) in the same class as the one which is being replaced’’ (Kunene 1958: 167, compare also Kleinewillinghöfer (1995: 128)). In the next section I review the various possible sources for the added word form. . What are the sources for the added form? The widest range of possible ways of arriving at an added parallel form can be found in the taboo registers. Sometimes derivations and circumlocutions are placed in the register of respect on par with the taboo word. Often words are borrowed into the respectful register. It also happens that words are modified in form. .. Near synonym Instead of daa ‘to come’ resembling the name Dawasso, Sidama women use higa ‘to return’ (Teferra 1987: 46). If the Iraqw (Cushitic, Tanzania) avoid the name Axweeso, which is based on a word meaning ‘evening’, and along with it the verb axwees ‘to talk’, this verb is replaced by oo’ ‘to say’ (Rekdal p.c.). A recurring type of (near) synonym is where the taboo word is replaced by an ideophone, as in Iraqw hhaawú for baha ‘hyena’, or Nyakyusa alya for imbene ‘goat’. These are onomatopoeic words for the sound that the animal makes. Van Rooyen (1968: 40) also reports on de-ideophonic hlonipha words in Zulu. .. Archaic word Another source from within the lexicon entails activating an archaic word in the register of respect. For example, the Zulu word for ‘sky’, izulu is replaced by the archaic word ibinga in hlonipha (Van Rooyen 1968: 42). .. Semantic extension of a common term A third strategy that also keeps within the lexicon is to make use of a common term but to extend its meaning. For example, the Yemsa word kushu ‘hand’ is used for ‘to give’ in the respectful variety, and the verb for ‘to chew’ is extended to mean ‘to eat’ (Lamberti 1993: 27). .. Circumlocution It is very common to use a circumlocution that is lexicalised. Among the Longuda of Northeastern Nigeria nyi tsanawa ‘thing which smells’ was chosen to

Paralexification in language intertwining

replace the word for ‘ant’ after a man named after the word for ‘ant’ had died (Newman and Newman 1977: 11, as quoted in Kleinewillinghöfer 1995: 127). .. Derivation A similar device is to nominalise a verb to replace a noun, as in Swati hlonipha: umbaso ‘thing for kindling’ from the verb ‘to kindle’ for umlilo ‘fire’ (Herbert 1990: 460). .. Modification of an existent form To avoid the taboo word form one of the simplest ways is to change a consonant or insert a syllable. These devices are very common in secret languages, such as Pig Latin, but are less common in registers of respect. In hlonipha one consonant is sometimes changed to another, to a glottal stop in Xhosa (Herbert 1990: 460), and often to a click in Zulu (Van Rooyen 1968: 38). The initial consonant is replaced by s or the initial syllable is replaced by som in Ballissha (Teferra 1987: 53). Some of the words in the Ma’a paralexicon are derived from Mbugu by replacing the last vowel with e and imposing a lowhigh tone pattern, e.g. ndaté is the Ma’a form of Mbugu ndata ‘stick’. Substitution by reversal of a sound change is reported for hlonipha: s is replaced by k, which is a reversal and over-application of the regular sound-change of k to s before i (Van Rooyen 1968: 39). Syllable reversal and awareness of historical changes were, for example, a factor in the making of Shelta (Grant 1994: 135–138). .. Borrowing The most common way of obtaining a parallel word form is to take one with a similar meaning from another language. In registers of respect this borrowing is often from a wide range of languages, not always neighbouring languages. In a mixed language such as Ma’a the borrowing is also from several sources, because the purpose is to be different from Mbugu. In Michif, the borrowing is basically from French because the purpose of Michif was to be French-like. .. Neologism Herbert (1990: 460) mentions coinage of words as a source for hlonipha words in Zulu. Several authors report that a number of words in the added lexicon are of unknown origin (Lamberti 1993) or possibly an arbitrary coinage (Teferra 1987: 47). Unless the actual coinage is witnessed, it remains difficult



 Maarten Mous

to prove; thus from a methodological point of view claims of coinage are best avoided. . Scope of paralexification Now that we have seen that the phenomenon of paralexification is not unique and in all contexts is essentially the same, we have to pay attention to the differences in scope of the phenomenon and to how these differences correlate with the function and social setting. Within paralexification the following factors may be of importance. .. Extent In the mixed languages such as Ma’a and Media Lengua, paralexification applied to a very large extent, in fact to most of the lexicon. In registers of respect, the number of words affected is sometimes large but not on the scale of the mixed languages. In slang, the number of words is smaller still. .. Part of lexicon The phenomenon of paralexification is sometimes limited to or concentrated in certain semantic domains of the lexicon. In a mixed language such as Ma’a we see that paralexification is nearly complete in the basic vocabulary but largely absent in various domains for specialised vocabulary, such as flora and fauna. Registers of respect, however, tend to affect all parts of the lexicon. Paralexification can even apply to function words, as is the case of Ma’a possessives (see (4) above) and Yemsa pronouns (see 2.2.1). Slang and ritual languages are restricted to one semantically or functionally coherent part of the lexicon. .. Conscious or not (function or no function) A very important factor is whether people are conscious about which form of the paralexis they are using. As soon as people are conscious about this, there is a function for the paralexicon. If a large part of the lexicon exists in parallel forms, decisions as to which form to use are probably always conscious and functional. In the case of Swahili intrusion on Bowe, people are not conscious about which form they use. People cannot decide to speak the one or the other register. In the case of Ma’a and Mbugu, there often is some code-switching between the two, but people are able to limit themselves to one of the two registers. That is to say, speakers are conscious about the fact that there are

Paralexification in language intertwining

two registers, already evident from the different names for the two registers, and speakers can choose which one to use in order to fulfill a certain function. .. Function of the extra set of forms A frequent function of the paralexicon is secrecy. Secrecy is the function of the extra set of words in the case of Shelta for example (Grant 1994), but also in numerous other cases. In the case of Ma’a, the function is the expression of ethnicity. In the cases of taboo on names and words similar to names of certain relatives, the function is respect. The paralexicon may serve a religious function, or it may signal in-group solidarity. .. Ways of getting the added word form Not all the strategies are appropriate in all circumstances. Derivation cannot be used when the added forms in the paralexification have the function of a secret code. In the cases such as Ma’a where the paralexicon is motivated by desire for group-distinctiveness, derivation is not enough. On the other hand, derivation is very common in the registers of respect where it serves the purpose of avoidance and not of group-distinctiveness. .. Number of language sources for the added word form If the paralexicon is used for ethnic reasons, borrowing is the most common source of the added word. The borrowing is from one language, as is the case for Michif and Media Lengua, if the paralexicon reflects a positive attitude to that language. Borrowing from a variety of languages either reflects a negative attitude to the basic language, as is the case in Ma’a, or serves the function of avoidance out of respect. .. Range of topics for communication If the added lexicon is limited in the set of domains in which it can be used, the range of the paralexicon will be limited too, as it is in Rotwelsch, or in a ritual language such as Callahuaya (Muysken 1994). In cases such as Ma’a where the paralexicon is adequate for any topic of conversation the paralexicon is larger, but it is still more limited than the standard lexicon. .. Individual or community phenomena, coded or not Imperfect informal second language acquisition is individual and the phenomena differ from individual to individual and from occasion to occasion. The other cases of paralexification mentioned in this chapter, however, are conven-



 Maarten Mous

tionalised for a community of speakers, both the registers for respect and the mixed languages. .. Acquired during childhood or not Sometimes the paralexicon is acquired during childhood as a first language, Ma’a for example. In many cases it is acquired later, such as registers of respect, initiation languages and languages created by adolescents, for example Petjo (Van Rheeden 1994) and Javindo (De Gruiter 1994).

. Final remarks The intertwining character of mixed languages is the result of a process of paralexification as I have shown for Ma’a (Mous to appear). When the intertwining register has become the only register, previous paralexification can still be recognised because of the morphological properties. Intertwined or mixed languages have undergone paralexification that applied to a large extent of the vocabulary; that is adequate for all topics of communication; that it is coded and shared by a speech community. Furthermore, paralexification can, for certain purposes, affect precisely basic vocabulary. In those cases, the basic vocabulary loses its aura of conservatism and should not be used as an argument in historical linguistic disputes.

Notes . In the introduction to a recent collection of papers on registers, one of the editors, Biber, gives a definition of ‘‘register’’ that excludes these ‘‘registers of respect’’ since he states ‘‘lexical choice itself does not typically mark a register’’ (Biber 1994: 34). In the case of the registers of respect, the choice of register may be more conscious and discrete, and therefore primarily lexical, compared to the examples in the collection. However, I consider the exclusion of lexical phenomena from register studies unfortunate and continue to use the term ‘‘register’’.

References Bakker, Peter. 1997. ‘A Language of Our Own’: The genesis of Michif, the mixed Cree–French language of the Canadian Métis. (Oxford Studies in Anthropological Linguistics, 10.) New York and Oxford: Oxford University Press.

Paralexification in language intertwining 

Bakker, Peter and Maarten Mous (eds.) 1994. Materials on Mixed Languages. Amsterdam: IFOTT Biber, Douglas. 1994. ‘‘An analytical framework for register studies.’’ In Sociolinguistic perspectives on register, D. Biber and E. Finegan (eds), 31–56. Oxford: Oxford University Press. De Gruiter, Miel. 1994. ‘‘Javindo, a contact language on pre-war Semarang’’. In Materials on mixed languages, P. Bakker and M. Mous (eds.), 151–159. Amsterdam: IFOTT. Grant, Anthony. 1994. ‘‘Shelta’’. In Materials on mixed languages, P. Bakker and M. Mous (eds.), 123–150. Amsterdam: IFOTT. Hamers, Josiane F. and Michel H. A. Blanc. 1989. Bilinguality and Bilingualism. Cambridge: Cambridge University Press. Haviland, John B. 1979. ‘‘Guugu Yimidhirr brother-in-law language’’. Language in Society 8: 365–393. Herbert, Robert K. 1990. ‘‘Hlonipha and the ambiguous woman’’. Anthropos 85: 455–473. Kleinewillinghöfer, Ulrich. 1995. ‘‘Don’t use the name of my dead father. A reason for lexical change in some Northwestern Adamawa languages (Northeastern Nigeria)’’. Afrika und Übersee 78: 125–137. Kolbusa, Stefanie. 1995. ‘‘Schwiegermeidung bei den Nyakyusa’’. MS, Universität Bayreuth. Kunene, D. P. 1958. ‘‘Notes on Hlonepha among the Southern Sotho’’. African Studies 17: 159–182. Lamberti, Marcello. 1993. Materialien zum Yemsa. (Studia Linguarum Africae Orientalis, 5.) Heidelberg: Universitätsverlag C. Winter. Mous, Maarten. to appear. ‘‘Ma’a as an ethno-register of Mbugu’’. Sprache und Geschichte in Afrika 15/16. Muysken, Pieter. 1994. ‘‘Media Lengua’’, ‘‘Callahuaya’’. In Materials on mixed languages, P. Bakker and M. Mous (eds), 201–211. Amsterdam: IFOTT. Osherton, Daniel N. and Howard Lasnik (eds.). 1990. An invitation to cognitive science, vol. i: Language. Cambridge, MA: MIT Press. Rayfield, J. R. 1970. The Languages of a Bilingual Community. (Janua linguarum, series practica, 77.) The Hague: Mouton. Salmons, Joe. 1991. ‘‘Register evolution in an immigrant language: The case of some Indiana German dialects’’. Word 42: 31–56. Smith, Norval. 1995. ‘‘An annotated list of creoles, pidgins, and mixed languages’’. In Pidgins and Creoles: An Introduction, J. Arends, P. Muysken and N. Smith (eds.), 331–374. Amsterdam: John Benjamins. Teferra, Anbessa. 1987. ‘‘Balissa: Women’s speech among the Sidama’’. Journal of Ethiopian Studies 20: 44–59. Van Rheeden, Hadewych. 1994. ‘‘Petjo: The mixed language of the Indos in Batavia’’. In Materials on mixed languages, P. Bakker and M. Mous (eds), 223–237. Amsterdam: IFOTT. Van Rooyen, Christian Stephanus. 1968. ‘‘A few observations on the hlonipha language of Zulu women’’. Limi 5: 35–42. Yilma, Aklilu. 1992. ‘‘The linguistic etiquette of Yemsa’’. Journal of Ethiopian Studies 25: 1–14.

Chapter 5

Pidginization, creolization and creoloids in Stockholm, Sweden Ulla-Britt Kotsinas

.

Introduction

Pidgins are often characterized as simplified and irregular contact varieties used by non-native speakers, containing no or few grammatical features and having a restricted lexicon. Creoles, on the other hand, are mostly characterized as languages with normal grammatical features and an expanded lexicon, developed by those who have grown up in an environment where a pidgin is the means of communication between different groups. As observed by many scholars, this creolization process may take place in quite a short time, i.e. in the course of a few generations. Mühlhäusler, however, remarks that a dramatic development of pidgins can take place in the absence of creolizing children (Mühlhäusler 1986: 182). Bickerton (1981) and Mühlhäusler (1986) both observe that the pidgin input plays an important role in creolization, and Sankoff (1980: 155) points out that: ‘‘nativization, that is, child acquisition, is not the only thing that can lead to grammatical expansion. Adult, second-language speakers already had done a pretty good job of it by the time native speakers of Tok Pisin came along’’. It has been suggested that pidgins contain certain lexical and grammatical contributions from the speakers’ native languages, and both pidgins and creoles are sometimes referred to as ‘‘mixed’’ languages. Mühlhäusler (1986: 182), however, rejects the opinion that substrate languages may have a decisive impact on the development of creoles, particularly as regards syntactic innovation. Instead, he points to universal principles of language development and pragmatic factors and remarks that there are significant parallels between pidgin expansion and second-language acquisition. Both Mühlhäusler and other scholars have pointed to structural similarities between pidgins, and the linguistic outcome of the immigration to Western Europe and Scandinavia

 Ulla-Britt Kotsinas

during the last few decades, the so-called Gastarbeiterdeutsch in Germany, and similar varieties in other countries (Clyne 1968; Meisel 1977a, b; Heidelberger Forschungsproject 1977; Holm 1989; Kotsinas 1985, 1989a, 1995). In the following, I discuss some apparently deviant features in pidginized varieties of Swedish, as used by adult first generation immigrants. These features are not primarily caused by interference from the speakers’ mother tongues. Instead, they are the results of compensatory strategies used by the speakers to fulfil their grammatical and pragmatic needs. Most of these strategies seem also to be utilized in pidgins, and just as in pidgins they represent the very first steps in an incipient creolization process in the course of which grammatical markers are developed. I will also briefly discuss the emergence of new linguistic varieties used by second generation immigrants in Sweden.

. Sweden today Sweden has long been considered a monolingual country, in spite of the fact that minority languages such as Lappish, Romany and Finnish have been spoken there for many centuries. Today, however, Sweden is extremely multilingual and multicultural. About 140 foreign languages are represented in the country, and in certain areas of Sweden the social and linguistic situation is remarkably similar to that in plantation areas, where the ‘‘classic’’ pidgins and creoles, such as Hawaiian Pidgin English, Tok Pisin, etc. once arose. Since the mid-sixties Sweden, like many other West European countries, has received a huge wave of immigrants from agricultural areas in southern Europe and, more recently, refugees from different parts of the world. Approximately 20 per cent of the population now consists of immigrants and their children, and in the age group 16–24 about every fourth person was either born abroad, or has one or two parents born abroad. In contrast to the normal situation in many other European countries, the immigrants in Sweden do not live isolated from each other in different ethnic neighbourhoods. Most of them live in suburbs, often called ‘‘concrete ghettos’’, on the outskirts of the big cities. Here 50–70 per cent of the inhabitants are of foreign origin and between 30 and 40, and in some cases up to about 100, languages are represented. Also, first or second generation immigrant children are in an overwhelming majority in kindergartens and schools, ranging from 60–100 per cent. In areas like these, the average income is low,

Pidginization, creolization and creoloids 

unemployment figures high, and recipients of social security benefits comparatively numerous. Many immigrants living there have very limited contacts with native Swedes, partly due to social barriers, partly due to the type of job — if any — available to them. Native Swedes as well as those immigrants who have lived in the country for many years tend to move away to more ‘‘Swedish’’ areas. They are constantly being replaced in these areas by more recent immigrants and refugees. Since no ethnic group, not even the Swedes, is in a majority, Swedish has to be used as a lingua franca in shops, cafeterias, post offices, schools, etc. The Swedish spoken by adult first-generation immigrants varies a great deal, ranging from broken varieties, used by newly arrived immigrants and refugees, to near-native varieties, used by those who have spent a longer time in Sweden and have good contacts with Swedes. However, there are also a number of immigrants, usually low-skilled workers, who have spent many years in Sweden but still use contact varieties which exhibit certain similarities to pidgins. So far, however, no stable pidgin seems to have crystallized, possibly due to the demographic instability of these areas. In the following discussion the term Immigrant Swedish (IS) is used as a common term for pidginized varieties of Swedish spoken by many immigrants and refugees who have lived in Sweden for a number of years. Depending on the speaker’s length of stay in Sweden, contacts with native Swedes, mother tongue, linguistic awareness and many other factors, these varieties display a great deal of variation, but there are also striking similarities between them as regards the solutions of linguistic problems. Most of the examples of Immigrant Swedish given below, are derived from a corpus consisting of recorded conversations among one Polish and five Greek informants (Kotsinas 1982, 1985). At the time of immigration these informants were 23–46 years old, and at the time of recording they had spent three to fifteen years in Sweden. They were all employed as cleaning personnel at Stockholm University. In addition, there are also a few examples from Finnish-, Turkish- and Spanish-speaking immigrants, who had spent between five and twenty-one years in Sweden. All the informants were unskilled workers with low educational qualifications, and few contacts with native Swedes, and they all lived in suburbs like the ones described above. None of them had attended Swedish language courses, i.e. they had acquired their Swedish informally. Examples similar to the ones quoted here are, however, also easily found in varieties used by immigrants with other mother tongues. For comparison, examples are taken from different pidgins and creoles,

 Ulla-Britt Kotsinas

and also from German Gastarbeiterdeutsch (IG) and the interlanguage of English learners (IE). See the list of abbreviations at the end of the chapter.

. Immigrant Swedish At first sight, pidgins may seem quite irregular, and certain pidginized immigrant varieties may give the same impression. In the following excerpt, the speaker, a 45-year-old Greek man who had spent nine years in Sweden at the time of the recording, complains about difficulties he had in getting the correct salary for his work. The variety used by this speaker deviates remarkably from native Swedish, and is quite difficult to understand for a native speaker. (1) Ja skriva kontratt, de system inte, skriva på kontratt, till exempel hundra kronir varje kväll, ja stanna en timmar eller två timmar, kommer exepetor stanna tre timmar. Mycke rum, eh mycke rum, kanske tre, fyra rum, dhiladhi rum, den e stor, den e, lite alla stor, förstår du mej? Fyra rum, bra bra, okej, varje kväll fyra rum, de samma jobba, men check me tvåtusenhundra, den andra check entusentvåhundra, också skatt, sjuhundra. Ja jobba fem fem år.Varför inte betala sju, sjuhundra kronir. Fem år Ulla, förstår du mej? En nästa gammal år, femtitre timmar fel, femtitre timmar, fel, ettusentrettihundrafemti kronir, fel. Den snälla kvinna på facket hjälpa, Dimitris, ja hjälpa, okej, tie dagar varsegod pengar. Varför inte ett år, ett år, ja? [. . .] Betala facket, betala skatt, varför inte hjälpa. Femtifyra kronir varje månad, varsegod. Tvåhundra betala hjälpa, varför? ‘I write contratt (=contract), it system not, write on contratt, for example, hundred crowns every night, I stay one hours or two hours, come supervisor stay three hours. Much room, eh much room, maybe three, four room, dhiladhi (Gr. = ‘that is’) room, it is big, it is, little, all big, you understand me? Four room, good good, okay, every night four room, it same work, but check with two thousand hundred, the other check one thousand two hundred, also tax, seven hundred. I work five five year. Why not pay seven-, seven hundred crowns, five year Ulla, you understand me? One next old year, fifty-three hours wrong, fifty-three hours, wrong, one thousand thirty hundred fifty crowns, wrong. The nice woman on union help, Dimitris I help, okay, ten days here you are money. Why not one year, one year, yes? [. . .] Pay union, pay tax, why not help. Fifty-four crowns every month. Here you are, two hundred pay help, why?’

Pidginization, creolization and creoloids 

Like most pidgins, IS-varieties like this one are characterized by a very restricted vocabulary with few or no synonyms. Grammatical morphemes such as inflections, articles, copulas, auxiliaries and dummy verbs tend to be omitted, and neither tense, aspect or number is morphologically expressed — cf. examples (2) to (6). Verbs, nouns and adjectives appear in invariant forms. With nouns the plural form may be used to denote both singular and plural, as in (4), or vice versa. Verbs usually appear in what looks like the infinitive form, as in (3) and (5), although sometimes present or past tense forms appear, as in (6). (2) IS: svenska svårt Swedish difficult ‘Swedish [is] difficult.’ (3) IS: ja skriva kontratt I write contratt ‘I [have] sign[ed] [a] contract.’ (4) IS: en timmar eller två timmar one hours or two hours ‘one hour or two hours’ (5) IS: ja hjälpa I help ‘I [will] help you.’ (6) IS: kommer jag en dag på en . . . come I one day on a . . . ‘One day I came to a . . .’

Conjunctions are learned comparatively late on, and subordinate clauses are rare. Instead, main clauses are juxtaposed in paratactic constructions, and the listener has to infer their connection, as in (7) and (8). Indirect speech is replaced by direct speech, as in (9) and (10). (7) IS: han säga hon född i by he say she born in village ‘he says [that] she [is] born in [a] village.’ (8) IS: ja inte kommer, ja jobba I not come, I work ‘I can’t come, [because] I am working.’

 Ulla-Britt Kotsinas

(9) IS: å han säjer men Maria, inte behöver göra här and he say but Mary, not need to.do here ‘and he said that I didn’t have to do (=clean) there.’ (10) IG: und dann ingeniör sagen, nächste monat nicht komme arbeit and then engineer say next month not come work ‘and then the engineer said that I didn’t have to come to work next month’

There are also features in pidginized varieties of Immigrant Swedish which at a first glance might seem deviant, or even unnecessary. At a closer glance, however, one finds that these are the effects of strategies used by the speakers to compensate for their restricted vocabulary, and the lack of grammatical markers. Such strategies are repetition, circumlocution, analytic decomposition, lexical over-use and semantic over-extension. . Repetition The frequent repetitions of words and phrases in pidginized varieties may give a monotonous impression, as if the speaker keeps saying the same thing over and over again. As in other natural languages, however, the repetitions have a wide range of functions (Kotsinas 1983), only with the difference that in some cases a repetition is the only means available to the speaker of a pidginized variety, while the native speaker has access to synonymous expressions. In example (11) the speaker uses the repetition to emphasize his indignation at being unjustly treated, and in (12) to (14) a native speaker might well have used intensifiers like really, very etc. In (15) to (17) the repetitions of the verb phrases have aspectual functions. In (15) the repetition of the durative verb vänta ‘wait’ marks extended duration, in native Swedish alternatively marked by the use of an adverb, such as länge ‘for a long time’. In (16) and (17), on the other hand, the repetition of the verb marks iteration. Similar examples are found in pidgins (18): (11) IS: femtitre timmar fel, femtitre, timmar fel fiftythree hours wrong fiftytree hours wrong ‘[I got] wrong [salary for] fifty-three hours [of work].’ (12) IS: bra bra good good ‘very good/excellent, etc.’

Pidginization, creolization and creoloids

(13) IS: gagster, gagster, bara gagster, samma me gagster gangster gangster just gangster same with gangster ‘[They are] real/bloody gangsters.’ (14) IS: vi kan inte lära så fort, svenska språk, vi kan inte we can not learn so quickly Swedish language we can not ‘We really can’t learn Swedish that fast.’ (15) IS: ja vänta vänta vänta I wait wait wait ‘I waited for a long time.’ (16) IS: skriva, skriva, skriva write write write ‘They wrote several times.’ (17) IS: kommer på torsdag, kommer på torsdag, kommer kommer come on Thursday come on Thursday come come kommer come ‘They said several times that the money would come next Thursday.’ (18) HPE: i stei tel mi, o neks wik, hi kamin, kamin ‘He kept telling me that he would come the following week.’

. Lexical over-use The impression of continual repetitions in pidginized immigrant Swedish varieties is reinforced by the frequent use of certain phrases, such as problem ‘problem’, det spelar ingen roll ‘it doesn’t matter’ (literally: ‘it doesn’t play any role’), till exempel ‘for example’, kanske ‘maybe’, and varför ‘why’. Normally these lexical items are used in approximately the same way by the IS-speaker as by native speakers, although much more frequently, i.e. they are over-used (Kotsinas 1984). Similar words and phrases seem to be frequent in Immigrant German (Meisel 1975), Russenorsk (Broch and Jahr 1981), and possibly in most pidgins. The frequency of phrases like these indicates that they fulfil communicatively and pragmatically important functions. Problem, for instance, occurs in contexts where the speaker wants to express that something is problematic, difficult, etc. or, when combined with a negation, the opposite, namely that someting is unproblematic, easy, irrelevant etc., cf. (19) to (21). Det spelar ingen roll is used to express the speaker’s lack of interest



 Ulla-Britt Kotsinas

or his indifferent attitude — real or pretended — towards something. The native speaker might use phrases meaning ‘it doesn’t matter’, ‘who cares’, ‘never mind’, ‘same thing’, etc., as in (22) and (23). Similarly, egal is used by Spanish and Italian immigrants in Germany much more frequently than in Standard German in meanings like ‘it doesn’t matter’, ‘same’, etc. (24) (Meisel 1975: 35). Both problem and, to a certain extent, det spelar ingen roll have equivalents in many of the immigrants’ native languages, for example Greek próvlima ‘problem’and dhen pézi rólo (lit. ‘it doesn’t play any role’) ‘it doesn’t matter’. (19) IS: ingenting problem me E. idag? nothing problem with E. today? ‘Have you had any problems/difficulties with E. today?’ (20) IS: ingen jag problem no I problem ‘I have no problems.’/‘It is OK with me.’ (21) IG: mir alles probleme me all problems ‘I have problems/difficulties.’ (22) IS: inte spela ingen roll säga du, spela mycke stora roll not play no role say you play very big role ‘You say that it doesn’t matter, but it does.’ (23) IS: kanske svensk, utlänning spela ingen roll maybe Swede foreigner play no role ‘Swedes and foreigners are equal.’ (24) IG: für mich ist egal for me is same ‘It doesn’t matter to me.’

Till exempel ‘for example’ refers to different kinds of exemplifications. In (25), hundra kronor ‘a hundred crowns’ is mentioned as an example of what the speaker might earn in one evening, and in (26), sju dar ‘seven days’ indicates how long time a certain activity might take. In these cases till exempel, thus, takes the place of modifying words meaning ‘approximately’, ‘about’, ‘roughly’ etc. In (27) till exempel modifies the whole following clause, and indicates that it is just a hypothetical example. A native speaker probably would have chosen a conditional clause instead, a solution not available to the immigrant speaker, who, in this case, doesn’t know the conjunction om ‘if ’.

Pidginization, creolization and creoloids

(25) IS: till exempel hundra kronir varje kväll for example hundred crowns every night ‘[Normally I get] about a hundred crowns every night.’ (26) IS: till exempel sju dar for example seven days ‘about a week’ (27) IS: till exempel inte läsa, på kasta, de inte bra for example not read on throw it not good ‘If for example you don’t read [the papers], but throw [them] away, maybe something unpleasant/bad will happen.’

Another marker of approximation is kanske ‘maybe’, as in (28) and (29). In pidginized varieties of German, vielleicht ‘maybe’ is used in a similar way, as in (30). (28) IS: kanske tjugofem grek maybe twenty-five Greek ‘about twenty-five Greeks’ (29) IS: kanske halv en timma kanske maybe half an hour maybe ‘about half an hour’ (30) IG: ich laufen [. . .] vielleicht halbe stunde laufen I run [. . .] maybe half hour run ‘[. . .] about half an hour’

In cases like these, kanske is placed in direct collocation with the modified phrase, usually preceding it, but sometimes following it, or, as in (29), both preceding and following the modified phrase. However, when placed sentenceinitially as in (31) (cf. also the Russenorsk example (32)), kanske usually marks the irreality of the whole proposition, i.e. it functions as a marker of epistemic modality. A native speaker would have had a choice among the following: a modal auxiliary, (a whole range of) adverbs, or a sentence meaning ‘I believe that . . .’ etc. In (33) the first clause could, like the first clause in (27), be expressed by a native speaker as a conditional clause. Both examples are uttered by the same speaker, the 45-year-old Greek man, from whom example (1) was also taken. There seems, however, to be a difference in this speaker’s use of till exempel and kanske in utterances like these. In (27) only the first clause is hypothetical



 Ulla-Britt Kotsinas

(till exempel inte läsa, på kasta ‘for example not read, on throw’), while the second (de inte bra ‘it not good’) is rather a statement of the consequence of the hypothetical event. In (33), however, kanske modifies both clauses, i.e. both the condition and the consequence are hypothetical. (31) IS: kanske den mycke gammal, mycke år den maybe it (=she) very old much years it (=she) ‘She may be very old.’/ ‘She is probably very old.’/’I think that . . .’ (32) RN: kanske tvoja rik man maybe you rich man ‘You are probably rich.’/‘I believe that you are rich.’ (33) IS: kanske spela du entusen kommer etthundra maybe play you one-thousand come one.hundred ‘If you bet 1000 [on slot machines] you may get 100.’

When placed clause-finally, on the other hand, kanske in Immigrant Swedish rather seems to be a marker of yes/no-questions, as in (34). Similar examples are found in Russenorsk, only with the difference that kanske ‘maybe’ is here placed sentence-intitially (35). (34) IS: din pojke kanske? your son maybe? ‘Is this your son?’ (35) RN: kanske kapitan på kontor? maybe captain on office? ‘Is the captain in his office?’

Finally, varför ‘why’ is frequently used in combination with inte ‘not’ as a marker of a rhetorical question expressing the opinion of the speaker, as in (36). A native speaker could also have used a main clause such as ‘I think’ in combination with a subordinate clause. In pidginized German warum ‘why’ is used in a similar way, cf. (37), (38). (36) IS: ja betala facket, betala skatt, varför inte hjälpa I pay union pay tax why not help ‘Why don’t they help me? I pay fees and taxes.’/‘I think that they ought to help me since I pay fees and taxes.’

Pidginization, creolization and creoloids

(37) IG: ich bezahl, warum ich niks drink? I pay why I nothing drink? ‘I pay, why am I not allowed to drink?’ (38) IG: ich eine pakete zigaret ein tag. Ja, warum ich niks rauche? I one packet cigarette one day Yes why I nothing smoke? ‘I smoke a packet of cigarettes every day. Why am I not allowed to smoke [here]?’

. Analytic decomposition Pidginized second language varieties are, like pidgins, mainly analytic, i.e. grammatical notions are expressed rather in terms of unbound morphemes than in terms of the inflections or stem alternations used in the target language. Analyticity also occurs at the lexical level, however. In the initial stages of language acquisition, semantically complex lexical items are difficult to learn. For this reason, the learner unconsciously concentrates on learning basic, unmarked words, i.e. lexical items with comparatively few semantic features. To express a semantically complex concept, for which the target language uses one word, the speaker has to combine basic lexical items. If he knows a word for positive evaluation, for instance good, he may combine this word with a negative element (‘not + good’) to replace words meaning ‘bad’, ‘evil’, and other such expressions of negative evaluation, as in (39), and in Immigrant German (40). If he wants to express the concept of ‘unmarried’ or ‘bachelor’ but doesn’t know these words but does know the word for ‘married’, he may use this word in combination with a negative element (41). He analyzes, as it were, the concept and decomposes it into its semantic components. This strategy, which may be called analytic decomposition, is economic, particularly in the inititial stages of learning, since it allows the learner to manage with only one of two antonymous words, at least temporarily. (39) IS: vi inte så bra svenska we not so good Swedish? ‘We don’t know Swedish very well.’/ ‘our Swedish is bad.’ (40) IG: niks gut Wetter nothing good weather ‘bad weather’



 Ulla-Britt Kotsinas

(41) IS: ja inte gift I not married ‘I am unmarried/a bachelor.’

Analytic decomposition in combination with modified repetition is a useful strategy to express approximation instead of using semantically vague modifiers with meanings such as ‘some’, ‘a little’, ‘a few’, ‘longer’, ‘about’ etc. In (42) the speaker indicates approximation by giving examples of sums of money. In (43) the same speaker wants to explain that someone is a teenager. Since he doesn’t know this word, he mentions a few examples of ages and, in order to emphasize further the approximate nature of his statement, puts the numerals in the wrong order. (In an earlier utterance he put these numerals in the correct order.) In (44), taken from the longer excerpt in example (1), the speaker wants to explain that normally he stays at work for a few hours, ‘‘one hour or two hours’’, but if the supervisor turns up he stays there longer, ‘‘three hours’’. In (45), the same speaker, who doesn’t know the pronoun ni ‘you-pl’, replaces this word with with a lengthy list of persons, including du ‘you- sing’ who are all said to know Swedish, in contrast to himself. (42) IS: tio kroni, tjugo kroni ten crowns twenty crowns ‘some money’ (43) IS: femton, sjutton, sexton fifteen seventeen sixteen ‘about sixteen years old’, ‘a teenager’, ‘very young’ (44) IS: ja stanna en timma eller två timma, kommer exepetor I stay one hours or two hours come supervisor stanna tre timma stay three hours ‘[Normally] I work a few hours, [but] if the supervisor comes I stay longer.’ (45) IS: du prata svenska, Ilias prata s., han prata s., Takis prata s. you speak Swedish Ilias speak S. he speak S. Takis speak S. ‘All of you speak Swedish [but I don’t].’

It is well known that circumlocutions of various sorts occur in pidgins and pidginized varieties. Some of these may be categorized as analytic decompositions, for example, when verbs are replaced by phrases such as ‘to make + N’,

Pidginization, creolization and creoloids 

for example mekim siga (‘make cigar’) for ‘smoke’, or noun phrases such as hand boots for ‘gloves’ (Mühlhäusler 1986: 146). In the Russenorsk example (46) below, kyrka ‘church’ is decomposed into ‘cottage [where they] speak about Christ’. In (47) the speaker wants to explain that he once wanted to buy clothes for his children. Clearly he doesn’t know the Swedish word kläder ‘clothes’ and decides to list his various garments instead. He starts off with byxor ‘trousers’ and intends to continue with a word meaning ‘shirt’ or ‘sweater’. When he discovers that he doesn’t know any of these words either, he has to take refuge in a third strategy, a gesture, pointing to his shirt. (46) RN: stova på Kristus spræk cottage on Christ speak ‘church’ (47) IS: köpa på barn byxor å den också buy on child trousers and that too [points to his shirt] ‘I wanted to buy clothes for my children.’

. Semantic over-extension So far, the discussed strategies are ‘‘natural’’ in the sense that they are also sometimes used by native speakers. However, there are also examples in pidginized speech of words used ‘‘incorrectly’’, i.e. in a non-idiomatic way. In most of these cases, the words are used in a wider sense than is possible in native speech, i.e. they denote meanings which they don’t have in native language. In Immigrant Swedish, for example, flicka ‘girl’ is used by some speakers for ‘woman’, i.e. flicka loses one of its semantic features, YOUNG, and retains only the features HUMAN and FEMALE. The word flicka is thus extended to cover a larger part of the semantic field than is normal in Swedish. Similarly, pojke ‘boy’ loses its feature MALE when it is used for ‘child’ as in (48), and only the features HUMAN and YOUNG are retained. Similar examples are kanal ‘channel’ used for ‘strait’, and stad ‘town’ used for ‘village’. (48) IS: inte pojke not boy ‘[I am] not a child.’/ ‘I am an adult.’

Semantic over-extensions are particularly frequent in the case of the semantic fields associated with certain verbs (Kotsinas 1984, 1995). With reference to the semantic field PERCEPTION, for example, many speakers use titta for

 Ulla-Britt Kotsinas

both titta ‘look’and se ‘see’, as in (49), and lyssna for both lyssna ‘listen’ and höra ‘hear’, as in (50). Titta and lyssna thus are assigned only the features SIGHT and HEARING, respectively, while more specialized semantic features are omitted. Similarly, fråga ‘ask’ is used by some speakers to replace other verbs of verbal communication. In particular, the speaker represented in example (1) uses fråga ‘ask’ to replace all other verbs within the field (with the exception for svara ‘answer’), i.e. words meaning ‘say’, ‘talk’, ‘chat’, ‘discuss’, ‘tell’, ‘request’ etc., as in examples (51) to (53). Fråga, thus, has only the feature VERBAL COMMUNICATION for this speaker. (As for (53), Swedish fråga ‘ask’ is not possible in this context, unlike English ask.) (49) IS: Jag titta en kille på fönstren I look a guy on windows ‘I saw a guy through the window.’ (50) IS: jag i S. aldrig inte lyssna lunch I in S. never not listen lunch ‘I never heard [the word] lunch in S.’ (51) IS: fråga kanske tio timmar ask maybe ten hours ‘we talked for ten hours (=a long while).’ (52) IS: kommer flicko mycke vacker, fråga me Takis come girl very beautiful ask with Takis ‘A beautiful girl came and talked to Takis.’ (53) IS: kommer på exepetor, fråga kom come on supervisor ask come ‘[I] went to the supervisor and told her to come.’

The same speaker uses komma ‘come’ (usually in the present tense form kommer, occasionally in the forms komma or kom) very frequently, and mostly in a very extended way, semantically. In fact, 25 per cent of his verb tokens consist of komma. He uses komma intransitively to denote all kinds of movement, i.e. with the meanings ‘come’, ‘arrive’ and also ‘go’, ‘walk’, ‘travel’, etc., in (52) to (54). In (53), the first instance of komma (kommer) is used for gå ’go’, while the second (kom) has the meaning ‘come’. He also uses komma transitively, i.e. where verbs meanings ‘receive’, ‘get’ and ‘give’ would be suitable, as in (55) and (56). In fact, he uses komma to denote any kind of change of location.

Pidginization, creolization and creoloids 

(54) IS: kommer på Nilsson come on Nilsson ‘[I] went to Mrs Nilsson.’ (55) IS: den flicka vill kommer cigarett that girl want come cigarette ‘The girl wanted to get/have a cigarette.’ (56) IS: kom en krona come a crown ‘give [me] a crown.’

The same speaker uses komma in an even more extended fashion, namely in cases where a native speaker would have chosen words meaning ‘become’ and ‘start’, as in(57) and (58), and ‘meet’, ‘get to know’, ‘become aware of ’, as in (59)and (60), i.e. to denote all cases of change of state. Similar examples are found in Immigrant German sentence (61). In short, this speaker uses komma ‘come’ whenever he requires a verb with the semantic feature CHANGE. (57) IS: den kommer tjugo år it come 20 year ‘She had become 20 years old.’ (58) IS: kommer på hemma come on home ‘[The disease] started at home.’ (59) IS: kommer semester, kommer Antinopoulos come vacation come Antinopoulos ‘I went on vacation and met Antinopoulos.’ (60) IS: kanske kommer fru var ja ringa maybe come wife where I call ‘My wife may understandwho I am calling.’ (61) IG: komme Sohn fümf Jahre kom in die Schule come son five years come in the school ‘My son became five years old and went to/entered school.’

Furthermore, stanna ‘stay, stop’ is also used by the same speaker in a generalized fashion, for instance with meanings such as ‘stay’, ‘remain’, ’live’, ’be alive’ etc., as in (62) and (63). In native Swedish it would not be possible to use stanna in any of these examples. In fact, stanna is used to denote any abstract state, as if it had only the feature REMAIN.

 Ulla-Britt Kotsinas

Finally, sluta ‘finish’ and, to some extent, färdig ‘finished, done’ are used by the same speaker to mark an end-point of an activity or state (63), in some cases even with the meaning ‘die’ (64). (62) IS: de kanske till exempel stanna sjuttifem år, stanna hundra år, it maybe for example stay 75 year stay 100 year de inte problem it not problem ‘One may live 75–100 years (=many years), if one does not have problems.’ (63) IS: ja alla toalett slut, färdig, stanna en toalett I all lavatory finish done stay one lavatory ‘I had finished [cleaning] all the toilets, only one was left.’ (About his job as a lavatory-cleaner) (64) IS: kanske han slut maybe he finish ‘he might be dead’

None of the above-mentioned extensions of the semantic fields of verbs have anything to do with the speakers’ mother tongue, in this case Greek, i.e. they are not the effect of interference. A great number of over-extensions are also to be found in the use of prepositions. Some speakers use på ‘on’ or, in some cases, i ‘in’ to denote location, direction and time and sometimes also for case-marking, i.e. in a way very similar to the well-known all-purpose prepositions in certain pidgins. The phrase kommer på Grekland ‘come on Greece’, for instance, may mean ‘I went to Greece’ or ‘I came from Greece’, depending on the context. In (65) på replaces genom ‘through’, and in (66) a native speaker would have used i ‘in’ to mark duration instead of på ‘on’, which in Swedish is used to mark a point on the time line. In (67) på seems to mark the object, and in (47) på in köpa på barn marks receiver instead of till or åt ‘to, for’. (65) IS: titta på fönstren en kille look on windows a guy ‘I saw a guy through the window.’ (66) IS: de första gången på mitt liv it first time on my life ‘It is the first time in my life.’

Pidginization, creolization and creoloids

(67) IS: betala på pengar pay on money ‘pay money’

. Compensatory strategies and natural language The use of compensatory strategies like the ones discussed above allows the speaker to express his basic linguistic needs by means of very restricted vocabulary and grammatical knowledge. At the lexical level, these strategies are useful when the speaker needs to express words and phrases which he has not yet learned. Intensifiers, for example, may be replaced by repetitions, and indefinite modifiers by kanske ‘maybe’, till exempel ‘for example’ or analytic decompositions. Semantically complex words may be replaced by analytic decompositions, and semantic over-extensions make it possible to use only one or a few words within a certain semantic field. The attitude of the speaker, finally, may be expressed by lexical over-use of phrases such as problem and det spelar ingen roll ‘it doesn’t matter’, and his opinion by constructions with varför ‘why’. Strategies like these occur not only in pidginized varieties but also in natural languages. The over-use of certain lexical elements such as problem, it doesn’t matter, etc., occurs more or less in all sorts of natural speech, and repetitions of different types are used in discourse to mark, for example, emphasis, iteration and duration (Tannen 1989: 69). Analytic decompositions are sometimes used when the speaker doesn’t know or remember the adequate word, and semantic over-extension is one of many causes of diachronic changes of word meaning. This indicates that the frequent use of strategies like these in pidginized speech is possibly caused or, at least, reinforced by the speakers’ communicative competence and linguistic experience of their own native language. . Grammar in pidginized varieties The view that grammatical markers are ‘‘missing’’ in pidginized varieties is true in the sense that base or target language grammatical markers are rarely used in pidginized varieties, particularly, of course, if these markers are opaque or irregular. However, there are ways in pidginized varieties, not only to compensate for lexical deficiencies, but also to mark grammatical relations.



 Ulla-Britt Kotsinas

Pidgins, particularly in the initial stages, are analytic in their structure, i.e. grammatical notions are expressed in the surface structure by the use of lexical morphemes, and this is clearly the case in Immigrant Swedish too. Number is, for instance, marked by numerals in cases where the linguistic or situational context is not enough. Locative, temporal and grammatical relations (case) are marked by one or a small number of prepositions meaning ‘on’, ‘in’ etc. Rhetorical questions are marked by varför ‘why’, yes/no-questions by the adverb kanske ‘maybe’, or an interjection, for instance ja ‘yes’, or va ‘what’, and hypothetical assumptions by kanske ‘maybe’, or till exempel ‘for example’. Essential categories in languages are the TMA-markers, expressing tense, modality and aspect. In natural languages markers of these categories usually form complex systems, very often consisting of a mixture of analytic periphrastic, and synthetic morphological methods, difficult to discover for the second-language learner, particularly if s/he acquires the language informally. As in pidgins, tense in Immigrant Swedish is marked by time adverbials such as igår ‘yesterday’, i morgon ‘tomorrow’ etc., as in (68). In sentence (69) the speaker even makes an attempt to mark sequencing by using gammal år ‘old year’ to indicate some past time, and nästa ‘next’ for a year later. (68) IS: förstår inte i dag, kanske i morgon förstår understand not today maybe tomorrow understand Approximately ‘Maybe I will understand better later/in the future.’ (69) IS: nästa gammal år 53 timmar fel next old year 53 hours wrong ‘The following year I got the wrong [salary for] 53 hours [of work].’

As for modality, epistemic modality, i.e. possibility or probability, is usually marked in Immigrant Swedish by the modal adverb kanske ‘maybe’, as in for instance (31) and (33). Deontic modality, i.e. obligation, necessity etc., is expressed by måste ‘must, have to’, used both as an auxiliary and as a main verb, as in (70) and (71). Måste seems to be learned very early, followed by the auxiliares vilja ‘want’ and kunna ‘be able to’, while temporal auxiliares are learned later. The same order of acquisition has been found in Immigrant German, where müssen ‘must’ is learned before wollen ‘want’ and können ‘be able to’ (Heidelberger Forschungsprojket 1977: 94). (70) IS: de måste fru jobba it must wife work ‘My wife has to work.’

Pidginization, creolization and creoloids 

(71) IS: vi måste paus nu we must pause now ‘We must [have] a break now.’

Another instance of deontic modality is the imperative. In Swedish, the imperative is morphologically marked in a number of different ways, depending on the conjugation of the verb, which makes this category opaque and difficult to learn. A solution to this problem is to mark a request by mentioning the addressee directly as in (72), or by using the politeness marker varsågod ‘please’, usually in sentence-initial position. The high frequency of varsågod in Immigrant Swedish indicates that it has rather a grammatical than a pragmatic function, cf. (73). Similar strategies are very frequent in Russenorsk, as in sentence (74). (72) IS: kom du tolk med mej come you interpreter with me ‘Come with me as interpreter.’ (73) IS: varsego, titta en fel please look one error ‘Look here, [I have made only] one error.’ (74) RN: værsego ju på moja skib vaskom! please you on me ship wash! ‘Clean the ship.’ (order given by a captain)

Modality and tense are interconnected, particularly as regards future tense (Traugott 1978). As pointed out by the Heidelberger Forschungsprojekt (1977b: 94) müssen ‘must’ in Immigrant German in some cases rather seems to mark future tense than modality and the same seems to apply for måste in Immigrant Swedish, as in (75). Even clearer is the connection between the marker of epistemic modality kanske ‘maybe’ and future tense in sentences (68) and (76). Similar examples are found in other pidginized varieties, in for instance the Immigrant English example (77), and also in Russenorsk, cf. (78). In (79), where a Spanish woman talks about her future plans, the irrealis nature of futurity is marked by three instances of kanske and one of I don’t know. (75) IS: lite lite, måste lite little little must little ‘[I] have to [learn] a little/some [Swedish] (in the future).’

 Ulla-Britt Kotsinas

(76) IS: kanske sextifem år kaputt finito maybe sixtyfive year kaputt finito ‘One may die at sixtyfive.’ (77) IE: maybe next time I understand more (78) RN: kanske moja på anner år kopom planka maybe I on other year buy plank ‘I may buy a plank next year.’ (79) IS: ja kanske spara lite penga å köpa lite hus å kanske I maybe save some money and buy little house and maybe ja gifta mej å kanske vet inte I marry and maybe know not ‘I may save some money and buy a little house and marry. I don’t know.’

With respect to aspect, Bybee and Dahl (1989: 85) suggest that this category is more basic in languages than the tense distinction. In child language, for example, aspect tends to be expressed early (Antinucci and Miller 1976; Bronckart and Sinclair 1973), and it is also noted that grammatical markers of aspect tend to be created at an early stage when pidgins creolize (Bickerton 1981; Traugott 1978; Mühlhäusler 1986). On the other hand, it is usually considered that no aspect markers exist at the pidgin stage. At first glance, no such markers seem to exist in Immigrant Swedish either. But as we have seen, repetitions sometimes are used to mark duration or iteration, i.e. to enforce or change the aktionsart of the verb. There are also other indications of an incipient aspect system, namely the semantic extensions of the verbs komma ‘come’, stanna ‘stay’ and sluta ‘finish’ described above, in (54) to (64), which are particularly obvious in the speech of the 45year old Greek man represented in example (1). As we have seen, komma ‘come’ is used by this speaker to denote all sorts of change. In fact, this single verb replaces all verbs with a perfective, ingressive aspect. Sluta ‘finish’ as used in this man’s variety, on the other hand, marks the end-point of an activity or a state, i.e. it functions as a marker of perfective, egressive aspect. Stanna ‘stay’, finally, is used for all abstract processes like ‘live’, ‘stay’ etc., i.e. it marks imperfective aspect. Only when it comes to denoting very concrete processes, is this man’s vocabulary more elaborated, consisting of about thirty verbs meaning ‘read’, ‘write’, ‘eat’, ‘drink’, ‘drive’, ‘clean’, ‘scrub’, ‘dance’, etc. It seems as if this speaker has

Pidginization, creolization and creoloids 

found a way to express aspect by using very few lexical items to denote states and abstract processes. (For further discussion, see Kotsinas 1989a, 1995). Similar extensions, particularly of movement verbs meaning ‘come’ or ‘go’, are found also in other Immigrant varieties.

. Grammaticalization and creolization When pidgins creolize, new grammatical markers are created out of lexical items, bleached of their original full meaning. Hopper and Traugott (1993: 97) point out that words that grammaticalize tend to have a basic, general meaning. Usually they are superordinate terms in their semantic fields, so-called hyponyms, for instance verbs like say, move and go. As a first step in this grammaticalization process, the meaning of the word becomes generalized and, at the same time, its use increases in frequency. The semantic reduction is usually accompanied by a phonological reduction. Later on, the original semantic meaning may be almost completely lost and the word becomes a function word, for instance an auxiliary. Finally the word may be cliticized, and even end up as an inflectional morpheme (Bybee and Dahl 1989: 63–67). The same process seems to be at work when pidgins creolize. In Hawaiian Pidgin English, for example, go was frequently used as a main verb and a preverbal modifier of ‘‘extremely indeterminate meaning and wildly fluctuating distribution’’ (Bickerton 1981: 31), while it has turned into a grammatical marker in Hawaiian Creole English (Bickerton 1981: 79). The lexical material used to create grammatical markers in creoles is essentially of the same type, independent of the lexifyier language and the speakers’ native languages. TMA-markers are often created from movement verbs meaning ‘come’ or ‘go’, and words meaning ‘stay’, ‘stop’, ‘finish’. In Hawaiian Creole English, for example, stei (stay) and go mark aspect and modality, respectively (Bickerton 1981: 26), while in Tok Pisin stop has developed into a locative copula (Sankoff 1980: 148). In many creoles, for instance Tok Pisin, verbs meaning ‘finish’ have become markers of perfective aspect. This tendency to create aspect and tense markers from lexical verbs involving a highly constrained set of semantic features is not restricted to pidgins and creoles. Traugott (1978: 373) remarks that there seems to be a constant tendency for tenses and aspects to be revitalized in language acquisition, language contact and language change. As demonstrated by

 Ulla-Britt Kotsinas

Traugott (1978), and Bybee and Dahl (1989), aspect and tense markers in unrelated languages are almost exclusively developed from originally locative verbs and particles, although future tense may also develop out of modals, such as desideratives meaning ‘want’ and ‘desire’, or verbs meaning ‘to owe’ or ‘to be obliged’. During this grammaticalization process, tense and aspect markers may develop into auxiliaries. Come is for instance used as an aspectual, ingressive auxiliary in Philippine Pidgin English, as in (80), like aller in French Guiana Creole (81). In Cameroon Pidgin go is used as a temporal/aspectual auxiliary, marking future tense, as in (82). There are a few similar examples in Immigrant Swedish. In (83) to (85), for instance, uttered by speakers with different native languages (Greek, Turkish and Finnish), komma is used as an aspectual auxiliary. (80) FPE: hi kam gro da paemili he came grow the family ‘The family was beginning to grow up.’ (81) FGC: li ka alé asas lui allé assasiner ‘He is going to hunt.’ (82) CP: i go kam he go come ‘He will come.’ (83) IS:

min fru kommer jobba klockan åtta my wife come work clock eight ‘My wife starts to work at eight o’clock.’

(84) IS:

sen ja komme vänta then I come wait ‘Then I began to wait.’

(85) IS:

jag semester kom å slut, jag kommer arbet I vacation come and end I come work ‘My vacation came to an end/ended, and I started to work.’

