Deconstructing Creole (Typological Studies in Language)

Deconstructing Creole

Typological Studies in Language (TSL) A companion series to the journal Studies in Language. Volumes in this series are functionally and typologically oriented, covering specific topics in language by collecting together data from a wide variety of languages and languages typologies.

General Editor

Assistant Editors

Michael Noonan

Spike Gildea

Suzanne Kemmer

Wallace Chafe

Matthew S. Dryer

Bernd Heine

Ronald W. Langacker

Doris L. Payne

Paul J. Hopper

Bernard Comrie

John Haiman

Sandra A. Thompson

Charles N. Li

Frans Plank

Andrej A. Kibrik

R.M.W. Dixon

Kenneth L. Hale

Dan I. Slobin

Andrew Pawley

Jerrold M. Sadock

University of Wisconsin

University of Oregon

Rice University

Editorial Board Santa Barbara San Diego Leipzig

Santa Barbara Canberra Canberra

Buffalo

Oregon St Paul

Konstanz

Cambridge, Mass.

Köln

Pittsburgh

Santa Barbara Moscow

Berkeley

Chicago

Volume 73 Deconstructing Creole Edited by Umberto Ansaldo, Stephen Matthews and Lisa Lim

Deconstructing Creole

Edited by

Umberto Ansaldo University of Amsterdam

Stephen Matthews University of Hong Kong

Lisa Lim University of Amsterdam

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 Deconstructing creole / edited by Umberto Ansaldo, Stephen Matthews and Lisa Lim. p. cm. -- (Typological studies in language, ISSN 0167-7373 ; 73) Includes bibliographical references and index. 1. Creole dialects. 2. Typology (Linguistics) I. Ansaldo, Umberto. II. Matthews, Stephen, 1963PM7831.D428   2007 417'.22--dc22 2007013803

isbn 978 90 272 2985 4 (Hb; alk. paper) © 2007 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents Acknowledgments Deconstructing creole: The rationale Umberto Ansaldo & Stephen Matthews 1 2 2.1 2.2 3 3.1 3.2 3.3 3.3.1 3.3.2 3.3.3 4

xi 1

On deconstruction  1 Deconstructing creole  3 Creole studies and linguistics  3 Introducing the volume  4 History of beliefs  8 A brief history of creole ideas  8 From the Language Bioprogram to the Creole Prototype  10 Creole myths  12 The myth of simplicity  12 The myth of decreolization  13 The myth of exceptional diachrony  13 Final remarks  14

Part 1 Typology and grammar Typology and grammar: Creole morphology revisited Joseph T. Farquharson 1 2 2.1 2.2 2.3 2.4 3 4 4.1 4.2

Introduction  21 Word-formation  22 Affixation  23 Reduplication  24 Compounding  25 Zero-derivation  26 Transparency  27 Inflectional morphology  28 Affixational inflectional morphology  29 Reduplicative inflectional morphology  30

21



Deconstructing Creole

5 5.1 5.2 6

Complex morphology  31 Complex morphology as inflectional (affixational) morphology?  31 Complexity and age  31 Conclusion  34

The role of typology in language creation: A descriptive take Enoch O. Aboh & Umberto Ansaldo 1 2 2.1 2.2 2.3 2.4 3 3.1 3.2 3.3 3.4 4 4.1 4.2 4.3 4.4 5

Introduction  39 Contact languages and ‘simple grammars’  40 Inflection and simplification  40 The Noun Phrase as a case study for competition and selection  42 The Feature Pool  44 Simplification again  46 Competition and selection in English, Gbe and the Suriname creoles  47 Properties of the noun phrase in English, Gungbe and the Suriname creoles  47 The function of determiners in the competing languages and the emerging creole  50 Intertwining syntax and semantics  52 Summary  56 Congruence, frequency and replication in Sri Lanka Malay  57 Morpheme sources  58 Structural features of case in SLM, Sinhala and Tamil  58 Functional alignments  60 Summary  62 Conclusions  63

Creoles, complexity and associational semantics David Gil 1 2 3 4 4.1 4.2 5 6

39

Creoles and complexity  67 Associational semantics  71 Associational semantics and complexity  75 Measuring complexity: The association experiment  79 Experimental design  81 Running the experiment  86 Results  88 Further questions: Why languages vary and why languages ‘undress’  90

67



Table of contents 

Admixture and after: The Chamic languages and the Creole Prototype Anthony P. Grant 1 2 3 4 5 6 7 8

9

The Creole Prototype  109 Introduction to the Chamic languages  111 Where the Chamic languages fit in genealogically  112 Influences on the Chamic languages: Whence and where  114 Lexical elements of unknown origin in Chamic  120 Aspects of Chamic typology: Phonology, morphology and syntax  121 Transfer of fabric in Chamic: The lexicon  126 How Indochinese Chamic languages ‘got this way’: The replication of the effects of the Creole Prototype as a dynamic diachronic process  130 Conclusions  136

Relexification and pidgin development: The case of Cape Dutch Pidgin Hans den Besten 1 2 2.1 2.2 2.3 2.4 2.5 3 3.1 3.2 3.3 4 5 6

109

141

Preliminaries  141 The CDP sentence: Relexification and stripping (and more)  142 SOV word order and the history of CDP  142 Relexification and stripping  144 Relexification and Pro-drop  147 Negation, temporal anchoring and ‘have’ and ‘be’  149 Looking ahead  151 CDP DPs: Relexification, stripping and adaptation  151 DP-internal Word Order  151 Petrified endings? Nominalizations?  154 Conclusion  155 CDP PPs  155 CDP clauses again  157 Conclusions  159

Part 2 Sociohistorical contexts Sociohistorical contexts: Transmission and transfer Jeff Siegel 1 2 2.1 2.2 2.2.1

Introduction  167 Transmission of the lexifier  167 Break in transmission  167 Normal transmission  169 Lack of evidence of a pre-existing pidgin  169

167

 Deconstructing Creole

2.2.2 2.2.3 2.3 3 3.1 3.2 4 4.1 4.2 4.3 5

Existence in some creoles of morphology from the lexifier  172 Conventional language change  172 Discussion  175 Transmission of substrate features  177 Language transfer  177 Substrate reinforcement  185 Associated ideologies  187 The development of post-colonial ideology in the ‘New World’  187 Discussion  191 ‘Imperfect’ learning  194 Conclusion  195

The sociolinguistic history of the Peranakans: What it tells us about ‘creolization’ Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene 1 2 2.1 2.2 2.3 2.4 2.5 3

Creoles and the notion of ‘creolization’  203 The Peranakan population and the genesis of Baba Malay  206 The non-traumatic birth of the Peranakans  207 Multilingualism and the nature of transmission  209 The Peranakans as privileged British subjects  210 Baba Malay features  212 Summary and reflections  218 Final remarks  220

The complexity that really matters: The role of political economy in creole genesis Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz 1 1.1 2 2.1 3 4

4.1 4.2 4.3 4.4 4.5

203

Introduction  227 Purpose  229 Interaction, not simply ‘access’  229 Correlating colonization and types of interaction  230 Beyond correlation: The descriptive and explanatory power of the Matrix of Creolization in relation to key debates in creolistics  231 Toward a typology of colonization and creolization: Political economy and the continua, matrix, and space of Afro-Caribbean creolization  234 Superstrate economies  239 Superstrate ideologies, cultures, and linguistics  242 Superstrate politics  245 Substrate economies  248 Substrate ideologies, cultures, and linguistics  251

227

4.6 5

Table of contents 

Substrate politics  255 Conclusion: The linguistic outcomes  258

Creole metaphors in cultural analysis Roxy Harris & Ben Rampton 1 2 3 4 4.1 4.2 5 Index

265

Introduction  265 Ideologies in creole linguistics  266 Creole language study and the shift in linguistics  270 Interaction as a site of ‘transcultural’ encounter  273 Interactional siting: Ritual and remedial interchanges  276 Processes of symbolic evocation: Historical consciousness in situated code-switching  278 Conclusion  281 287

Acknowledgments Our thanks go to everyone who has supported this project for the past two years. First and foremost, we thank the contributors to the volume for agreeing to be part of the general theme, and for their hard work in revising their chapters and reviewing papers, as well as other colleagues who offered their comments and suggestions to improve our contributions: Ana Deumert and Daniel Nettle. We are particularly grateful to Enoch Aboh for the many discussions and suggestions in the course of compiling this volume, and Nick Faraclas for his insightful advice and constant support from the early days of the project. We also owe our thanks to Kees Vaes at John Benjamins for his support of the project, and to Mickey Noonan, the series editor, for his efficiency and for wanting to include this volume in this series. Finally we wish to express our appreciation to Michelle Li for initial formatting of the papers and Martine van Marsbergen and Patricia Leplae for their expeditious handling of the production of this volume. UA & SM & LL Amsterdam & Hong Kong, March 2007

Deconstructing creole The rationale* Umberto Ansaldo* & Stephen Matthews** *University of Amsterdam **University of Hong Kong

1

On deconstruction

The term ‘deconstruction’ is notoriously difficult to grasp and has often been defined in negative terms, i.e. in reference to what it is not, rather than what it is (e.g. Derrida 1985). It is usually accepted that, in deconstructing, we critically analyze texts and concepts and do not merely engage in a perverse destruction of them. The sense in which we (loosely) interpret the deconstruction of ‘creole’ in this volume is indeed a constructive one: we engage with the conceptual foundations of the notion of creole (not with the texts themselves), in order to critically assess the present state of our knowledge. Crucially, the chapters of this book all offer novel constructive approaches as viable alternatives to the conceptual frameworks being questioned. This is in line with at least one interpretation of deconstruction as consisting essentially of two phases: reversal and displacement (Derrida 1981). This is possible because the critical work of much of the ‘creole paradigm’ (see section 3 below) has already been presented elsewhere (DeGraff 2001b, 2003, 2005; Mufwene 2001); it is therefore time to move on and shift our perspective towards new horizons in the study of language contact and change, of which creole studies are part. In doing so, we hope to contribute to a better integration of creole studies within the field of language creation (or genesis), and to raise the general awareness of all it has to offer to the field of general linguistics. An intrinsic aspect of scientific enquiry is the formulation of theories or models of the world to help us understand the objects under investigation. In theory building, systems of thoughts are established which can then be shared by all the scientists who subscribe to, or believe, in a given theory, i.e. those who work towards the advancement of the theory. While Popper (1959) showed us that the way to advance science was to test a theory and discard it as soon as it was found to be inappropriate, what we can observe in the case of linguistics (among other disciplines) is how the theory turns into a paradigm in the Kuhnian sense, a Weltanschaung that conditions the individuals within it *

We thank Hugo Cardoso and Lisa Lim for comments on this chapter.



Umberto Ansaldo & Stephen Matthews

(see section 3.2). Such conditioning expresses itself in the tendency to always work within the paradigm for the advancement of the paradigm, rather than seriously consider the criticism addressed toward it. This behavior, essentially a type of pre-scientific belief (Gray 2004), is often presented as a positive strategy for the advancement of ‘Theory’, but can also be seen as a manifestation of the evocative power that a paradigm can accumulate (Hinzen 2006). That power is quantifiable in terms of the number of active believers a theory may have, the public impact it receives in the form of citations and popularity (converts), as well as the academic weight it carries in terms of publications devoted to and controlled by it. According to Feyerabend (1975), Galileo’s views prevailed not because of their being scientifically sound, but because of his arguing in the right language of the time (Italian) and because of his having understood the connection between a new trend in society and the ideology for representing it (Copernican astronomy vs. old scholasticism); in other words, he was part of a stronger paradigm. In referring to the term ‘creole’ as a paradigm, we want to point out that many of the orthodox views underlying creole studies are only partly due to appropriate scientific enquiry and viable theory-building. Much of the paradigm is constructed and upheld because of the social and political (power-based) dynamics that underlie it, striving for preservation and propagation rather than critical introspection and analysis (see DeGraff 2001b, 2005). Perhaps the most powerful force in creating the creole paradigm within creole studies was Bickerton’s Language Bioprogram Hypothesis (LBH, Bickerton 1981, 1984), which exploited an even more powerful paradigm, namely Chomsky’s generative enterprise. Note however how, from a strictly generative perspective, the notion of an LBH not identical to UG is untenable, a point clearly realized by Chomsky (e.g. Chomsky 1986, see also e.g. DeGraff 2001a; Rizzi 2001). To this date, obviously, it is extremely difficult not to mention Bickerton in talking about creoles, notwithstanding the fact that the LBH has systematically been proven wrong (Roberts 1998, 1999, 2000; Siegel 2007). More recently, McWhorter (1998, 2001, 2005) offers a new paradigm, the ‘Creole Proptotype’: based on the observation of shared typological features, McWhorter describes creoles as ‘simple grammars’, i.e. devoid of aspects of complexity that allegedly only emerge over time. The popularity of such models is reflected in the fact that a number of papers in this volume can be seen as critical and/ or alternative views on McWhorter’s latest paradigm and ideas associated with it, such as complexity and simplicity. The central aspect of our ‘deconstruction’ is therefore to investigate how much of the creole paradigm is what Popper describes as cultural-historical conjecture, and how much of it is actually empirically sound. As Woody Allen’s character Harry puts it in the movie Deconstructing Harry, “between the Pope and air-conditioning, I take airconditioning”. So do we.



Deconstructing creole: The rationale

2 Deconstructing creole 2.1

Creole studies and linguistics

This volume consists of a collection of articles offering novel and critical perspectives of the fundamental ideas that have defined and supported the notion(s) of creole uniqueness to this date. It is our hope that, in providing a series of challenges to the ideologies and theoretical ideas around which notions of ‘creole’ are constructed, this volume will offer a comprehensive assessment of the state of the art. The ultimate goal is to overcome the artificial dichotomy between creole and non-creole languages, in order to integrate the study of creolization phenomena into mainstream linguistics, i.e. the study of language variation and language creation. The field of contact linguistics and creolistics has in recent years developed from a rather small, closed circle into an area of interest for linguists in general and, beyond that, sociologists, historians and anthropologists (Palmié 2006). A milestone in these developments was Thomason and Kaufman (1988), who opened up the field to historical linguistics in particular and treated creoles and contact varieties on a continuum rather than as separate linguistic objects. The follow-up was however slow (though there have been repeated calls to that effect, cf. e.g. Mufwene 1990), and it was only in 2001, with a special issue of Linguistic Typology dedicated to the creole debate, that the linguistic community at large took a serious interest in the integration of studies of creolization and contact into general linguistic theory. From a typological perspective, once ‘creoles’ and contact languages become a unified object of study on a par with other subfields of linguistics, they lose the special status that had been awarded them a priori by traditional creolists. The fundamental idea behind this is that it is not creoles as such that constitute a unique type of language warranting a separate field of enquiry, but rather the way in which at least some creolists have approached the study of language that is unique. However, in combining careful sociocultural observation, ideological reflection, and synchronic as well as diachronic structural analysis, creolists have much to contribute to general linguistics, typically still locked within traditional disciplinary boundaries that limit the scope and impact of empirical as well as theoretical observations. Creolists tend to be unlike linguists working in other fields, whether defined in terms of language areas, families or topics, who typically do not assume any uniqueness to these languages. Rather, non-creolists typically work on the assumption that their work will contribute somehow to the field of linguistics as a whole (e.g. sign linguistics). In the field of pidgin and creole languages, there seems to be a belief that the objects of study are unique (as argued by McWhorter, e.g. 2001) and/ or provide a privileged perspective on language (as argued by Bickerton 1981, 1984). To admit that no such uniqueness actually holds would be seen as failure, as much of the discussion posted on the CreoLIST in the early 2000s illustrates (Morrissey 2002). This has re1.

A rare example of this is DeGraff (2001a).





Umberto Ansaldo & Stephen Matthews

sulted in pidgin and creole linguistics remaining in the isolation which Bickerton denounced as early as 1976 (see also Jourdan 1991). For example, a recent book on grammars in contact (Aikhenvald & Dixon 2006: 11) explicitly excludes pidgin and creole languages from the scope of its enquiry as a ‘special’ case. The aim of this volume is to present a series of studies that depart from and move beyond ‘creole exceptionalism’ (DeGraff 2001b, 2005), i.e. the idea that creoles are special or abnormal linguistic creations. There are at least three parallel lines of enquiry which, together, have supported exceptional accounts of the creation of new grammars and as such form the ‘creole paradigm’: 1. Creole grammars are structurally exceptional; 2. Creoles develop in an exceptional acquisitional environment; 3. Creoles are different from languages that have been created long ago. Each of these points can be refuted by careful linguistic and sociohistorical analysis of the formation of creole languages (e.g. Arends 2001; DeGraff 2001b, 2003, 2005). Once this is achieved, one can dissolve the notion of ‘creole’ as a particular type of language and accept the fact that one is looking at products of high-contact environments in specific sociohistorical settings. This answers calls such as that of Muysken (1988) for clear theoretical grounding of otherwise subjective notions such as simplicity and complexity, and of Mufwene (1990) for the reintegration of creole studies into mainstream linguistic theory, as these languages have much to contribute to general and historical linguistic studies, in particular regarding the role of contact in language change and the significance of social factors for the nature of language transmission. To date, the impact that creole languages have had on current linguistic theory falls far short of their potential in this regard, largely due to the ‘exceptional’ status that has been traditionally assigned to them.

2.2

Introducing the volume

The study of ‘creole’ languages offers many interesting dimensions of enquiry, ranging from sociohistorical to structural as well as anthropological domains. What constitutes perhaps the most exciting and challenging aspect of the field, from our point of view, is the possibility of reflecting on the creation of a new grammar and its diffusion in a population. Because of their relatively recent formation, in studying creoles, we can in a sense ‘observe’ language genesis as it happens. It is in this sense that the lessons 2. In introducing the chapters in this section, the various contributions are not presented according to the order in which they actually appear in the volume; the Table of Contents is realized to provide clarity in an overview of contributions, as well as orientation for less-initiated readers, students, etc. In the spirit of this volume, however, here we discuss the chapters not in linear order but as part of a narrative. In approaching this volume then, it is up to the individual reader to pick and choose among the chapters either according to the classification provided by the TOC or along the line of reasoning that follows.



Deconstructing creole: The rationale

learned from the study of creole formation may have a significant impact on the field of general linguistics. But in order to do so they have to shed the shell of exceptionalism they have so far inhabited, and enter a wider world of enquiry. The studies in this volume attempt to do just that: they critically revisit problematic notions or false dichotomies of the exceptionalist paradigm (see sections 3.1 and 3.3), and they offer new and broader areas of linguistic enquiry within which creole studies can be situated, in particular linguistic typology and sociolinguistic historiography. A central issue in the creation of new languages has revolved around the problematic notion of ‘special’ transmission attributed to creole genesis (Siegel this volume). The nature of transmission has often been described as ‘imperfect’ or ‘broken’ in the development of contact languages such as creoles, mixed languages, etc. However, within multilingual speech communities where no standard varieties are imposed on speakers, we find acquisitional routines very much like the ones posited for creoles, once again obviating the need for invoking ‘exceptional’ creole development scenarios. Different interpretations of transmission and transfer and their ideological underpinnings are the topic discussed by Siegel. He shows how differences between different views of transmission can be seen as involving matters of degree rather than irreconcilable positions; crucially, however, the different interpretations carry with them major ideological consequences in postcolonial societies of the New World. This chapter shows how problematic notions such as that of ‘imperfect’ transmission need to be reevaluated against current second language acquisition theories as well as current ideologies of ‘creolization’; in this sense, Siegel suggests, it may be worth shifting our focus to the role of creativity and agency in new language formation. In this connection, it can be shown how an unusually high degree of contact is present in the formation of numerous varieties across the globe that do not require any exceptional account of their genesis; this is the focus of the study by Ansaldo, Lim and Mufwene, who show how new languages can emerge in ecologies where no break in transmission, no social violence nor poverty of input are present. In the case of the Peranakan communities of Southeast Asia, a high degree of multilingualism combined with informal acquisition as well as the emergence of a new ethnic community provided the right environment for the evolution of a new language characterized by Sinitic-Malay admixture. Applying a view of acquisition that takes targeted, guided and normative instruction as ‘normal’ is historically untenable as these concepts only arise in Western European nation states and are extremely recent historical constructs. Casual, non-normative and often linguistically heterogenous input in acquisition is the norm and this is very much the type of scenario that applies to creole ecologies. As comprehensively shown by DeGraff (e.g. 2005), arguing that creole speakers are failed target learners is ideologically grounded as well as structurally biased (also section 3.3 below). This bias is the focus of deconstruction in the contribution by Farquharson, who analyzes the historical linguistic currents that lie beneath the perception of creole morphology as morphologically simple. He shows how more accurate morphological analysis of the Jamaican Creole word-formation process, semantic transparency and





Umberto Ansaldo & Stephen Matthews

inflectional morphology may reveal the presence of morphologically ‘complex’ features that early creolists were unaware of. Farquharson identifies a certain reluctance on the part of some creolists towards a serious engagement with data, and calls for more theoretically grounded discussions of morphology, time-stability and age of languages. It is often the case that careful structural analysis as well as rigorous application of theories of language change reveal that emergency strategies or innate programmes (à la Bickerton 1981 or McWhorter 1998, 2001) do not need to be invoked to account for creole grammar. This also resolves the much invoked, but scantily supported, pidginto-creole evolutionary scenario, as pidgins are no longer required as the sine qua non for the emergence of creole languages (see Roberts 1998, 1999, 2000; Siegel this volume, to appear). The role of pidgin in language creation is treated by den Besten, who argues that – in those rare cases where a pidgin is indeed historically attested in the prehistory of a creole – the development from one to the other is not necessarily linear. In particular, he shows that Cape Dutch Pidgin can only be taken as one of the contributing languages, alongside other codes available in the contact environment, to the development of Khoekhoe Afrikaans, and that other languages, such as Dutch, also contributed to the development of the new variety. Moreover, den Besten also revisits different takes on the role of relexification and evaluates their explanatory potential by exposing their limitations in accounting for diachronic processes. As we suggested above, in accounting for the creation of new grammars, one can show that new grammatical features that emerge in creole and contact languages are derived from a combination of substrate and superstrate features (as in Mufwene’s 2001 feature pool), as well as general patterns of language change (such as those described in various accounts of grammaticalization, for example). This approach is taken by Aboh and Ansaldo, in an evolutionary, typological take on language creation. The authors argue against a generic metric of complexity by showing that serious typological investigation allows us to understand the input-output relation in language creation. Comparing the evolution of Surinamese creoles and their input languages, they show how a process of feature competition and selection accounts for the structural output. Likewise, based on a typological analysis of Sri Lanka Malay, they show that ‘complex’ morphology can indeed emerge in radical contact environments, as long as the feature pool supports it. Complexity and simplicity are indeed extremely difficult notions to discuss and may be more fruitfully investigated within specific subdomains of grammar. In his chapter, Gil shifts our current understanding of complexity to the field of associational semantics, in order to provide an empirically-based test for the assumed relationship between complexity and age, as understood within the Creole Prototype (McWhorter 1998). Gil shows how simplicity in this sense can be found to higher degrees in some non-creole languages, such as Sundanese and Minangkabau, undermining the idea that creoles – assuming they share common structural properties – may be treated as generally ‘simpler’ than other languages. This also leads to the conclusion that what appear to be processes of ‘simplification’ may simply be random outcomes of language histories, not necessarily related to external factors such as adult second language acquisition.



Deconstructing creole: The rationale

Grant investigates the relationship between language contact and typology further, showing how the former influences the latter in the evolution of the Chamic language family. He demonstrates that the observed process of apparent simplification can be ascribed to the role of a lingua franca in the history of a language family, without invoking a specific process of pidginization. While Gil expresses a healthy scepticism regarding the reconstruction of external ecologies in absence of appropriate linguistic and historical data, Grant shows that where diachronic data is available, typology goes a long way toward explaining the outcome of processes of admixture. It seems clear that ‘complexity’ of structure and the ‘age’ of languages are parameters that can only be understood in a relative sense within strict linguistic models (Musyken 2001), rather than as universal indicators of language type (Ansaldo & Nordhoff forthcoming). In diachronic terms, complexity does not apply cross-linguistically: The problem here becomes immediately evident when one considers that reduction and expansion or complexification refer to logical sequences on a classificatory or taxonomic plane that cannot possibly be directly diachronized let alone mapped onto (empirically insufficiently known) historical processes. (Palmié 2006: 445)

In typological terms, many ‘old’ languages can be seen as less complex than creole languages. A more productive idea of complexity can be found outside the structural domain; with this in mind, Faraclas, Walicek, Alleyne, Geigel and Ortiz show what type of complexity really matters in understanding language creation and diffusion, in an in-depth investigation of the correlation between creole languages and political-economic systems across the Caribbean. The authors convincingly argue for the fact that understanding the political economy of a community can effectively explain the type of ‘­creole’ that emerges in such an environment, showing how differences in politicaleconomic systems are reflected in the differential manifestations of African substratal features in Caribbean creole varieties. By doing so, they provide us with a realistic, historically-supported and culturally-nuanced explanation of the distribution of, among others, Spanish creoles (cf. McWhorter 2000). All in all, these chapters end up relativizing the notions of complexity and simplicity by showing that these are usually somewhat biased notions deriving from subjective or theoretically-specific perspectives, and should be limited to working definitions for experimental purposes, within specific sub-domains of linguistic structure. The final chapter can be seen as a sort of epilogue that opens up a whole new direction for future investigation. Harris and Rampton revisit the assumptions behind creole ‘metaphors’, discussing the ideological foundations of a range of ‘creole beliefs’. At the same time, they offer new analytical insights and new empirical domains within which the notion of ‘creole’ may be significant, such as the study of interaction in trans­ national contexts within sociolinguistic micro-ethnogaphy. As such, they end this vol-





Umberto Ansaldo & Stephen Matthews

ume on a positive, constructive note, by showing how methodologically sound analysis of rich ‘creole’ data, freed from speculative ‘must haves’ of past creolists’ assumptions, may pioneer future areas of linguistic enquiry within which creole varieties play a fundamental role.

3 History of beliefs 3.1

A brief history of creole ideas

The field of Creole Studies is rightly associated with Western colonial expansion and the resulting slave trade, and covers the linguistic history of predominantly the Atlantic and Pacific between the 16th and 20th centuries. During the explorations of the 17th century, we find many accounts of contact varieties reported by travellers, missionaries etc.: French mixed with English, Spanish and Dutch in Martinique; MalayoPortuguese in the East Indies; Delaware Jargon in North America (Holm 1988). In these early records, the languages tend to be described as autonomous systems independent of any lexifier. This continues throughout the 18th century with much work done by Moravian missionaries, especially on Caribbean varieties. At this point, languages such as Greek and Latin were still considered exemplary grammars against which to measure other varieties, which were in comparison usually found deficient (Holm 1988; Mühlhäusler 1995). The discovery of Sanskrit, which was to launch the comparative enterprise, would exacerbate this trend. It is in the 19th century that we find the first missionary publications in Creole languages; interestingly, these were criticized for using a corrupt medium in press. At the same time, there were voices in defense of the autonomy of creole linguistic systems (e.g. Sranan in Greenfield 1830), which stressed that the origins of creoles were similar to those of other languages. The Romantic movement turned the attention of linguists to the people and their dialects, and this reinforced the interest in creoles (French-based varieties in particular). As discussed in Meijer and Muysken (1977), ideologies of creole languages from the 19th century are closely related to racial assumptions about Black Africans. For example, Bertrand-Bocandé (1849) implies that a civilization and its language are equally complex, and argues that, to Africans, the intricate morphological properties of European languages would be too complicated and would need therefore to be simplified in acquiring them. Van Name, a librarian at Yale, in Contributions to Creole Grammar (1870), put forward the first ‘scientific’ study of Creoles, which focused on substrate languages and the common structural properties of Creoles. Van Name suggested the relationship between a pidgin stage and the creole, as well as the idea of creolization as rapid language change. He also recognized the innovative character of 3. It is important to note how the first creolists were not trained scientists; see section 2.2 and DeGraff (2005).



Deconstructing creole: The rationale

creole languages, rather than merely focusing on structural decay. Along similar lines Adam (1883), a French magistrate in Guinea, traced the first relationships between West African languages and French Creoles, arguing that creoles are simply non-European languages with European lexical influences. Furthermore, Adam suggested that a similar analysis could be applied to Romance, which he saw as mixtures of Latin and substratal vernaculars. On the other hand, the notion of universal tendencies underlying creole structure arose in Coelho (1880–86), of the Geographical Society of Lisbon. In his view, universal strategies of second language acquisition play a role in emergency situations of contact (an early view of creolization along the lines that Bickerton would later develop in his LBH). Schuchardt, one of the fathers of creolistics, took over Coelho’s work and was instrumental in ‘elevating’ creole studies to an academic discipline. According to Meijer and Muysken (1977), with Schuchardt, we see the establishment of creole studies as separate from the interests of creole communities. With him, the ‘problematic’ genetic position of creoles became obvious, i.e. the fact that, due to their mixed grammars, they could not be clearly identified along a genetic lineage, as all other languages. Important notions developed by Schuchardt include: (i) pidginization as a process involving a simplified contact language that develops to facilitate communication under the influence of external factors; (ii) pidgin to creole development, in order for the slaves to have one medium of inter-linguistic communication; (iii) decreolization and variation with the creole. Additionally, Schuchardt had a keen interest in substratal analysis, not only regarding the role of African languages in Caribbean Creoles but also for example in respect to the role of Malay in Asian creoles (Meijer & Muysken 1977). It was Reinecke, an American settled in Hawai‘i (1937), who launched the modern field of Creole Studies, engaging in large-scale classification and documentation of restructured languages and describing over 40 different varieties. He focused on sociolinguistic patterns of creolization, introducing distinctions between different types of Creole societies (e.g. settlers vs. plantation vs. maroon) and the internal relation to the substrate varieties as influential for the restructuring process. Hall (1940) also compared creoles across different regions and linguistic groupings and pioneered the idea of spontaneous pidgin creation as well as the pidgin-creole life-cycle. It is difficult not to see the relationship between the ideas of the time and the theories developed by the (fore)fathers of creolistics. As noted in Palmié (2006: 444), the concerted efforts of people like Van Name, Schuchardt and Coelho did eventually contribute towards “classifying the languages of the subaltern colonial populations and isolating them (from the rest of supposedly ‘normal’ languages) as a theoretically salient anomaly”. The (racially-based) idea of Black Africans as ‘simpler’ is extended to their languages, based on the equation between language, culture and race. From then on, to this day, creole languages have been regarded as somewhat deficient – from not afford4. End 19th to early 20th century. 5.

And, simplifying, old = complex > young = simple (cf. Schleicher in DeGraff 2001b: 219).





Umberto Ansaldo & Stephen Matthews

ing the sophistication of European varieties, to being simplified, unmarked or ‘pure’ instantiations of UG. Most recently these assumptions can be seen in the assumption that creoles lack a certain type of complexity, typically identified mainly with morphology, as this was (and is) a central domain of enquiry in the comparative and historical study of Indo-European languages (a similar prejudice can be found in accounts of non-IndoEuropean languages, cf. DeGraff 2003: 393; Farquharson this volume). Note that it is from the study of Indo-European languages that some of the more robust linguistic theories to this date are derived. The violent and inhuman conditions associated to slavery justify the notion of ‘emergency’, which is eventually associated with the acquisition strategies of slaves, i.e. ‘normal’ acquisition is denied to them. This corroborates the necessity of simplification. Finally, the Romantic movement and Rosseau’s ideas of naturalism lend philosophical material to the idea of the creole universal (Meijer & Muysken 1977). What we are left with are the pillars of the ‘creole paradigm’, namely (1) the idea of simplicity, (2) the idea of broken or special transmission, (3) the idea of universal strategies peculiar to creolization. Notwithstanding the clear ideological bias of these notions, and the overt criticism of them in recent literature (see next section), creolists’ research agendas are still predominantly preoccupied with (1) – (3).

3.2

From the Language Bioprogram to the Creole Prototype

Controversial proposals such as Bickerton’s LBH and McWhorter’s Creole Prototype (CP) are often considered to have been particularly thought-provoking in the history of creole studies. Bickerton’s LBH was based around four assumptions regarding the creation of Hawai‘i Creole English (HCE): (i) rapid formation, (ii) impoverished input, (iii) shared features of creole languages, and (iv) lack of substrate input. As already mentioned above, much evidence against these assumptions, both in the formation of HCE as well as in the linguistic histories of other creole languages, has been put forward (see Jourdan 1993 for a brief account; Mather 2006 for a recent take). In particular, work by Roberts (1998, 1999, 2000) and by Siegel (2000, to appear) have shown that (i) and (ii) do not hold, as HCE emerged gradually over a couple of generations and from a pidgin that already contained a number of features found in HCE (i.e. it was not so impoverished). Moreover, (iii) and (iv) have been repeatedly undermined by studies showing that many of the alleged similarities between creole languages are only superficial in nature, and that real similarities can be accounted for by the similar typologies that influenced many creole languages (superstrates and substrates; see Muysken 2001; Aboh & Ansaldo this volume). Note also how, from a UG-based theory of language, it is logically possible to conceive of but one type of language creation, 6. As noted in DeGraff (2003), a few exceptions apply among creolists, such as Greenfield (1830), Muysken (1988), Mufwene (2001), who argue for a uniformitarian approach to genesis. 7. We, among others (e.g. Siegel to appear), find it surprising that the LBH is still given so much air in creole publications (e.g. Mather 2006; Singler 2006).



Deconstructing creole: The rationale

equal for all human beings. It follows that the concept of creolization as a special type of UG-manifestation is a contradiction in terms (see DeGraff 2001a, b, Rizzi 2001). In an attempt to put forward a different type of universalist genesis of all creole languages, McWhorter (1998, 2005) identifies three traits unique to creole languages (or features that, if found, will identify a language as a creole), which comprise the Creole Prototype (CP), namely: (1) minimal use of inflection, (2) lack of tone used to contrast monosyllables or make grammatical distinctions, and (3) semantically regular derivation. This conjunction of features can only be found in young languages, according to McWhorter, as older languages tend to accumulate signs of old age and depart from the CP. In McWhorter’s sense, creole grammars are therefore simple. Tones, inflection and derivational non-compositionality are by-products of language change but are not necessary for basic communication as they are not inherent to UG (McWhorter 2005: 10). Therefore, they only grow over time and are an indication of emergence of complexity. There are at least two problems in the CP. One, as pointed out in Dahl (2004), is the fact that the hypothesis is not falsifiable. In discussing pidgins and creoles, he notes that: …in order to be defined as a creole, a language must have as its primary historical source a language which has a sufficiently simplified grammatical structure [a pidgin]. No grammatical property of a language can therefore be a counterexample to the thesis that creoles have the world’s simplest grammars, because in order to be a creole, the language has to originate from an earlier language state which did not have that property (a pidgin), and if it has it there are only two logical possibilities: either that stage did not exist, in which case it is not a creole, or the property has been acquired later, in which case it is not a counterexample either, since it just means that the language is on its way to losing its creole character. (2004: 111)

The other problem comes from typological-evolutionary approaches to language creation (Croft 2000; Mufwene 2001). In these approaches, language change, and therefore language creation, is approached as a matter of competition and selection of the features from different languages that are seen as competing in a language contact situation. As suggested in Ansaldo (2005, to appear), and argued in Ansaldo and Aboh (this volume) and Ansaldo and Nordhoff (forthcoming), the input-output relation can usefully be seen within the feature pool-based framework where, assuming we have a proper understanding of the typological dynamics in competition, we can predict what types of features will emerge from a given contact ecology. Ansaldo and Nordhoff (forthcoming), based on the analysis of ‘complexity’ in Sri Lanka Malay, a young language by McWhorter’s standards, note that McWhorter (2005: 317) has a particular explanation for young languages with signs of old age: “Creoles with a moderate degree of inflection…have long existed in intimate contact with inflected superstratal or adstratal languages”. Based on this intuition, they suggest that we can generalize as follows:





Umberto Ansaldo & Stephen Matthews

Creoles with clearly isolating morphology have long existed in intimate contact with isolating superstratal or adstratal languages Such a realization suggests that there is no need to call upon UG-features, especially in the light of the lack of clarity and agreement over which properties really constitute UG, and undermines the necessity of the notion of CP in accounting for creole formation. As already noted in Muysken (2001), as well as in general in substratist approaches (e.g. Alleyne 1971; Lefebvre 1986; Muysken & Smith 1986; Arends, Kouwenberg & Smith 1995), the study of West African languages, for example, can help account for a significant number of common structural features of creoles.

3.3

Creole myths

In a series of papers, DeGraff (2001b, 2002, 2003, 2005) argues against the set of ideas sketched above which portray creole languages as products of exceptional development, i.e. the set of ‘creole myths’. DeGraff ’s work is a theoretical deconstruction of the creole paradigm in the strict sense, and provides an enlightening critique of the problematic aspects of the field. As stated in the opening of this chapter, our goal in the remainder of the volume is to take creole deconstruction as a point of departure and constructively move on towards the integration of creole studies into larger theoretical and empirical paradigms. Therefore, in order for our foundations to be solid, we here revisit the deconstruction already advocated within the field. 3.3.1 The myth of simplicity According to DeGraff, the notion that creoles are structurally impoverished variants, or degenerate offshoots, of their European norms is an explicit case of minoration linguistique resulting from the historical and ideological assumptions of the 17th – 19th centuries already presented above. In creole-related writings of the colonial area, the ‘programmes of perception’ were congruent with race theories that provided philosophical justification for the New World Slavery. The creole theories were ordered by Europe’s ‘normative gaze’ vis-à-vis the non-White and non-Christian world (in 1894, Poyen-Bellisle in a University of Chicago dissertation describes how slaves try to imitate [civilized man’s speech] but fail because they do not belong to the same race, their vocal tracts and their lips are different). Moreover, in the age of discovery, race- and evolution-related myths can be seen as follow-ups to these early ideas, that result in the congruence between the evolution of races and that of languages typical of 19th and 20th century historical linguistics (in 1924, Meillet describes Spanish Creole or French Creole as varieties of Spanish and French deprived of almost all their grammar, weakened in their pronunciation and re8. Summarizing succinctly DeGraff ’s positions is by no means an easy task; the interested reader can find a thorough exposition of his views in particular in DeGraff (2005).



Deconstructing creole: The rationale

duced to a small lexicon; our emphases). The power of the culture = race = language equation is still alive today, as creole genesis falls outside the scope of the Comparative Method, given that creoles are regarded as non-genetic ‘orphans’ outside the family tree of human languages (see also Mufwene 2001). What is worse, in DeGraff ’s view, is that such myths are still alive today.10 3.3.2 The myth of decreolization The prestige that early creolists attribute to the lexifier can be further seen in the notion of ‘decreolization’. The basic assumption here is that whenever Creole languages are spoken alongside their lexifier, Creole languages, being impoverished, would naturally tend to assimilate ‘back’ to it (and would otherwise constitute a threat to the purity of the lexifier). In this respect DeGraff notes how, while all languages undergo change,11 only within creolistics do we find concepts applying the de- prefix (e.g. ‘deLatinization’, ‘de-Africanization’ do not apply). If language change is assumed to follow a number of general tendencies (or universal diachronic patterns), then we should account for these changes as moving towards some same direction, i.e. all language change is creation or birth (see Mufwene 2004). The myth of decreolization is also linked to the idea of abrupt or broken transmission criticized above (section 2.1) and to the notion of simplicity or simplification (addressed in 2.1 and 3.3). 3.3.3 The myth of exceptional diachrony If we assume that language is a basic human faculty that develops along the same lines for all human beings, it is surprising to note how so many different exceptional theories have to be put forward to account for creole genesis. DeGraff sees the idea of strict relexification (Lefebvre 1998) as supporting the notion that creole creators follow different interlanguage strategies than other humans (in particular, they cannot ‘escape’ their native grammars, something untenable within a uniformitarian interpretation of acquisition). On similar grounds, Bickerton’s early idea of pidgin as a ‘linguistic fossil’ and associated notions of rapidity of change are untenable as creole grammars can be shown to fall within developmental patterns typical of ‘regular’ language change and there is evidence to suggest that creoles did not develop more rapidly than other languages (see also Mufwene 1996; Ansaldo to appear). A similar critique of the idea of pidginization as implying specific diachronic patterns common to only one type of 9. These views all share the common assumption that the ultimate goal of the slave would be to fully master the lexifier (target language); since the outcome of the creole ecology was not found to be an identical replica of the lexifier, the hypothesis had to be put forward to account for such ‘failure’. 10. For example, a recent news item described Haitian Creole as “Haiti’s language of broken French” (Gardner 2004). 11. Obviously this includes types of change whereby a creole variety changes from basilectal to acrolectal features.





Umberto Ansaldo & Stephen Matthews

people (i.e. the creole speakers) is criticized as dehumanizing and basically neo-Darwinian (in the negative sense of Darwinism), underlying McWhorter’s Creole Prototype (see 3.2 above).

4 Final remarks The only ideologically plausible and empirically grounded difference between creole and non-creole languages arising from such a deconstruction is, according to DeGraff, a sociohistorical one, a conclusion already put forward in Muysken (1988, 2000), den Besten, Muysken and Smith (1994), and Mufwene (1996, 1998, 2000). Individual speakers in contact ecologies, whether in Creole, Romance or Germanic languages, would have made use of the same mental process in the formation of their respective new languages. Therefore, in the linguistic domain, exceptionalist scenarios that have led to the construction of the creole paradigm can only be accounted for as ideological constructs that viewed speakers of creole languages as having failed in one way or another, in respect to language evolution, language acquisition and language creation. Considering the historical, theoretical and critical reflections presented above, we believe it is fair to suggest the following conclusions: a. creole languages do not form a typological or otherwise structurally unique class distinguishable from non-creole grammars; b. there is no clear distinction between ‘normal’ change and creole formation; c. language contact and language creation are ubiquitous; creole grammars are the product of the linguistic ecology and general diachronic (or internal) patterns of language; d. creole exceptionalism is a set of sociohistorically-rooted dogmas, with foundations in (neo-)colonial power relations, not a scientific conclusion based on robust empirical evidence. It is thus in sociohistorical terms that we view creole languages as an object of linguistic enquiry. Along lines already explored in Ansaldo and Matthews (2001), we assume that a certain degree of admixture is present in language change at large (Muysken 2000) and that creoles are particularly heterogeneous admixtures due to their sociolinguistic histories of contact and multilingualism. There is no dichotomy between normal and special transmission here (Thomason 1997) as language is seen, in the light of Mufwene (1998), as a communal construct never entirely possessed by a speaker. Genetic affiliation will be a matter of counting and quantifying features, ultimately a statistical exercise as shown in Thomason and Kaufman (1988) for the evolution of the English language. Nor does speed of change enter the picture, as acquisition of I-language can only proceed at one and the same speed (Rizzi 2001), while diffusion of Elanguages is correlated with size and type of network (and considerations of frequency



Deconstructing creole: The rationale

and ‘markedness’: Croft 2000; Ansaldo to appear; Ansaldo & Nordhoff forthcoming). As we hope to have shown above, and as aptly stated in Palmié (2006: 448):12 … “creolization theory” is ultimately a mere reflex of the very conditions it seeks to denounce and supercede – and so, once properly conceptualized, might itself be more profitably regarded as an object of, rather than a tool for, (anthropological) enquiry.

References Adam, L. 1883. Les Idiomes Négro-Aryen et Maléo-Aryen: Essai d’Hybridologie Linguistique. Paris: Maisonneuve. Aikhenvald, Y. & R. Dixon. 2006. Grammars in Contact: A Cross-Linguistic Typology. Oxford: Oxford University Press. Alleyne, M.1971. Acculturation and the cultural matrix of creolization. In Hymes, D. (ed.). Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. 169–186. Ansaldo, U. 2005. Typological admixture in Sri Lanka Malay: The case of Kirinda Java. ms. University of Amsterdam Ansaldo, U. to appear. Revisiting Sri Lanka Malay: Genesis and classification. In Dwyer, A., D. Harrison & D. Rood (eds). A World of Many Voices: Lessons from Documenting Endangered Languages. [Studies in Language.] Amsterdam/ Philadelphia: John Benjamins. Ansaldo, U. & S.J. Matthews. 2001. Typical creoles and simple languages. The case of Sinitic. Linguistic Typology 5 (2/3). 311–326. Ansaldo, U. & S. Nordhoff. forthcoming. Complexity and the age of languages. In Aboh, E.O. & N. Smith (eds) Complex Processes in New Languages. Amsterdam/ Philadelphia: John Benjamins. Arends, J. 2001. Social stratification and network relations in the formation of Sranan. In Smith, N. & T. Veenstra (eds). Creolization and Contact. Amsterdam/ Philadelphia: John Benjamins. 291–308. Arends, J., S. Kouwenberg & N. Smith 1995. Theories focusing on the non-European input. In Arends, J., P. Muysken & N. Smith (eds). Pidgins and Creoles: An Introduction. Amsterdam/ Philadelphia: John Benjamins. 99–110. Bertrand-Bocandé, E. 1849. Notes sur la Guinée portugaise ou Sénégambie méridonale. Bulletin de la Société de Géographie 12. 57–93. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Bickerton, D. 1984. The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7. 173–203. Chomsky, N. 1986. Knowledge of Language: Its Nature, Origin and Use. New York: Praeger. Coelho, A. 1880–86. Os dialectos românicos ou neolatinos na África, Ásia e Ámérica. Boletim da Sociedade de Geografia de Lisboa 2. 129–196; 3. 451–478; 6. 705–755. [Reprinted in Morais-Barbosa, J. (ed.). 1967. Estudios Linguísticos Crioulos. Lisboa: Academia International de Cultura Portuguesa.] Croft, W. 2000. Explaining Language Change: An Evolutionary Perspective. Longman. 12. Our brackets.