In many pidgins verbs meaning ‘finish’ have turned first into aspect markers and later on into temporal auxiliaries, often placed in sentence-final position as in the New Guinea Pidgin sentences (86) and (87). In Immigrant Swedish slut in a few cases seems to have a similar function, as in (88).

Pidginization, creolization and creoloids 

(86) NGP: em i ritim dispela buk pinis he i read this-fellow book finish ‘He has read this book.’ (87) NGP: mi tokim ju pinis I talk you finish ‘I told you.’ (88) IS:

de han gå ut, slut it he go out finish ‘He went out of/left the room.’

There are also many examples of verbs with an original meaning of ‘stay’, ‘stop’, etc., which have developed into a locative copula, as in Tok Pisin and Hawaiian Pidgin English, as in (89) and (90). A similar example in Immigrant Swedish is (91), where Swedish stanna (contrary to English stay) is incorrect. (89) NGP: hous bilong you i stap we? ‘Where is your house?’ (90) HPE: hi ste Maui ‘He is in Maui (temporarily).’ (91) IS:

den tjugo år stanna Joannina it twenty year stay Joannina ‘She was/lived in Joannina (a town) for 20 years.’

Not only verbs, however, provide material for new grammatical markers in pidgins and creoles. Prepositions may, for instance, develop into complementizers. The New Guinean Pidgin ‘‘all purpose’’ preposition long has the function of a purposive complementizer in (92) and Russenorsk på/po ‘on’ is used in the same way in (93). In Immigrant Swedish there are a few similar examples, such as (94) and (95), produced by speakers with Spanish and Greek as their respective mother tongues. (92) NGP: em i go long kisim sut ‘He went to get an injection.’ (93) RN: gå på slipom go on sleep ‘Go to sleep.’

 Ulla-Britt Kotsinas

(94) IS:

stoppa dom tre fyra timmar tåget på undersöka barne stop they three four hours train on examine child ‘They stopped the train for some hours in order to examine the child.’

(95) IS:

den barn sex månar kommer på skriva på kyrka it child six months come on write on church ‘The child had become six months old and we went to register him at church.’

. Adults as creolizers The Immigrant Swedish examples above are taken from speakers with different native languages, and the speakers have achieved different degrees of creolization or progress in acquiring Swedish. The frequency of certain compensatory strategies and grammatical solutions in some of these varieties may, of course, obstruct understanding on the part of the native speaker, particularly since they usually occur in combination with deviations in pronunciation and word order, and incorrect use or loss of Swedish grammatical morphemes. This inter-individual variation, i.e. the differences between the speakers’ varieties, may lead the native listener to the conclusion that every immigrant speaker uses his own idiosyncratic variety and that there are no rules whatsoever. It is, however, a common experience among teachers of Swedish as a second language that speakers of pidginized Immigrant Swedish, even those who have different native languages, often understand each other very well, while the Swedish teacher herself do not understand the conversation going on in the group. This lack of comprehension on the part of the teacher may have its background in the fact that s/he does not expect Swedish to be used in a pidginized way. On the other hand, the students’ understanding of each other’s varieties supports the assumption that the strategies and grammatical devices discussed above are based on universal phenomena, and that there exist common solutions to linguistic problems, irrespective of both the speakers’ native languages and the target language. The development of grammatical markers when a pidgin turns into a creole has traditionally been regarded as a result of children’s creativity. As we have seen, though, this process may be initiated even by first generation speakers. Taking into consideration that Immigrant Swedish speakers, from whom most of the examples above are taken, have developed their language with very little

Pidginization, creolization and creoloids 

contact with the Swedish society and with practically no theoretical knowledge of grammar, it is amazing how elaborated the grammatical devices are in these varieties. This elaboration is only accidentally a result of transfer from the speakers’ mother tongue. Neither is it to any important degree a result of transfer of a pre-existing pidgin from ‘‘older’’ immigrants to newcomers, i.e. an effect of learning. Instead, the elaborations are probably the effects of numerous interactions, in which the participants try their best to express their linguistic intentions and make themselves understood, i.e. the results of a collective process, as it were. In the light of these assumptions, the frequently observed intra-individual irregularities in pidginized speech can be explained as emerging from the speaker’s attempts to be clear and expressive. Even a speaker of the most ‘‘fossilized’’ variety is probably working continually to develop his language into a more effective means of communication. During his efforts to make himself understood, he tries out new words, forms and grammatical constructions, with the consequence that his lexical and grammatical systems are changing — although in some cases very slowly — either in the direction of ‘‘correct’’ target language forms or towards a creole-like language. In doing so he takes advantage of his own experience of how languages work, as well as of his human linguistic ability and universal semantic and communicative strategies.

.

New varieties?

An interesting question is what happens to the Swedish language spoken by children and adolescents who are born and raised in areas like the ones described above. Evidently, these children are exposed to a linguistic input quite different from that of children in more monolingual Swedish-speaking areas. Are they going to develop a Swedish-based creole or do they learn idiomatic Swedish? So far, no creole has emerged. Such a development is counteracted both by the strong influence of school and mass media and by the fact that no stable pidgin exists. This, however, does not mean that second generation immigrants use Swedish in exactly the same way as native Swedish young people do. In fact, many teenagers with an immigrant background use varieties of Swedish, especially in informal conversations, which deviate markedly from the varieties used by native Swedish adolescents. Terms like Rinkebysvenska

 Ulla-Britt Kotsinas

‘Rinkeby-Swedish’, Albysvenska ‘Alby-Swedish’, Gårdstenska ‘Gårdstenish’ and Rosengårdssvenska ‘Rosengård-Swedish’ are used for varieties spoken in suburban areas of the big cities where immigrants live. Jokingly, even such terms as Kebabspråk ‘Kebab language’ and Spaggesvenska ‘Spaghetti-Swedish’ occur. Varieties like these contain both typical Swedish features and features that deviate from native Swedish. Normally, most phonemes are pronounced as in the local variety of young people’s language. The most deviant feature is the intonation, often described as ‘‘choppy’’ or ‘‘uneven’’ by the speakers themselves. Exactly what gives this impression is not clear. One reason may be the non-occurrence or smaller frequency of certain reductions and assimilations. Another reason may be changes in vowel length towards less marked features, namely that short vowels are a little prolonged, long vowels a little shortened. A certain ‘‘unSwedish’’ intonation pattern containing an unexpected rise of tone may be yet another reason. As for grammar, certain deviations may occur, particularly concerning word order, gender and agreement, and the prepositional system. Usually, the deviations involve a preference for a less marked form instead of a more marked form. For example Swedish has two genders, n-gender and t-gender (en boll ‘a ball’, ett bord ‘a table’). Since no simple rules exist for the distribution of gender, this has to be learnt separately for each word. In varieties like the ones discussed the n-form, which is the most frequent one in Swedish, often replaces the t-form. Another example is the use of prepositions. Swedish has a great number of prepositions, and in the Swedish of second generation immigrants, less marked prepositions such as på ‘on’ and i ‘in’ are often preferred to more marked ones. Some ‘‘incorrect’’ prepositional phrases seem to be so fixed that they may be regarded as features of the local dialects, rather than‘‘errors’’, for instance läsa på tidningen ‘read on the newspaper’, where på ‘on’ replaces idiomatic Swedish ‘in’. Very frequent are also certain types of word order deviations. As for vocabulary, some frequent Swedish words are used in a semantically extended way. The distinction between gå ‘go, walk’ and åka ‘travel by vehicle’, for instance, is obligatory in Swedish, while verbs meaning ‘go’ in English and many other languages are used for both meanings. In immigrant adolescent language, however, gå is used in contexts like Jag ska gå (=åka) till Grekland i sommar ‘I will go to Greece this summer’, which is unidiomatic from the Swedish point of view. Like the use of certain prepositional phrases, this phrase seems to form part of the local dialect.

Pidginization, creolization and creoloids

Even more striking, however, are the many slang words, originating from the many different languages spoken in the area, most of which are quite incomprehensible to a native Swede. For ‘girl’, for instance, kopella and munara (Greek), giz and orospo (Turkish), conyo (Spanish) and harra (Arabic) can be heard, and for ‘boy’ raklo (Romany), silo (Turkish), kusipää (Finnish) and malakka (Greek). Many of these words originally had a derogatory or obscene meaning. Frequently used words are abou, said to be of Arabic origin, used in exclamations like Abou what a car! ‘What a nice car!’ and lan (Turkish), when addressing someone, Har du parra lan? ‘Do you have money, man?’, where parra is also imported (Turkish). Finally, there are translations into Swedish of expressions like ‘I swear’, ‘Fuck your mother’ and ‘I will kill you’, phrases that normally do not occur in Swedish. Lately, many English words originating in various American youth cultures, particularly that of hip-hop, have become frequent. Also new speech styles have come in, namely verbal duels very similar to the ones used in Turkey (Dundes, Leach and Öskök 1986), and in Black English (Labov 1972). Taken together, features like these may easily give the impression that teenagers raised in certain suburban areas don’t know Swedish. All these features — i.e. grammatical deviations, aspects of pronunciation, and slang words — are used by most of the second generation teenagers in a very similar way. In fact, it is impossible for an adult Swede to guess the socalled home language of individual adolescents, a fact that indicates that the deviations have very little to do with interference from their parents’ languages. Instead, the features may be interpreted as markers of local identity. In fact, the use of varieties like these do not differ in any fundamental way from the use of most other varieties spoken by young people, particularly those which, like Black English as described by Labov (1972), occur in socially deprived environments. As in these varieties, the use of certain slang words and other linguistic elements regarded as ‘‘incorrect’’, are to a certain extent used to give an impression of toughness and opposition to mainstream society. Some speakers even mention that some slang words are kept secret in interactions with the police and other authorities. On the other hand, there is a certain variation in the use of this variety. The most elaborate variety is used in spontaneous conversation with friends, particularly by boys, while girls use the variety in a less explicit form. There is also an obvious variation in terms of situation, conversational partner, and topic of conversation, a fact that indicates that these varieties are not, at least not exclusively, an effect of insufficient knowledge of Swedish. As a matter of fact, many individuals are bidialectal, i.e. they are able to shift between almost





Ulla-Britt Kotsinas

perfectly normal local Stockholm, Gothenburg etc. dialect varieties, and the second-generation immigrant variety. Remarkably, many Swedish teen-agers born in areas like the ones described above, master these varieties too to different degrees, i.e. ‘‘not as perfectly as us’’ as an immigrant boy expressed it. The emergence of varieties like these may thus be interpreted partly as a consequence of the multilingual and multicultural society in which these adolescents are raised, and partly as strong markers of local identity and solidarity. No doubt, the Swedish language used by many teenagers in these areas contains a certain grammatical instability, resulting in a preference for unmarked structures and semantic extensions, similar to, but not as far reaching as the ones used by first-generation immigrants. Incidentally, one finds also mistakes in the understanding and use of Swedish vocabulary, mainly due to insufficient contact with native Swedes outside these suburbs, and a lack of experiences of Swedish customs and way of life. One 15-year-old boy, for example, believed that the word suburb meant ‘a place where many immigrants live’, which, of course, was a by-product of his own experience. The many slang words from the languages in the area, on the other hand, should rather be regarded as innovative (non-simplified) features of a kind that is normal in all kinds of teenage language, which, among other things, are a marker of a common identity in areas like these. Varieties like these where creole-like features occur, although no pidgin ancestors exist, have been called creoloids, and are well-known from other parts of the world, where cultural and linguistic contact is at hand, for example patois or British Black English in England (Le Page and Tabouret-Keller 1985; Hewitt 1986; Edwards 1986; Rampton 1991), or Chicano English on the AmericanMexican border (Penfield and Ornstein-Galicia 1985). Typically, ‘‘odd’’ intonation patterns, less marked grammatical features, and new slang words, are utilized, and like the Swedish varieties, these varieties are mainly used by young people. (For further discussion, see Kotsinas 1988, 1989b, 1992.)

. Conclusion The Swedish used by many first-generation immigrants, displays many features, typical of pidginization, such as a preference for unmarked grammatical forms, semantic extension, and analytic solutions of lexical and grammatical problems. There is also evidence for the assumption that a creolization process may already be operating in the language of first-generation immigrants. Contrary

Pidginization, creolization and creoloids

to common popular belief, interference from the mother tongue plays a minor role in the formation of these varieties, with the exception of pronunciation, where such interference is usually obvious. Instead, universal linguistic and communicative strategies seem to play an important role. The varieties of Swedish used by second-generation immigrant teenagers in certain areas contain features which could be ascribed to the lack of linguistic contact with native speakers of Swedish. Here, too, the preference of less marked features is evident. However, there are also obvious nonsimplified innovative features in these varieties, particularly as regards vocabulary and, to a minor extent, speech style. Different from first-generation varieties, second-generation varieties are comparatively homogeneous. On the other hand, there is important inter- and intra-individual variation with respect to the extent to which these varieties are used. This, in turn, has to do with the marking of solidarity with parents and newly arrived friends, group identity, and with opposition to or acceptance of the norms of Swedish society and plans for the future.

Abbreviations CP FPE HPE IE IG NGP RN

Cameroon Pidgin Philippine Pidgin Hawaiian Pidgin English Immigrant English Immigrant German, Spanish or Italian immigrants New Guinean Pidgin/Tok Pisin Russenorsk

References Antinucci, F. and R. Miller. 1976. ‘‘Guyanese: A French creole’’. Journal of Child Language 3: 167–189. Bickerton, D. 1977. ‘‘Pidginization and creolization: Language acquisition and language universals’’. In Pidgin and creole linguistics, Albert Valdman (ed.), 49–69. Bloomington and London: Indiana University Press. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma Publishers. Broch, I. and E. H. Jahr. 1981. Russenorsk. Oslo: Novus Forlag. Bronckart, J. P. and H. Sinclair. 1973. ‘‘Time, tense and aspect’’. Cognition 2: 107–129. Bybee, J. and Ö. Dahl. 1989. ‘‘The creation of tense and aspect systems in the languages of the world’’. Studies in Language 13: 51–103.



 Ulla-Britt Kotsinas

Clyne, M. 1968. ‘‘Zum Pidgin-Deutsch der Gastarbeiter’’. Zeitschrift für Mundartforschung 35: 135–139. Dittmar, N. 1981. ‘‘On the verbal organization of L2 tense marking in an elicited translation task by Spanish immigrants in Germany’’. Studies in Second Language Acquisition 3: 136–163. Dundes, A., J. Leach, and B. Öskök. 1986. ‘‘The strategy of Turkish boys’ verbal dueling rhymes’’. In Directions in Sociolinguistics, J. Gumperz. and D. Hymes (eds.), 130–160. Oxford: Blackwell. Edwards, V. 1986. Language in a Black Community. Clevedon: Multilingual Matters. Heidelberger Forschungsprojekt 1977. Die ungesteuerte Erlernung des Deutschen durch spanische und italienische Arbeiter. Osnabrück: Osnabrücker Beiträge zur Sprachtheorie. Hewitt, R. 1986. White Talk Black Talk. Interracial friendship and communication amongst adolescents. Cambridge: Cambridge University Press. Holm, J. 1988. Pidgins and creoles, vol. i. Cambridge: Cambridge University Press. Holm, J. 1989. Pidgins and creoles, vol. ii. Cambridge: Cambridge University Press. Hopper, P. and E. Traugott. 1993. Grammaticalization. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press. Klein, W. and N. Dittmar. 1979. Developing Grammars. Berlin: Springer. Kotsinas, U.-B. 1982. Svenska svårt. Några invandrares svenska talspråk. [Swedish difficult. Some immigrants’ spoken Swedish]. (MINS 10). Doctoral dissertation, Stockholm University. Kotsinas, U.-B. 1983. ‘‘Repetition in immigrant Swedish’’. Indian Journal of Applied Linguistics 9: 69–88. Kotsinas, U.-B. 1984. ‘‘Ask maybe ten hours. Semantic over-extension and lexical overuse’’. Scandinavian Working Papers on Bilingualism 2: 23–42. Stockholms Universitet, Institute of Linguistics, Department of Research on Bilingualism. Kotsinas, U.-B. 1985. Invandrare talar svenska. [Immigrants speak Swedish]. Lund: Liber. Kotsinas, U.-B. 1988. ‘‘Immigrant childrens’ Swedish — a new variety?’’. Journal of Multilingual and Multicultural Development, 129–140. (Special Issue.) Kotsinas, U.-B. 1989a. ‘‘Come, stay, finish. On the development of aspect markers in interlanguage and pidgin/creole languages’’. In Proceedings of the Second Scandinavian Symposium on Aspectology, L.–G. Larsson (ed.), 33–48. (Studia Uralica et Altaica, Acta Universitatis Uppsaliensis, 19.) Uppsala: Almqvist and Wiksell. Kotsinas, U.-B. 1989b. ‘‘Errors in the language of immigrant and Swedish pre-school children’’. In Language Learning and Learner Language, B. Hammarberg (ed.), 70–82. Scandinavian Working Papers on Bilingualism, 8). Stockholm: University of Stockholm, Institute of Linguistics, Department of Research on Bilingualism. Kotsinas, U.-B. 1992. ‘‘Immigrant adolescent’s Swedish in multicultural areas’’. In Ethnicity in youth culture, C. Palmgren, K. Löfgren and G. Bolin (eds.), 43–62. Stockholm: Youth Culture at Stockholm University. Kotsinas, U.-B. 1995. ‘‘Aspect marking and grammaticalization in Russenorsk compared with Immigrant Swedish’’. In Arctic Pidgins: Northern Perspectives on Language Contact, Ingvild Broch and Ernst Håkon Jahr (eds.), 123–154. Berlin: Mouton de Gruyter. Labov, W. 1972. Language in the inner city. Philadelphia: Philadelphia University Press.

Pidginization, creolization and creoloids

Le Page, R. and A. Tabouret-Keller. 1985. Acts of identity. Cambridge: Cambridge University Press. Meisel, J. 1975. ‘‘Ausländerdeutsch und Deutsch ausländischer Arbeiter’’. Zeitschrift für Literaturwissenschaft und Linguistik, 18: 9–53. (Sprache ausländischer Arbeiter.) Meisel, J. 1977a. ‘‘Linguistic simplification: A study of immigrant workers’ speech and foreigner talk’’. In The Notions of Simplification, Interlanguages, and Pidgins and Their Relation to Second Language Pedagogy, P. Corder and E. Roulet (eds.), 88–98. Genève: Droz. Meisel, J. 1977b. ‘‘The language of foreign workers in Germany’’. In Deutsch im Kontakt mit anderen Sprachen, C. Molony, H. Zobl and W. Stölting (eds.), 184–212. Kornberg/Ts: Scriptor. Mühlhäusler, P. 1986. Pidgin and Creole Linguistics. Oxford: Blackwell. Penfield, J. and J. Ornstein-Galicia. 1985. Chicano English: An ethnic contact dialect. Amsterdam and Philadelphia: John Benjamins. Rampton, B. 1991. ‘‘Interracial Panjabi in a British adolescent peer group’’. Language in Society, 20: 391–422. Sadler, W. 1973. Untangled New Guinea Pidgin. Madang: Kristen Pres. Sankoff, G. 1980. ‘‘Variation, pidgins and creoles’’. In Theoretical Orientations in Creole Studies, Albert Valdman and Arnold Highfield (eds.), 139–164. New York: Academic Press. Schumann, J. H. 1978. The Pidginization Process. Rowley, MA: Newbury House. Tannen, D. 1989. Talking Voices. (Studies in Interactional Sociolinguistics, 6.) Cambridge: Cambridge University Press. Traugott, E. 1978. ‘‘On the expression of spatio-temporal relations in language’’. In Universals of human language, vol. iii, Word Structure, Joseph Greenberg (ed.), 370– 400. Stanford: Stanford University Press.



Chapter 6

The origin of creole languages The perspective of second language learning* Pieter Muysken

.

Introduction

The relation between the study of creoles and of second language acquisition has been a complex one; on the one hand, it is probably an open door to say that there should be more contact between these fields, but on the other hand there has long been a bi-directional symbiotic exchange. There are many cross-currents, some tending in one direction, some in the other. Creoles have contributed a number of concepts and ideas to the study of second language acquisition. Conversely, results from the study of second language acquisition have contributed and can contribute further to creole studies. In fact, one of the earliest researchers working on creoles, Coelho (1880–1886), derived many features of West-African Portuguese creoles from universals of second language acquisition. This paper , dealing with this symbiotic relationship, was originally written in response to Ulla-Brit Kotsinas (this volume). While Kotsinas investigates second language varieties with respect to their pidgin/creole features, I will take the opposite point of view here: to what extent can the grammatical properties of creole languages plausibly be atttributed to what we know of the processes of L2 acquisition of their lexifier languages. I will take Negerhollands, the now extinct Dutch-lexifier creole language of the Virgin Islands (notably St Thomas and St John) as an example, and contrast features of this language with what we know of the acquisition of Dutch as a second language. In exploring the relationship between second language acquisition and creole genesis I am not advocating that the process of creole genesis was the result of attempts by slaves to unconditionally acquire the language of their masters. I am sympathetic to the position taken by Baker, e.g. in this thoughtful (1996) review of Chaudenson (1992), that the colonial languages are not

 Pieter Muysken

really the target that slaves are aiming at. This does not mean, however, that processes of second language learning were not important, but only that the construction of the new grammar was not necessarily target-directed. Section 2 is devoted to the connections between the two disciplines, and in section 3 I list some of the dimensions relevant to pidgin genesis and stabilization. In section 4 I briefly discuss some relevant historical details of the genesis of Negerhollands, and in section 5 some of its properties. Section 6 is devoted to the issue what light the variation in Negerhollands throws upon the L2 acquisition process, and section 7 summarizes the evidence on the main question: which features of Negerhollands can be seen as the result of the acquisition of Dutch as a second language and which cannot.

. Connections between second-language research and creole studies There are a number of areas of contact between second language research and creole studies. These include: A. Basilect. In creole studies Bailey (1973) has proposed the terms ‘‘basilect’’, ‘‘mesolect’’, and ‘‘acrolect’’ to mark the distance of creole varieties from the basic creole and the colonial European variety. The metaphor involved in these terms has been interpreted as representing different stages in the acquisition process, where the early stages have been termed ‘‘basilect’’ etc. B. Interlanguage. Creoles may offer support for the notion of interlanguage as a structured system with its own rules (Selinker 1972). If they are systematic in themselves, and the direct reflex of second language learning, the latter may also be highly regular (Wekker 1995). C. Form/meaning. The fields share a heigthened awareness of the need to study the meaning and use of forms (e.g. tense and aspect markers) carefully and not to focus exclusively on their morpho-syntax. Thus we find studies of the meaning and use of tense/aspect markers such as been (bin) both in second language learning and in creoles (Bickerton and Odo 1972; Huebner 1980). D. Social factors. The fields also share an increased sensitivity to the role of social factors in determining progress in L2 development, e.g. in the use of the term Pidgin to refer to migrant workers’ interlanguages in the European context (e.g. in the Heidelberg project as reported on in Klein and Dittmar 1979) and in John Schumann’s Pidginization Hypothesis (1978).

The origin of creole languages 

E. Similar outcomes. Creole forms often resemble the actual linguistic outcome of incomplete L2 acquisition in different stages in the L2 development of e.g. English or French. The pre-verbal position of invariant negation in Creoles is matched with a similar position in an early stage in the L2 acquisition of English (Andersen 1983). F. Transfer. The study of L2 learning provides evidence on the potential role of interference and transfer, relevant for the substrate issue in creole studies. This is relevant for possible scenarios accounting for the preservation of African language patterns in the Caribbean creoles (Alleyne 1980). G. Grammaticalization. The study of L2 learning provides evidence on potential paths or chains of grammaticalization. Thus the fact that in German migrant interlanguage a form such as vieleicht ‘perhaps’ has come to be used as a modal marker could be interpreted as evidence for its susceptibility to a process of grammaticalization in general (Klein and Dittmar 1979; Kotsinas, 1995). In spite of these many areas of shared conceptual frameworks, terminology, and research questions between the study of creoles and of second language acquisition, many questions remain unanswered, and some of these are illuminated in the chapter by Ulla-Britt Kotsinas: H. Conventionalization. How does ‘‘conventionalization’’ of interlanguage forms proceed? What is the relation between the individual and the group in this respect? I. Hybridization. Do we need ‘‘tertiary hybridization’’ (which necessarily involves speakers of at least three languages) as claimed by Whinnom (1971), or are two groups enough? J. Imitation. Is there a process of ‘‘imitation of learner varieties’’ by speakers of the dominant language, as claimed by Bloomfield (1933)? K. Speed of grammaticalization. How rapidly can grammaticalization proceed? The chapter by Kotsinas suggests quite rapidly, particularly in the case of discourse particles. The same is claimed in Bruyn (1995) for creoles. However, in historical linguistics a much slower process is often assumed. L. The amount of input needed. How much input is needed for L2 acquisition to be succesful? M. Ethnolects. When we try to establish parallels between the results of L2 learning processes and creoles, often we think of terms of ethnolects of the

 Pieter Muysken

colonial languages emerging on the plantations. Can we use the metaphor of ‘‘ethnolects’’ when discussing creoles in statu nascendi? The transition from first generation to second generation patterns, as in Kotsinas’ urban youth culture [Rinkeby Swedish] is not clear yet.

. Scenarios for pidgin genesis I am committed to some version of the Uniformitarian Hypothesis for language change, but this does not mean we can simply assume that the social conditions for creole genesis and second language acquisition are comparable. We will never know to what extent present day situations resemble or not the context of enslavement. A number of factors have been proposed to determine whether or not an incipient pidgin will emerge from a second language learning situation, for example, in migration contexts: – – – –

–

the wish of migrants to return to their homeland may hinder adequate acquisition of the new language a limited number of target language speakers and limited acquisition opportunities the use of Foreigner Talk by target language speakers limited access to more highly valued social roles linked to the new language, due to a caste system, religious differences, or racial, social, legal discrimination absence of educational opportunities

Features of resulting pidgins may include simplification, impoverished target language morpho-syntax and lexicon, strong native language-influenced pronunciation, and a limited stylistic range per speaker. However, before a pidgin can be creolized, it must need to be accepted as a separate norm. The crystallization of the pidgin and the stability of a pidginized variety may be furthered by: – – – – –

a socially homogeneous community speaking the pidginized variety reciprocal native language/target language community communication in the pidginized variety use of the pidginized variety in communication between different groups of non-target language speakers the variety is learned by newcomers as a system the pidginized variety is mutually intelligible with the target language.

The origin of creole languages

Notice that simply pointing out regularities in the L2 system does not give sufficient evidence for its pidgin and system status. Similar syntactic/semantic constraints on variability may hold for individual learner systems as for conventionalized community languages.

. Brief summary of the early history of the Virgin Islands In 1493 Columbus gave the Virgin Islands their present-day name and met some Amerindians on St. Croix. The St. Croix Tainos were subsequently decimated by genocide and epidemics. From 1600 onward the islands were populated by Europeans from different descent and the slaves imported from Africa. In 1653 the Danish West-Indian Company was founded, and henceforth the (late) entry of Denmark into the European colonizing efforts was a fact. This resulted in 1665 in a first attempt by the Danish to settle on the island of St. Thomas. In 1671 the Danish West-Indian Company obtained a monopoly over St. Thomas, and in 1672 the Danish colonization proper of the island began, with 113 inhabitants. Shortly afterwards a group of Dutch planters who had fled from St. Eustatius (to escape the English) settled on St. Thomas and, according to Goodman (1985), they possibly brought a Dutch pidgin or creole with them, spoken by their slaves. Sabino (1990) argues that the number of slaves brought along was probably very limited, however. In 1688 there were 422 slaves in St. Thomas and 317 whites, who consisted of: 66 32 20 18 3 3 1 1

Dutch households English households Danish households French households German households Swedish households Holstein households Portuguese households

This array confirms that the dominant input probably was Dutch, with Danish and English as secondary components. And indeed, the lexicon of Negerhollands is mostly Dutch, with some English and Danish elements. In 1692 one fifth of the slave population consisted of children born in St. Thomas. Between 1672 and 1739 the majority of slaves imported may have



 Pieter Muysken

consisted of Twi-speaking Akan. However, Akan has not left any specific linguistic traces in the creole, as far as we know. By 1725 the slave population had increased to 4490, and the slave society was fully constituted. Most of what we know about Negerhollands comes from missionaries. One of these, the German missionary Auerbach (1774) writes in the creole, which by then had become very important in the colony: Die hab well twee drie onder die swart Volk, die sender a leer voor verstaan beetje van die hollandisch Taal, as sender woon na die Stadt, en hoor die ider Dag van die Blanko, maar die Plantey-Volk no kan vor verstaan die soo. Doch, die no sal maak een Verhinder, as die lieve Broeer will skriev eenmaal na sender, maski die ben Hollandisch of na die Hoogduytsch, soo die sal maak sender moeschi bli, en ons sal lees die Brief voor sender na Creol. Na St. Croix die hab meer van die Negers, die sender kan verstaan English, as na St. Thomas en St. Jan, maar doch sender English Praat ka mingel ook altoeveel met die Creol- en Guinee-taal. . . Da Neger-English die ben. ‘‘There are two or three among the black people who have learned to understand a bit of the Dutch language, as they live in town, and hear it every day from the whites, but the plantation folk cannot understand it. This should not be an impediment if the dear brethren will write to them once, albeit in Dutch or High German, for this will make them very happy, and we will read the letter for them in Creole. On St. Croix there are more blacks that can understand English than in St. Thomas and St. John, but still their English speech is mixed too much with Creole and Guinea language. It’s Negro-English.’’

Before closing off this historical section it is appropriate to consider briefly the applicability to St. Thomas of some demographic models that have been proposed as crucial for determining a creole’s features, and which are based on assumptions about acquisition. These models have come under criticism because they overstress speaker numbers at the expense of more complex social factors of group formation, communicative networks, and stratification. Nonetheless, Baker (1996) continues stressing demographic history as a key to understanding creole development. In his earlier work, Baker (1982) distinguished three important demographic events in the development of a creole: Event 1: the moment at which there are as many slaves as speakers of the colonial language; Event 2: the moment at which there are as many slaves born in the colony as speakers of the colonial language; Event 3: the end of the importation of slaves and contract labourers. For Negerhollands, this would amount to:

The origin of creole languages 

Start: Event 1: Event 2: Event 3:

1672 1680: as many slaves as whites ±1700: as many locally born slaves as whites ±1800: end of slave imports

Baker (1982) thus would predict a quite rapid creolization process, and hence a creole remote from its lexifier. Taking Baker’s work as his point of departure, Bickerton (1984) proposed a Pidginization Index (PI): PI =Y ×

P , R

where Y=number of years between the moment of colonization and event 1; P= slave population at event 1; R=yearly average of imported slaves after event 1 For Negerhollands this index would yield: PI = 7 years × (200 slaves / (4000 slaves imported in 40 years; but there was also considerable mortality, hence > 100 slaves per year). Again, the PI would predict a heavily pidginized creole, rather remote from its lexifier. Recall, however, that by 1692 one fifth of the slave population consisted of children born in St. Thomas. This suggests a rather stable setting, with many opportunities for socialization in the new culture and hence the possibility that an early form of the creole not so distant from the lexifier became the norm. This is on the assumption that early learners had more of an opportunity to learn the creole than later learners.

. Relevant features of Negerhollands Having explored some of the early history of Negerhollands, I will now illustrate some basic features of the language with proverbs from Magens (1770): (1) Pampuen no kan parie Kalbas pumpkin not can give.birth.to calebas ‘A pumpkin cannot give birth to a calebas.’

Example (1) demonstrates the fixed Subject–Negation–Verbal Complex–

 Pieter Muysken

Complement word order of Negerhollands. It contrasts with that of Dutch, where the auxiliary kan occupies the second position, and the main verb voortbrengen occurs at the end of the sentence, preceded by the object (cf. (2)). There is also a difference with respect to the position of the negation marker. Generically used nouns, common in proverbs, do not get an article. Notice also the occurrence of both Spanish or Portuguese (together labeled as Ibero-Romance) elements: parie ‘give birth to’ (< parir), and Dutch ones: kan ‘can’. The form no ‘not’ could be either English or Spanish, but the latter is more probable. Structurally, the crucial features are pre-verbal negation no, and a modal + verb cluster in pre-object position. Negerhollands is quite different from Dutch in this respect, which would have: (2) a.

kan niet een kalebas voortbrengen can not a calabash give.birth.to b. kan de kalebas niet voortbrengen can the calabash not give.birth.to

or

(3) Negerhollands: Neg–Modal–Verb–Object Dutch: Modal–Neg–Object–Verb Modal–Object–Neg–Verb

In Dutch the option of object-negation is presumably due to leftwards scrambling, and limited to definite objects. Leftwards scrambling is a feature of OV languages, which Negerhollands is clearly not. The fact that negation precedes the whole verbal cluster, including the modal, is due to the fact that there is no rule placing the finite verb in a pre-negation agreement position, as in Dutch. (4) Branmier val na Malassie, da sut hem ha vind Ant fall loc molasses, because sweet 3sg pst find ‘He gets what he deserves.’

Sentence (4) contains the all-purpose locative preposition na (probably < Port. na < em a ‘in the (fem)’; cf. also Du naar ‘to’). There is an example of fronting for the purpose of focus or emphasis. In a construction with da, sut ‘sweet’is placed early in the subordinate clause and is understood as emphasized (da sut). The particle ha (< Du. had ‘had’ or a dialectal form of the verb ‘have’) is used to mark tense. Notice that hem, the stressed non-subject form of the pronoun in Dutch, is used for the subject, the direct object and the indirect object in Negerhollands.

The origin of creole languages 

(5)

Mie jammer Ju tee mie kries Ju, tee mie neem Steen veeg mie Hogo. 1sg bewail you till 1sg becry you till 1sg take stone wipe 1sg eye ‘I pity you to the point of crying for you, of wiping my eyes with a stone.’

In sentence (5) it is striking that the first person pronoun expressing the subject here, mie, is derived from Du. mij ‘me’, a non-subject form; it is furthermore also used possessively. The preposition or conjunction tee may well derive from Port. até ‘until’ and we see that the verbs jammer and kries can be used transitively (with a human object in this case), which they cannot in the language from which they are derived, namely Dutch. Many originally Dutch verbs thus have acquired other syntactic properties. A remarkable feature is also the serial verb construction neem . . . veeg . . . ‘take . . . wipe . . .’, in which the object of the verb neem is marked as an instrument. Many researchers connect these constructions with West-African languages. Various authors have tried to argue against the existence of serial verb constructions in Negerhollands. Serial verb constructions do not actually seem to have been used very often in earlier Negerhollands. (6) Hunder wil si Kikkentje alteveel. chicken love 3poss chicken much ‘(S)he loves her/his children.’

In (6) we encounter an example of the invariant possessive pronoun si (Negerhollands has no grammatical gender) (< Du. masc. sg. zijn) and of the (pidgin-like) periphrastic adverb alteveel ‘all too much’ as a degree marker. Hesseling (1933) argues that the use of wil (
Example (7) is given here to illustrate two phonological features: the replacement of the (marked, i.e. difficult vowel) /ø:/ by /e/ in deer and the appearance of an extra vowel in hogo, which yields an unmarked CVCV-pattern, which is easy to pronounce. Such vowels are termed ‘‘epithetic’’. (Incidentally, the initial h in hogo, absent in standard Du oog, may well be due to Zealandic or Flemish influence.)

 Pieter Muysken

Consider now another eighteenth-century proverb: (8) Gras le gruj na Dootman sie Door. Gras hab grow loc dead.man 3pposs door ‘Nobody takes care of widows and orphans.’

Here we notice a typical creole possessive construction with both the possessor dootman preposed to the noun and the resumptive third person possessive sie. While in an earlier example we had deer ‘door’ from Du. deur, here we have a form door possibly derived from English, or alternatively from Low German. Notice also that the compound dootman contains an invariant adjective doot, where Dutch would have an inflected adjective dode. The same thing occurs in (9), where klein remains uninflected and Dutch would have kleine. (9) Als die Vier ka yt, klein Kint le jump na die Hassesje. When the fire prf out little child hab jump loc the ashes ‘They do with him/her as they please.’

Example (9) shows that the particle yt ‘out’ can be used as a verb, unlike Dutch. This type of reinterpretation from a particle to a verb is typical of the relation between a creole language and its lexifier. It is furthermore preceded by the perfective aspect marker ka (< Port. acabar ‘finish’), an element which occurs in many creole languages in one form or other. The article (when present) is invariant die ‘that’. It is not unusual in creole languages, and for that matter, in other language families, that the article is derived from the demonstrative pronoun. Notice that in the second clause there is no inversion of subject and verb, as in Dutch (where we would have had springt het kleine kind ‘jumps the little child’); in this respect the Subject-Verb-Complement order of Negerhollands is very strict. In both of the previous examples there is a particle marking duration or habitual, le, possibly from Dutch leggen ‘lay’. Some of the particles used to mark Tense, Mood and Aspect in Negerhollands derive from verbs. As a result there is a small class of frequently used homophonous verbs and particles. The element lo (assumed to derive from Dutch loop ‘walk’) belongs to this class, and it is not always clear in what capacity it is used. (10) Twee slem no kan kook Boontje na een Pot two smart not can cook bean loc one pot ‘No reason to get in each other’s way.’

The origin of creole languages 

A final proverb illustrates the optionality of plural marking, both after a numeral (twee slem) and with a plural object that can be understood as a mass (boontje). With specific non-quantified plural nouns there is the possibility of affixing sender (3rd plural pronoun) or a related form to the noun phrase. In the next example, from the same period but not a proverb, we see the possibility of combining pre-verbal tense/aspect particles (a ka): (11) vordaarom mi a ka kom voor doop met Water. therefore 1sg pst prf come for baptize with water ‘Therefore I have come to baptize with water.’

(±1780)

Notice also the use of voor as a purposive complementizer, very general in the creole though not in the varieties of Dutch that provided the L2 input to Negerhollands. Finally, note the absence in the following two examples of the Dutch proclitic oblique pronoun er: (12) a.

wagut ju wil du mit di what you want do with it (cf. Du. ‘wat wil je ermee doen?’) b. am no weet een gut fan di he not know a thing of it (cf. Du. ‘hij weet er niets van’)

Where Dutch would have a form with er, either directly to the left of the preposition or somewhere earlier in the clause, Negerhollands has the deictic element di (derived from the Dutch demonstrative die).

. Variation in Negerhollands There is considerable variation present in the different varieties of Negerhollands, some of which can be interpreted as the frozen reflex of different stages in the L2 learning process. . Passive An important locus of variation is the passive. Magens (1770) suggests that the passive in Negerhollands was used only rarely. Passives marking an action in progress were avoided. Nonetheless a passive could occur, and then mostly in

 Pieter Muysken

white speech. The passive auxiliary verbs were wort, bin, and wees. From the examples it is clear that the passive, when used at all (and one must wonder to what extent this was simply to translate a Dutch passive) was expressed by a passive auxiliary verb and a Dutch perfect participle: (13) a.

Mie bin vervolgt. ‘I have been persecuted.’ b. Mie wort vervolgt. ‘I am persecuted.’

There was also a passive in the general creole of both whites and blacks, albeit only in the perfective form and without overt passive morphology on the verb. The perfective aspect marker ka regularly occurs in the creole instead of the passive auxiliary verb bin. (14)

Die Hus ka bou. ‘The house has been built.’

. Prepositions A rich area for variation studies concerns the use of complex prepositions, while Dutch postpositions are lacking. In one text we find the following variants: (15) a.

boven op die Berg above on det mountain b. nabovo die Berg loc=above det mountain c. na bovo op die Berg loc above on det mountain d. naboven op die Berg loc=above on det mountain e. boven na die Hemel above loc det heaven

There is variation: in the appearance of na; in the final vowel of the locative adverb, either a schwa (boven) or an assimilated vowel (bobo); and in the subsequent use of another locative element. The interpretation of Dutch locative prepositions such as boven as nominal in nature may be attributed to substrate influence (Bruyn 1995).

The origin of creole languages 

. Particles Particles may occur separately from the verb in Verb–Particle constructions, as in (16), or together with the verb, as in (17): (16) a.

da mi sa due mi Parik en Deegen an emp 1sg fu do 1sg wig and sword on ‘I will put my wig and sword on.’ b. ons ka kom for bid hem an 1pl prf come for pray sg on ‘We have come to adore him.’

(17) a.

loop dan due an ju klaer walk then do on 2sg clothes ‘go and put your clothes on’ b. en ha bid an die Kind and pa adore det child ‘and adored the child’

So far, no cases of V + particle + pronoun have been encountered. It is not clear how this variable is embedded socially. We may hypothesize that combinations such as due an and bid an in (17) are closer to the basilect, but this needs further investigation. The order in (16) reflects the Dutch order.

.

Conclusions

A number of features of Negerhollands may well be explicable as deriving from the acquisition of Dutch as a second language. There is no general synthesis of the acquisition process so far; we must rely on the case studies of Extra (1978), van Helvert (1985), Lalleman (1986), Vermeer (1986), Broeder (1991), and Coenen and Klein (1992): (18) a. b. c. d. e. f.

rigid SVO order the absence of postpositions pre-verbal negation the absence of tense and person marking on verbs periphrastic possessives of the type (Mary her/his book) the absence of Dutch er forms in Negerhollands

 Pieter Muysken

g. h. i. j.

the use of tense/mood/aspect auxiliaries the absence of inflection on the adjective the loss of passive morphology in basilectal Negerhollands contiguous verb + particle constructions

While there has been no systematic study of this, anecdotal observations confirm that the loss in Negerhollands of the Dutch marked vowel [ø:] is also explicable in these terms. The same holds for the presence of epithetic or paragogical vowels, with the proviso that this depends on the first language of the speakers involved. In the above survey, lexical features of Negerhollands that may be due to second language acquisition processes have not been discussed. Intriguing is the alternation between kom (presumably more creole) and word (more Dutch) for ‘become’: (19) si Wief a kom swanger 3poss wife pa come pregnant ‘his wife became pregnant’ (20) die Saak a ka word bekent det affair pa prf become known ‘the affair had become known’

The use of the verb ‘‘come’’ for CHANGE is noted by Kotsinas too as a pidgin feature. The lexical origin of some of the grammaticalized Negerhollands preverbal tense/aspect markers is not fully clear. The form lo ‘future’ may derive from Portuguese logo ‘soon’, while lo ‘durative’ and le ‘durative’ may be derived from Dutch loop ‘walk’ and leg ‘lie, lay’, respectively. Not as easily explicable as due to L2 learning are some other characteristics: (a) In Negerhollands we find invariant pronouns based on strong oblique forms in the target. Broeder (1991) shows that initially in L2 Dutch the pronouns are indeed invariant, but they are based on the nominative forms of the pronoun rather than an oblique form. (b) The selection, and semantic features, of the tense/mood/ aspect particles and of auxiliary verbs in creoles have been claimed to be very specific to the creole languages (Bickerton, 1980), although Kotsinas shows that many of the meanings expressed by them, such as perfective, durative, and irrealis, are also semantic features expressed in Immigrant Swedish interlanguages and similar systems.

The origin of creole languages

(c) As such, the use of serial verbs is not characteristic of interlanguage systems in general, even though the use of analytic forms to express complex meanings (one of the uses of serial verb constructions) may well be. (d) The use of na as a general purpose locative and oblique preposition in Negerhollands may well have an African origin, as is also the case for the interpretation of Dutch locative prepositions such as boven ‘above’ as nominal in nature. At the same time, Kotsinas remarks on the overextension of locative prepositions to cover a wide range of oblique meanings in Swedish interlanguage. (e) Finally, the use of forms like sender ‘they’ as a nominal plural marker is not particularly characteristic either of interlanguage systems. Neither has a specific African source been identified. Some of the non-interlanguage features reflect West-African patterns such as serial verbs and the use of nominal locative forms, others may derive from general principles of language construction. It would then be a mystery, however, why these do not show up in L2 Dutch interlanguage.

Note * Some of the material in this contribution was presented orally at the workshop on Language Contact and Creole Genesis, held at the University of Amsterdam in June 1995. The analysis of the Negerhollands data was presented in a lecture at the School of Education, Harvard University, in November 1995, with the support of a Language Learning lectureship. The Negerhollands data were collected and put into a computer data-base with the support of the Netherlands Research Foundation (NWO) by Cefas van Rossem and Hein van der Voort (cf. van Rossem and van der Voort (1996)).

References Alleyne, Mervyn C. 1980. Comparative Afro-American. Ann Arbor, MI: Karoma. Andersen, Roger W. (ed.). 1983. Pidginization and Creolization as Language Acquisition. Rowley, MA: Newbury House. Bailey, Charles James N. 1973. Variation and Linguistic Theory. Arlington, VA: Center of Applied Linguistics. Baker, Philip and Chris Corne. 1982. Isle de France Creole: Origins and affinities. Ann Arbor, MI: Karoma. Baker, Philip. 1996. ‘‘Pidginization, creolization, and français approximatif ’’. Review article of Chaudenson (1992). Journal of Pidgin and Creole Languages 11.1: 95–120.



 Pieter Muysken

Bickerton, Derek. 1981. Roots of Language. Ann Arbor, MI: Karoma. Bickerton, Derek. 1984. ‘‘The language bioprogram hypothesis’’. The Brain and Behavioral Sciences 7: 123–221. Bickerton, Derek and Carol Odo. 1976. ‘‘General phonology and pidgin syntax. Change and variation in Hawaiian English’’. (Vol. 1 (mimeo) U. of Hawaii.) Bloomfield, Leonard. 1933. Language. London: Allen and Unwin. Broeder, Peter. 1991. Talking About People. A multiple case study on adult language acquisition. Amsterdam: Swets & Zeitlinger. Bruyn, Adrienne. 1995. Grammaticalization in Creoles. The noun phrase in Sranan. Amsterdam: IFOTT. Doctoral dissertation, University of Amsterdam. Bruyn, Adrienne, Pieter Muysken and Maaike Verrips. 1998. ‘‘Double object constructions in the creole languages: Development and acquisition’’. In Creole Acquisition and Diachrony, Michel DeGraff (ed.). Cambridge, MA: MIT Press. Chaudenson, Robert. 1992. Des îles, des hommes, des langues. Paris: L’Harmattan. Coelho, Adolpho. 1880–1886. ‘‘Os dialectos românicos ou neolatinos na África, Ásia e América’’. Boletim da Sociedade de Geografia de Lisboa 2, 129–96; 3, 451–78; 6, 705–55. Coenen, Josée and Wolfgang Klein (1992) ‘‘The acquisition of Dutch’’. In Utterance Structure: Developing grammars again, Wolfgang Klein and Clive Perdue (eds.), 189– 224. Amsterdam and Philadelphia: John Benjamins. Extra, Guus. 1978. Eerste- en tweede-taalverwerving: De ontwikkeling van morfologische vaardigheden. Muiderberg: Coutinho. Graves, Anne. 1977. The present state of the Dutch creole of the Virgin Islands. Dissertation, University of Michigan. Helvert, Korrie van. 1985. Nederlands van en tegen Turkse kinderen. Doctoral dissertation, Nijmegen University. Hesseling, Dirk Christiaan. 1905. Het Negerhollands der Deense Antillen. Bijdrage tot de Geschiedenis der Nederlandse Taal in Amerika. Leiden: Sijthoff. Hinskens, Frans and Cefas van Rossem. 1995. ‘‘The Negerhollands word ‘sender’ in eighteenth century manuscripts’’. In: The Early Stages of Creolization, Jacques Arends (ed.). Amsterdam and Philadelphia: John Benjamins. Huebner, Thom. 1982. From Topic to Subject dominance in the interlanguage of a Hmong speaker. Doctoral dissertation, University of Hawaii. Josselin de Jong, J. P. B. de. 1924. ‘‘Het Negerhollandsch van St. Thomas en St. Jan’’. MKNAW, Deel 57, Serie A, no. 3, 55–71. Josselin de Jong, J. P. B. de. 1926. Het Huidige Negerhollandsch (Teksten en Woordenlijst). VKAW, Nieuwe Reeks, Deel 26, no. 1. Klein, Wolfgang and Norbert Dittmar. 1979. Developing Grammars. Berlin: Springer. Kotsinas, Ulla-Britt. 1995. ‘‘Aspect marking and grammaticalization in Russenorsk compared with Immigrant Swedish’’. In Arctic Pidgins: Northern Perspectives on Language Contact, Ingvild Broch and Ernst Håkon Jahr (eds.), 123–154. Berlin: Mouton de Gruyter. Lalleman, Josine. 1986. Dutch Language Proficiency of Turkish Children Born in the Netherlands. Dordrecht: Foris.

The origin of creole languages 

Muysken, Pieter and Hein van der Voort. 1991. ‘‘The binding theory and creolization: Evidence from eighteenth century Negerhollands reflexives’’. In Development and Structures of Creole Languages, Essays in honor of Derek Bickerton, Francis Byrne and Thom Huebner (eds.), 145–158. Amsterdam and Philadelphia: John Benjamins. Nelson, Frank. 1936. Fieldnotes on Negerhollands. Published in Van Rossem and van der Voort (1996). Rossem, Cefas van and Hein van der Voort. 1996. Die Creol Tael. 250 years of Negerhollands texts. With contributions by Frank Nelson, Robin Sabino and Gilbert Sprauve and the assistance of Hans den Besten, Pieter Muysken, and Peter Stein. Amsterdam: Amsterdam University Press. Sabino, Robin. 1988. ‘‘The copula in vernacular Negerhollands’’. Journal of Pidgin and Creole Languages 3, 199–212. Sabino, Robin. 1990. Towards a phonology of Negerhollands: An analysis of phonological variation. Dissertation, University of Pennsylvania. Schumann, John. 1978. The Pidginization Process. Rowley, MA: Newbury House. Selinker, Larry. 1972. ‘‘Interlanguage’’. International Review of Applied Linguistics 10, 209–231. Sprauve, Gilbert A. 1990. ‘‘Dutch Creole–English Creole distancing: Historical and contemporary data considered’’. International Journal of the Sociology of Language 85, 41–50. Stein, Peter. 1985. ‘‘Die Anfänge der Verschriftung einer Kreolsprache: Das Negerhollands im 18. Jahrhundert’’. In Entstehung von Sprachen und Völkern, Glotto- und ethnogenetische Aspekte europäischer Sprachen, Akten des 6. Symposions über Sprachkontakt in Europa, Mannheim 1984, P. Sture Ureland (ed.), 437–57. Tübingen: Max Niemeyer Verlag. Stein, Peter. 1986. ‘‘Les premiers créolistes: Les Frères Moraves à St. Thomas au XVIIIe siècle’’. Amsterdam Creole Studies 9, 3–18. Vermeer, Anne. 1986. Tempo en struktuur van tweede-taalverwerving bij Turkse en Marokkaanse kinderen. Doctoral dissertation, Tilburg University. Voort, Hein van der, and Pieter Muysken. 1995. ‘‘Eighteenth century Negerhollands reflexives revisited’’. In Creolization: The Early Stages, papers presented at the workshop held in Amsterdam, 4–5 September 1992, Jacques Arends (ed.), 25–42. Amsterdam and Philadelphia: John Benjamins. Wekker, Herman. 1995. Creolization and Language Acquisition. Berlin: Mouton de Gruyter. Whinnom, Keith. 1981. ‘‘Linguistic hybridization and the ‘special case’ of pidgins and creoles’’. In Pidginization and creolization of languages, D. Hymes (ed.), 91–115. Cambridge University Press.

Chapter 7

Koine formation and creole genesis* Jeff Siegel

.

Introduction

The term ‘‘koine’’ has a confoundingly wide variety of interpretations in sociolinguistics (as pointed out in Siegel 1985, 1993a), but among scholars working in language contact, a definition such as the following would be widely accepted: A koine is a stabilized contact variety which results from the mixing and subsequent levelling of features of varieties which are similar enough to be mutually intelligible, such as regional or social dialects. This occurs in the context of increased interaction or integration among speakers of these varieties.1

Descriptions of koines as defined here include the original Greek Koine¯ (Thomson 1960; Bubenik 1993), the ancestor of modern Arabic (Ferguson 1959), Australian English (Trudgill 1986), Dhuwaya [Australian Aboriginal] (Amery 1993), and varieties of Overseas Hindi (Mohan 1978; Gambhir 1981, 1988; Damsteegt 1988; Mesthrie 1991, 1993; Siegel 1987, 1988b). In this chapter, I first illustrate the processes involved in koine formation (or koineization, as it is sometimes called), using Fiji Hindi as an example. Then I discuss the mechanisms of levelling in more detail. Next, I defend the concept of ‘‘koine’’ and distinguish it from other language contact varieties. Finally, I show how the processes involved in koine formation may be relevant to pidgin/creole genesis.

. The processes of koine formation Fiji Hindi is an example of an ‘‘immigrant koine’’ — one which develops when speakers of different dialects move to another location and form a new community. This is opposed to a ‘‘regional koine’’ which remains in the area where

 Jeff Siegel

the contributing dialects are spoken. Using Chaudenson’s (1977) terms, we can say an immigrant koine is exogenous and a regional koine is endogenous. From 1879 to 1916, over 45,000 people from North India were brought to Fiji as indentured labourers. The vast majority spoke regional dialects of Hindi, mainly various subdialects of Bhojpuri (32.5 per cent), subdialects of Eastern Hindi — especially Avadhi (30.0 per cent), and subdialects of Western Hindi (7.6 per cent). Most of the immigrants were also familiar with Hindustani, the spoken lingua franca of North India, especially with Bazaar Hindustani of Calcutta. Evidence clearly shows that the labourers continued speaking their different dialects among themselves in Fiji (Siegel 1987: 40–46). Thus, the regional dialects and varieties of Hindustani provided a pool of linguistic variants for the developing linguistic community in Fiji, and initially this pool was used for everyday interaction. This is referred to as mixing. However, only some of the variants from the pool became features of the new contact variety; most of the forms in the original mixture are no longer used. This process of reduction or attrition of variants is what is usually referred to as levelling. Also, compared to the individual varieties in contact, fewer and more regular forms remain. This reduced complexity has led to another notion often associated with koineization: simplification (Trudgill 1986). The immigrant koine which developed, Fiji Hindi (FH), is therefore a composite dialect, showing a mixture of features of the main varieties that were in contact. This can be clearly seen in the verb morphology shown in Table 1, which is derived from Bhojpuri (B), Avadhi (A), and (Bazaar) Hindustani (H). Table 1. Fiji Hindi verbal suffixes and their origins 1sg, pl 2sg, pl 3sg 3pl

Imperfect

Perfect (trans)

Definite future

-ta¯ -ta¯ -e -e

-a¯ -a¯ -is -in

-ega¯ -ega¯ -ı¯ -ı¯

(A, H) (A, H) (B) (B)

(A, H) (H) (A) (A)

(H) (H) (A, B) (A, B)

There is also evidence of mixing in FH in some ‘‘intermediate’’ or ‘‘interdialect’’ forms (Trudgill 1986: 62), such as tuma¯r ‘your (singular)’ — cf. Bhojpuri and Avadhi toha¯r and Bazaar Hindustani tum(h)a¯ra¯. An example of levelling of various forms once in use can be seen in Table 2. The table also illustrates the reduced complexity of FH in comparison to the two most common regional dialects in contact, in terms of fewer

Koine formation and creole genesis 

Table 2. Hindi dialects and Fiji Hindi perfective suffixes (transitive) Bhojpuri

Avadhi

Fiji Hindi

1sg (masc) (fem)

-lõ, -lı¯

-eu¯, -a¯ -iu¯,

1pl (masc) (fem)

-lı¯, -lı -lyu

-a¯, -an, -en -ı

2sg (masc) (fem)

-le, -lâ, -las -lı¯, -lis

-es, -is, -au -is(i)

2pl (masc) (fem)

-lâ(h) -lu¯

-eu, -u¯, eo -ı

3sg (masc) (fem)

-las, -le(s) -lı¯

-is, -es, -ai -ı¯, -isi

-is

3pl (masc) (fem)

-lan(i) -lin

-in, -en, aı˜ -ı , -ini

-in

-a¯

variants, fewer categories (e.g., the elimination of gender) and the loss of inflections (some for person and number). Reduced complexity can also be seen in the more regular and semantically transparent pronoun system of FH, compared to that of Avadhi, for example, as shown in Table 3. (The FH plural marker is derived from the noun log meaning ‘people’.) Another process involved in levelling is called ‘‘reallocation’’ (Trudgill 1986: 110). Variants which are not eliminated in levelling may be reallocated to new functions, such as stylistic or social markers. For example, in Fiji Hindi, certain forms are considered rustic and are used by ‘‘country bumpkin’’ Table 3. Avadhi and Fiji Hindi nominative pronouns 1sg 1pl 2sg (intimate) 2sg (familiar) 2pl (familiar) 2sg (formal) 2pl (formal) 3sg (proximate) 3pl (proximate) 3pl (remote) 3pl (remote)

Avadhi

Fiji Hindi

maı˜ ham taı˜, tu¯ tum, tu tum, tu a¯p(u) a¯p(u) ı¯ in u¯ un

ham ham log – tum tum log a¯p a¯p log ı¯ ı¯ log u¯ u¯ log

 Jeff Siegel

characters in dramas or comedy routines. These include many previously common items from Bhojpuri dialects, such as the third person possessive okar (instead of the usual uske) and the emphatic suffix -wa¯, as well as some alternative forms of borrowings from English, such as moaka¯ ‘car’ versus moar The following example is from a humorous newspaper column by Tirlok Tivari (Shanti Dut, 13 November 1986, p. 4): (1) okar beauwa¯, jinna¯, kharı¯d-is pura¯na¯ moaka¯ his son:emph Jhinna buy-perf old car ‘His son, Jhinna, bought an old car.’

In socially unmarked Fiji Hindi, this would be: (2) uske bea¯ jinna¯ kharı¯d-is pura¯na¯ moar his son Jhinna buy-perf old car

. Mechanisms of levelling Varieties of Hindi such as Avadhi and Bhojpuri have been in contact for centuries without the process of levelling occurring. But in Fiji, the social circumstances were right for koineization to take place: ‘‘the need for unification among speakers of different dialects in a new environment.’’ (Domingue 1981: 150). Using terms from LePage and Tabouret-Keller (1985), Trudgill (1986) describes how initial contact in a new environment may lead to a ‘‘diffuse’’ linguistic situation with a great deal of mixture and variability. But a diffuse situation may later become ‘‘focussed’’ — with less variation, generally accepted norms and the emergence of a new recognizable variety. This may happen when the contact variety becomes the sole language of a community of people who have unified and acquired a new independent identity, as indentured Indian labourers did in Fiji. Under these conditions, levelling (and associated reallocation) may occur in two different ways, or in a combination of both. First, there is the gradual levelling resulting from a ‘‘complex series of accommodation processes’’ (Trudgill 1986: 96). According to the speech accommodation theory of social psychology (Giles 1977), now also called communication accommodation theory (Gallois et al. 1988), people may, again under certain social conditions, modify their speech by adapting to the speech of others. Thus, dialect differences are reduced as speakers acquire features from other varieties as well as avoid features from their own variety that are somehow different. This may

Koine formation and creole genesis 

occur over several generations until a stable compromise dialect develops, as Trudgill (1986: 96) shows for transplanted Norwegian dialects in the new industrial town of Høyanger. Alternatively, the processes may be more rapid as children are born into the community where they hear not only the dialects of their parents but also the mixture being used around them. Here the children acquire only some features and not others, and thus levelling and stabilization are accelerated, as in Fiji (Siegel 1987: 203). A key question involves which features are levelled out and which remain. In describing the development of Guyana Bhojpuri, Gambhir (1988: 77) notes that in the levelling process, ‘‘relatively idiosyncratic features’’ of each dialect are eliminated and the most common shared features have the best chance for remaining. In other words, certain features are ‘‘reinforced’’ by the frequency in which they occur. This view is shared by Trudgill (1986: 98), who defines levelling as ‘‘the reduction or attrition of marked variants’’. Here Trudgill uses the term ‘‘marked’’ to mean ‘‘unusual’’ or ‘‘in the minority’’ in the specific contact situation, and not in any universal sense. What is implied is that in gradual koineization, the least common variants go out of use and in nativization the most common variants are the ones acquired by the children. There is some clear support for this point of view, if we look at other countries where Indian indentured labourers were imported, and where varieties of Overseas Hindi (OH) developed, shown in Table 4. As pointed out by Mesthrie (1991), the earliest recruitment in the Indian indentured labour system, which began in 1834, was concentrated in the state of Bihar where the Bhojpuri dialect is spoken. Later, recruiting moved gradually westward into the area which is now the eastern part of the state of Uttar Table 4. Countries importing Indian indentured labour, and varieties of Overseas Hindi (from Siegel 1988a) Name of country

Years of Labourers importation imported

Current Indian population

Mauritius Guyana Trinidad

1834–1900 1838–1916 1845–1916

453,063 238,909 143,939

670,000 500,000 600,000

South Africa

1860–1911

152,184

1,000,000

Suriname

1873–1916

34,304

120,000

Fiji

1879–1916

60,695

350,000

Name of variety of Overseas Hindi Mauritian Bhojpuri Guyanese Bhojpuri Trinidad Bhojpuri/Hindi South African Bhojpuri Suriname Hindustani/Sarnami Fiji Hindi/Hindustani

 Jeff Siegel

Pradesh, where Avadhi is spoken. The countries which had earlier recruiting, such as Mauritius and Guyana, had the largest percentage of Bhojpuri speakers and the varieties of OH there have the greatest number of Bhojpuri features. On the other hand, the varieties of OH in the countries which had later recruiting, such as Suriname and Fiji, have fewer Bhojpuri and more Avadhi features. However, frequency in the environment does not explain every feature which is retained. For example, in Fiji Hindi, the word che ‘six’ (instead of the more common cha) occurs only in the Chattisgarhi dialect whose speakers made up only 2.6 percent of the North Indian labour force (Siegel 1988a: 13). Thus, factors other than frequency must be responsible for the retention of some features and the levelling out of others. Trudgill (1986: 102), for example, suggests that ‘‘naturalness’’ or lack of typological markedness may also be a factor. This issue is discussed later in this chapter.

. Koines and other contact varieties The distinction between koines and other language contact varieties is often based on the difference between dialect contact and language contact. However, Mufwene (1997) makes several criticisms of this notion. First, he points out (pp. 44–45) that there are problems distinguishing languages and dialects, and the criterion of mutual intelligibility is not always sufficient. I also make this point in my 1985 article — referring to Baegu, Fataleka, To’aba’ita and Baelelea (North Malaita, Solomon Islands) which are mutually intelligible but considered separate languages by their speakers, and to Bhojpuri and Rajasthani which are not mutually intelligible but considered dialects of Hindi (p. 365). In order to avoid the problems associated with the terms ‘‘language’’ and ‘‘dialect’’, I characterized koine formation (or koineization) as the result of the mixing of linguistic subsystems, such as regional dialects, literary dialects and sociolects, defined as follows (p. 365): Two or more different linguistic varieties may be considered subsystems of the same linguistic system if they are genetically closely related and thus typologically similar enough to fulfil at least one of two criteria: (1) they are mutually intelligible or (2) they share a superposed, genetically related linguistic system, such as a national standard or literary language.

Mufwene (1997: 45–46) quotes this definition and criticises it on several fronts. Firstly, he says that ‘‘the correlation between genetic and typological kinship is questionable’’, noting that genetic kin, including dialects, may be typologically

Koine formation and creole genesis

different in one way or another. However, it is not stipulated that any correlation is necessary for koineization — rather, that the varieties considered subsystems must be typologically similar enough to be able to meet one of the two criteria. In other words, the varieties that can undergo koineization are restricted to those with sufficient shared features to enable them to be used reciprocally for successful communications in certain social circumstances. As shown before (Siegel 1985: 366), subsystems can be in contact for many years without any koineization taking place. It only occurs following certain political, social, economic or demographic changes which cause either increased interaction among speakers of the subsystems or decreased inclination to maintain linguistic distinctions. A typical example of such a change is migration. The second criticism Mufwene makes is (p. 46): ‘‘The condition of sharing a superposed system in the form of a ‘national standard or literary language’ would preclude applying the term ‘koine’ to many varieties that developed out of contact of unwritten language varieties.’’ However, these written varieties are mentioned only as examples of superposed systems. The example of ‘‘a regional lingua franca’’ is given elsewhere (Siegel 1987: 23), and it is clear that I include vernacular Hindustani in India as the common denominator among mutually unintelligible dialects, such as Rajasthani and Bhojpuri. Concerning Kituba, Mufwene (p. 46) says that according to my criteria, it would be considered a koine. It developed out of contact between varieties of Kikongo which some linguists call ‘‘dialects’’. But since they are not mutually intelligible, they are not linked by a common variety, and their speakers do not consider themselves as speaking dialects of the same language, these varieties do not qualify as ‘‘subsystems’’ of one system according to my definition. Thus, I fully agree with Mufwene’s view that Kituba is not a koine. Lastly, Mufwene misinterprets my definition when he says (p. 46) that it ‘‘makes allowance for genetically and typologically related languages to produce koinés’’. He also incorrectly attributes to me the point of view that ‘‘distinct languages that are closely related genetically and typologically may produce koines, rather than a pidgin or creole, by mixing features’’ (p. 47). Again, in my 1985 article, I spend three pages (367–370) discussing why the concept of koineization should not apply to the result of contact between distinct languages. I conclude (p. 376): Some linguists have extended the use of the term koineization to include pidginization and other types of language mixing, such as fusion and convergence. However, it would seem more profitable to restrict its use to the mixing of linguistic subsystems.



 Jeff Siegel

This point of view is reiterated in a later article (Siegel 1993b: 118). Nevertheless, in light of these criticisms and misinterpretations, I have revised my definition to that given at the beginning of this chapter, and simply used the term ‘‘mutually intelligible’’. Of course, mutual intelligibility itself is not absolute, and as pointed out by Mufwene (1997: 46) and other authors, it depends on factors such as attitudes, beliefs and goodwill (e.g., see LippiGreen 1994). My definition is that two varieties are mutually intelligible if their speakers can use them with each other for successful everyday communication. Some varieties may be mutually unintelligible in certain circumstances, for example when speakers want to maintain separate linguistic identities, but mutually intelligible in other circumstances, for example when speakers have the need to communicate. Of course, only varieties with enough similarity or a common link have the potential to be mutually intelligible. And under the social circumstances in which koineization occurs, this potential will be taken advantage of. In more general terms, Mufwene (1997: 48) asks: ‘‘[W]hat really distinguishes pidgins and creoles from koines?’’. I will now try to answer this question, starting with an examination of the sociolinguistic conditions which lead to their genesis. Firstly, we have already seen that the type of language contact leading to koine development is unique because of its continuity. Speakers can continue using their own linguistic varieties in everyday interaction with speakers of the other similar varieties. Thus, there is no need for second language learning or code-switching, and no ‘‘imperfect’’ language transmission (Thomason and Kaufman 1988). There is also no transfer (or interference) — i.e. speakers unconsciously carrying over features from one language (usually their first) when speaking (or trying to speak) another language. This kind of contact is quite different from that involved in the development of pidgins and creoles, which at some stage involves the necessity to acquire a new language, and to code-switch to this language for intergroup communication. As a result, in the case of koine formation, there is no ‘‘target variety’’ to be learned, and thus also no substrate-superstrate distinction. The features of one dialect may predominate in the final koine, but that does not mean it was the ‘‘target’’. Unlike in development of other contact languages, this predominant variety itself undergoes changes due to the influence of the other varieties in contact, as Attic Greek itself was affected by Ionic (Bubenik 1993: 14). Thus, it is not accurate to say that Attic Greek was the target variety in the development of the original Greek Koine¯ (Mufwene 1997: 48). In addition, there is no

Koine formation and creole genesis 

one ‘‘lexifier’’; several of the varieties in contact share in providing the lexical and morphological content. Also contrary to Mufwene, in koine formation there is no restricted contact between any of the contributing varieties, as we assume there is in pidgin and creole development. Because of the social changes that lead to koineization as described above, the speakers of the contributing varieties belong to the same newly established speech community. In fact, koine formation requires intimate social interaction between speakers of the different varieties in contact. As a result of all these factors, there is very little if any of the restructuring that comes from attempting to learn a target that one has little access to. Restructuring, if it is involved in koine formation, consists mainly of reduction in the number of grammatical distinctions and forms and some regularization of existing rules. It rarely involves significant innovations or the development of new rules, as we find in pidgins and creoles. In koines, there is mostly mixing of forms from different varieties in contact, not mixing of different rules. So, if we assume that language contact varieties can be placed on a continuum on the basis of the degree of restructuring, koines would be on the opposite end from pidgins and creoles. Consequently, as pointed out first by Nida and Fehderau (1970: 152) and later by Gambhir (1981: 185), koines, as opposed to pidgins and creoles, are not structurally discontinuous from their contributing varieties. As Pasch (1993: 290) observes, koines show systematic correspondences with their ‘‘base language(s)’’ in all four linguistic domains: phonology, morphology, syntax and lexicon, whereas pidgins and creoles show a ‘‘break’’ in one or more of these domains. Because of this break, Thomason and Kaufman (1988) and other scholars believe that pidgins and creoles have no genetic affiliation with any of the languages which were in contact to create them. On the other hand, koines are still genetically closely related to the varieties which were in contact. Thus, pidgins and creoles are generally considered to be new languages whereas koines are considered to be new dialects. It is true that there is no clear linguistic means of distinguishing ‘‘language’’ from ‘‘dialect’’. This leads Mufwene (1997: 48) to deduce that the distinction is not worth maintaining and, therefore, that the distinction between koines and pidgins has ‘‘no particular analytic benefit’’. He concludes (p. 53): Koine is a term/concept which may as well be abandoned. The kinds of contactbased language varieties so-labeled show no particular structural features or language change mechanisms which distinguish them from pidgins and creoles.

 Jeff Siegel

The little bit of motivation for keeping the term . . . would lie in the distinction language versus dialect. However, this distinction seems to be of little use even in subfields of linguistics that are more sensitive to politics and ideology, viz., sociolinguistics, sociology of language, and the ethnography of communication.2

LePage and Tabouret-Keller (1982) similarly point out that without idealization, there is often no clear linguistic means of distinguishing one language from another (just as there is no biological means of distinguishing different races). But we would not want to abandon the term language just because in some cases it is difficult to decide linguistically where the dividing line is between two varieties. Similarly, the prototypical distinctions between dialects and languages, and between koines and pidgin/creoles, are generally clear, especially according to sociolinguistic criteria, and so the terms are still useful. To abandon them for the reasons given by Mufwene would be like abandoning ‘‘black’’ and ‘‘white’’ because of ‘‘grey’’. I am not saying that for every contact variety it is ‘‘black or white’’ whether it is a koine or pidgin/creole. The result of contact between Spanish and Portuguese on the Uruguay-Brazil border (Trudgill 1986) is an example of a grey area. In fact, the results of various kinds of language contact are graded phenomena — for example, the continuum of the degree of restructuring. But it is useful to have sociolinguistic terms characterizing various parts of the contact variety continuum, just as it is to have colour terms for the spectrum.

. Implications for creole genesis One point on which Mufwene and I agree completely is that many of the same processes are involved in the creation of different language contact varieties (Mufwene 1997: 48; Siegel 1997). Although I emphasized the difference between koines and pidgins/creoles earlier in this chapter, I would now like to show how some of the processes and principles of koine formation are relevant to creole genesis. . Mixing, levelling and reallocation The first process is mixing. In arguing against substrate influence in creoles, Bickerton (1981: 49) uses the term ‘‘cafeteria principle’’ (attributed to J. L. Dillard) to poke fun at the idea that a language can be made up of various linguistic features from different language varieties, like a tray of items chosen

Koine formation and creole genesis

for lunch at a cafeteria. He claims that it is ‘‘absurd to propose that a creole could mix fragments’’ of various languages. This point is reiterated in Bickerton (1984: 184): Substratophiles have never attempted to compare whole systems, but have picked out and compared isolated rules and features from creole and substratum languages. Implicit in this operation is the belief that languages can be made by throwing together a heterogeneous set of items.

But looking at the development of Fiji Hindi and other koines, we have seen that it is indeed possible for a new contact variety to develop out of a mixture of features from several contributors. There seems to be no reason why this kind of mixing should not also be possible for pidgins and creoles. A link between koine formation and pidgin/creole development is found in one of the earliest accounts of koineization: Blanc’s (1968) description of the ‘‘koineizing process’’ involved in the emergence of modern Israeli Hebrew. He shows how it arose from the mixture of various literary dialects and ‘‘several substrata’’ (pp. 237–238), meaning versions of Hebrew spoken as a second language by the mainly Eastern European immigrants to Palestine. Because of the transfer and restructuring found in imperfect second language learning, these versions of Hebrew were strongly influenced by the various Eastern European languages in phonology, vocabulary and grammar (p. 249). However, these modified versions of Hebrew were for the most part mutually intelligible and used for everyday interaction by the immigrants, providing a pool of variants from which the new variety of Hebrew emerged. This situation differs from the ones we have been examining so far in that it involved more than one language and some second language learning. Unlike in typical koine formation, there was a target or superstrate, Hebrew, as well as substrate languages, the Eastern European first languages of the immigrants. Thus, varieties used for communication were not established dialects of Hebrew but modified versions of it resulting in part from incomplete second language learning. In this way, the starting point for the development of Israeli Hebrew differs from that of koines, such as Fiji Hindi. But after this point, the processes affecting the formation of the new contact variety were basically the same: levelling and reallocation. Blanc (1968) also gives an account of the gradual levelling that occurred in the development of Israeli Hebrew. Overall it took place over three generations, but occurred more quickly among the initial native speakers and their peers. He describes the situation after levelling had been completed as follows (p. 246): ‘‘There is a relatively stabilized usage, variant forms having by now



 Jeff Siegel

been selected out, some for informal use, some for formal, some neutral, some rejected.’’ This is an indication that some reallocation was also involved in the process, with some variants being allocated new stylistic functions. In summary, contact between speakers of Eastern European languages and literary Hebrew led to various spoken second language versions of Hebrew. But after this stage, when these versions became used for everyday communication among immigrants in Palestine, the situation was similar to what we have seen before with koineization. There was no longer any target language. Day to day interaction was possible without any further second language learning in the newly emerging community. Mixture of the features of these versions of Hebrew then followed, and finally levelling, with some features selected, some reallocated stylistic functions, and others dropping out of use. This was a gradual process, speeded up somewhat by nativization. This kind of mixing and levelling may similarly be important in the genesis of stable pidgin and creole languages, and may help to explain how these varieties develop from the variable pre-pidgin (or jargon) stage. In the initial stages in the development, at least two mutually unintelligible languages are involved. For various reasons, one of these languages is chosen as the medium of communication, and this becomes the target or superstrate. In attempting to communicate in this language, speakers of the other (substrate) language may produce modified versions of the target as a result of the transfer and restructuring typically found in second language acquisition. Model simplification on the part of the superstrate speakers may also lead to some modification. Unlike in some other contact situations, there is restricted access to the target and thus no corrective mechanism to bring these modified versions closer to it. A situation then arises where there is a need for communication among speakers of different substrate languages, and the only common medium they have is the limited knowledge they have of the superstrate language. Their versions of the superstrate may be quite different from those of native speakers, but they are similar enough to each other to facilitate communication. At this stage, the superstrate is no longer a ‘‘target’’ in the normal sense of this term, as pointed out by Thomason and Kaufman (1988) and Baker (1990). The goal is not communicating with superstrate speakers, but with other substrate speakers. As far as substrate speakers are concerned, they are already speaking the same language. At this point, the situation is similar to that of koine development and the emergence of new dialects, such as Israeli Hebrew. The modified versions of the superstrate are used for everyday communication and provide a pool of variants for a new contact variety.

Koine formation and creole genesis 

The processes of levelling and reallocation then commence. Some features are eliminated while others are reassigned stylistic or social functions. The process is gradual, but may be accelerated by nativization, as children acquire only some of the features in the mixture. We have seen similar processes occur with Israeli Hebrew, but because of the diversity of the variants involved in this kind of situation, the majority structurally distinct from the superstrate, it is a new language — a pidgin or creole — not a new dialect, which emerges. The degree of structural distance from the superstrate, however, depends on the degree of restructuring in the initial stage. An example can be found in the development of Melanesian Pidgin (MP). Imperfect versions of English were picked up by speakers of different Melanesian languages through restricted contact with Europeans in sandalwood and beche-de-mer trading in the early and mid 1800s. These versions were then used for communication in more extended contact when plantation labour recruiting began in the 1860s. Crowley (1990) shows that at this early stage there was a great deal of variation. For instance, both modifier + head constructions (e.g., Tana man ‘man from Tanna’) and head + modifier constructions (e.g., man Tana) were found in early MP, but the former was eventually levelled out and only the later remains in modern Bislama, the Vanuatu dialect of MP. Also, the third person plural pronoun had these various forms: (3) hemfala olgeta olgetafala ol

But olgeta is the form that remains in Bislama. Another example of levelling is in the preverbal desiderative marker. The following forms were found in early MP: (4) laek wan wante(m)

Only wante(m) remains with this function in Bislama. However, laek is recognized as the the form used in the Papua New Guinea dialect, Tok Pisin. Thus, one of the forms that has been levelled out of common use in Bislama has been reallocated the function of regional marker. Examples of levelling in creole formation are found in the development of Kriol, the English-lexifier creole spoken by Aborigines in the Northern

 Jeff Siegel

Territory of Australia, as described by Harris (1986, 1991). From 1880 through the turn of the century, with the discovery of gold and the expansion of the pastoral industry, there was extensive contact between Aboriginal people, Chinese immigrants and Europeans from other parts of Australia. Varieties used for communication included various dialects of English, second language varieties of English, and some established contact varieties, such as Chinese Pidgin English and Southeast Australian Pidgin English. In the first decade of the twentieth century, Aborigines in the eastern part of the Northern Territory were nearly annihilated by pastoralists. In 1908 the Anglican church established a mission on the Roper River (now called Ngukurr) where Aborigines from at least eight different language groups took refuge. Here a new community began to form, and the various forms of English, or English-lexified contact varieties known by the Aboriginal people, were used for intergroup communication. This is when a new contact language, Kriol, began to emerge. Children growing up in this environment were exposed to these varieties, as well as to standard English in the mission school. Levelling rapidly took place as children acquired some of the features in the mixture being used around them, but not others. The extensive data from historical texts, presented by Harris (1986), reveals a great deal of variation, both lexical and grammatical, before the emergence of Kriol. For example, two different numeral markers were used, one from Chinese Pidgin English, as in one piecee, and another from Southeast Australian Pidgin English, as in one fella. Only the latter remains in Kriol, as in wanbala ‘one’. Also, transitivity was sometimes unmarked and sometimes marked with a transitive suffix on the verb, either -im or -it: (5) a.

‘make’ (transitive): meik meigim meigit b. ‘give’ (transitive): gib gibim gibit

As the result of levelling, such variation does not occur in Kriol. The transitive form of ‘make’ is meigim, following the more usual pattern with the -im suffix, while that of ‘give’ is gibit, retaining the rarer -it suffix. The variation that is found in Kriol is systematic, and results from reallocation, where earlier variants are now socially marked. Native speakers of Kriol

Koine formation and creole genesis 

command a continuum of styles ranging from what they describe as ‘‘heavy’’ to ‘‘light’’, heavy being typically closer to Aboriginal languages and light closer to English. For example, in the heavy style the third person pronouns are im (singular), dubala (dual) and olabat (plural), while in the light style they are im (singular) and dei/dem (plural). (See Sandefur 1979: 86–87.) The heavy style is generally used in Aboriginal contexts, while the light style is used for communication with non-Aborigines.3 To summarize, it is usually assumed that in creole genesis, and especially in radical or abrupt creolization, linguistic restructuring of the superstrate language, widespread nativization and the emergence of a creole all happen simultaneously. However, the viewpoint presented here suggests that substantial contact-induced restructuring occurs before the social conditions are right for the emergence of a new stable contact variety. When the conditions are right, then these already modified forms of the superstrate language become part of the mixture of variants which are used for communication. When a new community begins to form, then there is levelling of these variants to form the new contact variety. Nativization is important only in that it may accelerate the levelling process. . Reinforcement Principles similar to those in koineization also appear to influence which features are retained in new dialects and in pidgins and creoles. With regard to Israeli Hebrew, in the early years of immigration to Palestine, speakers of Eastern European languages were the vast majority. Typological similarities among these languages resulted in similar kinds of transfer and restructuring and consequently similar second language versions of Hebrew. The frequency of these transferred features in the pool meant that they were reinforced and ultimately retained, and thus there is a clear Eastern European substrate influence in modern Hebrew. So here we see, as we did in koineization, the reinforcement of certain features in the levelling process according to the numbers of speakers who use them — only in this case it is reinforcement of substrateinfluenced features, not dialect features. We can call this substrate reinforcement. In the development of pidgins and creoles, as with Israeli Hebrew, when substrate languages are typologically similar, their speakers’ second language versions of the superstrate language will also be similar in terms of transfer and restructuring. Thus, when levelling occurs, substrate-influenced features will be common, and therefore reinforced, having a good chance of being

 Jeff Siegel

retained in the emerging contact language. When the substrate languages are typologically heterogeneous, then second language versions will be more varied. Since no substrate-influenced features are very common, and since none will have strong majority support, most of them will be levelled out. This point of view converges with the notion that the extent of substrate influence in creoles is directly proportional to the degree of homogeneity in the substrate languages (e.g., Sankoff 1984; Mufwene 1986, 1990; Singler 1988; Thomason and Kaufman 1988). However, it is clear that factors other than frequency affect retention of features in pidgins, creoles and other contact varieties. For example, in Fiji Hindi, the word kauncı¯z or kauncı¯ ‘what’ was retained even though the only regional dialect in which it occurs is Magahi, which in Fiji was spoken by only 3.8 percent of the North Indian immigrants (Siegel 1987: 141). This form is also found in other varieties of Overseas Hindi. It is also significant that this is a bimorphemic question word, derived from which or what thing, as often found in pidgins and creoles: (6) Fiji Hindi (and other OH varieties): kauncı¯ (kaun ‘which/who’, cı¯z ‘thing’) Chinese Pidgin English wat ting Cameroon Creole wetin

There are other similarities as well between koines such as Fiji Hindi and other contact varieties. For example, a pronominal plural marker derived from a word meaning ‘person’ or ‘people’ or ‘a group of people’, similar to the Fiji Hindi log, is found in Tok Pisin and Kriol, and in some varieties of Aboriginal English and Fiji English. This is shown in Figure 1. (The Tok Pisin and Kriol suffixes -pela and -bala are derived from the English word fellow.) Also, in many contact varieties, including koines, there is post-verbal periphrastic marking of completive aspect with grammaticalization of a word meaning ‘finish’ or ‘already’: (7) Fiji Hindi: Ham kha¯na¯ kha¯ya¯ khala¯s. (foreigner talk register) I food eat finish ‘I’ve eaten.’ Tok Pisin: Singapore English:

Mi kaikai pinis. The blood clot already.

‘I’ve eaten.’ ‘The blood has clotted.’ (Williams 1987: 103)

As I have pointed out before (Siegel 1997), we need to explain such linguistic similarities among contact varieties in general, not just among different

Koine formation and creole genesis

Source varieties Degree of restructuring/transfer

Pool of variants

Mixing Levelling Australian English Koine

Singapore English

Caymanian

Indigenized Semi-Creole new dialect

Jamaican Creole

Saramaccan

Pidgin/Creole

Figure 1. Continuum of contact varieties

pidgins and creoles. The Language Bioprogram Hypothesis (Bickerton 1984, 1988) has been put forward to explain similarities among creoles, but it cannot apply to koines, new dialects and pidgins because children’s first language acquisition is not necessarily involved in the emergence of these varieties. Principles of second language acquisition have also been invoked to account for similarities (Goodman 1985; Thomason and Kaufman 1988; Mufwene 1991) but they cannot apply to koines where no second language is involved. The explanation seems to be that, in addition to frequency, there are other common factors which determine which features are retained during levelling or, as some linguists put it, which features are ‘‘selected’’ by the emerging contact variety (Mufwene 1990). Many such factors have been proposed, including linguistic factors such as markedness (Mufwene 1990, 1991), semantic transparency (Seuren and Wekker 1986), regularity (Williams 1987) and salience (Mufwene 1991), and social factors such as prestige, political power and cultural values and attitudes of the groups using particular variants (Siegel 1993b: 118). Mufwene (1990: 12) suggests a weighting of these factors to come up with a set of ‘‘selection principles’’ which will predict for each contact situation which features in the mixture will be eliminated and which will be retained. Mufwene (1990) has shown that markedness is relative depending on the contact situation, and is related to other factors such as frequency. Leaving these factors aside, there are two others that seem to be relevant. The first is perceptual salience; it is generally assumed that unbound morphemes, especially those that are stressed, are perceptually more salient than bound or



 Jeff Siegel

unstressed morphemes. Goodman (1985: 132) notes the fact that ‘‘pidgins as well as creoles have the same tendency to select free rather than bound morphemes from the target language . . .’’ Consistency (or uniformity or invariance) is another factor. Andersen’s ‘‘One-to-one Principle’’ (1984: 79, 1990: 51) was designed to account for the forms selected in interlanguage. It also refers to perceptual salience, but its main claim has to do with consistency: one form, one meaning. It states that ‘‘an interlanguage system should be constructed in such a way that an intended underlying meaning is expressed with one clear invariant surface form (or construction)’’. Andersen (1984: 80) compares his One-to-One Principle to Naro’s (1978: 340–341) ‘‘Factorization Principle’’ which shapes native speakers’ Foreigner Talk: ‘‘Express each invariant, separately intuited element of meaning by at least one phonologically separate, invariant stress-bearing form . . .’’ This combination of salience and consistency makes a feature more transparent. Thus in the levelling process, when there is no clear majority or unmarked form, structures with the greater chance of being retained may be those composed of stressed free morphemes, each having one consistent meaning or function. The features common to several contact varieties shown above for the most part shared these characteristics earlier in their development. Plural pronouns made up of the singular plus a plural marker are more transparent than completely separate, unrelated forms. A periphrastic, stressed postverbal perfective marker is more transparent than a synthetic construction. A question word originally made up of two separate words may also be more transparent, especially if it is part of paradigm such as what thing ‘what’, what place ‘where’ and what time ‘when’.

. Conclusion In general, the preceding analysis presents the view that similar processes may operate in the emergence of different language contact varieties, here exemplified by koine formation and creole genesis, and also including the development of new dialects such as Israeli Hebrew and Fiji English. We see that under similar social conditions, involving unification or focussing of a previously diverse community, the same processes of mixing and levelling seem to be in operation. This may explain the general similarities among contact varieties, such as the apparent mixture of features from

Koine formation and creole genesis 

varieties in contact and continua of social and stylistic variation. In addition, the similarities among more specific features of different contact varieties may be due to the same linguistic principles affecting selection of features in levelling, the most important of these being frequency and a combination of salience and consistency. The broad linguistic differences among contact varieties are due to the nature of the pool of variants that undergoes mixing and levelling. In the case of koines, the pool is comprised mainly of already existing varieties of a language. But in other varieties, such as pidgins and creoles, the pool contains a large proportion of newly modified versions of a language resulting from transfer and restructuring in second language learning and from model simplification. The degree of modification depends on both linguistic factors, such as the typological distance between the languages involved, and social factors, such as the degree of access to the language being modified. Thus, with regard to input into the pool of variants, there is a continuum ranging from very little restructuring to drastic restructuring. This continuum is reflected in the contact varieties which emerge after levelling, ranging from koines at one end and pidgins and creoles at the other. Of course, individual creoles also vary in the degree of their restructuring depending on conditions in the initial stage (Bickerton 1984, Schneider 1990). In the middle, in the grey areas, are the semi-creoles, such as Caymanian and Réunionnais (see Figure 1.). Although the different contact varieties range along a continuum according to linguistic features, we have seen that it is possible to distinguish them according to sociolinguistic criteria. In other words, we should not worry about trying to come up with purely linguistic definitions of what are primarily sociolinguistic phenomena. However, we have also seen that different types of contact varieties, such as koines and creoles, may have gone through some of the same processes in their development. We should now concentrate on understanding both the linguistic and social factors involved in these processes.

Notes * Thanks go to Diana Eades for comments on earlier drafts of this chapter. . Here I am referring to naturally developing koines, not to consciously planned koines, such as those described by Mühlhäusler (1993) and Hancock (1993). . However, Mufwene himself finds it difficult to dispense with the dialect/language distinction, as he uses it several times in discussion later in his article (pp. 57, 64).

 Jeff Siegel

. Some creolists (e.g. Alleyne 1971; LePage 1977; Mufwene 1986, 1988; Valdman 1991) have disagreed with the established point of view that the typical ‘‘post-creole’’ continuum, as found in Jamaica and Guyana, develops from ‘‘decreolization’’ as a result of renewed contact between a creole and its superstrate language. They believe instead that the social and stylistic variation in creoles are present from early in their development, and point to evidence, for example, in travellers’ accounts. This would be expected if we assume that part of the levelling process includes reallocation of certain variants to social and stylistic markers, as occurred in Australia.

References Alleyne, Mervyn C. 1971. ‘‘Acculturation and the cultural matrix of creolization’’. In Pidginization and creolization of languages, Dell Hymes (ed.), 169–186. Cambridge: Cambridge University Press. Amery, Rob. 1993. ‘‘An Australian koine: Dhuwaya, a variety of Yol u Matha spoken at Yirrkala in North East Arnhemland’’. International Journal of the Sociology of Language 99: 45–64. Andersen, Roger W. 1984. ‘‘The one to one principle of interlanguage construction’’. Language Learning 34.4: 77–129. Andersen, Roger W. 1990. ‘‘Models, processes, principles and strategies: Second language acquisition inside and outside the classroom’’. In Second language acquisition–foreign language learning, Bill van Patten and James F. Lee (eds.), 45–68. Clevedon, Avon: Multilingual Matters. Baker, Philip. 1990. ‘‘Off target?’’. Journal of Pidgin and Creole Languages 5.1: 107–119. Barz, Richard K. and Jeff Siegel (eds.). 1988. Language transplanted: The development of Overseas Hindi. Wiesbaden: Harrassowitz. Bickerton, Derek. 1981. Roots of Language. Ann Arbor, MI: Karoma. Bickerton, Derek. 1984. ‘‘The Language Bioprogram Hypothesis’’. The Behavioral and Brain Sciences 7: 173–221. Bickerton, Derek. 1988. ‘‘Creole languages and the bioprogram’’. In Linguistics: The Cambridge survey, vol. ii, Linguistic theory: Extensions and implications, Frederick J. Newmeyer (ed.), 268–284. Cambridge: Cambridge University Press. Bickerton, Derek. 1992. ‘‘The sociohistorical matrix of creolization’’. Journal of Pidgin and Creole Languages 7.2: 307–318. Blanc, Haim. 1968. ‘‘The Israeli koine as an emergent national standard’’. In Language problems in developing nations, Joshua Fishman, Charles A. Ferguson and Jyotirindra Das Gupta (eds.), 237–251. New York: Wiley. Bubenik, Vit. 1993. ‘‘Dialect contact and koineization: The case study of Hellenistic Greek’’. International Journal of the Sociology of Language 99: 9–24. Chaudenson, Robert. 1977. ‘‘Toward the reconstruction of the social matrix of creole language’’. In Pidgin and creole linguistics, Albert Valdman (ed.), 259–276. Bloomington, IN: Indiana University Press.

Koine formation and creole genesis 

Crowley, Terry. 1990. Beach-la-Mar to Bislama: The emergence of a national language of Vanuatu. Oxford: Clarendon Press. Damsteegt, Theo. 1988. ‘‘Sarnami: A living language’’. In Language transplanted: The development of Overseas Hindi, Richard K. Barz and Jeff Siegel (eds.), 95–120. Wiesbaden: Harrassowitz. Domingue, Nicole. 1981. ‘‘Internal change in a transplanted language’’. Studies in the Linguistic Sciences 4.2: 151–159. Ferguson, Charles A. 1959. ‘‘The Arabic koine’’. Language 35.4: 616–630. Gallois, Cynthia, Arlene Franklyn-Stokes, Howard Giles and Nikolas Coupland. 1988. ‘‘Communication accommodation in intercultural encounters’’. In Theories in intercultural communication, Y. Y. Kim and W. B. Gudykunst (eds.), 157–185. (International and Intercultural Communication Annual, 12.) Newbury Park: Sage. Gambhir, Surendra K. 1981. The East Indian Speech Community in Guyana: A sociolinguistic study with special reference to koine formation. Ph.D. dissertation, University of Pennsylvania. Gambhir, Surendra K. 1988. ‘‘Structural development in Guyanese Bhojpuri’’. In Language transplanted: The development of Overseas Hindi, Richard K. Barz and Jeff Siegel (eds.), 69–94. Wiesbaden: Harrassowitz. Giles, Howard. 1977. ‘‘Social psychology and applied linguistics: Towards an integrated approach’’. ITL Review of Applied Linguistics 35: 27–42. Goodman, Morris. 1985. ‘‘Review of Roots of Language by Derek Bickerton’’. International Journal of American Linguistics 51.1: 109–137. Hancock, Ian. 1993. ‘‘The emergence of a Union dialect of North American Vlax Romani, and its implications for an international standard’’. International Journal of the Sociology of Language 99: 91–104. Harris, John W. 1986. Northern Territory pidgins and the origin of Kriol. (Pacific Linguistics C-89). Canberra: Australian National University. Harris, John W. 1991. ‘‘Kriol — the creation of a new language’’. In Language in Australia, Suzanne Romaine (ed.), 195–203. Cambridge: Cambridge University Press. LePage, Robert. 1977. ‘‘Processes of pidginization and creolization’’. In Pidgin and Creole Linguistics, Albert Valdman (ed.), 222–255. Bloomington, IN: Indiana University Press. LePage, Robert and Andrée Tabouret-Keller. 1982. ‘‘Models and stereotypes of ethnicity and language’’. Journal of Multilingual and Multicultural Development 3.3: 161–192. LePage, Robert and Andrée Tabouret-Keller. 1985. Acts of Identity: Creole-based approaches to language and ethnicity. Cambridge: Cambridge University Press. Lippi-Green, Rosina. 1994. ‘‘Accent, standard language ideology, and discriminatory pretext in the courts’’. Language in Society 23: 163–198. Mesthrie, Rajend. 1991. Language in Indenture: A sociolinguistic history of Bhojpuri-Hindi in South Africa. London: Routledge. Mesthrie, Rajend. 1993. ‘‘Koineization in the Bhojpuri-Hindi diaspora — with special reference to South Africa’’. International Journal of the Sociology of Language 99: 25–44. Mohan, Peggy. 1978. Trinidad Bhojpuri: A morphological study. Ph.D. dissertation, University of Michigan. Mufwene, Salikoko S. 1986. ‘‘The universalist and substrate hypotheses complement one

 Jeff Siegel

another’’. In Substrata versus universals in creole genesis, Pieter Muysken and Norval Smith (eds.), 129–162. (Creole Language Library, 1). Amsterdam and Philadelphia: John Benjamins Mufwene, Salikoko S. 1988. ‘‘Starting on the wrong foot’’. Journal of Pidgin and Creole Languages 3.1: 109–117. Mufwene, Salikoko S. 1990. ‘‘Transfer and the substrate hypothesis in creolistics’’. Studies in Second Language Acquisition 12: 1–23. Mufwene, Salikoko S. 1991. ‘‘Pidgins, creoles, typology and markedness’’. In Development and structures of creole languages: Essays in honor of Derek Bickerton, Francis Byrne and Thom Huebner (eds.), 123–143. (Creole Language Library, 9). Amsterdam and Philadelphia: John Benjamins Mufwene, Salikoko S. 1997. ‘‘Jargons, pidgins, creoles, and koines: What are they?’’ In The structure and status of pidgins and creoles, Arthur Spears and Donald Winford (eds.), 35–70. Amsterdam and Philadelphia: John Benjamins. Mühlhäusler, Peter. 1993. ‘‘German koines: Artificial and natural’’. International Journal of the Sociology of Language 99: 81–90. Muysken, Pieter and Norval Smith (eds.). 1986. Substrata Versus Universals in Creole Genesis. Amsterdam and Philadelphia: John Benjamins. Naro, Anthony J. 1978. ‘‘A study on the origins of pidginization’’. Language 54:2: 314–347. Nida, Eugene A. and Harold W. Fehderau. 1970. ‘‘Indigenous pidgins and koines’’. International Journal of American Linguistics 32:2: 146–155. Pasch, Helma. 1993. ‘‘Phonological similarities between Sango and its base language. Is Sango a pidgin/creole or a koiné?’’. In Topics in African linguistics, Salikoko S. Mufwene and Lioba Moshi (eds.), 279–292. Amsterdam and Philadelphia: John Benjamins. Sandefur, John R. 1979. An Australian Creole in the Northern Territory: A description of Ngukurr-Bamyli dialects (Part 1). (Work Papers of SIL-AAB Series B, Volume 3.) Darwin: Summer Institute of Linguistics. Sankoff, Gillian. 1984. ‘‘Substrate and universals in the Tok Pisin verb phrase’’. In Meaning, form, and use in context: Linguistic applications, Deborah Schiffrin (ed.), 104–119. Washington, DC: Georgetown University Press. Schneider, Edgar W. 1990. ‘‘The cline of creoleness in English-oriented creoles and semicreoles of the Caribbean’’. English World-Wide 11.1: 79–113. Seuren, Pieter A. M. and Herman Wekker. 1986. ‘‘Semantic transparency as a factor in creole genesis’’. In Substrata versus universals in creole genesis, Pieter Muysken and Norval Smith (eds.), 57–70. (Creole Language Library, 1). Amsterdam and Philadelphia: John Benjamins. Siegel, Jeff. 1985. ‘‘Koines and koineization’’. Language in Society 14: 357–378. Siegel, Jeff. 1987. Language Contact in a Plantation Environment: A sociolinguistic history of Fiji. Cambridge: Cambridge University Press. Siegel, Jeff. 1988a. ‘‘Introduction’’. In Language transplanted: The development of Overseas Hindi, Richard K. Barz and Jeff Siegel (eds.), 1–19. Wiesbaden: Harrassowitz. Siegel, Jeff. 1988b. ‘‘The development of Fiji Hindustani’’. In Language transplanted: The development of Overseas Hindi, Richard K. Barz and Jeff Siegel (eds.), 121–149. Wiesbaden: Harrassowitz. Siegel, Jeff. 1993a. ‘‘Controversies in the study of koines and koineization’’. International

Koine formation and creole genesis 

Journal of the Sociology of Language 99: 1–8. Siegel, Jeff. 1993b. ‘‘Dialect contact and koineization: A review of Dialects in contact by Peter Trudgill’’. International Journal of the Sociology of Language 99: 105–121. Siegel, Jeff. 1997. ‘‘Mixing, levelling and pidgin/creole development’’. In The structure and status of pidgins and creoles, Arthur Spears and Donald Winford (eds.), 111–149. Amsterdam and Philadelphia: John Benjamins. Singler, John Victor. 1988. ‘‘The homogeneity of the substrate as a factor in pidgin/creole genesis’’. Language 64.1: 27–51. Thomason, Sarah Grey and Terrence Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Thomson, George. 1960. The Greek Language. Cambridge: W. Heffer & Sons. Trudgill, Peter. 1986. Dialects in Contact. Oxford: Blackwell. Valdman, Albert. 1991. ‘‘Decreolization or dialect contact in Haiti?’’. In Development and structures of creole languages: Essays in honor of Derek Bickerton, Francis Byrne and Thom Huebner (eds.), 75–88. (Creole Language Library, 9). Amsterdam and Philadelphia: John Benjamins. Williams, Jessica. 1987. ‘‘Non-native varieties of English: A special case of language acquisition’’. English World-Wide 8.2: 161–199.

Chapter 8

Koineization and creole genesis* Remarks on Jeff Siegel’s contribution Frans Hinskens

.

Introduction

The debate regarding Bickerton’s bioprogram hypothesis (1984) has — among other things — led to an increased interest in socio-historical aspects of the coming of age of creole languages. This seems to have coincided with the growing general conviction that the emergence and development of creole languages are processes which are only in certain respects unique. Some authors have even gone so far as to suggest that creole languages, apart from their relatively short history and maybe certain aspects of their social position, from the linguistic point of view (a) do not constitute a group, and (b) are not essentially different from older languages (Muysken 1988). In this light, research and reflection on the question of to what extent creole linguistics and the study of language contact overlap (the theme of the 1995 Amsterdam Creole Workshop) should definitely be welcomed. For years, Jeff Siegel’s work has focussed on this question, more in particular on the relevance of insights into processes of koineization to the development of creole languages. His work concentrates on the sociolinguistic situation in the Fiji Islands, an archipelago in the Pacific Ocean. The present chapter presents some comments on Siegel’s contribution, and in particular on some of his data and ideas regarding the relationship between koineization and creole genesis. Siegel’s findings and proposals will be looked at from four angles: (1) a. conceptual and methodological aspects (Sections 2 and 3, the second part of Section 4, and Section 7), b. the currently occurring processes of levelling and koineization among traditional dialects of languages spoken in the Old World (the first part of Section 4),

 Frans Hinskens

c. insights into the approach to koineization in historical linguistics (Section 5), as well as d. the study of creole languages, language contact and language acquisition (Section 6). In each of these respects, the discussion will be brief and sketchy. It will not be strictly confined to Siegel’s contribution to the present volume only. The main aim of this discussion is to point out possibilities of deepening research in the direction Jeff Siegel’s chapter seems to indicate.

. Terminology, definitions and the conceptual framework One of the reasons why the notion of koine seems so elusive is that different authors stress different aspects of its nature. As Siegel (1993a: 5) points out, some writers focus on linguistic aspects. An example is Dillard’s (1972: 302) definition of a koine as ‘‘a ‘common’ dialect which lacks prominent features of the more conventional dialects’’. Other writers emphasize more functional aspects, defining a koine as a regional lingua franca. Koines result from prolonged contact between related linguistic systems, usually more or less contiguous dialects of the same language, sometimes closely related languages (cf. Siegel 1993b: 118–119), i.e. between linguistic systems which are sufficiently similar to be mutually intelligible (cf. Siegel, this volume). However, continuous contact between related systems as such does not necessarily lead to the formation of a koine. What is needed is sufficient ‘‘integration or unification of the speakers of the varieties in contact’’ (Siegel 1985: 369). This may result from ‘‘some large-scale political, economic or demographic change in society which causes increased interaction among speakers of different dialects and decreased inclination to maintain linguistic boundaries.’’ (Siegel 1992: 110 — cf. 1986: 456; 1993b: 116–117). In short, social and socio-psychological conditions have to be favorable. The process of koine formation itself can be roughly defined as structural convergence between closely related linguistic systems, eventually leading to the stabilization of some compromise variety. In part, this rough definition concurs with an earlier article by Siegel (1985: 370) in which he states that ‘‘[k]oineization leads to the development of a new compromise variety with features of the contributing varieties, whereas convergence leads to changes in the contributing varieties themselves without development of a new variety.’’ 1

Koineization and creole genesis 

Suppose, however, that the original dialects disappear (as seems to be the case of Avadhi, Bhojpuri, Braj2 as well as Bazaar Hindustani, the main ancestors of Fiji Hindi, the language Siegel’s work focuses on) and that in the longer run only the new, blended variety remains. Wouldn’t that be called a koine then? This point is of some importance, since there is an obvious need for a clear definition of the notion of koine, and such a definition will have to cover the basic insights into how a koine comes into existence. However, the essence of the phenomenon is still partly unknown. In consequence, the definition will inevitably be a stipulative (rather than a descriptive) one, i.e. a reasonable agreement on what is meant by the notion of koine. In his 1985 article (p. 365), the author states that the development of a koine, i.e. of a new, compromise variety starts with ‘‘the mixing of features of [. . .] different dialects’’. In Siegel 1988b (p. 122) koineization is defined simply as dialect mixing. Interestingly, this mixing sometimes leads to the development of interdialect forms, i.e. new variants which are formally intermediate between the variants of different dialects. Siegel (1993b: 114) demonstrates this with some examples for the contact between several dialects of North Malaitan transplanted from the Solomon Islands to Fiji.3 In the introduction of Siegel’s contribution to the present volume, the author defines koines as stabilized contact varieties resulting from the mixing and subsequent levelling of features of different varieties. In his account, the first part of the definition, the mixing of features of different varieties, seems to have been pushed somewhat to the background. This may have happened partly under the influence of Trudgill’s Dialects in contact from 1986, to which the author refers extensively. As Trudgill observes in this monograph, the contact between dialects may lead to an enormous amount of linguistic variability in the early stages. Koineization is a complex of processes to reduce the abundant formal variation. Trudgill apparently does not consider the outcome, a koine, as a blend of different dialects. In Trudgill’s view, koineization, which is a focusing process (Le Page and Tabouret-Keller 1985), is a combination of dialect levelling and simplification. In this view, levelling occurs between dialects, whereas simplification is a process taking place within a single dialect. However, there seems to be no reason why the notion of levelling should be restricted to the reduction of interdialectal variation alone. It cannot be maintained that the intra-systemic effects of contact with another dialect always amount to simplification. In other words, it should be assumed that levelling can affect both inter- and intra-systemic variation and, intra-systemically, may or may not result in simplification. There is, in fact, much evidence

 Frans Hinskens

to support this view. The following example serves to illustrate the latter point. Limburg dialects of Dutch traditionally have a system of noun pluralization which is much more complex than that of other varieties of Dutch, including the standard language. Recent research on a specific rural Limburg dialect shows that the levelling which is taking place in the noun pluralization system concerns the parts of morphology where the differences between the dialect and other varieties of Dutch (again, including the standard) are relatively small, leaving intact both the parts of the system that are most different from other dialects and the parts that are most complicated. At a certain place within the system there even seems to be a development in the direction of more complexity (Hinskens 1992: Section 6.3.18). A final point regarding the terminology and the framework concerns Siegel’s (this volume) observation that there are immigrant and regional koines. The question arises as to whether there are reasons to assume that they are different or behave differently. Are there any essential differences between these two types in the way they come into being or in their further development? Both develop in situations of ‘‘increased interaction and [. . .] the need for unification among speakers of different dialects’’ (Siegel 1992: 110; Domingue 1981: 150). However, in contrast to regional koines, immigrant koines are not regional lingue franche or standard languages and frequently they function as the main language in the immigrant communities, thus Siegel (1993a: 6). But do these sociological differences generally cause any structural differences or differences in the way the two types of koines evolve? Do the processes leading to the development of a regional koine occur gradually compared to those resulting in an immigrant koine? Does the development of immigrant koines proceed more rapidly because nativization is much more urgent in this situation? Siegel’s conception of the role of nativization in processes of koineization will be discussed in Section 5 below. As suggested in Auer and Hinskens (1996: Sections 3.2 and 3.4), processes of convergence between related dialects in communities of dialect speaking migrants might differ to some extent from those taking place in the non-migrant type of situation. After all, in the former type of situation the dialect speakers are usually no longer in direct contact with related dialects (including the standard language) which are either not represented or are only weakly represented in the migrant community, whereas in the latter type of situation the standard is present. In immigrant situations, isn’t it the lack of access to the ancestral standard languages which prompts rapid koineization rather than nativization as such? This is one of the issues on which future research could focus.