Umberto Ansaldo & Stephen Matthews Dahl, Ö. 2004. The Growth and Maintenance of Linguistic Complexity. Amsterdam/ Philadelphia: John Benjamins. DeGraff, M. (ed.). 2001a. Language Creation and Language Change. Creolization, Diachrony and Development. Cambridge, MA: MIT Press. DeGraff, M. 2001b. On the origins of Creoles: A Cartesian critique of Neo-Darwinian linguistics. Linguistic Typology 5 (2). 213–230. DeGraff, M. 2002. Relexification: A re-evaluation. Anthropological Linguistics 44 (4). 321–414. DeGraff, M. 2003. Against Creole exceptionalism. Language 79 (2). 391– 410. DeGraff, M. 2005. Linguists’ most dangerous myth: The fallacy of Creole Exceptionalism. Language in Society 34, 533–591. den Besten, H., P. Muysken & N. Smith. 1994. Theories focusing on the European input. In Arends, J., P. Muysken & N. Smith (eds). Pidgins and Creoles. An Introduction. Amsterdam/ Philadelphia: John Benjamins. 87–99. Derrida, J. 1981. Positions. [Trans. A. Bass.] Chicago: University of Chicago Press. Derrida, J. 1985. Letter to a Japanese friend. In Bernasconi, R. & D. Wood (eds). Derrida and Differance. Warwick: Parousia Press. 1–8. Feyerabend, P. 1975. Against Method: Outline of an Anarchistic Theory of Knowledge. London: Verso. Gardner, S. 2004. AIDS stalks Haiti’s children. Reuters. 9 April 2004. Gray, J. 2004. Heresies: Against Progress and Other Illusions. London: Granta Books. Greenfield, W. 1830. A Defense of the Surinam Negro-English Version of the New Testament. London: Bagster. Hall, R.A. 1966. Pidgin and Creole Languages. Ithaca: Cornell University Press. Hinzen, W. 2006. Mind Design and Minimal Syntax. Oxford: Oxford University Press. Holm, J. 1988. Pidgins and Creoles, Volume 1: Theory and Structure. Cambridge University Press Hymes, D. (ed.). 1971. Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. Jourdan, C. 1991. Pidgins and creoles: The blurring of categories. Annual Review of Anthropology 20. 187–209. Lefebvre, C. 1986. Relexification in creole genesis revisited: The case of Haitian Creole. In Muysken, P. & N. Smith (eds). Substrata versus Universals in Creole Genesis. Amsterdam/ Phiadelphia: John Benjamins. 279–300. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. McWhorter, J. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74. 788–818. McWhorter, J. 2000. The Missing Spanish Creoles: Recovering the Birth of Plantation Contact Languages. Berkeley/ Los Angeles: University of California Press. McWhorter, J. 2001. The world’s simplest grammars are creole grammars. Linguistic Typology 5 (2/3). 125–166. McWhorter, J. 2005. Defining Creole. Oxford: Oxford University Press. Mather, P-A. 2006. Second language acquisition and creolization: Same (I-) processes, different (E-) results. Journal of Pidgin and Creole Languages 21 (2). 231–274.



Deconstructing creole: The rationale 

Meijer, G. & P. Muysken. 1977. On the beginnings of Pidgin and Creole studies: Schuchardt and Hesseling. In Valdman, A. (ed.). Pidgin and Creole Linguistics. Bloomington: Indiana University Press. 21–45. Meillet, A. 1924. Introduction à la classification des langues. In Meillet, A. & M. Cohen (eds). Les Langues du Mondes. Paris: Champion. In A. Meillet 1951. Linguistique Historique et Linguistique Générale. Vol. 2. Paris: Klincksieck. 53–69. Morrissey, B. 2002. What is a creolist? Journal of Pidgin and Creole Languages 17 (1). 111–115. Mufwene, S. 1990. Creoles and universal grammar. Linguistics 28 (4). 783–807. Mufwene, S.S. 1996. The founder principle in creole genesis. Diachronica 13. 83–134. Mufwene, S.S. 1998. What research on creole genesis can contribute to historical linguistics. In Schmid, M., J. Austin & D. Stein (eds). Historical Linguistics 1997. Amsterdam/ Philadelphia: John Benjamins. 315–338. Mufwene, S.S. 2000. Creolization is a social, not a structural, process. In Neumann-Holzschuh, I. & E.W. Schneider (eds). Degrees of Restructuring in Creole Languages. Amsterdam/ Philadelphia: John Benjamins. 65–84. Mufwene, S.S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mufwene, S.S. 2004. Language birth and death. Annual Review of Anthropology 33. 201–222. Muysken, P. 1988. Are Creoles a special type of language? In Newmeyer, F. (ed.). Linguistics. The Cambridge Survey. Vol. 2. Linguistic Theory: Extensions and Implications. Cambridge: Cambridge University Press. 285–301. Muysken, P. 2000. Bilingual Speech: A Typology of Code-mixing. Cambridge: Cambridge University Press. Muysken, P. 2001. Creolization. In Haspelmath, M., E. König, W. Oesterreicher & W. Raible (eds). Language Typology and Language Universals. An International Handbook. Vol. 2. Berlin/ New York: Walter de Gruyter. 1656–1668. Muysken, P. & N. Smith (eds). 1986. Substrata versus Universals in Creole Genesis. Amsterdam/ Philadelphia: John Benjamins. Mühlhäusler, P. 1995. Pidgins, creoles and linguistic ecologies. In Baker, P. (ed.). From Contact to Creole and Beyond. London: University of Westminster Press. 235–250. Palmié, S. 2006. Creolization and its discontents. Annual Review of Anthropology 35. 453–456. Popper, K. 1959. The Logic of Scientific Discovery. [Die Logik der Forschung 1934.] Routledge. Poyen-Bellisle, R. 1894. Le Sons et les Formes du Creoles des Antilles. Ph.D. dissertation, University of Chicago. Reinecke, J. 1937. Marginal Languages: A Sociological Survey of the Creole Languages and Trade Jargons. Ph.D. dissertation, Yale University. Rizzi, L. 2001. Broadening the empirical basis of universal grammar models: A commentary. In DeGraff, M. (ed.). Language Creation and Language Change. Creolization, Diachrony and Development. Cambridge, MA: MIT Press. 453–472. Roberts, S.J. 1998. The role of diffusion in the genesis of Hawaiian creole. Language 74. 1–39. Roberts, S.J. 1999. The TMA system of Hawaiian Creole and diffusion. In Rickford, J. & S. Romaine (eds). Creole Genesis, Attitudes and Discourse: Studies Celebrating Charlene J. Sato. Amsterdam/ Philadelphia: John Benjamins. 45–70. Roberts, S.J. 2000. Nativization and genesis of Hawaiian Creole. In McWhorter, J. (ed.). Language Change and Language Contact in Pidgins and Creoles. Amsterdam/ Philadelphia: John Benjamins. 257–300. Siegel, J. 2000. Substrate influence in Hawai‘i Creole English. Language in Society 29. 197–236.



Umberto Ansaldo & Stephen Matthews Siegel, J. 2007. Recent evidence against the Language Bioprogram Hypothesis: The pivotal case of Hawai‘i Creole. Studies in Language 31 (1). Singler, J.V. 2006. Yes, but not in the Caribbean. Journal of Pidgin and Creole Languages 21 (2). 337–358. Thomason, S. 1997. Contact Languages: A Wider Perspective. Amsterdam/ Philadelphia: John Benjamins. Thomason, S. & T. Kaufman 1988. Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Van Name, A. 1870. Contributions to creole grammar. Transactions of the American Philological Association 1. 123–167.

Part 1

Typology and grammar

Creole morphology revisited* Joseph T. Farquharson University of the West Indies, Mona & Max Planck Institute for Evolutionary Anthropology, Leipzig

1

Introduction

The American linguist Dwight Whitney states that “[i]f we desire to understand the forces which are at work in language, we must be willing to examine their operations in petty and prosaic detail” (Whitney 1871: 38). Surprisingly (or not), those who have based grand theories on the simplicity of creole morphology have not undertaken thorough descriptive studies to see what morphological processes are actually there. It appears that some researchers over the years have been content with merely repeating the slogan that Creoles have “little or no morphology” without really testing it for themselves; a situation which has led them to concentrate more on what Creoles lack than on what they actually possess. However, with the work of the past decade or so, e.g. Brousseau, Filipovich and Lefebvre (1989), Lefebvre (2002), Kouwenberg and LaCharité (2001a, 2001b, 2004), Plag (2003), Bhatt and Plag (2006), we are coming to realize that the frequently repeated phrase that Creoles have “little or no morphology” is seriously flawed. Only continued neglect of the data can lead one to say that they have none, and whether they have only a little can only be seen in a comparative light. Then we will be faced with the problem of deciding which language should be used as the yardstick. The studies which are now appearing on these languages suggest that the claim must be seen as signaling not a lack on the part of languages designated as Creoles, but the failure of generations of researchers (driven by popular, non-empirical wisdom) to devote attention to this particular phenomenon (on this point see Muysken 2004: 1656). We now know that it is not safe to generalize over Creoles since there is now copious counter-evidence to many of the previous claims. For example, Nubi, an Ara* This paper was first presented under the title ‘Simplicity versus complexity: Issues in the study of creole morphology’ at the Seventh Creolistics Workshop held at the Justus-Liebig Universität, Gießen, 6-8 April 2006. I would like to thank the participants at that workshop for their feedback. I would also like to thank the editors and an anonymous reviewer for their very useful comments which have been incorporated as far as possible. Needless to say, all remaining faults are my own.



Joseph T. Farquharson

bic Creole, has retained prefixing, suffixing, and compounding patterns from Arabic (Heine 1982). It is probable that the sense of lack which was felt in the past was owing to the refusal of researchers to explore and engage the data. For example, Bailey (1966) only devotes two pages to morphology in her 163-page book on the grammar of Jamaican Creole. As a case in point, even though reduplication was stereotyped as one of the hallmarks of Creole languages, no comprehensive and systematic descriptions of the process existed for many Creole languages until the appearance of the volume edited by Kouwenberg (2003) which was devoted to the subject. As far as I am aware, only two substantial works (both articles) existed on word-formation in Jamaican Creole prior to the 1990s: Cassidy (1957), DeCamp (1974), and their concentration was on reduplication. In addition, the lack of descriptions of compounding in Pidgins and Creoles is somewhat at variance with the perception that these languages make extensive use of this process as a means of compensating for not having more ‘complex’ patterns. The ramifications of the important explorative and descriptive work on the morphology of Creoles which has been going on especially over the past two decades can be neatly summed up in Gil’s pronouncement that “all known isolating languages still have some morphology – affixation, compounding, or other kinds of processes such as reduplication, stem alternation, and so forth” (2006: 92). This leads us to one very poignant question: If we were mistaken about these things is it possible that we could be mistaken about other things too? The aim of this chapter is not to join the debate about whether Creole languages are complex or not with regards to their morphological module (see Aboh & Ansaldo, Gil, this volume). It can be taken as the skeleton for a kind of Descriptivist’s Manifesto, calling for a change in our thinking about Creoles in terms of what they lack, or what we assume they lack, and a move toward describing how they actually implement particular aspects of Grammar. Some of the issues discussed in this chapter have already been taken up by various authors such as DeGraff (2001), in response to McWhorter (2001). Most of my examples are drawn from the Atlantic Creoles because those are the varieties with which I am most familiar.

2 Word-formation This section is a very brief look at the types of word-formation processes to be found in Creoles (affixation, reduplication, compounding, and zero-derivation). The treatment must not be taken to suggest that all processes appear in all Creoles, but that the process named has been reported in some or at least one Creole language.



2.1

Creole morphology revisited 

Affixation

Of all the areas of Creole morphology, affixation has been the most sorely neglected. All the data now being unearthed continue to demonstrate the blinding effect that ageold biases can have on serious study. There are several Creole languages which have been described as possessing (derivational) affixational morphology. Brousseau and Nikiema (2004) have found 28 productive affixes, 7 semi-productive affixes, and 6 non-productive affixes in St. Lucian French Creole, and my most recent count based on work still in progress puts the number of derivational affixes in Jamaican Creole at around 15. Steinkrüger’s (2003) study of Chabacano reveals 20 derivational affixes: 8 prefixes and 12 suffixes. Haitian Creole has a diminutive prefix -ti (1) which designates “a small noun”, or when the base refers to an animate entity, “a young noun” (Lefebvre 1998: 306) (1)

Haitian Creole (Lefebvre 1998: 306) ti-chat ‘kitten’ chat ‘cat’ ti-mounn ‘child’ mounn ‘person’ ti-dlo ‘pond, brook’ dlo ‘water’ ti-wòch ‘pebble’ wòch ‘stone’

There is also attestation of suffixation. Kouwenberg (1994) reports that Berbice Dutch Creole (BD) makes use of affixational, specifically suffixational morphology. Example (2) shows the BD nominalizer which Kouwenberg describes as being productive over adjectives, possessive pronouns and nouns which occur in attributive position. (2)

Berbice Dutch Creole (Kouwenberg 1994 :233–234) fεtε ‘fat’ fεtεjε ‘fat one’ eʃti ‘first’ eʃtijε ‘the first one’ jεrma ‘woman’ jεrmajε ‘the female type’

Also, Jamaican Creole shows the creation of a new suffix, which is more than likely related to the second element in names such as Kimiesha, Nickiesha, Taneisha. The suffix attaches to nouns, and verbs. If the assumption that this suffix is etymologically from the latter part of these proper (female) names is correct, we would have an interesting case of the native development of an affix. Note that it attaches to more verbs than nouns, and that for most of the derived forms, the feature associated with the suffix has shifted from feminine to person. (3)

Jamaican Creole (Farquharson 2006) DERIVATIVE BASE begiisha beg ‘to beg’ doniisha don ‘to (be) finish(ed)’ laafiisha laaf ‘to laugh’ maniisha man ‘man’

GLOSS ‘a woman who is always begging’ ‘an over-sexed woman’ ‘a person who laughs a lot’ ‘an easy woman’



Joseph T. Farquharson



nyamiisha taakiisha tekiisha singiisha skieriisha

nyam ‘to eat’ taak ‘to talk’ tek ‘to take’ sing ‘to sing’ skier ‘to scare’

‘a glutton’ ‘ a person who talks too much’ ‘a person who likes to take/ receive’ ‘a person who likes to sing a lot’ ‘a very ugly person’

This last feature also renders the affix etymologically opaque since the feminine meaning which it inherited from its putative source form only appears in three of the nine forms presented here. Cross-linguistically, it is more common for morphemes designating male to get lexicalized into person markers than for those designating female to do so. Another case of the development of an affixal person marker is attested for Sranan. Van den Berg (2003) presents a convincing analysis based on good documentary evidence to show that at least as early as the 1760s Sranan possessed a gender-neutral suffixal person marker -man. The early appearance of this affixational process has implications for discussions on how long it takes for natural languages to develop this type of morphology. (4)

18th century Sranan (Van den Berg 2003: 243) gakuman [stutter]V -person ‘stutter’ konkroman [trick]V -person ‘trickster’ koliman [cheat]V -person ‘a cheat’

A language may gain a new affix either by borrowing, as in Berbice Dutch jε, through reanalysis of an old affix(-like form), as in Jamaican Creole -iisha, or by converting free morphemes into bound forms through grammaticalization. The latter, which is generally believed to take a long time to develop is apparently what occurred with Sranan -man. Even though the Sranan case discussed here is a clear case of derivational morphology, it also has implications for the correlation between the ‘age of a language’ and the morphological patterns it exhibits. This issue is taken up in section 5.2.

2.2

Reduplication

This section only deals with the derivational use of reduplication. In section 4.2, I will mention the inflectional use of reduplication in Creoles. It is generally believed that contact languages such as Creoles are more likely to utilize iconic processes (see Seuren 2001: 430), and since reduplication is widely associated with iconicity, then it is assumed that all Creoles will possess this word-formation strategy. For example, Samarin (1971) considers reduplication a universal feature of Pidgins and Creoles. However, Bakker has shown that while reduplication is employed as a productive word-formation strategy in most Creoles, it is “virtually absent in Pidgins” (2003a: 37). If the presence of reduplication in Creoles is due to the ‘communicative pressures’ arising from the contact situation, then we would expect that non-expanded Pidgins would make



Creole morphology revisited 

more use of reduplication than Creoles/ expanded Pidgins. However, Bakker’s study of some thirty non-expanded Pidgins finds it almost absent. Most would agree that the reduplication of nouns to denote plurality is iconic, since the increase in form is indicative of ‘more of the same meaning’. There is evidence to demonstrate that it is not even safe to assume that more iconic patterns will be cross-linguistically more frequent. Dryer (2005), working with a sample of 957 languages, found that only 8 of them coded nominal plurality by means of complete reduplication. In fact, we cannot even make this generalization over languages designated as Creoles since Bollée (2003) reports that Seychelles Creole has reduplication of adverbs, adjectives, and verbs, but no attested cases of reduplicated nouns, and Velupillai (2003) shows that reduplication as a word-formation device is absent from Hawai‘i Creole English. Moving now to another point, recent research also proves that it is not wise to conclude that Creoles will only possess iconic/marked patterns. For example, take the Saramaccan forms in (5) below. A purely iconic reading would have dictated that more of the same form would result in more of the same meaning. But here, reduplication actually results in less of the meaning of the base. (5)

2.3

Saramaccan Creole (Bakker 2003b: 76) geli ‘yellow’ geli-geli ‘yellow/ yellowish’ guun ‘green’ guuun-guuun ‘green/ greenish’ weti ‘white’ weti-weti ‘white/ whitish’ baafu ‘soup’ baafu-baafu ‘soup/ souplike’ wata ‘water’ wata-wata ‘water/ watery’

Compounding

In his pioneering book on (the lexicon of) Jamaican Creole, Frederic Cassidy points out that “[n]ew words are formed in the folk speech by compounding, derivation, back formation, and the like, much as in Standard English” (1961: 69). Despite this pronouncement, derivation and compounding as word-formation processes are not treated by Cassidy. Also we cannot be certain whether the phrase “much as in Standard English” speaks to the existence of these processes in both languages, or he means that with regard to compounding, derivation, and back formation, Jamaican Creole and English exhibit the same behavior. If his intention was the latter, then he would have been guilty of helping to propagate the claim that if Creoles have any morphology at all, it is fossilized and identical to one or more of their source languages, especially its lexifier. Only recently have we started to see decent descriptions of compounding in Creoles such as (Early) Sranan (Braun 2005) and Haitian Creole (Brousseau 1988). 1. Note however, that there were languages in Dryer’s sample which coded nominal plurality by partial reduplication, but these were treated as languages which employ prefixation. Nevertheless, the prefixing languages (partial reduplication included) only account for 118 out of 957.



Joseph T. Farquharson

An example from Jamaican Creole will serve to demonstrate that compounding in Creoles is a fruitful area of investigation. While English has [V-N]N compounds, it does not appear to be really productive, and only a handful of mostly old compounds referring to people (e.g. pickpocket) and objects (e.g. drawbridge) have been constructed on this pattern (see Lieber 2005: 378). On the contrary, the process appears to be quite productive in Jamaican Creole forming nouns referring to persons (6a) and tangible entities (6b), but also creating action nouns (6c), a property which is lacking in the English case.

(6)

Jamaican Creole a. Persons brok-vaibz chat-mout fala-lain juk-maka lego-biis nyam-daag sok-fingga waak-fut wash-beli

(break-vibes) (chat-mouth) (follow-line) (pierce-horn) (let.go-beast) (eat-dog) (suck-finger) (walk-foot) (wash-belly)

‘a kill-joy’ ‘a gossip’ ‘a stranger’ ‘a cunning fellow’ ‘an unruly person’ ‘pej. epithet for a Chinese’ ‘one who sucks his finger(s)’ ‘a pedestrian’ ‘the last child born’



b. Tangible entities mad-dem (to mad-me) tai-ed (tie-head)

‘a type of pudding’ ‘a scarf which is used to tie the head’





‘a quarrel’ ‘a look of contempt’ ‘a manner of pointing with the lips’ ‘a taste’

2.4

Zero-derivation

c.

Action nominals chru-wod (throw-word) kot-yai (cut-eye) push-mout (push-mouth) ties-mout (taste-mouth)

It is very likely that the multifunctionality that exists in several Creoles may be the result of extensive use of zero-derivation (also called conversion). However, the pervasiveness of this feature is not matched by the number of studies on it. Among the few which exist must be mentioned Braun’s (2005) dissertation on word-formation in Early Sranan which devotes a whole chapter to the subject. In Haitian Creole, zero-derivation derives nouns and adjectives/ participles from verbs. However, there are restrictions on the type of verb which can form the input for this process: “unaccusative verbs (e.g. ale ‘to go’), unergative verbs (e.g. krache ‘to spit’), and the intransitive versions of verbs which show a transitive/ intransitive alternation (e.g. bwòte ‘to move’)” (Lefebvre 2002: 44). For Papiamentu, Kouwenberg and Murray (1994) state that some monosyllabic verbs mostly of Dutch origin, but a few from English, undergo zero-derivation to



Creole morphology revisited 

nouns (7). They explain that this particular pattern of zero-derivation usually creates instruments, instrumental end-products, events, and a remaining set with miscellaneous interpretations (Kouwenberg & Murray 1994: 24). (7)

Papiamentu (Kouwenberg & Murray 1994: 24) bor ‘to drill’ > ‘drill’ klèsh ‘to clash’ > ‘a dispute’ bas ‘to blow’ > ‘balloon’

3 Transparency Before we move on to inflectional morphology, we turn briefly to a slightly different but related matter – semantic transparency in Creole morphology. McWhorter (2001: 156) argues that the tendency for Creoles to employ compounds where their lexifiers use unitary expressions is a reflection of their Pidgin past. However, this argument excludes the substratum languages and how they lexicalize these same concepts. As a case in point, while several of the terms for body-parts in the English-lexified Creoles of the Atlantic are unanalyzable expressions copied from their lexifiers (e.g. Jamaican Creole fut ‘foot’, ed ‘head’), we cannot ignore the presence of compound expressions for some body-parts which are either not lexicalized in the lexifier or are unanalyzable words. Surely, before one claims pidginization or universals of semantic representation, it is important to see how well these compounds match the structure in substrate languages and how frequent it is cross-linguistically for languages to employ these same strategies for realizing the concept. For example, why did Sranan use aifutu (eyefoot) to designate ‘ankle’ rather than some other combination? Along those same lines we can recognize the presence of compounds whose semantics are not that transparent, at least from a European perspective. Jamaican Creole as well as several other Atlantic Creoles (both English- and French-lexified) concatenate an adjectival concept with a body-part in order to designate certain human characteristics and emotions (8). It has been pointed out before that this resembles similar patterns in West African languages such as Àkán and Ìgbo (see Alleyne 1980: 115–116). (8)

Jamaican Creole def-iez (deaf-ear) ‘deafness, a deaf person’ haad-bak (hard-back) ‘mature’ haad-iez (hard-ear) ‘obstinate, obstinacy’ haad-yai (hard-eye) ‘disobedient, disobedience’ jrai-yai (dry-eye) ‘audacious’ lang-beli (long-belly) ‘gluttony, gluttonous’ lang-got (long-gut) ‘gluttony, gluttonous’ red-yai (red-eye) ‘jealous, jealousy’ swiit-mout (sweet-mouth) ‘flattering, flattery’



Joseph T. Farquharson

There is also a view that when morphology does exist in a Creole language it tends to be semantically transparent (see McWhorter 1998: 797; Seuren 2001). A counter-example is presented by what Kouwenberg, LaCharité and Gooden (2003) refer to as “X-like reduplication” in Jamaican Creole (9). There are several factors which militate against viewing this process as transparent. First, the process applies to input that have no more and no less than two syllables. For monosyllabic bases to be eligible they must first undergo phonological alteration via an epenthetic vowel (-i) before they can form the input for the reduplicative process. The synchronic requirement for this binarity is opaque, although Kouwenberg and LaCharité (2004) hazard a diachronic explanation. Second, the process applies to multiple lexical categories: Adjectives (9a), Nouns (9b), and Verbs (9c). Third, it has been observed that Adjectival bases which undergo this type of reduplication have an intensive or attenuated reading. These two readings are diametrically opposed to each other, since one signals the characteristic presence of a feature while the other suggests that it is only there in bits. (9) Jamaican Creole (Kouwenberg et al. 2003: 107) a. big ‘big’ bigi-bigi ‘biggish’ swiit ‘sweet’ swiiti-swiiti ‘having sweet contents’ yala ‘yellow’ yala-yala ‘yellowish, yellow-spotted’ b. buk ‘book’ bwai ‘boy’ huol ‘hole’

buki-buki ‘bookish, liking to read’ bwayi-bwayi ‘boyish’ huoli-huoli ‘perforated, having many holes’

c. juk ‘to pierce’ laaf ‘to laugh’

juki-juki ‘prickly’ laafi-laafi ‘inclined to laughter’

Consider also the reduplicative pattern in Saramaccan which is exemplified in (10). These are reduplicated versions of animal names, which have the same meaning as the base (in those cases where simplex bases do exist). We do not get the normal plural or multiplicative interpretation we would expect. In fact, this pattern seems to be reserved for animals, insects which are normally found in large groups. (10) Saramaccan (Bakker 2003b: 74) mosimosi ‘mouse’ wasiwasi ‘wasp’

4 Inflectional morphology It was once the received view that Creoles do not make use of inflectional morphology. That view is slowly changing into one which says that they make minimal use of the process (cf. McWhorter 1998: 792). While this may be closer to the truth, it should not deter us from approaching each Creole with an open mind, to see whether or not



Creole morphology revisited 

inflectional morphology is present as opposed to assuming that it has none. In addition, while McWhorter (1998) makes it clear that he is dealing with affixational inflectional morphology, in McWhorter (2005) where he addresses inflectional morphology and Universal Grammar, he only mentions inflectional affixes. As far as the study of the Atlantic Creoles are concerned, it is likely that many researchers have been looking for affixes, and by doing so neglect everything else, and they have probably been expecting suffixes for this is the form in which many inflectional categories find exponence in the lexifier languages of these Creoles (see Booij 2004: 361). On this matter, Kihm (2003) provides a very good discussion of many of the presuppositions which have forestalled the study of inflectional morphology in Creole languages. Concurring with my observations, he notes that the belief that Creoles make no use of inflectional morphology is generally a statement about the presence/ absence of “overt bound (i.e. affixal) marking for a given set of features” (2003: 333). However, his concern is in showing that Creoles do have inflectional categories and that they implement these in the morphology which is the only place they can be implemented. A part of his intention in that work is to sever the a priori connection between inflectional categories and affixal morphology which has prejudiced the treatment of Creoles. Using that work as a launching pad, I would like to go further in stating that while not all Creoles have been shown to make use of affixal morphology to realize inflectional categories, some do. We should also recognize the alternative morphological strategies which are used by some Creoles such as clitics and reduplication.

4.1

Affixational inflectional morphology

Steinkrüger (2006) argues that the Tense, Mood, and Aspect (TMA) markers of Chabacano behave morpho-phonologically like prefixes rather than free morphemes. However, in Mauritian Creole the TMA markers are independent morphemes or clitics. This fact reiterates the need to heed the call which asks that we treat each Creole language on its own terms. Kouwenberg (1994) shows that Berbice Dutch Creole employs inflectional morphology to encode perfective and imperfective Aspect (11). (11) Berbice Dutch Creole (Kouwenberg 1994: 256) a. poktεkε an ɔ dektεk like-PF=1SG and 3SG take-PF=1SG ‘(He) liked me, and he took me.’ (i.e. ‘He married me.’)

b. εkε ma dεki ju mwa, titεkε fultε moi do 1SG IRR take 2SG go-IPF time=1SG feel-PF good though ‘I will take you (there) when I feel well (again) though.’

2. Admittedly, McWhorter is free to focus on any topic that he so chooses, but by not dealing with the other morphological methods which some languages use to mark inflection, he is probably adding to the belief that inflectional morphology is synonymous with affixational morphology.



Joseph T. Farquharson

Also, I submit that the marker of Progressive Aspect de in Jamaican Creole should be treated as affixational since it is obligatory for eventive verbs (12a), cannot receive independent stress, and cannot be separated from the verb by other elements. The last of these criteria is illustrated in (12). Separating the Progressive marker from the verb renders the utterance ungrammatical (12c). (12) a. Di piipl-dem *(de) baal. DET people-PL PROG bawl ‘The people are crying.’ b. Di piipl-dem aalwiez de baal. DET people-PL always PROG bawl ‘The people are always crying.’ c. *Di piipl-dem de aalwiez baal. det people-pl prog always bawl



In Jamaican Creole the comparative and superlative affixes, -a and -is respectively, which are derived from English appear to be productive, although there are phonotactic constraints which apply (13). That this is a case of inflection rather than derivation is still up for discussion in the field of morphology (see Bauer 1983: 40). Booij (2004: 365) does point out that inherent inflection such as that encoded by comparatives and superlatives is probably closer to derivation than contextual inflection (e.g. Person, Case), however, seeing that these morphological processes do not change word class and are required by the syntax, then they can be treated as inflection. (13) a. badbase base + -aComparative = bada ‘worse’ base + -isSuperlative = badis ‘worst’

b. nofbase base + -aComparative = nofa ‘more plentiful’ base + -isSuperlative = nofis ‘most plentiful’

4.2

Reduplicative inflectional morphology

Bybee, Perkins and Pagliuca (1994: 169–172) propose that the employment of reduplication for Aspectual purposes commonly exhibits grammaticalization from iterative to intransitive, but the grammaticalization process may follow two different subpaths to the same end point:

(a) iterative > continuative > progressive > imperfective > intransitive;



(b) iterative > frequentative > habitual > imperfective > intransitive.

Parkvall (2003: 21) informs us that reduplication is employed for Aspectual purposes in Jamaican Creole, Miskito Coast Creole, St. Kitts Creole, Sranan, Ndyuka, Saramac-



Creole morphology revisited

can, Krio, Nigeria English Creole, Cameroon English Creole, São Tomé Portuguese Creole, Angolar Portuguese Creole, Annobón Portuguese Creole, Berbice Dutch Creole, and Papiamentu. An example of one such use is given in (14) from Jamaican Creole. (14)

Continuative Dem sing-sing so-tel dem taiyad. 3pl red-sing until 3pl tired ‘They sang [i.e. continued singing] until they were tired.’

5 Complex morphology 5.1

Complex morphology as inflectional (affixational) morphology?

McWhorter (2001: 139) states that Saramaccan is simple because it “has neither inflectional morphology nor free equivalents such as noun class classifiers”. This ignores morphological processes such as reduplication which we have seen in use in several Creoles. There is no a priori reason why adding an affix to a stem is more complex than reduplicating the stem – both may have morphophonemic repercussions.

5.2

Complexity and age

Close to the end of the nineteenth century Otto Jespersen claimed that “simple things are pretty often of quite recent growth” (1993 [1894]: 61). This statement is a fitting precursor of sorts to McWhorter (2001) who believes that Creoles have simple morphology because they have not existed long enough in order to develop more complex structures: Let us assume for these purposes that tens of millennia of drift would leave all grammars existing during that timespan equal in terms of the amount of complexity accreted beyond the bounds of the genetic specification for language. This stipulation predicts, then, that one subset of the world’s natural languages, creoles, would differ from the rest of the world’s natural languages in displaying less of this kind of needless complexity. (2001: 132)

This idea that complex morphology – normally equated with (inflectional) affixational morphology – is the product of age is not novel. Bickerton (1990), who claims that Pidgins and Creoles are reminiscent of human proto-language presents similar arguments on the correlation between morphology and language development: “Protolanguage will seldom if ever have any kind of inflection – any -ings, -’ss, -eds, any numberor person-agreement, and so on” (1990: 126). Also, note Whitney (1871: 46) who proposes that “[e]very formative element, whether prefix or suffix, was once an inde-





Joseph T. Farquharson

pendent vocable, which first entered into composition with another vocable, and then, by a succession of changes of form and of meaning...gradually arrived at its final shape and office”. The chief argument presented by these authors is that inflectional morphology is a product of age in languages. This leads to the implicit conclusion that a language with inflecting morphology is of necessity old. I will provide evidence later to show that this thesis is flawed, but first I will take up the matter of language age. First, any claim that one language is older than another is a direct appeal to the vague concept of E[xternal]-language (see Chomsky 2001). All I[nternal]-languages are equal in age, because each speaker has to construct their own grammar based on experience. The baby learning English today has no access to the grammar of Chaucer; the only raw materials it has are those in its immediate environment which provide the primary linguistic data (PLD) it uses to construct its grammar. This is not to deny that many aspects of grammar remain stable over time and are passed from one generation to the next, rather it stresses the point that in the matter of grammar construction, we cannot see past our PLD providers. This view finds support in Hale (2007: 33) who argues that: It is apparent that language does not ‘change’ in the same sense as, e.g., the physical structure of the universe. In the latter case, we are dealing with the modification, under a variety of sources, of essentially the same substance over long periods of time. There are no discontinuities (though there may be catastrophic events of various types which change gross morphological features in some particularly salient way). By contrast, in the case of language change, we must confront the fact that there is, in a very real sense, a different object (a different grammar) with each new generation.

The confusion exists because of a failure to distinguish between linguistic change– which we assume to be instantaneous – and the diffusion of that change to the rest of the speech community, which may take several centuries. A good illustration comes from Sri Lanka Malay (SLM) where at least one speaker is using what was once a free pronoun as a cliticized agreement marker. None of SLM’s adstrate languages is known to have a similar structure, thus, contact is an implausible explanation. The examples in (15) show one speaker carrying out a linguistic operation which is normally said to take centuries to happen. (15c) demonstrates that there is no attested intermediate stage of double marking which can account for the change. 3. I would like to thank Sebastian Nordhoff for providing these SLM examples which were drawn from his joint presentation with Umberto Ansaldo, ‘Complexity and the age of languages’, presented at the Seventh Creolistics Workshop held at the Justus-Liebig Universität, Gießen, 6-8 April 2006. The anonymous reviewer expressed disquiet with my using one example of change without propagation as empirical evidence. That the speaker made consistent use of the form is an indication that it is not a mistake, and its being an invention by this speaker cannot negate the fact that it is a linguistic change, at least in this person’s grammar. The change has already taken place. As I am arguing here, we should not confuse the change itself with its propagation throughout the community.



Creole morphology revisited 

(15) a. Se piisang ara-makang. 1sg banana prog-eat ‘I am eating a banana.’

b. Piisang s-ara-makang. banana 1sg-prog-eat



c. *Se piisang s-ara-makang. 1sg banana 1sg-prog-eat

The presence of languages such as (Mandarin) Chinese, Yoruba, etc. give us enough proof that old E-languages do not have to possess inflectional (affixational) morphology. However, we still need to answer the question of whether the possession of inflectional morphology implies that a language is old. Gil (2001) shows that Riau Indonesian, while old (from a social perspective) does not make use of inflectional morphology, and Ansaldo and Nordhoff (forthcoming) question the correlation between age and the presence of inflectional morphology. It is obvious that the thesis needs to be reworked. A revised version of the argument might run something like this: while older languages may differ with regard to their (non-)possession of inflectional morphology, younger languages – such as Creoles – have not been around long enough to acquire this type of morphology. We have seen from the discussion above that even this more conservative version runs into problems. Some (but probably not all) Creoles do mark some grammatical relations morphologically. Another counter-argument is provided by Sign Languages (SL). From an E-language perspective, most known SLs are of quite recent growth. To prove a point, if we agree for a moment with those who equate language age with external history, then we would have to say that SLs are young languages. However, SLs exhibit sequential (in other words concatenative) and simultaneous morphology. In simultaneous morphology the base sign is altered not by adding extra material to it, but by changing the direction, rhythm, or path shape of the base sign (Aronoff, Meir & Sandler 2005: 309). Simultaneous morphology tends to be inflectional (prefixation or affixation) and is productive across sign languages. If we look at Romance or Germanic languages, their inflectional morphology is mostly of the concatenative type. SLs provide cases of so-called young languages which possess inflectional morphology and use non-concatenative methods to realize it. SLs possess complex verb agreement systems (affixal), and even home signs which could be considered the Pidgin version of SLs show rudimentary agreement systems (Aronoff et al. 2005: 315). Aronoff et al. (2005: 302) note that: Depending on the particular analysis, a single verb may include five or more morphemes. For example, the American Sign Language verb LOOK-AT may be inflected for subject and object agreement as well as for temporal aspect, and it 4. For a very good discussion of morphology in an SL (American Sign Language), see Frishberg and Gough (2000).



Joseph T. Farquharson

could be accompanied by a grammatical nonmanual (e.g. facial) marker that functions as an adverbial. Such a verb, meaning, for example, ‘he looked at it with relaxation and enjoyment for a long time’, consists of five morphemes.

They point out that the morphology demonstrated here is reminiscent of heavily inflecting languages and that morphological realization of “verb agreement, classifier constructions, and verbal aspects have been found in all well studied sign languages”. Therefore, in this respect, it seems advisable to follow the caution of Whitney who states that “[n]o single trait or class of traits however fundamental may be its importance, can be admitted as a definite criterion by which the character of a language shall be judged, and its rank determined” (1967: 234).

6 Conclusion This chapter rides on a fairly recent wave in Creole studies, of subjecting Creole languages to closer scrutiny to ascertain whether they do or do not possess morphology. The investigations conducted so far reveal that many Creole languages contain more morphology than they were previously credited with. Examples from various Creole languages of word-formation processes (affixation, compounding, reduplication, conversion) have been given as supporting evidence. The same is done for inflectional morphology, in a section which signals the need for a broadening of our perspective in treating not only the traditional pattern (i.e. affixation) but also the alternative morphological methods such as reduplication which are used by numerous languages, including Creoles. Therefore, the somewhat bold pronouncement made by Seuren and Wekker (2001: 424) that “the absence (or extreme poverty) of morphology in Creole languages seems to be a solid datum, and a highly significant one”, appears to be not as solid a datum as previously believed. In the second part of the chapter, we looked at the view forwarded by McWhorter that Creoles do not possess inflectional morphology because inflectional systems take very long to develop and Creoles have not been around long enough to give rise to this phenomenon in their grammars. First the chapter tackled the notion of ‘dating/ aging a language’, arguing that from a language-internal perspective and from the perspective of language learners all languages are equal in age. This conclusion in no way suggests that there are not certain aspects of grammar which are stable over time, but their time-stability is not relevant to the language learner. Second, and finally, the chapter presents results from recent research on sign languages which shows that they have inflectional morphology, a point which puts into question the suggestion that inflectional morphology only exists in ‘old’ (E-)languages. In many ways, the discussions which have taken place so far attempt to show that Creoles are complex because they have the features McWhorter (2001) claims they lack. My aim in this paper is not to join the academic tug-o’-war in trying to prove that



Creole morphology revisited 

Creoles are simple or complex, but to encourage researchers to revisit aspects of Creole grammar without the prejudices which typified earlier works in the field. Only then can we be certain that our conclusions are based on solid empirical evidence.

References Alleyne, M.C. 1980. Comparative Afro-American. Ann Arbor: Karoma. Ansaldo, U. & S. Nordhoff. forthcoming. Complexity and the age of languages. In Aboh, E.O. & N. Smith (eds). Complex Processes in New Languages. Amsterdam/ Philadelphia: John Benjamins. Aronoff, M., I. Meir & W. Sandler. 2005. The paradox of sign language morphology. Language 81 (2). 301–344. Bailey, B.L. 1966. Jamaican Creole Syntax: A Transformational Approach. Cambridge: Cambridge University Press. Bakker, P. 2003a. The absence of reduplication in Pidgins. In Kouwenberg, S. (ed.). Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. 37–46. Bakker, P. 2003b. Reduplication in Saramaccan. In Kouwenberg, S. (ed.). Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. 73–82. Bauer, L. 1983. English Word-formation. Cambridge: Cambridge University Press. Bhatt, P. & I. Plag (eds). 2006. The Structure of Creole Words: Segmental, Syllabic and Morphological Aspects. Tübingen: Max Niemeyer Verlag. Bickerton, D. 1990. Language and Species. Chicago: University of Chicago Press. Bollée, A. 2003. Redupliation in Seychelles Creole. In Kouwenberg, S. (ed.). Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. 219–225. Booij, G. 2004. Inflection and derivation. In Booij, G., C. Lehmann, J. Mugdan, W. Kesselheim & S. Skopeteas (eds). Morphologie/ Morphology. Berlin: Walter de Gruyter. 360–369. Braun, M. 2005. Word-Formation and Creolisation: The Case of Early Sranan. Ph.D. dissertation, Universität Siegen. Brousseau, A. 1988. Triptyque sur les composés: Les noms composés en Français,Fongbe et Haitien en regard des notions de tete et de percolation. [Number 2 in Travaux de recherche sur le créole haïtien.] Montréal: Université de Québec Montréal. Brousseau, A., S. Filipovich & C. Lefebvre. 1989. Morphological processes in Haitian Creole: The question of substratum and simplification. Journal of Pidgin and Creole Language 4. 1–36. Brousseau, A. & E. Nikiema. 2004. Towards a derivational inventory of St. Lucian: Methodology and analysis. Paper presented at the biennial meeting of Society for Caribbean Linguistics, Curaçao. August 2004. Bybee, J., R. Perkins & W. Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: Chicago University Press. Cassidy, F.G. 1957. Iteration as a word-forming device in Jamaican folk speech. American Speech 32 (1). 49–53. Cassidy, F.G. 1961. Jamaica Talk: Three Hundred Years of the English Language in Jamaica. London: Macmillan.



Joseph T. Farquharson Chomsky, N. 2001. Language and problems of knowledge. In Martinich, A.P. (ed.). The Philosophy of Language. New York: Oxford University Press. 581–599. DeCamp, D. 1974. Neutralizations, iteratives, and ideophones: The locus of language in Jamaica. In DeCamp, D. & I. Hancock (eds). Pidgins and Creoles: Current Trends and Prospects. Washington, D.C.: Georgetown University Press. 46–60. DeGraff, M. 2001. On the origin of creoles: A Cartesian critique of Neo-Darwinian linguistics. Linguistic Typology 5. 213–310. Dryer, M.S. 2005. Coding of nominal plurality. In Gil, D., M. Haspelmath, M.S. Dryer & B. Comrie (eds). The World Atlas of Language Structures. New York: Oxford University Press. 138–141. Farquharson, J.T. 2006. A prolegomenon to the study of derivational affixation in Jamaican Creole. Paper presented at Leipzig Spring School Students Conference. 25–26 March 2006. Frishberg, N. & B. Gough. 2000. Morphology in American Sign Language. Sign Language and Linguistics 3 (1). 103–131. Gil, D. 2001. Creoles, complexity, and Riau Indonesian. Linguistic Typology 5. 325–371. Gil, D. 2006. Early human language was Isolating-Monocategorial-Associational. In Cangelosi, A., A.D.M. Smith & K. Smith (eds). The Evolution of Language. New Jersey: World Scientific. 91–98. Hale, M. 2007. Historical Linguistics: Theory and Method. Malden: Blackwell.. Heine, B. 1982. The Nubi Language of Kibera: An Arabic Creole. Berlin: Dietrich Riemer. Jespersen, O. 1993 [1894]. Progress in Language with Special Reference to English. Amsterdam/ Philadelphia: John Benjamins. Kihm, A. 2003. Inflectional categories in Creole languages. In Plag I. (ed.). Phonology and Morphology of Creole Languages. [Linguistische Arbeiten 478.] Tübingen: Max Niemeyer Verlag. 333–363. Kouwenberg, S. 1994. A Grammar of Berbice Dutch Creole. [Mouton Grammar Library 12.] Berlin: Mouton de Gruyter. Kouwenberg, S. (ed.). 2003. Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. Kouwenberg, S. & D. LaCharité. 2001a. The iconic interpretations of reduplication: Issues in the study of reduplication in Caribbean Creole languages. In Fischer, O. & M. Nanny (eds). Iconicity in Language and Literature. Special issue of the European Journal of English Studies. 59–80. Kouwenberg, S. & D. LaCharité. 2001b. The mysterious case of diminutive yala-yala. In Christie, P. (ed.). Due Respect: Essays on English and English-Related Creoles in the Caribbean in Honour of Robert LePage. Kingston: University of the West Indies Press. 124–134. Kouwenberg, S. & D. LaCharité. 2004. Echoes of Africa: Reduplication in Caribbean creole and Niger-Congo languages. Journal of Pidgin and Creole Languages 19 (2). 285–331. Kouwenberg, S., D. LaCharité & S. Gooden. 2003. An overview of Jamaican Creole reduplication. In Kouwenberg, S. (ed.). Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. 105–110. Kouwenberg, S. & E. Murray. 1994. Papiamentu. [Languages of the World 83.] München: Lincom Europa. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press.