Koineization and creole genesis 

. Data In the contribution at issue, Siegel presents the contours of an overarching model in which processes of koineization and creole genesis are related to each other. The emphasis is on the structural tendencies in Fiji Hindi4 and their socio-historical motivations, as well as on Siegel’s ideas for a more general model for the similarities and differences between several types of contact varieties. The few findings regarding structural tendencies in Fiji Hindi discussed in Siegel’s contribution mainly serve illustrative purposes, so one should not be too critical as to the scarcity of data and findings presented — the more so since the author has published several other papers as well as a monograph on the sociolinguistic history of Fiji in which more findings have been discussed in greater detail. Yet, a few questions do seem pertinent. In Table 1 and annexes, the author discusses the results of the mixture of features of Bhojpuri (B), Avadhi (A) and Bazaar5 Hindustani (H) in the Fiji Hindi system of verbal suffixes marking person and number. Most of the logically possible types of mixture seem to occur: Fiji Hindi has suffixes originating in A, in B, and in H, suffixes occurring in both A and B, and mixtures of A and H suffixes; it even has a possessive pronoun second person singular which constitutes a mixture of the A, B and H forms. (These mixed variants qualify as interdialectal variants, although Siegel does not use this terminology in this connection.) But obviously the Fiji Hindi paradigms for the marking of person and number in verb morphology and in the system of personal pronouns does not contain any forms which are common to B and H or mixtures of B and H features. Is this an accidental or a systematic gap? In other words, are there no areas in the grammar of Fiji Hindi at all in which B and H features coincide or have been mixed? If, indeed, there are none, then the obvious next question is why? Is there no overlap between B and H in these provinces of morphology? Are there internal constraints ruling out the cooccurrence of the B and H variants in Fiji Hindi or is there a sociolinguistic motivation? Siegel’s Table 3 shows the Avadhi and the ‘reduced’ Fiji Hindi systems of nominative personal pronouns. Interestingly, in Fiji Hindi the noun log, ‘people’, has been grammaticalized into the pronominal plural marker log — and, as a matter of fact, into the ‘‘general pluralizer’’ (Siegel 1987: 195–196, 1988b: 145). In as far as number is marked through affixation, the existence of one single generic plural marker for both nominal and pronominal NP’s is very common for creole languages; in most of those cases, however, the affix is etymologically related to the third person plural personal pronoun of the

 Frans Hinskens

lexifier language. An example is Negerhollands sender which, in its heyday, had no less than six different, though related, grammatical functions (cf. Hinskens and van Rossum 1995). The provenance of the Fiji Hindi general pluralizer is similar to that of creole languages such as Berbice Dutch. The Berbice Dutch generic pluralizer -apu is derived from the Eastern Ijo [+human] plural nominalizer apu , which still has nominal properties (Kouwenberg 1994: 230ff.; Arends, Kouwenberg and Smith 1995: 105). As far as personal pronouns are concerned, the Fiji Hindi pluralizer reminds one of colloquial varieties of American English, where folks, guys, people occur in you guys etc., to disambiguate second person plural you; cf. also all in you all, y’all. Another aspect of the data presented in Table 3 leads to the question whether the Fiji Hindi nominative pronoun system only displays a ‘‘reduction of complexity’’? After all, Avadhi does not seem to formally distinguish between singular and plural second person pronouns, whereas Fiji Hindi does.

. Levelling processes In his discussion of levelling, one of the processes involved in koineization, Siegel (this volume) points out that one of the central questions is which features are levelled out and which remain. He quotes Gambhir (1988), who observed that in the levelling process the more idiosyncratic features of the dialects involved are disposed of whereas the most frequently occurring shared features have the best chance for surviving. A very similar finding emerges from several relatively recent sociodialectological studies of Old World dialects, specifically of dialects of Swedish (Thelander 1980, 1982), Spanish (Holmquist 1988), Dutch (Scholtmeijer 1992; Hinskens 1992) and German (Auer 1997). From all of these studies it appears that dialect features tend to get levelled out earlier and more rapidly the more limited their geographical distribution and, conversely, that dialect features tend to be more resistant the larger their areal spread. At closer consideration, this general finding constitutes a hen-and-egg problem: are dialect features less resistant because of their smaller areal spread, or is their areal spread smaller because they are less resistant? In the latter case, their vulnerability is obviously related to (an)other factor(s). Further research is needed to answer this question. Scholtmeijer’s (1992) work is interesting, in part because the case he describes seems to lie somewhere in between Siegel’s regional versus immigrant situation types. Scholtmeijer studied standard Dutch in three new polders in the

Koineization and creole genesis 

IJsselmeer, a lake to the east of Amsterdam. These polders were settled from 1930, ‘42 and ‘57 onward, respectively. The settlers came from almost all parts of the Netherlands. Of course, all of these people, many of whom were farmers, brought their own dialect with them. Scholtmeijer investigated the settlers’ varieties of standard Dutch, limiting his study to phonology and phonetics, i.e. to what he refers to as accents. He obviously was not interested in what happened to the settlers’ original dialects, nor in the question whether or not a koine emerged. Nevertheless, some of his findings seem to be relevant to the question which types of dialect features survive levelling processes. Scholtmeijer’s findings show generally that those accents appear to win out which are best represented in terms of numbers of speakers. This finding is in line with Siegel’s conclusions for koineization; the number of speakers is also the factor underlying substrate reinforcement in the case of Modern Hebrew, sketched in Section 5 of Siegel’s contribution.6 To a large extent it is comparable to the factor of the relative geographical spread of dialect features. A more general point needs to be made with respect to the issue of the sifting of dialect features by processes of levelling, i.e. with respect to the question which dialect features are going to be ‘‘dumped’’ (Markey 1986: 12) and which ones are allowed to start a new life in the emerging koine. One of the major questions in this connection seems to be whether it is at all possible to distinguish between external and internal factors7 or principles accounting for the transfer and retention of dialect features. As regards external factors, one should think first and foremost of sociolinguistic and sociological parameters as well as mechanisms of language acquisition (cf. Section 6 below). If this question is answered in the affirmative, the logical next question is: how can the relevant external and internal factors guiding processes of koineization be identified and distinguished? Only after it has been established which external and internal factors play a role in a given case of koine formation, can one start studying and, eventually, weighting them; according to Mufwene (after Siegel, this volume), the latter is an important task for researchers in this area. In order to be able to weigh the factors which play a role in a specific case of koine formation, one needs vast amounts of quantitative data.8

. Koineization and beyond In his discussion with Mufwene (1997) on what distinguishes koines from pidgins and creole languages, Siegel points out general differences in the

 Frans Hinskens

sociolinguistic conditioning leading to their genesis (among other things). One of these differences is that ‘‘in the case of koine formation, there is no ‘target variety’ to be learned, and thus also no substrate-superstrate distinction.’’ But, one might ask, was the Fiji situation that balanced? In other words, were there no differences in prestige between, say, Avadhi, Bhojpuri, Braj and Bazaar Hindustani? An equal prestige among the contributing varieties is one of the three criteria for koineization advanced by Hock (1991: 485). Two other criteria which are often applied in historical linguistics are that the linguistic systems involved be closely related languages or mutually intelligible dialects, and that there be no link language. The first of these two additional conditions seems to be met in the Fiji case. This does not hold for the last condition, in other words there was a link language. Out of Hindustani, ‘‘the language of wider communication in India’’ where it was in use as a lingua franca (Siegel 1988b: 124 and 1992: 93), a pidginized variety developed, which is referred to as Bazaar Hindustani or Basilectal Hindustani (note 5 above). Initially this Bazaar or Basilectal Hindustani served as some sort of a link language between the members of the different ethnic groups among the Indian immigrants,9 the Europeans, the Fijians and other Pacific Islanders in Fiji (Siegel 1988b: 127). It was a major ingredient in the mixture, levelling, and structural elaboration leading to the development of Fiji Hindi (Siegel 1992: 110). An aspect that does not seem to be very relevant to the immigrant situation of koine formation at Fiji is discussed in Cortelazzo’s (1993) sketch of specific sociolinguistic developments in the history of Italian. For a long period after the unification of Italy, an overwelming majority of the population continued to speak their local dialect only. There was, however, some social class-bound variation in dialect use, the elite speaking a type of regional koine, i.e. varieties of the dialect from which the areally limited, and especially the local, dialect features were (tendentially or completely) removed. This delocalized, cultivated type of dialect has meanwhile been supplanted by Italian — insofar as anything like a national Italian standard language is in common oral usage (cf. Berruto 1989: 8–9, Trumper 1989: 31–35 and the references in Sgroi 1989: 120–21). According to Sobrero (1996: 107–108), nowadays koine formation in Italy is more complicated. The author distinguishes three types of koineization, an active one, concerning the spread of the koine of ‘‘a strong urban centre into the neighbouring territory’’ (e.g. Milanese and Neapolitan), a passive one, which levels out dialectal differences under the pressure of the standard

Koineization and creole genesis 

language, and a third type, which concerns ‘‘the reinforcement and expansion’’ of transition zone dialects. This last type ‘‘can be compared to the ‘passive’ one.’’ While the social basis of active, urban koineization can be characterized as bourgeois, the third type is proletarian in nature, according to Sobrero. In Siegel’s conception, the life-cycle of a koine may look as follows.10 At the start of koineization, there will be a period of instability, characterized by heavy mixture, relative formal simplicity due to levelling, and an almost complete lack of formal consistency. Siegel refers to this unfixed stage as pre-koine. Little by little, sets of lexical, phonological and grammatical norms may develop on the basis of the varieties in contact, with grammar, especially morphology, undergoing reduction. The result is referred to by Siegel as a stabilized koine. Once this stage has been reached, levelling is completed. Subsequently the use of this stabilized koine may be extended to other domains besides the contact situations. This will probably trigger elaboration, both grammatically and stylistically, resulting in an expanded koine. Once this expanded koine acquires native speakers, there will be further linguistic expansion; Siegel labels the outcome a nativized koine. ‘‘[I]n gradual koineization, the least common variants go out of use and in nativization the most common variants are the ones acquired by the children’’ (Siegel, this volume), so nativization plays a key role not only in the elaboration of a koine, but also in the selection of dialect features in the levelling process; more on this aspect in Section 6 below. Fiji Hindi is a nativized koine. This developmental continuum parallels that of pidgins and creole languages; Siegel’s terminology is analoguous to that coined by Peter Mühlhäusler (1980: 32, 1981: 37) for the evolution of creole languages: jargon > stabilized pidgin > expanded pidgin > creole. The nativization of a koine can occur after any of the first three phases — just as jargons and stabilized pidgins can be creolized.11 If a pre-koine (or a jargon) nativizes, stabilization and expansion will be part of the nativization. The similarities in the life-cycles of koines and creole languages do not stop here. ‘‘To a very limited extent, the reintroduction of Standard Hindi/ Urdu has affected Overseas Hindi so that there has been wide variation forming a ‘post-koine continuum’’’ (Siegel 1990: 94). Standard Hindi or Urdu have had stabilizing effects on certain features already in Fiji Hindi, and helped eliminating residual dialectal forms. This elimination process had the effect that several dialect features are now generally considered ‘‘old’’, ‘‘archaic’’ or even ‘‘rustic’’ and are going out of use. Examples are the -wa¯ definite forms of nouns, the -it imperfective suffix, and the -wa¯ perfective suffix (see Siegel 1987: 208–209. Cf. 1988b: 135, 1988b: 131 and 1993b: 114

 Frans Hinskens

respectively). At the same time, Standard Hindi/Urdu have helped to eliminate specific originally Pidgin Hindustani features of Fiji Hindi, such as the generic nonimperative use of the -o verb forms (Siegel 1987: 209; 1992: 101, 107–108). But dekoineization pressures are not just exerted by Standard Hindi/Urdu. A large part of the Fiji Hindi lexicon comes from English.12 The English loan words have been phonologically and morphologically adapted. Insofar as there were already Hindi words available, many of them have been ousted by their English equivalents. Interestingly, ‘‘several English loan words have two forms — an older one from when the word was first borrowed, probably during the indenture era, and a newer one, closer to English in pronunciation’’ (1992: 107). Examples include older sa — newer Set, ‘shirt’, older ibil — newer eibal, ‘table’, older buru¯s — newer braS, ‘brush’. Cases such as these appear to indicate a separate dimension in the ‘‘post-koine continuum’’,13 which Siegel (1987: 209–210) mentions, but does not give examples of. An important issue in the study of creole languages concerns the question whether creole continua ‘‘are made up of coexistent systems or are seamless wholes where boundaries between systems are impossible to establish’’ (Winford 1997b: 309). The latter view has led to the development of the polytectal grammar model in the early days of creole studies, the former — and more recent — one explains variation in creole languages as a product of ‘‘the interaction of autonomous though not completely discrete grammars’’ (Winford 1997b: 310; Hinskens 1998: 174–177). Is there an interesting parallel to this question in the present-day Fiji Hindi koine situation? As much of Winford’s (as well as much of Rickford’s) work demonstrates, quantitative methods can be appropriate to answer questions regarding the nature of creole continua — and koine continua, it would seem.

. Koineization versus creole genesis In the second half of the previous section, some of Jeff Siegel’s insights into the emergence and development of koines were discussed. On the face of it, these could give rise to the assumption that the life-cycles of koines on the one hand and those of creole languages on the other are highly similar or even identical. And Siegel’s terminology definitely suggests such similarities. However, although (a) the outcomes of the processes of koineization and pidginization are alike (Siegel 1988b: 146) and (b) the two types of developments can both

Koineization and creole genesis 

apply in the same situation (1988b: 122), the author is reserved when it comes to equating the two. As a matter of fact, here and in earlier work Siegel very explicitly points out a number of differences between koine formation on the one hand and pidginization and creole genesis on the other. These differences can be summarized in three points. First, whereas pidginization and creolization result from situations of limited social interaction between the speakers of different languages in contact (not in the last place limited access of the substrate languages’ speakers to the superstrate language), koineization presupposes free and close interaction between speakers of different language varieties. As a result, in situations of pidginization and creolization, speakers need to create and acquire a new language for intergroup contact, whereas in situations of koineization, speakers can go on using their own dialects for intergroup contact. Second, pidginization and creolization are generally thought of as relatively rapid and sudden processes, whereas koineization is usually a gradual, continuous process which takes place during a long period of sustained contact.14 Third, as a consequence of the differences in the genetic relatedness between the linguistic systems involved in the contact, the social contexts of the contact, and the duration of the catalysis, compared with their source languages, pidgins show a much higher degree of formal simplification and restructuring than koines. Unlike koines, pidgins and creole languages are structurally discontinuous from and consequently usually not mutually intelligible with their source languages (cf. Siegel 1988b: 147, 1993b: 119). As Muysken and Smith (1995: 5–6) put it, in the coming into existence of koines, ‘‘no linguistic or social violence is involved.’’ In sum, to a large extent the differences between pidginization and creole genesis on the one hand and koineization on the other are due to differences in the values of a small number of language-related, social and demographic variables (cf. Hymes 1971: 83), in other words differences in the quantity of the ingredients that make up the cocktails — though essentially the ingredients themselves do not differ. In connection with the third difference, Siegel (this volume — Section 3) adds that koines, in contrast to pidgins and creoles, rarely undergo innovations or rule change. ‘‘In koines, there is mostly mixing of forms from different varieties in contact, not mixing of different rules.’’ Now does this imply that rule mixing does occur in creolization? If so, then the question is whether the author is thinking of the mixing of corresponding individual rules of the languages involved in the contact, or rather of the mixing of rule sets. The first type of mixing does not seem very likely.

 Frans Hinskens

Afrikaans developed out of processes of koineization of several different Dutch dialects, but (according to e.g. Muysken and Smith 1995: 5; Den Besten, Muysken and Smith 1995: 93; Mühlhäusler 1997: 7) it also has ‘‘creoloid’’ characteristics. To the extent to which the diachronic study of Afrikaans is feasible, this language then seems to provide us with unique possibilities for testing Siegel’s claims regarding the essential differences between koineization and creolization. Siegel’s considerations summarized above are perfectly valid from the point of view of any ‘‘catastrophe’’ theory of creole genesis, i.e. any theory in which abruptness and rapid nativization play a major role. However, how essential would these differences, and especially the second one, be from the point of view of gradualist models?15 In gradualist models, creolization is not necessarily an instantaneous process, but rather a gradual development extending over several generations of speakers. Siegel points out that (2) [i]n radical or abrupt creolization [. . .] linguistic restructuring [. . .] widespread nativization and the emergence of a creole all happen simultaneously. However, the viewpoint presented here suggests that substantial contact-induced restructuring occurs before the social conditions are right for the emergence of a new stable contact variety.16 The following questions arise: (3) a. is this suggestion intended as a principle or a possibility? b. to which extent is the suggested scenario specific to the development of immigrant koines — to the exclusion of regional koines? c. to which extent is this idea in line with the theory that certain Atlantic creole languages originate from pidgins that had already developed in West Africa? More in particular with monogenesis/relexification theories of the development of Atlantic creoles out of hypothetical systems such as the one that used to be referred to as West African Pidgin Portuguese?17 The alternative scenario for creole genesis suggested by Siegel seems to make much room for imperfect second language learning of the superstrate language.18 But at this point the scenario needs further elaboration and robust empirical support. Subsequently, Siegel observes that nativization by children is commonly seen as the crucial step both in abrupt and in gradual creolization. In his scenario for koineization, however, it can only accelerate the levelling and stabiliza-

Koineization and creole genesis

tion processes. The linguistic input of children born in a new community consists of their parents’ first language and the developing contact language. Their input ‘‘includes the variants in the mixture which have not been levelled out at an earlier stage and, again, they acquire some features but not others.’’ Clearly, insights into the nature of the filter between linguistic input and intake in language acquisition19 and linguistic socialization could help reach a deeper understanding of this aspect of the mechanisms involved in koineization. Most of the sifting of dialect features in processes of levelling seems to take place during language acquisition and linguistic socialization. In the conclusions to his contribution to this volume, Siegel presents a figure in which some of his insights and proposals are visually represented in their mutual relationships. One of the main ideas is that there may be a continuum ranging from koines on the one hand to creole languages on the other, with indigenized varieties and semi-creole languages (neither of which were discussed in Siegel’s contribution) somewhere in between, and with the degree of restructuring/transfer gradually increasing from koines over indigenized varieties and semi-creoles to creoles. Several questions arise: (4) a. where in this scheme should pidgins be placed? Higher up in the scheme — somewhere in the intersection of source varieties and restructuring? In the top-right corner only? b. where would mixed languages (Bakker and Mous 1995) fit in — if at all? c. would immigrant koines, which according to Siegel (a) frequently serve as the dominant language in the immigrant communities, and (b) may develop more rapidly than regional koines, be placed somewhat nearer to the creole pole of the continuum?

.

General discussion

The author should be thanked among other things for clearly fencing off the notion of koine from related notions such as pidgin and creole, and sketching the contours of a sociolinguistic model in which basic similarities, differences of degree as well as essential differences (e.g. the role of nativization) between koineization and creole genesis are insightfully ordered relative to each other. In the present contribution, Siegel develops the outlines of his model mainly on the basis of:



 Frans Hinskens

(5) a. some of his findings for Fiji Hindi (briefly discussed in Section 1 of his contribution), b. the theory proposed in Trudgill 1986 (sketched in Sections 1 and 2 of Siegel’s contribution), c. replications to criticism of earlier work by Siegel articulated in Mufwene 1997 (in Section 3), and, d. some ‘‘discussion’’ with Bickerton (in Section 4). But how empirical is the model Siegel outlines here? This is difficult to asses since the author presents only few data to support his insights (but cf. Section 3 above), and no hard, testable hypotheses are derived. In connection with the former point of criticism: it’s hard to see exactly which role the excursus on modern Israeli Hebrew (in Section 4 of Siegel’s contribution) plays in the line of reasoning. Modern Israeli Hebrew can probably not be considered as a genuine product of koineization, since the features from the original pool of variants which ‘‘survived’’ the levelling processes were Slavic ‘‘substrate features, not dialect features’’, as the author points out. These substrate features stem from ‘‘the involvement of second language learning’’ early in the development of the language. But after this stage, when the various spoken second language versions of Hebrew became used for everyday communication among immigrants in Palestine, ‘‘the situation was similar to what we have seen before with koineization.’’ So it seems that in Siegel’s model, Modern Hebrew, which is ‘‘a new language, not a new dialect’’, would thus rather be placed somewhere inbetween (immigrant) koines and creoles. Another reason why it is hard to assess the argumentative weight of the case of Modern Hebrew is that no data are presented for this language at all. Therefore it is also difficult to compare Siegel’s interpretation of (Blanc’s account of) the development of Modern Hebrew with the one which was recently proposed by Wexler (1995). According to Wexler, Modern Hebrew arose through the relexification of Yiddish (which he analyzes grammatically as Sorbian, and more generally Slavic) with vocabulary from Classical, i.e. Biblical Hebrew. So apparently Blanc/Siegel and Wexler agree that pidginization and subsequent creolization did not occur in the history of Modern Hebrew. But unlike Blanc/Siegel, Wexler thinks that Modern Hebrew developed very rapidly, probably even in the course of one generation. Koineization is one of the roads which can lead to the development of what is usually referred to as a standard language. For the sake of a solid empirical underpinning of the model Siegel is in the course of developing, it

Koineization and creole genesis 

might be rewarding to study the paradigm cases of koine formation, i.e. he¯ koiné diálektos of Attic/Ionic dialects of Hellenistic Greek (Hock 1991: 485– 488; Joseph 1992: 87; Bubenik 1993) and the ancestral Arabic koine, which developed in the early days of the spread of Islam, from which all of the modern spoken Arabic dialects eventually emerged (Ferguson 1959) — insofar as indepth studies of these dead koines are at all feasible. Apart from these or any of the 35 other languages listed in Siegel 1985 (p. 359) which have been called koines, research might focus on Modern Persian and contemporary Standard Igbo, which have equally been said to have evolved out of koines (by e.g. Windfuhr 1992: 184, and Williamson and Nolue Emenanjo 1992: 196, respectively). For Jeff Siegel the most obvious next step would be to test and further develop his model on the basis of data from Fiji Hindi. For the testing of certain aspects of the model, quantitative sociolinguistic methods seem to be called for (cf. the final paragraphs of Sections 4 and 5 above). Apart from the question how the model should be further investigated, also the question what can and should be studied ought to be considered. With one exception (1988b: 127), the part of Siegel’s work that has been cited here invariably deals with lexical items and, more so, grammatical traits; in this respect, his orientation is very similar to that of the vast majority of pidgin and creole studies (cf. Singh and Muysken 1995). But how about the phonology?

Notes * The realization of this chapter has been made possible by a Fellowship of the Royal Netherlands Academy of Arts and Sciences. . On convergence see the contributions by Kouwenberg and Arends to this volume. . A Western Hindi dialect (cf. Siegel 1988b: 123). . Examples from dialects of other languages can be found in e.g. Samuels (1972: 98–100), Chambers and Trudgill (1980: 132–142), and Hinskens (1992: §12.2.1). . ‘‘[S]ometimes called Fiji Hindustani’’ (Siegel 1992: 94). . Or Basilectal Hindustani (cf. Siegel 1988a: 12, 1992: 94), ‘‘the basilectal form of the Indian lingua franca’’, which may have been pidginized (1988b: 122). To the present author it is not clear whether the pidginization is supposed to have occurred in India or at Fiji and how the product relates to the ‘‘pidginized form of plantation Hindustani’’ which developed in Fiji (1988b: 126). . The number of speakers exerts only a tendential, though highly probable, effect. As Siegel puts it, ‘‘[t]he size of the group is an ambivalent factor.’’ There are exceptions in two

 Frans Hinskens

directions: in terms of numbers of speakers, ‘‘the country with the smallest Indian population, Suriname, has one of the thriving varieties of Overseas Hindi’’, namely Sarnami, whereas ‘‘in two other countries where the Overseas Hindi is dying, Guyana and Trinidad, the proportion of Indians is over 50 percent.’’ (1990: 100; cf. 1993b: 115–116). The anonymous reviewer points out that the position of Sarnami is due to ‘‘‘the ghetto effect’, reinforced especially by racial attitudes’’, while the majority situation in Guyana and Trinidad leads the — upwardly socially mobile — Indians to stick to the prevailing community vernacular. . In the final paragraphs of Section 3 of his contribution, Siegel presents an admirably brief overview of some of the factors which have been studied in this context. . See Hinskens, Van Hout and Wetzels (1997: 21–22) for considerations regarding the disentanglement of internal and external factors in the study of language variation and change. Hinskens and Van Rossem (1995: 66–67) reflect on the methodology of creole linguistics and, more in particular, the need for and feasibility of quantitative approaches. . The vast majority of the South Indians, who started to arrive in Fiji after 1903, did not speak any variety of Hindi/Hindustani (Siegel 1988b: 126). . What follows is a summary of Siegel (1985: 373–375, 1986: 456, 1988b: 147–148, and 1993a: 6–7). . Cf. Hymes’ (1971: 84) definition of creolization as ‘‘that complex process of sociolinguistic change comprising expansion in inner form, with convergence, in the context of extension in use’’. Compare Siegel (1987: 16, 17, 201–203). The conviction that nativization plays a crucial role in creolization fails to account for situations such as those in West Africa or Papua New Guinea, ‘‘where there is no clear structural difference between first and second language varieties of the same contact vernacular’’ (Winford 1997a: 2). . Many examples are given in Siegel (1992: 103–107). . The present author alone is responsible for the interpretation of this variation in the phonetic realization of English loan words proposed here. . Cf. Samuels’ (1972: 92) distinction between ‘‘two main types of contact: Type A: stable and continuous contact between neighbouring systems that are adjacent on either the horizontal (regional) or the vertical (social) axis; Type B: sudden contact, resulting from invasion, migration or other population-shift, of systems not normally in contact hitherto.’’ Needless to add that in Siegel’s view koines develop in the A Type, and pidgins and creoles in the B Type contact situation. . E.g. as represented in the work of Arends — cf. Arends (1993), Arends and Bruyn (1995). . With respect to the six main varieties of Overseas Hindi (spoken in Mauritius, Guyana, Trinidad, South Africa, Surinam and Fiji), Siegel observes that the ‘‘important question [. . .] where did koineization take place — in India or Overseas?’’ has been almost ignored (1988a: 15). . Cf. Den Besten, Muysken and Smith (1995: 87–89), and Arends, Muysken and Smith 1995: 320, for references and discussion. . For some references to literature regarding the imperfect L2 learning hypothesis of creole genesis, cf. Den Besten, Muysken and Smith (1995: 97), and Arends, Muysken and Smith (1995: 320).

Koineization and creole genesis

. See Johnson and Johnson (1998: 173) for a brief definition of and some references regarding the notion of ‘intake’.

References Arends, Jacques. 1993. ‘‘Towards a gradualist model of creolization’’. In Atlantic Meets Pacific, Francis Byrne and John Holm (eds.), 371–380. Amsterdam and Philadelphia: John Benjamins. Arends, Jacques and Adrienne Bruyn. 1995. ‘‘Gradualist and developmental hypotheses’’. In Jacques Arends, Pieter Muysken and Norval Smith (eds.), 111–120. Arends, Jacques, Silvia Kouwenberg and Norval Smith. 1995. ‘‘Theories focusing on the nonEuropean input’’. In Jacques Arends, Pieter Muysken and Norval Smith (eds.), 99–109. Arends, Jacques, Pieter Muysken and Norval Smith. 1995. ‘‘Conclusions’’. In Jacques Arends, Pieter Muysken and Norval Smith (eds.), 319–330. Arends, Jacques, Pieter Muysken and Norval Smith (eds.). 1995. Pidgins and Creoles: An introduction. Amsterdam and Philadelphia: Johns Benjamins. Auer, Peter. 1994. ‘‘Führt Dialektabbau zur Stärkung oder Schwächung der Standardvarietät? Zwei phonologische Fallstudien’’. In Standardisierung und Destandardisierung europäischer Nationalsprachen, Klaus Mattheier and Edgar Radtke (eds.), 129–162, Frankfurt, etc: Lang. Auer, Peter and Frans Hinskens. 1996. ‘‘The convergence and divergence of dialects in Europe. New and not so new developments in an old area’’. In Peter Auer, Frans Hinskens and Klaus Mattheier (eds.), 1–30. Auer, Peter, Frans Hinskens and Klaus Mattheier (eds.). 1996. Convergence and Divergence of Dialects in Europe. (Sociolinguistica 1996.) Bakker, Peter and Maarten Mous (eds.). 1994. Mixed Languages. Fifteen case studies in language intertwining. Amsterdam: IFOTT. Barz, Richard and Jeff Siegel (eds.). 1988. Language Transplanted. The development of Overseas Hindi. Wiesbaden: Harrassowitz. Den Besten, Hans, Pieter Muysken and Norval Smith. 1995. ‘‘Theories focusing on the European input’’. In Jacques Arends, Pieter Muysken and Norval Smith (eds.), 87–98. Berruto, Gaetano. 1989. ‘‘Main topics and findings in Italian sociolinguistics’’. In Elisabetta Zuanelli Sonino (ed.), 7–30. Bickerton, Derek. 1984. ‘‘The language bioprogram hypothesis’’. The Behavorial and Brain Sciences 7: 173–188. Bright, William (ed.). 1992. International Encyclopedia of Linguistics. New York: Oxford University Press. Bubenik, Vit. 1993. ‘‘Dialect contact and koineization: The case of Hellenistic Greek’’. In Jeff Siegel (ed.), 9–23. Chambers, Jack and Peter Trudgill. 1980. Dialectology. Cambridge: Cambridge University Press. Cortelazzo, Manlio. 1993. ‘‘Koinés dialectales italiennes après le XVIe siècle’’, followed by ‘‘Discussion’’. In Écriture, langues communes et normes. Formation spontanée de koinès



 Frans Hinskens

et standardisation dans la Galloromania et son voisinage, Pierre Knecht and Zygmunt Marzys (eds.), 227–234. Neuchâtel etc.: Faculté des Lettres, Université de Neuchâtel; Librairie Droz. Dillard, J. 1972. Black English: Its history and usage in the United States. New York: Random House. Domingue, Nicole. 1981. ‘‘Internal change in a transplanted language’’. Studies in the linguistic sciences 4.2: 151–159. Ferguson, Charles. 1959. ‘‘The Arabic koine’’. Language 35.4: 616–630. Gambhir, Surendra. 1988. ‘‘Structural developments in Guyanese Bhojpuri’’. In Richard Barz and Jeff Siegel (eds.), 69–94. Hinskens, Frans. 1992. Dialect levelling in Limburg. Structural and sociolinguistic aspects. Ph.D. dissertation, University of Nijmegen. [A revised and abridged version has been published under the same title by Niemeyer, Tübingen in 1996.] Hinskens, Frans. 1998. ‘‘Variation studies in dialectology and three types of sound change’’. In Linguistics of Variation, Sociolinguistica 12, U. Ammon (ed.), 155–193. Hinskens, Frans, Roeland van Hout and W. Leo Wetzels. 1997. ‘‘Balancing data and theory in the study of phonological variation and change’’. In Variation, change and phonological theory, Frans Hinskens, Roeland van Hout and W. Leo Wetzels (eds.), 1–33. Amsterdam and Philadelphia: John Benjamins. Hinskens, Frans and Cefas van Rossem. 1995. ‘‘The Negerhollands word sender in eighteenth century manuscripts’’. In The early stages of creolization, Jacques Arends (ed.), 63–88. Amsterdam and Philadelphia: John Benjamins. Hock, Hans Henrich. 1991. Principles of historical linguistics. Berlin and New York: Mouton de Gruyter. [Second edition, revised and updated.] Holmquist, Jonathan. 1988. Language Loyalty and Linguistic Variation. A study in Spanish Cantabria. Dordrecht: Foris Publications. Hymes, Dell. 1971. ‘‘Introduction to section III’’. In Pidginization and creolization of languages, Dell Hymes (ed.), 65–90. Cambridge: Cambridge University Press. Johnson, Keith and Helen Johnson (eds.). 1998. Encyclopedic Dictionary of Applied Linguistics. A handbook for language teaching. Oxford: Blackwell. Joseph, Brian, 1992. ‘‘Greek’’. In William Bright (ed.), vol. 2, 86–93. Kouwenberg, Silvia. 1994. A Grammar of Berbice Dutch Creole. Berlin and New York: Mouton de Gruyter. Le Page, Robert and Andrée Tabouret-Keller. 1985. Acts of Identity: Creole-based approaches to language and identity. Cambridge: Cambridge University Press Markey, Thomas. 1986. ‘‘When minor is minor and major is major: Language expansion, contraction and death’’. Paper delivered at the Third International Conference on Minority Languages, University College, Galway, Eire, 21–26 June, 1986. Mufwene, Salikoko. 1997. ‘‘Jargons, pidgins, creoles, and koinés: What are they?’’. In Arthur Spears and Donald Winford (eds.) Mühlhäusler, Peter. 1980. ‘‘Structural expansion and the process of creolization’’. In Theoretical orientations in creole studies, Albert Valdman and Arnold Highfield (eds.), 19–55. New York: Academic Press.

Koineization and creole genesis 

Mühlhäusler, Peter. 1981. ‘‘The development of the category of number in Tok Pisin’’. In Generative studies on creole languages, Pieter Muysken (ed.), 35–84. Dordrecht: Foris Publications. Mühlhäusler, Peter. 1997. Pidgin and Creole Linguistics. London: University of Westminster Press. (Expanded and revised edition.) Muysken, Pieter. 1988. ‘‘Are creoles a special type of language?’’. In Linguistics; the Cambridge survey, vol. 2, Frederick Newmeyer (ed.), 285–301. Cambridge: Cambridge University Press. Muysken, Pieter and Norval Smith. 1995. ‘‘The study of pidgin and creole languages’’. In Jacques Arends, Pieter Muysken and Norval Smith (eds.), 3–14. Samuels, Michael. 1972. Linguistic Evolution. With special reference to English. Cambridge: Cambridge University Press. Scholtmeijer, Harrie. 1992. Het Nederlands van de IJsselmeerpolders. Ph.D. dissertation, University of Leiden. Sgroi, Salvatore. 1989. ‘‘A selected critical bibliography of Italian sociolinguistics’’. In Elisabetta Zuanelli Sonino (ed.), 109–166. Siegel, Jeff. 1985. ‘‘Koines and koineization’’. Language in Society 14, 357–378. Siegel, Jeff. 1986. ‘‘Wai: A Malaitan language in Fiji’’. In FOCAL II: Papers from the Fourth International Conference on Austronesian Linguistics, Paul Geraghty, Lois Carrington and Stephen Wurm (eds.), 435–463. (Pacific Linguistics, C-94.) Siegel, Jeff. 1987. Language Contact in a Plantation Environment: A sociolinguistic history of Fiji. Cambridge: Cambridge University Press. Siegel, Jeff. 1988a. ‘‘Introduction’’. In Richard Barz and Jeff Siegel (eds.), 1–22. Siegel, Jeff. 1988b. ‘‘The development of Fiji Hindustani’’. In Richard Barz and Jeff Siegel (eds.), 121–149. Siegel, Jeff. 1990. ‘‘Language maintenance of Overseas Hindi’’. In Learning, keeping and using language: Selected papers from the Eighth World Congress of Applied Linguistics. Sydney, 16–21 August 1987, Michael Halliday, John Gibbons and Howard Nicholas (eds.), 91–113. Amsterdam and Philadelphia: John Benjamins. Siegel, Jeff. 1992. ‘‘Language change and culture change among Fiji Indians’’. In Culture Change, Language Change — case studies from Melanesia, Tom Dutton (ed.), 91–113.. (Pacific Linguistics, C-120.) Siegel, Jeff. 1993a. ‘‘Introduction: Controversies in the study of koines and koineization’’. In Jeff Siegel (ed.), 5–8. Siegel, Jeff. 1993b. ‘‘Dialect contact and koineization’’. In Jeff Siegel (ed.), 105–121 Siegel, Jeff, (ed.). 1993. Koines and koineization (International Journal of the Sociology of Language, 99). Singh, Rajendra and Pieter Muysken. 1995. ‘‘Column. Wanted: A debate in pidgin/creole phonology’’. Journal of pidgin and creole languages 10.1: 157–169. Sobrero, Alberto. 1996. ‘‘Italianization and variations in the repertoire: The koinaí’’. In Peter Auer, Frans Hinskens and Klaus Mattheier (eds.), 105–111. Spears, Arthur and Donald Winford (eds.). 1997. The Structure and Status of Pidgins and Creoles. Amsterdam and Philadelphia: John Benjamins.

 Frans Hinskens

Thelander, Mats. 1980. De-dialectalisation in Sweden (FUMS rapport 86). Uppsala: Uppsala University. Thelander, Mats. 1982. ‘‘A qualitative approach to the quantitative data of speech variation’’. In Sociolinguistic variation in speech communities, Suzanne Romaine (ed.), 65–83. London: Arnold. Trudgill, Peter. 1986. Dialects in Contact. Oxford: Blackwell. Trumper, John. 1989. ‘‘Observations on sociolinguistic behavior in two Italian regions’’. In Elisabetta Zuanelli Sonino (ed.), 31–62. Wexler, Paul. 1995. Relexification in creole and non-creole languages: Prolegomena to a research. Paper presented in the OSU Linguistics Speakers Series. Linguistics Department, Ohio State University, Columbus, October 19, 1995. Williamson, Kay and E. Nolue Amenanjo. 1992. ‘‘Igbo’’. In William Bright (ed.), vol. 2, 195–199. Windfuhr, Gernot. 1992. ‘‘Persian’’. In William Bright (ed.), vol. 3, 183–188. Winford, Donald. 1997a. ‘‘Introduction: On the structure and status of pidgins and creoles’’. In Arthur Spears and Donald Winford (eds.), 1–31. Winford, Donald. 1997b. ‘‘Column. Creole studies and sociolinguistics’’. Journal of Pidgin and Creole Languages 12: 303–318. Zuanelli Sonino, Elisabetta (ed.). 1989. Italian Sociolinguistics: Trends and issues (International Journal of the Sociology of Language, 76.)

Chapter 9

Convergence and explanations in creole genesis Silvia Kouwenberg → .

Introduction

My aim is to evaluate the potential relevance of the notion of convergence for explanations of creole genesis. This will be done against the background of its conventional use in historical linguistics, which is the subject of the first section below. In Section 2.1, we will note that there is no coherent theory of convergence in historical linguistics. Nor is there a unified class of phenomena which go under the label of convergence, as will become evident in Section 2.2. Obviously, this lack of clarity seriously detracts from its explanatory potential, and leads one to wonder what, if anything, it could contribute to explanations outside its original jurisdiction. On the other hand, some convergence studies note its potential for explanations of creole genesis, and the notion of convergence is now rapidly gaining in popularity among creole linguists. This cannot be explained from renewed interest in the relevance of insights from historical linguistics for creole history alone, since, as we shall see in Section 3, its use in the creole literature is markedly different from that in historical linguistics publications, despite some superficial similarities. Its popularity may have more to do with the present deadlock in creole linguistics, where models for which superior explanatory power has been claimed appear to have reached their limits. It turns out that convergence is used by creolists wishing to take a compromise position between substrate, superstrate, universalist and other possible explanations in creole genesis by simply allowing a combination of these. As such, its application is ad hoc and not particularly revealing, concealing the need for creolists to go back to the drawing-board.

 Silvia Kouwenberg

. Convergence in historical linguistics The received view of convergence in historical linguistics can be characterized somewhat loosely as referring to change in languages in contact where this stems from mutual accomodation, and where it leads to increasing equivalence of structure or intertranslatability. It is thought to take place through borrowing and/or diffusion of features. It accounts for an observed correlation between geographical distance and linguistic distance. For instance, Boas (1929: 6) noted that ‘‘in a considerable number of native languages of the North Pacific Coast we find, notwithstanding fundamental differences in structure and vocabulary, similarities in particular grammatical features distributed in such a way that neighbouring languages show striking similarities’’. The fundamental differences in structure and vocabulary rule out (demonstrable) genetic relationships. Convergence by diffusion of properties over contiguous areas is then assumed to account for the observed similarities in particular grammatical features. In the following, we will examine the implications of this characterization of convergence in Section 2.1, to turn to a consideration of some of the phenomena which have been claimed to represent cases of convergence in Section 2.2. . The theory .. Convergence and bilingualism As a contact-induced phenomenon, convergence is intimately connected with bilingualism. At least since Weinreich (1953), the bilingual individual is seen as ‘‘the ultimate locus of contact’’ (6), and, as he puts it, ‘‘contact breeds imitiation and imitation breeds linguistic convergence’’ (viii). The assumption in much of the literature is, then, that convergence originates in the linguistic behaviour of bilinguals. Thus, Hock (1986: 492 f.) postulates a scenario in which convergence stems from the development of interlanguage varieties of the languages in contact (say, languages A and B) by bilingual individuals, which, due to continued interaction, display a build-up of increasingly mixed features until finally, the interlanguage varieties of A and B are structurally completely mixed (A-B). It is obvious that for this scenario to work, a bilingual who attains less-than-perfect command of the other language is required. But exactly what level of bilingual competence will do the trick? And is it a prerequisite that bilingualism is widespread? Or that the speech of bilinguals is normative for the wider community? For instance, Hock asserts that for

Convergence and explanations in creole genesis 

convergence to take place in the manner described, all speakers of the languages involved must be bilingual (493), whereas Foley (1986: 25) claims that convergence in the cases which he has investigated does not require extensive bilingualism, merely enough bilingual speakers to innovate and propagate the adoptions, and Gumperz and Wilson (1971) deal with a sociolinguistic context in which monolingual home groups are vital for the continued existence of language diversity, which provides the linguistic context for convergence. In addition to the question of the role of these sociolinguistic variables, a major problem with the presumed central role of bilingualism is in the lack of convergence studies which refer to insights from second language acquisition studies. With the linguistic behaviour of bilinguals as the presumed source of convergence, the mechanisms whereby interlanguage varieties are created and the properties of these varieties would seem to be of crucial interest for students of convergence. The central role of bilingualism in this process remains a stipulation until it can be shown that interlanguage varieties have the properties which arise in convergence cases. Interlanguage varieties are notorious for having properties which are neither target language properties, nor first language properties (see, for instance, the discussion in Appel and Muysken 1987: 82–92), whereas convergence studies assume that convergence has a model in at least one of the languages in contact. Such issues are simply not addressed in the convergence literature. As we shall see below, studies of convergence tend to be concerned with cataloging the similarities found in a particular area, not with establishing the mechanisms by which these came about. .. Directionality Convergence is typically but not exclusively regarded as a bidirectional or multidirectional process. For instance, McMahon (1994: 213) distinguishes convergence from borrowing, the former bidirectional, the latter unidirectional. In most instances of convergence then, we should expect to find structural borrowings in language A from language B alongside structural borrowings in language B from language A. But in the discussion in Section 3.2 we shall also see cases of unidirectional change, and Hock (1986: 492), for instance, distinguishes between mutual and unidirectional convergence. The term diffusion (of features) — which would appear to refer to a unidirectional process — is frequently encountered in relation to and instead of convergence (as in Boas 1929; Campbell et al. 1986; Dryer 1989; Emeneau 1956), as is the notion of wave theory (Wellen-hypothese), again with seemingly unidirectional implications (Jakobson 1931; Jeffers and Lehiste 1979; Bynon 1977). We should note,

 Silvia Kouwenberg

however, that individual changes are always assumed to be unidirectional, i.e. in the direction of one of the available languages. True bidirectionality, where resultant properties are not ascribable to any individual source language, but rather to the interaction between the languages in contact, is not what is meant when convergence studies speak of mutual or bidirectional influence. For instance, Gumperz and Wilson, in their (1971) case study of the Kupwar varieties of Urdu, Kannada and Marathi, identify changes in all possible — and sometimes unexpected — directions (see below). Much of the literature is concerned with establishing the ultimate source of convergent features. As we shall see, however, it is not always obvious that a shared property has its source in any of the contact languages. A related assumption is that convergence makes use of what languages have in common. Civ ¹jan (1965: 15), quoted in Joseph (1983: 200), claims that convergence moves towards ‘‘an eventual model, including the most important structural similarities and creating as a result the optimal conditions for the easiest possible conversion from language to language’’. Weinreich (1953), like Civ’jan, sees convergence as solving communication problems, and in his claim that ‘‘[i]f the phonic or grammatical systems of two languages are compared and their differences delineated, one ordinarily has a list of the potential forms of interference in the given contact situation’’ (3), there is a strong suggestion that where there are no differences, there will be no interference (in the speech of bilinguals), hence no convergence (recall that interference in the speech of bilinguals is seen as the source of convergence). We should therefore expect to see convergent change only where there is disagreement between languages in contact. However, it is not at all evident that convergence is restricted in this manner. Thus, the celebrated case of infinitive loss in the Balkans involves the loss of what was essentially a shared property of the languages in contact. Also, we shall note several instances in the Kupwar languages where two out of the three languages in contact share a particular property, yet appear to have changed towards the minority model provided by the third language. What determines the direction of change is, then, entirely unclear. Furthermore, there is the issue of the level of analysis at which structural similarity should be established. Masica (1976) asserts that ‘‘[i]t is surface order that seems to pattern areally, not ‘deep’ order’’ (14), but surface and deep order remain murky notions as there are no references to any particular model of grammar. Gumperz and Wilson (1971) treat as equally important the loss of gender marking in favour of person agreement on the past auxiliary in Kupwar Urdu — apparently under the influence of Kannada and Marathi —

Convergence and explanations in creole genesis 

and the wholesale replacement of grammatical gender by semantically determined gender in Kupwar Marathi and Urdu — under the influence of Kannada, while change in word order in Kupwar Urdu is only mentioned in passing and not counted at all. Their concern is with convergent tendencies in the surface expression of morphological categories, rather than with such tendencies in relation to structural properties. .. A typology of relationships between languages The term areal linguistics has become popularized as refering to studies of the results of diffusion of structural features across linguistic boundaries (see for instance Campbell et al. 1986: 530). Trubetzkoy’s proposition 16, proposed and accepted at the First Linguistics Conference, held in The Hague in 1928, gave formal recognition to this phenomenon by distinguishing language groups based on genetic relationships from those based on typological similarities: Proposition 16. Jede Gesamtheit von Sprachen, die miteinander durch eine erhebliche Zahl von systematischen Übereinstimmungen verbunden sind, nennen wir Sprachgruppe. Unter den Sprachgruppen sind zwei Typen zu unterscheiden: Gruppen, bestehend aus Sprachen, die eine grosse Ähnlichkeit in syntaktischer Hinsicht, eine Ähnlichkeit in den Grundsätzen de morphologischen Baus aufweisen, und eine grosse Anzahl gemeinsamer Kulturwörter bieten, manchmal auch äussere Ähnlichkeit im Bestande der Lautsysteme, -dabei aber keine systematische Lautentsprechungen, keine Übereinstimmung in der lautlichen Gestalt der morphologischen Elemente und keine gemeinsamen Elementarwörter besitzen,solche Sprachgruppen nennen wir Sprachbunde. Gruppen, bestehend aus Sprachen, die eine beträchtliche Anzahl von gemeinsamen Elementarwörter besitzen, Übereinstimmungen im lautlichen Ausdruck morphologischer Kategorien aufweisen und, vor allem, konstante Lautentsprechungen bieten, -solche Sprachgruppen nennen wir Sprachfamilien.’’ (Trubetzkoy 1928: 18)1

Convergence cannot be understood properly except in relation to Trubetzkoy’s notion of Sprachbund. We can characterize the relation between them as that between a process (convergence) and its result (Sprachbund). Trubetzkoy’s proposition was based on the observation that genetically unrelated but geographically adjacent languages are sometimes characterized by structural correspondence to a degree which is unexpected from the point of view of their genetic distance. The shared features which characterize a Sprachbund are assumed to have spread through the area — rather than having originated

 Silvia Kouwenberg

in every language independently. Well-known cases which have been claimed to constitute such areal groupings include the languages of the Balkans (see any number of publications in areal linguistics), the Indian subcontinent or South Asia (Emeneau 1956; Masica 1976), areal groupings of North American Indian languages (Swadesh 1971) and of Meso-American languages (Campbell et al. 1986), Europe (Hock 1986) and Africa (Hock 1986). In the most extreme case, the linguistic distance between languages in geographic proximity may be reduced to the point of complete intertranslatability, as has been claimed for the language varieties of the Kupwar area (Gumperz and Wilson 1971). Instead of the term Sprachbund, linguistic area is the rather intransparent term most frequently encountered in the literature (for instance in Campbell et al. 1986; Emeneau 1956; Masica 1976); the more litteral translation language alliance has not made it. Henceforth, I will use the terms Sprachbund and linguistic area interchangeably. Not only is the origin of observed correspondences between languages in a Sprachfamilie and a Sprachbund claimed to be different, but in addition, such correspondences are not of the same kind. Thus, while we may encounter similarities in the sound inventories of the languages involved, we ought not to find systematic sound correspondences in a Sprachbund, whereas this is what we should expect to find if languages are genetically related. While words may spread through a Sprachbund, in particular culture-related vocabulary, shared basic vocabulary is unexpected, unlike the situation in a Sprachfamilie. In short, a Sprachfamilie can be said to be primarily characterized by material likeness of words and/or morphemes — note that there is no mention in Trubetzkoy’s proposition of similarities in syntax and morphological structure as necessary characteristics of a Sprachfamilie!, whereas a Sprachbund is primarily characterized by structural similarity and a lack of material resemblances (terminology borrowed from Jakobson 1939). Table 1 offers an idealized representation of types of relationships between languages. The shared features of linguistic areas constitute isoglosses. One of the questions which have figured in the literature is how many isoglosses suffice for a Sprachbund, and whether they should bundle consistently, or whether overlap and diffuse boundaries are admissible. A single isogloss is sufficient for Dryer’s (1989) large linguistic areas, but one should note that his view of what constitutes a linguistic area corresponds to Trubetzkoy’s Sprachgruppe rather than Sprachbund, as it does not preclude the possibility that the diagnostic property is shared due to partly or largely genetic causes (266). In general, authors insist on several isoglosses, but allow for a core versus periphery

Convergence and explanations in creole genesis 

Table 1. A typology of relationships between languages (à la Trubetzkoy) Material likeness

Structural similarity

Type of relationship

+

+

+

−

−

+

−

−

Sprachfamilie: the expected case for languages which are genetically related. Sprachfamilie: structural similarity is lacking perhaps as a result of divergent developments; in any case, it is not a requirement of Trubetzkoy’s proposition 16. Sprachbund: similarities between geographically proximate languages result from convergent developments. No genetic or other relationship.

distinction, such that a linguistic area is more strongly defined by its diagnostic properties in the core, where isoglosses cluster, than in its periphery (e.g. Masica 1976). This approach works well only where the diffusion of features can be portrayed in a wave model of concentric and ever-expanding circles from a single source. In that case, some features have simply spread further than others. The area closer to the source constitutes the core, outlying areas the periphery. Unfortunately, it is not always possible to identify a source for diffused features, and even where this is possible, there is not usually a single source for all features, nor do isoglosses pattern in the way predicted by the core-periphery model. We will consider these issues again when we examine the most famous Sprachbund of all, that of the Balkans. .. The methodology The process of convergence, then, results in structural similarity despite a general lack of material likeness among neighbouring languages. But Bynon (1977: 244) warns that isomorphism of structure is not necessarily diagnostic of convergence, as this depends not on the nature of the shared features but on their relative chronology, i.e. their position in the chain of innovatory changes that constitutes the historical grammar of each of the languages concerned. A similar point is made by Trubetzkoy (1936: 82) where he refers to P. Kretschmer (no reference included), who argued ‘‘dass zwischen Entlehnung und Verwandtschaft nur ein chronologischer Unterschied besteht.’’2 Sebeok (1949) points out that shared properties within Ural-Altaic may point towards genetic relationships as well as areal grouping as a result of convergence. The upshot is that well-documented language histories or the reconstruction of these from comparison with related languages elsewhere are a prerequisite if we are to confidently ascribe similar structural features to convergence. In the

 Silvia Kouwenberg

absence of such histories, we cannot rule out alternative possibilities such as shared inheritance and/or independent development. A related problem is that of the nature of the diagnostic features: the more common these are, crosslinguistically, the more likely it is that properties have arisen from languageinternal developments or were shared to begin with. Dryer’s (1989) methodology has a built-in strength requirement, such that his large linguistic areas are defined only by typologically uncommon traits, but, as pointed out before, he does not rule out the possibility of inheritance. Campbell et al. (1986) propose a strength-scale for diagnostic properties, pointing out that marked traits are more valuable in defining a linguistic area than are unmarked traits, and positing a continuum of linguistic areas from those weakly defined to much stronger areas. Based on a strength-evaluation of several linguistic areas they conclude, for instance, that the Balkans is weakly defined. A major weakness of convergence studies is in the simple fact that convergence can only be diagnosed after the fact, since, as we shall see later, some claimed instances of Sprachbund phenomena seem to have involved changes which have not been modelled on any available pattern in the contact situation. Also, developments in neighbouring languages are rarely simultaneous, as shown for instance for the spread of infinitive loss in the Balkans, which has taken place anywhere between the tenth century (Greek) and the present (Albanian) (Joseph 1983). In other words: the process of convergence can be recognized only by its endresult, which may be separated from its initial state by hundreds if not thousands of years. At the same time, that endresult is ascribed to the convergent process. But this puts us in danger of circular reasoning along the lines of ‘‘we see result X, therefore process Y must have obtained; because process Y was at work, we got result X¹¹ (substitute Sprachbund and convergence for X and Y). Careful reading of the literature reveals widespread circularity of this type. The first step in convergence studies is the observation of areally shared features/correspondences between languages which are not demonstrably genetically related, or if they are, the correspondences cannot be accounted for by the genetic relationships. These languages are then said to form a Sprachbund. This conclusion can be reinforced by showing that the shared properties differ from those expected from languages of their particular genetic affiliation, for instance by comparison with cognate languages which are spoken outside the area. It is then assumed that this Sprachbund developed due to influence which the pertinent languages exerted on each other. In other words, convergence has been at work.

Convergence and explanations in creole genesis 

Campbell et al. (1986: 534) refer to this as the circumstancialist approach and point out that the actual borrowing or diffusion is not demonstrated. The credibility of this approach rests largely on acceptance of an underlying assumption that the existence of languages in (close) contact is an inherently unstable state of affairs, and that bi/multidirectional influence (mediated by bilingualism) is inevitable in such a situation. Moreover, such influence is destined to be in the direction of a Sprachbund. In other words, the process itself is assumed to be unidirectional, in the sense that it predictably (and perhaps irreversibly) leads to shared properties and eventually to intertranslatabality. In this way it is predicted that every contact situation, given enough time, will result in a Sprachbund. Maintenance of separate lexicons — although very likely containing a shared subset — ensures continued usefulness of the languages in contact as cultural or political identifiers. Thus, Civ’jan (1990) states: ‘‘les langues balkanique, avec leurs nombreuses affinités structurelles, avec un système grammatical commun, gardent leur pleine individualité’’.3 Similarly, Gumperz and Wilson (1971) claim that despite a high degree of intertranslatability between languages spoken in the Kupwar village, ‘‘[s]peakers can validly maintain that they speak distinct languages corresponding to distinct ethnic groups’’ (164f.); see Figure 1. Based on the assumption of the inevitability of change towards intertranslatability, almost any formal similarity is ascribable to convergence. But consider Foley’s (1986) work on the Papuan languages of New Guinea, where he raises the question why, with all the tendencies that he notes toward cultural and linguistic convergence, a homogeneous culture and unifying

No interference → →

Partial isomorphism → →

Intertranslatability of structure Lexicon Lexicon A B

Language Language A B

Language Language A B

Coincidental overlap between grammars

Increasing number of features is drawn into shared grammar

Grammar A–B

Single grammar, two (or more) lexicons

Figure 1. Contact between (genetically unrelated) languages

 Silvia Kouwenberg

language has not emerged from all the linguistic diversity of the area (26f.). Surely, a time-depth of several thousand years would have been sufficient for such a development? The picture that he sketches suggests that every community constitutes its own core, so to speak, and that despite diffusion over large areas, there are also divergent tendencies. This ‘‘results in languages converging toward each other only so far, and the remaining differences between them being accentuated’’ (9). Obviously, viewing this immense area as having as many cores as there are communities is not what convergence studies can handle comfortably. And yet, Foley’s evidence strongly suggests that to regard diffusion as essentially unidirectional does not do justice to the complexities of the sociolinguistic situation, and that despite diffusion of linguistic traits over large areas, linguistic diversity not only remains, but may even be increased (Foley 1986). Note also that for the Kupwar case, Gumperz and Wilson are quite clear about the fact that this situation has come about as a result of a combination of factors including strict ethnic separateness of home life, the association of language with this separateness, and public interaction between all ethnic groups through Marathi, a language which is viewed as socially neutral. The presumed instability here appears to have been caused by widespread bilingualism in the public domain and the normative linguistic status of male bilinguals in the home groups. If social conditions influence the stability of bi/multilingual societies, some social conditions might favour long-standing maintenance of separate grammars as well as separate vocabularies, as perhaps in the Papuan languages. .. Summary To conclude this section, in trying to construct a theory of convergence, and a methodology for the identification of convergence cases, a major problem is that of the mechanism by which languages presumably converge. Such a mechanism has simply been postulated, not demonstrated. The cause-andeffect relations that are assumed to hold between the phenomena of convergence and Sprachbund have been established by circular reasoning rather than by testing against a model. As a result, any and all correspondences between geographically adjacent languages not demonstrably resulting from genetic relationships can be treated as possible results of convergence. Entirely different historical outcomes have been claimed to be instances of convergence, simply because they are characterized by formal similarity. In the following we will review some of these.

Convergence and explanations in creole genesis 

. The phenomena .. Unidirectional convergence Anttila (1989: 172ff.) ascribes the loss of phonemic distinctions in Baltic Finnic to the filtering of its sound system through the Baltic and Germanic ones. For example, Modern Finnish lacks the contrast between /š, s, s´/ of preBaltic Finnic. Baltic has /s, š/, which presumably led to the filtering out of /s´/, while Germanic only has /s/, filtering out /š/. According to Anttila, although all the individual phonological changes between Pre-Baltic Finnic and Late ProtoBaltic Finnic are commonplace, it is the cumulative effect which makes convergence a plausible explanation (173). Similarly, Hock (1991: 493f.) describes Czech word-initial accent (unlike other Slavic languages), certain unusual developments in its vowel system, and the elimination of a historical contrast between palatalized /ly/ and velarized // (which has gone to [l], indistinguishable from [ly]) to the influence of German. Thus, the relevant German dialects had diphthongization of the old high vowels (u¯ > au, ı¯ > ei). Significantly, the German changes predate the analogous development which took place in Czech. It appears from this example that convergence need not involve mutual adaptation. This is a clear case of the-result-which-determines-what-we-callthe-process. We should be aware however, as pointed out before, that individual changes that are isolated in cases of convergence in the literature are always unidirectional. The discussion of the Kupwar case below will further demonstrate that. .. Multidirectional convergence: Kupwar The bi/multidirectional nature of convergence becomes apparent only when looking at a number of changes, where some may have gone in one direction, others in the reverse direction. Gumperz and Wilson (1971) claim that longstanding multilingualism in Kupwar has obliterated the differences which may be observed elsewhere between dialects of Kannada (Dravidian), Urdu and Marathi (both Indic), resulting in virtually complete intertranslatability. Linguistic competence for many Kupwar residents is said to comprise a single grammar and three separate lexicons. Retention of each native lexicon ensures continued usefulness of the languages as social/ethnic identifiers. Comparing the Kupwar dialects of Urdu, Kannada and Marathi with Standard varieties of these languages, Gumperz and Wilson (1971) count the number of changes in each direction. They calculate the changes as 11 involv-

 Silvia Kouwenberg

ing Kupwar Urdu (which has changed about equally towards Kannada and Marathi), 6 involving Kupwar Marathi towards Kannada, and 6 involving Kupwar Kannada towards Marathi. They conclude that most changes have been towards Marathi and Kannada. They also conclude that all changes are convergences involving the three varieties as a set, being changes either of one toward the other two, or of two toward the other one: (1) Urdu, Marathi → Kannada 6 Kannada → Urdu, Marathi 5 Urdu → Kannada, Marathi 4 Urdu, Kannada → Marathi 1 (to the left of the arrow is the affected language, to the right is the language which provides the model; after Gumperz and Wilson 1971: 163)

It must be pointed out that their calculations are based on a small number of features, and may not be reliable indicators of the overall direction of change. For instance, word order is not discussed, but change in that area is mentioned in passing. As noted in 2.1.2, convergent change is normally assumed to eliminate disagreement between languages in contact. Where several languages are involved, we would expect change to be towards a majority pattern. It turns out however, that change is towards a minority pattern in several instances. Based on Gumperz and Wilson’s description of the convergent changes (155ff.), the picture below emerges. Changes 1 and 8.b involve change in two languages towards a third. Affected languages are mentioned first, the arrow points towards the language which is assumed to have provided the model for change. (2) 1. 2. a. b. c. d. e. 3.

replacement of grammatical gender by semantically determined gender: Marathi, Urdu → Kannada Agreement: loss of multiple agreement positions: Urdu → Kannada, Marathi loss of gender- in favour of person-agreement on the past auxiliary: Urdu → Kannada, Marathi loss of person- in favour of tense marking in future verb constructions: Urdu → Kannada loss of case marking and agreement with NP2 on the verb: Urdu, Marathi → Kannada loss of number/gender agreement on noun modifiers: Urdu, (Marathi) → Kannada loss of nominalizing suffix: Kannada → Urdu, Marathi

Convergence and explanations in creole genesis 

4. 5. 6. 7. 8. a. b. 9. a. b. c. 10.

loss of accusative marking in certain contexts: Kannada → Urdu, Marathi use of past non-finite verb forms in certain contexts: Urdu → Kannada, Marathi use of copula: Kannada → Urdu, Marathi replacement of ‘say’ complementation by ‘that’ complementation: Kannada → Urdu, Marathi interrogative verbal suffixes: replacement of interrogative suffix by ‘what’: Kannada → Urdu, Marathi invariable final position of ‘what’: Urdu, Kannada → Marathi semantic change: change in purposive constructions: Urdu → Marathi, Kannada use of genitive suffix in certain contexts: Urdu → Marathi, Kannada adoption of inclusive/exclusive distinction for 1Pl pronouns: Urdu → Marathi, Kannada mentioned in passing: word order change: Urdu → Kannada, Marathi

From the above it appears that Kupwar Urdu is most severely affected by contact with Kannada and Marathi. More importantly, Kannada and Marathi appear to determine the direction of change about equally. But with Marathi as the ‘mediator language’, change towards Kannada is unexpected. Recall that inter-ethnic communication takes place through Marathi, a language which is viewed as socially neutral. Change towards a non-Marathi model, in this case Kannada, does not concur with the diglossic situation described here. Contact, after all, is between Urdu and Marathi and between Kannada and Marathi. The possible influence of Kannada on Urdu should at least be mediated through Marathi: if Marathi should change towards a Kannada model in contact with that language, it could conceivably be the source of a similar change in Urdu. But there is no indication in Gumperz and Wilson that this is necessarily the case. Again, the question of the mechanism by which convergent change takes place arises: does the direction of communication determine the direction of change? .. The Balkans The Balkan Sprachbund area constitutes the single most prominent convergence case in the literature. The languages which form the Balkan Sprachbund

 Silvia Kouwenberg

— Greek, Albanian, Macedonian (South Slavic), Bulgarian (South Slavic), Serbo-Croatian (South Slavic), and Romanian (Romance) — are said to share several areal features which set them apart from surrounding, genetically related languages. The most celebrated of these is that of the finite (personal) construction which has replaced the infinitive. In the following, we will first discuss some of the problems surrounding infinitive loss, then turn to some more general issues. Joseph (1983) chronicles the Balkan infinitive loss. It is a rather sobering study. He points out that the typological status with respect to finiteness of the Balkan languages without an infinitive is no different from their status before the loss of the infinitive. His claim is that all that has changed really in the verbal system of a language like Greek is the extent to which the finite/ nonfinite distinction is realized morphologically in the verbal system (p. 215). Joseph also points out repeatedly that the divergences among the Balkan languages concerning infinitive loss deserve as much emphasis as the general convergence which has received so much attention (e.g. p. 253). Thus, infinitive loss is not categorical in any of the Balkan languages (except perhaps Greek), and the infinitive has been revitalized in several. In Greek the infinitive was essentially lost by the tenth century, and while there are systematic traces of an infinitive up to the 15th or sixteenth century, there are no such traces afterwards (p. 81). In contrast, Albanian displays a fairly recent receding of the infinitive in the Tosk dialect, and a countertendency of revitalization of the infinitive (p. 100). Romanian has come to restrict the use of the infinitive, but has retained it for the complements of a putea ‘can, be able’, and a sti ‘know how (to)’ (p. 162). Joseph also states that even though language contact seems to have played an important role in starting the infinitive replacement process in the Balkan languages, each language essentially carried the process through on its own, and pre-existing tendencies in certain languages where finite and nonfinite forms were in competition, combined with other language-particular developments such as a phonetic merger of the infinitive with a third person finite form (p. 187 ff.) appear to have been instrumental in this process. Bynon (1996: 49) adds to this the general tendencies of spoken language varieties to use rather fewer infinitive constructions than do their literary counterparts. An additional point which deserves to be made is that all of these languages possessed an infinitive construction before, and that its loss is an innovation for each one of these languages (Bynon 1996: 48). Its loss therefore can be said to have resulted in uniformity (but note again that infinitive loss

Convergence and explanations in creole genesis 

is not uniform, with partial retentions in far from identical contexts in the participating languages), but not necessarily in greater uniformity than that which existed before; rather, convergence has resulted in different similarity from that which existed in the pre-change situation. Clearly, and counter to expectations, convergence has not here exploited shared or similar properties, in which case we would have expected to see levelling in the contexts of use of the infinitive, not loss. A more general problem with the Balkan case is that none of the Balkan languages have all of the areal features which are said to define this area (Lehiste 1988: 64). For instance, Campbell et al. (1986) point out that one of the strongest Balkan features, the postposed article, is not in Greek (p. 561). They conclude that the concept of an areal core from which isoglosses expand outward (see Masica 1976: 170f; Joseph 1983: 245) seems inappropriate for the Balkans, as the language with the greatest number of areal features is Rumanian, on the northern border of the area, whereas Macedonian, the language considered most typically Balkan, lacks several of the areal traits (561). All in all, they judge the Balkans to be a weakly defined linguistic area. A final problem is that several of the areal features have no local precursor. In other words, these features could not have been incorporated into the grammars of these languages by mutual accomodation as there was apparently no model to accomodate. McMahon (1994: 220) claims that even the loss of the infinitive, which is usually acribed to Greek — where it is said to have been an innovation which then spread to languages under its influence — cannot be traced to Greek with any confidence. It has been suggested that ‘‘the ideal model, to which the Balkan Linguistic Area aspires, is located not in the past, but in the future’’ (Joseph 1983: 199, who quotes Civ¹jan 1965; this position is also referred to, with apparent agreement, by Jeffers and Lehiste 1979: 146). What we have here is a Science Fiction scenario of a linguistic conspiracy whose objective is complete cross-language uniformity of structure. What this again demonstrates is that failure to define an explicit model of convergent change has yielded a set of conclusions which are in fact unfalsifiable assumptions. .. Summary In the preceding we have seen that, despite all appearances, convergence is not necessarily bidirectional, does not necessarily utilize what converging languages have in common, and does not even necessarily have a model in any of the languages in contact. Our conclusion is that convergence in historical

 Silvia Kouwenberg

linguistics really characterizes a result, rather than the process that brings about this result. As such, it is a descriptive rather than explanatory concept. As we shall see below, the reverse is true of the recent application of this concept to creole issues: convergence is taking on a different identity in its new creole employment.

. Convergence in creole studies . Introduction The relationship between creole language studies and historical linguistics has been an uneasy one from the earliest attempts by scholars such as Hugo Schuchardt to undertake the former (DeCamp 1977; Meijer and Muysken 1977), but recently, both sides have shown renewed interest in the other. Thus, Thomason and Kaufman (1988) present a theory of creolization which is an integral part of a theory of genetic linguistic relationships, and from their end, several creolists now approach creolization as a historical process which can profitably be examined with the tools provided by theories of language change developed in historical linguistics (e.g. Bruyn 1995; articles in Baker and Syea 1996). This development includes a rapid increase in references to convergence and convergent developments in relation to creole genesis.4 Consider for instance Mufwene’s (1993) Africanisms in Afro-American Language Varieties. Contributors to that volume appeal to convergence to account for such wideranging phenomena as those in (3); of these, Singler and Thomason use the term congruence to refer to the same types of phenomena referred to as convergence by the other authors.5 (3) – – –

selection of lexical or functional items (Hazaël-Massieux, Baker, Robertson) syntactic patterns (Hancock) general (Boretzky, Holm, Mufwene, Singler, Thomason)

Rozencvejg’s (1976) model of convergence involves some notion of optimalization of grammars through simplification which he claims to be characteristic of linguistic change under contact conditions, and which shows up in particular in creoles (26). He thus draws an explicit parallel between creole formation and cases where convergence has taken place, as in the Balkans (see also Bynon 1996). Today’s creolists are painfully aware of the problematic

Convergence and explanations in creole genesis 

nature of the notion of simplification (see Muysken 1988; note that Bynon 1996 does not appear to be hindered by such an awareness). Also, Rozencvejg rather jumps to conclusions when he states that ‘‘[a] consistent and radical simplification of this nature [viz. involving the replacement of infinitives by finite constructions] has taken place in the Creole French languages’’ (46), as work such as Koopman (1986), and DeGraff (1993, 1994) shows. Nonetheless, the view that ‘‘normal’’ contact-induced phenomena and those resulting in creolization are instances of a single type of change, as also espoused by Weinreich (1953) and Thomason and Kaufman (1988), is worth investigating. But, as we shall see below, creolists use convergence in ways markedly different from those we have reviewed in the preceding sections. The difference may be explainable in terms of a general difference in emphasis: historical convergence studies appear to be concerned with particular types of change which can be identified by their historical results. Creolists on the other hand, are interested in mechanisms of change, and in demonstrating how these mechanisms account for known results. Historical convergence studies are not strong on mechanisms, as we have seen. . The phenomena .. Similarity of form Convergence is most commonly appealed to in relation to the selection of lexical items in creole genesis, where it refers to the choice of items which show accidental similarity of form in several contributing languages. Thus, Baker (1993), discussing the African contribution to the lexicon of Frenchbased Creoles, points to the possibility of convergence between lexical items from different contributing languages, viz. between words from Bantu and non-Bantu African languages, or between an African and a French word (140), as in the examples in (4). (4) a.

bunda, buda, bõda ‘behind, anus’ (Haiti, Guadeloupe, Dominica, Martinique, St.Lucia): Kimbundu mbunda ‘nadégas’; also: Manding buu-daa ‘anus’, Bambara boda ‘anus’ (Baker 1993: 142) b. senga ‘boisson alcoolisée’ (Seychelles): Nyanja nsenga ‘beer sediment’; also: Wolof senga ‘palm wine’ (Baker 1993: 145)

Exactly the same phenomenon is referred to as conflation by Kihm (1987), who takes it a step further in suggesting that this phenomenon helps in the

 Silvia Kouwenberg

retention of particular items in creole genesis. He defines conflation as follows: Given that a fortuitous formal similarity of really or apparently comparable elements from possibly very different languages is an attested and, after all, inevitable fact, one may expect spontaneous learners of a second language to grab at such elements and conflate them in their minds, by virtue of this principle [. . .] that you more easily learn what you think you already know. (p. 113)

Concretely, chance similarities of form and function of functional items in Portuguese and the languages of Guiné-Bissau helps explain why some Portuguese elements were retained in Kriol rather than some others, and why they occur in a somewhat dissimilar form or function as compared to the corresponding Portuguese items. The more striking cases are set out in Table 2. Obviously, coincidence of form in these cases has results which go beyond the mere retention of an item: as Kihm also points out, his lexical items are associated with semantic and syntactic properties which are of wider relevance for the grammar of Kriol. Note that Kihm chooses to treat the items under discussion as modified Portuguese retentions, whereas they might equally well be said to constitute modified substrate retentions. The relevant point appears to be that it is the interaction between superstrate and substrate languages which accounts for the presence of these forms in Kriol, and for their properties. Simons (1985) presents comparable cases in Solomon Islands Pijin — although she does not refer to these as instances of convergence. Solomon Islands Pidgin nao can equally well be related to English ‘now’ as to a series of Table 2. (After Kihm 1987: 114–128) Kriol

Portuguese

Substrate (selection)

Preverbal negation ka

nunca ‘never’

Manjaku dika/rika unaccomplished negation, Balanta ke negative (modified by conditional)

1sg clitic n

oblique 1sg mi, mim

Balanta n, Mandinka n´ (subject), Manjaku -in, Mandinka -n

Past enclitic ba

imperfect indicative -va or acabar ‘finish’ > Kriol kaba

Manjaku ba ‘finish’ expresses past

Definite punctual future vem ‘she/he/it comes’ bin; also adverb ‘and then’

Manjaku bi, Mankanya bi, Balanta ben ‘come’ also used to express future and ‘and then’

Causative -ntV / ndV

Manjaku -andan, Mandinka -ndi causative

transitive -ntar

Convergence and explanations in creole genesis 

Malaitan forms. It performs the functions illustrated in (5), which correspond to Malaitan functions performed by forms which display remarkable coincidence of form with the presumed English etymon of nao. (5) a.

nao is sentence-final and marks completive: Hemi had tumas nao. 3sg difficult very nao corresponds to Malaitan na’a/no’o ‘It is very difficult.’ b. nao marks a topic: Deswan nao mi no save. this nao 1sg neg know corresponds to Malaitan na/ne’e ‘But this I don’t know.’ c. nao is a discourse sentence connector: Nao mi talem hem. nao 1sg tell 3sg corresponds to Malaitan ma/na ‘And then I told him.’

In the same vain, Kouwenberg (1992) proposes that the Berbice Dutch Creole perfective suffix -tε can be profitably treated as the result of bidirectional linguistic accomodation, the retention of the Kalabar  i perfective suffix -té having been made possible by its similarity of form and function with the Dutch past imperfective suffix -də/-tə (of the Eastern-Ijo dialects which have contributed to the formation of Berbice Dutch Creole, Kalabar  i is assumed to have been the most important; see Smith et al. 1987). .. Similarity of structure For Hazaël-Massieux (1993), as for Kihm (1987), it is the coincidence of aspects of different linguistic systems which constitutes convergence/conflation. Such coincidence consists of some measure of (superficial) similarity of form or structure. In the selection of forms and patterns in the formation of the incipient creole, ‘‘converging’’ forms and patterns are thought to be favoured. Hazaël-Massieux develops a scenario of creolization in which speakers in contact overgeneralize the rules they infer from the norm to which they are exposed, and reinforcement of such rules takes place where these coincide with some aspect of either their own system or their interlocutor’s system (p. 110 f.). This type of reinforcement he calls filtering, a term reminiscent of Anttila’s (1989) Sprachbund filters, and with quite comparable results. A concrete example which he examines in some detail is that of syllable structure:

 Silvia Kouwenberg

French, on the whole, allows syllables of the following types: V, CV, CCV, CCCV, VC, VCC, VCCC, VCCCC (cf. hie ‘rammer’, si ‘if ’, tri ‘sorting’, strie ‘scratch’, as ‘ace’, halte ‘halt’, arbre ‘tree’, [d]extre ‘right hand’). All the syllables possible in Guadeloupean Creole are included in this list. However, even taking into account the most acrolectal variants of this creole, some syllabic patterns are always excluded. In word-initial position, consonant clusters are broken by either vocalic prosthesis or epenthesis, for example, istilo ‘fountain pen’, estati ‘statue’, ispo ‘sport’. In final position, the tendency, which is also attested in popular French [. . .], is to reduce consonant clusters to single consonants. Consonant clusters are thus allowed only word medially. It may be argued, quite plausibly, that this reduction, going in the direction of being less marked, was triggered or favored by the convergence of preferred syllabic structure in popular French and in African and Amerindian languages. (p. 113)

In this manner, African and Amerindian languages which disfavour consonant clusters are assumed to have filtered the syllabic structure of French, itself favourably predisposed, so to speak, by a similar tendency in popular French. But note that both Hazaël-Massieux’s filtering scenario and Kihm’s conflation are unidirectional, involving the modification of French and Portuguese, respectively, just as Anttila’s treatment of Baltic Finnic convergence with Baltic and Germanic is unidirectional. Hancock’s (1993) position is that ‘‘where syntactic patterns happen to be shared by some, or all, of the input languages, nothing new needs to be learned, and convergence of functions will help to ensure their retention in the emerging linguistic system’’ (187). He does not, unfortunately, illustrate this claim, and the vagueness of its formulation tempts one to repeat some of the objections raised against convergence treatments in historical linguistics: at what level of grammatical analysis do we consider syntactic patterns to be ‘‘shared’’, and if ‘‘learning’’ plays such an important role, how exactly do we incorporate insights from second language acquisition studies in our views of creolization? Singler’s and Thomason’s congruence is similarly undefined and unillustrated. However, some of the Berbice Dutch Creole properties examined in Kouwenberg (1992) could conceivably illustrate Hancock’s convergence, and Singler’s and Thomason’s congruence. We will briefly consider postpositions here. At first sight, (6) provides a striking structural parallel between Kalabar  i and Dutch, the substrate and superstrate languages of Berbice Dutch Creole: both employ postpositions in a context where a prepositional structure is used in English. Presumably, it is this convergence or congruence of syntactic patterns which accounts for the appearance of postpositions in Berbice Dutch Creole, a marked structure for an SVO language. After all, nobody needed to ‘‘learn’’ anything ‘‘new’’ here.

Convergence and explanations in creole genesis 

(6) Kalabar  i wárı¯ mé bí  o¯ house the inside ‘in the house’

Berbice Dutch di wari ben the house inside ‘in the house’

Creole Dutch het huis binnen the house inside ‘into the house’

But we should not allow ourselves to be deceived by appearances — the resemblance observed here is very superficial indeed. Dutch is predominantly prepositional, in constrast to Kalabar  i, which is exclusively postpositional. Dutch forms such as binnen ‘inside’ are postpositional only when heading a goal argument, prepositional where a location argument is involved. Berbice Dutch Creole postpositions are always spatial in reference (in contrast with those of Kalabar  i), and not restricted to goal arguments. In addition, Berbice Dutch has several non-spatial prepositions. Its postpositions are nominal, like those of Kalabar  i, but also have some clearly adpositional properties. More divergences between the Kalabar  i and Dutch models and the end-result of their contact are discussed in Kouwenberg (1992). The properties of Berbice Dutch Creole postpositions are hardly a common denominator of those of Kalabar  i and Dutch (see also Kouwenberg 1994 for an in-depth discussion of the properties of Berbice Dutch Creole postpositions). In all, an awful lot of new stuff appears to have been learned despite — or because of? — the superficial coincidence of syntactic patterns. It is obvious that retention is too simple a concept to cover the cases at hand. In his discussion of the possibility that say-complementation may have found a partial model in colloquial English quotatives, Mufwene (1996) also demonstrates this. He cites such constructions from the Genesis books, whose discourse style is thought to reflect spoken language in some ways, and from folk narratives, as in (7). He points out that the Gullah constructions in (8) and (9) must be considered in relation to substrate models as well as in relation to such a partial superstrate model. He points out that in nonstandard French, quotative uses of dire ‘say’ comparable to those of say in nonstandard English are rare, a difference which may explain why we do not find constructions similar to (8) and (9) in French Creoles, despite the equally significant presence of speakers of more or less the same West African languages among the substrate populations that developed them (Mufwene 1996: 15). (7) A man came to the door, says, ‘Hello stranger. What’ll ye have?’ (8)  æks m sε wε yi mama də I ask him/her say where you mother be a. ‘I asked him/her where your mother was.’ [N.B. translation corrected] b. ‘I asked him/her: ‘Where is your mother’?’

 Silvia Kouwenberg

(9)  hiə sε hi mama s k I hear say he/she mother sick ‘I hear/heard that his/her mother is/was sick.’ (Mufwene 1996: 12)

.. Convergence and divergence It is obvious that accounts of creole genesis which exploit similarities between substrate and superstrate languages are increasingly popular. We find statements of the form: ‘‘superstrate A and substrate B converge on grammatical property X, and that is why we get X in the creole’’ (see above for examples). There is a strong suggestion in this approach to creole genesis that where the substrate and superstrate are typologically similar, the creole will end up looking like a common denominator of the two. I have argued elsewhere, based on my work on the relation between Berbice Dutch Creole and its lexifiers Kalabar  i and Dutch, that this is a fallacy (Kouwenberg 1992). In fact, what is problematic here is the very imprecise notion of similarity. There is no agreement on what level of representation we should refer to to determine that domains of the grammars of languages A and B are similar. Where speakers perceive superficial similarities between surface structures of languages A and B, they are likely not to have access to more substantive similarities or dissimilarities at underlying levels of representation. Thus, convergence interpreted as coincidence of forms or patterns will yield the right results only where surface properties are fairly straightforward reflections of underlying properties. Where this is not the case, acting on perceived surface similarities may result in increasing divergence from the contributing languages. Application of the creole notion of convergence may thus result in dissimilarity between a creole language and the contributing substrate(s) and/or superstrate(s). This observation is based on a reconstruction of the development of Berbice Dutch Creole, typologically an SVO-language, in relation to its lexifiers, Kalabar  i (substrate) and Dutch (superstrate), both SOV-languages. Surface variation and typological inconsistency in both Kalabar  i and Dutch made both SOV and SVO-type orders available in the contact situation. Extra-linguistic considerations determining the types of exchange between speakers in contact, and thus the text-frequency of certain constructions and the surface orders of these, appear to have tilted the balance in favour of SVO-type orders (Kouwenberg 1992, 1996a). So although coincidence of their surface orders can be said to have resulted in the selection or continuity of certain patterns, the final result is typological distance between the creole language and its lexifiers. In Kouwenberg (1996b) I similarly argue that interplay between properties of

Convergence and explanations in creole genesis 

contributing languages may yield results which match properties of none of the languages in contact. .. Multiple causation In addition to the creole use of convergence to denote cases of similarity in form or structure between contributing languages in creole formation, there is another use of this notion which comes down to what we might call multiple causation: where several possible explanations of a particular fact about a creole language are available, reinforcing influence of all of these is assumed. For instance, Singler (1993), discussing the reasons adduced for the supposedly more important role of Kwa as opposed to Bantu languages in the formation of Haitian, notes that two possible explanations are available: the comparatively early arrival of Kwa speakers, and the relative suitability of Kwa grammar being largely analytic (rather than agglutinating, as in Bantu). The two explanations, he notes, converge, i.e. have similar results (p. 248 f.). Holm (1993) too uses convergence in this sense, as when he adopts the position that the phonology of creole languages ‘‘is likely to reflect the influence of not only their substrate but also their superstrate and sometimes their adstrate languages, as well as creole-internal innovations, universals of adult second language acquisition, and the convergence of all or some of these factors’’ (p. 317). Thus, the origin of certain palatals in the Atlantic creoles is ascribed by him to a combination of factors consisting of African substrate languages, language universals, and superstrate features that are now archaic or regional. For instance, he notes that Caribbean varieties of creole French tend to palatalize French /t/ and /d/ before a front high glide, and /k/ and /g/ before any front vowel, as in the forms in Table 3, a pattern which he claims to be partially parallel to Twi dialect variation between /ky/ and /ty/, on the one hand, and /gy/ and /dy/ on the other, but which can also be related to French palatalized /k/ and /g/ which were standard in the seventeenth century (322f.). Table 3. Palatalization in Caribbean Creole French (Lesser Antilles, French Guyana, Haiti). From Holm (1993: 322), after Stein (1984: 24–25) French

Gloss

Affricates

tiens bien coeur diable gueule

‘hold on well’ ‘heart’ ‘devil’ ‘snout’

tʃε˜ be tʃe dab dɔl

Palatalized alveolars

Palatalized velars

Depalatalized velars

tjɔ djab

kjε˜ nbe kjε gjab gjɔl

kε˜ be kε

 Silvia Kouwenberg

As both substrate explanations and superstrate explanations are available, converging influence of these is assumed. This second use of convergence represents an extended use of the notion; as in its first creole use, the emphasis is on similarity, but here the coincidence is not one of forms or patterns, but one of possible outcomes of competing explanations. Where different explanations are available to account for a particular result, these are assumed to have reinforced each other. Some authors, such as Mufwene (1993), use convergence in both the first and second senses — i.e. with reference to the possible coincidence of aspects of different source languages, and with reference to similar outcomes of different processes called upon as explanations; he also employs the term in the more historical sense of change in language contact situations (the latter with reference to Afro-American Vernacular English in relation to Standard American English). .. The return of the Cafeteria Principle Considering the recent literature, one will realize that the cases discussed in the preceding are by no means exceptional, and that convergence is rapidly gaining in popularity in creole linguistics. By now, it should be abundantly clear that the creole use of convergence does not correspond to its historical use. In fact, it seems to be going the way a number of other terms in creole studies which have been borrowed from that source have gone, that is to say, it is starting to lead a life of its own. This is hardly surprising in view of the fact, documented above, that no coherent theory of convergence is available in historical linguistics. A major difference with its historical use is that in the creole case, no change of existing languages toward a common model (real or imagined) is involved. In fact, it is an unstated — and perhaps unwarranted — assumption that the African and European source languages of creole languages remained largely unchanged in the contact situation. Its creole use refers either to superficial similarities between languages in contact (and is thus similar to Kihm’s 1987 conflation), or to the similar results of different possible explanations in creole genesis. From describing an observed result (and the process leading to this result) in historical linguistics, convergence has become an explanatory principle, accounting for the selection of particular forms or patterns in creole genesis. The only common ground between the historical and creole use of the term is in the reference to similarity; this however constitutes the endpoint of a development for the former, the starting point for the latter: convergence in historical linguistics leads to shared or similar

Convergence and explanations in creole genesis 

properties, and does not necessarily exploit initially shared or similar properties of languages, whereas convergence in creole linguistics (in the first of the two senses in which it is used there) depends entirely on initially shared or similar properties. In other words, while both traditions crucially refer to comparability of form or structure, they do so at different ends of a historical development. In its second creole sense, convergence functions as a kind of cover-all. In the best possible case, convergence of explanations provides cumulative evidence for a particular development, but in fact, this is not how it works. Many creolists now profess to support a position on creole genesis which incorporates a whole rainbow of factors of relevance, including the substrate, universals, the superstrate, second language acquisition strategies, salience, markedness, etc. Convergence of several or all of these serves to mitigate the indeterminacy that follows from such a position. Obviously, this allows for the selection of features into the incipient creole as a completely eclectic process, hinging only on (coincidental) similarities or putatively similar outcomes of processes. Thus we witness the return of the Cafeteria Principle through the back door.6 The question therefore is that of explanation: does convergence qua multiple causation in fact account for anything? Are we not in danger of adopting a new vague and indeterminate explanatory principle just because substratists, superstratists and universalists alike appear to have reached a stalemate in explaining creole genesis?

. Concluding remarks Having shown that convergence in historical linguistics cannot be diagnosed until after the fact, and that it is equated with its result in a circular and unrevealing exercise, we may conclude that convergence in historical linguistics suffers from a lack of explanatory power; it is thus reduced to being a purely descriptive concept. On the other hand, its employment in relation to creole issues, where it is interpreted as referring to multiple sources of creole forms or patterns and/or to multiple causation, has taken on explanatory force. Unfortunately, the present desire to leave all options open and allow for any and all possible factors to contribute to explanations in creole genesis seriously detracts from its validity, and may even mean the return of the Cafeteria Principle. In addition, as shown for one case, that of Berbice Dutch Creole, reliance on superficially observed similarities may just result in change away

 Silvia Kouwenberg

from an available model, which is the opposite of what appears to be intended with the creole use of convergence. At this point, it is doubtful that convergence yields interesting generalizations over the processes at work in creole genesis. After all, we can only be clear about the significance of convergence if we are clear about the mechanisms of substrate and superstrate transfer, first and second language acquisition etc. which are claimed to converge. In the absence of such clarity, convergence will merely serve to cover a general reluctance to formulate strong and testable hypotheses about these mechanisms. This is a pessimistic evaluation of its potential for explanation in creole genesis. On the other hand, where strong hypotheses are available, the cumulative evidence of converging explanations has the potential to strengthen the case for a particular scenario of creole development. In short, creole linguists need to get back to the drawing board.

Notes . My translation: Proposition 16. We shall refer to any group of languages united by a number of systematic similarities as ‘Sprachgruppe’. We may distinguish two types of Sprachgruppe. (1) Groups consisting of languages which display much similarity in syntax, as well as similarity in the basic morphological structures, which offer a large number of shared cultural vocabulary items, and which often also display similarity in sound inventories, but which do not have systematic sound correspondences, and which lack similarity of morphological forms and shared basic vocabulary. Such Sprachgruppen will be called ‘Sprachbunde’. (2) Groups consisting of languages which have a large number of shared basic vocabulary items, and which display similarities in the sound shape of morphological categories, and which, most of all, display systematic sound correspondences. Such Sprachgruppen will be called ‘Sprachfamilien’ (language families). . My translation: what distinguishes borrowing from kinship is only their chronological order. . My translation: despite their numerous structural similarities and their shared grammars, the Balkan languages have retained their individual identities. . Silverstein (1972) employed a convergence model to account for the genesis of Chinook Jargon, but Thomason (1982) demonstrates his assumption that Chinook Jargon arose out of contact between Chinook and English to be false, as Chinook Jargon can be shown to predate colonial contacts. Other early references to convergence include discussions by Hymes (1971), sparked by Gumperz and Wilson’s contribution to that volume on the Kupwar linguistic situation, and occasional use of the term in relation to creole continuum situations, where the basilect is sometimes thought to converge or have converged to the acrolect. . Robertson’s use of the term convergence in relation to the development of the clause-

Convergence and explanations in creole genesis 

final negator of Berbice Dutch Creole is distinctly idiosyncratic: it refers to the combination of a substrate-derived form *ka and a superstrate-derived form *nε in the creole form kanε. As his represents an unusual application of the concept, I shall not consider it here. . But note the recent defense of the Cafeteria Principle in Mufwene (1996).

References Anttila, Raimo. 1989. Historical and Comparative Linguistics (2nd edn). Amsterdam and Philadelphia: John Benjamins. Appel, René and Pieter Muysken. 1987. Language Contact and Bilingualism. London: Edward Arnold. Baker, Philip. 1993. ‘‘Assessing the African contribution to French-based creoles’’. In Mufwene (ed.), 123–155. Baker, Philip. 1995. ‘‘Motivation in creole genesis’’. In From Contact to Creole and Beyond, Philip Baker (ed.), 3–16. (Westminster Creolistics Series 1). London: University of Westminster Press. Baker, Philip and Anand Syea (eds.). 1996. Changing Meanings, Changing Functions: Papers relating to grammaticalization in contact languages (Westminster Creolistics Series, 2). London: University of Westminster Press. Boas, Franz. 1929. ‘‘Classification of American Indian languages’’. Language 5: 1–7. Boretzky, Norbert. 1993. ‘‘The concept of rule, rule borrowing, and substrate influence in creole languages’’. In Mufwene (ed.), 74–92. Bruyn, Adrienne. 1995. Grammaticalization in Creoles: The development of determiners and relative clauses in Sranan. Amsterdam: IFOTT. (Studies in Language and Language Use, 21.) Bynon, Theodora. 1977. Historical Linguistics. Cambridge: Cambridge University Press. Bynon, Theodora. 1996. ‘‘Convergent change: Some recent rethinking’’. In Baker and Syea (eds.), 47–52. Campbell, Lyle, Terrence Kaufman and Thomas C. Smith-Stark. 1986. ‘‘Meso-America as a linguistic area’’. Language 62: 530–570. Civ¹jan, T.V. 1965. Imja sušcˇestvitel’noe v balkanskix jazykax. Moscow: Nauka. (Cited in Jeffers and Lehiste, 1979.) Civ¹jan, T.V. 1990. ‘‘Résumé of La carcasse linguistique du modèle du monde balkanique (MMB)’’. In Lingvisticeskie osnovy balkanskoj modeli mira. Moscow: Nauka. [No page number.] DeCamp, David. 1977. ‘‘The development of pidgin and creole studies’’. In Pidgin and Creole Linguistics, Albert Valdman (ed.), 3–20. Bloomington and London: Indiana University Press. DeGraff, Michel. 1993. ‘‘Is Haitian Creole a pro-drop language?’’ In Atlantic meets Pacific: A global view of pidginization and creolization, Francis X. Byrne and John Holm (eds.), 71–90. Amsterdam and Philadelphia: John Benjamins. DeGraff, Michel. 1994. ‘‘The syntax of predication in Haitian’’. Amsterdam Creole Studies 11: 1–15.

 Silvia Kouwenberg

Dryer, Matthew S. 1989. ‘‘Large linguistic areas and language sampling’’. Studies in Language 13: 257–292. Emeneau, Murray B. 1956. ‘‘India as a linguistic area’’. Language 32: 3–16. Foley, W. 1986. The Papuan Languages. Cambridge: Cambridge University Press. Gumperz, John and Robert Wilson. 1971. ‘‘Convergence and creolization: A case from the Indo-Aryan/Dravidian border in India’’. In Pidginization and Creolization, Dell Hymes (ed.), 151–167. Cambridge: Cambridge University Press. Hancock, Ian. 1993. ‘‘Creole language provenance and the African component’’. In Mufwene (ed.), 182–191. Hazaël-Massieux, Guy. 1993. ‘‘The African filter in the genesis of Guadeloupean Creole: At the confluence of genetics and typology’’. In Mufwene (ed.), 109–122. Hock, Hans Heinrich. 1986. Principles of Historical Linguistics. Berlin and New York: Mouton de Gruyter. Holm, John. 1993. ‘‘Phonological features common to some West African and Atlantic creole languages’’. In Mufwene (ed.), 317–327 Hymes, Dell. 1971. ‘‘Introduction to Part III, General conceptions of process’’. In Pidginization and Creolization, Dell Hymes (ed.), 65–90. Cambridge: Cambridge University Press. Jakobson, Roman. 1931. ‘‘Sur la théorie des affinités phonologiques entre les langues’’. Reprinted in Nikolai S. Trubetzkoy 1949. Principes de Phonologie, 351–365. Paris: G. Klincksieck. Jakobson, Roman. 1939. ‘‘Franz Boas’ approach to language’’. International Journal of American Linguistics 10: 188–195. Jeffers, Robert J. and Ilse Lehiste 1979. Principles and Methods for Historical Linguistics. Cambridge, MA: MIT Press. Joseph, Brian D. 1983. The Synchrony and Diachrony of the Balkan Infinitive: A study in areal, general, and historical linguistics. Cambridge: Cambridge University Press. Kihm, Alain. 1987. ‘‘Conflation as a directive process in creolization’’. In Beiträge zum 4. Essener Kolloquium, Norbert Boretzky, Werner Enninger and Thomas Stolz (eds.), 111–138. Bochum: Brockmeyer. Koopman, Hilda. 1986. ‘‘The genesis of Haitian: Implications of a comparison of some features of the syntax of Haitian, French, and West African languages’’. In Substrate versus Universals in Creole Genesis, Pieter C. Muysken and Norval S. H. Smith (eds.), 231–258. Amsterdam and Philadelphia: John Benjamins. Kouwenberg, Silvia. 1992. ‘‘From OV to VO. Linguistic negotiation in the development of Berbice Dutch Creole’’. Lingua 88: 263–299. Kouwenberg, Silvia. 1994. A Grammar of Berbice Dutch Creole. (Mouton Grammar Library, 12.) Berlin and New York: Mouton de Gruyter. Kouwenberg, Silvia. 1996a. ‘‘Grammaticalization and word order in the history of Berbice Dutch Creole’’. In: Baker and Syea (eds.), 207–218. Kouwenberg, Silvia. 1996b. ‘‘Substrate or Superstrate: What’s in a name?’’ Journal of Pidgin and Creole Languages 11.2: 343–347. Lehiste, Ilse. 1988. Lectures on Language Contact. Cambridge, MA: MIT Press. Masica, Colin P. 1976. Defining a Linguistic Area: South Asia. Chicago and London: University of Chicago Press.

Convergence and explanations in creole genesis 

McMahon, April S. 1994. Understanding Language Change. Cambridge: Cambridge University Press. Meijer, Guus and Pieter Muysken. 1977. ‘‘On the beginnings of pidgin and creole studies: Schuchardt and Hesseling’’. In Pidgin and Creole Linguistics, Albert Valdman (ed.), 21–45. Bloomington: Indiana University Press. Mufwene, Salikoko S. (ed.). 1993. Africanisms in Afro-American Language Varieties. Athens, GA: University of Georgia Press. Mufwene, Salikoko S. 1993. ‘‘Introduction’’. In Mufwene (ed.), 1–31. Mufwene, Salikoko S. 1996. ‘‘Creole genesis: A population genetics perspective’’. In Caribbean Language Issues, Old and New, Pauline Christie (ed.), 163–196. Barbados, Jamaica and Trinidad and Tobago: The University Press of the West Indies. Muysken, Pieter C. 1988. ‘‘Are creoles a special type of language?’’. In Linguistics: The Cambridge survey, vol ii, Frederick J. Newmeyer (ed.), 285–301. Cambridge: Cambridge University Press. Robertson, Ian E. 1993. ‘‘The I jo element in Berbice Dutch and the pidginization/ creolization process’’. In Mufwene (ed.), 296–316. Rozencvejg, Victor Ju. 1976. Linguistic Interference and Convergent Change. The Hague and Paris: Mouton. Sebeok, Thomas. 1949. ‘‘The meaning of ‘Ural-Altaic’ ’’. Lingua 2: 124–139. Silverstein, Michael. 1972. ‘‘Chinook Jargon: Language contact and the problem of multilevel generative systems I–II’’. Language 48: 378–406; 596–625. Simons, Linda. 1985. ‘‘Malaitan influence on two grammatical particles in Solomon Islands Pijin’’. Papers in Pidgin and Creole Linguistics 4: 53–65. (Pacific Linguistics Series A No. 72.) Singler, John V. 1993. ‘‘African influence upon Afro-American language varieties: A consideration of sociohistorical factors’’. In Mufwene (ed.), 235–253. Smith, Norval S.H., Ian E. Robertson and Kay Williamson. (1987). ‘‘The I jo element in Berbice Dutch’’. Language in Society 16: 49–90. Stein, Peter 1984. Kreolisch und Französisch. Tübingen: Max Niemeyer. Swadesh, Morris. 1971. The Origin and Diversification of Language, Joel Sherzer (ed.). London: Routledge & Kegan Paul. Thomason, Sarah G. 1982. ‘‘Chinook Jargon in areal and historical context’’. Language 58: 820–870. Thomason, Sarah G. 1993. ‘‘On identifying the sources of creole structures. A discussion of Singler’s and Lefebvre’s papers’’. In Mufwene (ed.), 280–295. Thomason, Sarah G. and Terrence Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley, CA: University of California Press. Trubetzkoy, Nikolai S. 1928. ‘‘Établissement et délimination des termes techniques’’. Actes du Premier Congrès de Linguistes, tenu à la Haye, du 10-15 Avril 1928, 17–18. Leiden: Sijthoff. Trubetzkoy, Nikolai S. 1936. ‘‘Gedanken über das Indogermanenproblem’’. Acta Linguistica 1: 81–89. [Posthumously published in 1939]. Weinreich, Uriel. 1953. Languages in Contact. Findings and Problems. New York: Linguistic Circle.

Chapter 10

Contact-induced language change and pidgin/creole genesis Sarah G. Thomason

.

Introduction

The topic of this chapter is the complex relationship between linguistic interference — that is, change in an existing language that results partly or entirely from the influence of another language — and processes of creole (and pidgin) genesis. This subject is not really distinct from the other topics addressed in the 1995 Amsterdam Creole Workshop: all contact phenomena are intimately connected, so much so that (for instance) constraints proposed for one area are unlikely to be valid if there are counterexamples in another area. Nevertheless, specialists in the different areas approach their research from quite divergent perspectives. Mine is that of a historical linguist. The data I study primarily comprise completed linguistic changes, on the one hand, and attested pidgins and creoles, on the other hand — namely, the results of linguistic events that took place in past contact situations. Successful analyses of these historical results must be consistent with data from ongoing processes such as codeswitching, koineization, and interlanguage formation (and vice versa). My starting point is this: under different social conditions, specifically in diverse language-contact situations, speakers’ linguistic behavior will differ widely. But overall, human linguistic behavior will not differ randomly in terms of strategies adopted for coping with diverse contact situations. This means that (among other things) theories of pidgin/creole genesis that require behavior unique to a particular type of social context will be convincing only if they include an explanation of why such unique behavior arises. I have seen no evidence that any particular contact situations demand unique linguistic behavior. The linguistic results of contact certainly differ dramatically from one situation to the next, but the behavior of individual speakers and groups of speakers does not, as far as I can tell. Comparing processes of ordinary

 Sarah G. Thomason

contact-induced change with pidgin/creole genesis is therefore a potentially fruitful enterprise: as in other scientific fields, a more general explanatory hypothesis — one that covers a broader range of phenomena — is preferable to a more restricted one, so a theory that accounts both for pidgin/creole genesis and for other results of language contact will, all other things being equal, be preferred over a theory that accounts only for pidgin/creole genesis. In the following sections I will argue that social and linguistic features shared by various outcomes of language contact offer some promise for the development of a unified theory of contact-induced language change and the emergence of three kinds of contact language: pidgins, creoles, and two-language mixtures. Section 2 compares borrowing (in a narrow sense) with pidgin/ creole genesis, Section 3 compares shift-induced interference with pidgin/creole genesis, Section 4 compares borrowing with the genesis of twolanguage mixtures, and Section 5 is a brief conclusion.