Creole morphology revisited 

Lefebvre, C. 2002. The emergence of productive morphology in creole languages: The case of Haitian Creole. In Booij, G. & J. van Marle (eds). Yearbook of Morphology 2002. Dordrecht: Kluwer. 35–80. Lieber, R. 2005. English word-formation processes: Observations, issues, and thoughts on future research. In Štekauer, P. & R. Lieber (eds). Handbook of Word-Formation. [Studies in Natural Language and Linguistic Theory 64.] Dordrecht: Springer. 375–427. McWhorter, J.H. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74 (4). 788–818. McWhorter, J.H. 2001. The world’s simplest grammars are creole grammars. Linguistic Typology 5. 125–166. McWhorter, J.H. 2005. Defining Creole. New York: Oxford University Press. Muysken, P. 2004. Pidginization, creolization, and language death. In Booij, G. & C. Lehmann, J. Mugdan, S. Skopeteas & W. Kesselheim (eds). Morphologie/ Morphology. Berlin: Walter de Gruyter. 1653–1661. Parkvall, M. 2003. Reduplication in the Atlantic Creoles. In Kouwenberg, S. (ed.). Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. 19–36. Plag, I. (ed.). 2003. Phonology and Morphology of Creole Languages. [Linguistische Arbeiten 478.] Tübingen: Max Niemeyer Verlag. Samarin, W. 1971. Salient and substantive pidginization. In Hymes, D. (ed.). Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. 117–140. Seuren, P.A.M. 2001. A View of Language. Oxford: Oxford University Press. Seuren, P.A. M. & H.C. Wekker. 2001. Semantic transparency as a factor in creole genesis. In Seuren, P.A.M. (ed.). A View of Language. Oxford: Oxford University Press. 421–431. Steinkrüger, P. 2003. Morphological processes of word formation in Chabacano (Philippine Spanish Creole). In Plag, I. (ed.). Phonology and Morphology of Creole Languages. [Linguistische Arbeiten 478.] Tübingen: Max Niemeyer Verlag. 253–268. Steinkrüger, P. 2006. The puzzling case of Chabacano: Creolization, substrate, mixing and secondary contact. Paper presented at the Tenth International Conference on Austronesian Linguistics. Puerto Princesa City, Palawan, Philippines. 17–20 January 2006. Available at: http://www.sil.org/asia/philippines/ical/papers.html. Van den Berg, M. 2003. Early 18th century Sranan –man. In Plag, I. (ed.). Phonology and Morphology of Creole Languages. [Linguistische Arbeiten 478.] Tübingen: Max Niemeyer Verlag. 231–251. Velupillai, V. 2003. The absence of reduplication in Hawai‘i Creole English. In Kouwenberg, S. (ed.). Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages. [Westminster Creolistics Series 8.] London: Battlebridge. 245–249. Whitney, W.D. 1871. Strictures on the views of August Schleicher respecting the nature of language and kindred subjects. Transactions of the American Philological Association (1869–1896) 2. 35–64. Whitney, W. D. 1967. Language and the study of language. In Lehmann, W.P. (ed. & trans.). A Reader in Nineteenth-Century Historical Indo-European Linguistics. Bloomington: Indiana University Press. 225– 256.

The role of typology in language creation A descriptive take Enoch O. Aboh & Umberto Ansaldo University of Amsterdam

1

Introduction

As part of the overall aim of this volume, namely to test to what extent notions of ‘creole’ and ‘creolization’ are actually necessary and useful in accounting for cases of language creation (Ansaldo & Matthews this volume), this chapter focuses on the role of typology in the restructuring process. Our aim is to not rely on exceptionalist explanations unless these are clearly required from the observation of the data. In other words, we do not assume a priori that a different set of explanatory principles is required to account for the formation of a new grammar in a contact environment, but rather that explanations grounded in general linguistic principles, historical linguistics and language change, can account for the emergence of new structure. The thesis presented here is simple: if we have sufficient information about the typological input in a contact environment, we are in a position to explain the structural output by looking at how features of the input varieties are selected, discarded and exapted into the new grammar. The chapter is structured as follows. Section 2 lays down the theoretical foundations of our analysis and shows how its implications for the study of contact languages differ from previous theories of genesis. The discussion there bears on the notion of simplicity (or complexity) indicating how it can be directly related to the Feature Pool (FP) generated by the source languages that new varieties emerge in. Building on this discussion, sections 3 and 4 present data from two different contact environments which, in the literature, would be assigned to different exceptional phenomena, namely creolization and admixture respectively. As we show, these labels are not useful in suggesting differential evolutionary processes, as the same principles apply to both cases of typologies in contact. This and other conclusions are discussed in the final section.



Enoch O. Aboh & Umberto Ansaldo

2 Contact languages and ‘simple grammars’ An idea commonly found in studies of contact languages (e.g. creoles) is that these involve a somehow simpler grammar if compared to their source languages. Various reasons have been put forth in the literature (e.g. pidginization; second language acquisition failure; UG) to explain such a process of simplification (see Ansaldo & Matthews this volume). In the next section, we address the issue of simplification in the context of inflection to see how it may be reformulated in a theory that adopts the FP as the main source of the emerging language.

2.1

Inflection and simplification

A recurrent characterization of creoles is that they display virtually no (verbal) inflection compared to their lexifiers (and sometimes to their substrates). The absence of inflectional morphology (but see Farquharson this volume) in these languages is often considered evidence of language development from more complex structures to simpler ones, where ‘simplification’, or loss of inflection, is analyzed as resulting from imperfect second language acquisition due to limited access to the lexifier. This generalization appears reasonable at face value, but it embeds two major shortcomings when considering creoles: i. Most studies on inflection reduction in creoles focus only on the verbal paradigm: a look at the nominal paradigm suggests a different picture, as we will show in sections 3 and 4. ii. The role of typological congruence in (dis)favoring the emergence of the inflectional paradigm in creoles has never been studied. With respect to the Atlantic creoles, for instance, the potential substrate languages (e.g. Kwa) generally lack inflectional morphology unlike the lexifiers (e.g. French, English). It could therefore be argued that the lack of inflectional morphology in the emerging creole results from properties of the FP (Mufwene 2001) and not from imperfect second language acquisition. Moreover, in other contact environments such as Sri Lanka Malay (SLM), the FP does allow for morphologization, or what is often regarded as emergence of complex structure (Ansaldo 2005; Ansaldo & Nordhoff forthcoming). This paper investigates the nominal systems in the Suriname creoles and in SLM, and shows that the lack of certain nominal inflection (e.g. in the Suriname creoles) is not a case of simplification but rather a consequence of the competition and selection process involving syntactic and semantic features of the languages in contact. We propose that



The role of typology in language creation 

only features with semantic content enter this competition. In this view, we take syntactic features such as topic, focus, specificity, modality, as well as other discourse-related specifications, to be more competitive than purely formal syntactic features (e.g. phi‑features­ in terms of Chomsky 1995 and much related work). Accordingly, purely nominal inflection (e.g. person and number) is analyzed as an agreement phenomenon conditioned by core syntax (e.g. a spec-head requirement for checking purposes). We claim that such pure agreement features are disfavored in a situation of language contact and are likely to fade out, giving the illusion of inflection reduction (see Aboh 2006a). Following this line of argumentation, we interpret the development of case morphology in Sri Lanka Malay (SLM) as a piece of evidence that (a) morpho-syntactic elaboration happens when supported in the typological make-up of the languages in contact, and (b) in some contact situations, certain features, which otherwise would be regarded as pure inflectional features (e.g. case), may become more competitive because they have a semantic function that makes them more prominent at the discourse level. Accordingly, these features are likely to be selected in the emerging language. In our approach therefore, inflection reduction in the classical sense is a matter of the competitiveness of the relevant inflectional feature, and the typological make-up of the competition pool. Put together, these factors suggest that, when possible, the FP allows development of new forms that would appear ‘more complex’ under traditional accounts that only consider morphological shape. What is crucial for this framework is that such new forms are merely recombinations of features already present in the FP. The two case studies presented illustrate the role of typology in two different contexts: in both cases, establishing the FP is necessary in order to understand what can emerge from the restructuring process. In the case of the Suriname creoles (SCs), the typological matrix of the source languages (e.g. Kwa, English, Portuguese) appears rather homogeneous in many respects (e.g. word order, absence of inflectional morphology). This homogeneity directly translates into the new languages (i.e., the creoles) which systematically intertwine features from the two typological poles: Kwa versus Germanic. The end result is the emergence of creoles (e.g. Sranan, Sarmaccan) which appear to be more mixed than often assumed in the literature (pace Bickerton 1981, 1988; McWhorter 2001). In the second case, the ‘ganging-up’ of two unrelated yet typologically close systems, i.e. case marking in Sinhala and Sri Lankan Tamil, leads to typological dominance in the pool, resulting in a radical typological shift from Malayic to Lankan grammar, in which transfer of the L2/3 feature is heavy, and innovation more limited. We believe that comparing such different yet related contexts is the first step towards 1. This is compatible with the view that only contentful lexical items of the source languages can be restructured in the emerging language (e.g. Muysken 1981; Chaudenson 2001, 2003). In this paper, we take this view to also hold at the level of syntactic features. 2. For earlier developments of these ideas, see Ansaldo (2004, 2005, to appear); Ansaldo & Nordhoff (forthcoming).



Enoch O. Aboh & Umberto Ansaldo

uncovering possible general tendencies of contact environments and their structural outcomes.

2.2

The Noun Phrase as a case study for competition and selection

Before turning to the role of typology in language creation, we need to preview the chosen dimension of structural analysis, namely the NP, and its significance for our understanding of language creation. Since the 20th century, a question that has been central to studies of creole languages is that of their genetic affiliation, namely, to what extent a creole Xn+1 could be identified as descending from a language or a family of languages X. The answer to this question of course bears on the related question of whether creoles are isolated from their source languages, and represent an identifiable typological class of new languages mainly characterized by the process by which they came into existence. As is traditionally the case in linguistic studies, one way to approach these two questions is to study creole morphosyntax and creole clause structure(s) in order to characterize creole grammar(s). This method of investigation in linguistics has led to the situation where most studies on creole genesis put a lot of emphasis on such major components of the clause such as the expression of verb phrase and its arguments, the realization of tense, aspect, and mood specifications, as well as the coding of the complementizer field. Various studies in this domain have led to the generalization that creole languages display, among others, the following tendencies (see also Farquharson this volume): – absence of inflectional morphology; the verb is almost never inflected; – expression of tense, mood, and aspect with free morphemes in a rigid TenseMood-Aspect order; – introduction of additional arguments by means of serial verb constructions; – interrogation without subject-verb inversion (e.g. unlike in Romance and Germanic) but with fronting of the question-word only; – sentence subordination by means of grammaticalized verbs of saying. Given that these properties are frequently found across creole languages of various sources (e.g. French-based vs. English-based), they are often taken as defining properties of creoles (e.g. McWhorter 1998), or else default manifestations of the human language capacity (e.g. the Language Bioprogram Hypothesis in terms of Bickerton 1981). Without getting into the details of the theoretical (in)adequacies of the creole typological class or the Language Bioprogram, it seems obvious to us that an approach that takes the mentioned tendencies as defining properties of creoles leaves very little

3. For this see e.g. Ansaldo & Matthews (2001), Slobin (2005), Ansaldo (to appear b), Siegel (to appear).



The role of typology in language creation 

room for variation across the creoles. In such views, creoles would differ superficially in form, but not in structure. Starting from a different perspective, we argue in this paper that creoles actually differ deeply in structure. We claim that the generally observed tendencies can be accounted for independently on the basis of the FP in which a creole emerges. In order to do so, we discuss the noun phrase in the SCs, SLM and their potential source languages. The choice of the noun phrase is motivated by the following reasons: i. There is not much empirical study on the noun phrase in creole literature. ii. In the case of the Caribbean creoles, the noun phrase represents the domain par excellence where these languages vary. A simple illustration of this variation is that while Sranan has prenominal determiners, Haitian has postnominal determiners only. This is so even though Sranan and Haitian are assumed to have the same substrate languages (e.g. Gbe) and both their lexifier languages, English and French, have prenominal determiners (e.g. the, le). Similar variation is observed across creoles of the same lexifier (e.g. French creoles). Assuming that these determiners encode deixis inside the noun phrase in a way similar to tense within the sentence domain, the variation just illustrated here is comparable to a situation where tense is pre-verbal in a creole X but post-verbal in another creole Y even though both X and Y have the same substrate, and their lexifiers, though different, mark tense by a suffix on the verb. Needless to say, such a variation is hardly discussed in the literature. iii. In the case of SLM, in the NP we find a full case system, which is absent from the lexifier. A number of (spatial) prepositions, a possessive marker and a marker of definiteness, all of Malayic origin, are reanalyzed as post-nominal case suffixes. The case semantics are typically Lankan, i.e. match very closely the case systems of Sinhala and Tamil, as can be seen in the Nominative-Dative alignment and in the prominent discourse role of so-called ‘Dative Subjects’. The degree of congruence varies, however, depending on specific semantic properties of each case as well as the retention of lexifier features, as will be discussed below. This shows that, within a FP in which morphology inflection for case marking is prominent, ‘morphologization’ can indeed occur as a result of contact-induced change (see section 4 and also Ansaldo 2005, to appear). iv. We believe variation within the noun phrase to derive from structural properties of the language and to represent a piece of evidence that creoles are not structurally isomorphic. Taking the variation inside the nominal domain seriously, we therefore show that the differences across creoles can be accounted for if one assumes a theory of genesis that primarily takes into account typological properties of the languages in contact as well as the process of competition and selection as the driving force in language creation.



Enoch O. Aboh & Umberto Ansaldo

2.3

The Feature Pool

In Croft’s (2000) view, languages are best understood as constantly evolving systems that defy taxonomic categorization. In this sense, language can be seen as a population of linguistic features and grammar as a combination of idiolects: communication thus entails interbreeding of different idiolects. This means that when speakers interact they exchange utterances; in this exchange, they may replicate the linguistic features of their environment (their community, network, tribe etc.) identically, or they may replicate them with alterations, innovations etc. An identical replication is what we might expect in an environment with a high degree of monolingualism and strong normative tendencies (note however that language change, often defined in terms of population migration, also occurs in environments where contact is not present – or salient – which means that altered replication happens in language transmission between speakers in a homogeneous environment). On the other hand, in a highly multilingual environment with low normative tendencies, an altered replication is quite likely (Ansaldo 2006). In other words, in a typical situation of language contact, where speakers afford multilingual competence and where negotiation of different linguistic codes is the norm, altered replication can be expected to occur with rather high frequency. It seems therefore clear that, in order to understand the structural outcome of contact environments, we need to focus our attention on what possibly determines the direction of processes of altered replication. In order to do so, the notion of FP (Mufwene 2001) is particularly relevant: a FP can be defined as the total set of linguistic variables available to speakers in a contact environment in which a process of competition, selection and exaptation takes place. Understanding the outcomes of language contact, we propose, entails understanding precisely these dynamics of competition and selection, and subsequent exaptation. As already noted in Mufwene (2001), issues of markedness may be significant factors in determining the outcome of this process. Moreover, there is also a dimension of discourse salience (e.g. ‘perceptual salience’ and ‘semantic transparency’ in Mufwene 1991) that plays a role in the selection process. In the spirit of these works, we propose that a FP may be taken to represent the population of utterances OR features available to speakers in a given contact environment. A FP can therefore stand for (a) the admixture of linguistic codes to which speakers of a given multilingual environment can be potentially exposed to (i.e. the FP at the population level), or (b) the admixture of linguistic codes which a speaker of a given multilingual environment has knowledge of (i.e. their idiolect). These form the FP at the speaker level, that is, elements which their idiolect(s) draws from. We believe that these two levels of analysis should be kept separate because the principles that apply to the FP at the speaker’s level might not extend to the FP at the 4. Lass (1997: 377) defines a language as “a population of variants moving through time and subject to selection”.



The role of typology in language creation 

population level. For the purpose of this paper, and for the sake of clarity, we focus on the population level, and claim that the mechanisms that regulate competition and selection there are to be found in the structural-typological make-up of the FP. In this regard, a relevant notion for our discussion is that of typological dominance, which we evaluate on the basis of ‘unmarkedness’ of features within the FP. Unmarkedness is intended here simply as a manifestation of frequency, regularity and salience in a FP (in the sense of Haspelmath 2006), and points out those features that are more competitive and therefore more likely to be selected over less frequent/ regular, non-congruent, or marked features. The argumentation goes as follows. Suppose, for instance, a scenario where two strictly SVO languages compete with one strictly SOV language. Our working hypothesis is that the congruent (or converging) SVO-type features will reinforce each other and therefore become more regular/ frequent in the FP. This, in turn, makes such features acquire a high competitiveness, which may favor their selection in the emerging language. Being marked in such a FP, however, the SOV-type features appear disfavored or less competitive because their selection (even though possible) is subject to the huge pressure of more competitive SVO-type features. As already mentioned in the introduction, this simple framework suggests that we should be able to explain the structural properties of the output (i.e. the emerging language) if we have a good understanding of the FP it emerged from (i.e. features of the input varieties that the language creators were exposed to). In the cases discussed in sections 3 and 4, our current knowledge of the sociohistorical situations in which these contact languages came into being, and the increasing number of detailed studies on the possible donor languages and the contact languages themselves, enable us to precisely identify which feature in the contact language pertains to which donor language, and was therefore active in the FP. Yet, there are cases of contact languages for which less is known about how they came into existence. In this case, we propose that one can still reconstruct the FP backwards by (i) isolating a set of relevant features in the new language, and (ii) identifying a set of potential and relevant donor languages that these features can be traced back to. This way, we can reconstruct the typological make-up of the FP, and hypothesize which types of features can be selected in the competition (presumably depending on factors such as frequency and congruence). Keeping to the input-output relation strictly, we suggest that in most cases, new contact varieties only recombine a subset of features that were already present in the

5. SLM offers a good illustration: while Malay is verb-medial, both Sinhala and Tamil are verbfinal languages and, consequently, SLM is predominantly verb-final: SVO + SOV + SOV > SOV. 6. See Mufwene (2001) for discussions on the Founder Principle.



Enoch O. Aboh & Umberto Ansaldo

FP. At this stage then, it is possible to say if a new language simply replicated (a subset of) features already present in the FP or adopted exogenous or new features. A possible objection here could be to say that the proposed approach is to some extent circular in the sense that in some cases, one first identifies what feature is present in the new language, and subsequently reconstructs it back to the FP. Setting aside the role of similar reconstruction in comparative approaches to language study since the beginning of modern linguistics, we maintain that the type of reconstruction employed in such cases is precisely the linguistic tool required to (i) firmly establish the role of typology in the emergence of contact languages, and (ii) identify precisely how competition and selection proceed.

2.4

Simplification again

The characterization of creoles and pidgins as ‘simple’ often assumes explicitly or implicitly that these languages lack certain properties of their source languages. A fact that is often put forward as illustration is that, while creoles generally derive their lexicon from their lexifier languages (e.g. French, Portuguese, English, Spanish, Dutch), they typically lack the inflectional morphology found in those languages. Accordingly, creoles tend to be isolating, while their lexifiers are not. The view of language simplification as an inherent part of creolization obviously makes sense in a theory of creole genesis that assumes creoles to develop from pidgins which in turn develop from language dilution (Bickerton 1988; McWhorter 2001). Without getting into a critique of this developmental scenario, it is worth mentioning that the analysis of creole languages from the perspective of simplification is misleading (see Farquharson this volume). Indeed, studies in this domain often give the impression that a given creole has one main source (whether the superstrate or the substrate) to which it can be compared. Needless to say, such a view of ‘one source, one creole’ reflects major theories of genesis such as the universalist approaches, the superstrate-oriented theories and the substrate-oriented theories. Yet, our discussion of data from the SCs and from SLM points to the fact that contact languages generally express various feature combinations derived from different source languages in the FP. Put another way, languages in contact form a FP from which the new language derives a successful combinatory. In this proposed multisource approach that takes as its only driving force the process of competition and selection, we further observe that features that form the FP are not equally competitive. For instance, granting that morphemes are overt expressions of bundle of features, it 7. We draw the reader’s attention to the fact that while this approach can predict which features are likely to be found in the new language as a consequence of being in the FP, it does not predict which specific feature, among equally competing varieties, will eventually win the competition. This type of prediction will have to take into account external factors such as sociopolitical situations (e.g. social prestige, cultural significance).



The role of typology in language creation 

appears that morphemes that are semantically active are highly competitive and make it to the emerging language, regardless of whether they are of a derivational or inflectional type. On the contrary, morphemes that are semantically vacuous, or light, are less competitive and fade out of the FP. We conclude from this that such less competitive morphemes are not retained in the new language, not so much because of their putatively difficult acquisition, but rather because they did not survive in the FP. A welcome implication of our analysis is that creoles, for instance, appear more mixed than often claimed in the literature. In the following sections, we illustrate the role of the typological make-up of the FP in the linguistic type of the emerging creole.

3 Competition and selection in English, Gbe and the Suriname creoles The first case study we present to support our view involves the SCs, Sranan and Saramaccan, and their source languages, Gbe and English. For ease of presentation, we propose three-way comparison tables that contrast the features of the source languages to those of the resulting creoles. We first start with general properties of the noun phrase and aspects of the sentence in English, Gungbe, and the SCs that are relevant for the discussion.

3.1

Properties of the noun phrase in English, Gungbe and the Suriname creoles

In this section, we propose a general description of word order patterns and certain properties of the noun phrase in the competing languages forming the FP. It is crucial to understand that what matters for this discussion are the patterns that the languages display. In this regard, both English and Gbe languages distinguish between singular and plural forms in their nominal system. In Gungbe, for instance, number is encoded by a marker/ determiner only, leaving the noun in its bare form (1).

(1) Wémá éhè lέ book Dem Num ‘These books’

In English, on the other hand, number is always marked on the noun and sometimes on both the noun and determiner.

8. The FP that led to the formation of Saramaccan also includes Portuguese but this aspect is not discussed here; see Smith 1987; Smith & Cardoso 2004 for details. 9. See Aboh (2006a, b, in press) for discussion on details of such patterns in the Gbe languages and the Suriname creoles.



Enoch O. Aboh & Umberto Ansaldo



(2) The books vs. these books

The absence of agreement between the Gungbe noun and number determiner extends to the verbal domain where there is no subject-verb agreement either. In English on the other hand, this minimally translates into the verbal system with regard to third person singular -s. If we consider these facts from the typological perspective, Gbe languages (e.g. Gungbe) and English appear relatively poor when one considers inflectional morphology. Observe, for instance, that there are fewer than ten regular inflectional suffixes in English (Smith & Cardoso 2004: 8), and none in Gbe (Aboh 2004a).

(3)

Verbal inflections /-z/ [3rd person.singular]  hates /-iŋ/ [progressive]  be hating /-d/ [perfective]/ [passive]/[perfect]  hated/be hated/have hated

With regard to the present discussion, this means that the FP that led to the emergence of the SCs would lack features of the sort that create subject-verb agreement or noundeterminer agreement. We therefore do not expect these creoles to show such features, because the emerging language can only realize a subset of features that were made available by the competing languages in the FP. In the case at hand, our observation is compatible with the fact that the SCs do not have inflectional morphology (e.g. as found in agglutinating languages). (4) di bakuba versus den bakuba mi njan di bakuba den njan di bakuba

 the banana vs. the bananas  I ate the banana  they ate the banana

In proposing that the lack of inflectional morphology in Sranan and Saramaccan is a direct effect of the absence of such features in the competing languages, we imply that creole formation should be understood as a process of feature competition in the input (i.e. FP) that leads to a particular feature-combination in the output (i.e. the creole). It follows from this view of language creation that the paucity of inflectional morphology observed in certain creoles (e.g. Atlantic creoles) is not necessarily the result of imperfect language acquisition, as is often put in the literature, but is also a property of the FP. That some creoles (with a Romance lexifier language) exhibit long versus short verb forms that are comparable to verbal inflection to some extent underscores our view, that inflection may be acquired when it is associated with some semantics. That creole formation essentially involves feature recombination is further illustrated by Table 1, which recapitulates the general properties of the noun phrase in the competing languages and the SCs.10

10. See Lardiere (2005) for similar views with regard to the morphology-syntax interface.



The role of typology in language creation 

Table 1.  General properties of the Noun Phrase in English, Gbe and the Suriname creoles Noun marked as definite

Generic bare nouns

(In)definite bare nouns

Discourse deixis

English





*

 [this/that]

Gbe

*





Sr/Sc

*





Prenom. Postnom. Demonstradeterminer determiner tive reinforcer 

*



*



*





*



 this man here

As one can see from the last row of Table 1, the SCs do not mark the noun for definiteness, and use (in)definite bare nouns similarly to the Gbe languages. Also similarly to the Gbe languages, the SCs only mark the noun phrase for specificity and therefore make a systematic distinction between specific and non-specific referents. As Aboh (2004a, b, 2006a) argues, the SCs parallel with the Gbe languages when it comes to the function of the noun phrase, because the semantics of the determiner in these creoles developed under the influence of the Gbe languages. With regard to syntax, however, it appears that the SCs follow the English pattern. Table 1 clearly shows this since both English and the SCs have pre-nominal determiners and demonstratives, as well as the so-called demonstrative reinforcer construction where the noun appears to precede the demonstrative or some adverbial-like locative element. Such noun phrases are thought to encode emphasis (Berstein 1997). We know by now that Gbe languages lack such word order variation inside the noun phrase that would reflect discourse-syntax interaction. It seems clear from this description therefore that the Gbe languages cannot have provided the trigger for such syntax in the Suriname creoles. Instead, we face a situation where the emerging languages clearly map semantics (or, say, discourse-semantics) from one language type (i.e. Gbe) onto syntactic patterns from another language-type (Germanic). Assuming that this is the right characterization, one may wonder what principle allows such apparently peculiar recombination in the emerging language. We propose that the answer to this question resides in the fact that the syntax and the function of functional categories (in this case, the locus of specificity) are subject to different constraints in a situation of competition. It is arguable that function will be constrained by semantic properties such as prominence, discourse-relevance, frequency, etc., while syntax implies such notions as parameter setting, formal licensing, economy principle, etc. Given this, it is conceivable that in a situation of language contact and therefore competition and selection, the syntax and the semantics of functional categories are disassembled and reassembled in various ways that do not necessarily match the combinations found in the source languages (Lardiere 2000). Provided that such recombinations comply with UG (i.e. human language capacity), they may survive and eventually dominate other competing recombinations in the FP. We claim that the SCs are just



Enoch O. Aboh & Umberto Ansaldo

an illustration of this process in that the English syntactic pattern won the battle of the syntax of the noun phrase, while the Gbe languages won the battle of semantics. This creates a noun system (call it a DP) that has the semantic properties of noun phrases in Gbe, but the syntax of English noun phrases (see Aboh 2004b, 2006a). We are therefore led to conclude that the phenotype of creoles (i.e. the expression of selected features including their syntax and semantics) cannot be said to derive from processes such as acquisition/ restructuring and loss/ reconstruction but rather from a general recombination of the linguistic features from the competing languages that made it to the FP. In the following sections, we discuss in more detail the results of the recombination of certain features into the SCs and we suggest that these creoles (and presumably other languages that arose in similar conditions) are more ‘mixed’ than usually admitted in the literature.

3.2

The function of determiners in the competing languages and the emerging creole

We mentioned in previous paragraphs that the SCs map the function of the determiners in Gbe onto the syntax of the determiner in English. Table 2 contrasts Gbe languages and English in their usage of the determiner. Table 2.  Feature combinations and determiner expression in Gungbe and English D-features

Gungbe

English

[+specific +definite, +plural]

l l

the

[+specific, +definite, -plural]

l

the

[+specific, -definite, +plural]

é l

some/ certain

[+specific, -definite, -plural]

é

a/ some/ certain

l ∅ [definite] ∅ [generic] ∅ [indefinite]

the

[-specific, +definite, +plural] [-specific, +definite, -plural] [-specific, -definite, +plural] [-specific, -definite, -plural]

the ∅/ any [generic] a/ some

Observe from this table that Gbe languages use distinct determiners post-nominally for specific definite and specific indefinite noun phrases as well as for number. However, noun phrases that are simply definite, indefinite or generic do not show any marking (see Aboh 2004a, b, 2006a, and references cited there for discussion).11 English, on the other hand, exhibits pre-nominal determiners (e.g. the, a) which are ambiguous with regard to specificity, but necessarily encode (in)definiteness. In Enç 11. See also Ionin (2006) for a similar analysis of specificity as a syntactic category distinct from definiteness.



The role of typology in language creation

(1991: 7), for instance, “definiteness feature […] determines the specificity of the NP by constraining the relation of the referent of the NP to other discourse referents”. This characterization often leads to the conclusion that all definites are specific in languages like English. This feature certainly makes English different from Gbe languages where definites are non-ambiguously non-specific. Finally, English and Gbe also differ in their usage of bare nouns. In English, but not in Gbe, a bare noun (i.e. a noun that has no modifier) is sensitive to count versus mass distinction (e.g. I bought wine vs. *I bought book). With this description in mind, let us now see what picture emerges from the Suriname creoles. Table 3, based on Sranan, summarizes the semantic combinations observed in these languages. Table 3.  Feature combinations in Sranan D-features

Sranan

[+specific, +definite, +plural] [+specific, +definite, -plural] [+specific, -definite, +plural] [+specific, -definite, -plural] [-specific, +definite, +plural] [-specific, +definite, -plural] [-specific, -definite, +plural] [-specific, -definite, -plural]

den na ∅ (wan-tu) wan den(?)/ ∅ ∅ [definite] ∅ [generic] ∅ [indef]

Compared to Table 2, the last three rows of Table 3 indicate that the SCs and the Gbe languages are quite close when it comes to the expression of the feature combinations [-specific, ±definite, ±plural]. All these languages use bare nouns that are interpreted in context. English on the other hand has various determiners (e.g. a, the, some, any) that realize these feature combinations. In addition, all remaining rows in the table more or less correlate with the situation in Gungbe, including the problematic combination of features [+specific, -definite, +plural], which in Gbe is realized by a combination of the specific indefinite marker and the number marker é l, but which in Sranan surfaces as wan-tu. Here the creole combines the specific indefinite wan to the numeral tu, which expresses number. The intended meaning of the complex form wan-tu is similar to that of some, certain in English. While one cannot claim that a form like certain is too complex to be properly acquired, the emergence of a variant like wan-tu that intersects with both Gbe languages and English in expressing specific indefinite plurals is a perfect case of the feature recombination argued for here. Overall, we therefore have the picture that Sranan retained the Gbe system for the function/ semantics of the determiner (e.g. specific versus non-specific distinction, bare nouns). Put differently, discourse deixis or discourse anaphora marking seems to





Enoch O. Aboh & Umberto Ansaldo

follow the Gbe pattern. The syntax of the noun phrase, however, does not seem to be so obviously of a Gbe type. The following section discusses this issue in some detail.

Intertwining syntax and semantics

3.3

While previous discussion suggests that the semantic properties of determiners in the SCs derived from the Gbe languages, the way the components of the noun phrase are put together to form bigger nominal constituents does not seem to follow from the grammar of Gbe languages as described in Kwa literature (see Aboh 2004a and references cited there for discussion). We first start with the expression of number in English, Gbe, and the Suriname creoles. Table 4 shows that unlike the Gbe languages, English does not mark number on the definite determiner the, which may also trigger specific reading, it does not have a dedicated post-nominal number marker that also encodes definite reading, and it never marks number on the determiner only. Instead, number is marked in both the noun and the determiner, and the latter occurs pre-nominally in English. In the Gbe languages however, number is the expression of a post-nominal dedicated determiner that also encodes definite reading, and there is no realization of the number feature on the noun. Table 4.  Number marking in English, Gbe and the Suriname creoles

English

Post-N Number on Number on definite + Det and N Det only number

Number on definite Det the

Number on deictic Det

Number inflection on noun

Pre-N deictic + number

*

 [these/ those]





*



*

Gbe

*

*

*

*



*



Sr/Sc

*



*



*

*



Looking at Sranan in the last row of the table, it is interesting to observe that this creole combines features from the preceding two rows, which we can now assume to be representative of the FP when it comes to nominal expression. In this regard, it is worth noting that it cannot be the lack of salience of the English plural inflection that disfavored its selection in the emerging creole, where plurality is marked on the determiner only along the lines of the Gbe languages. Indeed, when plural morphology has some semantic weight, it has been maintained by the creole speakers. The following examples indicate that nouns that may be argued to embed ‘collective reading’, in a loose sense, surface with plural morphology.

(5) English Gungbe shoes àfkpà news xógbè

Saramaccan Ndyuka Sranan súsu susu susu njúnsu nyunsu nyunsu



The role of typology in language creation 

In his discussion of these forms, Smith (2006: 60) states that “it was not an inability to parse plural meanings out of plural forms that was at issue for the early creole formers – they ignored plural suffixes where the unmarked meaning was singular, and kept them for collective nouns, where the unmarked meaning was ‘plural’. The real reason for the loss of a meaningful plural suffix was presumably the availability of more salient pre-head structures”. In our terms, this would mean that plural inflection on the noun was lost because it is semantically vacuous and because a pre-nominal deictic determiner den could express plurality, as displayed by the competing languages. Recall that English demonstratives this and that have plural forms (these vs. those), while Gbe languages use no marking for singular nouns, but resort to a definite plural marker that follows plural nouns. We now reach the conclusion that even though it is arguable that English syntax prevailed as to the formal licensing of pre-nominal determiners (Aboh 2004b, 2006a), some aspects of these determiners conform to the pattern of both the substrate and the lexifier. With regard to inflection proper, the discussion on plural morphology in the SCs suggests that only semantically active inflection is visible for selection in a situation of language contact. The loss of possessive inflection, which we now turn to, further supports this view. Possessive constructions in English and Gbe differ in a number of interesting ways. In English one observe five general patterns in possessive constructions, as described below: 1 The Possessor precedes the Possessee, the two being related by the Saxon genitive marking (e.g. John’s book); 2 The Possessee is an indefinite noun phrase that precedes a preposition, which in turn precedes the Possessee (a friend of John); 3 Simple juxtaposition of Possessor-Possessee as in certain compounds (e.g. a horse leg vs. my leg); 4 The Possessee is an indefinite noun phrase that precedes a preposition, which precedes the possessor marked by Saxon genitive (e.g. a friend of John’s); 5 In Old English, the Possessor precedes a possessive pronoun that precedes the Possessee (e.g. Jesus Christ his sake). The Gbe languages on the other hand display three major possessive patterns, two of which correspond to the general English patterns in (1) and (3), respectively. In addition to such Possessor-Gen-Possessee and Possessor-Possessee sequences, Gbe languages also exhibit constructions of the type Possessee-Possessor-Gen, where the Possessee precedes the Possessor to which the genitive marking attaches. Put together, we can therefore say that the FP available to the creole speakers included the English patterns in (1) to (5) plus an extra Gbe pattern not covered by those five. Table 5 summarizes these patterns.



Enoch O. Aboh & Umberto Ansaldo

Table 5.  Possession marking in English, Gbe and the Suriname creoles 1 PossessorGenPossessee

English

3

4

5

6

PossesseePrepPossessor

PossessorPossessee

PossesseePrepPossessorGen

Dislocated Possessive PossessorProPossessee

  John’s book a friend of John

 a horse leg my leg

 a friend of John’s

 s f

*

*  and this we beg for Jesus Christ his sake. 1667 PEPYS Diary 12 Aug. */?  wémà Jan tn wémà cè [book my] *  a moy frigi en tere

Gbe  [Gungbe] Jan sín wémà

Sr/Sc

2

*

*

  (n)a buku fu datra oso mi mi oso

*

PossesseePossessorGen

As one may see from the last row representing the Suriname creoles, it appears that these languages retained patterns (2), (3), and (5) only. However, the creoles can be said to differ from their source languages in displaying no genitive inflection. In this regard, the absence of genitive inflection in the SCs enables these languages to develop a system that perfectly combines the two major patterns in Gbe and English, that is, [Possessor]-[Possessee] versus [Possessee]-[Possessor] without semantic loss. We take the loss of genitive inflection in the SCs to be additional evidence that only semantically active inflectional morphology is visible and (maybe) subject to transfer in a situation of language contact. Supporting evidence for this approach to the status of inflection in language contact comes from the case system in these languages. English and Gbe mark case in their pronominal system only, and can be said to belong to languages where case morphology is not semantically relevant, unlike, for instance, SLM as discussed in section 4. Given this situation, we predict that contact between these languages is not likely to produce a new language that has extensive case morphology. This is indeed what we find in Table 6, which illustrates the Nominative versus Accusative distinction in English, the Gbe languages, and the creoles. It appears here that English distinguishes between first person and third person singular, as well



The role of typology in language creation 

as first person and third person plural. The Gbe languages, on the other hand, exhibit case morphology only on first, second, and third person singular. With regard to the SCs, these make one distinction in the third person singular only. Table 6.  Case marking in pronouns in English, Gbe and the Suriname creoles English

Gungbe

Sranan

Saramaccan

NOM

ACC

NOM

ACC

NOM

ACC

NOM

ACC

I you s/he, it

me you him, her, it

ùn à é

mi ju/i a

mi i en

mi ju/i a

mi i en

we you they

us you them

mí mì yé

mì wè è/ì/ù (phonological alternation) mí mì yé

u unu den

u unu den

u unu de

u unu de

4 distinctions

3 distinctions

1 distinction

1 distinction

At first sight, one could be tempted to claim a perfect scenario of inflection loss in this case, since the SCs make only one distinction, while Gbe languages have three forms, as opposed to English which has four forms. Close scrutiny indicates, however, that the pronominal forms that were competitive in the FP were not weak pronouns as illustrated in Table 6 but rather strong forms as shown in Table 7. Table 7.  Case marking in strong pronouns in English, Gbe and the Suriname creoles English

Gungbe

Sranan

Subject/ object in cleft and dislocated sentences me you him, her, it us you them

Subject/ object in cleft Subject/ object in cleft and dislocated and dislocated sentences sentences mi ny ju j en é wi mí l unu mì l den yé l

Subject/ object in cleft and dislocated sentences

0 distinction

0 distinction

0 distinction

0 distinction

Saramaccan

mi ju hn wi unu de



Enoch O. Aboh & Umberto Ansaldo

What appears from this table is that, when it comes to strong forms, none of the languages involved in this situation of contact displays case distinction. In both English and Gbe languages, pronominal strong forms do not show any morphological distinction whether they are used in subject or object position. This last observation leads us to conclude that the loss of inflection is not related to language acquisition but instead to the nature of the inflection itself. When inflection simply reflects a syntactic configuration, such as subject-verb or verb-object configuration, it may not be competitive enough in a situation of language contact to participate in the FP from which the emerging language derives viable combinatories. However, when inflection has some semantics (e.g. intricate relation between nominative case and topicality), it may participate in the competition and selection process and emerge in the new language. As mentioned earlier, a crucial aspect of our argumentation is that the phenotype (or surface form) of new languages or creoles derives mainly from a process of combination of features from the FP, rather than a process of reduction, which has often misled researchers to think that creoles are somehow simple or simplified versions of the lexifier or substrate languages. It is also important to observe that this approach to creoles clearly reveals their mixed nature, since it focuses on how various features from the competing languages (e.g. superstrate vs. substrate) combine into a new language. In this regard, it appears that typological similarities or differences play a crucial role in framing the emerging languages. For instance, previous discussion has shown that languages with poor inflectional morphology like English and Gbe (in general) cannot give rise to a new language that has the morphology of, say, Turkish. If this is the right characterization, we then expect a situation of contact between highly inflected languages to produce a new language that has more inflectional morphology than not just English and Gbe, but also their related creoles. Section 4 on SLM explores this hypothesis.

3.4

Summary

The discussion in previous sections shows that weak and semantically vacuous inflection is disfavored in a situation of competition and selection. Given that inflection is often associated to functional categories, we conclude from this discussion that functional categories, when weak and semantically light, are less visible at the syntax-discourse interface and therefore are unlikely to be selected in the emerging language. In this view, inflection as such cannot be used as a metric for language acquisition, because it is not always visible to the learners. This means that a simple calculation of number of distinct inflectional morphology in the source languages and in the new language is a dubious method of evaluating simplicity or complexity. Finally, we have shown that in a contact situation where the competing languages show typological congruence, the emerging language is often more mixed than it would seem at first sight.



The role of typology in language creation 

4 Congruence, frequency and replication in Sri Lanka Malay Sri Lanka Malay (SLM) is the term used for several varieties of heavily restructured Malay spoken in Sri Lanka (Ansaldo & Lim 2004; Ansaldo 2005, to appear). These varieties can be described as mixed languages of trilingual base with predominantly Malay lexical material and Sinhala/ Tamil or ‘Lankan’ (see Slomanson 2006) grammar. SLM has been variously described as a ‘creole’ (Smith, Pauuw & Hussainmiya 2004), an ‘intertwined language’ (Bakker 2000), but as we consider these labels mere epiphenomena resulting from the application of a priori exceptionalist accounts, we do not engage in this debate further (for more see Ansaldo to appear). SLM clearly shows what under a certain view could be described as ‘complex’ features in the NP, in that it has a fully-fledged case-system as shown in Table 8. What is crucial is to realize that the lexifier of SLM is an Austronesian language typically ‘poor’ in morphology and in fact lacking a case system altogether. SLM therefore presents us with a rare case of morphologization, development of morphological material, as opposed to the more commonly observed reduction of it in contact environments. Moreover, SLM is a rare case in terms of genesis, as it offers us a case study of a language that retains original lexical items but completely shifts in grammar (Thomason & Kaufman 1988). Both apparently rare aspects of this language find however logical explanation through a FP-based analysis: by looking at the composition of the FP and considering the principles of competition and selection, we can explain how such a development can take place. In the following sections we discuss the FP of the nominal domain in the SLM variety of Kirinda Java (see Ansaldo 2005). SLM shows a full set of post-nominal case markers in the NP, a unique feature among Malay restructured varieties considering that Malay languages are typically isolating to mildly agglutinative and do not show nominal case morphology. On the other hand, case systems are common in Dravidian languages, to which Tamil belongs, as well as in Indo-Aryan grammars, to which Sinhala belongs. Both languages show agglutinative morphology with fusional tendencies. Moreover, because of over a millennium of intense contact, Sinhala and Tamil grammars have converged typologically and show substantial similarities (Masica 1976; Emeneau 1980). As the rest of this section shows, SLM has developed agglutinative morphology with incipient fusional tendencies in the nominal domain that suggest a typological shift away from the isolating type, as briefly illustrated in (6):12

(6) ni aanak-naŋ baek buku-yaŋ attu aada This student-DAT good book-ACC one have ‘This student has a good book.’

12. This can also be observed in other aspects of the grammar, see Ansaldo & Nordhoff (forthcoming).



Enoch O. Aboh & Umberto Ansaldo

4.1

Morpheme sources

As is typical for contact languages of the Malay world, also referred to as ‘Pidgin Malay Derived’ (PMD) varieties (Adelaar & Prentice 1996), the lexicon of SLM is heavily influenced by material of generic Malay origin. The authors do not mean to suggest the existence of a Malay pidgin is an ancestor of all Malay-based contact languages because this is historically not attested. PMD is just a convenient term to refer to what were most likely trade codes spoken in different harbors and markets of the region (still found today characterized by a mix of Eastern and Western Indonesian features, and not comparable to Standard Malay, see also Ansaldo 2005, to appear).13 Table 8 lists the SLM case markers and their probable etymology: Table 8.  Etymology of case markers in SLM Case

Marker

Etymology

Dative Accusative Possessive Locative Instrumental/ Ablative Comitative Nominative

-nang -yang -pe -ka -ring -le Ø

Malay nang, ‘towards’ Malay nya, +def marker pe < PMD punya Malay directional ke Jakarta bikin de < Malay dengan14

4.2

Structural features of case in SLM, Sinhala and Tamil

As this section shows, syntactic and semantic features of SLM are predominantly of the Lankan type. Table 9 shows the functions of the two Lankan systems, while Table 10 compares the function of SLM case with the adstrates.

13. The use of ‘Malay’ as a lexifier in the evolution of PMDs needs to be interpreted in a very broad sense, implying Malayic varieties spoken across the Indonesian archipelago. As is the case for the general profile of SLM, the Malayic element in the SLM nominal domain is predominantly present in the lexicon; most case markers have indeed Malayic etymology, and first and second person pronouns show the classic PMD, Hokkien-derived. 14. Note that the conflation of Experiencer, Benefactive, Goal and Possession is in fact not peculiar nor unique to South Asian languages, but can be seen as a universal tendency of Dative case marking, as clearly shown in Blake (1994: 145), who describes Dative as the main non-core case used to mark complements.



The role of typology in language creation 

Table 9.  Case systems in Sinhala and Tamil Sinhala

Tamil

Nominative

Agent (+Animate)

Nominative

Agent

Dative

Experiencer Goal of Vintr Beneficiary Possessor

Dative

Accusative (optional) GenitiveLocative

Human/ animate Def. goal of Vtrans Location Temporary possession

Accusative

Experiencer Goal of Vintr Beneficiary Possessor Def. goal of Vtrans

Instrumental-Ablative

Instrument Source

Ablative

Location Path Temp. poss. Source

Vocative (+anim)

Addressee

Associative/ Instrumental

Associate Instrument

GenitiveLocative

Table 10.  Thematic roles in SLM and adstrates Sinhala and Tamil SLM NOM DAT ACC POSS LOC INSTR

Agent Patient Experiencer Goal Possession Location Instrument Source

Sinhala

Tamil

    

     and Location

*

*

 

 *

Comparing Tables 9 and 10, we can make the following observations: i. Prototypical Agents in SLM are unmarked, as in the adstrates Sinhala and Tamil. ii. Experiencers and Goals in SLM are marked identically; they correspond to Dative case in the adstrates. Lack of Volition or Control, just as in Sinhala, is the key semantic feature here. iii. Accusative marking parallels the adstrate typology. The SLM definite object marker (ACC, cf. ex. 9 below) shares the feature [+definiteness] with Tamil and is optional as in Sinhala. iv. SLM shares the Instrumental-Ablative syncretism with Sinhala. v. In SLM, Genitive and Locative receive different marking.

 Enoch O. Aboh & Umberto Ansaldo

It is clear that the SLM system is a hybrid system that emerges based on the acquisition of the Sinhala and Tamil grammatical systems and reanalysis of Malay-derived adpositions; Malay lexical items are retained, but the grammar is substantially altered in the process.

4.3

Functional alignments

While syntactically case is realized post-nominally in SLM, congruent with Lankan typology, semantically there are aspects in which it can be considered innovative. In what follows we focus on the core cases of SLM and the grammatical relations they mark. Arguably the most prominent function of case in Sinhala and Tamil is the marking of Control as can be seen in the high frequency of ‘Dative subjects’ or Experiencers in Table 11. Table 11.  The Control feature NOM-DAT

alignment along the feature of Control

NOM DAT

Agent [+Control] Experiencer [-Control] Psych Vs; modal predication

This alignment is clearly present in the grammar of SLM: dative case is assigned, as in the adstrates Sinhala and Tamil, as shown in Table 12. Table 12.  Assignment of Dative in SLM and adstrates Sinhala and Tamil Thematic role Experiencer Goal/ Benefactive Possession/ Location

Kirinda Java (SLM)

Sinhala

Tamil

  

  

  

Table 12 shows a typical case of South Asian Dative case, in which several functions conflate.15 For most speakers of SLM, the absence of the Dative marker is not acceptable in most syntactic environments, particularly when in the function of Experiencer as opposed to Goal (or Beneficiary). The Agentive-Experiencer/ Goal opposition – i.e. zero-marking which appears to be reserved for prototypical Agents, and the Dative marker which covers first arguments in non-agentive roles as well as a range of Patient/ Goal roles – is the most consistent marking found in the NP in terms of frequency and distribution. Turning to the functions of Accusative in SLM, we find a less congruent picture, as shown in Table 13. 15. Not treated in this study.



The role of typology in language creation 

Table 13.  Assignment of Accusative in SLM and adstrates Sinhala and Tamil Kirinda Java (SLM)

Sinhala

Tamil

+Def +/-Anim -Obligatory Emphatic marker

+/-Def +Anim -Obligatory N.A.

+Def +/-Anim +Obligatory on +Anim Ns N.A.

Definiteness is undoubtedly a function of Accusative case in both Sinhala and Tamil, though they differ in the constraints that apply. In Sinhala, Accusative is only used with animate objects, so that definiteness can only concur with animacy. Moreover, even with animate nouns, Accusative is highly optional in Sinhala (Gair & Paolillo 1997), and we can therefore say that it is not obligatory to mark definiteness in Sinhala. In Tamil, Accusative case is obligatory for a human, direct object. In non-human direct objects, this case is only used to indicate definiteness. In SLM, animacy appears not to be marked, though it may be premature to draw this conclusion at this stage; the pronominal system shows a differential case marking for Dative first and second person singular, a typical manifestation of animacy distinction captured in Silverstein’s (1976) hierarchy as shown below (see also Ansaldo & Lim 2004; Ansaldo 2005): (8)

go-dang lu-dang dia-nang kitang-nang lorang-nang derang-nang

I-DAT you-DAT he-DAT we-DAT you.PL-DAT they-DAT

Like in Tamil, Accusative sometimes marks definiteness in SLM; like in Sinhala, it is highly optional. Moreover, Accusative marking sporadically also occurs for what appears to be emphasis (9), therefore adding a novel function to its repertoire. Interestingly, in Colloquial Malay varieties, an emphatic marker -nya is found that closely resembles the Accusative marker -yang (or -ya). Different features are therefore combined in what can be seen as an optional marker of prevalently definite objects (Ansaldo 2005).

(9) inni kendera-yaŋ bapi This chair-ACC take:go ‘Take this chair away’.

This is an interesting outcome from the point of view of the FP: we can say that the functions of Accusative in the adstrates compete more or less equally, as there is no congruence between the adstrates nor overwhelming regularity in either of them to be found.



Enoch O. Aboh & Umberto Ansaldo

Moving our analysis outside the core cases, two more interesting observations can be made. In SLM, we find a case of Instrument/ Ablative syncretism, meaning that the instrumental case marker –ring may indicate Source: (10) market-ring ais tra baaru ikkang billi bawa market-ABL ice NEG new fish buy bring ‘Get me some fresh fish from the market.’ This parallels the functions of Instrument in Sinhala, but not in Tamil, in which Instrument and Ablative are clearly distinct (Silva 2003; Ansaldo 2005, to appear). A plausible explanation for this pattern could be found in the fact that, of the two adstrates, Sinhala is quantitatively (numerically) and qualitatively (prestige) the dominant one; where the adstrates diverge, SLM seems to follow the dominant pattern. This is however not necessarily the case: in the realization of Possessive, a different outcome is found. As Table 13 shows, SLM has two distinct cases for Locative and Possession; in Sinhala and Tamil however, Location and Possession conflate (Schiffman 1999; Silva 2004). The SLM possessive case suffix is pe, a derived form of the Malay punya ‘to possess’: (11) goppe tumman go-yang e-tolak I:POSS friend I-DOM past-push ‘My friend pushed me.’ Reduced variants of punya are well attested as distinguishing features of contact-Malay varieties (PMD) such as Bazaar Malay, Baba Malay, etc. It is most likely that this feature was maintained from the original vernaculars of the SLM community and its exaptation led to the development of a new case that distinguishes SLM from its adstrates.

4.4

Summary

So far it has been shown that, when typologically congruent features overwhelmingly dominate the FP because of semantic prominence and discourse frequency, the structural output is mapped onto these features. On the other hand, the lack of congruence between competing systems seems to leave more room to the new grammar to adopt a pattern from the competing languages (presumably the one that scores higher on parameters such as discourse saliency, semantic transparency). Alternatively, the emerging language may develop a hybrid system, combining various aspects of the competing features, and thus creating novel structures. In the core case system, on which section 4 has focussed, we notice two different trends. (a) In the case of DAT, Tamil and Sinhala show basically total congruence in the FP of what is a typical South Asian DAT; this is replicated identically in the grammar of SLM and constitutes an exaptation from the point of view of a speaker of a colloquial Malay variety which completely lacks case morphology. The typological salience and local unmarkedness of DAT in the FP combine to win over possible alternatives. (b) In the case of ACC however, there is less congruence between Sinhala and Tamil, since the



The role of typology in language creation 

former is more sensitive to Animacy than Definiteness. Also, ACC appears less prominent, in terms of discourse semantics being irregular and mostly optional. In this case, the SLM ACC emerges as an adaptive innovation, which only partially combines dominant features of the FP and also exhibits a novel function. Finally, in the non-core cases we see two interesting patterns emerging: (i) Where there is a competition between the adstrates, as for the Instrument/ Ablative case, SLM patterns with the dominant one. However (ii) an early possessive morpheme retained from the ancestral language leads to the creation of a new case marking, as is shown for the case of Possessive (a Founder Principle effect, Mufwene 1991 and Ansaldo, Lim & Mufwene this volume). These different outcomes show that typological input can explain much of what goes on in the restructuring process, but not necessarily everything.

5 Conclusions This study is a first step in the application of evolutionary frameworks to language creation. The case studies presented here lead to a number of interesting reflections. In order to sensibly explain the output of contact situations, it is crucial to consider the structural-typological composition of the multilingual environment in which a new grammar emerges. The concept of FP helps us define the object of enquiry and suggests a number of relevant principles in the competition and selection process: (a) syntax-discourse prominence; (b) local (un)markedness; (c) frequency. Moreover, a FP-based approach shows that SCs can be described as more mixed than previously suggested; this indicates that the competition and selection process is to a large extent FP-specific (i.e. a feature selected from a specific variety in a contact situation may not necessarily be selected in a different contact situation). In the proposed analysis, what were once considered ‘radical creoles’ appear to have emerged from a FP of Germanic/ Romance languages and Kwa languages, which share a significant number of typological properties. It is not surprising therefore that the emerging language displays similar typological properties (e.g. lack of exuberant inflection, word order; see also Ansaldo & Nordhoff forthcoming). We conclude from this that the alleged similarity of typical Creoles requires no exceptional explanation if it is to be found in the FP. Also, a FP-based approach shows that SLM can likewise be accounted for without referring to exceptional scenarios, be they along the lines of creolization or along the lines of admixture, intertwining, etc. (see also Grant this volume). It also shows that morphological ‘abundance’, or any other feature of grammar that might be assigned the label of ‘complex’, can be selected if the FP allows for it. The competition and selection processes described in this study can be seen as cases of admixture where lexical, syntactic and semantic features of different grammars recombine into new grammatical (and typological) profiles (see Ansaldo to appear).



Enoch O. Aboh & Umberto Ansaldo

While this approach to language creation appears promising, a number of principles that govern the competition and selection of linguistic features need further elaboration. In particular, further study is needed to better understand the relationship between the mechanisms underlying this process at speaker and population levels, respectively. We intend to return to this issue in future work.

References Aboh, E.O. 2004a. The Morphosyntax of Complement-Head Sequences: Clause Structure and Word Order Patterns in Kwa. Oxford/ New York: Oxford University Press. Aboh, E.O. 2004b. Toward a modular theory of creole genesis. Paper presented at the International Joint Conference of the SCL-SPCL-ACBLPE on Caribbean and Creole Languages. 11–15 August 2004. Aboh, E.O. 2006a. The role of the syntax-semantics interface in language transfer. In Lefebvre, C., L White & C. Jourdan (eds). L2 Acquisition and Creole Genesis: Dialogues. Amsterdam/ Philadelphia: John Benjamins. 221–252. Aboh, E.O. 2006b. Complementation in Saramaccan and Gungbe: The case of c-type modal particles. Natural Language and Linguistic Theory 24 (1). 1–55. Aboh, E.O. 2007. La genèse de la périphérie gauche du saramaka: Un cas d’influence du substrat? In Gadelii, K. & A. Zribi-Hertz (eds). Grammaires Créoles et Grammaire Comparative. Saint Denis: Presses Universitaire de Vincennes. 73–97. Adelaar, K.A. & D.J. Prentice. 1996. Malay: Its history, role and spread. In Wurm, S.A., P. Mühlhäusler & D.T. Tryon (eds). Atlas of Languages of Intercultural Communication in the Pacific, Asia and the Americas. Berlin: Mouton de Gruyter. 673–693. Ansaldo, U. 2004. The evolution of Singapore English: Finding the matrix. In Lim, L. (ed.). Singapore English. A Grammatical Description. Amsterdam/ Philadelphia: John Benjamins. 127–149. Ansaldo, U. 2005. Typological convergence and admixture in SLM. ms. University of Amsterdam. Ansaldo, U. 2006. Social and structural principles of contact languages: An evolutionary framework. Paper presented at the Department of Linguistics seminar, University of Cape Town, South Africa. 25 April 2006. Ansaldo, U. 2007. Review of John McWhorter, 2005, Defining Creole, Oxford University Press. Journal of Pidgin and Creole Languages 22 (1). Ansaldo, U. to appear. Revisiting SLM. Genesis and classification. In Dwyer, A., D. Harrison & D. Rood (eds). A World of Many Voices: Lessons from Documented Endangered Languages. [Studies in Language.] Amsterdam/ Philadelphia: John Benjamins. Ansaldo, U. & L. Lim. 2004. Kirinda Java and the Malay Creoles of Sri Lanka. Paper presented at the Curaçao Creole Conference, Curaçao, Netherlands Antilles. 11–15 August 2004.  Ansaldo, U. & S.J. Matthews. 2001. Typical creoles and simple languages. The case of Sinitic. Linguistic Typology 5 (2/3). 311–326. Ansaldo, U. & S. Nordhoff. forthcoming. Complexity and the age of languages. In Aboh, E.O. & N. Smith (eds) Complex Processes in New Languages. Amsterdam/ Philadelphia: John Benjamins.



The role of typology in language creation 

Bakker, P. 2000. Rapid language change: Creolization, intertwining, convergence. In Renfrew, C., A. McMahon & L. Trask (eds). Time Depth in Historical Linguistics. Cambridge: McDonald Institute for Archeological Research. 585–620. Bernstein, J. 1997. Demonstratives and reinforcers in Romance and Germanic languages. Lingua 102. 87–113. Bickerton, D. 1981. Roots of Language. Ann Arbor, MI: Karoma. Bickerton, D. 1988. Creole languages and the Bioprogram. In Newmeyer, F. (ed.). Linguistics: The Cambridge Survey 2. 268 –284. Blake, B. 1994. Case. Cambridge: Cambridge University Press. Chaudenson, R. (revised in collaboration with S.S. Mufwene). 2001. Creolization of Language and Culture. Routledge. London. Chaudenson, R. 2003. La Créolisation: Théorie, Applications, Implications. Paris: L’Harmattan. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Croft, W. 2000. Explaining Language Change. An Evolutionary Approach. Edinburgh: Longman. Emeneau, M.B. 1980. Language and Linguistic Area. Stanford, California: Stanford University Press. Enç, M. 1991. The semantics of specificity. Linguistic Inquiry 22. 1–26. Gair, J. & J. Paolillo. 1997. Sinhala. München: Lincom Europa. Haspelmath, M. 2006. Against markedness (and what to replace it with). Journal of Linguistics 42 (1). 25–70. Ionin, T. 2006. THIS is definitely specific: Specificity and definiteness in article systems. Natural Language Semantics 14. 175–234. Lardiere, D. 2000. Mapping features to forms in second language acquisition. In Archibald, J. (ed.). Second Language Acquisition and Linguistic Theory. Oxford: Blackwell. 102–129. Lass, R. 1997. Historical Linguistics and Language Change. Cambridge: Cambridge University Press. Masica, C. 1976. Defining a Linguistic Area: South Asia. Chicago: The University of Chicago Press. McWhorter, J. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74. 788–817. McWhorter, J. 2001. The world’s simplest grammars are creole grammars. Linguistic Typology 5 (3/4). 125–166. McWhorter, J. 2005. Defining Creole. Oxford: Oxford University Press. Mufwene, S.S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Muysken, P. 1981. Halfway between Quechua and Spanish: The case of relexification. In Highfield, A. & A. Valdman (eds). Historicity and Variation in Creole Studies. Ann Arbor: Karoma. Schiffman, H.F. 1999. A Reference Grammar of Spoken Tamil. Cambridge: Cambridge University Press. Siegel, J. 2007. Recent evidence against the Language Bioprogram Hypothesis: The pivotal case of Hawai‘i Creole. Studies in Language 31 (1). Silva, A.W.L. 2003. Tamil. Sri Lanka: Pubudu Printers Padana. Silva, A.W.L. 2004. Sinhalese. Sri Lanka: Pubudu Printers Padana. Silverstein, M. 1976. Hierarchy of features and ergativity. In Dixon, R.M.W. (ed). Grammatical Categories in Australian Languages. [AIAS Linguistic Series 22.] Canberra: Australian Institute of Aboriginal Studies; New Jersey: Humanities Press. 112–171



Enoch O. Aboh & Umberto Ansaldo Slobin, D. 2005. From ontogenesis to phylogenesis: What can child language tell us about language evolution? In Parker, S.T., J. Langer & C. Milbraith (eds). Biology and Knowledge Revisited. From Neurogenesis to Phylogenesis. Mahwa, NJ: Erlbaum. Slomanson, P. 2006. SLM morphosyntax: Creole or mixed? In Deumert, A. & S. Durrleman (eds). Structure and Variation in Language Contact. Amsterdam/ Philadelphia: John Benjamins. 135–158. Smith, N. 1987. The Genesis of the Creole Languages of Surinam. Ph.D. dissertation. University of Amsterdam. Smith, N. 2006. Very rapid creolization in the framework of the restricted motivation hypothesis. In Lefebvre, C., L. White & C. Jourdan (eds). L2 Acquisition and Creole Genesis: Dialogues. Amsterdam/ Philadelphia: John Benjamins. 49–65. Smith, N. & H. Cardoso. 2004. A new look at the Portuguese element in Saramaccan. Journal of Portuguese Linguistics 3. 115–147. Smith, I., S. Pauuw & B.A. Hussainmiya. 2004. SLM: The state of the art. In Singh, R. (ed.) Yearbook of South Asian Languages 2004. Berlin/ New York: Mouton de Gruyter. 197–215. Thomason, S. & T. Kaufman. 1988. Language Contact, Creolization and Genetic Linguistics. Berkeley/ Los Angeles: University of California Press.

Creoles, complexity and associational semantics* David Gil Max Planck Institute for Evolutionary Anthropology, Leipzig

1

Creoles and complexity

Can one tell that a language is a creole just by looking at it, that is to say, by examination of its synchronic grammatical properties, without reference to its history? Given that the definition of a creole language is couched exclusively in historical terms, this question is a substantive one, which in recent years has attracted an increasing amount of attention. One frequently-proposed answer to this question appeals to the notion of complexity. Specifically, it is claimed that one can indeed tell whether a language is a creole just by looking at it, namely by evaluating its overall level of complexity. This latter * My thinking on creoles and complexity has profited greatly from an ongoing dialogue with John McWhorter, who, in the best spirit of true scientific debate, has forced me, again and again, to re-examine and refine my positions on the issues that interest us both and on which we do not always agree. My understanding of the issues discussed in this paper has also benefited, in ways that I am not always conscious of myself, from discussions with numerous other colleagues, including Balthasar Bickel, Matthew Dryer, Jeff Good, Claire Lefebvre, Mikael Parkvall, and numerous other participants at various conferences and seminars where aspects of this material were presented. The construction of the Association Experiment would not have been possible without the assistance of Yokebed Triwigati, who drew the beautiful pictures, and of the many colleagues and friends who provided me with their expertise on individual languages: Mark Alves (Vietnamese), Margot van den Berg (Sranan), Walter Bisang (Yoruba), Brice Davakan (Fongbe), Robert Early (Bislama), Claire Lefebvre (Fongbe), Robby Morroy (Sranan), Pian (Sundanese), Kofi Saah (Twi), Remi Sonaiya (Yoruba), and Wernher Suares (Papiamentu). The running of the Association Experiment was made much easier by the many people who helped me with obtaining subjects and with other technical matters: L. O. Adewole (Ile Ifẹ), Jonas N. Akpanglo-Nartey (Winneba), Gabriella Hermon (Newark), Tali Konas (Tel Aviv), Ofra Moscovici (Tel Aviv), Omoniyi Emannuel Olanrewaju (Ile Ifẹ), Pian (Rangkas Bitung), Yeshayahu Shen (Tel Aviv), Remi Sonaiya (Ile Ifẹ), and Yusrita Yanti (Padang). Finally, I am hugely grateful to the Department of Linguistics at the Max Planck Institute for Evolutionary Anthropology for providing the infrastructure and support necessary for the conduct of such a large-scale cross-linguistic experimental study.



David Gil

position has been argued for most forcibly by John McWhorter, who, in a series of recent publications (McWhorter 1997, 1998, 2000, and 2001b, c, reproduced in McWhorter 2005), has put forward the claim that all creole languages are simpler than all non-creole languages, or, as suggested in the title of one of his articles, “the world’s simplest grammars are creole grammars”. McWhorter’s claim has triggered a considerable amount of controversy, with counterarguments proceeding in several distinct directions. One common objection calls into question the validity of the notion of complexity as a global property of languages or grammars. Some argue that it simply does not make any sense to try and characterize the overall level of complexity of a language (see Aboh & Ansaldo this volume). At the same time, some hold to the belief that all languages, creoles and otherwise, are of roughly equal overall complexity, that is to say, there are no such things as simpler grammars or more complex ones. In fact, it would seem as though the conventional wisdom is sympathetic to both of the above positions, even though they are mutually contradictory. (If you can’t measure overall complexity, you can’t then say that different grammars have to be of equal complexity.) However, recent years have witnessed an increasing interest in the notion of complexity, as well as a concomitant readiness to entertain the possibility that languages may in fact differ from one another with respect to their overall levels of complexity (Comrie 1992; Romaine 1992; Shosted 2006; and the papers in Karlsson, Miestamo & Sinnemäki to appear). Assuming that it is indeed possible to engage in meaningful cross-linguistic comparisons of linguistic complexity, McWhorter’s claim may then be put to the test of substantive empirical typological investigation. To date, the most extensive cross-linguistic examination of the creoles-are-simpler hypothesis is that of Parkvall (to appear). Drawing upon the typological database of WALS (World Atlas of Language Structures, Haspelmath, Dryer, Gil & Comrie 2005), and making use of 53 linguistic features ranging over all areas of grammar, Parkvall defines a numerical measure of complexity ranging from 0 (maximally simple) to (1 maximally complex), applying it to 153 non-creole languages from WALS, 2 creole languages from WALS, and also an additional 23 creoles and 7 pidgins and expanded pidgins from his own work. According to his calculations, the complexity of the 153 non-creole languages ranges from a minimally complex 0.18 for Pirahã to a maximally complex 0.62 for Burushaski. In contrast, the complexity of the 25 creole languages ranges from a minimally complex 0.08 for Hawaiian Creole to a maximally complex 0.33 for Papiamentu. Based on these figures, Parkvall concludes that the hypothesis is right, that creole grammars are indeed simpler than the grammars of other, non-creole languages. Indeed, examination of Parkvall’s figures show that the middle and upper ranges of grammatical complexity – from 0.34 to 0.62 – are inhabited exclusively by non-creole languages, 122 out of the 153 non-creole languages in the sample. In contrast, however, the lower ranges – from 0.18 to 0.33 – evince a substantial degree of overlap between non-creole and creole languages. In fact, the simplest non-creole language, Pirahã, is less complex than 20 out of the 23 creole languages in Parkvall’s sample,



Creoles, complexity and associational semantics 

while even Standard Indonesian, at 0.26, is less complex than 7 out of the same 23 creole languages. Parkvall’s calculations thus suggest that McWhorter is half right. Specifically, whereas creole languages are, without exception, of a relatively low level of complexity, not all languages with a low level of complexity are creoles: other older languages may also exhibit the simplicity characteristic of all creole languages. In Gil (2001a), I present a detailed contrastive analysis of a non-creole language, Riau Indonesian, and McWhorter’s stock example of a creole language, namely Saramaccan, showing that according to McWhorter’s own criteria, Riau Indonesian is every bit as simple, in fact even simpler, than Saramaccan. On the basis of this comparison, it is argued that his proposed bidirectional universal governing creoles and complexity needs to be replaced by a weaker, unidirectional implicational universal, to the effect that if a language is a creole, it is necessarily simple, but not vice versa. Thus, whereas many non-creole languages are indeed more complex than all creole languages, some non-creole languages, such as Riau Indonesian, may also turn out to be simple. In response to this argument, McWhorter (2001c, 2005: 68–71, to appear) concedes that Riau Indonesian is as simple as Saramaccan, but, in defense of his original claim, suggests that Riau Indonesian is in fact also a creole language. Moreover, in face of the claim, in Gil (2001a), to the effect that other varieties of colloquial Malay/Indonesian, such as Siak Malay, are akin to Riau Indonesian with respect to complexity, McWhorter makes the even more far-reaching claim that many or all colloquial varieties of Malay/Indonesian are creole languages. Given our lack of knowledge concerning the external circumstances surrounding the historical development of most colloquial varieties of Malay/Indonesian, one cannot rule out the possibility that some such varieties might have undergone the kind of radical restructuring that is characteristic of creole languages. In fact, given the importance of Malay as a major language of empire, commerce, literature and religion throughout the preceding 1500 years, it would be rather surprising if no Malay-based creoles had ever arisen throughout the archipelago. Indeed, there are a number of examples of Malay-based creoles, perhaps the most well-known of these being Baba Malay (Pakir 1986; Lim 1988; Ansaldo & Matthews 1999). However, with respect to a majority of contemporary varieties of colloquial Malay/Indonesian, there is, to the best of my knowledge, not a shred of independent historical evidence to support the claim that they are creole languages or the descendants thereof. In particular, given that a location in east-central Sumatra, near where Riau Indonesian and Siak Malay are spoken, is the probable homeland of the Malay language (Tadmor 2002), and the indigenous population throughout the region has been Malay-speaking for the last 1500 years at least, without historical evidence for any kind of massive social disruption, it is hard to imagine what might motivate such speakers to switch in sudden fashion from an ordinary Malay dialect to a Malay-based creole. It must therefore be concluded that colloquial varieties of Malay/Indonesian do indeed provide bona fide examples of non-creole languages that are every bit as simple as typical creole languages.



David Gil

In recent work, McWhorter has been moving toward a more nuanced typology of language contact and concomitant simplification, arguing that a variety of situations of imperfect transmission may give rise to differing degrees of grammatical reduction (2001a, 2006). In particular, he claims that major world languages such as English, Persian and Mandarin, by dint of their adoption and use by diverse populations over broad geographical areas, have undergone a certain amount of simplification in comparison to related languages in the same genealogical families, Germanic, Iranian and Sinitic respectively; for such languages, he introduces the label Non-Hybrid Conventionalized Second-Language Variety, or NCSL. Indeed, as a major world language with a grammatical structure that is simpler than that of its closest Malayo-Polynesian relatives, Malay/Indonesian may also merit the characterization as an NCSL – as indeed is suggested by McWhorter. Where McWhorter goes one step too far, however, is in suggesting that not some but rather all of the simplicity of Malay/Indonesian is due to its extraneous sociolinguistic circumstances. He writes: If I were presented with a language whose history did not involve acquisition being more often by adults outside of a school setting than by children, and this language were nevertheless as underspecified as Riau Indonesian, then I would readily concede that even an older language can attain a level of relative simplicity akin to [a typical creole language] (2005: 71).

But Riau Indonesian is that language, as there is good reason to believe that, for many generations back, it (or its ancestor) was being commonly acquired in a naturalistic home setting by very young children. Moreover, if the role of Riau Indonesian as an interethnic contact language renders it less than optimal as an example of a simple non-creole language, then it can readily be replaced by some of the varieties of Malay spoken in the same region but in different social settings. For example, Siak Malay is grammatically almost as simple as Riau Indonesian, however, sociolinguistically it lies at the other end of the scale, as a small rurally-based language associated with a single ethnic group and used primarily for intraethnic communication. In fact, as suggested in Gil (2001a), the grammatical simplicity of Riau Indonesian is a characteristic feature of the geographical region in which it is located, encompassing parts of Sumatra and adjacent Java. In this region, a similar degree of simplicity is exhibited not only by the local varieties of Malay/Indonesian, but also by other completely different languages, such as Minangkabau and Sundanese, especially in their more colloquial registers. Accordingly, languages such as Minangkabau and Sundanese may provide even clearer examples of run-of-the-mill uncontroversially older languages whose grammars are every bit as simple as those of typical creole languages. This chapter takes some steps toward integrating Minangkabau and Sundanese into the discussion of creoles and complexity. Whereas Gil (2001a) presented a con-



Creoles, complexity and associational semantics 

trastive analysis of Riau Indonesian and Saramaccan encompassing phonology, morphology, syntax and semantics, the present chapter examines a wider range of languages but with respect to a single albeit very significant semantic property. The languages examined are eight uncontroversially older languages, English, Hebrew, Twi, Fongbe, Yoruba, Vietnamese, Minangkabau and Sundanese, and three creoles, Papiamentu, Sranan and Bislama. These languages are examined with respect to their compositional semantics: the rules that govern the ways in which the meanings of complex expressions are derived from the meanings of their constituent parts. Languages may differ with respect to the complexity of their compositional semantics. Of the eight non-creole languages examined, English and Hebrew are the most complex, while Minangkabau and Sundanese are the simplest. But what of the three creole languages, Papiamentu, Sranan and Bislama? If “the world’s simplest grammars are creole grammars”, then they should be even simpler than Minangkabau and Sundanese with respect to their compositional semantics. However, it turns out they are not: their compositional semantics is actually significantly more complex than that of Minangkabau and Sundanese, though for the most part simpler than that of languages such as English and Hebrew. Thus, the results of this chapter provide further support for the arguments presented in Gil (2001a) against the bidirectional correlation between language age and grammatical complexity, by showing that some noncreole languages can be every bit as simple as creole languages, at least with respect to one very important semantic property.

2 Associational semantics At the heart of the compositional semantics of all languages is a basic and very simple rule, that of associational semantics. However, the nature of associational semantics may perhaps be more readily understood with reference to an example drawn not from natural languages but rather from an artificial one, the language of pictograms, those iconic signs that can be found at airports, railway stations, the sides of roads, and other similar locations. Consider the following example:

(1)

The above example is formed by the juxtaposition of two simple signs, each with its own ‘lexical semantics’. The arrow means something like ‘thataway’, ‘over there’, or ‘go in that direction’, while the sign to its right is a clear iconic representation of the meaning ‘bicycle’. So much is straightforward; but what we are interested in here is the meaning of the example as a whole, that is to say, the interpretation of the juxtaposition of the two signs, its compositional semantics.



David Gil

Readers in many European cities will be familiar with the above example as a semi-conventionalized traffic sign designating a special lane for bicycles. However, in other contexts, the above example can also be seen as providing directions to a bicycle shop or exhibition. Thus, the above sign has a broad range of possible interpretations, including, among others, ‘bicycles go thataway’ and ‘go thataway for bicycles’. Now at this point an overzealous grammarian might be tempted to describe the facts in traditional terminology, along the following lines. The first sign is a verb ‘go thataway’, the second sign is a noun ‘bicycle’, and the verb may assign either of two different thematic roles to its nominal argument, resulting in the observed ambiguity: theme for ‘bicycles go thataway’, goal for ‘go thataway for bicycles’. Such a description, though, is an overextension of grammatical analysis to a domain in which it is of little or no relevance. In actual fact, pictograms provide little or no evidence for the existence of parts of speech such as verbs and nouns, and little or no reason to posit grammatical rules that make reference to thematic role assignment. In particular, in the above example, there is little or no evidence to support the existence of ambiguity involving two distinct interpretations. Rather, the construction is more appropriately characterized as vague with respect to its various possible interpretations, possessing, instead, a single general underspecified meaning, along the lines of ‘something to do with thataway and bicycles’. The notion of ‘something to do with’ may be put on a more rigorous footing with the following definition of the Association Operator A:

(2) The Association Operator A: Given a set of n meanings M1... Mn, the Association Operator A derives a meaning A ( M1... Mn ), or ‘entity associated with M1 and... and Mn’.

(The Association Operator is defined and discussed in Gil 2005a, b.) Two subtypes of the Association Operator may be distinguished, the Monadic Association Operator, in which n equals 1, and the Polyadic Association Operator, for n greater than 1. In its monadic variant, the Association Operator is familiar from a wide variety of constructions in probably all languages. Without overt morphosyntactic expression, it is manifested in cases of metonymy such as the often cited The chicken left without paying, where the unfortunate waiter uses the expression the chicken to denote the person who ordered the chicken. Using small upper-case letters to represent the meanings of individual expressions, we can represent the meaning of chicken in the above sentence by means of the Monadic Association Operator as A (chicken), or ‘entity associated with chicken’. The nature of the association between the entity and the chicken is left open by the Association Operator, but may be filled in by the context, which, in the case of a restaurant, is the obvious one involving a dishonest or forgetful customer. In other cases, the Monadic Association Operator is overtly expressed via a specific form, which is commonly referred to as a genitive, possessive or associative marker. Consider, for example, the English possessive enclitic ’s. Application of ’s to John yields the expression John’s, which has the interpretation A (john), or ‘entity associated with John’, where the nature



Creoles, complexity and associational semantics 

of the association is unspecified. Some idea of how unconstrained the association is can be obtained by comparing the obvious meanings of phrases such as John’s father, John’s nose, John’s shirt, John’s birthday, John’s suggestion and so forth, or by considering the range of meanings of a single phrase such as John’s book, which could denote the book that John owns, the book that John wrote, the book that’s about John, or, in more specific contexts, the book that John was assigned to write a review of, and so forth. In its polyadic variant, the Association Operator provides a basic mechanism of compositional semantics in which the meaning of a complex expression is derived from the meanings of its constituent parts. In accordance with the Polyadic Association Operator, whenever two or more expressions group together to form a larger expression, the meaning of the combined expression is associated with, or has to do with, the meanings of each of the individual expressions. Thus, when applying to example (1) above, the Association Operator assigns the juxtaposition of signs the interpretation A (thataway, bicycle), or ‘entity associated with thataway and bicycle’: a unitary underspecified interpretation which, in the right context can be understood as ‘bicycles go thataway’, ‘go thataway for bicycles’, or in any other appropriate way. In general, this is how the compositional semantics of pictograms works: when two or more signs are juxtaposed the Polyadic Association Operator applies to produce a single underspecified interpretation associated in an indeterminate fashion with the interpretations of each of the individual signs. The language of pictograms may thus be characterized as possessing associational semantics. So much for pictograms; but what of natural languages? As argued in Gil (2005a), there is good reason to believe that the Polyadic Association Operator is a universal cognitive mechanism underlying the compositional semantics of all languages. However, whereas with pictograms the Polyadic Association Operator stands alone as the sole means of deriving compositional semantics, in the case of natural languages it provides for a basic and very general interpretation which, in most cases in most languages, is subsequently narrowed down by a variety of more specific semantic rules making reference to any number of morphosyntactic features such as linear order, nominal case marking, verbal agreement, and so forth. Hence, in order to observe the Polyadic Association Operator in action, it is necessary to dig a little harder. Three possible domains in which one can search for associational semantics are phylogenetically, back in evolutionary time; ontogenetically, back in developmental time; and typologically, across the diversity of contemporary adult human languages. Phylogenetically, as argued in Gil (2005a, 2006), the Polyadic Association Operator may be observed to apply in the novel sequences of signs occurring in the artificial languages mastered by captive great apes such as the bonobo Kanzi and the orangutan Chantek; accordingly, the cognitive ability of associational semantics may be reconstructed back to our common ancestor some ten million years ago – long before humans and full-fledged human language had evolved. Ontogenetically, as argued in Gil (2005a), the Polyadic Association Operator may be seen to apply in the utterances of young infants in an early multiple-word stage of first-language acquisition; accordingly, associa-



David Gil

tional semantics may be posited for a developmental stage prior to the acquisition of most of the other grammatical machinery characteristic of most other languages. The evolutionary and developmental facts are both reflected in the architecture of adult human grammar, with the Polyadic Association Operator lying at the core of the compositional semantics. However, whereas for great apes and young infants, associational semantics is all that there is, in adult human grammar the Polyadic Association Operator constitutes the foundation for additional more specific rules of compositional semantics, making reference to a wide variety of morphosyntactic features such as linear order, case marking, agreement, and so on. Although many of these rules might themselves be universal, numerous others are specific to individual languages. Accordingly, languages may differ in the extent to which their compositional semantics makes use of such additional rules. This variation results in a typology of associational semantics. At one end of the scale are compositionally associational languages, in which the number of additional semantic rules is relatively few and their combined effect relatively insignificant, while at the other end of the scale are compositionally articulated languages, in which the number of additional semantic rules is relatively large and their combined import much more substantial. As argued in Gil (2005a, b), Riau Indonesian provides an example of a compositionally associational language. Consider the following simple sentence in Riau Indonesian:

(3) Ayam makan chicken eat

Sentence (3) might be glossed into English as ‘The chicken is eating’; however, such a translation, or for that matter any other translation that one might come up with, cannot but fail to adequately represent the wide range of interpretations that is available for the Riau Indonesian sentence, wider than that of any simple sentence in English. To begin with, the first word, ayam ‘chicken’, is unmarked for number and definiteness, while the second word, makan ‘eat’, is unspecified for tense and aspect. But the semantic indeterminacy is more far reaching yet. The thematic role of ayam ‘chicken’ is also unmarked: in addition to agent, it could also be patient (‘Someone is eating the chicken’); moreover, given an appropriate context, it could assume any other role, for example benefactive (‘Someone is eating for the chicken’), comitative (‘Someone is eating with the chicken’), and so forth. Moreover, the ontological type of the entire construction is itself underspecified: in addition to an event, it could also denote an object (‘The chicken that is eating’), a location (‘Where the chicken is eating’), a time (‘When the chicken is eating’), and so on. Thus, sentence (3) exhibits an extraordinarily large degree of semantic indeterminacy, in which many of the semantic categories encoded in the grammars of most other languages are left underspecified. Rather than being multiply ambiguous, sentence (3) is thus vague, with a single unitary interpretation, most appropriately represented in terms of the Polyadic Association Operator, as A (chicken, eat), or ‘entity associated with chicken and eating’. As suggested by these facts, the compositional semantics of sentence (3) works just like the juxtaposition of signs in



Creoles, complexity and associational semantics 

example (1). Accordingly, Riau Indonesian may be characterized as a compositionally associational language. However, Riau Indonesian differs in an important respect from the language of pictograms: although largely compositionally associational, it still possesses a number of additional grammatical rules whose effect is to constrain the degree of semantic indeterminacy. To cite just one example, in (3) above, although ayam ‘chicken’ may assume any thematic role whatsoever, there is a significant preference for it to be understood as agent (‘The chicken is eating’) as compared to patient (‘Someone is eating the chicken’). In Gil (2005b), facts such as these are accounted for in terms of additional semantic rules making reference to head-modifier structure, iconicity and information flow. It is rules such as these which justify the characterization of Riau Indonesian as a largely rather than purely compositionally associational language. Still, as evidenced by facts such as those in (3) above, the combined effect of these rules is much less far reaching than in other, more compositionally articulated languages. Contrast Riau Indonesian with a more compositionally articulated language such as English. In English, in a sentence such as The chicken is eating, there is obligatory grammatical encoding for all the semantic categories that are left unspecified in Riau Indonesian: number, definiteness, tense, aspect, thematic role, ontological type and others. As a result, its range of possible interpretations is greatly reduced. Even in English, though, associational semantics is clearly in evidence: it is the Polyadic Association Operator that ensures that the interpretation of The chicken is eating has to do with ‘chicken’ and ‘eating’, and that it cannot mean, say, ‘John bought the book’. But in English associational semantics is the foundation for a wealth of additional rules of semantic representation, most or all of which, quite naturally, make reference to the considerably greater morphosyntactic elaboration of English, in comparison to Riau Indonesian. Thus, with respect to their compositional semantics, languages may range from compositionally associational to compositionally articulated, in accordance with the extent to which the Polyadic Association Rule is supplemented with further, more specific rules of semantic interpretation.

3 Associational semantics and complexity Is there any relationship between associational semantics and overall complexity? Clearly, within the domain of compositional semantics, compositionally articulated languages are more complex than their compositionally associational counterparts. But what, if anything, can be said about their overall level of complexity? A widely held position holds that languages with more impoverished morphosyntactic structures compensate with more elaborate rules of pragmatics. In accordance with this view, languages lacking overt expression of various semantic categories force their speakers to fill in the missing pieces, as it were, by appealing to linguistic and extra-linguistic context. For example, in the domain of anaphora and the so-called



David Gil

‘pro-drop parameter’, Ross (1982), Huang (1984) and others distinguish between ‘hot languages’ such as English, in which pronouns are obligatory, and ‘cool languages’ such as Chinese, in which they are optional, and in which, accordingly, speakers have to exercise their pragmatic competence in order to figure out the missing elements. Following such reasoning, the burden is merely shifted from the morphosyntax to the pragmatics, while the overall level of language complexity remains the same. In particular, in the case at hand, compositionally associational languages would make up for their morphosyntactic simplicity with greater pragmatic complexity, and accordingly, their overall level of complexity would end up roughly the same as that of compositionally articulated languages. The trouble with this position is that it presupposes the existence of a universal meaning representational system with a fixed set of semantic features shared by each and every one of the world’s languages, into which every utterance in every language must be decoded, if not by purely grammatical means then with the additional assistance of the pragmatics. However, the existence of such a universal semantics is problematical on both practical and principled grounds. In practical terms, when linguists say things like “the grammar doesn’t encode it, so speakers have to figure it out from the context”, the ‘it’ in question is, more often than not, a distinction that is absent from the target language but one that the linguist expects to find, because it is prominent in English and other European languages. The mistake is to assume that this absence is a lacuna that needs somehow to be filled in by the pragmatics: in most cases, there is no such need whatsoever. Consider, for example, the category of number in Riau Indonesian. As pointed out in (3) above, the word ayam ‘chicken’ is unmarked for number, and can be interpreted as either singular or plural. Does a speaker of Riau Indonesian, every time he or she hears the word ayam, wonder whether it refers to a single chicken or to a plurality of chickens? In a few specific contexts probably yes, in a large majority of everyday situations surely no. Unless of course the speaker is entrusted with the task of translating from Riau Indonesian into a language such as English with an obligatory singular/ plural distinction. To assert that the pragmatics of Riau Indonesian must compensate for the language’s impoverished morphological number marking by forcing the speaker to figure out whether each and every word is meant to be singular or plural is tantamount to imposing an English grammatical category onto the semantics of Riau Indonesian. Analogous observations hold with respect to all of the other categories – definiteness, tense, aspect, thematic role, ontological type, and so forth – that are encoded in the grammars of English and 1. In fact, in some of my earlier writing, I leaned toward a similar position with regard to the compositional semantics of Riau Indonesian, for example in Gil (1994: 195), where, after presenting a precursor to the associational analysis outlined above with reference to example (3), I wrote that “the precise nature of the relationship between the constituent meanings, and who did what to whom, is determined by context and by extra-linguistic knowledge of the world.” As will become evident below, I no longer adhere to this position.



Creoles, complexity and associational semantics 

other European languages but left unspecified in Riau Indonesian. Thus, when linguists say that the pragmatics of a language such as Riau Indonesian must work overtime in order to clear up the vaguenesses or ambiguities left behind by an excessively underdifferentiating grammar, they are lapsing into a prescriptive mode, asserting that all languages have to aim for the precise level of clarity and specificity that just happens to be characteristic of English and other familiar European languages. Quite clearly, the choice of English and other European languages as the basis for the putative set of universal semantic features is arbitrary and unmotivated. Consider, once again, the example of number. In Arabic, in addition to singular and plural there is also a dual number. Now if a Eurocentric linguist can suggest that in Riau Indonesian the pragmatics must disambiguate singular from plural, what is then to stop an Arabic grammarian from proposing that in impoverished English, where dual and plural are conflated, whenever a speaker encounters a plural noun, he or she must stop to figure out whether it is dual or three-or-more? But speakers of Arabic also have a problem, because in Larike (an Austronesian language of Indonesia), in addition to singular, plural and dual there is also a trial number. So would a Larike linguist have speakers of Arabic scrambling to determine whether each three-or-more noun is actually trial or four-or-more? (And in the meantime, our Riau Indonesian speaker is also now struggling to figure out whether ayam is singular, dual, trial, or four-or-more.) Moreover, why stop with trials? If quadrals are of dubious existence, and quintals and beyond still unattested in human languages, who is to say that the concepts that they represent should be excluded from the set of universal semantic features into which linguistic utterances are decoded by means of our grammatical and pragmatic competence? Once again, analogous observations hold with respect to just about any other grammatical category: if there is a language that is less grammatically differentiated than English with respect to the category in question, there is also bound to be one that is more grammatically differentiated than English with respect to it. Hence, a Eurocentric linguist cannot impose English-like semantic features on a language that does not encode such features without being willing to accept a similar imposition of more highly differentiated features from some other language onto the grammar of English. And the result is a potentially infinite regress. What this shows, then, is that there is a serious practical problem with the presupposition of a universal meaning representational system with a fixed set of semantic features shared by each and every one of the world’s languages: we simply do not know what these features are, and in our ignorance, opt for the convenient but unwarranted assumption that they are similar to those of English and other European languages. However, in addition to the above practical problem, there is also a more principled reason to question the presupposition of a universal meaning representational system with a fixed set of semantic features accessed by each and every one of the world’s languages. Quite simply, there is no a priori reason to assume that such a universal semantics should exist; and in the absence of such reason, the most appropriate default assumption is that no such universal semantics does exist. After all, languages



David Gil

differ from each other in just about every other domain, from phonetics through phonology, morphology and syntax to discourse structure. Also, the things that we talk about differ from language to language, be they cultural artefacts or aspects of the physical environment. In view of such variation, it would certainly be very surprising if it turned out that there did exist a semantic representational system and a set of semantic features that was invariant across languages. Of course, there will be lots that is universal in semantics, just as there are universal properties within every other domain of language, as well as commonalities shared by all cultures and physical environments; the point is that in addition to such universals we should expect to find lots of semantic features that one language encodes in its grammar while another ignores, both in its grammar and in its rules of pragmatic inference. Thus, for example, in the absence of positive evidence to the effect that speakers of Riau Indonesian make regular and systematic use of their pragmatic competence to disambiguate singular, plural, dual, trial and so on, considerations of parsimony point to the conclusion that the meaning representational system for Riau Indonesian does not distinguish between categories of number. In everyday parlance, vagueness has mildly negative connotations; however, vagueness is one of the central design features of linguistic semantics, and for good reason. The number of forms, grammatical or lexical, at our disposal, will always be several orders of magnitude fewer than the number of things we can imagine, and then talk about with these forms: plural number will be vague with respect to the distinction between 234 and 235, a noun such as chicken will be vague with respect to dimensions such as weight, height, color, texture of plumage, and so on and so forth. Without such vagueness, language could simply not function. For the most part, we do not notice this vagueness; it generally attracts our attention only in the comparative crosslinguistic context, when we encounter an unfamiliar language in which a distinction that we expect to find turns out not to be there. How can a language ‘do without’ such and such a category, we wonder. But it is important to keep in mind that all languages are massively vague, and that in comparison, the differences between languages such as Riau Indonesian and English are small indeed. Moreover, less vagueness, or greater specificity, is not always of greater functionality. Sometimes it is useful to have the right form at one’s disposal; but in many other cases, obligatory grammatical encoding constitutes an inconvenience: witness the number of times in this section where I have had to refer to the generic speaker with the awkward expression “he or she”, for lack of a gender-neutral third-person singular pronoun in English. To a speaker of Riau Indonesian, or any other language without masculine and feminine pronouns, having to always distinguish between “he” and “she” may seem a strange burden to have to deal with. Vagueness is a necessary feature of language: it is there for a very good purpose, to expedite communication. Thus, it just does not make sense to suppose that whenever a speaker encounters an instance of vagueness, he or she immediately makes use of pragmatic rules in order to fill in an arbitrary large number of additional semantic features. Of course, pragmatics is there to flesh out utterances with



Creoles, complexity and associational semantics 

whatever additional information happens to be relevant in the given context. But it is not there in order to translate all utterances in all languages into some universal meaning representation system with a shared set of semantic features. In particular, it is not reasonable to assume that whenever a speaker of Riau Indonesian encounters a sentence such as Ayam makan in (3) above, he or she automatically embarks on some long and arduous pragmatic path of ‘figuring everything out’: assigning number and definiteness to ayam, tense and aspect to makan, thematic role to ayam, and ontological type to the construction as a whole – essentially translating Ayam makan into English. Rather, the speaker simply fills in what is necessary in the situation at hand, and does not bother with anything else: just like other speakers dealing with other utterances in other languages. In summary, then, there is little reason to believe that languages with more impoverished morphosyntactic structure compensate for their grammatical simplicity with more complex rules of pragmatics. Accordingly, languages with more impoverished morphosyntactic structures provide plausible prima facie candidates for the characterization as simpler languages. In particular, all else being equal, compositionally associational languages are simpler, while compositionally articulated languages are more complex. Thus, the Association Typology provides a window into the cross-linguistic exploration of the notion of linguistic complexity.