. Why borrowing is irrelevant to pidgin/creole genesis I believe that, to understand language contact phenomena, it is necessary to distinguish between two quite different processes through which linguistic material is transferred from one language to another. Terrence Kaufman and I have called these two interference processes borrowing and shift-induced interference (see Thomason and Kaufman 1988 for detailed discussion). Borrowing, in our framework, is the incorporation of features by native speakers into their own language from another language. A few hedges are needed to avoid some obvious pitfalls with this definition: the people who do the borrowing could, in principle, be non-native speakers with native-like fluency in the borrowing language, or they could be people with more than one native language, or they could be fully fluent speakers of a fully crystallized trade pidgin that has no native speakers. The crucial point of the definition is that the people who are responsible for the interference are fluent in the borrowing language, so that no question of imperfect learning of that language arises. In shift-induced interference, by contrast, the people who are responsible for the interference are non-native, and in fact not fully fluent, speakers of the language that is undergoing contact-induced change; they have not learned the target language perfectly, and their lack of knowledge results both in carryover of features from their original first language (L1) into the target language (TL) and in the non-appearance in their version of the TL of certain features of the

Contact-induced language change 

TL. Here too a hedge must be added: it is quite possible that shifting speakers might choose to carry over features of their original L1 into their version of the TL even when they have learned the relevant TL structures and know that those features are not part of the TL. That is, they might introduce interference features deliberately, for instance to maintain a grammatical distinction from their original L1 that is lacking in the original TL. In order for the results of imperfect learning to become a fixed part of the TL, either the shifting group must be relatively isolated from the original TL speech community (as with Indian English), or original TL community members must eventually borrow — in my narrow sense — some or all of the interference features that appear in the shifting group’s version of the TL. In the first case only the isolated TL variety will have been influenced by the shifting group’s original L1; in the second case the entire TL, or at least the speech of a subcommunity that includes both original TL speakers and members of the shifting group, will have been influenced by the shifting group’s original L1. My reason for emphasizing this distinction in considering processes of interference is that the two processes are quite different in nature, and my reason for emphasizing the distinction in considering results of interference is that their results also differ, in principle and (in many cases) in fact. The major difference is that borrowing begins with words and may or may not include structural features as well (that depends on the intensity of the contact); shift-induced interference, by contrast, begins with phonology and syntax and may or may not include words. (Here again, see Thomason and Kaufman 1988 for detailed argumentation and examples.) Given this picture, it might seem reasonable at first glance to expect both borrowing and shift-induced interference to be important components of processes of pidgin/creole genesis, since the most salient result of pidgin/creole genesis is a language whose lexicon and structures cannot be traced back to a single parent language — that is, it is a mixed language, from the viewpoint of a historical linguist.1 Borrowing and imperfect group learning also result in languages with lexicon and structure from more than one source, so it looks as if all these things go together in a straightforward way — albeit with differences in degree, since, although all languages contain material from more than one source language, most languages are not mixed in the strong sense. (That is, most languages comprise lexical and structural subsystems that can mostly be traced back to a single parent language, leaving a modest amount of residue that reflects foreign interference.)

 Sarah G. Thomason

But in fact the process of borrowing cannot possibly be a major factor in a process of pidgin/creole genesis. The reason is that borrowing requires, by my definition, an already-existing language into which features from some other language (or languages) can be incorporated. If there is no language, there is nothing to incorporate foreign features into, and no fully fluent speakers to do the incorporating; and since pidgin/creole genesis, also by definition, is a process by which a new language is created, the starting point is not a language.2 After a pidgin or creole exists as a language — a crystallized speech form with its own grammar, which must be learned like any other language — of course things can be borrowed into it; but before the contact language exists, there is no ‘‘there’’ there, no structure to which features from other languages can be added. This is not a mere issue of terminology, a position I’m forced to adopt because of my peculiar definition of ‘‘borrowing’’. It actually has empirical consequences. Processes of pidgin/creole genesis — which, though they haven’t been directly observed, can be inferred in part from the structures of the resulting contact languages and in part from partially analogous processes in second-language acquisition and other types of contact — typically result in languages in which the lexicon comes primarily from a single source, and always result in languages in which the grammar comes neither from the lexifier language nor from any other single language.3 If borrowing were part of the process of pidgin/creole genesis, we would have to posit something like the following scenario. Suppose we start with three languages, A, B, and C, in a new contact situation. Suppose further that, for some social reason(s), A turns out to be the lexifier language for a pidgin/creole-to-be. If borrowing is one (or the only) process through which the new pidgin or creole emerges, then the scenario necessarily involves specific processes that are radically different from borrowing in other kinds of contact situations, i.e. those which do not result in pidgins or creoles. To compare the processes at all, we would need to assume that speakers of all three languages are incorporating features from the other two languages into their L1; some kind of convergence of these three different sets of borrowings would eventually produce a single contact language (except that it’s not at all clear how this convergence would come about, if the separate groups are borrowing things independently). It’s easy enough to see how the lexical part of this scenario would work for B and C speakers, because they could import from A all the lexical items they needed for their versions of the emerging pidgin or creole. But this kind of

Contact-induced language change 

‘‘borrowing’’ would be very different in kind from borrowing in less extreme contact situations, because it would include basic vocabulary from the beginning; in less extreme contact situations, non-basic vocabulary is always borrowed first, and basic lexical items are borrowed only later, if at all. So this would be a special type of borrowing, not obviously comparable to what goes on in more ordinary contact situations. Moreover, If A speakers participate in the creation of the new contact language — they don’t always, especially in abrupt creolization, but sometimes they do, and certainly their participation can’t be ruled out in principle — then they would be borrowing structure without any vocabulary at all, a quite radical departure from what goes on in less extreme borrowing situations (that is, borrowing in situations which do not involve language shift by groups of speakers). And for all speakers, the kinds of structural changes in their native languages would be very different indeed from structural changes that come about through borrowing in ordinary contact situations. The most striking difference is that the incorporated structures (still on my hypothetical scenario) are not very likely to be just like structures in the source languages: A speakers, for instance, do not simply replace or augment their inflectional morphology with B or C morphology, or even with modified B or C morphology. Instead, in the typical pidgin or creole, they end up with no inflectional morphology at all, no matter how complex the inflectional systems of B and C, and of A itself, are. More strikingly, none of the three languages keeps sizable portions of its own original structure intact; the only subsystem of the new contact language that could be said to remain largely intact is the lexicon of A, and even that is likely to be severely distorted phonologically and structurally in the pidgin or creole.4 An additional point is that, if borrowing were the only process involved in pidgin/creole genesis, the result for original B and C speakers would be two different languages, one with A words and B structure and another with A words and C structure. It is possible to imagine such an initial stage, with later grammatical (including phonological) convergence of these separate languages into a single language with one grammar,5 but no such two-step process is attested, as far as I know, in any pidgin or creole community. If there is no direct evidence for such a series of developments, and if an equally successful hypothesis that does not require an additional unattested step is available, then an important methodological criterion of historical linguistics can (and must) be invoked: all other things being equal, the simpler of two historical hypotheses is to be preferred over the more complicated one. Of course, the second ‘if ’

 Sarah G. Thomason

is crucial here; I will argue below that an equally successful hypothesis is indeed available, one that is both simpler and more general than the hypothetical two-step process described above. Overall, then, the problem with comparing borrowing and pidgin/creole genesis is that the structures of prototypical pidgins and creoles do not show the kinds of features, with the kinds of implicational universals, that result from borrowing in contact situations that do not produce pidgins or creoles. So either the processes through which pidgins and creoles emerge are not borrowing or my definition of borrowing is wrong. But redefining borrowing won’t eliminate the need to distinguish these different processes, so I see no advantage in changing the terminology. Focusing on the process of pidgin/creole genesis rather than on the linguistic outcome of the process makes the difference even more evident, though it’s less easy to prove, because we can’t point to direct observations of the creation of a pidgin or a creole. Still, I do think we have to drop at least one of our assumptions about the behavior of speakers of A, B, and C: I do not believe that speakers of each of the three languages will create their own pidgin or creole by incorporating features from the other two languages into their L1. That scenario leaves too little room for the kinds of accommodations that can be observed in contacts between individuals who don’t share a language, and for the creativity that can be observed in individual linguistic behavior both in and out of contact situations. In other words, I think it is much too simplistic to assume that people create a contact language merely by changing their native languages (see note 5 for further discussion). That, in short, is why I don’t believe that borrowing has played an important role in pidgin/creole genesis. In the next section I will consider the type of contact-induced language change that does seem to me to be relevant to a historical account of processes of pidgin/creole genesis.

. Shift-induced interference and pidgin/creole genesis Shift-induced interference, the other major process through which languages influence each other, is more obviously relevant to the emergence of abrupt creoles — creoles that arise quickly, without a fully-crystallized pidgin stage — than to pidgin genesis. So I’ll start with the creoles. To oversimplify somewhat, imagine a new contact community that involves numerous languages and is comprised entirely of monolingual spea-

Contact-induced language change 

kers who (therefore) do not share a lingua franca. Imagine further that there is one socially dominant group in the contact community, and that its members, though powerful, are both illiterate and greatly outnumbered by speakers of other languages. The other people in the contact situation will therefore find it difficult to learn the powerful group’s language as a whole (because access will be too limited), but they can learn some of that group’s vocabulary. Finally, imagine that speakers of the other languages can no longer use their own languages for community-wide purposes (because no one else knows them), and that they must perforce shift away from those languages — immediately if they are truly isolated within the community, later if there are other speakers of their L1 present. This does not mean that they will shift to the dominant group’s language, of course, because they don’t have enough opportunity to learn that language as a whole; but they will have to give up their native languages,6 and this sociolinguistic circumstance makes the process of creole genesis partly analogous to other cases of shift by one group of people to another group’s language. As in more ordinary cases of language shift — that is, cases in which the shifting group does learn the entire target language, though quite possibly with some structural modifications — what happens in pidgin/creole genesis contexts is a process of language learning, specifically, for adults, secondlanguage learning. The special kind of learning that must be posited for abrupt creolization will certainly result in imperfect learning of the lexifier language, the ‘‘TL’’ (which is unlikely to be a true target language for most of the shifting speakers). That’s not all there is to the process: there is also creation of new grammatical structures, through a process of (implicit) negotiation among all the speakers involved in the genesis of the new creole. This part of the process will ensure that (at least in a hypothetical scenario in which all substrate speakers are monolinguals with different L1s) the creators of the creole will not carry over features from their original L1s into the emerging creole unless those features either are very easy to learn (i.e. unmarked) or are shared among most or all of the substrate languages. By contrast, in ordinary shift situations there is no negotiation among shifting speakers, because they all share the same L1; and if they learn the TL imperfectly, they are likely to make similar or identical ‘‘errors’’. These ‘‘errors’’ may include (among other things) marked features of their original L1.7 Not so in creole genesis: unless some particular substrate group is especially prominent in the new contact community, no marked features from substrate languages are likely to turn up in the resulting creole.

 Sarah G. Thomason

Of course real-life abrupt creole genesis is not exactly like my hypothetical scenario: social factors of various kinds, most notably the probable presence in the new contact community of groups of people who share the same L1, skew the picture. But I believe that the idealization bears a close enough relationship to real life that the actual emergence of an abrupt creole can reasonably be viewed as a variation on the same theme. Pidgin genesis is less obviously connected with language shift, since no one is shifting; the prototypical pidgin is strictly a second language. Nevertheless, imperfect learning plays the same kind of role in pidgin genesis as in abrupt creole genesis: the creators of the pidgin do not learn any other language in the contact situation fully (if they did, there would presumably be no need for a contact language — not that linguistic creativity can safely be assumed to be constrained by need). As with abrupt creoles, there is a process of implicit negotiation that involves cross-language compromise and serves to limit the number and kinds of marked features that are likely to appear in the pidgin. Unlike abrupt creoles (at least in my idealization) and cases of ordinary language shift by groups of people, however, new pidgins serve only limited functions. Therefore, many, most, or all the pidgin creators also continue using their native languages, and the various native languages remain available as a potential source of new linguistic material as the contact language crystallizes. One result is a greater number of marked (especially areal) features in the grammar of a prototypical new pidgin than in the grammar of a prototypical new abrupt creole. The salient common factor in ordinary shift-induced interference, creole genesis, and pidgin genesis, then, is imperfect learning of a target language by a group of people.8 There are also significant differences. First, ordinary shiftinduced interference typically (though not necessarily) takes place in a twolanguage contact situation, while both creole and pidgin genesis typically (though not necessarily) involve more than two languages. Second, in ordinary shift-induced interference an entire TL is learned, with relatively minor changes, while in abrupt creole and pidgin genesis no full TL is learned. One consequence of this latter fact is that there is a greater role in pidgin and abrupt creole genesis for creative responses to the need to develop a systematic grammar for the emerging contact language. Pidgin genesis differs from abrupt creole genesis primarily because a new pidgin is used as a second language only; the main linguistic correlate of this difference is a greater number of marked features in typical pidgin grammars than in typical creole grammars.

Contact-induced language change 

. The relationship between linguistic interference and language mixture I believe that, as indicated in section 2 above (and argued in detail in Thomason and Kaufman 1988), there are just two major types of linguistic interference: borrowing (in my narrow sense) and shift-induced interference. These are not mechanisms of interference; rather, they are sociolinguistic processes with linguistic results that are to some extent predictable. I have argued in the preceding sections that borrowing is not relevant to pidgin/creole genesis, but that the imperfect learning which characterizes shiftinduced interference underlies both pidgin genesis and abrupt creole genesis as well as less extreme shift situations. But though borrowing is irrelevant to pidgin/creole genesis, it is directly connected with the third type of language mixture, the emergence of twolanguage mixtures; and this connection is in important ways analogous to the connection between shift-induced interference and the emergence of a new pidgin or an abrupt creole. Crucially, all three types of language mixture involve creativity. In pidgin/creole genesis the results of creativity are seen in the cross-language compromise of the pidgin/creole grammar, and in twolanguage mixtures the results of creativity are seen most obviously in the particular choices of combination — lexicon from Spanish and grammar from Quechua in the Media Lengua, Russian loanwords and finite verb inflection in an otherwise Aleut structure (including nonfinite verb inflection) in Mednyj Aleut, French noun phrases in an otherwise Cree structure in Michif, retention primarily (and apparently, by now, only) of non-Bantu vocabulary in heavily bantuized Ma’a, and so forth.9 With all this creativity, however, all three types of mixed language still reflect the two basic processes of interference: shift-induced interference in pidgins and abrupt creoles, borrowing in two-language mixtures. That is, as noted above, the crucial distinction between imperfect group learning, on the one hand, and full (or at least extensive) bilingualism, on the other, is reflected in the structures of the two basic types of mixed language as well as in less extreme cases of linguistic interference. I have already discussed pidgins and creoles; to show the symmetry and the parallels between type of interference process and type of contact language, I will describe two-language mixtures briefly in this section. The amount of readily available sociohistorical and linguistic information on two-language mixtures is still very limited indeed, so that attempting to generalize about the entire set of languages is risky in the extreme. Neverthe-

 Sarah G. Thomason

less, the few examples seem to fall into two distinct sets: gradually-developing mixtures, in which the process takes place over several centuries and affects all grammatical subsystems in more or less equal measure, until nothing is left but some vocabulary, including much basic vocabulary; and abrupt mixtures, in which the mixing takes place over a short period and is uneven in its effects on the system, often being compartmentalized into a single subsystem, e.g. the lexicon, or the finite verb morphology, or noun phrases (Thomason 1995 presents arguments and evidence in support of this analysis and the sketch in the following paragraphs). In both types of two-language mixture there is likely to be a large amount of borrowing of non-basic vocabulary, from the viewpoint of the dominant language (see below). In both borrowing and two-language mixtures, only two languages are typically involved,10 and there is at least extensive bilingualism on the part of the borrowers or creators of the mixed language. Because of this bilingualism, linguistic material entering the borrowing language or the bilingual mixture is not severely distorted by comparison to corresponding source-language structures. Even in cases that look like deliberate abrupt creations, there seems to be a dominant language for the creators of the mixture, and material from the non-dominant language is incorporated into that L1.11 But, as with ordinary shift-induced interference vs. pidgin/creole genesis, the process by which a two-language mixture arises cannot simply be equated with borrowing. In abrupt two-language mixtures the element of deliberate creativity is more prominent than in most ordinary borrowing situations, and the result is the emergence of a new ethnic group or subgroup as well as a new language. In slow-developing two-language mixtures, by contrast, the conscious or unconscious determination to avoid total assimilation to a socially dominant group is more prominent than in most ordinary borrowing situations: the element of linguistic creativity comes from the choice to preserve vocabulary, especially basic vocabulary, from the original ethnic language.

. Conclusion The point I would like to close with is this. From one perspective, the relationship between ordinary shift-induced change and ordinary borrowing, on the one hand, and pidgin/creole genesis and emerging two-language mixtures, on the other, is a difference in kind: the social situations are different (less ‘‘extreme’’ vs. radically disruptive sociolinguistic circumstances), and so are the linguistic

Contact-induced language change 

results (ordinary, non-disruptive contact-induced changes vs. radical disruptions in normal processes of change). But from another perspective, the one I would prefer to adopt, the difference is one of degree, or rather of varying quantities of the same ingredients: the same social processes (e.g. change from one ethnic group to another and adoption of some of another group’s identifying features) are operative in all kinds of contact-induced language change, and the linguistic processes — creativity as well as actual changes — are also universal in this domain. The most extreme cases — pidgins, creoles, and two-language mixtures — are unusual only in their ways of combining the ingredients.

Notes . See Thomason 1997 for arguments in favor of defining ‘mixed language’ in historical terms, and for a discussion of the crucial notion of fuzzy boundaries in historical linguistics: of course no human language can be traced entirely back to a single parent language; it’s a matter of degree. Some creolists (notably Le Page 1994) have recommended throwing the historical linguist’s baby out with the fuzzy bathwater, but such arguments rest on an imperfect understanding of the principles and especially the methods of historical linguistics. All historical sciences must wrestle with the ‘becoming’ of states, and therefore with indeterminate borderlines between states, e.g. between the end of Latin and the beginning of Old French. And historical scientists such as evolutionary biologists, like historical linguists, must deal with problems of identifying and defining ‘things’ (species, languages) when any sizable real-life data complex presents a much messier picture, one in which criteria apply neatly only some of the time. But successful historical sciences — and historical linguistics is widely acknowledged to be a highly successful one — achieve testable and fruitful results in spite of such complications. The messiness of real data makes the task of unraveling the regularities of historical developments more difficult, but it does not, in the many favorable cases, make it impossible. This situation is, of course, analogous to the difficulties in defining and identifying ‘‘a language’’ in order to describe its grammatical patterns: the existence of grammatical regularities, which leads (for instance) to a high level of predictability of the results of child first-language acquisition in the vast majority of speech communities, is ample justification for abstracting the notion of ‘‘a language’’ from the extreme variability of real-life language use. . Some creolists might still argue that a process of pidgin/creole genesis is merely a process of simplifying a lexifier language, but I don’t think most specialists accept such a scenario nowadays. There are certainly some borderline cases between creole and changed lexifier language (see Thomason 1997 for discussion), but those are not prototypical creoles, and their grammars cannot be fully accounted for by simplification alone. . My comments about creole genesis throughout this chapter pertain to abrupt creolization — the emergence of a creole in a social context that lacks a fully crystallized pidgin — and not to the development of a creole from a fully crystallized pidgin. The nativization

 Sarah G. Thomason

of a well-established expanded pidgin is not, in my opinion, crucially connected with any kind of linguistic interference; it is mainly a social process. (This statement of course begs the question of whether nativization is part of the definition of creolization, but that is a topic for another paper.) . The fuzzy boundary issue complicates this picture too, of course. Languages on the borderline between contact languages and languages that have developed by normal genetic descent, such as, perhaps, Reunionese or acrolectal varieties of certain Caribbean creoles, have sizable portions of some grammatical subsystems of the lexifier language. . This two-step process is close to the historical scenario envisioned by Claire Lefebvre in numerous writings; see, for instance, Lefebvre 1993 and also her chapter in this volume. But Lefebvre has presented no evidence of radically different initial-stage creole grammars to support her proposal that Haitian Creole arose in the first instance through relexification in Fon (and also, presumably, in other relevant African languages, such as the Bantu languages that were also present in the earliest creole-forming community in Haiti) from French, which would be language A in my hypothetical scenario. More seriously, research on second-language learning in general and interlanguage in particular, which is comparable to early pre-pidgin or pre-creole formation, shows that learners do not simply replace native morphemes with target-language morphemes while keeping native-language grammar intact, as Lefebvre’s hypothesis requires. Instead, they develop new grammatical patterns from the start, some of which do not approximate either their L1 or the TL. Several good examples can be found in Ulla-Britt Kotsinas’ chapter on Immigrant Swedish in this volume, for instance the use of the Swedish politeness marker varsågod ‘please’ in sentence-initial position to express an imperative, or the grammatical and semantic extensions in usage of the Swedish verb komma ‘come’. One could, of course, claim that individual linguistic behavior in pidgin/creole genesis contexts is completely unlike individual linguistic behavior in other language-contact/language-learning contexts; but that would be a very strong claim, which would therefore require very strong arguments before it could be convincing. To date, no argumentation or evidence has been presented in support of a claim that speakers behave very differently in contact situations that do give rise to pidgins or creoles and contact situations that don’t. Lefebvre’s focus on relexification as ‘‘the central mental process involved in the creation of creole languages’’ (this volume) does raise a very important question, however: how are the patterns of change and creation in the genesis of a pidgin or creole organized in any one speaker’s brain? But the stilltentative state of research on second-language acquisition, including among other things the continuing controversy about the extent to which adult second-language learners have access to Universal Grammar, seems to me to dictate extreme caution in theorizing about mental processes in former pidgin/creole genesis contexts, which are not available for experimental study. This doesn’t mean that no future progress can be foreseen in this domain: future advances in second-language acquisition theory and the increasing body of information available about speakers’ behavior in current contact situations should eventually make it possible to draw solid inferences about individuals’ behavior in past contact situations, including pidgin/creole genesis situations. . This scenario is of course not original; see e.g. Alleyne 1971 and Thomason and Kaufman (1988: 149–150). It is, however, hypothetical in the sense that there may not have

Contact-induced language change 

been many, and possibly not any, contact settings in which it actually occurred as described. As I will argue below, however, the basic scenario is relevant to pidgin genesis as well, in spite of the complicating factors that must be taken into account there, most obviously the absence of language shift. The unifying thread is group second-language acquisition in a (typically) multilingual social context that does not encourage, or in some cases permit, full acquisition of a TL grammar. . This isn’t the whole story for ordinary shift-induced interference. In some cases, as noted above in section 2, the version of the TL spoken by members of the shifting group is learned by original members of the TL speech community, which means that the interference features introduced by shifting speakers will be fixed in the TL as a whole. A more complex developmental process is probably much more common, however: original TL speakers acquire some, but not all, of the shifting group’s interference features, and the interference features that they do acquire may be modified rather than identical to those in the shifting speakers’ version of the TL. In such a case, although the TL has indeed been changed through the influence of imperfect learning by the shifting speakers, the final linguistic outcome differs from the shifting group’s learners’ errors as well as from the original TL. The last step in this complex sequence of events — the accommodation between the shifters’ new version of the TL and the original TL speakers’ original version — resembles koinéization (see Jeff Siegel’s chapter in this volume) rather than contact-induced language change involving sharply divergent linguistic systems. . The term ‘target language’ must be interpreted very loosely here, to include partial targets (in practice, lexicon only) as well as full targets in the sense of effort on the part of the learners to acquire the entire TL. As has often been pointed out (see e.g. Thomason and Kaufman 1988: 152, 352, n. 4), there is no good reason to suppose that the lexifier languages were full TLs in most pidgin/creole genesis contexts. . It will be evident from this formulation that I do not accept Peter Bakker’s definition of two-language mixtures as languages with ‘‘the grammar from one language and the lexicon from another’’ (1994: 25; see also Anthony Grant’s definition, this volume). Attested twolanguage mixtures display more diversity in their linguistic make-up than Bakker’s definition allows for, and I have not seen any convincing arguments that linguistic creativity in this domain is (almost) always limited to dividing lexicon from everything else. Bakker acknowledges that Mednyj Aleut is an exception to his generalization, but he feels that Michif is not in fact significantly different from other two-language mixtures. His argument here is that, because of the polysynthetic nature of Cree verbal structure, it is ‘‘‘impossible to combine [Cree grammar with French lexicon] in the same way as in the other cases’’ (1994: 21). But this argument is weakened by his own examples (examples 1 and 2, p. 20), which show the insertion of French and English stems into a Cree verbal complex; if it can happen in these cases, it can happen elsewhere too. Moreover, the French noun phrases in Michif are neither characterized exclusively by Cree grammar nor grammatically inert: they have active French grammatical features, such as masculine vs. feminine gender. Michif has both French and Cree lexicon and both French and Cree grammatical features, though the Cree grammatical features predominate. . It does seem odd to add the hedge ‘‘typically’’ to the statement that a two-language mixture involves only two languages. I maintain the term ‘‘two-language mixture’’ in order

 Sarah G. Thomason

to avoid even clumsier locutions; but the crucial point is extensive bilingualism, not the number of languages, and in principle more than two languages could be present, with extensive multilingualism instead of mere bilingualism on the part of the creators of the contact language. In such a case, I would expect the genesis process to resemble borrowing rather than shift-induced interference. I know of no examples of such cases, but they must be allowed for in principle. As for borrowing in less extreme contact situations, it does occur in multilingual rather than mere bilingual contexts; again, the relevant factor is extensive bilingualism/multilingualism, as opposed to monolingualism. . Unfortunately, ‘‘dominant’’ is not an easy term to define in a non-circular way. This observation about one language being dominant therefore cannot, at least for now, be elevated into a theoretical principle. And it may turn out to be wrong: I know of no principled means of excluding the possibility that a group of speakers would create a mixed language with evenly balanced components from each of two source languages.

References Alleyne, Mervyn C. 1971. ‘‘Acculturation and the cultural matrix of creolization’’. In Pidginization and Creolization of Languages, Dell H. Hymes (ed.), 169–186. Cambridge: Cambridge University Press. Bakker, Peter. 1994. ‘‘Michif, the Cree-French mixed language of the Métis buffalo hunters in Canada’’. In Mixed Languages, Peter Bakker and Maarten Mous (eds.), 13–33. Amsterdam: Institute for Functional Research into Language and Language Use. Lefebvre, Claire. 1993. ‘‘The role of relexification and syntactic reanalysis in Haitian Creole: Methodological aspects of a research program’’. In Africanisms in Afro-American Language Varieties, Salikoko S. Mufwene (ed.), 254–279. Athens: University of Georgia Press. Le Page, R.B. 1994. ‘‘The notion of ‘linguistic system’ revisited’’. International Journal of the Sociology of Language 109: 109–120. Thomason, Sarah Grey. 1995. ‘‘Language mixture: Ordinary processes, extraordinary results’’. In Spanish in Four Continents: Studies in language contact and bilingualism, Carmen Silva-Corvalán (ed.), 15–33. Washington, DC: Georgetown University Press. Thomason, Sarah Grey. 1997. ‘‘A typology of contact languages’’. In The Structure and Status of Pidgins and Creoles, Arthur K. Spears and Donald Winford (eds.), 71–88. Amsterdam: John Benjamins. Thomason, Sarah Grey, and Terrence Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley, CA: University of California Press.

Chapter 11

Yiddish as a contact language Ellen Prince

.

Overview

By any conventional definition of ‘‘creole language’’, Yiddish is not a creole, not in its genesis and not in its form. For one thing, while there is some disagreement on exactly where the original Yiddish speakers came from (Weinreich 1980 vs. Faber and King 1984), there is no evidence whatsoever that they were from diverse geographic and linguistic backgrounds and there is no evidence whatsoever that there was ever a pidgin stage. On the contrary, Modern Yiddish is clearly descended from Old Yiddish, a highly inflected sister language of Middle High German.1 In fact, the line between ‘‘Judeo-German’’, i.e. Middle High German with a ‘‘Jewish accent’’, and Old Yiddish, a separate language from Middle High German, is often not clear. At the same time, Yiddish has, since its inception, been a language in contact. In fact, it is safe to assume that the original speakers, no matter where they hailed from, were in a language contact situation before the development of Yiddish insofar as, while they clearly did not hail directly from their original homeland in the Middle East, they had always maintained some degree of Semitic contact wherever they went. Once arrived in Germanic-speaking lands, this Semitic contact remained in varying degrees, as it has to the present. Furthermore, when they began to migrate east several centuries later, new contacts arose with Balto-Slavic, Hungarian, and Romanian, the result of which was Modern (Eastern) Yiddish. Not surprisingly, the results of these diverse contacts are manifest in Yiddish at virtually every level of linguistic description. While presumably every language has in some way been influenced by some other languages, Yiddish ‘‘wears its history on its sleeve’’, so to speak, and lends itself remarkably well to studies of contact effects. (I do not believe it is a coincidence that Uriel Weinreich was both a Yiddishist and a theorist on contact phenomena!) Thus it provides a fertile field for investigating what sorts of changes occur in a language in contact.2

 Ellen Prince

In this chapter, I shall first sketch what I take to be the motivation for the vast majority of contact effects in Yiddish: not an imperfect competence in the Germanic system on the part of the speakers but a particular semantic or pragmatic intent: to exploit the formal possibilities of Yiddish in order to express in it concepts which were expressed in some different but ‘‘analogous’’ way in the contact languages. Furthermore, there is no reason to think that Yiddish is not unique: quite possibly, the overwhelming majority of cases of language contact phenomena involve importing new meanings to old forms, and the borrowing of new forms, at least at the syntactic level, is highly unlikely. Therefore, after giving a brief overview of the types of contact phenomena found, I shall look more closely at a few of these, in an effort to see more clearly what exactly has changed and why the changes may be seen as having had a semantic/pragmatic motivation. Finally, I shall discuss a case where the form may have in fact changed as a result of language contact, but where I believe the contact effect on form is indirect, the direct contactinspired change being at the semantic level, the formal change being languageinternal, following a change in the distribution and frequency of the affected form due to the semantic change. . Motivation for contact effects in Yiddish First, there is the issue of whether the contact effects evident in Yiddish were the result of the Yiddish speakers’ imperfect competence in Germanic causing them to misanalyze the Germanic data they heard, or, on the other hand, were the result of their having competence in both (Germanic) Yiddish and some other language and importing items from that other language into Yiddish fully in accordance with the grammar of Yiddish. I believe the second is the case in that, in the vast majority of cases, we must conclude that the Yiddish speakers responsible for the various changes were fully competent Germanic speakers who were simply borrowing items from other languages they knew, all the while maintaining and fitting these items into the Germanic system. . Overview of contact effects While, at first blush, Yiddish appears to show effects of contact at every level of linguistic description, we find that, if we distinguish primary effects, i.e. the cases of direct influence, from secondary effects, internal changes within Yiddish subsequent to these primary effects, the primary effects — the real

Yiddish as a contact language 

effects of contact — are in fact restricted in very large part to two levels, lexical and semantic/pragmatic.3 Not coincidentally I believe, these two levels form a natural class: they are the levels at which form-meaning correspondences are specified.4 That is, the lexical level pairs phonological representations of morphemes with lexical meanings; the semantic/pragmatic level pairs syntactic representations of phrases/clauses with discourse understandings.

. Lexical changes due to contact Not surprisingly, Yiddish shows massive lexical borrowing. Traditional estimates are that 20% of the vocabulary are Semitic in origin, 20% Slavic, plus a handful of 3 items from other languages. As would be expected, the borrowings are largely of open class items and abound in the expected domains: e.g. Semitic in religious domains and Slavic in food domains. At the same time, they appear in other domains as well, exemplified in (1a), and also in closed class items, exemplified in (1b): (1) a.

Semitic: ponim ‘face’, gvir ‘rich man’, meshuge ‘crazy’, moyre ‘fear’. Slavic: tate ‘papa’, khmare ‘cloud’, modne ‘strange’, khapn ‘grab’. b. Semitic: beshas ‘while’, afile ‘even’, iz ‘so’, lemoshl ‘for example’. Slavic: tsi ‘whether’, abi ‘so long as’, khotsh ‘although’, i . . . i . . . ‘both . . . and . . .’

Furthermore, there is some borrowed bound morphology from Slavic, all derivational and nearly all affective, exemplified in (2a), and one Semitic inflectional suffix, the masculine plural -im, on a handful of Germanic nouns that were presumably reanalyzed as Semitic, shown in (2b):5 (2) a.

-inke -nik -atsh -zhe

adj. diminutive marker, e.g. tayerinke ‘dear-dim’ agentive marker, e.g. shlimazlnik ‘never-do-well’ pejorative marker, e.g. yungatsh ‘brat’ intensifying marker on imperatives and interrogative Wh-words (e.g. Zayt-zhe nisht kayn nar ‘Don’t be a fool!’, Vu-zhe bin ikh? ‘Where on earth am I?’) b. nar naronim~naroim ‘fool’ poyer poyerim ‘peasant’ doktor doktoyrim ‘doctor’ tayvl tayvolim ‘devil’ faktor faktoyrim ‘broker’

 Ellen Prince

Interestingly, there is no borrowed inflectional verbal morphology. In fact, the borrowing of verbs is complex and takes several forms. The simplest — and rarest — is the borrowing of a Semitic root plus vocalic pattern or of a Slavic stem as a verb stem, exemplified in (3a). More often, a verbal infix is affixed to a Semitic root plus vocalic pattern or to a Slavic stem to form a verb stem, exemplified in (3b). For Semitic, however, the most common pattern is where a nominal form of the target Semitic verb is borrowed as the complement of a Germanic helping verb, the resulting periphrastic verb having the syntax of Germanic verbs plus separable prefixes, as shown in (3c): (3) a.

Semitic peyger-n kholem-en zikh shekhtn Slavic khap-n nudzhe-n tulye-n b. Semitic harg-en-en Slavic zhal-ev-en rat-ev-en hor-ev-en c. Semitic mekane zayn moykhl zayn maskim zayn khasene hobn moyre hobn poter vern

‘die (as of animals)’ ‘dream’ ‘slaughter’ ‘grab’ ‘bother’ ‘cuddle’ ‘kill’ ‘regret’ ‘save’ ‘toil’ ‘envy’ ‘forgive’ ‘agree’ ‘marry’ ‘fear’ ‘be rid of ’

Perhaps the most interesting type of lexical borrowing involves the borrowing of features, not of forms. By this I mean the pairing of Germanic closed class items with the ‘‘analogous’’ Slavic ones, to be used ‘‘analogously’’ to the Slavic, e.g. the phenomenon exemplified by the extension of the originally thirdperson reflexive pronoun to all persons under the influence of the invariant Slavic reflexive pronoun, illustrated in (4a), the pairing of certain Germanic separable verbal prefixes with Slavic aspectual prefixes and used analogously, (Talmy 1982), illustrated in (4b–c), and the pairing of the Germanic inanimate neuter Wh-word with the Slavic and used as the factive that-complementizer, analogous to Slavic, illustrated in (4d).6 I shall return to this matching phenomenon below.

Yiddish as a contact language 

(4) a.

Earlier: Ikh vash mikh Later: Ikh vash zikh ‘I wash self’ Ya st¹irayu-s¹ b. Earlier: hern ‘hear’ Later: hern; onhern zikh ‘hear a lot of ’ na-slishat¹-s¹a c. Earlier: lakhn ‘laugh’ Later: lakhn; onlakhn zikh ‘laugh one’s fill’ na˛smia˛c sie˛ d. Earlier: S’iz gut az er kumt. Later: S’iz gut vos er kumt. ‘It’s good that he’s coming.’ Khorosho shto on prikhod´it.

Slavic: Russian

Slavic: Russian

Slavic: Polish

Slavic: Russian

In sum, there is a great deal of lexical borrowing in Yiddish, which I have divided into two types, the borrowing of a form and the borrowing of a feature. The borrowing of a form is most common among open class words, though it does occur in certain closed class items, including derivational morphology, and in one extremely limited inflectional affix in the nominal paradigm. Verbal inflectional morphology is entirely native in form. In contrast, feature-borrowing appears concentrated in the closed class lexicon and often in paradigmatic material, but here again, when verbal morphology is considered, we find feature-borrowing limited to derivational items.

. Putative syntactic changes due to contact I shall now turn to some contact effects that have been or could be construed as syntactic. However, as I shall try to show, all can be seen as occurring at the level of discourse/pragmatics, any syntactic change being in fact internal to Yiddish and secondary. . The clearest case: Dos-sentences Consider the Yiddish sentence in (5):

 Ellen Prince

(5) Dos hot Leyb gezen Erike-n. this-neut has Leonard-masc.nom seen Erica-fem.acc ‘It’s Leonard who saw Erica.’

The dos-sentence in (5) is composed entirely of Germanic morphemes; however, so far as I know, it has no obvious analog in any other Germanic language; in fact, there is no attested correlate even in Old Yiddish, the Yiddish spoken in the Middle Ages by Jews living in German-speaking lands before migrating to the Slavic-speaking lands of Eastern Europe. Now consider the Russian sentence in (6): (6) Eto L¹eon¹id uv¹id¹el Er¹iku. this-neut Leonard-masc.nom saw Erica-fem.acc ‘It’s Leonard who saw Erica.’

The Russian eto-sentence in (6), which has analogs in other Slavic languages, has a great deal in common with the Yiddish sentence in (5): both are simple sentences (i.e. without subordinate clauses),7 each has an expletive NP in initial position, in both cases the expletive NP is the neuter demonstrative pronoun, and both are translated by the English it-cleft. Thus we have what seems to be a case of borrowing — but what exactly has been borrowed? While traditional Yiddishists considered this a case of syntactic borrowing, it is perhaps better described as the borrowing of a discourse function, here originally associated with some Slavic syntactic form and then borrowed into Yiddish, where it becomes associated with some already existing Yiddish syntactic form, distinct from the Slavic syntactic form. To see that the Slavic and Yiddish syntactic forms are in fact distinct, we must consider briefly the syntax of each. First, let us consider the syntax of the Yiddish dos-sentence in (5). Typical of Germanic, Yiddish obeys the Verb-Second Constraint, whereby, in canonical declarative sentences, the finite verb must occupy the second position in the sentence. Thus, if nothing is topicalized, for example, the subject may occur preverbally, as in (7a), but, if something is topicalized, the subject must occur after the finite verb, as seen in (7b, c):8 (7) a.

(Ikh veys az) Leyb hot gezen Eriken. (I know that) Leonard has seen Erica b. (Ikh veys az) nokhdem hot Leyb gezen Eriken. c. *(Ikh veys az) nokhdem Leyb hot gezen Eriken.

Note that Yiddish, unlike other continental Germanic languages but like

Yiddish as a contact language 

Icelandic, obeys the Verb-Second Constraint in subordinate as well as main clauses (Santorini 1989, among others). Second, it is relevant to consider the nature of the position occupied by the expletive NP dos in (5) above. Two plausible possibilities are some complementizer position or some ‘‘Topic’’ position. However, when we consider sentences like (8), where the sentence is embedded and follows an overt complementizer, we see that the expletive NP cannot itself be in the complementizer position: (8) Ikh veys az dos hot Leyb gezen Eriken. I know that this has Leonard seen Erica ‘I know that it’s Leonard who saw Erica.’

Thus, we infer a structure roughly on the order of (9) for the sentence in (5):9 (9) [CP [C’ [IP dos [I’ hoti [VP Leyb ti gezen Eriken]]]]]

Turning now to the syntax of the Russian eto-sentence in (6), we find two important differences. First, note that the finite verb is indeed in third position. That is, the subject remains in preverbal position as in the canonical variant of (6), shown in (10), which is not surprising, since Russian, unlike Yiddish, is not a Verb-Second language: (10) L¹eon¹id uv¹id¹el Er¹iku. Leonard saw Erica

Similarly, note that the verb would be in third position if something, e.g. an adverb, were topicalized to first position, as in (11): (11) Potom L¹eon¹id uv¹id¹el Er¹iku. then Leonard saw Erica

Second and more surprisingly, if we check the possibility of the eto-sentence in (6) in an embedded context, we find that it is ungrammatical, as shown in (12), suggesting that the demonstrative expletive NP eto itself occupies the complementizer position:10 (12) *Ya znayu shto eto L¹eon¹id uv¹id¹el Eriku. I know that this Leonard saw Erica ‘I know that it’s Leonard that saw Erica.’

Until now, we have considered only focused subjects. However, it turns out that Russian eto-sentences may syntactically mark any constituent as focus, not just the subject, as shown in (13):

 Ellen Prince

(13) Eto Eri¹iku L¹eon¹id uv¹id¹el. this-neut Erica-fem.acc Leonard-masc.nom saw ‘It’s Erica that Leonard saw.’

In (13), we see that the focused constituent is in fact marked by being topicalized, although of course string-vacuously in the case of focused subjects. Thus, possible structures for the Russian eto-sentence in (6) and (13) are ones along the lines of (14a) and (14b), respectively:11 (14) a. [CP [C¹ eto [CP L¹eon¹idj [C¹ Ø [IP tj [I¹ uv¹id¹eli [VP tj ti Er¹ikuk ]]]]]]] b. [CP [C¹ eto [CP Er¹ikuk [C¹ Ø [IP L¹eon¹idj [I¹ uv¹id¹eli [VP tj ti tk ]]]]]]]

To return for a moment to Yiddish, we note another syntactic difference here, one that is entirely predictable from the differences between the syntax of the two languages: whereas any constituent may be focused in Russian eto-sentences, only the subject may be syntactically focused in Yiddish dos-sentences, as shown in (15):12 (15) a.

Dos Eriken hot Leyb gezen. this Erica has Leonard seen b. *Dos hot Eriken Leyb gezen.

This difference between Russian and Yiddish of course follows from the differences in their syntax: since the expletive fills the landing site for topicalized constituents in Yiddish, nothing else may be topicalized, while, in the Russian case, the expletive occupies Comp, leaving the landing site for topicalized constituents free to accommodate them. I shall now return to the question of what exactly has been borrowed in this case. Clearly, there has been no obvious syntactic borrowing, since the syntax of Yiddish dos-sentences is not at all the same as the syntax of Russian eto-sentences. At the same time, there were no dos-sentences in Yiddish prior to Slavic contact. From where then did they come? In fact, it turns out that, while dos-sentences as such are of recent vintage in Yiddish, their syntax is entirely native to Yiddish and is well-entrenched in Germanic in general. Consider another syntactic form, the Yiddish es-sentence, shown in (16), cognate of English there-sentences and of the even more similar German es-sentences: (16) a.

(Ikh veys az) es hot Leyb gezen Eriken. (I know that) it has Leonard seen Erica ‘Leonard saw Erica’ or ‘There saw Leonard Erica.’

Yiddish as a contact language 

b. [CP [C’ [IP es [I’ hoti [VP Leyb ti gezen Erica]]]]

A comparison of the syntactic structure of Yiddish es-sentence in (16) with that of Yiddish dos-sentence in (5) and (9) shows that the two are structurally identical. The only difference is a lexical one: the lexical realization of the initial expletive NP is the neuter singular personal pronoun in the es-sentences, while it is the neuter singular demonstrative pronoun in the dossentences. Thus, there has in fact been a formal change, but the change was in the lexicon, not in the syntax: dos ‘this/that’ has acquired an expletive feature.13 Thus there has been a lexical borrowing of sorts, the borrowing from Slavic into Yiddish of a lexical feature associated with the lexical item having the sense of the neuter singular demonstrative pronoun.14 Of course, concomitant with — and as the motivation for — this lexical change, there has been a pragmatic change, the borrowing of a discourse function: the discourse function of Slavic eto-sentences was borrowed into Yiddish and associated with a native Yiddish syntactic form, albeit one crucially containing the result of a lexical borrowing. That is, associated with Slavic etosentences is the discourse function of marking the proposition conveyed as a focus/focus-frame information structure, where the focus-frame represents an open proposition assumed to be known to the hearer and where the focus supplies the (new) instantiation of the variable in that open proposition. In more familiar terms, we have here more or less the discourse function associated with English it-clefts (Chomsky 1971, Prince 1978, Horn 1981, Delin 1994, among others), though, pace Gundel (1977), not at all their syntactic structure. And this is the discourse function associated with Yiddish dos-sentences. This is sketched in (17): (17)

Discourse function of Slavic eto-sentences and Yiddish dos-sentences: Slavic eto-sentences and Yiddish dos-sentences structure the proposition they represent into a focus and a focus-frame, where the focusframe is an open proposition and the focus identifies the instantiation of its variable. b. In eto-sentences, the focus is represented by the topicalized constituent. In dos-sentences, the focus is represented by the subject or, for at least some speakers, any other constituent that is prosodically prominent. c. The open proposition is marked as being assumed to be already known to the hearer, and the instantiation of its variable is marked as being new to the hearer.

a.

 Ellen Prince

Thus, the Yiddish dos-sentence in (5) and the Russian eto-sentence in (6) are marked by their form as having the information-structure sketched in (18): (18) Information-structure for (5) and (6): Focus-frame, known: X saw Erica. Focus, new: X = Leonard

One might now ask what sort of discourse function is associated with Yiddish es-sentences, the syntactic source for dos-sentences. Interestingly, it is unrelated to the discourse function of dos/eto-sentences. As I have argued elsewhere (Prince 1988), Yiddish es-sentences have the discourse function sketched in (19): (19) Discourse function of Yiddish es-sentences: Yiddish es-sentences mark the their subject NP as not representing an entity that is a member of the preceding set of Forward-looking Centers (Grosz, Joshi, and Weinstein 1983, among others), i.e. that has already been evoked in the current discourse-segment.

From this it follows that es-sentences but not dos-sentences may (and in fact often do) occur with no special prior context, as in (20a), since es-sentences, unlike dos-sentences, do not mark a focus/focus-frame information-structure, and it also follows that dos-sentences but not es-sentences may have a pronominal subject, as in (20b), since dos-sentences, unlike es-sentences, may have a subject which has already been evoked in the current discourse segment:15 (20) a.

Men redt take in Ginshprik un Reb Todres redt in Berlin un me hert. Ir farshteyt? Es/#Dos geyt alts azoy geshvint, az me ken gor nit gloybn. (RP:237) ‘You’re talking in Ginshprik and Mr. Todres is talking in Berlin and you hear [him]. You understand? Everything is going so fast/#It’s everything that’s going so fast that you just can’t believe [it].’ b. Dos/#Es shlogst du di puter? (GF) ‘It’s you that’s churning the butter?/#There churn you the butter.’

In sum, I have tried to show that Yiddish has borrowed from Slavic a discourse function, the structuring of a proposition into a focus and focus-frame functionally equivalent to English it-clefts. In Slavic, this discourse function is associated with a certain syntactic form, one in which an expletive demonstrative NP appears to occupy the complementizer position and in which the focused NP appears to be moved to the ‘‘topic’’ position; in Yiddish, the same discourse function has come to be associated with a grossly different syntactic

Yiddish as a contact language 

form, one in which an expletive demonstrative NP appears to occupy the ‘‘topic’’ position and where the focused NP presumably remains in situ within the clause, having undergone no special movement. Elsewhere (Prince 1992a), I have sketched a possible account of how speakers in a contact situation ‘‘match up’’ syntactic forms in the two languages; suffice it to say here that, however it is to be accounted for, such matching occurs and is, I believe, the driving force behind a good deal of contact effects. . More speculative: [NP NP-NP] I should now like to turn to a case where I believe there has been semantic borrowing posing as a syntactic one but where the evidence is less clear. Consider (21): (21) a.

a sheyne meydl ‘a pretty girl’ a fayner bokher ‘a fine lad’ di grine oygn ‘the green eyes’ b. a meydl a sheyne a bokher a fayner di oygn di grine

Alongside of the expected prenominal adjectives in (21a), Yiddish has postnominal adjectival modification of the form NP-NP, where the second NP is formed from the adjective. The whole NP is intoned as a single constituent, without a pause or break between the two constituent NPs, akin to English Catherine the Great or Nathan the Wise. However, unlike them, it is not limited to having a proper name as the first NP. In all the cases I have found, such postnominal modification is restrictive rather than appositive, though prenominal modification may be either. Not surprisingly, any NP may in fact occur postnominally, whether it is formed from an adjective or not, so long as it can be understood restrictively, as illustrated in (22): (22) eyner a yid ‘one [who is] a Jew/guy’ a yid a melamed ‘a Jew/guy [who is] a teacher’ a melamed a kabstn ‘a teacher [who is] a pauper’ eyner a yid a melamed a kabstn ‘one [who is] a Jew/guy [who is] a teacher [who is] a pauper’

 Ellen Prince

So far as I know, this is not found in other Germanic languages. However, note that NPs consisting of [NP NP] are indeed well-formed in Germanic; in addition to the Catherine the Great-type, Germanic has so-called appositive genitives, which are also grammatical in Yiddish, as shown in (23): (23) a shtik fleysh a glezl vaser der taytsh isho

‘a piece [of] meat’ ‘a glass [of] water’ ‘the translation [into Yiddish] [of] isho’

The obvious question is why early Yiddish speakers would have used the syntax of Catherine the Great-type NPs or appositive genitives for ordinary restrictive modification of common nouns. One extremely speculative — but entirely plausible — answer might be the influence of Judeo-French, the language that is conventionally taken to be the early Yiddish speakers’ native language immediately before settling in German-speaking lands (Weinreich 1980, but see also Faber and King 1984).16 That is, French has both prenominal and postnominal adjectival modification, with prenominal typically being appositive and postnominal restrictive. Further support for this comes from the fact that certain restrictive adjectives which always occur prenominally in French also do in Yiddish, exemplified in (24): (24) a.

l’autre homme ‘the other man’ la première fois ‘the first time’ la seule personne ‘the only person’ la vraie histoire ‘the real story’ b. der anderer man ‘the other man’ dem ershtn mol ‘the first time’ der eyntsiker mentsh ‘the only person’ di emese mayse ‘the real story’

*l’homme autre

French

*la fois première *la personne seule l’histoire vraie ‘the true story’ ?*der man der anderer

Yiddish

*dem mol dem ershtn *der mentsh der eyntsiker di mayse di emes(dik)e ‘the true story’

Thus I should like to propose the following very speculative story. Perhaps the early Yiddish speakers wished to make the restrictive/appositive modifier

Yiddish as a contact language 

distinction they had made in Judeo-French and wished to make it positionally, as they had in Judeo-French. In principle, if all sorts of borrowing are equally possible, they could have simply borrowed the Judeo-French syntax, producing phrases like those in (25): (25) *a meydl sheyne *a bokher fayner *di oygn grine

‘a pretty girl’ ‘a fine lad’ ‘the green eyes

However, some things are much more ‘borrowable’ than others, with syntax being borrowable only with the greatest difficulty, if at all. What the early Yiddish speakers may have done instead was to attempt to approximate the positional difference they were seeking by means of a form already grammatical in Germanic, the NP-NP form found in both the Catherine the Greattype NPs and also in appositive genitive NPs. Interestingly, that this was a logically possible form for ordinary modified NPs would have been reinforced by their knowledge of Hebrew, which in fact has postnominal modification of the form NP-NP as its normal form of modification, illustrated in (26): (26) a.

na’arah yafah ‘enayim yeruqot b. ha-na’arah ha-yafah ha-’enayim ha-yeruqot

[girl pretty] [eyes green] [the-girl the-pretty] [the-eyes the-green]

‘a pretty girl’ ‘green eyes’ ‘the pretty girl’ ‘the green eyes’

Note that, while the phrases in (26a) look like simple bare nouns with a following adjective, in fact it is reasonable to parse them as NP-NP, since the Hebrew indefinite article is null; if the noun phrase is definite, the NP-NP structure is unambiguous, as seen in (26b). Thus it is possible that the early Yiddish speakers, aiming for Romance-type postnominal modification for restrictive adjectives and hearing Catherine the Great-type NPs in Germanic and of course not hearing any negative data, assumed that this was a Germanic option for postnominal modification even with common noun NP heads, this conclusion supported by their knowledge that such was the case in Hebrew. Note that the Germanic appositive genitive NP-NP would have simply provided another piece of evidence that Germanic was like both Romance and Hebrew, since Old French and therefore presumably JudeoFrench and also Hebrew had appositive genitives of the form NP-NP, as in (27):

 Ellen Prince

(27) Hôtel Dieu bet el kos mayim yom kipur rosh khodesh

[house God] [house God] [glass water] [day atonement] [head month]

‘house of God’ French ‘house of God’ Hebrew ‘a glass of water’ ‘day of atonement’ ‘head of the month, i.e. first day of the month’

. Even more speculative: Es-sentences The third case of what I take to be putative syntactic change and actual semantic/pragmatic change involves postposed subjects, illustrated in (28): (28) a.

Es iz geshtorbn Truman. it is died Truman ‘Truman died.’ b. Es derlangt a fal arop a driter rod! it serves a fall down a third wheel ‘A third wheel suddenly falls off!’ c. Iz eyn mol, iz tsun im gekumen a yid, azoy farnakhtlekh, ven es so one time is to him come a guy, so late, when it hobn already burned the lamps have shoyn gebrent di lompn. ‘So once a guy came to him late at night, when the lamps were already lit.’ d. Es iz nit aroysgekumen keyn bilet. it is not out-come no ticket ‘No ticket came out.’

Note, canonical word order would have the subject in initial position, so long as nothing else occupies that position, as in (29): (29) a.

Truman iz geshtorbn. ‘Truman died.’ b. A driter rod derlangt a fal arop! ‘A third wheel suddenly falls off!’ c. Iz eyn mol, iz tsun im gekumen a yid, azoy farnakhtlekh, ven di lompn hobn shoyn gebrent. ‘So once a guy came to him late at night, when the lamps were already lit.’

Yiddish as a contact language 

d. Keyn bilet iz nit aroysgekumen. ‘No ticket came out.’