4 Measuring complexity: The association experiment In order to compare the relative complexity of creole and non-creole languages, one may thus examine them with respect to the Association Typology. If, as argued by McWhorter, “the world’s simplest grammars are creole grammars”, then as a corollary of this hypothesis we would expect to find that creole languages are more compositionally associational than non-creole languages. Is this indeed the case? Right away, we come up against a methodological stumbling block which might be dubbed ‘the typologists’ dilemma’. Linguist A describes language X as being like this; linguist B has written a grammar of language Y arguing that it is like that; while linguist C insists that language Z is different again. Along comes the typologist wishing to compare languages X, Y and Z – and indeed, based on the reported evidence, they look quite different from each other. But are these differences real differences between the languages, or do they instead reflect different approaches to three similar languages adopted by three different linguists? Perhaps linguist A is a generativist, linguist B an adherent of Basic Linguistic Theory, and linguist C an iconoclastic free thinker; in which case the typologist wishing to compare languages is instead actually comparing linguists or linguistic theories. Ideally, the typologist would go to the field and collect all of the necessary data first-hand, but human life-spans, not to mention research budgets and energy levels, are limited, and this is simply not practical.



David Gil

An alternative approach to typology is provided by the experimental method. Experiments are limited in their efficacy in that they can only address very specific questions; no single experiment or even set of experiments can provide an adequate overall profile of a language. However, the great advantage of experiments lies in their objectivity and uniformity; they provide the typologist with the possibility of looking at a large number of languages through the same eyes, with a reasonable degree of confidence that the cross-linguistic results are indeed comparable, and not subject to the kinds of distortion that are introduced by the mediation of diverse linguists and linguistic theories. The Association Experiment is designed to measure the degree to which languages are associational, that is to say, the extent to which their compositional semantics relies on the Polyadic Association Operator to the exclusion of other more specific rules of semantic interpretation. The Association Experiment focuses on languages with the following two typological properties: (a) isolating; and (b) apparent SVO basic word order. The motivation for this choice of focus is as follows. To begin with, there is a priori reason to believe that, as a linguistic type, isolating languages exhibit greater diversity with respect to the associational typology than do other linguistic types. In agglutinating, synthetic and polysynthetic languages, the elaborate morphology typically supports many additional rules of semantic interpretation on top of the Polyadic Association Rule, resulting in a high degree of articulation of the compositional semantics. In contrast, isolating languages have fewer morphosyntactic devices at their disposal, and are therefore more likely to exhibit a range of values from compositionally articulated to compositionally associational. More specifically, however, the isolating SVO profile has also been attributed to creole languages, and in addition it is characteristic of Riau Indonesian and other closely related languages such as Minangkabau and Sundanese. Thus, it provides an appropriate backdrop for the comparison of creoles with other typologically similar non-creole languages, which is the main goal of this chapter. Any attempt to measure, by means of objective experimentation, the extent to which a language is compositionally associational faces an immediate problem: how do we know which rules are responsible for a given construction being interpreted in a certain way if not through the mediation of linguistic analysis conducted by a linguist within a particular theoretical framework? The answer of course is that we don’t, at least not for sure. However, at the level of prima facie plausibility, it seems possible to distinguish between interpretations probably attributable to the application of various 2. These two properties are related by an implicational tendency, to the effect that isolating languages are usually SVO, but not vice versa; see, for example, Siewierska and Bakker (1996). The qualification of “SVO” with “apparent” is intended to avoid the presupposition that all of the languages examined have well defined categories of subject, verb and object, taking cognizance of the fact that at least one language appearing to have SVO basic word order has been argued to be lacking in subjects, verbs and objects, see Gil (1994, 2000, 2001b, 2002) on Riau Indonesian.



Creoles, complexity and associational semantics 

specific rules of semantic interpretation making reference to morphosyntactic features such as linear order, case marking, agreement and the like, as opposed to interpretations which may more plausibly be attributed directly to the Polyadic Association Operator. Interpretations of the latter kind are referred to here as Apparently Associational Interpretations. The goal of the Association Experiment is thus to compare the availability across languages of such Apparently Associational Interpretations.

4.1

Experimental design

The Association Experiment is a truth conditional experiment, in which subjects are asked to judge whether certain states of affairs constitute possible interpretations of particular sentences. Each stimulus presents a written sentence in the target language, beneath which are two pictures. Subjects are asked to read the sentence and then look at the two pictures, after which they must respond to the question: Which of the two pictures is correctly described by the sentence? Subjects may provide any of four different responses: “the picture on the left”, “the picture on the right”, “both pictures”, or “neither picture”. In each stimulus, one of the two pictures is a designated test picture, whose availability as a possible interpretation for the given sentence is what is being tested by the experiment, while the other picture is an alternative picture whose availability as a possible interpretation is, for the most part, not of interest. The Association Experiment consists of 4 training stimuli followed by 40 actual experimental stimuli. Of these 40 experimental stimuli, 34 are real test stimuli, while the remaining 6 are distractors, designed to identify subjects whose responses are not reasoned and systematic, because they have not properly understood the experimental task, or for any other reason. The distractor items are stimuli for which there is only one sensible response. Of the 6 distractors, one was designed to elicit the response “picture on the left”, one to elicit “picture on the right”, two to elicit “both pictures” and two to elicit “neither picture”. However, it turned out that a significant proportion of otherwise clearly competent subjects failed to respond as intended to the two distractors designed to elicit the response “both pictures”. Accordingly, these two distractors were ignored, while the remaining four were employed as originally intended. Any subject that failed to respond appropriately to one or more of the four remaining distractors was discarded, while only subjects that responded appropriately to all four were taken into consideration. In general, depending on the population being examined, between 90% and 95% of experimental subjects performed appropriately on the

3. This was due in part to some accidental properties of the actual pictures that were used, and in part to a general tendency for subjects to treat the experimental task as a forced-choice task, even though it was not presented as such: subjects consistently preferred to choose exactly one of the two pictures, rather than both or neither, even when other semantic considerations might have led to a different response.



David Gil

four distractor stimuli, permitting their responses to be taken into account in subsequent analysis. Of the 34 test stimuli, two are designed to test the subjects’ ability to accept interpretations that are semantically anomalous: a banana running, and a woman eating a car. These are included because many of the alternative pictures that are paired with the test pictures are of a similar anomalous nature, and it is of interest to ascertain whether subjects exhibit any reluctance to accept anomalous interpretations if and when these are suggested by the stimulus sentences. In general, subjects exhibited little reluctance to accept such anomalous interpretations. These two stimuli are not discussed any further here. Remaining, then, are 32 test stimuli designed to examine the availability of Apparently Associational Interpretations. These 32 stimuli consist of two parallel series of 16 stimuli, series A and series B, with similar sentences and similar test pictures. Where they differ is with respect to the alternative picture. Whereas in series A, the alternative picture represents an interpretation that is completely impossible, involving participants that are not referred to in the stimulus sentence, in series B, the alternative picture represents an interpretation that is more readily available than the test interpretation. In principle, an ‘ideal’ subject might be expected to respond identically to similar test pictures in either series, irrespective of the different alternative pictures. After all, if the test picture constitutes an available interpretation of the sentence, then the subject should choose it in both series: on its own in series A, together with the alternative picture in series B. Conversely, if the test picture does not constitute an available interpretation of the sentence, then the subject should reject it in both series: rejecting both pictures in series A, choosing the alternative picture in series B. However, subjects do not behave in this ‘ideal’ way. Consistently, and in overwhelming numbers, when presented with a test picture that constitutes an available interpretation of the sentence, subjects accept it when presented with an impossible alternative picture in series A, but reject it when presented with a more highly preferred alternative picture in series B. Again, the reason for this behavior is the general tendency of subjects to select one of the two pictures rather than both of them, even when both should be available choices. Thus, series A and B present strikingly different pictures of the availability of Apparently Associational Interpretations, with series A exhibiting much higher rates of availability than series B. In fact, in series B, the rates of availability of Apparently Associational Interpretations are so uniformly low that many of the interesting and important patterns of variation, both between constructions within languages, and for the same constructions across languages, are neutralized. Accordingly, series A provides a much better picture of the relative availability of Apparently Associational Interpretations across constructions and across languages. For this reason, then, as well as for reasons of space, in this chapter we shall be concerned only with the 16 test stimuli of series A. A schematic representation of the 16 series A test stimuli is given in Table 1.



Creoles, complexity and associational semantics 

Table 1.  Test Sentences (Series A) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

clown drink book Clown drinking while reading book clown buy happy Clown buying fruit with happy face clown eat river Clown eating by river soldier cut axe Soldier cutting wood with axe coffee laugh Person spilling coffee, onlooker laughing money happy Man holding money with happy face table dance People dancing on tables chair jump People jumping over chairs bird eat Cat eating bird tiger afraid People fearing tigers dog draw Man drawing dog hat sew Men sewing hats mouse chase cat Cat chasing mouse mouse bite snake Snake biting mouse car push woman Woman pushing car elephant look tiger Tiger looking at elephant

Bare Peripheral Following

Bare Peripheral Preceding

Bare Patient Preceding

Bare Patient Preceding plus Agent

In Table 1, each test stimulus is described in terms of the written sentence, represented schematically in small caps, and the test picture, for which a simple verbal description is given. (For reasons of space, the alternative picture is not represented.) The schematic small-cap representation of the written sentence shows the expressions denoting basic concepts belonging to the major ontological categories of thing, property and activity, in the order in which they occur in the sentence. For the most part, it thus provides a representation of the content words common to all of the languages examined, to the exclusion of all of the additional grammatical items, lexical and morphological, which vary from language to language. For example, in stimulus 1,



David Gil

clown drink book represents, among others, the English test sentence The clown is drinking the book, and the Minangkabau test sentence Badut minum buku. Whereas for English, clown drink book leaves out the definite article the and the combination of auxiliary is and gerundive suffix -ing present in the test sentence, for a strongly isolating language such as Minangkabau, the same schematic representation happens to constitute a perfect interlinear gloss for the test sentence, consisting as it does of three content words with no additional grammatical marking. In a few cases, however, expressions denoting basic concepts may consist of more than a single content word. For example, in stimulus 1, clown drink book also represents the Bislama test sentence Fani man i dring buk, in which the basic concept clown is expressed by means of a compound consisting of two content words fani ‘funny’ and man ‘man’. The 16 test sentences for each of the eleven languages are given in the Appendix, in Tables A1 to A11. For all of the languages but Hebrew, the sentences are shown in the conventional orthography in which they were presented to the subjects. (For Hebrew, they are shown in a phonemic transcription, whereas they were presented to the subjects in Hebrew script.) Beneath each test sentence in Tables A1–A11, an interlinear gloss is provided. In each of the languages examined, the test sentences are the simplest naturalsounding and stylistically-neutral grammatical sentences allowed for by the language, in accordance with the schematic representations in Table 1 above. As evident from Tables A1-A11, the languages examined differ considerably with respect to the degree to which the shared content words must be supported by additional grammatical markers. For example, the marking of definiteness (or specificity) is obligatory in English, Hebrew, Papiamentu, Sranan, Twi and Fongbe; accordingly, for these languages all of the referential expressions in the test sentences are marked with a definite article (or an article marking specificity). However, in Bislama, Yoruba, Vietnamese, Minangkabau and Sundanese, such marking is optional; hence, in these languages, no articles are present in the test sentences. Similarly, the marking of tense-aspect is obligatory in English, Hebrew, Papiamentu, Sranan, Twi and Yoruba; accordingly, in these languages, some kind of present and/ or progressive maker is present in the test sentences, as appropriate in the context of describing a picture portraying an activity. In contrast, such marking is optional in Bislama, Vietnamese, Minangkabau and Sundanese; hence,

4. For English and Hebrew, the test sentences were constructed by myself. For the remaining languages, the test sentences were constructed by various combinations of myself, linguists familiar with the languages, and native speakers, and were then doublechecked with native speakers before the experiment began.



Creoles, complexity and associational semantics 

in these languages, tense and aspect marking is absent from the test sentences. Thus, given the different grammatical properties of the respective languages, the test sentences are by no means structural equivalents of one another. Rather, the test sentences in different languages might best be thought of as constituting ‘content-word equivalents’ of each other, sentences making use of similar words arranged in superficially similar patterns. As to whether they are semantically equivalent: this is precisely what the Association Experiment is constructed to find out. As indicated in the rightmost column of Table 1, the 16 stimuli divide into two groups of 8, each of which in turn divides into two groups of 4. The first group of 8 stimuli test for Apparently Associational Interpretations in constructions of the Bare Peripheral type. Bare Peripheral constructions involve the juxtaposition of an expression denoting an activity and another expression denoting an entity which cannot be construed as filling a slot in the semantic frame of the former, activity expression. The term ‘peripheral’ is thus intended as an antonym of the term ‘core’ as in ‘core argument’, while the term ‘bare’ takes cognizance of the fact that the peripheral expression bears no marking to indicate the semantic nature of its relationship to the activity. Consider, for example, stimulus 1, and, in particular, the collocation drink book. As inanimate but solid objects, books have no place in the semantic frame of drinking; accordingly, book is characterized as peripheral. In a compositionally articulated language with, say, rigid SVO order, subjects should judge the sequence clown drink book as semantically anomalous, since it requires an interpretation in which book is the patient of drink; accordingly, they should reject the test picture of a clown drinking while reading a book. In contrast, in a compositionally associational language, subjects should be able to interpret the sequence clown drink book as A (clown drink book), that is to say, as meaning essentially anything involving a clown, a drinking, and a book, and in such a case, they should accept the test picture of a clown drinking while reading a book as a possible interpretation of the sentence. Thus, stimuli 1–8 test for Apparently Associational Interpretations of the Bare Peripheral type. However, they do so in a number of distinct but related ways. Whereas in stimuli 1–4 the bare peripheral expression follows the activity expression, in stimuli 5–8 the bare peripheral expression precedes it. A further, somewhat more subtle distinction can be made with respect to the nature of the semantic relationship between the bare peripheral expression and the activity expression. In general, the se5. A somewhat special case is presented by Fongbe, as in this language, tense and aspect are coded by, among other things, an alternation between VO and OV word order: whereas the present progressive makes use of a participial construction with OV order, a simple VO order without overt marking of tense-aspect is generally assigned a past-tense interpretation (Lefebvre & Brousseau 2002), or one that involves perfective aspect (Enoch Aboh p.c.). In order to make the Fongbe test sentences comparable to those in other languages, no marking of tense and aspect was included; even though some of the test sentences might have thus had a past or perfective interpetation, subjects did not seem to have any trouble relating them to test pictures portraying activities in the here and now.



David Gil

mantic relationship is looser in stimuli 1, 2, 5 and 6 than it is in stimuli 3, 4, 7 and 8. Whereas in the former cases, the semantic relationship can only be characterized as associational, in the latter cases, the semantic relationship is of a type that is typically referred to as a subcategory of oblique, either locative or instrumental. The second group of 8 stimuli test for Apparently Associational Interpretations in constructions of the Bare Patient Preceding type. Bare Patient Preceding constructions involve an expression denoting an activity preceded by an expression denoting an entity understood as the patient of that activity. Consider, for example, stimulus 9. In a compositionally articulated language with rigid SVO order, subjects should interpret the sequence bird eat as entailing that bird is the agent of eat; accordingly, they should reject the test picture of a cat eating a bird, since the bird is assigned the wrong thematic role, that of patient. In contrast, in a compositionally associational language, subjects should be able to interpret the sequence bird eat as A (bird eat), that is to say, as meaning essentially anything involving a bird and an eating, hence, in particular, they should accept the test picture of a cat eating a bird as a possible interpretation of the sentence. Thus, stimuli 9–16 test for Apparently Associational Interpretations of the Bare Patient Preceding type. In terms of S, V and O, these stimuli thus test for OV order (in languages that are supposed to be of SVO basic word order). Stimuli 9–16 also fall into two distinct subgroups. Whereas stimuli 9–12 contain constructions involving only a patient expression followed by an activity expression, stimuli 13–16 contain constructions in which a patient expression and an activity expression are in turn followed by an expression referring to the agent of the activity.

4.2

Running the experiment

The Association Experiment is conducted on a laptop computer, making use of a database in FileMaker Pro. Subjects are tested individually, one after another. Subjects are presented with the stimuli one at a time, each stimulus occupying the entirety of the computer screen. Different subjects receive different randomized orders of the various stimuli. Each response, either verbal, such as “the one on the left”, or gestural, such as pointing to the chosen picture, is recorded immediately by the experimenter, by clicking on a button located beneath the picture chosen, or, in the case of “both” or “neither” responses, between the two pictures. Whereas in some situations, I was able to conduct the experiment on my own, in other cases I was helped by a local assistant, whose role was to round up subjects and/ or help me communicate with them. The Association Experiment is an ongoing project, still in progress; at the time of writing, over 1,200 subjects had been tested in around 20 languages. The general aim is to examine a wide range of languages in combination with a set of sociolinguistic variables: age, level of education, place of residence, place in which the experiment is conducted, and others.



Creoles, complexity and associational semantics 

Often, psycholinguistic studies compare exotic tribes people in their villages to American students on college campuses, as a result of which it is not always possible to tell whether observed patterns of variation are due to differences in the respective languages or in the experimental situations and the sociolinguistic circumstances of the subjects. In order to facilitate a fair cross-linguistic comparison unaffected by extraneous sociolinguistic factors, a Baseline Sociolinguistic Profile is defined, consisting of subjects who are uneducated, low-to-middle class, over 12 years of age, living in a community where the test language is spoken, and tested in their home community, in a public or semi-public area. For each language examined, the experiment is run on a population of subjects conforming to this Baseline Sociolinguistic Profile. In addition, within individual languages, the experiment is also run on additional populations outside the Baseline Sociolinguistic Profile. Thus, for all languages where this is feasible, the experiment is run on a population of university students. And for selected languages, the experiment is also run on other populations: children, migrants living in other regions and countries, and subjects in different test situations. This chapter represents an initial report on a limited subset of the experimental results obtained so far. Eleven languages are reported on: English, Hebrew, Papiamentu, Sranan, Bislama, Twi, Fongbe, Yoruba, Vietnamese, Minangkabau and Sundanese. For each language, data is presented from a population of subjects conforming to the Baseline Sociolinguistic Profile.

6. For English, the subjects were people in a shopping mall and a farmers market in Newark, Delaware, USA. For Hebrew, the subjects were people in a shopping mall and in cafés in low-tomiddle-class neighborhoods of Tel Aviv, Israel. For Papiamentu, the subjects were mostly people in a McDonald’s restaurant in Willemstad, Curaçao, Netherlands Antilles. For Sranan, the subjects were people in a food court, clerical staff in an office, and people in a residential neighborhood in Paramaribo, Surinam. For Bislama, the subjects were street vendors and passers-by in an open-air restaurant in Port Vila, and people hanging out at the main village store in Lakatoro on the island of Malekula, both in Vanuatu. For Twi, the subjects were passers-by in the street outside my hotel in Kumasi, Ghana. For Fongbe, the subjects were vendors and other workers at the voodoo market in Lomé, Togo. For Yoruba, the subjects were people in a squatter area adjacent to the university in Ile Ifẹ, Nigeria. For Vietnamese, the subjects were strollers around Hoàn Kiếm lake in the centre of Hà Nội, and people in open-air restaurants in Sầm Sơn, Thanh Hóa, both in Vietnam. For Minangkabau, the subjects were mostly people in roadside cafés and footstalls in Padang, West Sumatra, Indonesia. And for Sundanese, the subjects were people in a foodstall outside a railway station and friends of a friend in Rangkas Bitung, Banten, Indonesia. Given the diverse nature of the societies in which the experiment is conducted, with different levels of development, degrees of education and affluence, and amounts of internal diversity and inequality, it is obviously not possible to come up with populations that are completely equivalent from a sociolinguistic point of view. The above test populations represent what was probably the nearest that one could reasonably get to a shared Baseline Sociolinguistic Profile.



David Gil

5 Results The results of the experiment for the eleven test languages considered in this paper, conducted on populations conforming to the Baseline Sociolinguistic Profile, are presented in Table 2. Table 2.  Results of the Association Experiment (Baseline Sociolinguistic Profile) Language Group

Non-Isolating Creole

West African

Southeast Asian

Language

English Hebrew Papiamentu Sranan Bislama Twi Fongbe Yoruba Vietnamese Minangkabau Sundanese

Number of Subjects

32 30 25 23 17 21 26 31 38 30 35

Availability (%) of Apparently Associational Interpretations Bare Peripheral

Bare Patient Preceding

7 10 14 29 51 22 56 68 67 74 76

4 4 5 12 15 12 19 26 15 57 49

In Table 2, the first two columns specify the language and the group to which it belongs. The third column shows the number of subjects conforming to the Baseline Sociolinguistic Profile who performed appropriately on the distractor items, and whose responses are therefore taken into account. The fourth and five columns present the results of the experiment: the availability, in percentages, of Apparently Associational Interpretations, averaging over the two major types of constructions, Bare Peripheral (stimuli 1–8) and Bare Patient Preceding (stimuli 9–16). The results in Table 2 show that for both of the non-isolating languages, English and Hebrew, the availability of Apparently Associational Interpretations is, as expected, very low, in the 0–10% range. The results of the experiment accordingly support the characterization of English and Hebrew as compositionally articulated languages.

7. For the purposes of the above calculations, a response is taken to reflect the availability of an Apparently Associational Interpretation in either of two cases: if the subject chose the test picture, or if the subject chose both the test and the alternative picture. Due to subjects’ general preference to choose exactly one of the two pictures, the first case is overwhelmingly more frequent than the second.



Creoles, complexity and associational semantics 

Thus, English and Hebrew provide a base point with respect to which the nine remaining languages, all of an isolating type, may be compared. Looking at the remaining languages, one is immediately struck by the large amount of variation between them. With respect to the availability of Bare Peripherals, the figures range from 14% in Papiamentu to 76% in Sundanese, while with regard to the availability of Bare Patients Preceding, the figures range from 5% in Papiamentu to 57% in Minangkabau. Thus, in terms of the availability of Apparently Associational Interpretations, isolating languages are clearly not of a single type. Of the languages examined, Minangkabau and Sundanese emerge as the most compositionally associational. Among the nine isolating languages, a consistent difference is evident in the relative availability of Apparently Associational Interpretations for the two construction types, with Bare Peripherals scoring significantly higher than Bare Patients Preceding in all languages. What this suggests is that in isolating languages, the domain of core arguments is more likely than other more peripheral domains to exhibit tighter grammatical structure, resulting in less flexible word order, and thereby setting the stage for construction-specific rules of compositional semantics making reference to linear order. In other words, core arguments tend to be more articulated, while peripheral expressions provide a more congenial environment for associational semantics. But what of the three creole languages, Papiamentu, Sranan and Bislama? To begin with, even these three languages exhibit a significant amount of variation, with Papiamentu displaying the lowest degree of availability of Apparently Associational Interpretations, Sranan occupying an intermediate position, and Bislama exhibiting the highest rate of availability of Apparently Associational Interpretations. At least Sranan and Bislama are significantly more compositionally associational than the baseline non-isolating languages English and Hebrew. However, in comparison with the non-creole isolating languages of West Africa and Southeast Asia, the three creole languages, Papiamentu, Sranan and Bislama, exhibit a relatively low degree of availability of Apparently Associational Interpretations. Thus, Papiamentu is less associational than all six non-creole languages, Sranan is in the same ballpark as Twi but less associational than the remaining five non-creole languages, while Bislama is almost as associational as Fongbe but less associational than Yoruba, Vietnamese, Minangkabau and Sundanese. In summary, all three creole languages, Papiamentu, Sranan and Bislama, are significantly less associational than the two non-creole languages of western Indonesia, Minangkabau and Sundanese. Thus, the results of the Association Experiment show that with respect to their compositional semantics, Papiamentu, Sranan and Bislama are substantially more complex than at least some non-creole languages, such as Minangkabau and Sunda8. Since the results reported on in this chapter represent work in progress, the figures have not yet been submitted to rigorous statistical analysis. Impressionistically, however, most or all of the generalizations discussed herein are so robust that it is very unlikely that they will not prove to be statistically significant.

 David Gil

nese. Admittedly, compositional semantics is just one aspect of language, but it is a very important one. Accordingly, the relative complexity of Papiamentu, Sranan and Bislama in this domain, in contrast to other older languages, casts further doubt on the claim that “the world’s simplest grammars are creole grammars”. In particular, the compositionally associational character of Minangkabau and Sundanese suggests that some older languages may be even simpler than many creole languages. The experimental results also make it possible to compare one particular creole language, Sranan, with both its main lexifier language, English, and its proposed West African substrate language Fongbe. The results show that with respect to the availability of Apparently Associational Interpretations, Sranan lies smack in the middle between English and Fongbe. If, in accordance with the relexification hypothesis (Lefebvre 1998), Sranan is considered to be relexified Fongbe, then since relexification it must have become less associational, moving away from Fongbe and toward English. Conversely, if, in accordance with a superstratist perspective (Chaudenson 1979, 1992; Mufwene 1997), Sranan is considered to be a direct descendant of English, then the results suggest that contact with Fongbe must have brought about an increase in the degree of associationality. Whatever the scenario, however, the fact remains that with respect to the rate of availability of Apparently Associational Interpretations, and hence with respect to one important aspect of complexity, Sranan occupies an intermediate position between its lexifier and substrate languages. Again, facts like these suggest that creole languages do not constitute a synchronic typological class to the exclusion of all older languages.

6 Further questions: Why languages vary and why languages ‘undress’ The results of the Association Experiment raise several further questions regarding the relationship between creoles and complexity. To begin, one may wonder why the three creole languages examined, Papiamentu, Sranan and Bislama, differ from each other in the way that they do with respect to the degree of availability of Apparently Associational Interpretations. The different properties of the respective substrate languges do not seem to shed much light on this question. For Papiamentu and Sranan the substrate languages are West African, exhibiting, at least in the case of Fongbe and Yoruba, a substantial degree of availability of Apparently Associational Interpretations. In contrast, for Bislama, the substrate languages are Oceanic, and at an earlier historical stage possibly also Australian – languages which are characterized by richer morphology and presumably also a lower degree of availability of Apparently Associational Interpretations. Thus, if substrate languages had an effect on the differences between Papiamentu, Sranan and Bislama, the prediction would be that Apparently Associational Interpretations should be more readily available in the two Atlantic creoles than in the one Pacific one. However, as shown by the results of the Association Experiment, the opposite is the case: Appar-



Creoles, complexity and associational semantics 

ently Associational Interpretations are more readily available in Bislama than in Papiamentu and Sranan. A more promising account for the differences between the three creole languages derives from consideration of the superstrate languages and their relationships to the respective creoles. Sociolinguistically, there would seem to be an inverse correlation between the amount of ongoing contact present between the creole and its lexifier language and the degree of availability of Apparently Associational Interpretations. In Curaçao, where Papiamentu has the lowest degree of availability of Apparently Associational Interpretations, Spanish has been exerting an ever-increasing influence, through commercial contacts with the nearby South-American mainland, missionary activity in Spanish, and mixed marriages with speakers of Spanish, all resulting in increasing bilingualism (Maurer 1986). In contrast, in Surinam, where Sranan has a higher degree of availability of Apparently Associational Interpretations, English has been largely absent from the sociolinguistic landscape for the last 300 years, with Sranan fulfilling an increasing range of functions, and Dutch assuming the role of the acrolectal language (Holm 1989: 436, citing unpublished papers by Eersel, Essed & Westmaas). And in Vanuatu, where Bislama has an even higher degree of availability of Apparently Associational Interpretations, English may still be present alongside French, as one of the official languages, however, its functions have been steadily decreasing, even as the use of Bislama, now the national language, has spread into more and more domains (Charpentier & Tryon 1982). Thus, the more extensive the contact between the lexifier language and the corresponding creole, the more likely it is that the lexifier language will impose its compositionally articulated semantics on the creole language, thereby decreasing the degree of availability of Apparently Associational Interpretations. Two plausible scenarios present themselves for such a process of imposition. The more straightforward scenario suggests that the degree of associationality can be transferred directly from source to donor languages. Bilingual speakers notice that in the lexifier language peripheral arguments cannot occur without prepositions, and objects cannot occur directly in front of their verbs, and so they copy this feature over from the lexifier language to the creole. However, a more complex scenario suggests that this transfer does not occur directly, but rather through the mediation of a more general linguistic property, namely the overall complexity of the grammar and semantics. As suggested in this paper, the availability of Apparently Associational Interpretations provides one important measure of the overall complexity of a language. However, there is more to a language than its compositional semantics, and no logical necessity that complexity with respect to compositional semantics be correlated with complexity in other grammatical domains. Nevertheless, such a correlation may indeed exist. Specifically, languages with a low rate of availability of Apparently Associational Interpretations tend to exhibit greater complexity across the board, even in domains that have no obvious bearing on the availability of Apparently Associational



David Gil

Interpretations. In other words, compositionally articulated languages tend to be of greater overall complexity than compositionally associational languages. Comparing Papiamentu, Sranan and Bislama provides strong support for this correlation. One indication of grammatical complexity, mentioned in Section 1, is the numerical measure defined by Parkvall (to appear). Parkvall’s figures for the three creole languages, Papiamentu 0.33, Sranan 0.24 and Bislama 0.13, show that decreasing grammatical complexity, as he defines it, provides an accurate predicator for increasing availability of Apparently Associational Interpretations. Whereas Parkvall’s measure encompasses all areas of grammar, from phonology through morphology and syntax to semantics, an alternative measure, the Articulation Index, is limited in its scope to the domain of semantics. The Articulation Index provides a measure of the semantic resolution of a language: how finely it carves up the semantic pie, how many semantic distinctions are encoded in the lexicon and grammar. The Articulation Index of a language is calculated on a sample text by counting the number of morphemes m occurring in the text, the number of basic concepts c belonging to major ontological categories, thing, property and activity, denoted in the text, as per the small-caps representation in Table 1, and then taking the ratio of these two numbers, m/c. The Articulation Index thus provides a measure of all of those additional semantic features, number, definiteness, tense, aspect, thematic role, and numerous others, that languages may choose to express with overt forms: separate words, clitics or affixes. As an illustration of how the Articulation Index is calculated, let us take, as a sample text, stimulus sentence 1, clown drink book. The number of basic concepts denoted is 3. In Papiamentu, the sentence E payaso ta bebe e buki contains 6 morphemes, hence the resulting Articulation Index is 6/3, or 2.00. In contrast, in Bislama, the sentence Fani man i dring buk contains 5 morphemes, and therefore the resulting Articulation Index is 5/3, or 1.67. As suggested by the above examples, the Articulation Index provides a measure of the extra grammatical baggage that a language always carries around with it, such as the obligatory articles and tense-aspect markings of Papiamentu, and the obligatory predicate marker of Bislama. As shown by the little calculation above, the Articulation Index of Papiamentu, at 2.00, is higher than that of Bislama, at 1.67. In work in progress, the Articulation Index is calculated by taking all 32 stimulus sentences as the sample text. In this calculation, the three creole languages emerge with the following values: Papiamentu 2.14, Sranan 2.00, and Bislama 1.74. Like Parkvall’s measure, the Articulation Index of these three languages thus provides an accurate predicator for the availability of Apparently Associational Interpretations:

9. However, the articulation index does not take into account the expresssion of semantic features by means of configurations of forms, such as, for example, linear order or c-command relations. And of course, it does not take into consideration the expression of semantic features by ‘covert’ forms, posited within various approaches such as generative grammar.



Creoles, complexity and associational semantics 

Papiamentu, with the highest Articulation Index, has the lowest degree of availability, while Bislama, with the lowest Articulation Index, has the highest. It is worthy of emphasis that the Articulation Index and the Association Typology are for the most part logically independent properties of languages. Admittedly, if one of the creole languages had nominal case marking, this would increase the Articulation Index while concomitantly decreasing the availability of Apparently Associational Interpretations. However, most of the grammatical markers present in Papiamentu, Sranan and Bislama are not like case marking; they have no inherent logical bearing on the availability of Apparently Associational Interpretations. Thus, the obligatory nominal articles in Papiamentu are not marked for case, and therefore should not have any effect on the assignment of thematic roles. But as suggested by the correlation between the Articulation Index and the Association Typology, their presence does indeed constrain the assignment of thematic roles, reducing the availability of Apparently Associational Interpretations. Work in progress suggests that the Articulation Index and the Association Typology constitute two manifestations of a single deeper property of languages, namely their overall level of semantic resolution, or overall articulation. Highly articulated languages have a high Articulation Index and a highly articulated compositional semantics, while languages of low articulation have a low Articulation Index and a highly associational compositional semantics. Overall articulation may thus be viewed as a broad and general reflection of the level of complexity in a language’s semantics. In this context, then, the varying degrees of availability of Apparently Associational Interpretations in Papiamentu, Sranan and Bislama may be viewed as consequences of their respective typological profiles and relative degrees of overall articulation. Of course, their different typological profiles may themselves be consequences of their different histories and present-day circumstances. As is often the case, the sociolinguistic and the typological explanations for the varying degrees of availability of Apparently Associational Interpretations in Papiamentu, Sranan and Bislama may both be valid. The notion of overall articulation as a typological property is of course relevant also for the non-creole languages to which Papiamentu, Sranan and Bislama are compared in this chapter. Whereas languages such as English clearly exhibit a higher level of overall articulation than any creole language, languages such as Minangkabau and Sundanese just as clearly exhibit a lower degree of overall articulation. This is vividly reflected in the Articulation Index. Taking stimulus sentence 1 once again as a sample text, in Minangkabau, Badut minum buku contains just 3 morphemes, ‘clown’, ‘drink’ and ‘book’; hence, the resulting Articulation Index is 3/3, or 1.00. Stimulus sentence 1 in Minangkabau thus yields the ideal maximally simple limiting value of the Articulation Index, 1.00, obtained in the absence of any grammatical elaboration, any lexical or morphological expression whatsoever for categories such as number, definiteness, tense, aspect, thematic roles and the like. Not all sentences in Minangkabau and Sundanese are quite like that; however, examination of any natural-



David Gil

istically occurring corpus will come up with lots that are, and lots more that come close. Using, as before, the 32 stimulus sentences as the sample text, the values of the Articulation Index that emerge are 1.02 for Minangkabau and 1.05 for Sundanese – clearly lower than those for Papiamentu, Sranan and Bislama. Thus, the Articulation Index joins force with the availability of Apparently Associational Interpretations in supporting the characterization of Minangkabau and Sundanese as exhibiting a lower degree of overall articulation than Papiamentu, Sranan and Bislama. Still, overall articulation is a semantic property, while the claim that “the world’s simplest grammars are creole grammars” pertains to all domains of grammar; so perhaps Minangkabau and Sundanese are endowed with greater complexity in other areas. This possible objection is countered in Gil (to appear). Building on Gil (2001a), which presents a contrastive analysis of Saramaccan and Riau Indonesian, Gil (to appear) provides a sequel contrastive analysis of Riau Indonesian and two other languages spoken in central Sumatra: Siak Malay and Minangkabau. Although there is no reason to believe that Riau Indonesian is a creole language or has undergone any kind of radical restructuring due to imperfect second-language acquisition by adults, its functions as a koiné and contact language render it suspicious to some; accordingly, Siak Malay and Minangkabau are chosen as closely related languages with impeccable sociolinguistic credentials, as ‘in-group’ languages identified with specific ethnicities and cultures, that is to say, as uncontroversially old languages. Nevertheless, as shown in Gil (to appear), Siak Malay and Minangkabau are typologically very similar to Riau Indonesian, and pretty much as simple, in all domains of grammar. Accordingly, they, too, are every bit as simple as Saramaccan and other creole languages. McWhorter (to appear) is justifiably concerned with the diachronic implications of such claims. “Why would a language undress?”, he asks. Quite convincingly, he shows that in a wide variety of cases, grammatical simplicity is a natural outcome of sociolinguistic circumstances, and in particular, a history of extensive second-language acquisition by adults. The problem arises when he attempts to make the leap from many or even most languages to all. In support, McWhorter appeals to general scientific methodology, proposing an analogy. With reference to the highly isolating languages of Flores about whose history almost nothing is known, he writes: “As geologists treat cracked quartz as a sign of volcanic eruptions in the past, linguists might treat the strange simplicity of Keo, Rongga, and Ngadha as evidence of social disruption in the past”. However, as suggested in this paper, it is not just a handful of languages in Flores that are problematical to his thesis. Moreover, the analogy to geology fails on more principled grounds. As an essentially observational field, geology is epistemologically rooted in the more theoretical discipline of material science, where, crucially, experiments can be conducted and then replicated. Thus, material scientists can subject quartz to extreme conditions in the laboratory and observe it crack, formulate generalizations about when it will and will not crack, and then hand these generalizations back to the geologist. However, historical linguists cannot conduct experiments in social disruption and second-language acquisition; hence, they have no choice but



Creoles, complexity and associational semantics 

to rely on their observations. Which, as things stand now, suggest that, in a handful of cases, languages may become simple without there being any evidence of such disruption. Admittedly, in some of these cases, such disruption may have occurred, only to then be lost in the mists of time. But in other cases, such as that of Riau Indonesian, Minangkabau and Sundanese, all of the available historical evidence points against such a scenario. McWhorter also questions the plausibility of simplification being brought about by internally-motivated developments, drawing attention to the relative scarcity of languages in which simplicity is not demonstrably due to language contact and radical restructuring: “if Riau Indonesian’s nature is a natural development, then we would expect that a map of Sumatra or Sulawesi would be dotted with at least occasional highly analytic languages that had happened along the same pathway...”. But in general, linguistics is replete with examples of rare linguistic phenomena, see for example the collection of papers in Cysouw, Gensler and Wohlgemuth (to appear), and it offers a plethora of reasons, of very different types, for why some linguistic phenomena might be rare, ranging from historical accident to Darwinian selection, from characteristics of our physical environment to innate properties of the human mind. Or, as might be the case here, to the laws of mathematical probability. To the extent that Riau Indonesian, or languages such as Minangkabau and Sundanese, are special, it is not so much any single individual grammatical property that makes them so, but rather the constellation of several independent properties, each of which on its own just happens to make these languages a little bit simpler in some particular domain. Crucially, taken one at a time, each of these independent properties has counterparts in other languages across the world. A small phonemic inventory? That of Hawaiian is smaller. No inflectional morphology? What about Mandarin Chinese. Flexible word order? Look at Latin, Georgian and Warlpiri. While in some cases, simplicity with respect to individual features such as these may indeed have resulted from language contact, in many other cases, such developments are surely more appropriately viewed as products of internal development. But if separate and independent internal developments can bring about simplification in each of several distinct domains of grammar, then it is only a matter of chance and probability until a language crops up in which, coincidentally and for diverse reasons, each of these developments happens to have taken place. It was Einstein who famously stated that God does not play dice with the universe, and McWhorter echoes this prejudice when he inveighs against viewing the existence of very simple languages as resulting from the “roll of the dice”. Unfortunately, in order to do away with dice, it is not sufficient to be a philosophical determinist; one also has to assume omniscience. However, in the real world, we do not know everything about the present or the past, and we cannot always provide meaningful answers to questions such as why English is English and how Riau Indonesian turned into Riau Indonesian. In our state of incomplete knowledge, there is much that can be explained by invoking the laws of probability. In particular, it makes perfect sense to view languages such as



David Gil

Riau Indonesian, Minangkabau and Sundanese as resulting from a fortuitous roll of the dice, the coincidental outcome of a number of independent processes of simplification, each of which is attested elsewhere in the world as an internally-motivated process, without recourse to extraneous historical factors such as social disruption, secondlanguage acquisition and radical restructuring.

References Ansaldo, U. & S. Matthews. 1999. The Minnan substrate and creolization in Baba Malay. Journal of Chinese Linguistics 27. 38–68. Charpentier, J.M. & D.T. Tryon. 1982. Functions of Bislama in the New Hebrides and Independent Vanuatu. English World-Wide 3. 146–160. Chaudenson, R. 1979. Les Créoles Français. Évreux: Nathan. Chaudenson, R.1992. Des Îles, des Hommes, des Langues. Paris: L’Harmattan. Comrie, B.1992. Before complexity. In Hawkins, J.A. & M. Gell-Mann (eds). The Evolution of Human Languages, Proceedings of the Workshop on the Evolution of Human Languages, Held August, 1989 In Santa Fe, New Mexico. Addison-Wesley, Redwood City. 193–211. Cysouw, M., O. Gensler & J. Wohlgemuth. (eds). to appear. Papers from the “Rara and Rarissima” Conference. Leipzig, 29 March -1 April, 2006. [tentative title] Gil, D. 1994. The structure of Riau Indonesian. Nordic Journal of Linguistics 17. 179–200. Gil, D. 2000. Syntactic categories, cross-linguistic variation and Universal Grammar. In Vogel, P.M. & B. Comrie (eds). Approaches to the Typology of Word Classes. [Empirical Approaches to Language Typology.] Berlin/ New York: Mouton. 173–216. Gil, D. 2001a. Creoles, complexity and Riau Indonesian. Linguistic Typology 5. 325–371. Gil, D. 2001b. Escaping Eurocentrism: Fieldwork as a process of unlearning. In Newman, P. & M. Ratliff (eds). Linguistic Fieldwork. Cambridge: Cambridge University Press. 102–132. Gil, D. 2002. The prefixes di- and N- in Malay/Indonesian dialects. In Wouk, F. & M. Ross (eds). The History and Typology of Western Austronesian Voice Systems. Canberra: Pacific Linguistics. 241–283. Gil, D. 2005a. Isolating-Monocategorial-Associational Language. In Cohen, H. & C. Lefebvre (eds). Categorization in Cognitive Science. Oxford: Elsevier. 347–379. Gil, D. 2005b. Word order without syntactic categories: How Riau Indonesian does it. In Carnie, A., H. Harley & S.A. Dooley (eds). Verb First: On the Syntax of Verb-Initial Languages. Amsterdam/ Philadelphia: John Benjamins. 243–263. Gil, D. 2006. Early human language was isolating-monocategorial-associational. In Cangelosi, A., A.D.M. Smith & K. Smith (eds). The Evolution of Language. Proceedings of the 6th International Conference (EVOLANG6). Singapore: World Scientific. 91–98. Gil, D. to appear. Creoles, complexity and the languages of Central Sumatra. In Klein, T. & N. Faraclas (eds). Simplicity and Complexity in Pidgin and Creole Languages. London: Battlebridge. Haspelmath, M., M. Dryer, D. Gil & B. Comrie (eds). 2005. The World Atlas of Language Structures. Oxford: Oxford University Press. Holm, J. 1989. Pidgins and Creoles, Volume 2: Reference Survey. Cambridge: Cambridge University Press.



Creoles, complexity and associational semantics 

Huang, C.-T. J. 1984. On the distribution and reference of empty pronouns. Linguistic Inquiry 15. 531–574. Karlsson, F., M. Miestamo & K. Sinnemäki (eds). to appear. Language Complexity: Typology, Contact, Change. Amsterdam/ Philadelphia: John Benjamins. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar. Cambridge: Cambridge University Press. Lefebvre, C. & A.-M. Brousseau. 2002. A Grammar of Fongbe. Berlin: Mouton de Gruyter. Lim, S.C. 1988. Baba Malay: The language of the Straits-born. In Steinhauer, H. (ed.). Papers in Western Austronesian Linguistics 3. [Pacific Linguistics A-78.] Canberra: Australian National University. 1–61. McWhorter, J. 1997. Towards a New Model of Creole Genesis. New York: Peter Lang. McWhorter, J. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74. 788–818. McWhorter, J. 2000. Defining “Creole” as a synchronic term. In Neumann-Holzschuh, I. & E.W. Schneider (eds). Degrees of Restructuring in Creole Languages. Amsterdam/ Philadelphia: John Benjamins. 85–123. McWhorter, J. 2001a. The Power of Babel, A Natural History of Language. New York: W.H. Freeman. McWhorter, J. 2001b. The rest of the story: Restoring pidginization to creole genesis theory. Journal of Pidgin and Creole Languages 17 (1). 1–48. McWhorter, J. 2001c. What people ask David Gil and why. Linguistic Typology 5. 388–412. McWhorter, J. 2005. Defining Creole. Oxford/ New York: Oxford University Press. McWhorter, J. 2006. Language Interrupted: Signs of Non-Native Acquisition in Standard Language Grammars. New York: Oxford University Press. McWhorter, J. to appear. Complexity and contact: Strange cases in Indonesia. In Karlsson, F., M. Miestamo & K. Sinnemäki (eds). Language Complexity: Typology, Contact, Change. Amsterdam/ Philadelphia: John Benjamins. Maurer, P. 1986. Le papiamento de Curaçao: Un cas de créolisation atypique? Etudes Créoles 9. 97–113. Mufwene, S.S. 1997. Jargons, pidgins, creoles and koines: What are they? In Spears, A.K. & D. Winford (eds). The Structure and Status of Pidgins and Creoles. Amsterdam/ Philadelphia: John Benjamins. 35–70. Pakir, A.G.-I.S. 1986. A Linguistic Investigation of Baba Malay. Ph.D. dissertation, University of Hawai‘i, Manoa. Parkvall, M. to appear. The simplicity of creoles in a typological perspective. In Karlsson, F., M. Miestamo & K. Sinnemäki (eds). Language Complexity: Typology, Contact, Change. Amsterdam/ Philadelphia: John Benjamins. Romaine, S. 1992. The evolution of linguistic complexity in pidgin and creole languages. In Hawkins, J.A. & M. Gell-Mann (eds). The Evolution of Human Languages, Proceedings of the Workshop on the Evolution of Human Languages. Redwood City: Addison-Wesley. 213–238. Ross, J.R. 1982. Pronoun deleting processes in German. Paper presented at the Annual Meeting of the Linguistic Society of America, San Diego. Shosted, R. 2006. Correlating complexity: A typological approach. Linguistic Typology 10. 1–40. Siewierska, A. & D. Bakker. 1996. The distribution of subject and object agreement and word order type. Studies in Language 20. 115–161.



David Gil Tadmor, U. 2002. Language contact and the homeland of Malay. Paper presented at the 6th International Symposium on Malay/Indonesian Linguistics, Bintan, Indonesia. 4 August 2002.

Appendix The interlinear glosses below make use of the following abbreviations: abstr abstract; acc accusative; ag agent; constr construct; clf classifier; def definite; depat depatientive; f feminine; gen genitive; m masculine; nmlz nominalizer; pl plural; prd predicate; prog progressive; prs present; sg singular; trns transitive. Table A1.  English test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

The clown is drinking the book The clown is buying happiness The clown is eating the river The soldier is cutting the axe The coffee is laughing The money is happy The tables are dancing The chairs are jumping The bird is eating The tigers are afraid The dog is drawing The hats are sewing The mouse is chasing the cat The mouse is biting the snake The car is pushing the woman The elephant is looking at the tiger



Creoles, complexity and associational semantics 

Table A2.  Hebrew test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

haleycan šote et hasefer def-clown drink:prs:sg:m acc def-book haleycan kone ošer def-clown buy:prs:sg:m happiness haleycan oxel et hanaħal def-clown eat:prs:sg:m acc def-river haħayal ħotex et hagarzen def-soldier cut:prs:sg:m acc def-axe hakafe coħek def-coffee laugh:prs:sg:m hakesef sameaħ def-money happy:sg:m hašulħanot rokdim def-table-pl:f dance:prs:pl:m hakisaot kofcim def-chair-pl:f jump:prs:pl:m hacipor oxelet def-bird eat:prs:sg:f hanmerim poħadim def-tiger-pl:m afraid:prs:pl:m hakelev mecayer def-dog draw:prs:sg:m hakovaʕim tofrim def-hat-pl:m sew:prs:pl:m aħrey haħatul haʕaxbar rodef def-mouse chase:prs:sg:m after-cnstr:pl def-cat haʕaxbar nošex et hanaħaš def-mouse bite:prs:sg:m acc def-snake hamxonit doxefet et haiša def-car push:prs:sg:f acc def-man-f hapil roe et hanamer def-elephant see:prs:sg:m acc def-tiger

 David Gil

Table A3.  Papiamentu test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

E payaso ta bebe e buki def clown prog drink def book E payaso ta kumpra felisidat def clown prog buy happy E payaso ta kome e riu def clown prog eat def river E sòldá ta kap e hacha def soldier prog cut def axe E kòfi ta hari def coffee prog laugh E plaka ta felis def money prog happy E mesanan ta balia def table-pl prog dance E stulnan ta bula def chair-pl prog jump E para ta kome def bird prog eat E tigernan tin miedu def tiger-pl have fear E kachó ta pinta def dog prog draw E pèchinan ta kose def hat-pl prog sew E raton ta kore tra’i dje pushi def mouse prog run behind gen-def cat E raton ta morde e kolebra def mouse prog bite def snake E outo ta pusha e muhé def car prog push def woman E olefante ta wak e tiger def elephant prog see def tiger



Creoles, complexity and associational semantics 

Table A4.  Sranan test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

A komediman e dringi a buku def:sg funny-man prog drink def:sg book A komediman e bay breyti def:sg funny-man prog buy happy A komediman e nyan a liba def:sg funny-man prog eat def:sg river A srudati e koti a aksi def:sg soldier prog cut def:sg axe A kofi e lafu def:sg coffee prog laugh A moni breyti def:sg money happy Den tafra e dansi def:pl table prog dance Den sturu e dyompo def:pl chair prog jump A fowru e nyan def:sg bird prog eat Den tigri frede def:pl tiger afraid A dagu e teyken def:sg dog prog draw Den musu e nay def:pl hat prog sew A alata e lonbaka a puspusi def:sg mouse prog run-behind def:sg cat A alata e beti a sneki def:sg mouse prog bite def:sg snake A oto e pusu a uma def:sg car prog push def:sg woman A asaw e luku a tigri def:sg elephant prog see def:sg tiger

 David Gil

Table A5.  Bislama test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Fani man i dring buk funny man prd drink book Fani man i pem glad funny man prd buy happy Fani man i kakae reva funny man prd eat river Soldia i katem akis soldier prd cut-trns axe Kofi i laf coffee prd laugh Mane i glad money prd happy Ol tebel oli danis pl table pl-prd dance Ol jea oli jam pl chair pl-prd jump Pijin i kakae bird prd eat Ol taega oli fraet pl tiger pl-prd afraid Dog i dro dog prd draw Ol hat oli somap pl hat pl-prd sew Rat i ronem puskat mouse prd run-trns cat Rat i kakae snek mouse prd bite snake Trak i pusum woman vehicle prd push-trns woman Elefen i luk taega elephant prd see tiger



Creoles, complexity and associational semantics 

Table A6.  Twi test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Dwoka no renom krataa no clown def prog-drink book def Dwoka no ret anigye clown def prog-buy eye-receive Dwoka no rewe asuo no clown def prog-eat.hard river def Sogyani no retwa akuma no def soldier-ag def prog-cut axe Tii no resere tea def prog-laugh Sika no ani agye money def eye receive Apono no resa table def prog-dance Nkonnwa no rehuri pl-chair def prog-jump Anomaa no redidi bird def prog-eat-intr Nseb no suro pl-tiger def be-afraid kraman no redr ade dog def prog-draw thing ky no repam ade hat def prog-sew thing Akura no taa kra no so mouse def chase cat def up Akura no reka w no mouse def prog-bite snake def Kae no repia baa no car def prog-push woman def sono no rehw seb no elephant def prog-look tiger def

 David Gil

Table A7.  Fongbe test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Fanfún´ ´ nù wèmá ´ clown def drink written.matter def Fanfún´ ´ x` awàjíjὲ clown def buy happy-on-come ù t` ´ Fanfún´ ´ clown def eat body.of.water def S´jà ´ sέn asy´ ´ soldier def cut axe def Tî ´ kò-nú tea def laugh-thing àwá Àkwέ ´ jὲ money def come happy Távò l ̂ ú-wè table def:pl eat-dance Zìnkpò l ̂ l´n chair def:pl jump Hὲ ví ´ ù nú bird small def eat thing Tígrù l ̂ ì-xὲsì tiger def:pl resemble-fear Àvùn ´ è dès ̂n dog def draw picture Gbàkún l ̂ t` hat def:pl sew Ajakà ´ nyà awǐì ´ mouse def chase cat def Ajakà ´ hὲn-à ú dàn ´ mouse def hold-tooth snake def M´tò ´ sísέ ny´nû ´ car def push woman def Ajinakú ´ kp´n tígrù ´ elephant def look tiger def



Creoles, complexity and associational semantics 

Table A8.  Yoruba test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Aláwàdà ń mu ìwé ag-have-jest prog drink book Aláwàdà ń ra ìdùnnú ag-have-jest prog buy abstr-sweet-inside Aláwàdà ń jẹ odò ag-have-jest prog eat river ´ọjà ń la àáké soldier prog cut axe Tíì ń rẹérìnín tea prog laugh-nmlz-laugh Inú owó ń dùn inside money prog sweet Àwọn tábìlì ń jó table prog dance pl Àwọn àga ń fò chair prog jump pl Ẹyẹ ń jẹun bird prog eat-thing gbọ òn Àw n àmọtẹékùn ń pl tiger prog tremble Ajá ń yàwòrán dog prog draw-picture Àwọn fìlà ń rán ọ hat prog sew-clothes pl Eku ń lé ológbò mouse prog chase cat Eku ń gé ejò j mouse prog cut snake eat ọ̀ ń ti obìnrin vehicle prog push woman Erin ń wo àmọ tẹ ̀ kùn ́ elephant prog look tiger

 David Gil

Table A9.  Vietnamese test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Anh hề uống sách elder.brother clown drink book Anh hề mua sự sung sướng elder.brother clown buy nmlz happy Anh hề ăn sông elder.brother clown eat river Người lính chặt rìu person soldier chop axe Cà phê cười coffee laugh Tiền sung sướng money happy Bàn nhảy table jump Ghế nhảy chair jump Chim ăn bird eat Hổ sợ tiger afraid Chó vẽ dog draw Mũ may hat sew Chuột đuổi mèo mouse chase cat Chuột cắn rắn mouse bite snake Xe ô tô đẩy người đàn bà vehicle car push person clf grandmother Voi nhìn hổ elephant see tiger



Creoles, complexity and associational semantics 

Table A10.  Minangkabau test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Badut minum buku clown drink book Badut bali sanang clown buy happy Badut makan sungai clown eat river Tentara potong kapak soldier cut axe Kopi galak coffee laugh Pitih sanang money happy Meja bajoget table depat-dance Bangku lompek chair jump Buruang makan bird eat Harimau takuik tiger afraid Anjiang lukih dog draw Topi jahik hat sew Mancik kaja kuciang mouse chase cat Mancik gigik ula mouse bite snake Oto tundo padusi car push woman Gajah caliak harimau elephant see tiger

 David Gil

Table A11.  Sundanese test sentences 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Badut nginum buku clown ag-drink book Badut meuli gembira clown ag-buy happy Badut dahar cai clown eat river Tentara potong kampak soldier cut axe Kopi seuri coffee laugh Duit gembira money happy Meja joged table dance Korsi ngajlot chair jump Manuk dahar bird eat Maung sieun tiger afraid Anjing gambar dog draw Topi jait hat sew Beurit udag meong mouse chase cat Beurit gegel oray mouse bite snake Mobil dorong awewe car push woman Gajah deuleu maung elephant see tiger

Admixture and after The Chamic languages and the Creole Prototype* Anthony P. Grant Edge Hill University

1

The Creole Prototype

In a series of works, John McWhorter proposed existence of what he calls the Creole Prototype (McWhorter 1998, 2001, 2005). The Creole Prototype is a clustering of three typological features which mark out as prototypical creoles those languages which possess all three of them. McWhorter (1998) cited illustrative examples from Saramaccan of Surinam, Mauritian, Fa D’Ambu or Annobon Creole Portuguese, and several others ‘prototypical creoles’. His roster of prototypical creoles includes languages such as Tok Pisin which are still usually regarded as pidgins by many, as most of their users are L2 users, but which are used for general everyday communication, just like creoles, and which resemble creoles structurally. The three features which are supposed to constitute this prototype are: a. The possession in the language of semantically compositional derivation only, b. the absence of lexical tone, c. and the absence of bound inflectional morphology. McWhorter asserts that the concatenation of these three features within a single nonpidgin language shows that the language derives from a pidgin, as pidgins tend to lack all three characteristics; McWhorter also claims that no language which has native speakers and which is of non-pidgin origin shares all three features. Creolists have come up with numerous ways of subclassifying creoles over the past few decades, at least in part in an attempt to isolate what (if anything) makes creole languages ‘special’. Whatever its merits (and these have been hotly debated) the Creole Prototype does at least address some of these ‘special’ issues, namely those about creoles’ intrinsic structural properties and about the nature of the historical relationship they may have with the languages which provided the bulk of their lexicons (in both cases, pidginization is placed at the heart of creole genesis). Prototypicality may be * I would like to thank the editors and two anonymous referees for comments, Bob Blust and Graham Thurgood for advice on Austronesian, and Paul Sidwell for advice on Mon-Khmer.

 Anthony P. Grant

added to the list of kinds of features that are distinctive of creoles and which creolists have found it useful to look for. Not all creoles are prototypical creoles in McWhorter’s sense. Berbice Dutch (Kouwenberg 1994), a mixed language of the interior of Guyana in which items of inflectional morphology and much of the basic vocabulary derives from the Nigerian language Eastern Ijo while most free grammatical morphs and most of the rest of the basic vocabulary derive from Dutch, certainly is not prototypical, because it possesses a small battery of productive inflectional morphemes. Many would regard Réunionnais (Chaudenson 1974) as a French-lexifier creole, but its grammar contains numerous sets of verbs which are inflected for at least two tenses or moods and sometimes more. Not all prototypical creoles are primordial creoles (these being those ‘first-order’ creoles which arose directly from pidgins without being derived from other creoles). Berbice Dutch may be a primordial creole but is not prototypical; Saramaccan is a prototypical creole in terms of its structure and its lack of inflection but it is not a primordial one, as it derives from an early form of Sranan, the major English-lexifier creole of Surinam. Mindanao Chabacano of the southwestern Philippines has its roots in creole Spanish varieties of Manila Bay, such as the creole dialect of the town of Cavite (Frake 1971). Angolar, the language of a maroon community on São Tomé in West Africa (Maurer 1995), derives from an earlier form of Saotomense Creole Portuguese suffused with more borrowings from Kimbundu, a Bantu language of Angola. Examples of such non-primordiality could be multiplied. Nor – and it is important to recognize this – is there any correlation between a creole’s adherence to the Creole Prototype and the presence of – or indeed the lack of – basic lexical admixture from two or more sources, as the cases of Angolar (Maurer 1995; Lorenzino 1998) and Saramaccan (Smith 1987) show. These can be contrasted with Mauritian, which is also prototypical but which has a minimal amount of nonFrench elements (<1%) among its function words and its Swadesh list vocabulary. We may also mention the case of Riau Indonesian, which is the structurally-remodeled Indonesian variety of the eastern Sumatran coast and the adjoining islands opposite the Malay Peninsula, on which David Gil has done much (see especially Gil 1994 and Gil this volume), and Kéo, an Austronesian (Central Malayo-Polynesian) language of Flores, Indonesia, which lacks bound inflectional morphology (Baird 2002). And we may argue that there is a genetically-defined group of such languages, many of which fit (or used to fit) the Creole Prototype criteria perfectly. These are the Chamic languages. Recent work on these, their origins, affinities, and differences, can be found in Grant and Sidwell (2005), with a very important account of Chamic lexicon and historical phonology in the works of Graham Thurgood, especially Thurgood (1999).



Admixture and after 

2 Introduction to the Chamic languages The Chamic languages are mainly spoken in the coastal regions of South East Asia, on the coast of southern Vietnam, not far from the Mon-Khmer languages which at first seemed to provide the most likely counter-exceptions to McWhorter’s claim that creoles are the only languages to share the three characteristics referred to in (a) to (c). Structurally they resemble many languages which would be included under the Creole Prototype. Most of these languages lack lexical tone, their derivational patterns are remarkably transparent (especially so from a synchronic point of view), and they have no productive inflectional morphology. What affixes they do possess seem to border between inflectional and derivational morphology. If McWhorter’s claims about the Creole Prototype are well-founded, and if the languages which possess these features are creole languages, these languages are a set of genetically-related previously-unrecognized creoles, and would constitute a clearly-defined creole language family in themselves. And their internal linguistic history is documented from an earlier period than is that of Riau Indonesian or that of any Malayic lect – or indeed that of any previously recognized creole. The Chamic languages are members of the Western Malayo-Polynesian antigroup within Malayo-Polynesian, and therefore they are Austronesian languages, just as Riau Indonesian is. As previously mentioned, they have received a considerable amount of renown of late, thanks to work by Graham Thurgood, whose research concentrated on historical phonology (especially Thurgood 1996, 1999). What is interesting for our present purposes is that it can be shown that the linguistic changes which lead many of these languages to resemble modern prototypical creoles in typological terms have taken place over the past two millennia. We should note from the start that some Chamic languages fit the Creole Prototype formula better than others do, and that some Chamic languages which may have fitted these criteria earlier in their histories no longer do so because of later changes, mostly induced by contact with surrounding non-Chamic languages, which have erased Creole Prototype features formerly present. Nonetheless, it is likely that, apart from the possession of a system of a couple of productive infixes, a kind of morpheme

1. I define an antigroup (or ‘anti-group’) as a paraphyletic taxon, that is collection of languages which descend from a mother language just as its sister-languages do, but which are not related to one another by having particular shared innovations which mark them off as a cladisticallyconfirmed subgroup. The members of the antigroup constitute the languages which are left over in the course of the process of subgrouping after subgroups confirmed by innovations have been extracted. There may be smaller subgroups in the antigroup, but their immediate superordinate relationships cannot be plotted securely.

 Anthony P. Grant

which typifies both Bahnaric languages and Proto-Austronesian, Proto-Chamic as currently reconstructed would have fitted the Creole Prototype well. The Chamic languages which Thurgood (1999) discusses in varying degrees of detail are the following: Cham (Western Cham, spoken in Cambodia and the Mekong Delta of Vietnam, and the Eastern Cham of the city of Phan Rang, Vietnam), Roglai (with Northern, Southern and Cat Gia Roglai/ Cac Gia Roglai varieties), Haroi, Chru (all southern Vietnam), Jarai and Rade (these two are spoken in the southern Vietnam highlands and are geographically separate from the other Chamic languages), and Tsat, or Hainan Cham. Tsat (the name derives from the Hainanese version of the name Cham) is spoken by a few thousand ‘Utsat’ Muslims in two villages at the extreme south of Hainan, China. Major sources on these languages are given in Thomas et al. (1977), Thomas (1998) and Thurgood (1999); note Thurgood and Li (2003) and Zheng (1997) on Tsat, and Thurgood (2005) on Phan Rang Cham, and papers in Grant and Sidwell (2005), not least Grant (2005a, b). The Chamic languages are related as a group to Acehnese, spoken by over two million people in the extreme north of Sumatra, in such a way that Acehnese is genetically equidistant from all the ‘Indochinese’ and Hainan Chamic languages. But Acehnese is clearly a structurally conservative derivative of Proto-Chamic, though it is one which has been strongly influenced by Malay after Acehnese and the other Chamic languages split up. The split of Acehnese from the Chamic languages of Indochina probably dates from about 982AD, with the fall of the northern Cham Empire to Vietnamese military forces, who were themselves being pushed southward by the Chinese. The split of Tsat from the more northerly Chamic communalects is of a similar or a somewhat later date; we are still not sure when it took place, or indeed whether the Utsat represent the descendants of one or more instances of migration from Champa separated by decades or centuries.

3 Where the Chamic languages fit in genealogically All the structural, phonological and lexical evidence which we have indicates that the languages most closely related to Chamic are Malayic languages, including Malay prop2. It should be noted that the phonological systems of Chamic languages are a considerable distance away from being optimally unmarked, and in this respect they differ from those of many creole languages. For instance, most Chamic languages have 8/9 vowel qualities plus a phonemically distinct feature of vowel length at least between /a/ and /aa/) rather than the 5- or 7-vowel systems we associate with ‘traditional’ creole languages. 3. The Mekong Delta Chams live in that part of Vietnam where Khmer rather than Vietnamese is the major local language, and only a minority of the recent loans into Mekong Delta Cham are from Vietnamese. They are all Muslim and their variety of Cham contains numerous loans from Malay.



Admixture and after 

er (and its varieties, including the aforementioned Riau Indonesian), Minangkabau, Kerinci, Iban and others. It appears that somewhat over two millennia ago there was a Malayo-Chamic linguistic node as a part of what we usually term Western MalayoPolynesian, from which Malayic and Chamic, at that time very similar, were to diverge and go their separate ways (although Malay later influenced some of those Chamic languages spoken by Muslims, mainly later forms of Cham). Blust (1992) reconstructs two dozen high-frequency contentive lexical forms which are shared between Malayic and Chamic and which are rarely if ever found elsewhere; *urang ‘person’ (Phan Rang Cham ra:ng, Malay orang) is one such form. Furthermore, most of the sound-changes which separate Proto-Malayic from PMP (Adelaar 1992) can be paralleled in, and are supported by, the lexical reconstructions of Proto-Chamic in Thurgood (1999). Of all these languages the one with the longest documented history is Cham, which in its Eastern variety is also the one which has strayed least from its traditional location since the Chams moved down from around Hue (where they had arrived after traveling from Borneo) and settled further south within southern Vietnam. The Cham empires, known collectively as Champa, which were founded around the end of the second century AD, were the home of a distinctively Cham culture. This belonged directly to the Indian-influenced ‘Indosphere’ of influence, rather than to the Chineseinfluenced ‘Sinosphere’ which shaped so much of Vietnamese culture and by which the Chams were influenced later and indirectly. But nowadays all the Chams in Cambodia and the Mekong Delta region of Vietnam, and many in eastern Vietnam, are Muslim as are the Utsat and Acehnese. The remainder of the Chamic groups mostly practise a syncretic religion which contains elements of Islam, Hinduism, Buddhism and the beliefs of the Chams’ Mon-Khmer-speaking neighbors. The first record of Cham, given in the Appendix, is a short inscription from Trakiêu near the city of Phan Rang or Panduranga in southern Vietnam (presented in Thurgood 1999: 3 and taken from Marrison 1975), which is written in an Indic-derived script, and which has been dated to the middle of the 4th century AD. The language in this inscription (presented bilingually with a Sanskrit translation) contains several loans from Sanskrit and already shows signs of the syncopation of certain disyllabic stems into monosyllables with initial consonant clusters (which were impermissible in Proto-Malayo-Polynesian or PMP). But otherwise it looks more like literary Malay than like anything else (indeed all the non-Sanskrit elements bar one can also be found in Malay), and it seems to reinforce the existence of a language ancestral to both Malayic and Chamic. This inscription is the first attestation of any Austronesian language and it predates all other attestations of Cham or other Austronesian languages (the next being Old Malay materials) by some three centuries. Later Cham inscriptions, carved in a language which is often mixed with Sanskrit words, date from the 9th and 10th centuries. There has been a literary form of Cham, referred to as Written Cham and most similar to a conservative form of Eastern Cham, in use for over a millennium. This too uses a distinctive script of Indic origin, still known to a diminishing number of male Chams.

 Anthony P. Grant

All this information lets us plot the approximate dates and some of the changes which have been wrought on Chamic through the better part of two millennia, although some important influences (for example the incorporation of words of Mon-Khmer origin) are not attested until the 10th century (borrowing forms of Mon-Khmer origin must almost certainly have begun earlier, but only elements of Austronesian or Malayic and of Sanskrit origin are found in the earliest Cham inscriptions). This is the middle of the period during which Cham, then a dialect linkage spoken in little coastal settlements along the south coast of Vietnam, was beginning to diversify and spread outward and inward. It is clear from the contents of the bilingual Cham and Chinese glossary preserved from the late 15th century (discussed in Edwards and Blagden 1940–1942) that the partial relexification of Cham with Mon-Khmer (specifically Bahnaric) lexical elements was an essentially completed process by the fall of the southern Cham Empire in 1471. For its part, Acehnese has a literary tradition spanning several centuries, and the Cham written tradition has already been mentioned, but linguistic material on the other Chamic languages seems to be nonexistent before the mid-19th century or later. To this day the coverage on Chamic languages is imbalanced, with a considerable number of dictionaries of varying size and quality but rather few collections of texts or grammars. Thurgood (1999) lists most important works produced up to 1995.

4 Influences on the Chamic languages: Whence and where The defining factor which shaped the Chamic languages more than any other is that of contact-induced language change and its effects. Table 1 lists the languages which have exerted major degrees of direct influences on some or all Chamic languages. All the Chamic languages, including Tsat and Acehnese, have acquired a number of elements from Sanskrit (Cham and Acehnese have more of these than other languages). Additionally, the lexica of Cham and Acehnese and presumably Tsat also include many elements from Arabic which may have been acquired from Malay. It is quite clear that external influence from various directions has been of immense importance in shaping the Chamic languages as we now know them; it is also in evidence even in the earliest inscriptions. It appears that the first generation of the populations in the Cham communities, which were settled by traders, craftsmen and pirates, and which moved south from about Hue on the coast of Vietnam, came from the intermarriage of exogenous Austronesian-speakers with their roots in Borneo, who we assume were mostly male, and endogenous speakers of Mon-Khmer languages, members of groups who would have furnished females with whom the Austronesian-speakers would have intermarried. The numbers of Austronesian speakers living there could not easily be replenished but there were always more speakers of Bahnaric and other Mon-Khmer languages in Indochina. The changes towards increased ‘prototypicality’ in the Chamic languages have occurred more specifically as the result of several waves of intimate language contact, mostly with Mon-Khmer languages. Some changes have later been undone as the result of further developments which are also contact-induced.



Admixture and after 

Table 1.  External linguistic influences on Chamic languages: Sources Proto-Chamic: many elements of all kinds from Bahnaric and Mnong languages, possibly some elements from Katuic languages (which may have exerted a strong early influence on Bahnar: Sidwell 2002). There is a sizeable tranche of common elements of unknown origins; it cannot be regarded as external influence but it has a big role in the history of Chamic. Acehnese only: possibly items from Aslian languages of eastern Malaya, many words from Malay (but there are some Malayisms in other Chamic languages, especially Western and Eastern Cham); many elements from Arabic, Sanskrit, Tamil, Portuguese, Dutch and English which have usually come in via Malay. Haroi: Hrê (N. Bahnaric), Bahnar; later forms taken from Vietnamese Written Cham: Pan-Chamic forms plus some more from Bahnar; some Khmer loans, many others from Sanskrit of which rather few may have passed to Western and/or Eastern Cham. Western Cham: predominantly Khmer (the source of a considerable amount of material), a little (late?) material from Vietnamese Eastern Cham: some cultural/intellectual items from Malay (with further items once being used in Cham, but now out of use: Thurgood 2002 ms), most influences from Vietnamese Vietnam Chamic lgs: the dominant influence is Vietnamese (these borrowed forms are fewest in Western Cham, where Khmer has been the dominant influence, but there are very many in Northern Roglai, Southern Roglai, Chru, Haroi, Jarai, Rade and Eastern Cham, and they include some free grammatical morphs) Tsat: Li/Hlai (the Tai-Kadai language indigenous to central Hainan, gives a few loans into Tsat); more significantly, Chinese languages: local Hainanese varieties; military Mandarin; school and state Putonghua, in that order.

The relevant features of the internal structural history of the Chamic languages can be seen by applying traditional diachronic and philological techniques, including the investigation of earlier stages of written Cham and the comparison of forms in one or more Chamic languages with corresponding forms in related languages. Some salient structural features of relevant related and contiguous languages, including those germane to examining the structural features comprising the Creole Prototype, are documented in Table 2. It will be seen from this that Roglai, Chru, Jarai, Rade, and Western Cham (and to a lesser extent Haroi) fit the confines of the Creole Prototype: they lack lexical tone, noncompositional derivation and inflectional morphology. The likeliest source for the Mon-Khmer loans (and other Mon-Khmer influence) that are common to the Chamic languages are the South Bahnaric languages, including Bahnar, a major local language of inland southern Vietnam from the South Central subgroup of the South Bahnaric languages. Mon-Khmer forms which are pan-Chamic include *cit/*sit ‘small’; *cim/cicim ‘bird’; *gop ‘other, guest’ (Acehnese: ‘he, she’). Mon-Khmer forms which are found in at least one Chamic language include: *jep ‘to drink’ (Acehnese); *cot ‘mountain, hill, cordillera’ (Acehnese, Phan Rang Cham). The impact of non-inherited vocabulary items (especially Mon-Khmer loans) on high-frequency vocabulary in two representative languages, Jarai and Acehnese, can be seen in Tables 3 and 4.

Present

Present

Present

Proto-Malayo-Chamic Present (assumed)

Two, productive

Fossilised

Fossilised

Fossilised

Fossilised

Fossilised

Fossilised

Fossilised

Fossilised

Present

Present

Present

Absent

Absent

Absent

Two

Two

Two

Two

Two

Two

Two

Two

Absent

Present

Present

Present

Absent

Absent

Absent

Written Cham

Western Cham

Phan Rang Cham

Haroi

Chru

Rade

Jarai

Absentrthern Roglai

Tsat

Bahnar*

Khmer*

Aslian languages*

Vietnamese*

Hlai*

Minnan Chinese*

indicates non-Austronesian languages.

Present and productive

Present

Acehnese

*

Fossilised

Present

Standard Malay

PMP (assumed)

Infixes

Prefixes

Absent

Absent

Absent

Absent

Absent

Absent

Absent

Absent

Absent

Absent

Absent

Absent

Fossilised

Fossilised

Absent

Absent

Present

Not known

Present

Suffixes

Six

Six

Six

Absent

Absent

Absent

Five

Absent

One allophonic contrast

Absent

Absent

Absent

Present

Present

Present?

Present

Present

Two

Three

Three

Three

Three

Three

Two

No; restructured register

Two

Two, maybe increasing

Present?

No; have become glottal stops

Absent

Absent

Absent

Ejective obstruents

Absent

Absent

Absent

Absent

Absent

Absent

Tones

Table 2.  Typological characteristics of Chamic and other relevant languages

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Unknown

Absent

Numeral classifiers

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Present

Originally though no longer

Absent

No; disyllables were the shortest permissible contentives

Never

Monosyllable default shape for contentive

 Anthony P. Grant



Admixture and after 

Table 3.  The glosses of elements on the Jarai versions of the Swadesh and Blust lists which derive from Mon-Khmer, other non-Austronesian or unknown sources Jarai is the single Chamic language which apparently contains the greatest number of nonAustronesian elements in its basic vocabulary, as far as we can tell; it has therefore been chosen for discussion here. (Even so, it does not contain a much greater layer of non-Austronesian elements than other Chamic languages seem to do.) These data are mostly taken from Thurgood (1999: 279–370); the list was completed by me with the inclusion of a few forms from Lafont (1968). Asterisked forms in the relevant sections are those whose Mon-Khmer credentials are not secure but which are assumed to exist by Thurgood, usually because of their un-Austronesian phonological characteristics. Those concepts which are expressed by adjectives in English are expressed by stative verbs in Jarai. Early Mon-Khmer borrowings (or assumed borrowings from Mon-Khmer languages) which are thus shared with other Chamic languages: Nouns: Head*; neck*; husband/male; man/male; wife (2 forms*); firewood*; leaf; grass*; sand; mountain range*; bird (also means: animal); a fly; rope; river; meat* Adjective equivalents: Old*; correct/right; other; black; white*; dry; warm; big*; narrow*; small*; good; round (2 forms) Verbs: To hold; to sing; to hit or strike (2 forms, 1 of which is of uncertain origin); to vomit; to swell; to scratch*; to eat*; to burn (2 forms); to dig*; to stand; to lie down; to sleep (2 forms); to climb; to pull (2 forms*); to flow or run; to split; to bite*; to break*; to spit (2 forms); to yawn; to steal; to say; to swim; to cut; to choose; to open (2 forms); to wash (2 forms, 1 of them being uncertain in origin); to weep*; to turn Other form classes: Not; what?*; near; and/with* Later Mon-Khmer loans (or possible loans) which are not common to most Chamic languages but which are found in more than one such Chamic language: Nouns: The back (anatomical); horn*; thunder; spider Adjectival equivalents: Dull or blunt; right side*; left side; dirty; heavy*; cold Verbs: To see; to smell; to squeeze*; to hide*; to flow; to hear; to fear; to pull; to wipe away Other form classes: we; thou Loans from other languages: Nouns: Seed (2 forms); person/human being; (possibly) salt (all of these are from Indic sources and all these forms are pan-Chamic or just post-pan-Chamic in distribution, but ‘salt’ may be PMP in origin) Items of unknown origin which are also attested in at least one other Chamic language (sometimes including Acehnese) and which can thus be used in arguments for subgrouping: Nouns: Female; roof; branch; root, water, forest; earthworm; cloud; night (or else the form is from Proto-Austronesian); house Adjectival equivalents: Full; dry (2 forms); thin; much Verbs: to hold; to blow; to laugh; to suck; to cook Other form classes: below; I (polite form); that; thou; because

 Anthony P. Grant

Table 4.  Glosses for words of Mon-Khmer origin in Acehnese (Cowan 1948; Durie & Banta 1995) Body parts: cheek; nostril; neck; stomach/guts; jaw/chin, arm, urine. Kin terms: nephew/niece; grandchild; old man; stranger; parents; older sister; older brother; baby; father; person; great-grandchild. Natural phenomena: hill; swamp; river; tree; coals; noon; dawn; mountain; ditch. Flora and fauna: citrus; cotton; eggplant; lizard; a bear; python; bird; straw; hawk; deer; a frog; a duck; a bird’s beak. Manufactured items: a match; a harrow; ladle; a stable; a mat; a card for a loom; rope; pillar/ post; handle; bowl (
Our knowledge of the history and present characteristics of other languages in the Malayo-Chamic group enables us to recognize that Acehnese is morphologically conservative in preserving productive infixes which Malay on the one hand, and the other Chamic languages on the other have lost. But we can see that Acehnese is innovative within Chamic in its treatment of ejective consonants (which it can yet be shown to have once possessed). On the other hand, Acehnese once had ejectives /b’, d’/, mostly occurring in items from Mon-Khmer languages (in addition to these sounds having developed in a few forms inherited from Proto-Austronesian which contained an /h/

4. Some of these infixes were multifunctional in role in earlier Austronesian languages. For instance /-in-/ or its reflexes marks patient in transitive sentences, is also used as a deverbative nominaliser in certain Austronesian languages (for instance in some used in the Philippines) and was a past-tense marker in others (e.g. Palauan). So drawing a line between purely derivational and purely inflectional affixes is difficult in this case.



Admixture and after 

after the vowel following the /b/ or /d/), but it has categorically changed them to /’/, while preserving original and inherited /b d/ unchanged. Acehnese has never developed lexical tone, which is a secondary and stepwise development in those Chamic languages (notably Phan Rang Cham and Tsat) which have acquired it. We may compare the state of affairs in Western Cham with the increasing use of tone, which occurs to a very slight extent and quite predictably in Jarai, then increases in presence slightly in Eastern or Phan Rang Cham, where it is also quite predictable in occurrence, and finally is an essential part of all morphs in Tsat. Nor has Acehnese ever had much inflectional morphology (nor have any of the Chamic or Malayic languages). What is less clear from the table above, though an examination of the features presented in the PMP row will make this clear, is that the possession of numeral classifiers is an innovative feature in Malayo-Polynesian languages (and it is not just found in languages which have been influenced by major cultural languages of South East Asia such as Chinese languages). Yet the mere possession of such numeral classifiers in an Austronesian language is not sufficient reason for us to attempt to classify together into one genetic unit all Austronesian languages which have them. Of any Chamic language, Acehnese has the greatest number of features which are characteristic of earlier stages in Austronesian and it also shows the smallest number of features characteristic of the Creole Prototype. This may be attributed to its relative conservatism, to a rather early split from the other Chamic languages (as Sidwell 2005 suggests), which still constituted a dialect chain, and to it being further away from strong areal influences from Mon-Khmer and other South East Asian languages (except for Malay, which has exerted a very strong influence on Acehnese lexicon). But Acehnese evidence is crucial to a reconstruction of Proto-Chamic and a deeper understanding of processes which caused the typological cast of Chamic languages (and the order in which these processes occurred).

5. Some Chamic languages also have an ejective palatal stop usually represented as /j’/, which is also characteristic of loans from Mon-Khmer languages, but this sound has not left any traces in Acehnese, and I do not know of any Malayo-Polynesian forms which could have given rise to /j’/. 6. The absence of lexical tone in a particular Chamic language does not mean that the historical development of that language’s phonological system is straightforward. Both Rade and Cat Gia Roglai lack lexical tone, but the historical developments of presyllables in the former, and of most morphs in the latter, are anything but straightforward or predictable, and for that matter the details of the developments differ greatly between the two languages. And in both cases it is impossible to find an external explanation (for instance, influence from another language) which would account for these phonologically very aberrant developments.

 Anthony P. Grant

5 Lexical elements of unknown origin in Chamic One stratum of significance in Chamic which is largely overlooked, but which is mentioned in Table 3, is that of forms which are of uncertain origin, which we may call incognita. There is a considerable tranche of forms in the lexicon of Chamic languages which have as yet no secure etymology, although dozens of them are also found in Acehnese and/ or Tsat as well as in other Chamic languages, so that they count as pan-Chamic elements by virtue of their distribution. Yet others may have occurred in earlier forms of Acehnese or Tsat but have been replaced by loans from Malay or Chinese respectively. The number of lexical incognita which separate Chamic languages from all others is much greater (maybe seven or eight times as great) than the stratum of lexical items, some two dozen or so, which are innovations peculiar to Malayo-Chamic languages. Phonologically and structurally these elements exhibit the features which are most especially found in the Mon-Khmer elements in Chamic languages, such as a high degree of frequency of ejective consonants, although they have not so far been provided with Mon-Khmer etymologies. Examples of such forms without clear cognates outside Chamic are: *ala ‘under, below’, *glay ‘forest’, *kumey ‘female, woman’, *sang ‘house’, *b’uh ‘to see’ (this is also used in Indochinese Chamic languages as the first part of a bipartite negative construction, Lee 1996; neither this construction nor the stem used has an equivalent in relevant Mon-Khmer languages). Each Chamic language has also developed a battery of lexical elements which are unique to it, and it is true that some apparent subgroups in Chamic (for example, Highland Chamic–Jarai and Rade) have special incognita exclusive to that subgroup. Nonetheless the existence of this pan-Chamic layer of incognita is notable, especially since it includes a number of high-frequency structural forms such as certain personal pronouns and several adpositions. Cham has also been a donor language, as well as a recipient language. Local MonKhmer languages of southern Vietnam have often borrowed quite extensively from Cham, which was an important regional lingua franca until the spread of Vietnamese further inland in the 20th century (and some speakers of Chru and Cat Gia Roglai are still bilingual in Phan Rang Cham). Many South Bahnaric languages such as Chrau have borrowed higher numerals from Cham, for instance. We may sum up the content of the Chamic lexicon as follows: Chamic languages contain hundreds of elements inherited from Malayo-Chamic, but also hundreds of borrowings from a variety of Mon-Khmer languages (primarily South Bahnaric, maybe Katuic and latterly often from Khmer and Vietic), a large number of words of unknown origin, and loans from Sanskrit, Arabic, Chinese etc. some of which may go back to Proto-Chamic. There are also more recent loans from French and English, often mediated via Vietnamese. The Chamic languages are reminiscent in their lexical diversity of mixed-lexicon creoles such as Saramaccan.



Admixture and after 

6 Aspects of Chamic typology: Phonology, morphology and syntax The results of the impact of other languages upon the Chamic languages are very considerable. In fact, the Chamic languages look typologically just like many other South East Asian languages. More especially they possess the features characteristic of the Mon-Khmer languages. This can be exemplified from any stratum of a Chamic language (see Alieva 1984 for a useful if partial catalogue of these areal features; other sources of areal features for South East Asia are Henderson 1964; Enfield 2005; and Bisang 2006; see also the discussion of Sinitic in Ansaldo & Matthews 2001). The consequence is that Chamic languages have a typological cast which is typical of their Mon-Khmer neighbors but which is also typical of the Creole Prototype. The fact that Chamic languages resemble some creoles in other typological or structural respects can be seen from examining Tables 5 and 6, which contain Chamic responses to groups of features which have been regarded as typifying many creoles of mostly European-lexifier background. The features in Table 5 are taken from Bickerton (1981). This is a work which is still regarded as the last word on creolistics by some non-creolists because of its attempt to infuse creolistics with the findings of generative grammar and Chomskyan ideas about language acquisition (rather than the reverse being attempted, as science would require), but which has rather fallen into disuse among creolists because of its inability successfully to account for the principles behind the origin of Hawai‘i Creole English, upon speculative accounts of whose genesis Bickerton’s Language Bioprogram Hypothesis was originally developed. Not all these features are found in all creoles of varying lexifier, but most are quite widespread among many creoles. Those features which are listed in Table 6 are from two works by Douglas Taylor (Taylor 1971, 1977) and they were drawn up to demonstrate structural similarities between creoles with different lexifiers (Taylor 1971) and also to highlight features found in many creoles and some West African languages, such as those of the Kwa group (Taylor 1977). Again, no creole contains all the polar (that is, the simply positive versus negative) features but many creoles contain most of them.

Realized and unrealized complements are kept distinct Relativization and subject copying ‘it has’ expresses both possession and existence Presence of bimorphemic question words There are equivalents of passive constructions Zero TMA marks simple past in action verbs and non-past in statives

Anterior TMA marks past before past in active verbs and simple past in statives Irrealis marks unreal time – future, conditional, subjunctive, etc. Nonpunctual marks both durative and habitual The only possible combination of TMA markers is Tense preceding Mood preceding Aspect

5 6 7 8 9 10a

10b 10c 10d 10e

4

Generic or non-specific zero article Fronting of noun phrases for focusing Distinction between attributive, locative-existential (and sometimes also equative) ‘be’-verbs Multiple negation

1 2 3

Bickertonian Language Bioprogram Prototype feature

Yes, in Chamic languages of Vietnam; see Lee (1996) No, there is no distinction Yes Yes Yes No, ‘actives’ are used Not in this way; the primary distinction between actives and statives does not affect Chamic TMA systems, since both kinds of verbs receive zero-marking Not in this way; see 10a Not in this way; see 10a Not in this way; see 10a Not in this way; see 10a

Yes No Yes

Realized in Chamic?

Table 5.  The distinctive features of the Bickertonian Creolistic Prototype (Bickerton 1981) and their realization in Chamic languages

 Anthony P. Grant

18

15 16 17

6 7 8 9 10 11 12 13 14

A form such as ma indicates ‘but’ A verb meaning ‘pass, surpass’ is used to mark the comparative Verbs meaning ‘come’ and ‘go’ include a feature of directionality when followed by an NP, serving also as ‘come to’ and ‘go to’ Double predication

The 3pl pronoun serves as noun plural-marker A combination of past and future markers marks conditional The word for ‘give’ also serves as a preposition meaning ‘for’ or ‘to’ Phrasal ‘what thing/person/place/time’ serve to indicate ‘what?, who?, where?, when?’ A prepositional phrase is employed to express the possessive absolute ‘mine, yours’ etc. A nominal phrase is employed to express the possessive absolute Demonstrative pronouns are postposed to referents Definite articles are postposed to their referents The pronominal determinant is postposed to its referent as in ‘my’, ‘your’, etc. ‘(my) body’ indicates ‘(my) self ’ The iterative/habitual function merges with completive The iterative/habitual is merged with the progressive The iterative/habitual is merged with the future A form such as na is a utility preposition

1 2 3 4

5

Feature

#

No

No Yes No – there is no definite article Yes No No Yes No No; inherited di and MK-derived ba are used in all Chamic languages incl. Acehnese No No Yes

Yes

No No No No

Occurs in Chamic?

Table 6.  Certain features of Caribbean creoles as manifested in Chamic languages (features from Taylor 1971, renumbered sequentially, with features from Taylor 1977 added)

Admixture and after 

 Anthony P. Grant

It should be noted that Malay and especially Tagalog (and to some extent Acehnese) do not have a similar constellation of ‘creole’ features: they preserve more inflectional morphology, which was lost after Acehnese split from the rest of Chamic. Chamic languages of Indochina have compulsory use of numeral classifiers (some of which are etymologically of Malayic origin and may also be found in Malay, others of Mon-Khmer and unknown origin), and they use verb serialization. Like certain Mon-Khmer languages of Vietnam, though not Vietnamese itself, and like modern French and the Gulf of Guinea Portuguese-lexifier Creoles, the Chamic languages which are used in Vietnam use bipartite negatives to negate predicates (Lee 1996 gives examples from five such Chamic languages). They lack bound inflectional morphology, instead marking grammatical relations either with free particles or with zero marking (that is, leaving grammatical relations not overtly marked). They also lack productive derivational morphology (but this is something that Mon-Khmer languages do not lack). Derivation of nouns in Indochinese Chamic languages is therefore compositional and resulting forms are usually transparent as being ‘the sum of their parts’. For instance the ethnic name Roglai means ‘forest people’: ra is from Proto-Chamic ra:ng ‘person’ and glay, a form maybe confined to Chamic, is ‘forest’. Furthermore, Chamic phonological systems have the same salient and typical features as neighboring Mon-Khmer languages (we find for instance the presence of ejective consonants, which are not found in Malayic lects, and we find richer vowel systems than one normally does in Malayic lects: though most have lost schwa, Chamic languages have phonemic mid-open vowels and falling diphthongs in addition to the vowel system of /i e a o u/ ansd schwa familiar from Malay). As the result of secondary developments, some of the Chamic languages now have tonal systems of the same kind as Vietnamese. There are other instances of phonological isomorphism and convergence toward neighboring Mon-Khmer languages too. For example, within the past century the range of word-final consonants permissible in Phan Rang Cham has shrunk in such a way that it now is identical with that which is available in the coterritorial and dominant language Vietnamese (although Written Cham still preserves some of the lost 7. Malayic languages also have numeral classifiers. Among the Chamic languages, some of the Malayic-derived numeral classifiers also serve as full nouns, while others are only used as classifiers. 8. The vast majority of Chamic words which include ejectives /b’, d’/ (and occasionally, in some Chamic languages, also /j’/) are of Mon-Khmer or unidentified origin; only two or three can be reliably attributed to Proto-Malayo-Chamic forms. Additionally Chamic languages have a word-initial glottal stop in non-Malayo-Chamic forms which also occurs elsewhere in the stem but which contrasts with purely vowel-initial forms within Chamic. The broadly diagnostic value of these sounds for marking out the non-inherited status of words in Chamic languages which contain these sounds extends also to certain vowel nuclei (and this is true of some single vowels and also some vowel combinations which each occur singly in inherited Malayo-Chamic forms) which are found in Chamic words. For instance, mid-vowels and nuclei such as /ua, ia, iau/ are rarely found in words of Malayo-Chamic vintage.