A second possibility is for the subject to occupy ‘‘Middle Field’’, the position immediately following the inflected verb; this is the usual subject position when something has been topicalized in V/2 Germanic languages: (30) a.

Plutsim iz Truman geshtorbn. suddenly is Truman died ‘Suddenly Truman died.’ b. Nokhdem derlangt a driter rod a fal arop! after-that serves a third wheel a fall down ‘After that third wheel suddenly falls off!’ c. Iz eyn mol, iz a yid tsun im gekumen, azoy farnakhtlekh, ven so one time is a guy to him come, so late, when dortn hobn shoyn di lompn gebrent. there have already the lamps burned ‘So once a guy came to him late at night, when the lamps were already lit.’ d. Avade iz keyn bilet nit aroysgekumen. of.course is no ticket not out-come ‘Of course no ticket came out.’

As I have argued elsewhere (Prince 1988), while Initial- and Middle-Field subjects are always grammatical, postposed subjects are extremely common when the subject NP does not represent an entity which has been already evoked in the current discourse segment, either because the NP represents a discourse entity that is new in that segment, as in (28a–c), or because it represents no discourse entity, as in (28d). Note that the type of verb that may have postposed subjects is not limited to existential or presentational. To my knowledge, no other Germanic language has quite this sort of subject-postposing. The question then arises: where did Yiddish subject-postposing come from? Here I am being even more speculative than before, but I believe that the story may be similar. More specifically, we know that Biblical Hebrew was VSO, with rampant topicalization into initial position of those subjects that were already evoked in the discourse segment (Moshavi, in prep.; cf. Shumacher 1980 for an analogous account of Arabic). Furthermore, it is widely known that Old French had rampant subject-postposing under similar discourse conditions, i.e. when the subject was new in the discourse segment. If this was also the case for Judeo-French, the early Yiddish speakers may have wished to convey that a

 Ellen Prince

subject was not Discourse-old (Prince 1992b) by having it occur postverbally. And we have some evidence that they in fact did. As pointed out to me by Beatrice Santorini, many old Yiddish texts begin with or include some Hebrew material, immediately followed by its Yiddish translation, and many are of the form of (31), where the uppercase items are in Hebrew, the rest in Old Yiddish: (31) AMRU KHKHMIM Es hbn gizagit unzri khkhmim in der tseyt ds SAID-3pl SAGES It have said our sages in the time that da iz vardn gibarn abrhm . . . prt is become born Abraham . . . ‘THE SAGES SAID. Our sages said in the time that Abraham was born . . .’ (Magen Abraham (Anon.), 1624, Lublin, p. 2.)

The obvious question here, of course, is why they did not instead choose Middle-Field position for Discourse-new subjects. The answer to that is, I believe, equally obvious, if one looks at the data: Germanic Middle-Field subjects are overwhelmingly Discourse-old and are in fact, at least in the Yiddish data, nearly always pronominal. Thus Middle-Field already had some clear and conflicting function. Furthermore, if the early Yiddish speakers were to have changed that function to marking Discourse-new subjects, they would have had no position for Discourse-old subjects in Topicalized sentences, given that only one constituent may occur preverbally in a V/2 language.17 Thus FinalField seems like a good choice for Discourse-new subjects, akin to Semitic and possibly Romance postverbal Discourse-new subjects. However, now we must ask whether having subjects in Final-Field constitutes a syntactic change in Yiddish. I believe it does not, as I shall now try to show. First, we know that Germanic objects could be postposed when ‘‘focused’’; we find this in Old English and earlier stages of New High German (Pintzuk 1991 and Kroch, personal communication). And we likewise find this, albeit rarely, in Yiddish, illustrated in (32), where finfuntsvantik kopikes ‘25 kopeks’ is focal information and is postposed past the postposed subject: (32) Frier hot a kvort bronfn gekost zibn kopikes. Dernokh hot men earlier has a quart booze cost seven kopeks. Later has one ayngefirt an aktsiz oyf bronfn, un es hot ongehoybn kostn a kvort introduced a tax on booze, and it has begun cost a quart bronfn finfuntsvantsik kopikes. (RP: 50) booze 25 kopeks ‘Earlier a quart of booze cost seven kopeks. Later they introduced a tax on booze and a quart of booze began to cost 25 kopeks.’

Yiddish as a contact language 

Second, it has been fairly widely accepted that subjects in languages like Yiddish originate VP-internally. This means that, aside from case and agreement, they are very much like objects. Given that objects can be postposed and given that case and agreement details can be worked out (which I leave to the syntacticians to describe), Final–Field becomes a possible site for subjects, as well as objects. . Summary of putative syntactic changes due to contact In sum, I have tried to show that language contact can bring about changes that have the superficial appearance of syntactic change but that in fact leave the syntax untouched, the real changes being lexical, semantic, and/or pragmatic. Of course, such a view requires us to make a sharp distinction between the syntax and the other components, a distinction which I take to be justified by how well it serves to explain changes of the sort I have sketched here. It goes without saying that the lion’s share of the research, even on the specific phenomena discussed here, remains to be done.

. Internal syntactic change due to contact-induced semantic change? I shall now turn to a case where a syntactic change has indeed occurred and where I believe it is a result of contact but only very indirectly so, the direct cause being internal to the language. More specifically, I shall propose a story in which there are two relevant changes in the language. The first, chronologically, is the result of contact and is semantic; this semantic change results in a dramatic drop in the frequency of some (native) form. The drop in frequency then results in a syntactic reanalysis of that form on the part of children acquiring the language, producing the syntactic change. The case involves the pluperfect tense, to which I shall now turn. . The Old Yiddish pluperfect: the facts Old Yiddish, like Middle High German and New High German, formed the pluperfect from the preterite of the auxiliary verb and the past participle of the main (lexical) verb, as illustrated in (33). Note that, also as in German, some verbs were conjugated with hobn ‘have’ while others were conjugated with zayn ‘be’.18

 Ellen Prince

(33) zj zjjtn ajm vvav er vvar gjvvezn avn- vvj zj ajn ajn dj gljan they told him where he was been and how they him in the valley hatn gjtragn had carried ‘They told him where he’d been and how they’d carried him into the valley.’ (Bovo-Bukh (79), 1507, from Smith 1968, Part I, p. 125; Part II, p. 348)

At some point subsequent to this stage but before Modern Yiddish, the preterite was lost, replaced by the present perfect (as in Bavarian German, with which Yiddish shares many features). This produced the expected change in the form of the pluperfect: it came to be formed from the present perfect of the appropriate auxiliary verb plus the past participle of the main verb, as illustrated by the example in (34), grammatical in certain Northeast dialects of Modern Yiddish: (34) Er iz geven geven oyfn barg un zey hobn im gehat getrogn in he is been been on-the mountain and they have him had carried in dem tol. the valley ‘He had been on the mountain and they had carried him into the valley.’

Note that this replacement of the preterite, wherever it occurred, by the present perfect preceded Slavic contact; the following example, collected by Beatrice Santorini, is from Western Yiddish, the variety of Yiddish spoken by those Yiddish speakers who never migrated to the east and never had Slavic contact: (35) Tsu Frankfurt hat er gihat eyn veyb ginumen. to Frankfurt has he had a woman taken ‘He had taken a wife in Frankfurt.’ (Court testimony, Worms, 1642)

. The Modern Yiddish pluperfect: the facts In most dialects of Modern Yiddish, we find pluperfects like the following: (36) a.

Ongekumen in a shtetl, farforn, vi geveyntlekh, in akhsanye. Un come into a village, lodged, as usual, in inn and in akhsanye hobn zikh shoyn gehat oyfgeklibn a sakh mentshn, in inn have self already had gathered a lot people, take face be the big rabbi mekabl ponim zayn dem groysn rov. (RP: 222) ‘Came into a village, lodged, as usual, in an inn. And in the inn a lot of people had already gathered to welcome the great rabbi.’

Yiddish as a contact language 

b. Tsvey yidn hobn zikh amol tsekrigt, eyner dem andem gezidlt, two Jews have self a-time fought, one the other cursed, biz eyner iz gevorn mole kas un hot aroysgerufn dem andern until one is become angry and has out-called the other oyf a duel. Zey hobn beyde gehat gedint in soldatn, hobn zey on a duel. They have both had served in soldiers, have they gekent shisn. could shoot ‘Two guys once fought, [each] one cursed the other, until one got angry and challenged the other to a duel. They had both served in the army, so they knew how to shoot.’ (RP: 51) c. Iz tsun im a mol gekumen a yid, a meshulekh fun epes a is to him a time come a Jew, a messenger from something a yeshive. Reb Anshl hot zikh shoyn gehat ongezen religious-school Mr. Anshl has self already had seen meshulokhim genug, hot er im nit gevolt tsulozn tsu zikh. messengers enough, has he him not wanted let to self ‘So once a guy came to him, a messenger from some sort of religious school. Mr. Anshl had already seen enough messengers, so he didn’t want to let him in to see him.’ (RP: 76) (37) a.

Iz gehat geshtanen oyf der same shvel fun shmad. is had stood on the very threshold of conversion [to Christianity] ‘[He] had been very close to converting to Christianity.’ (A. Tseytlin, cited in Mark 1978: 281) b. Zi iz nokh nit gehat aroys[gekumen] fun yener velt. she is still not had out[come] of that world ‘She had still not gotten out of the other world [beyond the grave].’ (Y. Opatoshu, cited in Mark 1978:281) c. S’iz gehat geblibn in im an akshones. it’s had stayed in him a persistence ‘A persistence had remained in him.’ (Y. Bashevis, cited in Mark 1978:281)

The examples in (36), all containing hobn-verbs, are, superficially at least, just what we would expect. However, those in (37) are quite surprising: although each verb is conjugated with zayn ‘be’, the first past participle is gehat ‘had’,

 Ellen Prince

not geven ‘been’. That is, we would expect to find the sentences in (38) instead of those in (37): (38) a.

Iz geven geshtanen . . . is been stood . . . ‘[He] had stood . . .’ b. Zi iz nokh nit geven aroys[gekumen] . . . she is not yet been out-come . . . ‘She had not yet gotten out . . .’ c. S’iz geven geblibn . . . it’s been stayed . . . ‘. . . had remained . . .’

In fact, the sentences in (37), where gehat ‘had’ occurs in the pluperfect with zayn-verbs as well as with hobn-verbs, exemplify most dialects of Yiddish, including the standard dialect.19 I shall now speculate on how this might have come about. . A plausible story I should now like to suggest a possible scenario where a semantic change in Yiddish resulting from Slavic contact affected the frequency of a form in a significant way; the new frequency would then have yielded a new analysis on the part of children aquiring Yiddish, yielding a syntactic change. If this scenario is correct, it is the closest thing I have seen to a real syntactic change resulting from contact; however, it would have come about indirectly, the only borrowing having taken place being at the semantic level. .. Sequence of tenses Before proceeding further, however, we must briefly describe another aspect of the Yiddish verbal system, so-called sequence of tenses. Old Yiddish had the expected Germanic system, whereby, for example, a main clause eventtime that is prior to the utterance-time is expressed by a past tense (preterite or present perfect), and a subordinate clause event-time that is prior to a main clause event-time which itself is prior to the utterance-time is expressed by the pluperfect. For example, consider again (33), repeated for convenience as (39):

Yiddish as a contact language 

(39) zj zjjtn ajm vvav er vvar gjvvezn avn- vvj zj ajn ajn dj gljan they told him where he was been and how they him in the valley hatn gjtragn had carried ‘They told him where he’d been and how they’d carried him into the valley.’ (Bovo-Bukh (79), 1507, from Smith 1968, Part I, p. 125; Part II, p. 348)

Here we see that the time of the telling-event in the main clause is prior to the utterance-time and so is expressed by the preterite; the time of the being-state and of the carrying-event of the subordinate clause are both prior to the time of the telling-event and so are expressed by the pluperfect. After Slavic contact, the semantics of the Yiddish tense system changed: it borrowed the Slavic system whereby the utterance-time remains the reference point for all the events in the sentence, those in subordinate clauses as well as those in main clauses. Thus, the Modern Yiddish equivalent of (39) is (40): (40) Zey hobn im gezogt vu er iz geven un vi zey hobn im getrogn they have him told where he is been and how they have him carried in dem tol. in the valley ‘They told him where he was and how they carried him into the valley.’

Note that, while the English translation of (40) is ambiguous between the intended (pluperfect) meaning and a meaning where all the events are simultaneous, the Yiddish sentence is unambigously pluperfect in meaning. To convey the sense of simultaneity, Yiddish would use the present tense in the embedded clauses, as in (41): (41) Zey hobn im gezogt vu er iz un vi zey trogn im in dem tol. they have him told where he is and how they carry him in the valley ‘They told him where he was [at that moment] and how they were carrying him [at that moment] into the valley.’

Some examples involving a main clause past tense are presented in (42): (42) a.

In a vokh nokh der khasene hob ikh shoyn derzen, vosara.min in a week after the wedding have I already seen what-kind-of klipe ikh hob gekrogn. shrew I have gotten ‘Within a week after the wedding I saw already what kind of shrew I had gotten.’ (RP: 32) [pluperfect sense=past form]

 Ellen Prince

b. Un er hot im gezidlt, vifl es iz in im arayn. and he has him cursed how-much it is in him in ‘And he cursed him, with however much was in him.’ (RP: 192) [past (cotemporaneous) sense=present form] c. An oyrekh iz amol gezesn ban a balebos un hot gevart, biz me a guest is once sat by a host and has waited, until one vet derlangen esn. will serve food ‘A guest once sat at in a host’s house and waited until one would serve food.’ (RP: 48) [future-in-the-past sense=future form]

.. Sequence of tenses and the pluperfect To return now to the pluperfect, we see that the semantic borrowing of the Slavic system of sequence of tenses produces a change in the distribution of the pluperfect, inter alia. In particular, the pluperfect goes from being fairly common in subordinate clauses to being practically nonexistent in subordinate, having been replaced there by the past tense (i.e. present perfect) when a pluperfect sense is to be expressed. Indeed, virtually the only occurrences of the pluperfect that we find today are, like those of (36) and (37) above, in main clauses, nearly all the exceptions being occurrences in preposed adjunct clauses, as in (43): (43) Az etlekhe vayber hobn gehat opgeredt, zogt er tsum when some women have had prt-talked says he to-the yungenman: . . . young-man . . . ‘When some of the women had finished speaking, he says to the young man . . .’ (RP: 155)

I believe that this change in distribution of the pluperfect following the borrowing of the Slavic-type sequence of tense system is relevant to the change in form of the pluperfect for the following reason: the narrowing of the distribution of the pluperfect entails a drop in its frequency. While I do not have a large corpus on which to do a full-fledged frequency analysis, I can report that, in the Old Yiddish Bovo Bukh (1507), the pluperfect occurs at least once on nearly every page, while, in the Modern Yiddish Royte Pomerantsn, nearly 200 pages long, the pluperfect occurs exactly six times, a drop in frequency of at least 97 percent.

Yiddish as a contact language 

.. Syntactic reanalysis of the pluperfect Once the above changes have taken place and the frequency of the pluperfect drops from being fairly high to extremely rare, the stage is set for a reanalysis of the syntax of the pluperfect on the part of the new generation of language learners. In particular, I believe they ceased to parse the pluperfect as had earlier generations, where the auxiliary verb is in the past (i.e. in the present perfect form), followed by the past participle of the main verb, as in (44a), and began to parse it as the past (i.e. the present perfect) of the main verb, with what had been the past participle of the auxiliary reanalyzed as some sort of intervening aspectual/temporal particle, as in (44b): (44) a.

Zey [[hobn gehat] geredt].

They [[have had] spoken] ‘They had spoken’ b. Zey [[hobn] gehat [geredt]] They [[have] prt [spoken]] ‘They had spoken’

One question that arises, of course, is what gehat is reanalyzed as. Unfortunately, it is not clear what the parse was before reanalysis, much less after, given the lack of precedent in the language for a past participle in a finite clause being part of the auxiliary rather than the verb phrase, and I must therefore leave an explicit syntactic description to further research. What is clear is that there was a syntactic reanalysis along the lines of (44) and that this reanalysis was purely language-internal, although the conditions which set the stage for it may well have resulted from the effects of language contact.

. Conclusion In this chapter, I have tried to give a sense of the range of contact effects on Yiddish, and I have tried to show that at least the direct contact effects are not distributed evenly or randomly across all linguistic levels. Rather, they cluster at the levels involving meaning: words and derivational affixes are borrowed into the lexicon, discourse functions associated with contact language syntactic forms are borrowed and come to be associated with native syntactic forms, and meanings associated with contact language tenses are borrowed and come

 Ellen Prince

to be associated with native tenses, replacing the native meanings previously associated with those tenses. Formal changes may ensue, but these are secondary and internal to Yiddish. I believe this is not a chance pattern and not peculiar to Yiddish; rather I take it to follow from two things: first, from the motivation for linguistic borrowing — the semantic/pragmatic intent on the part of speakers to express some concepts in one language that are expressed in some ‘analogous’ way in a contact language — and, second, from some inherent difficulty in altering the formal systems of a language, given that, if this were not the case, the simplest way to achieve such a semantic/pragmatic intent would presumably be simply to borrow the contact language form. Whether this difficulty is due to the modularity of an autonomous syntax or to something else, I do not know, but certainly the independence of form from function seems very clear.

Notes * This chapter owes a great debt to Tonjes Veenstra and Norval Smith for the opportunity they gave me to work on these phenomena and to learn about analogous phenomena in other languages and for their advice, patience, encouragement during my writing of this chapter. A special thanks is due Tony Kroch, Beatrice Santorini, and Meyer-Leyb Wolf for all the time and energy they have spent answering my many questions, and I also thank Bernard Katz, Norman Miller, Jerrold Sadock, Mordkhe Schaechter, and Moshe Taube for their help and support. . More specifically, Old Yiddish refers to the Yiddish spoken by Jews in German-speaking lands before the seventeenth century. Western Yiddish refers to the Yiddish spoken after that in German-speaking lands and in Alsace, the Low Countries, and Switzerland. What is generally called simply ‘Yiddish’ is technically Eastern Yiddish, the Yiddish spoken in Slavic- and Baltic-speaking lands and in Romania. (See Weinreich 1980.) . Parenthetically, it should be noted that Max Weinreich, perhaps for socio-political reasons, perhaps simply because he lacked notions of genetic classification, dubbed Yiddish a ‘fusion’ language, a notion which survives among most traditional Yiddishists to the present day and which sees Yiddish not as a Germanic language but as a ‘fusion’ of three ‘components’, Germanic, Semitic, and Slavic. In a somewhat different vein, Wexler (1991) broke with his fellow ‘fusionists’ in claiming, on phonological and lexical grounds and with no syntactic evidence, that Yiddish is in fact Slavic, a ‘relexified Sorbian’. I do not consider either of these claims serious and will not discuss them further. . The phonological level is perhaps a third; it will not be discussed here. . I should note that I am taking the level of morphology to involve purely formal paradigms, the actual phonological realizations of those paradigms being specified at the lexical level.

Yiddish as a contact language 

. However, see Perlmutter (1986) for compelling arguments that in fact Yiddish masculine plural nouns of Semitic origin were borrowed as unanalyzed wholes and only later were reanalyzed as stem + plural suffix, at which point this plural suffix, now a Yiddish morpheme, was available for combination with other Yiddish noun stems. . However, the Slavic Wh-word complementizer, shto ‘that’, is in fact used for both factive and nonfactive complements. I know of no account of the Yiddish distinction of factive and nonfactive complementizers. . Gundel (1977) argues that Russian eto-sentences are complex. However, her arguments are based on positing totally idiosyncratic syntactic properties of such sentences, e.g. deletion of the complementizer and case-marking of the focused NP with respect to the putatively subordinate verb rather than with respect to the putative matrix (zero) copula. This is clear if an object NP is focused, as in (i), where the focused item is accusative and not, as Gundel’s account would predict, predicate nominative: (i) Eto Er¹iku L¹eon¹id uv¹id¹el. this Erica-acc Leonard-nom saw ‘It’s Erica that Leonard saw.’ . The subject may also occur postposed to the end of the VP (Prince 1988); the point here is that it may not occur preverbally if something else does. . In fact, I do not believe that (9) is correct in that I do not believe that the subject forms a constituent with the VP. However, this is not relevant to the present discussion. . One of my informants, with no training in formal syntax, rejected sentences like (12) on the grounds that ‘‘shto and eto would have to be in the same place.’’ . Here and in (9), the exact nature of the syntactic structure is not relevant; what is relevant is that, given the distributional facts, any syntactic theory would have to posit different syntactic structures for Russian eto-sentences and Yiddish dos-sentences. . In Prince (1992a), I claimed that only subjects may be focused in dos-sentences, in accordance with the judgments of my informants. Since then, Moshe Taube (p. c.) has pointed out that, for some speakers at least, any constituent may be focused, the focus marked only prosodically. Furthermore, even in the written language where prosodic cues are absent, the presence of the unfocusable men ‘one’ produces an unambiguous nonsubject focus, as in (i): (i) Dos hot men mame-loshn geredt. this has one mother-tongue spoken #’It’s they/one that spoke Yiddish.’ ‘It’s Yiddish that they/one spoke.’ Thus, it remains the case that, although nonsubjects may be focused in dos-sentences, only subjects may be focused syntactically. . Dos has been glossed as ‘this/that’ since Yiddish, unlike both German and Slavic, lacks a proximal-distal distinction in demonstrative pronouns. . That what was borrowed was the feature [+expletive] rather than the actual Slavic lexical item eto ‘this’ is not surprising, given the difficulty of borrowing or switching closed class items in language-mixing situations (Weinreich 1952, Joshi 1985, among others).

 Ellen Prince

. Note that, in the case of dos-sentences, it is the identification of the instantiation that is new, i.e. the equation X=[discourse entity]. The particular discourse entity in the equation may of course be new or old to the hearer or new or old in the discourse. If it is Hearer-old and salient, it may be represented by a pronoun, as would be expected. In contrast, since essentences mark that the subject is Hearer-new, it follows that they may not have pronominal subjects. . Note that, if Faber and King (1984) are correct in positing Italy as the immediately preceding homeland of the early Yiddish speakers, the story is not affected as (I believe) Italian is similar to French in all the relevant respects. . Furthermore, Yiddish (like Bavarian German) lost the preterite fairly early, at least for most verbs, replacing it with a compound tense, which would mean that a Middle-Field subject would often be post-Infl but not really postverbal as in Hebrew (or Romance). . I am reproducing Smith 1968’s transliteration conventions (with the exception of the use of a for @), in contrast to the YIVO transcription conventions used in the rest of the chapter and following which (33) would look something like (i), presented here only to make (33) more legible: (i) Zi zeytn im vu er var givezn un’ vi zi in in di glian hatn getragn. . Just as this paper was about to be completed, the following headline appeared in the major Yiddish newspaper in the United States: (i) Mashinist fun gekrakhter ban iz gehat adurkhgeforn royte signaln engineer from crashed train is had through-gone red signals ‘Engineer on crashed train had gone through red lights.’ (Forverts, 16 February 1996) I thank Walter Golman for pointing this out to me.

References Chomsky, N. 1971. ‘‘Deep structure, surface structure, and semantic interpretation’’. In Semantics: An interdisciplinary reader in philosophy, linguistics, and philosophy, D. Steinberg and L. Jakobovits (eds.), 183–216. New York: Cambridge University Press. Delin, J. 1994. ‘‘On processing Clefts’’. Arbeitspapiere des Sonderforschungs-bereichs 340. Bericht Nr. 50. Faber, A. and King, R. 1984. ‘‘Yiddish and the settlement history of Ashkenazic Jews’’. Mankind Quarterly 24: 393–425. Goldvaser, M. 1974. ‘‘Di langfargangene tsayt in yidish’’. [The pluperfect tense in Yiddish] Working Papers in Yiddish and East European Jewish Studies 5. YIVO. Grosz, B. J., Joshi, A. K., and Weinstein, S. 1983. ‘‘Providing a unified account of definite noun phrases in discourse’’. In Proceedings of the 21st Annual Meeting, Association for Computational Linguistics. Cambridge, MA., 44–50. Gundel, J. 1977. ‘‘Where do cleft sentences come from?’’, Language 53: 343–59.

Yiddish as a contact language 

Hall, B. 1979. ‘‘Accounting for Yiddish word order, or what’s a nice NP like you doing in a place like this?’’ In Linear order and generative theory J. Meisel and M. Pam (eds.), 263–87. Amsterdam and Philadelphia: John Benjamins. Horn, L. 1981. ‘‘Exhaustiveness and the semantics of clefts’’. In Papers from the11th Annual Meeting of the North Eastern Linguistic Society, V. Burke and J. Pustejovsky, (eds.), 125–42. Amherst: University of Massachusetts, Department of Linguistics. Joshi, A. K. 1985. ‘‘Processing of sentences with intrasentential code-switching’’. In Natural language parsing: Psychological, computational, and theoretical perspectives, D. Dowty, L. Karttunen and A. Zwicky (eds.), 190–205. Cambridge: Cambridge University Press. Katz, D. 1987. Grammar of the Yiddish language. London: Duckworth. Mark, Y. 1978. Gramatik fun der yidisher klal-shprakh. [Grammar of Standard Yiddish.] New York: Congress for Jewish Culture. Moshavi, A. In prep. Topicalization in Biblical Hebrew. Ph.D. dissertation, Yeshiva University. Perlmutter, D. 1986. ‘‘The split morphology hypothesis: Evidence from Yiddish’’. Paper presented at the Milwaukee Morphology Meeting, April 1986. Pintzuk, S. 1991. Phrase structures in competition: Variation and change in Old English word order. Ph. D. dissertation, University of Pennsylvania. Prince, E. F. 1978. ‘‘A comparison of WH-clefts and IT-clefts in discourse’’. Language 54: 883–906. Prince, E. F. 1988. ‘‘The discourse functions of Yiddish Expletive ES + Subject-Postposing’’. Papers in Pragmatics 2: 176–94. Prince, E. F. 1992a. ‘‘On syntax in discourse, in language contact situations’’. In Text and context: Cross-disciplinary perspectives on language study, C. Kramsch and S. McConnellGinet (eds.), 98–112. Boston: D. C. Heath. Prince, E. F. 1992b. ‘‘The ZPG letter: Subjects, definiteness, and information-status’’. In Discourse description: Diverse analyses of a fund raising text, S. Thompson and W. Mann (eds.), 295–325. Amsterdam and Philadelphia: John Benjamins. Santorini, B. 1989. The generalization of the verb-second constraint in the history of Yiddish. Ph.D. dissertation, University of Pennsylvania. Shumacher, L. 1980. ‘‘Pragmatic factors in the change from VSO to SVO in Arabic’’. Paper resented at the How to Start a Sentence Workshop, Harvard University, March 18, 1980. Smith, J. C. 1968. Elia Levita’s Bovo-Buch: A Yiddish romance of the early sixteenth century. Ph.D. dissertation, Cornell University. Distributed in two volumes by University Microfilms, Inc., Ann Arbor, MI. Talmy, L. 1982. ‘‘Borrowing semantic space: Yiddish verb prefixes between Germanic and Slavic’’. Proceedings of the 8th Annual Meeting, Berkeley Linguistic Society. Weinreich, M. 1980. History of the Yiddish language. Transl. by J. Fishman, and S. Noble. Chicago: University of Chicago Press. Weinreich, U. 1952. Languages in contact. The Hague: Mouton. Wexler, P. 1991. ‘‘Yiddish — the fifteenth Slavic language. A study of partial language shift from Judeo-Sorbian to German’’. International Journal of the Sociology of Language 91: 9–150.

Chapter 12

Social stratification and network relations in the formation of Sranan* Jacques Arends

.

Introduction

While in recent years important progress has been made in the investigation of sociohistorical factors in creole genesis at the macro level, especially with regard to the historical demographics of plantation societies (cf. Arends 1995a; Baker 1982a; Jennings 1993; Singler 1990, 1992, 1993a, 1995), remarkably little attention has been devoted to sociohistorical factors at the micro level, more particularly the social structure of plantation communities. (A notable exception is Singler’s work (1993b, 1993c) on the French Caribbean.) One of the reasons for this is probably the alleged scarcity of relevant sociohistorical information that is available. Although for a long time this kind of information was rather scarce for Surinam as well (cf. Oostindie 1987), the situation has improved significantly since the mid-eighties. Well-known works from the late 1940s and early 1950s, such as Van Lier (1977 [1949]) and Rens (1953), have recently been followed up by a number of highly informative studies, such as Lamur (1987), Oostindie (1989), G. Brana-Shute (1990), Van Stipriaan (1993), Muyrers (1993), and Beeldsnijder (1994). The fact that most of these works are in Dutch increases the risk of their going unnoticed by most creolists. Apart from their inherent quality, this provides an additional reason for summarizing their main findings here. Apart from these studies, which are devoted specifically to (aspects of) the history of Surinam’s plantation society at the micro level, some other works have appeared recently, such as Beeldsnijder (1991), R. Brana-Shute (1989), Lamur (1985, 1990), and McLeod-Ferrier (1993), which touch upon it only marginally, but still contain interesting information on this topic. The fact that both the latter studies and those mentioned before are largely based on primary evidence (i.e. archival documents) adds significantly to their reliabil-

 Jacques Arends

ity. This is important because contemporary historical works, which have determined the image of plantation life for the past few centuries, have a strong tendency to neglect or distort aspects of the world of the slaves, if they provide any original information at all rather than simply plagiarizing their predecessors. It may also be important to note that, while some of these works (e.g. Oostindie 1989; Muyrers 1993) are in-depth case-studies, based on data for only one or two plantations, other studies (especially Van Stipriaan 1993 and Beeldsnijder 1994) are based on data for a representative selection of plantations, thereby providing a more generally valid view.1 Before proceeding to the actual discussion, some preliminary remarks may be in order regarding the relevance of sociohistorical evidence for creole formation. First of all, as will be shown below, this evidence may serve to correct some assumptions about the plantation system which are widely held among creolists, even though they are seldom based on actual historical evidence. One of these assumptions is the stereotypical view of plantations as extremely isolated, strictly bi-stratal micro-societies, which this chapter shows to be incorrect for the case of Surinam. Referring to nineteenth- rather than eighteenth-century Surinam (which is the focus of the present chapter), Lamur (1985: 52–53) writes that ‘‘the pluriformity of the slaves’ social life differed fundamentally from the image of uniformity that emerges from the current literature.’’ Second, while it is not possible at this moment to make any direct links between social structure on the one hand and linguistic structure on the other, it is still useful to get as precise a picture as possible of the social context in which creole formation took place. For instance, the existence of social contacts between slaves from different plantations or between slaves and Maroons is directly relevant for the question to what extent language contact played a role in creolization. Similarly, information on the social stratification of the slave community may allow us to identify more precisely groups of speakers who played an important role in this process. The evidence presented in this chapter suggests that the black overseers may have formed such a group. As is announced in its title, this chapter will discuss two aspects of the social structure of the plantation system: (1) the internal social stratification of the plantation community and (2) the external network relations (i.e. contacts outside the plantation) maintained by the slaves. Since the focus is on social life at the plantation — the primary locus of creole formation — and because the social history of Paramaribo (Surinam’s capital) has hardly been investi-

Social stratification and network relations 

gated as yet (R. Brana-Shute 1990: 121), the urban setting will only be touched upon marginally. The discussion will be restricted to what Bickerton (1989: 17) has called the ‘‘growth phase’’ in the development of plantation societies, i.e. in the case of Surinam roughly the 1700–1775 period. The reason not to include the equally (or even more) important pre-1700 years in this study is the fact that, apart from Rens (1953), hardly any evidence for this period is available. In other words, the social history of this important stage in the development of the Surinam creoles still largely remains to be written. Until this has been done, it will not be possible to get a fully comprehensive picture of the formation of these languages.

. Social stratification The social stratification of plantations’ free and enslaved populations is represented schematically in Tables 1 and 2, which are based on Lamur (1987), Oostindie (1989), Van Stipriaan (1993), and Beeldsnijder (1994). The major variable on which these tables are based is the division of labor, a socioeconomic phenomenon which — contra Lamur’s (1985) suggestion that it did not emerge until the nineteenth-century — existed at least as early as the 1720s (Oostindie 1989: 449–451; Beeldsnijder 1994, Ch. 6). As can be seen from these tables this variable correlates to a large extent with annual wage (see Table 1) and estimated market value (Table 2), respectively. It should perhaps be stressed that the representation in both tables, especially Table 2, is a schematic one, showing tendencies rather than absolute truths.

Table 1. Social stratification of the free population on eighteenth-century plantations in Surinam Professional category

Social/ethnic group

Average annual wage (guilders)

Average number per plantation (50–100 slaves)

Owner Administrator Manager Skilled worker

White White White White, free Black, free Mulatto White White

n.a. n.a. 700–1500 200–800

1 1 1 0/1

250 100

0/1 2

Bookkeeper White overseer

 Jacques Arends

Several aspects of this table require some explanation: –

–

–

–

– –

–

Temporary personnel, such as barbers/surgeons, who visited plantations only occasionally, and artisans, who were hired for limited periods of time, are not included in Table 1. The identification of social/ethnic groups from which people were selected for particular tasks does not exclude the possibility that in some cases people from other groups performed these tasks. Thus, in the literature mention is made sometimes of mulatto or even black ‘‘white’’ overseers, managers, and even plantation owners, but this seems to have been rare, especially at the beginning of the eighteenth century. According to Lamur (1987: 43), black and mulatto skilled workers became a more frequent phenomenon in the course of the eighteenth century. Annual wages are not given for plantation owners and administrators, since they derived their income directly from the profits of their plantation(s). Within the category of skilled workers, wages varied from 200 guilders for coopers to 800 for carpenters (Beeldsnijder 1994: 92). According to the same author (p. 89), usually one (non-captive) carpenter resided permanently on sugar plantations in the 1730–1750 period. The bookkeeper’s task was often performed on the side by one of the other white personnel (Beeldsnijder 1994: 93). The low position of the white overseer (blankofficier) in the plantation’s hierarchy also appears from regulations for plantation personnel (dating from 1686 and 1725), discussed in Beeldsnijder (1994: 90). In the course of the eighteenth century, especially from the 1750s onwards, many plantations got into the hands of overseas owners, such as banks and other financiers following bankruptcies. According to Beeldsnijder (1994: 63), more than 70 per cent of the plantations was owned by absentee planters from 1770 onwards. Similarly, Oostindie (1993: 34 n. 90) estimates the percentage of absentee owners at 70–80 per cent in the 1780–1830 period. Although Beeldsnijder claims that large-scale absentee ownership may have begun much earlier, according to Van Stipriaan (1993: 293) around 1750 only 20 to 25 per cent of the plantation owners were living outside Surinam. This phenomenon is known as ‘‘absenteeism’’ (Van Lier 1977), i.e. the situation where the actual, Europe-based owner of a plantation was represented in Surinam by an administrateur (agent), who resided in Paramaribo rather than on the plantation itself.

Social stratification and network relations 

–

The actual day-to-day management of the plantation was in the hands of a directeur (plantation manager). During the growth phase, with the important exception of the Jewishowned plantations, white women and children were rarely present on plantations. Beeldsnijder (1994: 41) has calculated, on the basis of head tax payments, that at least half of the planters coming to Surinam were unmarried. Jewish planters, however, often lived on their plantations with their families (p. 247). White plantation personnel, especially white overseers, were mostly recruited from among unmarried former soldiers and sailors.

Table 2. Social stratification of the enslaved population on eighteenth-century plantations in Surinam Professional category

Social/ethnic group

Estimated market value (guilders)

Approximate percentage of enslaved population

Black overseer Skilled slave

Creole/African black/mulatto Creole, African black/mulatto Creole, African African, black Creole all groups

500–800 300–1000

2% 10%

300–500

4%

Domestic slave Unskilled slave, field slave Unproductive slave

300–500

25–60%

<100

20–40%

Again, some additional information is in order: –

–

–

The terms ‘‘Creoles’’ and ‘‘Africans’’ refer to locally-born and Africanborn slaves, respectively. While Creoles were generally preferred for skilled labor and for domestic tasks (Beeldsnijder 1994: 124–125, 139, 149), this was apparently not the case to the same extent for the position of black overseer (basya). Even though the number of locally-born male slaves exceeded the number of basya positions by far, less than half of these were occupied by Creoles (p. 124, 154).2 Within the category of skilled slaves there was a sub-hierarchy, with carpenters (in a wide sense, i.e. including sawyers and coopers) at the top. And even within this subcategory, a further hierarchy was obtained with carpenters (in a narrow sense) at the top, followed by coopers, sawyers, and apprentices (Oostindie 1989: 105). The number of mulattos mentioned in the plantation inventories studied by Beeldsnijder (1994: 125) is surprisingly low, namely 1.5 per cent of the

 Jacques Arends

–

–

–

slave population. This may be due to the fact that not only those who were manumitted by their white fathers but also those who were sent to Paramaribo to learn a trade (p. 139) were not included in the inventories. The category of unproductive slaves includes persons who were unable to work, either due to age or some physical or mental defect. Slave children started to perform light duties around the age of ten to twelve; around the age of eighteen they were considered to be fully grown workers (p. 180). As noted by Beeldsnijder (p. 116), the average life expectancy of a mideighteenth-century Caribbean slave who had survived his first year in the colony was 30 years. Some of these positions, e.g. those of basya and skilled slave, could be occupied by one and the same person. Also, slaves could be promoted or demoted from one position to another. According to Beeldsnijder (1994: 16), switching between different tasks was the rule rather than the exception on Surinamese plantations. This means that the hierarchy was less fixed than may be suggested in Table 2. There is some evidence that the categories of black overseers, skilled slaves and domestic slaves formed a kind of elite within the enslaved population. Beeldsnijder (1994: 157) has found some evidence in legal records to support this. This suggests that the stratification within the slave population was primarily a dichotomous one: elite slaves versus all the other slaves.

In what follows, one category within the enslaved population, the black overseer, will be discussed in some more detail. An important fact about plantation residence patterns is that the white personnel was by far the least continuous segment of the population. Surinam’s eighteenth-century white population is characterized by Beeldsnijder (1994: 45) as ‘‘a white pioneer community consisting of many single men, who had no intention to stay permanently’’ (translation mine, J. A.). Towards the end of the growth phase (around 1775), the average duration of a manager’s residence on a plantation was 5.5 years, while that of a white overseer was as short as 1.1 years (Van Stipriaan 1993: 284–285). In contrast, slaves usually spent large parts of their lives on one and the same plantation. The absence of continuity among white personnel, especially among white overseers, undoubtedly contributed to a situation where the position of the black overseer in the hierarchy, in spite of his color and his status as a slave, in some respects equalled that of his white counterpart, the blankofficier. Other factors also contributed to this state of affairs. Many of a basya’s daily activities, such as allocating tasks, supervising

Social stratification and network relations 

work, and executing punishments, made him into an intermediary between manager and slaves. Combined with his knowledge of Afro-Surinamese culture, this gave the basya a degree of power among the slaves which sometimes may have surpassed that of the white overseer. At the same time, however, the basya’s role as a buffer between masters and slaves may have had unfavorable consequences for his hierarchical position as well (Van Deursen 1975: 217; Oostindie 1989: 165). In addition to this, there are some indications that the black overseer not only performed the role of priest in Afro-Surinamese religion but that he played a central role in subversive activities, such as rebellions and escapes, as well (Lamur 1990: 112–113; Van Stipriaan 1993: 282; Beeldsnijder 1994: 155). Although Beeldsnijder (1994: 157) is not certain that the basya occupied a socially privileged place within the slave community, he adds that in at least one legal document basya’s are referred to as ‘‘chiefs’’ among the slaves. The same author (p. 232) mentions a case of a rebellion in 1750 where five out of sixteen conspirators were black overseers. The convergence of the roles of basya and priest on Surinamese plantations has been noted by Lamur (1985: 26), who quotes a Moravian missionary’s report referring to ‘‘. . .den ersten Bastian welcher zugleich der erste prister ist . . .’’ [the highest-ranking basya who is also the highest-ranking priest (translation mine, J. A.)]. Documents referring to a religious conflict between missionaries and slaves that took place on the plantation Vossenburg around 1850 show that it was especially basya’s who were involved in this. Lamur (1990: 112–113) concludes from this that the basya’s ‘‘were looked upon by the fellow slaves as functionaries responsible for maintaining the religious system of the slaves’’ Although the evidence is still rather sketchy, it suggests that black overseers were central figures among the black populations of the plantations. The question is, of course, whether the basya’s seemingly pivotal role on the plantation had any linguistic consequences, and if yes, what these consequences were. Unfortunately, at this stage it is only possible to speculate. One possible avenue through which this question could be pursued is the social networks approach developed by Leslie and James Milroy (L. Milroy 1980). Due to his intermediary position, inbetween the white and black segments of the population, the basya probably had weaker (though not fewer) network ties within the black community than other slaves had. At the same time, his contacts with the few Whites on the plantation were stronger and more frequent than those of most other slaves (except perhaps the domestic slaves who worked in the planter’s house). Also, his verbal interactions with white

 Jacques Arends

personnel must have involved a more extended vocabulary, including technical terminology relating to the production of sugar and coffee. The idea that this vocabulary was Sranan (rather than Dutch) is supported by references to the importance of managers and artisans knowing Sranan, which can be found in eighteenth-century plantation manuals such as Van Dyk (c.1765; see also Beeldsnijder 1994: 86). From the perspective of the social network model of language variation and change (J. Milroy 1992), this combination of features adds to the basya’s potential role as a linguistic innovator. At the same time his prestige as a political and religious leader contributes to his position as a linguistic role model, especially for the African-born slaves, who formed a majority among the black population until the end of the eighteenth century (Price 1976: 12). The question, of course, is whether it is justified to apply this model, which was designed for ‘‘normal’’ language change, to the process of creole formation. A major difference between the two is that in the latter case we are dealing with the creation of a linguistic system rather than changes within an already existing language. Nevertheless, it would be interesting to know whether black overseers, many of whom were African-born, were selected from specific ethnolinguistic groups, since in that case the languages spoken by those groups may have played a disproportionately large role in the formation of Sranan. Unfortunately, no evidence on this matter is available with regard to Surinam. All we have are occasional remarks in contemporary works regarding personality traits of different ethnic groups, the reliability of which is difficult to assess. What is needed is the kind of primary evidence on ethnic affiliation of basyas that is sometimes found in plantation inventories or similar archival documents. Some evidence of this kind is available for another colony, namely Cayenne (French Guiana). Singler (1993b: 206–207), based on detailed archival documents from 1690 concerning a sugar plantation in this colony, states that among the African ethnolinguistic groups present on the plantation (including Atlantic, Mande, Akan, Gbe, Yoruba, and Bantu) the Gbe and Akan groups received the highest rankings in terms of productivity. Typically, it was a ‘‘Juda’’ (an African transported from Whydah (Ouidah) on the Slave Coast, i.e. probably a Gbe-speaker) who served as a black overseer at this plantation. Similarly, Alleyne (1971: 176), referring to ‘‘certain psychological and cultural traits possessed by the Coromanti [Akan, J. A.],’’ states that ‘‘they certainly seem to have been everywhere leaders of slave revolts,’’ but unfortunately he does not mention any sources to substantiate this claim. Alleyne’s claim is in line with the general impression gathered from the literature regard-

Social stratification and network relations 

ing Surinam to the extent that the Akan were favored among planters, while Loango slaves were considered second choice (Beeldsnijder 1994: 108, 122–123). This is supported by the fact that the average price for a Loango slave was some 20 per cent below that for other slaves (p. 110). Although much more historical research is needed to substantiate the idea that speakers of particular Kwa languages were preferred as overseers, if such a preference were found, it would be in line with the general impression that the influence of Kwa languages on Sranan and other Surinam creoles is stronger than that of Bantu languages.

. External network relations Before we turn to the issue of the external contacts maintained by plantation slaves, it should be emphasized that during the growth phase on average less than half of a plantation’s black workforce actually worked in the fields (Beeldsnijder 1994: 160, 276; see also Table 2). This means that the stereotypical view of the plantation as a place where the bulk of the slaves belonged to an anonymous labor force working the fields with little or no contact at all with anybody except fellow-slaves is incorrect. Due to the technical complexity of sugar-cane processing, which required a considerable portion of the slave population to work at the sugar-mill rather than in the field, this seems to be especially true for sugar plantations (Klein 1986: 61–62). The significance of this fact becomes clear when it is realized that a significant part of the formative period of Sranan took place in the context of sugar production; coffee, the other major crop, was not introduced until the second quarter of the eighteenth century (Van Stipriaan 1993). The primary source for information on external network relations is Muyrers (1993), which is entirely devoted to this subject. Much of Muyrers’ data is based on primary evidence regarding only one plantation (Catharina Sophia), which was owned by the colonial government and which therefore may have been a kind of ‘‘model’’ plantation. In addition to this, her data refer to the mid-nineteenth rather than the eighteenth century, which is the focus of this chapter. Nevertheless, Muyrers claims, based on data from secondary eighteenth and nineteenth century sources, that her data may be considered representative for other plantations from the same and earlier periods.3 This is confirmed by other studies, such as Keller (1982), R. Brana-Shute (1989), Oostindie (1989, 1993), Van Stipriaan (1993), and, especially, Beeldsnijder (1994), which show that virtually all types of external contacts mentioned by

 Jacques Arends

Muyrers for the mid-nineteenth century were common a hundred years earlier. (Studies such as Fleischmann (1984), on the French Caribbean, and Speedy (1995), on Louisiana, show that the existence of relatively frequent external contacts among slaves was by no means limited to Surinam.) The types of contact listed by Muyrers (p. 99–108) are subdivided here into four categories according to the context in which these contacts took place: work, trade, leisure, and resistance. The list is supplemented with two additional types of contact which are not mentioned by Muyrers. (If no particular source is mentioned in this list, the reference is Muyrers 1993.) . Work –

– –

– –

– – –

Transport of persons, products, necessities, and correspondence by boat to and from Paramaribo, enabling communication with slaves on other boats, on other plantations, and in Paramaribo. Blacks used for this work were mostly skilled slaves, either from Paramaribo or from the plantation. Communication with other Blacks often took place through — sometimes secret — songs. Hiring of slaves from other plantations or from Paramaribo. Hiring of skilled workers or contractors from Paramaribo to supervise certain jobs done by local skilled slaves on the plantation (Oostindie 1989: 73). Apprenticeship with a craftsman in Paramaribo (Beeldsnijder 1994). Taking part in the so-called commando service, the compulsory service to the colonial government comprising such tasks as digging out creeks, building fortifications,4 and taking part in military patrols against Maroons. Around the middle of the eighteenth century, military tasks began to get more professionalized and slaves were used less to perform them (Beeldsnijder 1994: 78, 83). In the 1770s, two regiments of free black soldiers were founded, whose special task it was to bring back run-away slaves and to fight newly formed Maroon tribes such as the Boni. Temporary or permanent transfer of whole workforces from one plantation to another (see also Beeldsnijder 1994: 219). Banishment of slaves to other plantations as a means of punishment. Working at a plantation’s provision grounds, which were often located behind the plantation itself (i.e. several miles away from the planter’s house), enabling contact both with slaves from other plantations and with Maroons and runaway slaves.

Social stratification and network relations 

–- Hunting and fishing by slaves, to supply both their masters and themselves with food. – Temporary residence of ailing and pregnant slaves in Paramaribo. This did not happen frequently until the last quarter of the eighteenth century. . Trade – –

Trading at the marketplace in Paramaribo. Trading on the plantations with Maroons, free Indians, and Whites. According to Beeldsnijder (1994: 87, 224), both free Indians and Europeans (soldiers, sailors, and smugglers) came to the plantations in order to trade with slaves. Although trading with Maroons was not officially permitted until the 1760s peace treaties, this seems to have happened quite frequently. A fact which has been neglected up to now is that, due to the intensive trade between North America and Surinam, North Americans were very well represented among sailors visiting the colony. The average number of North American ships visiting Surinam annually rose from 25 in 1710 to almost 90 in 1790 (Johannes Postma, unpublished data). The presence of North Americans in Surinam is especially interesting since many of them presumably spoke English, the main lexifier language of the Surinam creoles, which largely disappeared from the colony with the departure of the English in the 1670s. The potential linguistic relevance of the North Americans’ presence in Surinam is enhanced by the fact that they frequently visited sugar plantations in order to buy molasses on the spot (Pares 1956: 20, 106–107).

. Leisure –

–

Visits of manumitted slaves from Paramaribo. Some plantations had special buildings where former slaves, who had been manumitted and were living in Paramaribo, could stay when they were paying a visit (Beeldsnijder 1994: 127, 304 n. 27). Religious gatherings, such as funerals, winti prees (religious sessions), and visits to shared sanctuaries, outside the plantation. While no figures are available for winti prees held on plantations, contemporary Moravian diaries show that eighteenth century Saramaccan prees could be attended by up to 500 visitors (Price and Price 1980: 169). Hoogbergen (1990: 70) notes that slaves owned dugout canoes, which increased their mobility

 Jacques Arends

–

–

–

significantly. Lamur (1990: 107), writing about the plantation Vossenburg in the late 1840s, says that ‘‘the cult of the slaves at Vossenburg was apparently so important that the place of worship in the bush where the icon of the Supreme God was located was also attended by slaves from the nearby plantations . . .’’ Another altar ‘‘was located at a distance of one hour’s walking from the center of the plantation.’’ Secular festivities, such as the annual distribution of food and goods around New Year, at which slaves from neighboring plantations would be invited, baljar (i.e. dance) parties, and musical performances by so-called du societies. A painting by Dirk Valkenburg from 1707, reproduced in Price and Price (1980: 166), presents iconographic evidence that baljar parties were known at least as early as the begining of the eighteenth century (see also Beeldsnijder 1994: 131). From the late eighteenth century onwards both free Blacks/Mulattos and slaves participated in du societies, which were devoted to composing and performing song, dance and music (Voorhoeve and Lichtveld 1975: 16). Festivities where slaves from different plantations were present are mentioned from 1765 onwards (Muyrers 1993: 105). Inter-plantation communication through talking drums, tutus (wooden horns) or bentas (finger pianos), used to announce a death or a rebellion or to warn a nearby Maroon community for an approaching military patrol. Sexual relationships between slaves from different plantations. Such relationships are reported as early as the beginning of the eighteenth century (Herlein 1718: 97; see also Muyrers 1993: 103–104).

. Resistance –

–

– –

Maroons’ and runaways’ temporary residence on plantations, either secretly or openly. Maroons did not always hide (e.g. in slaves’ barracks) when visiting plantations; according to Lamur (1985: 11), they were sometimes present quite openly. Maroons’ visits to their former plantations. This seems to have happened especially with female Maroons who had been abducted from their plantation by force. Consultations between slaves and Maroons preceding rebellions, escapes, and plantation raids.5 Slaves’ temporary residence in Maroon communities (petit marronnage). As

Social stratification and network relations 

observed by Hoogbergen (1990: 84, 93), ‘‘even after the peace settlement (in the 1760s, J. A.) slaves continued to flee from the Surinamese plantations, while the Maroons themselves avoided the extradition of new figitives as much as possible . . . It often happened that the fugitives were only delivered after several years of pressure from the colonial authorities.’’ Not mentioned by Muyrers are two additional types of contact: contacts through imprisonment and contacts through foreign residence. As to the first, since the number of crimes committed by slaves was relatively low and since the great majority of those slaves who were tried in court received the death penalty (Beeldsnijder 1994: 249), the number of former prisoners among plantation slaves cannot have been very high. As to foreign residence, both slaves, accompanying white families, and free Mulattoes sometimes stayed in Europe temporarily, although the latter only in very small numbers. In the 50 odd years between 1729 and 1781 some 750 Blacks from Surinam (only a few percent of the total black population) visited the Netherlands, where they usually stayed for only a short time (Oostindie 1990: 232). Sometimes, however, their residence lasted longer, as in the case of Elisabeth Samson, a wealthy, free black woman, who lived in Holland for some three and a half years, after which she returned to Surinam (McLeod-Ferrier 1993: 46–47). Lenders (1994: 131) mentions the case of Scipio, a baptized Saramaccan Maroon, who by the end of the eighteenth century accompanied a Moravian missionary on his trip to Holland (paying his own fare). Since European residence generally did not last very long and involved only small numbers of people, its linguistic impact cannot have been very strong. Generally speaking, however, we may conclude that slaves in eighteenth and nineteenth century Surinam had quite a number of opportunities for establishing and maintaining contacts with people outside their plantation, not only with other slaves, but with Indians, Maroons, free Mulattoes, and Whites as well.