Admixture and after 

distinctions, for instance that existing between /-n/ and /-l/). In the meantime, the role of voicing in the articulation of pulmonic egressive stops has diminished in Phan Rang Cham to the point where voicing is no longer phonemically distinctive. The rudiments of a two-tone system have arisen, but tone is still entirely predictable as it occurs on the vowels after what are now voiceless consonants when these used to be voiced (the glottalic ejective stops are still voiced). Chamic adopted the Mon-Khmer tendency to turn a full disyllable into a sesquisyllable, more specifically a stressed monosyllable with a presyllable (the part which will exhibit a restricted and reduced range of vowels and consonants), by moving the stress onto the second half of the word, then reducing or dropping the vowel in the first half, and thus creating a morpheme-initial consonant cluster (these were unattested in PMP). This is already apparent in the Tra Kiêu inscription (see the Appendix). Reduction of disyllabic contentives to sesquisyllables and often then to monosyllables continued apace throughout the history of the Chamic languages, manifesting itself in different ways in the various Chamic languages. This culminates in the level of change which is found in Tsat (Zheng 1997), where under centuries of Hlai and latterly Chinese influence most former disyllables have become restructured as monosyllables plus tone-marking. Examples of the effect of such areal influences can be provided from several other facets of these languages. In fact, there was a very considerable degree of isomorphism or metatypy (Ross 1996) between Chamic languages (especially those spoken in Indochina: I will have cause to refer to these as ‘Indochinese Chamic’) and Mon-Khmer and especially Bahnaric languages. In simple terms, the same constructions using words with the same meanings were used in creating phrases and clauses in both Mon-Khmer and Chamic languages. This isomorphism was aided by the fact that Malayo-Chamic and Bahnaric languages were apparently already using similar typological constructions for certain syntactic purposes in the period before the two groups of languages came into contact. Both groups of languages favored Subject-Verb-Object constituent order in sentences, for example, and both used Noun-Genitive and Preposition-Noun constructions. Both sets of languages originally had a few infixes which were of similar shape and purpose, though much of the morphological apparatus which is found in some Austronesian languages appears to have been eroded in Malayo-Chamic languages in the period before Chamic split from Malayic: Just like Bahnaric languages, Chamic languages mark many kinds of morphosyntactic relations by using free-standing morphemes which are often lexical morphemes (and which thus acquire secondary grammatical roles in addition to their primary lexical roles) or by apposition, or by zero-marking. Quite a lot of this loss of inflections and their replacement by other 9. Despite the perennial popularity of an ‘Austric’ hypothesis, which links Austroasiatic (Munda plus Mon-Khmer) and Austronesian genetically, such a unity if it existed would have been dissolved several millennia ago, and we cannot appeal to that assumed unity to explain the strong typological similarities between the Bahnaric and Chamic languages.

 Anthony P. Grant

structural means (for example in the verb group) seems to have taken place in the language before the composition of the Tra Kiêu inscription before 400 AD. Admixture with Mon-Khmer elements has escaped few classes of Chamic words. Even the personal pronominal system in many Chamic languages is a mixture of forms of Austronesian or Malayo-Polynesian, Mon-Khmer and uncertain origin. For instance in Phan Rang Cham git ‘we’ is traceable back to Austronesian, bih ‘you’ to MonKhmer and tàhla’ ‘I’ is of unknown origin. Overall there was a very high degree of morph-by-morph intertranslatability at phrase, clause and sentence level between Indochinese Chamic languages and the Bahnaric languages which surrounded them. In stratal terms this has occurred because Bahnaric languages had been both a substratal and an adstratal influence upon the Chamic superstrate (that is, Bahnaric languages had already been present in southern Vietnam when Proto-Chamic was imported and remained available as a source of borrowings for Chamic languages), while Khmer and Vietnamese had also served as powerful and influential adstrates. Indeed Bahnaric languages have never been out of the picture for speakers of Indochinese Chamic for the past two millennia. If we compare the features of the Creole Prototype with the major features of South East Asian languages such as Chinese or Thai, we see that noncompositional derivation and the possession of meagre or nonexistent inflectional morphology are features which are shared by both groups of languages, prototypical Creoles as well as South East Asian languages. The major difference rests on the presence (in many South East Asian languages) or absence (in most creoles) of lexical tone. In short, areal linguistic pressures from long-standing members of the South East Asian Sprachbund on these languages have been immense (and current pressures, especially from Khmer, Vietnamese and Chinese, still are strong on certain members of the Chamic family), and as a result the Chamic languages do not stand out much typologically in the context of those geographical areas where the major languages are Mon-Khmer. Our records of Chamic languages, and especially those of Cham proper, which stretch over maybe sixteen centuries, show that this influence has not been immediate in its impact but has been gradual yet cumulative, and that it continued over centuries or even millennia. Not all kinds of influence have been equally long-lasting. Relexification of much Malayo-Chamic vocabulary with elements from South Bahnaric languages appears to have been a process complete by the 15th century, as Edwards and Blagden (1940–1942) shows, but phonological change of some Chamic languages towards their Mon-Khmer neighbors continues today.

7 Transfer of fabric in Chamic: The lexicon Not all the Mon-Khmer and other influence on Chamic languages has been merely typological. What we may call ‘fabric’ – the actual morphemic stock of these languages, be these morphemes bound, free or whatever – has been affected by the incursion



Admixture and after 

of forms from Mon-Khmer languages just as much as ‘pattern’ (the sequences of morphs used in a language to create compounds, phrases and clauses), especially morphological and syntactic typology) has been. This can most immediately be seen from an examination of the contentive morpheme stock of the Chamic languages, or what is generally regarded as the lexicon. The only productive elements of Malayo-Chamic origin which remain in some of the modern Chamic languages are a dwindling number of items of the lexicon. And all elements which are of Austronesian or PMP origin in Chamic languages are also found in the modern Malayic languages, with the exception of a very few forms which still occur in Chamic but which have been replaced over the past centuries by loans in Malayic, where the original forms had earlier presumably occurred. There is no linguistic evidence to suggest that Chamic languages subgroup more closely with any other Austronesian group than they do with Malayic. Chamic languages contain a very high proportion of lexical and free morpheme elements from earlier stages of Bahnaric languages – and to a very great extent these are the same words from the same sources in each Chamic language, which means that they reconstruct back to the parent language Proto-Chamic. Indeed it is likely that elements of Mon-Khmer origin outnumber those of PMP origin in the lexicon as a whole (Alieva 1986 suggests that 50–60% of the lexicon of Chamic languages is loaned, although she gives no source for her estimate and she does not further decompose the various strata of Mon-Khmer and other elements). The figures in the partial reconstruction of Proto-Chamic lexicon in Thurgood (1999), revised by Blust (2000), are as follows: he finds 282 forms from PMP, 275 forms certainly or probably from Mon-Khmer languages (of which 120 have been confirmed as coming from this source) and 169 pan-Chamic forms of uncertain origin, in addition to 24 forms which are early and pan-Chamic loans from extra-South East Asian or other languages, mostly from Sanskrit. If we had a greater amount of well-etymologized Chamic material, I suspect (on the basis of examining the contents of dictionaries of Chamic languages such as Lafont 1968) that the proportion of Chamic forms which had PMP roots would be smaller. The impact of Mon-Khmer languages on the Chamic lexicon is analogous in size, proportion and depth of penetration to that of Portuguese on the lexicon of Saramaccan. The impact of these languages on the shape and form of Chamic morphosyntax and segmental and canonical phonology is even greater, since in many ways Chamic languages nowadays simply look like Mon-Khmer languages which have typically Mon-Khmer phonological word-shapes but which happen to use a basically Malayic lexicon. The only Chamic form class in which we do not find loans from Mon-Khmer present, and also the only one in which pan-Chamic items of unrecognized etymology are absent, is that of the numerals. All the other form-classes – demonstratives, locative adverbs, high-frequency verbs, personal pronouns, numeral classifiers – contain items from at least these sources, Malayo-Chamic, Mon-Khmer (specifically Bahnaric) and items of unknown origin.

 Anthony P. Grant

Indeed, despite the recurrent idea that nouns are the form class into which it is easiest to borrow and incorporate forms from other languages, there are more MonKhmer elements in Thurgood’s list which are not nouns than there are nouns of MonKhmer origin there. Some of the commonest verbs in Chamic languages are of MonKhmer origin, and many other Chamic verbs cannot be shown to be loans but have no cognates elsewhere in Austronesian. The fact that there is minimal bound inflectional verbal morphology in Chamic languages into which loaned verb forms need to be slotted, and that what inflectional morphology there is available in these languages is often synchronically transparent, cannot have hindered the wholesale absorption of such verbs into the lexica of these languages. The fact that both Malayic and Mon-Khmer languages make use of verbal infixes with broadly similar shapes and broadly similar roles may explain why infixed wordforms from Mon-Khmer languages were incorporated easily and wholly into Chamic languages. The absorption of Mon-Khmer items into Proto-Chamic and its daughters is principally an instance of (partial) relexification rather than a process of adding new labels to the lexicon for naming new concepts, or rather than a process of adlexification (the modification of the structure of semantic fields by means of borrowing words which introduce a means of expressing previously undifferentiated concepts), since few of the items which have been borrowed name items or concepts which were hitherto unfamiliar to the Chams. The same can be said of the status and role of many of the widespread elements of uncertain etymological origin. A vivid demonstration of this can be seen by an analysis of the typical Chamic version of a 200-item Blust list (presented in Blust 1993), a widely-used apparatus which is an adaptation of the Swadesh 100- and 200-word lists to the cultural and semantic characteristics of Malayo-Polynesian languages, and in which 31 Swadesh list items have been replaced by items which for semantic or other reasons are of more relevance to Malayo-Polynesian languages (the Blust list is available on the World Wide Web). I have collated such lists for all available Chamic languages and have examined the distribution of inherited and other forms therein across the relevant languages, including Acehnese, and also in PMP and standard Malay (Grant 2005b). The average Blust list gloss is represented by only 1.8 different forms across the nine Chamic languages which I surveyed. Even the average range of different forms per gloss among the Malayic lects using the same list is much higher, showing a ratio of about 2.45 forms per gloss across eight such lects. The etymological breakdown of the average Chamic language translation of the Blust list is as follows: forms of PMP origin number 105 items (including a few forms which show special Malayo-Chamic phonological or morphological developments), forms of Mon-Khmer origin but which are common Chamic in status account for 42 items, unknown origin but pan-Chamic 11, post-Common Chamic but widespread forms 35, elements peculiar to a particular language 3–5, items of later and external origin 2. All in all 115 items on the PMP list of 200 reconstructed forms have direct reflexes in Acehnese or any other Chamic language; the vast majority of these forms are



Admixture and after 

also found in Malay. Although the lists used differ in a number of entries, these figures compare well with the figure for the Saramaccan Swadesh 200-item list given in Smith (1987) in which 49.94% of the entries derive from English, the major lexifier, as opposed to 34% from Portuguese, while the proportion of items of non-Portuguese origin on the Swadesh 200-item list drawn from Maurer (1995) for Angolar is at most 25%. Very few glosses on the Blust list are expressed by an inherited word in one Chamic language and a word of Mon-Khmer derivation in another, so that wholesale relexification of the basic vocabulary toward Mon-Khmer has not been a major factor actuating linguistic change in the time after the period of the breakup of the Chamic dialect cluster a little over a millennium ago.10 What may come as a surprise to students of Austronesian languages is that, using the metric of the Blust list, Chamic languages (and this is true also of Acehnese) are among the most lexically conservative languages in the whole of Malayo-Polynesian. Only the Malayic languages themselves, which are the very genetic sisters of Chamic languages, are more conservative in this respect, but structurally and typologically they (and of course Chamic languages) are less representative of what we assume to have been the state of affairs in PMP than (say) the lexically less conservative Tagalog or other Philippine languages. This odd fact in and of itself does not rule out the possibility that Chamic languages could be seen by some observers to be creolized. Creolization and basic lexical mixture need not go hand in hand, as we have seen in section 1. Could not Chamic languages then be better regarded as mixed lexicon creoles like Saramaccan or Angolar, even if there is no concrete evidence of a pre-Proto-Chamic pidgin stage? Immediately one contrast with the situation of Riau Indonesian suggests itself. Despite its lack of inflectional morphology Riau Indonesian does not seem to have a high proportion of lexical elements which cannot also be found in Jakarta Malay, nor forms which are also missing in other Malay varieties which derive from a Hokkieninfluenced Bazaar Malay (a prototype discussed and illustrated in Adelaar and Prentice 1996). By its admixture with elements of other origins, the vocabulary of the Chamic languages is quite different from that of Riau Indonesian in this respect. The question of language contact and its effects also hangs over the varying aetiologies of the structural changes in Riau Indonesian and the Chamic languages. The proliferation of zero-marking in Riau Indonesian as a means of marking several kinds of relations, which is the major structural feature which sets it apart from other Malayic lects (which mark the same relations more overtly), has not come about as the result of intimate language contact with, and influence by, speakers of other Austronesian languages in Sumatra; for instance Batak. Instead, the increase in zero-marking in Riau Indonesian is internally-driven (zero-marking of previously overtly-marked 10. The direction of linguistic contact has been two-way and some Bahnaric languages, such as Bahnar proper and Chrau, have borrowed many elements of acculturational vocabulary (for instance, the words for ‘1000’ and higher numerals, and also words such as ‘king, ruler’) from Cham.

 Anthony P. Grant

grammatical relations is also found, though to a lesser extent, in other varieties of Indonesian). In the case of the Chamic languages, where zero-marking also abounds, we cannot be sure that internal change is the cause for the spread of zero-marking. And as Chamic infixes such as the denominalizing infix /-an-/ derive from Mon-Khmer languages, we may be justified in expressing such caution.

8 How Indochinese Chamic languages ‘got this way’: The replication of the effects of the Creole Prototype as a dynamic diachronic process The structural characteristics of the Chamic languages, Acehnese apart, seem quite clearly to show the salient features of the Creole Prototype as it is depicted in McWhorter (1998; see also Tables 5 and 6 for further lists of creole-like features in Chamic languages). On those occasions when they lack Creole Prototype features it is not as a result simply of loss of Creole Prototype features which they formerly had, but instead because these have been eroded by the effects of contact-induced language change on particular languages after the split up of the Chamic languages, which have added features to these languages which they did not previously possess. It is therefore probable that there was a period in the history of Chamic languages (and this occurred after Acehnese split off) when the three criteria of the Creole Prototype were obeyed in all the languages, just as they still are in Rade, Northern Roglai (and probably other Roglai varieties too), Haroi, Chru, and Western Cham, and were until recently in Phan Rang Cham. The rise of tones in Phan Rang Cham and Tsat means that they conform less to the Creole Prototype than they previously did. Most of these languages lack distinctive lexical tone and they always have done (just as, we assume, did Proto-Chamic; Malayic languages do not have lexical tone and nor does Acehnese). Meanwhile those which now have lexical tone can be shown to have acquired it through secondary developments of other phonological phenomena in the course of their history. These languages also have a small battery of items of clear, readily segmentable, and mostly non-bound derivational morphology, which are used with a rather small number of stems whose number is no longer increasing. Additionally, such morphological and syntactic relations as are overtly indicated morphemically in Chamic languages, and they are not many (certainly they are fewer than those which are generally encountered in European languages) are carried out by using a small number of free grammatical morphs, some of which have subsequently acquired a status as grammaticalized items after (or at the same time as) their previous status as purely lexical items. Having explored the ‘what’ and the ‘how’ of these features, namely the nature of these features and their causation, we should look at the ‘why’, and, where possible, also the ‘when’. The last of these is the question which can be answered most easily, at least



Admixture and after 

in terms of broad detail. A broad-brush examination of Proto-Chamic inflectional morphology makes this clear, as this is the best way into the answer to the question. It is clear that Proto-Malayo-Chamic possessed the productive use of infixes, and also of some verbal prefixes, since this use has been continued somewhat in the Chamic language Acehnese, where the few infixes which are in productive use reconstruct back to Proto-Austronesian. As such Proto-Malayo-Chamic was a rather conservative language in broad structural terms, preserving much of the structure which has been reconstructed for Proto-Austronesian and Proto-Malayo-Polynesian. Its descendants are not usually so conservative structurally, but they have each lost different features of this system and each of them did so in different orders. Acehnese has kept all of the older features which have been profiled in the table, whereas in Malay there are only relics of the previous infixal system, and these have been lexicalized into forming part of the very few stems in which they occur. Some of these lexicalizations are in fact of Proto-Malayo-Chamic vintage and they suggest that the morphological structure of this language was already undergoing changes over two millennia previously, with some of the affixes then slowly losing their content. There is some Malay linguistic evidence to support this. For instance the Malay form (me)mimpi ‘to dream’ is related to Acehnese lumpĕy ‘to dream’. Both these words go back to an inflected PMP *h-in-ipi, a Malayo-Chamic stem formation deriving from a PMP root *hipi ‘to dream’. The incorporation of the infix -in- into the morph as what is now an undetachable part of a stem is probably a formation of Malayic vintage. There are only rare relics of such inherited infixes within words, but these were to last as far down as the modern form of Written Cham, and these remain as unanalyzable lexical forms into modern Cham. The loss of productive infixation in Chamic languages may postdate the period of loss of such infixes in Malayic, since no Malayic language has this kind of productive infixation,11 but this loss may have occurred after Malayic and Chamic went their separate ways. Of necessity it postdates the splitting off of Acehnese, a language which itself had been strongly marked by the various structural, lexical and phonological effects of Bahnar and similar languages. No other Chamic language has preserved as much of the inherited Proto-Chamic inflectional morphology as Acehnese has – indeed they have all behaved in this respect in the same way as modern Cham has, and this structural change must have its seeds in the period before the Vietnam Chamic languages split up. These languages are all equal-

11. If it does, then it runs counter to the evidence from the Blust list, which suggests that Chamic remained unified as a group for longer before splitting up than the Malayic lects did. Probably the Acehnese divergence from the other Chamic languages in respect of basic lexicon is at least in part the result of later heavy borrowing from Malay.

 Anthony P. Grant

ly simple or equally complex in their morphological apparatus, and in this way they all differ from the morphologically conservative (and thereby more complex) Acehnese.12 This loss in itself suggests a continued period of post-Acehnese or extra-Acehnese Chamic unity, one which included the sharing of lexical and structural innovations (including among these innovations some structural ‘losses’ such as the loss of productive infixation). All this suggests a rudimentary time-frame, of a millennium or less (and probably much less), for the elimination of so much inherited inflectional (and derivational) morphology from the then disunited Chamic languages, which had split into Acehnese and the others, and for its replacement for the most part with techniques of zero-marking of the categories which were formerly indicated morphologically through infixation (if we can truly call this replacement). As we have seen, zero-marking is a technique which in most Chamic languages has replaced the marking of aspectual relations, diathesis, valency and several other features which are encoded on the verb by the use of inherited elements of affixation in other Austronesian languages. The loss of prefixes and of detectable infixes in Tsat should not be seen as an instance of post-migration pidginization of the inflectional structure of Proto-Tsat under the influence of Hlai and later on of Hainanese, both of them being languages with minimal bound morphology. (It is likely, though, that Hlai may have been the first language of the women with whom the Cham settlers on Hainan intermarried in the first years of Utsat life on Hainan, if as we assume the Cham colony there, consisting largely of Muslim merchants fleeing from the fall of the northern Cham capital at Indrapura, was predominantly male; many Utsat still know Hlai). The reason for the loss of such morphemes is much simpler than that, and it has to do with a different sort of contact-induced change. The striking effects of Hainan Chinese segmental and canonical phonology on Tsat meant that the stressed stem syllable (or a portion of it) was the part which survived the reshaping of ‘Proto-Tsat’ syllables away from the typically Chamic sesquisyllable (with its format of a pre-syllable, using a reduced roster of both the inherited vowels and consonants, followed by a full syllable) to a Southern Chinese-style monosyllable which is generally a Hainanized version of the original stressed syllable in the Chamic word. Prefixes and also non-prefixed initial or pretonic syllables which had remained in Proto-Chamic were generally lost without trace in Tsat (except on those occasions when their remnants melted into the body of the newly-construct-

12. Acehnese also contains fewer lexical elements of Mon-Khmer origin than other Chamic languages do, but we can only speculate about the extent to which elements of Mon-Khmer origin which were present in earlier stages of Acehnese have been replaced by innovated forms or by the massive wave of loans from Malay in more modern forms of the language.



Admixture and after 

ed stem).13 Further phonological modifications which happened at a time when Tsat was separated from other Chamic languages meant that infixed morphemes within Chamic words tended to be melted into the onsets of newly-shaped Tsat syllables (with original medial -l- changing into a palatal glide, for instance), or to be dropped altogether. Tsat therefore has no productive bound affixes of any sort. Acquisition of lexical tone means that Tsat can no longer be included under the Creole Prototype. But acquisition of such a feature also shows that Chamic languages are certainly not immune to adopting other areal features which operate, say those which are found in Hainanese but not in Khmer, Vietnamese, Bahnar or Hrê, and which may serve to make individual Chamic languages look increasingly different from one another. We have some insights into some of the processes by which Proto-Chamic became more and more like its neighbouring Bahnaric languages in ways which go beyond gross morphemic borrowing. Thurgood (2001) discusses issues of propositional processing in Phan Rang Cham and explains how the ease of processing propositions in this language, of working out who is doing what to whom, has enabled this language to be easily learnable because of its heightened semantic transparency. Additionally, Thurgood suggests that this consideration (and the typological structure of the resulting language) has made it easier for Cham, both in previous centuries and nowadays, to be used as a lingua franca by people who speak other languages, mostly Mon-Khmer ones. The end result also obeys the principles of the Creole Prototype. Thurgood’s hypothesis is certainly likely to be the explanation for which Chamic turned out the way it did, but I would like to approach matters from a slightly different angle. It is significant that the characteristic structural features from Mon-Khmer languages which have been most successfully replicated in Chamic languages (and most of the syntactic features among these are first attested in written and more formal, and thus more archaic, registers of Cham proper) are those which are also supported by their occurrence in, or which at least have already had a toe-hold in, Malayic languages such as Standard Malay. This is true of the nature of the morphological systems of these languages as consisting of sparse, largely transparent and also largely derivational morphological elements (especially prefixes), and of the honorific and other distinctions which suffuse and even dominate the structure of the personal pronominal system. This commonality across genealogical boundaries is additionally true of such other features as the optional use of free-standing preverbal tense-aspect markers, of obligatory use of numeral classifiers with count nouns (classifiers are much attested in Austronesian languages outside Malay, for instance in languages of Sarawak: Bob Blust, 13. This has also happened in other Chamic languages, which makes the use of certain procedures of the investigation of historical phonology very difficult to practise in Chamic languages. For instance the word for ‘five’ in Malay is lima, in Cham and Northern Roglai lima, and in Jarai rima, but in Rade it is ema and in Tsat ma33. All these changes are perfectly regular and very pervasive. On a sheer count of basic vocabulary Indochinese Chamic languages appear to be dialects of a single language. Because of the operation of sound changes most of them are not mutually intelligible.

 Anthony P. Grant

personal communication, but are not reconstructible to Proto-Malayo-Polynesian and are not found in the major languages of the Philippines), and so on. These two psycholinguistic considerations interacted with one another fruitfully and dynamically, as follows: 1. most morphs in Cham, just like most morphs in Khmer or Bahnar, had a clear lexical semantic content, with their more clearly grammatical use (that is, their use as a grammatical element), when it occurred, being secondary to this primary clear lexical sense; and 2. most typological features and overtly-expressed categories of the (basically derivational rather than inflectional) morphology, and to some extent also the syntax of Cham, Bahnar and Malay, were already similar across these genetic linguistic boundaries, and these structural similarities tended to be expressed by using similar kinds of morphological devices, even if the forms used differed considerably. These pre-existing factors would have made all the more easy the task of increasing and of bringing about the eventual alignment of the typological features of the prestigious but exogenous and demographically-outnumbered Cham language to those of the endogenous and demographically more numerous and more widely-spread, if less prestigious, Bahnaric languages. And indeed this is just what has happened. Typologically Chamic languages resemble Bahnaric languages in more respects than they resemble the Malayic lects, even the structurally simplified Malayic ones, which are undoubtedly their closest genetic relations. The end results of the operation of these processes may look like creoles, but that does not mean that the languages underwent creolization, and certainly not that of the abrupt kind (Thomason & Kaufman 1988: 147–166). The evidence of Acehnese, and sundry philological evidence, and more so a large body of comparative linguistic evidence, may make us sure of that. Structurally and lexically Chamic languages look at first glance like South East Asian editions of mixed creoles, especially Saramaccan and Angolar. But to assume that these languages all arose in the same way is to embrace the abductive fallacy, which states that because a set of processes A result in an end-product B, that another end-product D, which resembles B in all respects, must be the result of the operation of the same set of processes A, when it could result from the operation of a different set of processes C. Were we to indulge in this we would, as Hamp (1992) puts it, be misusing similarity. If Chamic languages are creoles in a diachronic sense, then they are extremely gradual ones, which have been developing as creoles for far longer than European-lexifier creoles have been around. Of course, although we do not fully understand the way in which the Cham communities were set up and how they spread throughout what is now Vietnam, the pattern of development of earlier Cham society in Vietnam was very different from those of any plantation society (of the kinds which were hotbeds for creoles), so that in that respect there is no good demographic reason why Bickerton’s Language Bioprogram Hypothesis, for instance, should apply to Chamic, and especially not to varieties which



Admixture and after 

spread and diversified after the departure of Acehnese. Both the history of Chamic and the history of plantation creoles (which is what Saramaccan and Angolar started out as, before marronage in both cases) require a period of language shift by members of subjugated (if numerically superior) populations, the shift occurring from Bahnar (and similar languages) to Cham in the first case and from various African (and sometimes also American Indian) languages to the European creoles in the other. Both Champa and the Caribbean and other European colonial plantation societies initially depended on in situ interethnic intermarriage for the perpetuation of the languages to which they gave rise. Nevertheless, in the first case it appears that it was the predominantly male composition of the Cham enclaves, areas which were soon largely divorced (at least in terms of military support) from their original societies in Borneo and which were therefore compelled to be largely autonomous, where the colonists intermarried with local women and exacted tribute from indigenous Bahnaric groups. The locally-born women who intermarried with the first Cham settlers (and who were not slaves, let it be said) would have learned Cham (or rather Proto-Chamic) as a second language, but it was one which was in everyday use in their new households, while the children of these unions would have probably been raised as bilingual in (Proto-) Cham(ic) and a widely-used local Bahnaric language, in addition to any other languages which they may have acquired. Furthermore, given the linguistic diversity among Mon-Khmer or even among Bahnaric languages in the area, even at that early period (c. 200BC–0AD), it is possible that the newly superordinate and politically dominant Cham language, which had culturally prestigious speakers scattered over a considerable area, had prestige and status as a lingua franca between speakers of (say) the earlier forms of Bahnaric languages such as Bahnar, Stieng and Chrau, a role which would have reinforced its use, status and position in the speech economy of the area. But all of this scenario is rather speculative, because the solid evidence is sparse. Although the earliest Cham inscription shows to some extent the impact of a MonKhmer syllable canon on Malayo-Polynesian words, we have to wait until inscriptions dating from the ninth century AD before any actual morphs of Mon-Khmer origin are attested in the Written Cham materials which have come down to us and which we can date. In contrast, about half the contentives found in the first Cham inscriptions (including the opening exclamation siddham! ‘good fortune!’ in the first one) are taken straight from Sanskrit, and several of these sanskritisms remain in use in the modern Cham language, and some remain also in other Chamic languages. And yet evidence of this kind enables us to see a sort of link between Chamic languages and creoles in the ways in which we know about their respective histories. We have a body of data, philological and comparative, which shows us a lot about the pathways of evolution of the Chamic languages over the past two millennia, and which shows the loss of inflections as a process which did not happen in a Big Bang, as abrupt creolization happens. It is evidence of a similar kind, mostly comparative and taken from Saotomense, Principense and Fa D’Ambu in the case on Angolar, and both com-

 Anthony P. Grant

parative (from Sranan and Ndyuka) and philological (dating from the late 18th century, and buttressed with Sranan data from that era) for Saramaccan, which allows us to understand so much about the history of these languages. We could wish for more. It would be marvelous to have further Cham inscriptions of an early vintage showing a language in which Mon-Khmer words were already in wide use, or Angolar data from before the 1890s, or more extensive Saramaccan data (including more data from the 19th century, a period from which there is little material, or from the barely attested variety called Djutongo). But we have much already. We do have evidence from present and past forms of the Chamic languages, including Acehnese, to show that they way in which these developed was not by way of pidginization with subsequent creolization, and that this means that languages can acquire a bundle of basic typological features which make them look like creoles, if the historical and areal linguistic circumstances are auspicious.

9 Conclusions This chapter has explored the role of admixture and has submitted McWhorter’s Creole Prototype to a test with material from a group of South East Asian languages, the Chamic languages, which superficially look like McWhorter’s prototypical creoles. Yet these are languages which have borrowed probably half their lexical morphemes, and certainly a very high proportion of their core lexicon, from other languages. We are fortunate that we have sufficient documentation of earlier stages of some of the Chamic languages, and sufficient knowledge of the languages which surround them and the ones with which they are related, to know the history behind the Chamic languages’ Sprachbund-inspired development as the result of massive structural and lexical influence from Bahnaric languages of southern Vietnam. In other scenarios in other parts of the world where similar languages of creole typological appearance are found, we are often not so lucky. What does the study of admixture in the Chamic languages tell us? Here are some preliminary conclusions: 1. The Chamic languages resemble prototypical creoles typologically because they have developed as the result of influence from Bahnaric languages upon a language from the Malayo-Chamic subgroup of Western Malayo-Polynesian. Typological similarity between this language and the Bahnaric languages results in the effects of contact-induced change being maximized through the operation of isomorphism and through the development of an early form of Cham as a lingua franca. 2. The typological similarity between many Bahnaric languages and many creoles makes languages which had been influenced by Bahnaric languages have a superficially creole appearance.



Admixture and after 

3. Philological evidence and supporting material from Acehnese, enables us to see that the way in which the Chamic languages evolved is the result of a set of circumstances which included general typological similarity between most of the languages involved and a long period of continued bilingualism in Cham and Bahnaric languages. Chamic languages are therefore not to be seen as creoles-asformer-pidgins, but rather as languages which emerged gradually over a course of centuries rather than over a course of years. 4. This case study suggests that the greater the amount and variety of diachronic data we have on individual contact situations, the clearer our understanding of those linguistic situations which superficially resemble creoles will be.

References Adelaar, K.A. 1992. Proto-Malayic: The Reconstruction of its Phonology and Parts of its Lexicon and Morphology. [Pacific Linguistics C-119.] Canberra: Australian National University. Adelaar, K.A. & D.J. Prentice. 1996. Malay: Its history, role and spread. In Wurm, S.A., P. Mühlhäusler & D. Tryon (eds). Atlas of Languages of Intercultural Communication in the Pacific, Asia and the Americas. Berlin/ New York: Mouton de Gruyter. 673–693. Alieva, N.F. 1984. A language union in Indo-China. Asian and African Studies [Bratislava] XX. 11–21. Ansaldo, U. & S.J. Matthews. 2001. Typical creoles and simple languages. The case of Sinitic. Linguistic Typology 5 (2/3). 311–326. Baird, L. 2002. A Grammar of Kéo: An Austronesian Language of East Nusantara. Ph.D. dissertation, Australian National University. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Bisang, W. 2006. Southeast Asia as a linguistic area. In Brown, K. (ed.). Encyclopedia of Languages and Linguistics 11. Amsterdam: Elsevier. 587–595. Blust, R.A. 1992. The Austronesian settlement of mainland Southeast Asia. In Adams, K.L. & T.J. Hudak (eds). Proceedings of the Second Annual Meeting of the South East Asian Linguistics Society. Tempe: Arizona State University Press. 25–83. Blust, R.A. 1993. Central and Central-Eastern Malayo-Polynesian. Oceanic Linguistics 32. 243–292. Blust, R.A. 2000. Review of Thurgood 1999. Oceanic Linguistics 39. 435–445. Chaudenson, R. 1974. Le Lexique du Parler Créole de la Réunion. Paris: Champion. Cowan, H.K.J. 1948. Aanteekeningen betreffende de verhouding van het Atjehsch tot de MonKhmer talen. Bijdragen tot de Taal-, Land- en Volkenkunde van Nederlandsch-Indië 104. 429–514. Durie, M. & A. Banta. 1995. Acehnese comparative. In Tryon, D.T. (ed.). Austronesian Dictionary. Volume 1. Berlin/ New York: Mouton de Gruyter. 407–420. Edwards, E.D.& C.O. Blagden. 1940–1942. A Chinese glossary of Cham words and phrases. Bulletin of the School of Oriental and African Studies 10. 53–91. Enfield, N.J. 2005. Areal linguistics and Mainland Southeast Asia. Annual Review of Anthropology 34. 181–206.

 Anthony P. Grant Frake, C.O. 1971. Lexical origins and semantic structures in Philippine Creole Spanish. In Hymes, D. (ed.). Pidginization and Creolization of Language. Cambridge: Cambridge University Press. 223–242. Gil, D. 1994. The Structure of Riau Indonesian. Nordic Journal of Linguistics 17. 179–200. Grant, A.P. 2005a. The effects of multidirectional linguistic contact and Chamic. In Grant, A.P. & P.J. Sidwell (eds). Chamic and Beyond: Mainland Austronesian Languages. [Pacific Linguistics 569.] Canberra: Pacific Linguistics. 37–104. Grant, A.P. 2005b. Norm-referenced lexicostatistics and Chamic. In Grant, A.P. & P.J. Sidwell (eds). Chamic and Beyond: Mainland Austronesian languages. [Pacific Linguistics 569.] Canberra: Pacific Linguistics. 105–146. Grant, A.P. & P.J. Sidwell (eds). 2005. Chamic and Beyond: Mainland Austronesian Languages. [Pacific Linguistics 569.] Canberra: Pacific Linguistics. Hamp, E.P. 1992. On misusing similarity. In Iverson, G.K. & G.W. Davis (eds). Explanations in Historical Linguistics. Amsterdam/ Philadelphia: John Benjamins. 95–103 Henderson, E.J.A. 1964. The topography of certain phonetic and morphological characteristics of South East Asian languages. Lingua 15. 400–434. Kouwenberg, S. 1994. A Grammar of Berbice Dutch Creole. New York/ Berlin: Mouton de Gruyter. Lafont, P.-B. 1968. Lexique Jarai: Parler de la Province de Plei Ku. Paris: Publications de l’Ecole Française. Lee, E.W. 1996. Bipartite negatives in Chamic. Mon-Khmer Studies 26. 291–317. Lorenzino, G.A. 1998. The Angolar Creole Portuguese of São Tomé: Its Grammar and Sociolinguistic History. Munich/ Newcastle: Lincom Europa. Marrison, G.E. 1975. The early Cham language and its relationship to Malay. Journal of the Malaysian Branch of the Royal Asiatic Society 48 (2). 52–59. Maurer, P. 1995. L’Angolar: Un Créole Portugais Parlé à São Tomé. Hamburg: Buske. McWhorter, J.H. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74. 788–818. McWhorter, J.H. 2001. The world’s simplest grammars are creole grammars. Linguistic Typology 5. 125–166. McWhorter, J.H. 2005. Revisiting the Creole Prototype: Signs of antiquity in older languages. Paper given at the conference Creole Languages between Substrate and Superstrate, MaxPlanck Institute for Evolutionary Anthropology, Leipzig. 3–5 June 2005. Ross, M. 1996. Contact-induced change and the comparative method. Cases from Papua New Guinea. In M. Durie & M. Ross (eds). The Comparative Method Reviewed: Regularity and Irregularity in Language Change. Oxford: Oxford University Press. 180–217. Sidwell, P. 2002. Genetic classification of the Bahnaric languages: A comprehensive review. MonKhmer Studies 32. 1–24. Sidwell, P.J. 2005. Acehnese and the Aceh-Chamic language family. In Grant, A.P. & P.J. Sidwell (eds). Chamic and Beyond: Studies in Mainland Austronesian Languages. [Pacific Linguistics 569.] Canberra: Pacific Linguistics. 211–246. Smith, N.S.H. 1987. The Genesis of the Creole Languages of Surinam. Ph.D. dissertation, Universiteit van Amsterdam. Taylor, D. 1971. Lexical and grammatical affinities of creoles. In Hymes, D. (ed.). Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. 293–296. Taylor, D. 1977. Languages of the West Indies. Baltimore: Johns Hopkins Press.



Admixture and after 

Thomas, D.D. (ed.). 1998. Studies in Southeast Asian Languages, no. 15: Further Chamic Studies. [Pacific Linguistics A-89.] Canberra: Australian National University. Thomas, D.D., E.W. Lee & D.L. Nguyen (eds.). 1977. Papers in Southeast Asian Languages, no. 4. Chamic Studies. [Pacific Linguistics A-48.] Canberra: Australian National University. Thomason, S.G. & T. Kaufman. 1988. Language Contact, Creolization and Genetic Linguistics. Berkeley/ Los Angeles: University of California Press. Thurgood, G. 1996. Language contact and the directionality of internal drift: The development of tones and registers in Chamic. Language 71. 1–31. Thurgood, G. 1999. From Ancient Cham to Modern Dialects: Two Thousand Years of Change. [Oceanic Linguistics Special Publication 28.] Honolulu: University Press of Hawai‘i. Thurgood, G. 2001. Learnability and direction of convergence in Cham: The effects of longterm contact on linguistic structures. Proceedings of the Eleventh Annual Western Conference on Linguistics. Fresno: California State University. 507–527. Thurgood, G. 2002. Crawfurd’s ‘Malay of Champa’. ms. [To appear in a Festschrift for P.J. Mistry.] Thurgood, G. 2005. A preliminary sketch of Phan Rang Cham. In Adelaar, A. & N. Himmelmann (eds). The Austronesian Languages of Asia and Madagascar. London: Curzon. 489–512. Thurgood, G. & F.-X. Li. 2003. Contact induced variation and syntactic change in the Tsat of Hainan. In Bradley, D., R. LaPolla, B. Michailovsky & G. Thurgood (eds). Language Variation: Papers on Language Variation and Change in the Sinosphere and the Indosphere in Honour of James A. Matisoff. Canberra: Pacific Linguistics. 285–300. Zheng Y.-Q. 1997. Huihuihua Yanjiu. Shanghai: Shanghai Yuandong Chuban She.

Appendix Transcription of the first Cham inscription, Tra-Kiêu, Vietnam, mid-fourth century AD (transcription and glosses from Thurgood 1999: 3, taken from Marrison 1975: 53). Siddham! Ni yang nāga punya putauv. Fortune! This REL serpent possess king. Ya urāng sepüy di ko, kurun ko jemā labuh nari ya person respect di him for him jewels fall from

svarggah. heaven.

Ya urāng paribhū di ko, kurun saribu thun davam di naraka, ya person insult di him, for one.thousand year remain di hell, dengan tijuh kulo ko and seven family him. (REL: relativiser; ya and di are an indefinite pronoun and a preposition respectively; Sanskritisms are italicized.)

Relexification and pidgin development The case of Cape Dutch Pidgin Hans den Besten University of Amsterdam & Stellenbosch University

1

Preliminaries

Cape Dutch Pidgin (henceforth CDP) may constitute one of the few instances of a documented pidgin preceding a natively spoken contact variety of a European language, in this case Khoekhoe Afrikaans. However, Khoekhoe Afrikaans shares with Standard and Cape Afrikaans the V2 rule of Dutch, which means that CDP has been (further) adapted to Dutch syntax because CDP was an SOV language without V2. Therefore, Khoekhoe Afrikaans – which is sometimes referred to as a Creole variety of Afrikaans – cannot serve as evidence for the Pidgin-to-Creole cycle that is sometimes invoked as a model for Creole genesis. A superficial comparison of CDP with the Khoekhoe languages Nama (Khoekhoegowab) and Korana makes clear that both CDP and Khoekhoe are SOV. Since CDP data are mainly attributed to Khoekhoen it does not seem unreasonable to assume that CDP has come about through relexification of Khoekhoe, as was suggested in den Besten (1978: 44, 1987: 25–27). Nevertheless, the following cautious question should be asked: To what extent does relexification explain the structure of CDP? As will be shown in this paper, relexification does not suffice to explain all of the syntactic facts of CDP. Furthermore, it should be noted that – in so far as word order and functional categories are concerned – the data discussed in this paper argue for a type of relexification that has more in common with the kind of relexification that created Media Lengua (Muysken 1981, 1988) than with relexification as advocated for creole genesis by Lefebvre (1998).

1. Also known as Orange River Afrikaans and Northwest Afrikaans. Older names such as Hottentot Afrikaans or Kleurling (‘Colored’) Afrikaans are shunned nowadays. 2. For a discussion of Lefebvre (1998) see DeGraff (2002).

 Hans den Besten

2 The CDP sentence: Relexification and stripping (and more) 2.1

SOV word order and the history of CDP

Consider the following CDP sentences:

(1) ‘tZa lustigh, duijtsman een woordt Calm, ons V kelum tza quiet, Dutchman-[Pl] one word say, 1Pl 2 cut-throat (1672; Franken 1953: 113)



(2)

gy dit Beest fangum zoo, en nu dood maakum zoo, 2 this animal catch Fut, and now kill Fut, is dat braa, wagtum ons altemaal daar van loopum zoo is that good, wait 1Pl all away run Fut (1705–1713; Kolbe 1727: 1.502)

Example (1) is one of the earliest CDP examples with SOV word order. However, an even earlier case of SOV word order (albeit with directional complements in preverbal position) has been recorded for Cape English Pidgin: 3. In principle, each pidgin example will be provided with an indication of the period, year or even date of recording followed by an indication of the manuscript or publication in which the example can be found, which is sometimes followed by a reference to a modern edition. – Grammatical abbreviations used in this article are: acc: accusative case – Adv: adverbial – comm: common gender – cop: Copula – Decl: declarative – Dep: dependent case – f: feminine – Fut: future – imp: imperative – m: masculine – neg: Negation – Nom: nominative case – Pl: plural – Pass: Passive – poss: possessive – refl: reflexive pronoun – RemPst: remote past – sg: singular – t: tense – 1/2/3: first/second/third person. – Lower case abbreviations set in bold case indicate traces (e.g. vp, adv). 4. V ‘2’ [= u] is the object form of gij ‘2’ and belongs to the written register of 17th century Dutch. Most probably the Khoekhoen who uttered this threat said jou ‘2’ or julle ‘2pl’. 5. Gy ‘2’ belongs to the written register of 17th c. Dutch and should be jij ‘2(sg)’. For similar reasons, nu ‘now’ (= [ny]) should rather be nou. However, it cannot be excluded that Kolb meant [nu] with the German grapheme since noe ‘now’ (= [nu]) is attested for Khoekhoe Afrikaans (van Zyl 1947: 4, 5, 22, 33, etc.), while it is known that Kolb (1719, 1727) used (= [u]) for (parts of) words that he considered un-Dutch. – Daarvan is a germanism (davon) and should be weg. It is unclear why Kolb did not use the latter expression, which is also German after all. This may throw some doubt upon the trustworthiness of this example, which nevertheless contains genuine material (dood makum ‘to kill [lit. ‘dead make’]’; braa < Du. braaf; ons ‘us’ as a nominative). Dit ‘this’ in dit Beest, which looks like the Dutch neuter DP dit beest, may in fact be attributive dat ‘that, those’ (cf. Afr. dit ‘it, that’ < Du. dat ‘that’), since there is evidence from CDP and early Khoekhoe Afrikaans for the use of dat as a variant for die. Cf. den Besten (2004: 92– 95), von Wielligh (1925: 109, 141, 163) and Jacobs (1942: 27, 47). 6. Like CDP CEP started somewhere between the 1590s (first Dutch and English callers at the Cape) and 1652 (founding of the Cape Colony). The last two sentences date from 1715: You Englishman, you no Hottentot (1715; Beeckman 1718; 1924: 117). CEP contributed to the lexicon of CDP. Cf. den Besten (1987).



Relexification and pidgin development 

(3) Coree home go, Souldania go, home go (1613/14; Terry 1622; 1967: 83)

Before the 1670s CDP sentences either are too short to prove SOV or they involve empty have or be or finite be in second position. Note that the latter may be due to attempts by Europeans to ‘improve’ verbless pidgin sentences, although that cannot be proven. However, a clear case of a pidgin sentence that as been tampered with is the following one:

(4) grote Capitain is boven quaad Great Captain is above angry (1712–1717; Büttner 1716?;1970: 41)

By inserting is to the left of boven, Büttner has broken up the NP/DP grote Capitain boven ‘[The] great Captain above’, a deity who is sometimes mentioned in opposition to the “Great Captain below” (Raven-Hart 1971: 21, 118, 147, 234, 269, 321). This having been said, let us return to the examples in (1) and (2). As I stated above, (1) is one of the oldest CDP sentences with SOV word-order. Furthermore it is the oldest CDP sentence with the notorious verbal suffix -e)m / -om / -um or -me. It should be noted, however, that this suffix never was obligatory: the Dutch infinitival ending -en was also possible and there are a few cases of what seem to be deflected verbs, as in the following example, which I interpret as indicated under (5b):

(5)

a Bravas com Kapiteyn, …  b Brav- as com Kapiteyn, … Good [it] [be] when come [the] Captain, … (1673; ten Rhyne 1686; 1933: 140) [Cf. den Besten 2003.]