. Conclusion With respect to the social factors discussed in this chapter, two forces can be discerned whose linguistic consequences were more or less opposed: an internal social stratification favoring linguistic differentiation, and an external social network system favoring homogenization. Early social stratification provides indirect support for the hypothesis, first formulated by Baker

 Jacques Arends

(1982b), that creole continua may have arisen quite early on in the formation of creole languages. In a socially stratified population, different groups of Blacks had differential access to the language spoken by the Whites as well as differential motivation to learn it. As a result, a spectrum of varieties may have developed from quite early on. At the same time, early external contacts contributed to the homogenization of creole varieties spoken on different plantations. As a result, one, more or less homogeneous, creole developed rather than a number of different creole ‘‘dialects’’, one for each plantation (cf. Speedy 1995: 103). Apart from these rather general inferences, it does not seem possible at this moment to make any direct links between this type of sociohistorical evidence and purely linguistic developments. Nevertheless, I think that any historically realistic theory of creole formation should take into account the kind of extralinguistic evidence presented here.

Notes * The paper on which this chapter is based was presented at the Meeting of the Society for Pidgin and Creole Linguistics in New Orleans, January 1995, and at the Second Creolistics Workshop at the University of Westminster in London, April 1995. I am grateful to the audiences present at these occasions for their questions and comments. A special word of thanks is due to Philip Baker, Robert Chaudenson, Salikoko Mufwene, Pieter Muysken, Gert Oostindie and an anonymous reviewer for their remarks and criticism. . Van Stipriaan (1993) is based on archival documents concerning some 200 plantations in the 1750–1863 period, i.e. almost 30 per cent of the total number of Surinam’s plantations throughout its existence as a plantation colony. Beeldsnijder (1994) is based on archival documents for 18 plantations, providing data about 2,062 slaves (12 per cent of the total slave population) in the 1730–1750 period. In both cases, the archival documentation consists primarily of plantation inventories (i.e. valuation reports); these data are supplemented by letters, legal records, and other archival documents. . This observation is based on a sample of 89 basyas. . Apart from archival documents about Catharina Sophia, Muyrers’ database includes some fifteen contemporary secondary sources. According to Muyrers, during the eighteenth century the number and intensity of contacts may have been less than in the nineteenth century, because sanctions were more severe and the solidarity among slaves from the same plantation was weaker, which enhanced the chance of being betrayed when leaving the plantation without the owner’s consent. To some extent, however, this may have been counterbalanced by the fact that in the eighteenth century ethnolinguistic identity and the relationship between sipis (slaves who had been transported on the same ship) were more important factors of solidarity than living on the same plantation. . The total number of slaves that planters were officially required to deliver in order to

Social stratification and network relations 

work at the fort in Paramaribo was 300, but this number was never reached by far (Beeldsnijder 1994: 81). . An example of this can be found in Van den Bouwhuijsen et al.’s (1988) analysis of the 1750s Tempati rebellion. As stated by Hoogbergen (1990: 76), ‘‘rebellion was often the result of collaboration between the slaves who had run away before and the ones who had stayed behind. After the revolt on the plantation of Palmeneribo on the Suriname River in 1758, it appeared that the slaves had been in contact with small groups of Maroons who had been staying near the plantation for seven months without the white people ever aware of it.’’ See also Beeldsnijder (1994: 214, 221).

References Alleyne, Mervyn. 1971. ‘‘Acculturation and the cultural matrix of creolization’’. In Pidginization and creolization of languages, Dell Hymes (ed.), 169–186. Cambridge: Cambridge University Press. Arends, Jacques. 1995a. Demographic factors in the formation of Sranan’’. In Arends (ed.), 233–285. Arends, Jacques (ed.). 1995b. The Early Stages of Creolization. Amsterdam and Philadelphia: Benjamins. Baker, Philip. 1982a. ‘‘On the origins of the first Mauritians and of the creole language of their descendants: A refutation of Chaudenson’s ‘Bourbonnais’ theory’’. In Isle de France Creole, Philip Baker and Chris Corne (eds.), 131–259. Ann Arbor, MI: Karoma. Baker, Philip. 1982b. The contribution of non-Francophone immigrants to the lexicon of Mauritian creole. Doctoral dissertation, University of London. Beeldsnijder, Ruud. 1991. ‘‘Op de onderste trede: Over vrije negers en arme blanken in Suriname 1730–1750’’. OSO 10: 7–30. Beeldsnijder, Ruud. 1994. Om werk van jullie te hebben: Plantageslaven in Suriname, 1730–1750. (Bronnen voor de studie van Afro-Suriname, 16.) Utrecht: Department of Cultural Anthropology, University of Utrecht. Bickerton, Derek. 1989. ‘‘The lexical-learning hypothesis and the pidgin-creole cycle’’. Wheels within Wheels, Martin Pütz and René Dirven (eds.), 11–31. Frankfurt a.M.: Peter Lang. Brana-Shute, Gary (ed.) 1990. Resistance and rebellion in Suriname: Old and new. (Studies in Third World Societies, 43.) Williamsburg VA: Department of Anthropology, College of William and Mary. Brana-Shute, Rosemary. 1989. ‘‘Approaching freedom: The manumission of slaves in Suriname, 1760–1828’’. Slavery and Abolition 10: 40–63 Brana-Shute, Rosemary. 1990. ‘‘Legal resistance to slavery in eighteenth century Suriname’’. In Gary Brana-Shute (ed.), 119–136. Fleischmann, Ulrich. 1984. ‘‘Communication et langues de communication pendant l’esclavage aux Antilles: Contribution aux problèmes de la genèse des langues créoles’’. Etudes Créoles 6: 29–46. Herlein, J. 1718. Beschrijvinge van de Volksplantinge Zuriname. Leeuwarden: Injema.

 Jacques Arends

Hoogbergen, Wim. 1990. ‘‘The history of the Suriname Maroons’’. In Gary Brana-Shute (ed.), 65–102. Jennings, William. 1993. The demographics of creole genesis: Implications of sociohistoric and population studies for the origins of Cayenne Creole. MS. Keller, Saskia. 1982. ‘‘De slavenplantage: Een totale institutie?’’ Antropologische Verkenningen 1: 1–45. Klein, Herbert. 1986. African Slavery in Latin America and the Caribbean. New York and Oxford: Oxford University Press. Lamur, Humphrey. 1985. De Kerstening van de Slaven van de Surinaamse Plantage Vossenburg, 1847–1878. Amsterdam: Department of Anthropology, University of Amsterdam. Lamur, Humphrey. 1987. The Production of Sugar and the Reproduction of Slaves at Vossenburg, Suriname 1705–1863. (Caribbean Culture Studies, 1.) Amsterdam: Centre for Caribbean Studies. Lamur, Humphrey. 1990. ‘‘Slave religion on the plantation Vossenburg (Suriname) and missionaries’ reaction’’. In Gary Brana-Shute (ed.), 103–117. Lenders, Maria. 1996. Strijders voor het Lam: Leven en werk van Herrnhutter broeders en zusters in Suriname, 1735–1900. Leiden: KITLV Uitgeverij. McLeod-Ferrier, Cynthia. 1993. Elisabeth Samson: Een vrije, zwarte vrouw in het achttiendeeeuwse Suriname. (Bronnen voor de studie van Afro-Suriname, 15.) Utrecht: Department of Cultural Anthropology, University of Utrecht. Milroy, Leslie. 1980. Language and Social Networks. Oxford: Blackwell. Milroy, James. 1992. Linguistic Variation and Change. Oxford: Blackwell. Muyrers, Sabine. 1993. Het Netwerk van de Slaaf: Een onderzoek naar contacten van Surinaamse plantageslaven in de achttiende en negentiende eeuw. M.A. thesis, Erasmus Universiteit Rotterdam. Oostindie, Gert. 1987. ‘‘Historiography on the Dutch Caribbean ( –1985): Catching up?’’. Journal of Caribbean History 21: 1–18. Oostindie, Gert. 1989. Roosenburg en Mon Bijou: Twee Surinaamse plantages, 1720–1870. Dordrecht: Foris. Oostindie, Gert. 1990. ‘‘Preludes to the exodus: Surinamers in the Netherlands, 1667– 1960s’’. In Gary Brana-Shute (ed.), 231–258. Oostindie, Gert. 1993. ‘‘Voltaire, Stedman and Suriname slavery’’. Slavery and Abolition 14: 1–34. Pares, Richard. 1956. Yankees and Creoles: The trade between North America and the West Indies before the American revolution. Cambridge, MA: Harvard University Press. Price, Richard. 1976. The Guiana Maroons: A historical and bibliographical introduction. Baltimore: Johns Hopkins University Press. Price, Richard and Sally Price. 1980. Afro-American Arts of the Suriname Rain Forest. Berkeley: University of California Press. Rens, Lucien. 1953. The Historical and Social Background of Surinam’s Negro-English. Amsterdam: North-Holland. Singler, John. 1990. ‘‘On the use of sociohistorical criteria in the comparison of creoles’’. In Pieter Seuren and Salikoko Mufwene (eds.), Issues in Creole Linguistics (Linguistics 28), 645–659.

Social stratification and network relations 

Singler, John. 1992. ‘‘Nativization and pidgin/creole genesis: A reply to Bickerton’’. Journal of Pidgin and Creole Languages 7: 319–333. Singler, John. 1993a. ‘‘African influence upon Afro-American language varieties: A consideration of sociohistorical factors’’. In Africanisms in Afro-American Language Varieties, Salikoko Mufwene (ed.), 235–253. Athens GA: University of Georgia Press. Singler, John. 1993b. The cultural matrix of creolization: Evidence from Goupy des Marets’s description of a 1690 sugar plantation in French Guyana. In John Singler (1993d), 187–224. Singler, John. 1993c. ‘‘The setting for creole genesis in France’s Caribbean colonies: Evidence from seventeenth-century Marie-Galante’’. In John Singler (1993d), 225–236. Singler, John. 1993d. The African Presence in Caribbean French Colonies in the Seventeenth Century: Documentary evidence. (Travaux de recherche sur le créole haïtien, 16–17.) Montreal: Groupe de recherche sur le créole haïtien, Université de Québec à Montréal. Singler, John. 1995. ‘‘The demographics of creole genesis in the Caribbean: A comparison of Martinique and Haiti’’. In Arends (ed.), 203–232. Speedy, Karin. 1995. ‘‘Mississippi and Tèche Creole: Two separate starting points for Creole in Louisiana’’. In From contact to creole and beyond, Philip Baker (ed.), 97–114. London: University of Westminster Press. Van den Bouwhuijsen, Harry, Ron de Bruin and Georg Horeweg. 1988. Opstand in Tempati, 1757–1760. (Bronnen voor de studie van Afro-Suriname, 12.) Utrecht: Department of Cultural Anthropology. Van Deursen, A. 1975. ‘‘De Surinaamse negerslaaf in de negentiende eeuw’’. Tijdschrift voor Geschiedenis 88: 210–223. Van Dyk, Pieter. [ca 1765]. Nieuwe en Nooit Bevoorens Geziene Onderwijzinge in het Bastert Engels, of Neeger Engels. Amsterdam: Jacobus van Egmont. Van Lier, Rudolf. 1977 [1949]. Samenleving in een Grensgebied: Een sociaal-historische studie van Suriname. Amsterdam: S. Emmering. Van Stipriaan, Alex. 1993. Surinaams Contrast: Roofbouw en overleven in een Caraïbische plantagekolonie 1750–1863. Leiden: KITLV Uitgeverij. Voorhoeve, Jan and Ursy Lichtveld (eds.). 1975. Creole Drum: An anthology of Creole literature in Surinam. New Haven: Yale University Press.

Index of languages

Adamawa 116 Akan 162, 298, 299 Akkadian 94 Albanian 83, 226, 232 Tosk dialect 232 Aleut: Mednyj Aleut 92, 93, 107, 108, 257, 261 Western 92, 93 Amarna-Akkadian 94 Angloromani 81, 90, 91, 94, 108 Arabic 93, 151, 175, 213, 277 Arawak 99, 108 Arawakan 86, 94 Armenian 83 Assiniboine 93 Atlantic 46, 47, 55, 56, 210, 241, 298 Austronesian 83–85, 87, 101, 105 Oceanic 83, 84 Polynesian 84 Avadhi 176–178, 180, 201, 203, 204, 206 Ayiwo 85 Baegu 180 Baelelea 180 Ballissha 119 Baltic 229, 238, 286 Balto-Slavic 263 Bambara 235 Bantu 56–58, 82, 86, 94, 95, 116, 117, 235, 241, 257, 260, 298, 299 Bengali 86 Berbice 98–100, 108, 109, 204, 237–240, 243, 245 Bhojpuri 176, 178–181, 201, 203, 206 Guyana Bhojpuri 179 Bini 86 Bisayan 100, 102–104, 106 Bislama 187 Bowe 117, 120 Braj 201, 206 Caddo 92 Callahuaya 94, 121 Caló 91 Cameroon Pidgin English 146, 153 Canaanite 94

Cant: English 91, 94 Scottish Traveller 91 Carib 94 Caviteño 101, 102, 105 Cayenne Creole French see Guyanese Creole French Caymanian 193 Cebuano 102, 103, 106, 108 Celtic 87 Chabacano 101–107 Chavacano see Chabacano Chamorro 83 Chattisgarhi 180 Chavacano 101 Chinese Pidgin English 188, 190 Chinook 53, 81, 244 Copper Island Aleut see Aleut, Mednyj Cotabateño 98, 100–102 Cree 87, 92, 93, 115, 257, 261 Creole French 82, 235 Caribbean Creole French 241 Dominican Creole French 108 Creolese see Guyanese Creole English Cushitic 93, 95, 118 Eastern Cushitic 95, 118 Czech 229 Danish 161 Dhuwaya 175 Djutongo 56, 58–60, 76 Dominican Creole French 108 Dutch 4, 53, 61, 87, 92, 94, 99, 108, 158, 161, 162, 164–171, 210, 239 IJsselmeer Standard 205–206 Flemish 165 Limburgian 202 Zeelandic 165 East Futunan 84 English: American 204, 242 Australian 175 Black 151, 152 British Black 152 Chicano 152 Hiberno 94, 95

 Index of languages

English (cont.) Indian 251 Modern 83 Old 83, 278, 289 Singapore 190 Ermiteño 101 Ewe 2, 9, 33, 34, 38, 65, 69, 70, 74 Fa d’Ambu 64 Farsi 92 Fataleka 180 Finnic: Baltic 229, 238 Pre-Baltic 229 Late Proto-Baltic 229 Finnish 126, 127, 146, 151, 229 Fon(gbe) 43, 44, 48, 53, 57, 65–67, 69–74, 97, 260 French 2, 9, 13–24, 26, 27, 29, 31–33, 35–38, 43, 44, 48, 65–68, 73, 82–84, 86, 87, 92, 93, 96, 97, 108, 115, 119, 146, 159, 161, 235, 238, 239, 241, 257, 259–261, 274–277, 288, 291, 298, 300 French 2, 16–38, 43, 44, 48, 66–68, 73, 82–84, 86–87, 92, 93, 96, 97, 108, 115, 119, 235, 238, 239, 257, 260, 261, 274 Judeo-French 274–275, 277 Old French 259, 275, 277 Popular French 238 Frisian 46, 87, 94 Town see Stedsk West 87 Gaelic: Irish 94–5 Scottish 46, 95 Gammon see Shelta Gastarbeiterdeutsch 3, 126, 128 Gbe 43–45, 53, 57, 58, 65, 69–71, 74–76, 97, 98, 298 Eastern Gbe 2, 43, 44, 53, 57, 58, 70, 71, 74, 75 German 84, 92, 116–117, 132, 133, 134, 204, 229, 270, 279, 287 Bavarian 280, 288 Immigrant 131, 135, 139, 142, 143; see also Gastarbeiterdeutsch Low German 166 Middle High 263, 279 New High 278, 279 Germanic 6, 83, 87, 229, 238, 263–266, 268, 270, 274, 275, 277, 278, 282, 286, 289 Greek 4, 64, 83, 127, 128, 132, 133, 140, 144, 146, 147, 151, 175, 226, 232, 233 Attic 182, 213 Hellenistic 213 Ionic 182, 213

Koine 4, 175, 180, 183, 199–201, 205, 208, 212 Gullah 239 Guyanese Creole French 62, 65, 146, 298 Haitian 2, 9–11, 13–38, 43, 44, 48, 65–70, 72–75, 97, 115, 241, 260 Hawaiian Creole English 49, 145 Hawaiian Pidgin English 64, 126, 145, 147, 153 Hebrew 92, 185–187, 189, 192, 205, 212, 275–278, 288, 289 Biblical 212, 277, 289 Israeli 185–187, 189, 192, 212 Hiberno-English see English Hiligaynon 100–108 Hindi 86, 175–180, 185, 190, 201, 203, 204, 206–208, 212–214 Hindi: Eastern 176 Fiji 175–178, 180, 185, 190, 201, 203, 204, 206–208, 212, 213 Overseas Hindi 214 Standard 207, 208 Western 176, 213 Hindustani 176, 181, 206, 214 Basilectal Hindustani 206, 213 Bazaar 176, 201, 203, 206 Pidgin Hindustani 208 Hlonipha 116, 118, 119 Hungarian 83, 263 Icelandic 269 Ilwana 93 Indo-European 82, 83, 87 Inuktitut 92 Iraqw 95, 118 Island Carib 94 Italic 82 Janjero 116 see also Yemsa Japanese 49, 84, 90 Javanese 87, 94 Javindo 87, 94, 122 Judeo-French 274, 275, 277 Kalaar 237–240 Kalabai  65, 237–240 Kalderash 83 Kannada 222, 223, 229–231 Kupwar 222–224, 227–231, 244 Kikongo 44, 71, 96–98, 181 Kikoongo 53, 65 Kimbundu 235 Kimwani 94 Kituba 96, 181

Index of languages

Korean 49 Korku 85, 86 Kriol/Kriyol (Northern Territory Creole English) 187, 188, 190 Kriol (Guiné-Bissau Creole Portuguese) 236 Kurux 86 Kwa 10, 43, 76, 86, 96, 98, 241, 299 Kwaio 10 Lappish 126 Latin 82, 83, 92, 119, 259 Longuda 118 Louisiana French Creole 108, 300 Ma’a 3, 95, 113–115, 119–122, 257 see also Inner Mbugu Maasai 95 Macedonian 232, 233 Magahi 190 Magindanaw 101, 102, 106 Magori 85 Maisin 85 Makonde 94 Malagasy 82 Malaitan 201, 237 Malay 94, 101 Malayo-Portuguese 101 Maltese 93 Mande 10, 298 Manding see Mandingo Mandingo 33, 235 Marathi 222, 223, 228–231 Kupwar 222, 223, 229 Martinican Creole French 235 Mauritian Creole French 62, 64, 82, 108 Mbugu 114, 115, 119, 120 Inner 95 Outer 95 Mbugwe 95 Media Lengua 45, 47–50, 89, 90, 96, 120, 121, 257 Melanesian Pidgin English 187 Michif 45, 87, 92, 93, 107, 108, 115, 119, 121, 257, 261 Milanese 206 Mindanao Creole Spanish 100–102, 105–109 Mon-Khmer 84 Muong 84 Nahali 86 Nanggu 85 Ndyuka 9, 57, 61, 63, 68, 69, 76 Neapolitan 206 Negerhollands 4, 157, 158, 161–167, 169–171, 204 see also Virgin Islands Creole Dutch New Guinean Pidgin 147, 153

see also Tok Pisin Nilo-Saharan 95 Northeast Brazilian Creole Portuguese 49 Norwegian 179 Nyakyusa 116, 118 Nyanja 235 Ojibwe 93 Omotic 116 Oriya 86 Orma 93 Oromo 95 Palauan 84 Palenquero (Creole Spanish) 97 Papiamentu 86, 97, 165 Papuan 227, 228 Para-Romani 91 Pare 95, 115 Persian 83, 213 Petjo 94, 122 Philippine Creole Spanish see Chabacano Philippine Pidgin English 146 Phla-Phela 43, 57, 65 Pig Latin 119 Portuguese 49, 50, 53, 58–61, 63, 64, 67, 75, 76, 86, 98, 101, 104, 157, 161, 164, 170, 184, 210, 236, 238 Proto-Sranan 49, 50, 53, 55, 64, 67, 68 Puquina 94 Quechua 47–50, 89, 90, 94, 257 Rajasthani 180, 181 Réunionnais 193, 260 Romance 82, 83, 164, 165, 232, 275, 278, 288, 289 Romani 83, 87, 90–92, 94, 126, 151 Romanian see Rumanian Romany see Romani Rotuman 83, 84 Rotwelsch 92, 121 Rumanian 82, 83, 232, 233, 263 Russenorsk 3, 131, 133, 134, 137, 143, 147, 153 Salishan 92 Samoan 84 Santa Cruz 85 Saramaccan 10, 43, 44, 49, 50, 55–64, 67–75, 97, 98, 301, 303 Sarnami 214 Semitic 6, 94, 263, 265, 266, 278, 286, 287 Senkyoshigo 90 Serbo-Croatian 232 Shelta 87, 91, 94, 95, 119, 121



 Index of languages

Sicilian 93 Sidama 118 Sinama 102, 106 Slavic 6, 212, 229, 263, 265–268, 270–272, 280, 282–284, 286, 287, 289 South 83, 232 Solomon Islands Pidgin (English) 236 Sorbian 6, 212, 289 Southeast Australian Pidgin English 188 Spanish 47–50, 76, 82, 84, 86, 89–92, 97, 100–109, 127, 132, 143, 147, 151, 153, 164, 184, 204, 257 Sranan 10, 44, 47, 49–55, 58–64, 67–69, 76, 98, 291, 298, 299 St. Kitts Creole French 62 St. Lucian Creole French 235 Stedsk 87, 94 Swahili 83, 93, 94, 116, 117, 120 Swedish 3, 91, 126–131, 134–138, 140, 142–144, 146–153, 161, 170, 171, 204, 260 Immigrant Swedish 127–153 Tacana 94 Tagalog 84, 101–103, 105–108 Tai 84 Takia 85 Tausug 102, 106 Tayo 108 Ternate 101 Ternateño Creole Spanish 101 Tiruray 102 To’aba’ita 180 Tok Pisin 125, 126, 145, 147, 153, 187, 190 see also New Guinean Pidgin Tongan 84

Turkish 81, 92, 127, 146, 151 Tuvaluan 87 Twi 65, 71, 162, 241 Ulithian 84 Ural-Altaic 225 Urdu 207, 208, 222, 223, 229–231 Kupwar Urdu 222, 223, 230 Vietnamese 84 Virgin 4, 157, 161 Waskia 85 Welsh 86, 87, 90, 91 West African Creole Portuguese 64, 104, 157 Wolof 235 Xhosa 119 Yakan 102, 106 Yapese 84 Yemsa 116, 118, 120 see also Janjero Yiddish 6, 83, 117, 212, 263–265, 267–275, 277–280, 282–289 Yiddish 83, 117, 263–288 Eastern/Modern 263, 280, 283, 284 Northeast 280 Old 263, 268, 278, 279, 282, 284, 286 Western 280, 286 Yoruba 33, 65, 76, 298 Zamboangueño Creole Spanish 98, 100–106, 108 Zulu 118, 119

Index of authors

Adam, Lucien 9, 38 Alleyne, Mervyn 9, 159, 194, 260, 298 Amery, Rob 175 Andersen, Roger W. 159, 192 Antinucci, F. 144 Anttila, Raimo 229 Appel, René 221 Arends, Jacques 47, 49, 51, 53–56, 62, 204, 213, 214, 291 Auer, Peter 202, 204 Auerbach, J. C. 162 Bailey, Charles James N. 158 Baker, Philip 62, 67, 157, 162, 163, 186, 234, 235, 291, 303, 304 Bakker, Peter 64, 81, 86, 87, 92, 93, 115, 211, 261 Bashevis, Y. 281 Beeldsnijder, Ruud 291–305 Berruto, Gaetano 206 Biber, Douglas 122 Bickerton, Derek 45–49, 52, 54, 55, 58, 59, 62, 64, 73, 74, 76, 125, 144, 145, 158, 163, 170, 184, 185, 191, 193, 212, 293 Biggs, Bruce 83 Blanc, H. 117, 185, 212 Blanc, M. 212 Boas, Franz 220, 221 Boretzky, Norbert 91, 234 Brana-Shute, G. 291 Brana-Shute, R. 291, 293, 299 Broch, I. 131 Broeder, Peter 169, 170 Bronckart, J. P. 144 Brousseau, Anne-Marie 18, 28, 32, 38, 72 Bruyn, Adrienne 52, 59, 69, 159, 168, 214, 234 Bubenik, Vit 175, 182, 213 Bybee, Joan 144–146 Bynon, Theodora 221, 225, 232, 235, 235 Byrne, Frank X. 59 Campbell, Lyle 221, 223, 224, 226, 227, 233 Capell, Arthur 85 Capo, Hounkpatin C. 43, 57, 65, 76 Carden, Guy 11 Chaudenson, Robert 157, 304

Chomsky, N. 13, 29, 271, 288 Civ’jan, T. V. 222, 227 Clough, James Cresswell 81 Clyne, Michael 126 Coelho, Adolpho 157 Coenen, Josée 169 Collins, Chris 18 Crofton, Henry Thomas 90 Crowley, Terry 187 Daeleman, Jan 57 Dahl, Östen 144–146 Damoiseau, Robert 34 Damsteegt, Theo 175 de Gruiter, Miel 122 DeCamp, David 234 Déchaine, Rose-Marie 34 DeGraff, Michel 27, 235 Delin, J. 271, 288 den Besten, Hans 210, 214 Dillard, J. 184 Dittmar, Norbert 158, 159 Domingue, Nicole 65, 178, 202 Dryer, Matthew S. 221 Dundes, A. 151 Edwards, V. 152 Emeneau, Murray B. 221, 224 Ervin, S. M. 117 Étienne, G. 27 Extra, Guus 74, 121, 165, 169, 240 Faber, A. 263, 274, 288 Faine, Jules 18 Fehderau, Harold W. 183 Férère, Gérard A. 27 Ferguson, Charles A. 175, 213 Filipovich, Sandra 72 Fleischmann, Ulrich 300 Flûtre, Louis-Ferdinand 19 Foley, William 221, 228 Fournier, Robert 14, 17, 18 Frake, Charles O. 101, 106 Gallois, Cynthia 178 Gambhir, Surendra K. 175, 179, 183, 204 Giles, Howard 178

 Index of authors

Goldvaser, M. 288 Golman, Walter 288 Goodman, Morris F. 9, 18, 20, 27, 32, 33, 38, 161, 191, 192 Gougenheim, Georges 32 Grant, Anthony P. 2, 53, 81, 86, 91, 95, 101, 119, 121, 128, 146 Grosz, B. J. 272, 288 Gumperz, John 221, 222, 224, 227–231, 244 Gundel, J. 271, 287, 288 Haase, A. 15 Hall, R. 27, 37 Hamers, Josiane F. 117 Hancock, Ian F. 47, 61, 91, 193, 234 Harris, John W. 188 Hartsinck, Jan J. 57 Haviland, John B. 116 Hazaël-Massieux, Guy 234, 237 Heidelberger Forschungsproject 126 Herbert, R. 116, 119 Herlein, J. D. 62, 302 Hesseling, Dirk Christiaan 165 Hewitt, R. 152 Hinskens, Frans 4, 199, 202, 204, 208, 213, 214 Hock, Hans Heinrich 206, 213, 220, 221, 224, 229 Holm, John 9, 126, 234, 241 Holmquist, Jonathan 204 Hoogbergen, Wim 301, 303, 305 Hopper, P. J. 13, 145 Horn, L. 271, 289 Huebner, Thom 158 Hull, A. 19 Huttar, Georges L. G. 9, 38, 57, 59, 63, 76 Huttar, Georges L. M. 9, 38, 57, 59, 63, 76 Igla, Birgit 91 Jahr, E. H. 131 Jakobson, Roman 221, 224 Jeffers, Robert J. 221, 233 Jennings, William 62, 65, 291 Johnson, H. 215 Johnson, K. 215 Joseph, Brian D. 27, 213, 222, 226, 232, 233 Joshi, A. K. 272, 287–289 Katz, D. 286, 289 Kaufman, Terrence 11, 82, 182, 183, 186, 190, 191, 234, 235, 250, 251, 257, 260, 261 Kayne, Richard S. 29 Keesing, Roger M. 10, 38 Keller, Saskia 152, 178, 184, 201, 299 Kihm, Alain 236–237 King, R. 76, 116, 263, 274, 288

Kinyalolo, Kasangati K. W. 18, 38 Klein, H. 299 Klein, W. 158, 159, 169 Kolbusa, Stefanie 116 Koopman, Hilda 10, 18, 38, 235 Kotsinas, Ulla-Britt 3, 125–127, 130, 131, 138, 145, 152, 157, 159, 170, 171 Kouwenberg, Silvia 4, 5, 98, 99, 204, 213, 219, 237–240 Kretschmer, P. 225 Kroch, Anthony 278, 286 Kuiper, Franciscus B. J. 85 Kunene, D. P. 116, 118 Labouret, Henri 76 Labov, William 7, 151 Lalleman, Josine 169 Lamberti, Marcello 118, 119 Lamontagne, Greg 29 Lamur, Humphrey 291–294, 297, 302 Le Page, Robert 152, 201, 259 Leach, J. 151 Lefebvre, Claire 2, 9–38, 44, 45, 48, 66, 67, 72–74, 97, 115, 260 Lehiste, Ilse 221, 233 Lenders, Maria 303 Lichtveld, Ursy 302 Lightfoot, David 13 Lippi-Green, Rosina 182 Lipski, John M. 102, 104–106 Lumsden, John S. 10–13, 18, 28–30, 33, 34, 37, 38, 73, 74 McKaughan, Howard P. 101 McLeod-Ferrier, Cynthia 291, 303 McMahon, April S. 221, 233 McWhorter, John H. 50, 53 Magens, J. M. 163, 167 Mark, Y. 62, 97, 105, 122, 140–145, 158, 164, 166, 269, 272, 281, 288, 289 Markey, Thomas 205 Masica, Colin P. 222, 224, 225, 233 Massam, Diane 14, 18 Meijer, Guus 234 Meisel, Juergen 126, 131, 132, 289 Mesthrie, Rajend 175, 179 Miller, R. 144, 286 Milner, J.-C. 15, 31 Milroy, James 297, 298 Milroy, Leslie 297 Mohan, Peggy 175 Moshavi, A. 277, 289 Mous, Maarten 3, 81, 88, 92, 93, 95, 113, 114, 122, 211 Mufwene, Salikoko S. 11, 96, 180–184, 190, 191, 193, 194, 205, 212, 234, 239, 240, 242, 245, 304

Index of authors

Mühlhäusler, Peter 125, 137, 144, 193, 207, 210 Muyrers, Sabine 291, 292, 299, 300, 302–304 Muysken, Pieter 2, 3, 11, 38, 45, 47, 48, 69, 86, 89, 121, 157, 199, 209, 210, 213, 214, 221, 234, 235, 304 Nida, Eugene A. 183 Nurse, Derek 93 Newman, Roxana Ma 119 Newman, P. 119 Odo, Carol 47, 64, 158 Oostindie, Gert 291–295, 297, 299, 300, 303, 304 Opatoshu, Y. 281 Ornstein-Galicia, J. 152 Osgood, C. E. 117 Öskök, B. 151 Pares, Richard 301 Pasch, Helma 183 Pawley, Andrew 83 Penfield, J. 152 Perl, Matthias 60 Perlmutter, David 287, 289 Pintzuk, Susann 278, 289 Post, Marike 2, 24, 54, 58, 64, 66, 67, 93, 104, 127, 190, 194, 207, 208, 238, 288 Postma, Johan 55, 76, 301 Price, Richard 55, 58, 298, 301, 302 Price, Sally 299, 301, 302 Prince, E. F. 5, 6, 263, 271–273, 277, 278, 287, 289 Rampton, B. 152 Rayfield, J. R. 117 Rekdal, Ole-Bjørn 118 Rens, L. L. E. 54, 55, 76, 291, 293 Riego de Dios, Sister Maria Isabelita 101 Ritter, Elizabeth 18, 33, 38 Rivas, Father Carlos 101 Rivet, Paul 76 Ross, Malcolm 84, 85 Salmons, Joseph C. 117 Sampson, John 90 Samuels, Michael 213 Sandefur, John R. 189 Sankoff, Gillian 13, 125, 145, 190 Santorini, Beatrice 269, 278, 280, 286, 289 Schneider, Edgar W. 193 Scholtmeijer, Harrie 204, 205 Schumann, C. L. 53, 58, 59, 63 Schumann, John H. 158 Sebeok, Thomas 225 Selinker, Larry 158

Seuren, Pieter A. M. 191 Sgroi, Salvatore 206 Siegel, Jeff 4, 12, 175, 176, 179–182, 184, 190, 191, 200–214 Silverstein, Michael 244 Sinclair, H. 144 Singh, Rajendra 213 Singler, John V. 65, 190, 234, 241, 291, 298 Smart, Bath Charles, 90, 166 Smith, J. C. 280, 283, 288 Smith, N. 1, 2, 43, 46, 47, 49–51, 53, 58, 59, 61, 64, 69, 72–75, 81, 86, 90, 97–99, 105, 115, 204, 209, 210, 214, 215, 237, 280 Smout, Kary D. 90 Sobrero, Alberto 206, 207 Speedy, Karin 300, 304 Stewart, William A. 11 Sunio, Delicia, 101 Swadesh, Morris 60, 106, 224 Syea, Anand 234 Sylvain, Suzanne 2, 9, 17, 20, 27, 32, 38 Tabouret-Keller, Andree 152, 178, 184, 201 Talmy, L. 266, 289 Tannen, D. 141 Taube, Moshe 286, 287 Thelander, Mats 204 Thomason, Sarah Grey 5, 6, 11, 82, 92, 95, 182, 183, 186, 190, 191, 234, 235, 244, 249–251, 257–261 Thomson, George 175 Tinelli, Henri 28 Traugott, E. 13, 143–146 Travis, Lisa 29 Trubetzkoy, Nikolai S. 223, 225 Trudgill, Peter 12, 175–180, 184, 201, 212, 213 Trumper, John 206 Tseytlin, A. 281 Valdman, Albert 18, 20, 27, 194 Valkenburg, Dirk 302 van den Bouwhuijsen, Harry 305 van der Voort, Hein 64 van Deursen, A. 297 van Dyk, Pieter 298 van Helvert, Korrie 169 van Hout, Roeland 214 van Lier, Rudolf 291, 294 van Rheeden, Hadewych 122 van Rooyen, Christian Stephanus 116, 118, 119 van Rossum, Cefas 204 Veenstra, Tonjes 1, 59, 64, 286 Vermeer, Anne 169 Voorhoeve, Jan 9, 302



 Index of authors

Warren, George 52, 56 Weinreich, Uriel 220, 222, 235, 263, 274, 286, 287, 289 Weinstein, S. 272, 288 Wekker, Herman 52, 158, 191 Westerman, D. 33 Wexler, Paul 212, 286, 289 Whinnom, Keith 12, 159 Williams, Jessica 190, 191

Williamson, Kay 99, 213 Wilson, Robert 221, 222, 224, 227–231 Windfuhr, Gernot 213 Winford, Donald 208, 214 Wooding, Chris J. 43 Wullschlägel, H. R. 58, 59 Wurm, Stephen A. 85 Yilma, Aklilu 116

Index of subjects

access 46, 304 accommodation 254 theory 178 acquisition: adult second language 241 first language 282 imperfect second language 210 L2 see acquisition, second language language 211 mechanisms 205 moment of 122 order of 142 second language 12, 117, 125, 157, 160, 170, 238, 243 adjectives 71 adlexification 90 ambiguity 283 analytic decomposition 130, 135, 136, 137, 141 analyticity 142–4 articles 164, 165 assimilation 258 phonological 150 attitude 121 avoidance 121 backslang 91 basilect 158 bilingualism 85, 220, 221, 228 additive 85 compound 117 full 257 replacive 85 black overseers 292, 297 borrowers 258 borrowing 83, 119, 121, 250, 251, 254, 264, 267 additional 114 core vocabulary 82 derivational 82 discourse function 268 feature 267, 271; see also feature inflectional 82 massive 265 phonemic 82 replacive 114, 117 semantic 282, 283, 284

structural 253 syntactic 268, 275 cafeteria principle 184, 242 calques 151 case marker 28 causation: internal 279 multiple 241, 243 change: contact-induced 249 demographic 200 distributional 284 economic 200 language 160 lexical 279 linguistic 13; see also language change mechanisms 235 phonological 64, 229 political 200 pragmatic 267, 276, 279 radical 45 rule 209 semantic 276, 279, 282, 283 structural 253 syntactic 267, 276, 278, 279, 282 children 296 circumlocution 118, 130, 136, 137 circumstancialist approach 227 cliticization 145 code-mixing 85 code-switching 120, 249 coincidence of form 235–7 communication: inter-ethnic 46, 47, 52, 206, 231 intergroup 186, 188 means 55, 59 needs 106, 182, 186 range of topics 121 complementation say- 239 complexity 202, 204 conflation 236, 236 congruence 234, 238 consistency 192 construction: periphrastic 13

 Index of subjects

construction (cont.) serial verb 98 contact 46 between dialects 201 dialect 180 language 175, 180, 279, 282, 283, 292 phenomena 249 situation 256 social see social contact two-language situation 256 variety 210 continuity 296 continuum: koine 208 post-koine 207, 208 conventionalization 159 convergence 200, 219, 220, 222, 227, 230, 232, 235, 238, 240, 241–3, 252, 253 in creole studies 234 mechanisms 228, 231 multidirectional 229 unidirectional 229 conversion 72 copying 11, 13, 28, 36 core-periphery model 225 creativist model 105 creole: continuum 304 dialects 304 genesis 9, 10, 11, 37, 45, 48–50, 51, 53, 58, 59, 63, 64, 81, 108, 125, 145, 148, 149, 157, 160, 175, 184, 187, 189, 199, 203, 209, 210, 219, 234, 251–3, 255, 256, 291, 298 radical 10 see also language, creole creole genesis 145, 249, 254 creolization 108 language 292 creolizers 148, 255 creoloid see creole, genesis cryptolect 87 see also language, secret culture 297 dekoineization 208 demography 161, 162 historical 47, 50, 51, 52, 55, 65, 291 determiners 17–19, 21, 23–6, 28 agglutinated 16 definite 14, 15, 66–8, 166 deictic 20, 67 partitive 15 postnominal 17, 66–8 dialect 68, 200, 204, 229 compromise 179, 200 cultivated 206

de-localized 206 levelling 10, 30, 201 mixing 201 new 183, 212 plantation 304 transition zone 207 dialect: levelling 12, 34 dialects 180 differentiation 303 diglossia 231 direct speech 129 directionality 221, 222 discourse 91 function 268, 271, 272, 276, 278 particles 102, 159 style 239 divergence 228, 232, 239, 240 DP 18, 67, 68, 73 elaboration 149 entry: lexical 3 epithetic vowel 165 ethnicity 52, 121 distinct 92, 227–9 dual 46 emerging 46, 52, 87 new 46, 96, 258, 259 ethnolect 159 event: 1 162 2 162 3 162 time see reference point expansion 125, 207 grammatical 125 expletive NP 268, 270 extension: metaphorical 114 factorization principle 192 feature: areal 256 borrowing see borrowing dialect 205 diffusion 220–3 idiosyncratic 179, 204 marked 256 mixing of 201, 203 morphological 114 semantic 12, 137–40 shared 181, 222, 227, 243 substrate 250 syntactic 12 filtering 229, 237, 238 Sprachbund 237

Index of subjects 

focus 164, 269, 271–2, 278 contrastive 72 Foreigner Talk 160 form/meaning 158, 235 frequency: increase 282 reduction 279 functional category 10, 13 morphemes 102

intermarriage 46 intermediary 297 interpretation: type 18 intertranslatability 220, 224, 227, 229 intertwining see language isogloss 224

genesis: see creole; mixed language; pidgin gesture 137 glottochronology 60 gradualist model 210 grammar: autonomous 208 polytectal 208 substrate 109 grammatical marker 145–9 grammaticalization 13, 63, 64, 82, 145, 146, 159, 190 speed of 159 group: dominant 255, 258 shifting 251 group-distinctiveness 121

koine 183, 205 bourgeois 207 expanded 207 formation 175, 182 gradual 207 immigrant 175, 202, 206, 210 nativized 207 proletarian 207 regional 176, 202, 206, 210 stabilized 207 urban 206 koineization 199, 201–3, 206, 210, 212, 249 demographic 200 gradual 179

highlighter 72 postposed 72, 74 preposed 73, 74 history 53, 65 creole 108 homogenization 303 hybridization 159 tertiary 159 ideophone 98, 118 idiosyncratic features see features imitation 159 in-group 90 information: content 17 structure 271, 272 innovation 209 innovator see linguistic(s) input 159 intelligibility: mutual 182 interaction: public 228 interdialect 201 interference 159, 222, 249 shift-induced 250, 251, 254, 256 interlanguage 158, 170, 171, 192, 220, 221 formation 249 migrant workers’ 158

jargon 53, 186

labiovelars 98 language: abrupt creole 256 abrupt mixed 258 adstrate 241 change 298, see also change colonial 157 contact 81, 175, 188, 191, 211, 252, 254, 263 creole 12, 46, 62, 81, 97, 107, 185, 205 dependent 45, 48 and dialect 183, 184 dominant 45, 48, 255, 258 dominant substrate 75 dominant superstrate 61 family 223, 224 gradually-developing mixed 258 immigrant 127 intertwined 91, 96, 97, 107, 108, 122 intertwining 81, 86, 88, 92, 94, 96–8, 100, 108, 109 jargon see jargon lexical donor 88, 108 lexical replacive 90 lexifier 4, 60, 82, 157, 163, 253 link 206 mediator 231 mixed 9, 45, 47, 49, 58, 59, 81–6, 101, 107, 113, 114, 115, 122, 125, 251, 257 mixed replacive 90 mixing 10, 85, 259 modified target 186

 Index of subjects

language (cont.) new 45, 96, 106, 183, 187, 188, 212, 252, 258 norm 160 original 250, 258 pidgin 46, 53, 59, 81, 158, 205 pidginized creole 163 primary 45, 53 respect 115 ritual 121 second 256 second first 55 secondary 45 secret 87, 91, 119, 121, see also cryptolect semi-creole 211 shift 100, 255, 256 single parent 251 source 258 spoken 239 standard 204, 206 structural donor 88, 108 substrate 9–11, 13, 45, 48–9, 51, 73, 75, 83, 96–7, 102, 103, 105, 107, 109, 125, 159, 165, 168, 171, 185, 189, 190, 209, 239–1, 243, 255 superstrate 9–13, 45, 48, 49, 61, 75, 84, 105, 185–6, 239, 240–1, 243 symbiotic mixed 90, 91 taboo 115 target 10–13, 158, 160, 170, 182, 185, 186, 206, 230, 242, 250, 255 transmission 53, 54, 55 variation 298 younger 45, 49 see also language learning Language Bioprogram Hypothesis 45, 47, 49–51, 53, 75, 191, 199 language contact: see contact language learning: first 46 imperfect 256 imperfect second 185, 186, 250, 251, 264 incomplete second 159 second 45, 157, 169, 185, 186, 191, 212, 255 levelling 176, 177–8, 185, 187–90, 192, 201, 204–7, 210 dialect see dialect gradual 178 lexeme: new 117 old 117 lexical: over-use 130–4, 141 lexical category superstrate 11 lexical entry 10–14, 17, 20, 113 functional category 10, 11, 13, 18, 19, 26, 27

lexical item closed class see lexical item, functional category functional category 47, 61, 64, 265 superstrate 17 Lexical Learning Theory 73 lexicon 61, 227 basic 253, 258 core 82, 86 cultural 58 non-basic 253 pidgin 53 relexified 10 source 99 stable 53 superstrate 109 unstable 53 lexicostatistics 60, 61 life expectancy 296 life-cycle: creole 208 koine 208 lingua franca 82, 176 regional 181, 200 linguistic(s): area 220, 223, 227, 228, 233 change see change consequences 297, 303 creativity 256, 258 diversity 228 historical 61, 63, 113, 159, 219, 249, 253 influence 299 innovator 298 results 259 socialization see socialization universals see universals loanshifts 117 loanwords 61, 83 loss: gender marking 222 infinitive 222, 226, 232 language 116 phonemic 229 markedness 180, 191, 243 marronnage 58 matching 273 meaning 114 method: quantitative 208, 213 migrant 202 migration 54, 59 mixed language creators 258 genesis 11, 81, 250 mixed language genesis 59, 101 mixed marriage 87, 92

Index of subjects

mixing 176, 186, 209 dialect see dialect language see language modification 119 postnominal 273, 275 monogenesis 210 morphology 100, 129 borrowed 265 bound 98, 265 derivational 265 inflectional 253, 265, 266 morphosyntax 61 motivation 304 multilingualism 229 nativization 47, 179, 186, 189, 202, 207, 210 naturalness 180 negation: marker 164 negotiation implicit linguistic 255, 256 neologism 119 nouns: generic 15 locative 168, 171 mass 15, 17, 18 null string 11, 115 phonological 13, 28 One-to-One Principle 192 optimalization: grammar 234 over-application: sound-change 119 overlap cohort 55 palatalization 241 para-Romani 91 paralexicon 90, 114, 120 paralexification 88, 90, 113, 114, 115, 117 parataxis 129 passive 167 periphrastic marking 165, 190, 192 periphrastic structure 266 phase: establishment 46 growth 46, 295, 296, 299 phonetic source 18, 26 phonetic string 11, 12 phonology 10, 12, 165 diachronic 83 pidgin: creators 256 crystallization 160 extended 62 genesis 175, 187, 209, 249, 251, 252, 253, 254, 256

grammarless 64 jargon see jargon stable 47, 149 trade 250 unextended 62 pidginization see pidgin genesis Pidginization Hypothesis 158 Pidginization Index 163 plural: generic 203, 204 marker 97, 167, 203, 204 possession 166 postpositions 238 pragmatics 11, 264, 267 pre-koine 207 predicate cleft 73 prenasalization 98 preposition: locative 164, 171 prepositions 168, 238 complex 168 prestige 206, 298 pronouns 164, 167 prototype 254 quantitative: method 205 question element 69 question words 69 analytic 69, 70 synthetic 69 questioned Semantic Unit 69 reallocation 177, 185–8 reanalysis 10, 13, 34, 35, 37, 279, 282 recreolization 55, 107 reduction: phonological 64, 68, 145, 150 semantic 67, 145 reduplication 70, 72 register 115, 120 avoidance 116 respect 115 regularity 191 reinforcement 189, 205, 207, 242 substrate 189 reinterpretation 166 relabeling 14 relabelling 11, 13, 28, 35 relationship: genetic 83, 84, 85, 183 relexification 10, 11, 37, 47, 48, 49, 66, 75, 115, 210, 212 creole 12 hypothesis 45 target 13 relexifiers 11, 12, 28, 37



 Index of subjects

repetition 130, 136, 141, 144 repidginization 55 replacement 208, 230 representation: phonological 10, 11 residence 296 respect 121 restructuring 183, 185–6, 189, 209, 210 retention 190, 205, 239 substrate 236 superstrate 236 reversal: rule 119 syllable 119 salience 67, 191 secrecy 121 selection 207, 242 selection principles 191 semantic features 10 semantic(s) 11, 264 borrowing see borrowing extension 118 features see features field 99, 265 over-extension 130, 138–41, 145, 150, 152 reduction see reduction transparency 177, 191, 192 shared features see features similarity 240 typological 84 simplification 108, 113, 125, 176, 186, 201, 209, 234, 235 slang 120, 151, 152 slaves 51–5 trade in 55–7 social class 48, 206 social contacts 52, 292, 303 foreign residence 303 imprisonment 303 leisure 301 resistance 302 work 300 social need 46 social network: external relations 299, 303 model 298 social stratification 292, 293, 296, 303 social structure 291, 292 socialization: linguistic 163, 211 sociodialectology 204 sociohistory 291 sociolinguistics 255, 258 solidarity: in-group 121 Sprachbund see linguistic area

stabilization 179, 200, 207, 210 standard language see language strategies 149 compensatory 130, 141, 148 structure: marked 238 source language 258 syllable 165, 237 subject-postposing 277, 278, 279 substrate approach 45 substratum see language substrate subsystems: linguistic 180, 181 symbiosis 108 synonyms 114, 129, 130 near 118 syntactic features 10 syntax 61 time-depth 228 TMA: anterior 62, 64 completive 62 definite future 62 durative 62 future 62, 284 future-in-the-past 284 immediate completive 62 indefinite future 62 irrealis 62 marker 61–4, 166, 167 markers 170 nonpunctual 64 past 62, 282, 283, 284 pluperfect 279, 280, 282, 283, 284 present 283, 284 present perfect 280, 282, 284 preterite 280, 282, 283 progressive 62 sequence of tenses 282, 284 system 34, 61–3, 164, 170 tone: lexical 98 transfer 82, 149, 159, 186, 189, 205 transparancy see semantics typology 82, 97 unification 178, 192, 200, 202 uniformitarian Hypothesis 160 universalist approach 45, 75 universals: linguistic 46, 171, 241, 243 second language learning 169 variants 186, 188, 189, 211 marked 179 spectrum of 304

Index of subjects 

variation 148, 151, 153, 167, 178, 187, 188, 201, 206 interdialectal 201, 203 variety: indigenized 211 verb: auxiliary 266 infinitive 129 serial 165, 171 verb-particle constructions 169 verbal interaction 297 vocabulary: restricted 129 see also lexicon

wave model 225 wave theory 221 word: archaic 118 function 120; see also lexical item, functional category taboo 118 unmarked 135 word order 12, 164, 240 functional category 12 lexical category 12 surface 240

In the CREOLE LANGUAGE LIBRARY series the following volumes have been published thus far or are scheduled for publication: 1. MUYSKEN, Pieter & Norval SMITH (eds): Substrata versus Universals in Creole Genesis. Papers from the Amsterdam Creole Workshop, April 1985. 1986. 2. SEBBA, Mark: The Syntax of Serial Verbs: an Investigation into Serialisation in Sranan and other languages. 1987. 3. BYRNE, Francis: Grammatical Relations in a Radical Creole: Verb Complementation in Saramaccan. With a foreword by D. Bickerton. 1987. 4. LIPSKI, JOHN M.: The Speech of the Negros Congos in Panama. 1989. 5. JACKSON, Kenneth David: Sing Without Shame. Oral traditions in Indo-Portuguese Creole verse. 1990. 6. SINGLER, John V. (ed.): Pidgin and Creole Tense/Mood/Aspect Systems. 1990. 7. FABIAN, Johannes (ed.): History from Below. The ‘Vocabulary’ of Elisabethville by André Yav: Text, Translations and Interpretive essay. 1990. 8. BAILEY, Guy, Natalie MAYNOR and Patricia CUKOR-AVILA (eds): The Emergence of Black English: Text and commentary. 1991. 9. BYRNE, Francis and Thom HUEBNER (eds): Development and Structures of Creole Languages. Essays in honor of Derek Bickerton. 1991. 10. WINFORD, Donald: Predication in Carribean English Creoles. 1993. 11. BYRNE, Francis & John HOLM (eds): Atlantic Meets Pacific: A global view of pidginization and creolization. Selected papers from the society for Pidgin and Creole Linguistics. 1992. 12. BYRNE, Francis & Donald WINFORD (eds): Focus and Grammatical Relations in Creole Languages: Papers from the University of Chicago Conference of Focus and Grammatical Relations in Creole Languages. 1993. 13. ARENDS, Jacques (ed.): The Early Stages of Creolization. 1995. 14. KIHM, Alain: Kriyol Syntax. The Portuguese-based Creole language of Guinea-Bissau. 1994. 15. ARENDS, Jacques, Pieter MUYSKEN and Norval SMITH (eds): Pidgins and Creoles. An introduction. 1995. 16. CLEMENTS, J. Clancy: The Genesis of a Language: The formation and development of Korlai Portuguese. 1996. 17. THOMASON, Sarah G. (ed.): Contact Languages. A wider perspective. 1997. 18. ESCURE, Genevieve: Creole and Dialect Continua. Standard acquisition processes in Belize and China (PRC). 1997. 19. SPEARS, Arthur K. and Donald WINFORD (eds): The Structure and Status of Pidgins and Creoles. 1997. 20. RICKFORD, John R. and Suzanne ROMAINE (eds.): Creole Genesis, Attitudes and Discourse. Studies celebrating Charlene J. Sato. 1999. 21. McWHORTER, John (ed.): Language Change and Language Contact in Pidgins and Creoles. 2000. 22. NEUMANN-HOLZSCHUH, Ingrid and Edgar W. SCHNEIDER (eds.): Degrees of Restructuring in Creole Languages. 2000. 23. SMITH, Norval and Tonjes VEENSTRA (eds.): Creolization and Contact. 2001.