7. Saldanha or Souldania Bay was the original European name for Table Bay. Coree, whom this sentence is attributed to, was a Cape Khoekhoe who had been abducted to London in order to be educated as an interpreter. 8. Only one example is long enough but it involves ‘wrong’ word order:

(i) Hette





Have [= be] here

Hie





(13 September 1655; Muller 1655; 1952: 417)

[Cf. Afr. het, the present tense of hê ‘have’.] This may either be Dutch V2 (‘(He) is here’) or Khoekhoe topicalization of a remnant VP plus Pro-drop: [VP advk have/be ]j – [3MSg] – herek – vp. [Cf. den Besten 2003.] Similarly, a short Cape English Pidgin sentence from the same source, Nosie ‘Neg see’ [= ‘Don’t know’] (14 Sept. 1655; Muller 1655; 1952: 417), is no evidence for SOV word order. – Note that I have excluded the ‘sentence’ eerst houthalen dan eeten ‘first wood-fetch then eat’ (1647–1648; Janssen & Proot 1649; 1963: 29) because this was a fixed way of speaking among a group of Khoekhoen who hardly knew any Dutch words – if at all. The originators of this expression, which is perfect Dutch, must have been the Dutch sailors for whom the pertinent Khoekhoen were willing to fetch wood (cf. Nienaber 1963: 29).

 Hans den Besten

From the early 1670s until 1720, we can find a reasonable number of SOV sentences, mainly in travelogues and/ or early modern anthropological writings, among which Kolb(e) (1719, 1727), from whom example (2) has been taken. After 1720 there is only a trickle of pidgin data and we have to wait until 1825 before we meet an SOV sentence again:

(6) och Seuer! […] waarom gij mij dan koop? Ah Master! […] why 2 1SG then buy? (1825; Teenstra 1830; 1943: 218)

This is the last SOV sine V2 sentence I am aware of. It is a sentence without the m-suffix: Kolb(e) (1719, 1727) was the last one to use that suffix in a sentence – unless Rackum ‘javelin’ (1733–1741; Mentzel 1787; 1944: 288–289) is not a nominalization (< raken ‘hit’) but a sentence (raak-um ‘hit-it’ – cf. section 3.2). However, the old pidgin was not completely extinct: it left traces in a new SVOlike pidgin that was due to the heavy influx of slaves from Mozambique (since 1775; cf. den Besten 2001) and in early Khoekhoe Afrikaans. The former variety is represented by the fictitious letter by “een ingesete van Stellebos” or ‘a resident of Stellenbosch’ (1831) and by the pidgin spoken by the Frenchman Isaac Albach as reported in Trigardt (1836–1838) and by ex-slaves as reported by von Wielligh (1925) whereas the latter variety is indirectly represented by the letters written by Carolus Bastert (1801) and Jan Bastert (1801) and by C.E. Boniface’s interview with the ‘Hottentot’ Hendrik Kok (1830).10

2.2

Relexification and stripping

The preponderance of SOV structures in CDP is not really surprising given the fact that the Khoekhoe languages are SOV.11 Consider example (7), which has been taken from Rust (1965):

(7) Tita ge ti naoba goro gurin ei-!â ge mû 1SgNom Decl 1SgPoss uncleDep five years ago RemPst see ‘I have seen my uncle five years ago.’

In order to derive a CDP sentence with SOV word order out of (7) by means of relexification, the following seems to be necessary: (i) substitute Dutch words for the 9. The author called himself Kolb (cf. the Dedication in Kolb 1719) but is called Kolbe in the Dutch edition (1727), which may be due to the oblique form Kolben on the title page of the German edition (1719). 10. Cf. den Besten (2001) and (2006b). 11. Cf. Rust (1965), Hagman (1973), Olpp (1977), Haacke (1976) and den Besten (2002) for Nama (Khoekhoegowab) and Meinhof (1930) and Maingard (1962) for Korana. – The connection between SOV in CDP and CEP and SOV in Khoekhoe was first pointed out in den Besten (1978: 42–44).



Relexification and pidgin development 

Khoekhoe content words and pronouns, and (ii) leave out all functional elements. The former process I will refer to as relexification (proper), the latter as stripping.12 Example (7) contains five elements that the sentence has to be stripped of: (a) the so-called pgn (or: person-gender-number) markers -b ‘3MSg’ and -n ‘3CommPl’ (and actually also -ta ‘1Sg’ in tita, but I will ignore that), (b) the dependent case marker -a, (c) the declarative marker ge, and (d) the remote past marker ge. (Declarative ge and temporal ge are differentiated by tone.) Note that relexification (proper) as I am using it here – with retention of substrate word order, that is – has more in common with the kind of relexification discussed for Media Lengua by Muysken (1981, 1988) than with the kind of relexification of major category elements in creole genesis assumed by Lefebvre (1998), which targets superstrate word order – which would (incorrectly) favor SVO for CEP and CDP, since English is SVO and simple mono-verbal Dutch declaratives and imperatives usually are (S)VO as well (due to the V1/V2 phenomenon).13 So, when I speak about ‘relexification (proper)’, I mean relexification of linguistic units in some language L (in this case Cape Khoekhoe) through replacement of the major category lexical items these units are made up of by lexical elements from another language (or other languages). Such a kind of relexification targets substrate word order. Whether, in such a scenario, substrate semantics is preserved (cf. Lefebvre’s ideas about relexification) is irrelevant. (And it may differ from word to word.) As for stripping, note that it is not a necessary concomitant of relexification proper. One could try to relexify the functional elements as well (extended relexification) or one could decide to keep them in place. Potential cases of extended relexification will be discussed elsewhere in this paper.14 As for relexification proper without stripping (the Media Lengua option, so to speak), a completely hypothetical example would be the following variant of (7):15

12. These notions can already be found in den Besten (1978: 44). In a sense den Besten (1987: 25–27) would have been nothing but a rehash were it not for my hypothesis concerning the origin of the -um suffix, which I now consider to be very weak and which I do not want to discuss in the present paper. 13. However, the SOV status of CDP may have a more complex background than substrate word-order retention only. Dutch infinitival speech, which is a native speaker’s way of pidginizing one’s own language, is also SOV. So native speakers of Dutch (and German) may have passively supported SOV word-order post genesin. Coree’s lament in (3), though, shows that such a kind of support was not even necessary during the creation phase. 14. I agree with Lefebvre (1998) (cf. her specific use of reanalysis to account for relexification of functional elements) that relexification of functional elements may be more complex than relexification of content words, because some (many? all?) of the words chosen from the donor language’s lexicon to replace functional elements of the substrate language may have to get assigned a new category. 15. For relexification in Media Lengua see Muysken (1981, 1988).

 Hans den Besten



(8)

(Ik ge [DP mijn oom -b -a ] [DP vijf jaar -n ] 1Sg Decl 1SgPoss uncle -3MSg -Dep five years -3CommPl geleden ge zien) ago RemPst see

[Or, more probably: oom-i-a > oom-a (Nama/Khoekhoegowab) or oom-ma-a > oomma (Cape Khoekhoe).] A somewhat problematic example of relexification without stripping in Khoekhoe is reported upon by Theophilus Hahn in a letter to Hugo Schuchardt:16 “Heeltemalse (adv) Natuura-χu bedorven-he (χu von) (he Suffix des Passiv)” [i.e. “Heeltemalse (adv[erb]) Natuura-χu bedorven-he (χu of/from) (he passiv suffix)”]. This was supposed to be the translation (during a sermon) of Du. “Van Natuur geheelenall bedorven” (lit.) ‘by nature completely depraved’ (spelling Hahn’s), which according to Hahn should have been: “Hoaragase (gänzlich ũba χu gau-gau-he” [gänzlich ‘completely’]. Although I have my doubts about the quality of the translation this is a good example of relexification proper without stripping:

(9) a Hoaraga All??

-se ũ -b -a χu gau-gau -he -Adv nature -3MSg -Dep from deprave -Pass

b Heeltemal -se Natuur Completely -Adv nature

-a χu bedorven -he -Dep from depraved -Pass

Unfortunately, the pgn marker -b has not been retained, but that may be due to the morpho-phonology of the pgn-markers in Nama (Khoekhoegowab), which requires the allomorph -i when 3MSg follows an r, while this allomorph has to contract with the dependent case marker -a (­­-i+-a > -a). Furthermore, note that the postposition has not been relexified (for which cf. section 4). Less problematic but also less spectacular are Hahn’s examples of relexified imperatives: – hā -tse [Khoekhoe] (10) a !gun -tse Run -2MSg come -2MSg

b Loop -tse – Kom -tse [Relexified Kh.] Run -2msg Come -2msg

This is reminiscent of the isolated case of the imperative Atré ‘eat!’, which de Flacourt (1658: 58)17 had elicited in response to a gesture for eating: “manger Atré” [Fr. manger ‘to eat’]. Atré consists of the stem of the Dutch verb eten ‘eat’ and the imperative/ horta-

16. Letter of February 21, 1882 (04346 in Wolf 1993). For a partial edition see den Besten (1986: 226). 17. The second “page 58” is meant.



Relexification and pidgin development 

tive particle -re. A(a)t- instead of Eet- is due to ~<e> allophony in Cape Khoekhoe.18 In view of the above, we may tentatively conclude that the Cape Khoekhoen had at least two models at their disposal: relexification proper with, and relexification proper without stripping. And we may wonder whether they always kept these models apart. As we will see in sections 3 and 5, they did not. Furthermore, it is not clear what stripping actually meant for a native speaker of Khoekhoe when speaking CDP. In my view it can be either of three things: (a) [weak version] Only the functional heads are removed. (b) [strong version] The functional projections are removed as well. (c) [mixed version] Both options are available to the speaker.

2.3

Relexification and Pro-drop

Before going into murkier aspects of the relationship between Khoekhoe and CDP, I would like to discuss some predictions that can be made on the basis of the relexification cum stripping model. Consider the following variant of example (6): (11) Ti naoba -tai ge (titai) goro gurin ei-!â ge mû 1SgPoss uncleDep -1Sg Decl (1Sg­Dep) five years ago RemPst see Whenever a non-subject is topicalized, the ‘inverted’ subject must be assigned dependent case and has to undergo clitic doubling. In this specific case dependent case cannot be seen (-taj … tita-aj > -taj … titaj) but with other pronouns such as îb ‘3MSg’ and with referential expressions the case marker is visible: -bi … îb-ai/Johaneb-ai. Now note that under such circumstances the pronominal subject may be left out. We may call this semi-Pro-drop because the clitic is still signaling the subject. However, if the clitic may count as a functional element relexification cum stripping predicts full Pro-drop for the pidgin. And this prediction is correct. Compare: (12) ..., wat maakum zoo? ..., what [3MSg/2FSg] do Fut? (1705 – 1713; Kolbe 1727: I, 520, II, 162) Interestingly, this pattern has survived into the 19th century. It shows up as a pidginism in the Afrikaans letters by Carolus and Jan Bastert, in Boniface’s rendering of Hendrik Kok’s Afrikaans and in the pidgin letter by the fictitious “resident of Stellenbosch”: (13) waar komt zegt hij de Mens … Where [3MSg] come, says 3MSg the human-being-[Pl] … (C. Bastert 1801: 98)

18. Cf. Nienaber (1963: 186 and sub eet II).

 Hans den Besten

(14) … om nouw mijn ook te zegge, dat19 verlooren zal gaan, as … … for now me also to say, that [1Sg] lost will go, if … ‘…, in order to also tell me now that I shall perish if …’ (J. Bastert 1801: 101) (15) Maar – wat zal nou doen, sjeur? But – what shall [1Sg] now do, master? (Hendrik Kok in Boniface 1830; Nienaber 1971: 38) (16) ik vrag wat wil hef I ask what [3MSg] want have (een ingesete van Stellebos 1831; Nienaber 1971: 54) Pro-drop can also help elucidating the structure of a construction that was used both by the Khoekhoen and by the slaves, in the early as well as in the late corpus: the maskie + Predicate construction. This is in fact maskie ‘even though’ (a loan from Creole Portuguese as spoken by the slaves) + empty pronoun (a pidginism due to the Khoekhoen) + Predicate. Compare the following examples, which are attributed to Khoekhoen:20 (17) a

Mashy [= Masky] doot21, Icke strack nae onse grote Even-though [1Sg] die, 1Sg Fut to our great Kapiteyn toe, die man my soon witte Boeba geme. Chief [i.e. a deity] to (he will give me white oxen.) (1673; ten Rhyne 1686; 1933: 140)



Ik zeg maskie gehoor, sjeur, dan het hulle voor 1Sg say even-though [2Sg] heard, master, (then they have lied sjeur een leuge wys gemaak, … to you, master, …) (Hendrik Kok in Boniface 1830; Nienaber 1971: 37)



b

[For similar cases among the slaves see Franken (1953)]. So much for Pro-drop of non-initial subject pronouns. Now note that in the late text corpus we also find the counterpart of Pro-drop, namely a pronominal subject in the object form, albeit with only one stray example:22

19. Clause-initial subordinators in Khoekhoe also cause subject doubling and semi-Pro-drop. Note that Khoekhoe renders the complementizer that by means of a clause-final embedding noun plus a pgn-marker (or even by means of the pgn-marker only). Therefore clause-initial dat plus Pro-drop is a combination of Dutch and Khoekhoe syntax plus stripping. 20. In Early Modern Afrikaans maskie could be used as a substitute for Du. al ‘even though’. Cf. WAT 10: 165. A variant is almaskie (WAT 1: 161). For subordinators and Pro-drop cf. n. 19. 21. Dood ‘die’ (< dood ‘dead’) is also non-standard Afrikaans. Cf. WAT 2: 245. 22. Note that this ‘rule’ could have been applied twice. Similarly for Pro–drop in example (13).



Relexification and pidgin development 

(18) overal waar hem komt leer hij die volk Everywhere where 3MSgAcc come teach 3MSgNom the people zo moij, … so beautifully, … (C Bastert 1801: 98) Also in CDP this is a scarce phenomenon, with – at best – two examples, one of which is given in (19):23 (19) … tot jou Husing de dubbeltjes betaalt hemme, … … until 2SgACC (to)-Husing the dimes paid have, … (1705–1713; Kolb 1727: I, 122) Note that example (19) may look like a German or older Dutch subordinate with an Indirect Object pronoun in the Wackernagel position. Yet, the context makes it crystal-clear that jou is the Subject. Pro-drop and accusative subject pronouns under inversion seem to support the idea of relexification plus stripping. I would now like to discuss some aspects of sentential structure in CDP that do not seem to lead to such clear conclusions.

2.4

Negation, temporal anchoring and ‘have’ and ‘be’

Consider the following example, which needs a slight revision: (20) Namaqua boeba kros mos coqua  Namaqua boeba kros mosco qua24 [The] Namaqua-[Pl] ox hide-[Pl] [have] very angry [be] (18 February 1661; van Meerhoff 1661; 1957: 484) Example (20) illustrates both have-drop and be-drop. The absence of be (for which we have several examples) is not really surprising. Yet, it could be made to follow from relexification plus stripping – at least if now extinct Cape Khoekhoe was like presentday Nama (Khoekhoegowab), which puts a Tense particle to the left of the nominal predicate and a copula after it except for the present tense, which requires a specialized Tense particle and an empty copula. Empty be may be due to raising of be to T: (21) … [T a ]j – AP – [Cop e ]j … If we apply relexification plus stripping to such a structure only AP will be audible.

23. The use of the object form ons ‘1PL’ as a nominative subject is not restricted to the inversion context. 24. The Namas this sentence is referring to were armed and were carrying shields made of dried ox hides. In view of other data I am inclined to interpret qua (< Du. kwaad ‘angry’) as ‘dangerous’ or ‘aggressive’.

 Hans den Besten

Now if we may assume that Cape Khoekhoe – unlike Nama (Khoekhoegowab) – did not lexically distinguish have and be we can – everything else being equal – predict empty have as well. Note that this is not mere speculation in view of the evident use of have in the sense of be in the following sentence: (22) Hette Hie Have [= be] here (13 September 1655; Muller 1655; 1952: 417) The Khoekhoe women who were saying this – while pointing to the sky – most probably wanted to convey that the numen of the deity they had been venerating just a minute ago was present right there where they were standing.25 It goes without saying that the above is speculative. But if it is correct the relexification hypothesis still holds. However, things are less clear if we take a look at tense/ aspect marking and negation in CDP. These functions are expressed by words in Khoekhoe, but these words, being functional elements, had to undergo stripping and so alternative means had to be looked for to recover these functions. For tense/ aspect marking, use was made of ‘tense adverbials’ such as (al) gedaan ‘(already) done’ and al ‘already’ for perfect, and strack/ strakjes ‘soon, later’ and Engl. soon for future.26 Unlike the tense/ aspect particles of Khoekhoe, these adverbs did not have to immediately precede (or sometimes follow) the verb.27 Compare: (23) Mashy [= Masky] doot, Icke strack nae onse grote Even-though [1Sg] die, 1Sg Fut to 1PlPoss great Kapiteyn toe, die man my soon witte boeba geme. Capain to, that man [= 3MSg] 1Sg Fut white ox-[Pl] give (1673; ten Rhyne 1686; 1933: 140) [= (17a)] For negation, a Dutch pattern was borrowed to compensate for the loss of expressive power caused by non-relexification and stripping of Neg in the Khoekhoe pattern … – V – [Neg tama / tite ] – … (with tama ‘not’, tite ‘negative future’). The corresponding Dutch pattern is (and was) … – [Neg niet ] – … – V – … (assuming an SOV cum V2 analysis for Dutch). This pattern was taken over into CDP, maybe partly on analogy 25. Cf. den Besten (2003). Also compare n. 7. – Note that example (20) is the sole extant CDP sentence with a possessive predicate. For a parallel example we have to turn to Cape English Pidgin: England God, great God, Souldania no God, i.e. ‘England has God, great God, Saldanha does not have God’ (1616; Terry 1622; 1967: 84). Cf. den Besten (2004: Appendix). 26. For (al) gedaan, strack/strakjes and soon see den Besten (1987). For al see den Besten (2004). Also cf. postverbal zoo ‘FUT’(< Du. zo ‘presently’) in (2). This tense adverb can only be found in Kolb (1719, 1727). Note that gedaan also occurs in the letter by “een ingesete van Stellebos” (1831). 27. Possibly a later development is the use of the Dutch / Afrikaans participial prefix ge- to indicate anterior events. This phenomenon can be found in Isaac Albach’s pidgin as reported by Trigardt (1836–1838). Cf. Roberge (2006). And compare gehoor in (17b).



Relexification and pidgin development 

with the CDP pattern resulting from relexification and stripping of the following Khoekhoe pattern for negative quantifiers: … – [XP +Neg ] – … – V – [Neg tama / tite ] – …28 Compare the following example: (24) waarom ons die goeds niet weder beitum en opvretum why 1Pl the bugs Neg in-our-turn bite and up-eat (1705–1713; Kolb 1727: II, 66) So we may conclude that relexification and stripping do not suffice as means for creating CDP. Linguistic creativity (tense adverbials) and adaptation towards Dutch syntax played a role as well.

2.5

Looking ahead

However, we also need an explanation for the CDP verbal ending -um/ -om/ -m/ -em/ -me, for which cf. (1), (2), (12), etc. In den Besten (1987) I suggested that this suffix may derive from C.Kh. -m and -mi, two of the many allomorphs of the (nominal) pgnmarker 3MSg of Cape Khoekhoe, which also served as a nominalization marker. This is an exception to stripping. However it can be argued that -um / -om etc. actually has various functions and that the evidence may be partly artificial. This will be briefly discussed in sections 3 and 5 – briefly because this is a topic for another paper. Furthermore, discussing all the relevant details in the present paper would unnecessarily inflate it.

3 CDP DPs: Relexification, stripping and adaptation 3.1

DP-internal Word Order

The Khoekhoe DP has the following form: (25) [KP [­DP … N – D ] – K ] where D is an enclitic personal pronoun, the so-called person-gender-number marker (pgn-marker). N can be preceded by a demonstrative, a possessive (i.e. DP (di)), a numeral and/ or an adjective. The Kase marker a is not realized if DP is in SpecAgrP, i.e. clause-initially, in the possessor position and in the Spec of the (extended) projection of a denominal postposition. Prenominal elements may show up postnominally by means of an appositive DP with an empty noun:

28. Cf. den Besten (1986) or Roberge (2000). In the meantime I have found a few more negative clauses in CDP and CEP.

 Hans den Besten

(26) [KP [DP N – Di ]: [DP … [N e ] – Di ] – K ] where the Ds have to agree in features, not necessarily in form: (27) a1 a2 b1 b2

gei ao-b – gei om-i big man-3MSg big house-3MSg [­DP ao-b ] : [DP gei [N e ] -b ] – [­DP om-i ] : [DP gei [N e ] -b ] man-3MSg big -3MSg house-3MSg big -3MSg [DP [DP khoe-b ] (di) ha-gu ] person-3MSg (Poss) horse-3MPl di [N e ] -gu ] [DP ha-gu ] : [DP [DP khoe-b ] horse-3MPl : person-3MSg Poss - 3MPl

Relexification plus stripping would yield both Dutch and un-Dutch word orders. However, in CDP we only find expressions of the type A–N. The other order N–A is not attested: (28) a grote Capiteyn (passim) – great Captain b *Capiteyn grote –

groot volcq (van Riebeeck 1658)29 great people *volcq groot

Yet, there is one example that seems to be evidence for the N–A order: Tubaccum tzicum (1670; Bolling 1678: 20). Nienaber (1963: 285) interprets this as ‘good tobacco’, due to the context which says “de kalte god Tuback Tubaccum tzicum” i.e. ‘they called good tobacco Tubaccum tzickum’. In view of the meaning of the adjective in other sources, this should rather be ‘bad tobacco’. However, given the context the Khoekhoen’s comment Tubaccum tzickum does not have to be an NP. It may very well be a sentence: ‘[The] tobacco bad [be]’, i.e. ‘The tobacco is bad’. (However, cf. section 5.) As for possessives, note that Boebasibier in (29) below – which ten Rhyne translates with Lat. lac ‘milk’ and which can be reconstructed as Boeba si bier – does not seem to fit the relexification plus stripping schema. Yet, in spite of Nienaber’s sound laws d ~ t and t ~ ts/s (Nienaber 1963: 178–179, 182), si cannot be Kh. di ‘Poss’, since Nienaber’s sound laws define separate allophonic clusters. Therefore it must be an as yet undiphthongized version of Afr. sy ‘his’ (or rather: of its Dutch counterpart and predecessor sy/ sij):30

29. Grote Capiteyn ‘great captain/chief ’ can be found in various sources (i.a. 1655; van Riebeeck 1655; 1952: 375 and 1673; ten Rhyne 1686; 1933: 140). Groot volcq ‘great people’ (1658; van Riebeeck 1658; 1955: 406) means ‘aristocracy’. 30. There is independent evidence for the use of older [i] instead of younger [εi]/[ïi] by the Khoekhoen of the early colonial period: altit ‘always’ [= altijd/altyd] (1708; Bövingh 1712: 9 and 1714: 30), brito [= brijpot ‘porridge pot’ rather than Fr. cloche ‘bell jar’] (1655; de Flacourt 1658: 5[6]) [t–apocope + Nienaber’s sound law p ~ t (Nienaber 1963: 183)]. There were still remnants of undiphthongized [i] in 20th century Khoekhoe Afrikaans. Cf. den Besten (2005: 213).

(29)

Relexification and pidgin development 

Boebasibier  Boeba si bier [read: C.Kh. bi ‘drink, milk’] Cow his beer/drink/milk (1673; ten Rhyne 1686; 1933: 154)

Therefore, si in (29) signals a case of adaptation to the Dutch possessive construction DPi – pron.i – N, as in die hond z’n neus (litt.) ‘that dog his nose’, i.e. ‘that dog’s nose’. Out of this usage arose the Afrikaans se possessive with the invariable particle se (in Early Modern Afrikaans also sy(n)), as in die brak se neus ‘the dog SE nose’ or die kinders se boeke ‘the children SE books’. Possessives with invariant particles are also attested for pidgin (or creole) speakers in the second half of the 18th century.31 Yet, the possessive structure that can be predicted on the basis of the Khoekhoe DP – DI – NP – D construction and relexification cum stripping, i.e. DP – NP, is also attested; witness examples (30–31) and a couple of other ones:32 (30) Goaso, Lochhoeve broeder – Loeckhoeve, Goaso broeder Goaso, Lochhoeve brother – Loeckhoeve, Goaso brother (1658; van Riebeeck 1658; 1955: 325) (31) Schacher (sijnde Caepman ’s capiteyn zoon) Schacher (being [the] Caepman-[Pl] ’s chief son) ‘Schacher (being the son of the chief of the Caepmans [= ‘Cape-men’])’ (1658; van Riebeeck 1658; 1955: 319–320) Now (31) also shows that CDP occasionally made use of the Dutch s genitive as well, which would again be a case of syntactic adaptation. Note that this does not necessarily imply that ’s – and si – should be seen as cases of relexification of the Khoekhoe particle di. The evidence available is too limited to make such a claim: we don’t know whether these elements followed the syntax of their Dutch counterparts or the syntax of di, except maybe for s, which seems to follow the rules of Dutch.33 Something similar holds for the following isolated case, however for different reasons: (32) en was de grootste om dat hy van voorouders kauwaup waren and was the greatest because he of forbears chief were ‘and he was the greatest [of the two chiefs of the Ogoquas] because his forebears had been chiefs too’ (1779–1780; Gordon 1779–1780; 1992: 39) 31. Cf. den Besten (2006a). 32. The last one being die Volk tolk ‘those people interpreter’ [= their interpreter] (1801; C. Bastert 1801: 99). 33. Dutch s can only be combined with names, except for river names, which are definite DPs, while in CDP s combines with personal names and with ethnic names that are unmarked for number and definiteness. Khoekhoe di on the other hand can be combined with ‘any’ DP, just like se in Afrikaans. With only one example of si being available for the period up to 1720 we are not in a position to judge what was going on in that period.

 Hans den Besten

The DP hij van voorouders as such could count as a case of full relexification of a Khoekhoe DP – DI – NP structure: the DP in such a structure must be in the nominative: îb ‘he’, not îba ‘him’. However, the semi-postpositional use of the Dutch preposition van ‘of ’ is unlikely to be a case of relexification since there is no evidence for relexified postpositions in CDP (see section 4). So the possessive structure in (32) may be due to an early acquirer’s misunderstanding of the structure of Dutch, not unlike the following example of Masbieker Afrikaans Pidgin: in (33a), which is supposed to be equivalent to the Afrikaans sentence in (33b): (33) a

Die kar van die os hy hette gedaan34 [Masbieker Afr.] The cart of the ox he has/is done/weak ‘The cart-ox is exhausted’ [Afr. gedaan ‘exhausted’, (lit.) ‘done’]



Die kar-os het flou geword [Afrikaans] The cart-ox has weak become ‘The cart-ox has become weak’ (1860s; von Wielligh 1925: 96)



b

In conclusion: the evidence for relexification at the level of the DP is very weak.

3.2

Petrified endings? Nominalizations?35

There is however evidence for some cases of Khoekhoe pgn-markers in CDP. A few appellatives and quite a few clan and tribal names in CDP end in a Khoekhoe pgnmarker. These are partly Khoekhoe loans such as karo-s ‘caross’ or Nama-qu-a ‘(a) Nama’ with -s ‘3FSg’ and -gu ‘3MPl’ (and the case marker -a), and partly Dutch loans into Khoekhoe borrowed back into CDP: e.g. Toback-um / Taback-um ‘tabacco’ (1682; Tappe 1704: 131–132) with -(u)m ‘3MSg’ and boe-b-a ‘ox, cow’ with -b­ ‘3MSg’ (and the case marker ­-a), which derives from Du. boe ‘moo’. (For karos, Namaqua and boeba, see examples (20), (17a)/(23) and (29) above.) However, petrified endings in loanwords are irrelevant for the issue at hand: they don’t prove anything about relexification. The same applies to what seems to be a case of a Khoekhoe nominalization of a Dutch verb: kortom ‘little bit/portion; the chief ’s bartering tax’ (1673; ten Rhyne 1686;

34. Note the use of hette (< Afr. het ‘(finite) have’) as a copula. Compare (22). 35. The topic of this subsection as well as section 5 will be more extensively dealt with in another paper, which should also reconsider the etymology of the pidgin endings -um/-om/-em and -me. These pgn-like endings may well be Khoekhoe reinterpretations of Dutch (and English?) enclitic pronouns, which reminded the Cape Khoekhoen of their 3MSg pgn-markers -m, -ma and -mi.



Relexification and pidgin development 

1933: 136, 152 and 1778; Wikar 1779; 1935: 26 – < Du. korten ‘cut back’).36 Similarly for Rackum ‘javelin’ (1733–1741; Mentzel 1787; 1944: 288–289) – < Du. raken ‘hit’.37 Now there happen to be a couple of V+N compounds in CDP whose verbal parts do but whose nominal parts do not end in a pgn-marker -um­, e.g. Rackum-stok ‘hitting stick, javelin’ (1705–1713; Kolbe 1727: II, 47).38 However, these compounds cannot be taken as evidence for relexification. First of all, V-pgn–N is not a possible morphological structure in Khoekhoe as against V–N-pgn (or N–N-pgn). Secondly, there are only two such ill-formed compounds, while Etom schaep (litt.) ‘eat-om sheep’ (1673; ten Rhyne 1686; 1933: 136) may be a sentence with an object clitic and a right-dislocated object: ‘eat-it/himi [the] sheepi’. The remaining V+N compounds lack the suffix -um. Therefore it may very well be the case that Rackum-stok should also be analyzed as a sentence with a topicalized VP: (34) a Rackum stok  b Raack -umi stoki Hit -it [the] stick – which would be a case of relexification with partial stripping. Note that sentences with (remnant) VP topicalization are a normal phenomenon in Khoekhoe.39

3.3

Conclusion

If we disregard the frozen pgn–markers the phrase-internal syntax of DPs in Cape Dutch Pidgin can only be explained if besides relexification plus stripping adaptation to Dutch syntax is taken into account. On the other hand, if the sentential analysis for V-um-N constructions is correct, there is new evidence for relexification – albeit with partial (or even without) stripping.

4 CDP PPs Khoekhoe is an SOV language with post-nominal adpositions only. However, in CDP we find Dutch prepositions and circumpositions and unmarked locatives, while there 36. Wikar’s kortom represents the last instance of -om/-um, be it a suffix or a clitic. 37. Unless Rackum is a sentence: [VP raak ]j – itk – [3Sg]k – vp. Cf. example (34) and section 5. 38. Also Germ. Rakums-Stecken ‘hitting stick, javelin’ (accusative) (1708; Bövingh 1714: 11). 39. In den Besten (1987: 32) it is argued that bijteman ‘bite-man’ and Taback-Teckemans ‘tabacco-steal-men [= tobacco thieves]’ may contain the ending -um. This may be true for bijteman (which should then be analyzed as a sentence), while Teckemans must be a V+N compound without -um witness the following variant: Toback Tackmans (Dagregister 1 April 1669, quoted by Godée Molsbergen 1916: 121).

 Hans den Besten

is late evidence for a postposition saam ‘with’, whose etymology excludes the possibility of deriving it via relexification. A few examples of Dutch prepositions and circumpositions in CDP are: (35) … nae onse grote Kapiteyn toe … … to our great Chief to … (1673; ten Rhyne 1686; 1933: 140) (36) a … by u blyven … … with 2 stay …

b … in de buik zuypen … into the stomach drink (1705–1713; Kolbe 1727: I, 121 & II, 526)

As for unmarked locatives, I will restrict myself to one late example, since the two examples from the pre-1720 period require extra discussion: (37) ons loop weer ander strat 1Pl walk again/yet other street (een ingesete van Stellebos 1831; 1971: 54) Unmarked locatives may derive – through relexification and stripping – from locative expressions with the locative suffix -ba (although I am having doubts about this hypothesis), while the examples in (35) and (36) are evidence for adaptation to Dutch syntax rather than evidence for relexification. Therefore, the postposition saam, as exemplified in (38a–b), is a curious exception, which has to be explained:40 (38) a. … nie bemoei ander mense sam [1SgNom] not interfere other people with (een ingesete van Stellebos 1831; 1971: 54)

b. … aldaa zij had mooij kapraat, mijn zaam … there 3FSgNom has/d nicely spoken, 1SgAcc with (Isaac Albach (a Frenchman) in Trigardt 1836–1838; 1977: 61)

It is clear that post-nominal sa(a)m is evidence for Khoekhoe substrate, but it cannot be evidence for relexification for the following reasons. First of all, Du. saam, a dialectal variant of samen, is not an adposition but an adverb that can co-occur with the preposition met ‘with’ in its comitative usage: samen met DP ‘together with DP’ or met DP sa40. Late attestations of this postposition can be found in Benjamin Kats’s glossings of his own texts in Korana in Engelbrecht (1936: 207, 209). Furthermore, the WAT IV: 77 mentions a regional word handsaam ‘with the hand’, which is a petrified pidgin PP (hand saam ‘[the] hand(-[Pl]) with’). Van Jaarsveld, Jenkinson and de Wet (2001: 152) mention for Griqua Afrikaans of the 1980s (a Khoekhoe Afrikaans variety) voetsam and handsam ‘with the feet or hands respectively’.



Relexification and pidgin development 

men ‘with DP together’. Apparently, a structure like (39a) has been reanalyzed as (39b), after which the new postposition saam has been ‘cut out’ of its original environment: (39) a [P met ] – b [P met ] – c

DP – [Adv saam ]  DP – [P saam ]  DP – [P saam ]

Calling this relexification would be inappropriate. Real relexification would involve the insertion of Dutch prepositions into Khoekhoe postpositional frames, yielding e.g. *DP – [P met ] (with met ‘with’), just like the verb go, which is used in VO frames in English, shows up in an OV frame in example (3) above: … home go, Souldania go. Note furthermore that sa(a)m in (38a–b) is used in environments where Dutch cannot use samen: (40) a … zich (*samen) met anderen (*samen) bemoeien [cf. (38a)] … Refl (*together) with others (*together) interfere

b … (*samen) met iemand (*samen) spreken41 … (*together) with somebody (*(together) talk

[cf. (38b)]

This means that met – DP – samen has also been semantically reanalyzed, which is confirmed by nonstandard Afrikaans in which saam can be combined with met ‘with’ irrespective of its function (met Jan saam ‘with John with’, met die hamer saam ‘with the hammer with’, etc.). Cf. the following case from the letter by the resident of Stellenbosch: (41) hy seg moet dood slan mit die hand sam he say [1Sg] must dead beat with the hand with (een ingesete van Stellebos 1831; 1971: 54) Most probably the change from met DP saam to DP saam is due to Khoekhoe substrate (although the Indo-Aryan and Dravidian languages of the slaves from India-Sri Lanka may have played a role as well), but the change as such is not a case of relexification.

5 CDP clauses again The discussion of V+N compounds in section 3.2 may have made clear that stripping does not have to be complete. Two additional examples of incomplete stripping (which were already discussed in den Besten 1987) are (42) and (43): (42) a Goo, goo reght [reght = [rεx]]  Go, go straight

41. (40b) is grammatical if spreken is interpreted as ‘give a talk’.

 Hans den Besten



b Goo, goo –re Go, go –imp (1655; Muller 1655; 1952: 416)

(43) a Was makom?  What (?) do?

b Wat –ts makom? What –2MSgj [2MSg]j do? (1673; ten Rhyne 1686; 1933: 140)

I would like to suggest that the pre-nominal attributive adjectives grotom ‘big’ and sickom ‘bad’, which can be found in ten Rhyne (1686 [1933: 154]) (and which are problematic from the point of view of Khoekhoe grammar) can be analyzed as cases of incomplete stripping too. Let us first discuss the traditional analysis of Grotom courcour ‘big bird, ostrich’: (44) Grotom courcour Big bird Pre-nominal attributive adjectives do not carry agreeing pgn-markers in Khoekhoe – for which see section 3.1. Furthermore the CDP adjective groot – when used as an attributive element – never carries the suffix -om/-um. The same applies to other adjectives – with the remarkable exception of sickom/sieckum ‘bad’ (< Eng. sick and/ or Du. ziek ‘sick, ill’) as in ten Rhyne’s Sickom courcour.42 However, if we take into account that the pgn-markers cannot really be distinguished from enclitic subject pronouns, (44) can be reanalyzed as a case of predicate topicalization plus subject doubling: (45) [AP Groot- ] -omi courcouri [AP Big - ] -3MSg [the] bird ap [be] Sickum courcour can be analyzed the same way. A parallel example would be “slechte Tabacq, sieckum Tabacq”, i.e. ‘bad tobacco, sieckum Tabacq’ (1668; van Overbeke 1678; 1998: 82). This suggests another (and more complex) sentential analysis of Bolling’s Tubaccum tzickum (cf. section 3.1): (46) tubac­cumi – tzi­ck –umi [The] tobaccoi bad –3MSgj [3MSg]j [be]

42. According to ten Rhyne’s Latin text Sickom courcour means ‘young of a bird, a bird’s young’ (avis foetus), and so ‘young bird’. Given its etymology Sickom should mean something like ‘bad’. On consulting the manuscript underlying ten Rhyne (1686) I found that ten Rhyne had originally written avis foetens ‘stinking bird’.



Relexification and pidgin development 

According to this analysis, Tubaccum has been left-dislocated so that the actual sentence is nothing but the ‘word’ tzickum, which consists of a topicalized adjective and a subject clitic.43 If this analysis can be upheld, new analyses for verbs ending in –um become available. Contrary to what I suggested in den Besten (1987), such verbs do not all have to be nominalizations. At least some of them may be verbs followed by enclitic subject (or object) pronouns. Cf. the following analysis of a short monologue in Germanized CDP: (47) Hollaender arbeitem [The] Dutchmanj – [VP work ]-emj [3MSg]j vp sterbem dem Hottentot [VP die ]-emj [3MSg]j then(?) vp – [the] Hottentotk – sterbem is [read: as] storben44 [VP die ]-emk [3MSg]k vp – when [3MSg]k died [has] krup der ard als ock Hollaender mann crawl [3MSg]k the? earth like [the] Dutchman man (1694, Langhanß 1715: 119)45 We could even go one step further and claim that part of the (written) evidence may be artificial in that colonists and European visitors – not understanding the underlying grammatical system – affixed -um to just any verb – at least in writing, and not consistently – and there is minimal but incontrovertible evidence supporting this hypothesis.46 Further research will have to show whether we may go to such extremes in the grammatical interpretation of our CDP data. For the time being, I do not see any negative consequences for the question with which I opened this article: To what extent does relexification explain the structure of CDP? – where relexification should be understood as relexification of linguistic units (in Cape Khoekhoe) through replacement 43. There are more cases of sieckum (in van Overbeke (1678 [1998: 82]) and Kolb (1719: 448, 1727: II, 4)). These too are amenable to a sentential analysis. 44. Storben, which looks like the Dutch past participle gestorven (Germ. gestorben) may be a misinterpretation of sterben ‘die’. 45. This monologue is rendered in a slightly corrupt form in Raven-Hart (1971): storben is quoted as storbem and ard has been left out. 46. Kolb (1719: 348) – quoting from Arnold (1672) – tacitly changed Hottentot Brockvva! (Arnold 1672: 1099) into Hottentottum Brockqua! – on the mistaken assumption that hottento(t) originally was a verb meaning ‘give’ (+ Broqua ‘bread’). However, most probably hottento(t) (or rather *hotanto) was a sentence (a dancing ditty – cf. Raven-Hart (1971) and Nienaber (1963, 1989)). Furthermore, in none of the sources where Hot(t)ento(t) brokwa (or a variant thereof) can be found – among which at least two eye-witness reports – do we find the ending -um. – Note that the final [t] in hottentot – which most of us only know as an ethnic name – must have been added to give hotanto/hottento a more Dutch appearance. Van Riebeeck (1652–1662) usually writes Hottento although the variant Hottentot was already around. In the course of the second half of the 17th century the latter variant became dominant.

 Hans den Besten

of the content words these units are made up of by similar words from another language (or other languages).

6 Conclusions The answer to the above question concerning the forces that shaped CDP can be relatively short: (a) Relexification [of content words] is not enough. We also need (partial) stripping [of functional elements]. But (b) relexification plus (partial) stripping does not suffice either, because there are two additional factors: adaptation to Dutch syntax and linguistic creativity. This holds for sentential negation, temporal anchoring, word order within the DP, the s-genitive and the choice for pre-nominal and circum-nominal adpositions. Therefore, we should not overestimate CDP’s SOV word order, Prodrop under inversion and Khoekhoe-like enclitic pronouns. Relexification (plus (partial) stripping) is not the whole answer.

References 1

Linguistic Literature

DeGraff, M. 2002. Relexification: A re-evaluation. Anthropological Linguistics 44. 321–414. den Besten, H. 1978. Cases of possible syntactic interference in the development of Afrikaans. In Muysken, P. (ed.). Amsterdam Creole Studies 2. Amsterdam: Instituut voor Algemene Taalweten­schap, Universiteit van Amsterdam. 5–56. den Besten, H. 1986. Double negation and the genesis of Afrikaans. In Muysken, P. & N. Smith (eds). Substrata versus Universals in Creole Genesis. Papers from the Amsterdam Creole Workshop, April 1985. Amsterdam/ Philadelphia: John Benjamins. 185–230. den Besten, H. 1987. Die niederländischen Pidgins der alten Kapkolonie. In Boretzky, N., W. Enninger & Th. Stolz (eds). Beiträge zum 3. Essener Kolloquium über “Sprachwandel und seine bestimmenden Faktoren” vom 30.9. – 2.10.198[6] an der Universität Essen. Bochum: Brockmeyer. 9–40. den Besten, H. 2001. What became of the Cape Dutch pidgin(s)? In Igla, B. & Th. Stolz (eds). “Was ich noch sagen woll­te...”. A Multilingual Festschrift for Norbert Boretzky on Occasion of his 65th Birthday. Berlin: Akademie Verlag. 205–222. den Besten, H. 2002. Khoekhoe syntax and its implications for L2 acquisition of Dutch and Afrikaans. In Mesthrie, R. & P.T. Roberge (eds). Focus on Afrikaans Sociohistorical Linguistics, Part II. Journal of Germanic Linguistics 14 (1). 3–56. den Besten, H. 2003. Deciphering three Pidgin Dutch statements concerning religion recorded by Muller, Ten Rhyne and Bövingh. Quarterly Bulletin of the National Library of South Africa 57. 149–155. den Besten, H. 2004. Where the historian and the creolist meet: The linguistic analysis of “quotations” and other data in the Cape Archives and in old travelogues. In Huigen, S. & J. Kommers



Relexification and pidgin development 

(eds). Interpretations of Colonial Representations. Reflections on Alterity, Social History, and Intercultural Contact. Saarbrücken: Verlag für Entwicklungspolitik Saarbrücken. 81–111. den Besten, H. 2005. Kloeke en het Afrikaans. Tijdschrift voor Nederlandse Taal- en Letterkunde 121. 211–224. den Besten, H. 2006a. The origins of the Afrikaans pre-nominal possessive construction(s). In Thornburg, L.L. & J.M. Fuller (eds). Studies in Contact Linguistics. Essays in Honor of Glenn G. Gilbert. New York/ etc.: Lang. 103–124. den Besten, H. 2006b. Neerlandismen, pidginismen en Afrikaans in brieven van twee Khoekhoen uit 1800. Tydskrif vir Nederlands en Afrikaans 12. 25–42. Engelbrecht, J.A. 1936. The Korana, an Account of their Customs and their History, with Texts. Cape Town: Maskew Miller. Haacke, W.H.G. 1976. A Nama Grammar: The Noun-Phrase. M.A. dissertation, University of Cape Town. Hagman, R.S. 1973. Nama Hottentot Grammar. Ph.D. dissertation, Columbia University. Jacobs, J.F. 1942. Die Grikwas en hul bure. Bloemfontein/ etc.: Nasionale Pers. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar. The Case of Haitian Creole. Cambridge: Cambridge Univerity Press. Maingard, L.F. 1962. Koranna Folktales. Grammar and Texts. Johannesburg: University of Witwatersrand Press. Meinhof, C. 1930. Der Korannadialekt des Hottentottischen. Berlin: Reimer. Muysken, P. 1981. Halfway between Quechua and Spanish: The case for relexification. In Highfield, A & A. Valdman (eds). Historicity and Variation in Creole Studies. Ann Arbor: Karoma. 52–78. Muysken, P 1988. Media Lengua and linguistic theory. The Canadian Journal of Linguistics 33. 409–422. Nienaber, G.S. 1963. Hottentots. Pretoria: Van Schaik. Nienaber, G.S. 1989. Khoekhoense Stamname, ’n Voorlopige Verkenning. Pretoria/ Kaapstad: Academica. Olpp, J. 1977. Nama-Grammatika, soos verwerk deur H.J. Krüger. Windhoek: Inboorlingtaalburo van die Departement van Bantoe-onderwys. Roberge, P.T. 2000. Etymological opacity, hybridization, and the Afrikaans Brace Negation. American Journal of Germanic Linguistics and Literatures 12. 101–176. Roberge, P.T. 2006. On reconstructing a linguistic continuum in Cape Dutch (1710–1840). In Cravens, Th.D. (ed.). Variation and Reconstruction. Amsterdam/ Philadelphia: John Benjamins. 179–200. Rust, F. 1965. Praktische Namagrammatik. Auf Grund der Namagrammatiken von H. Vedder und J. Olpp. Cape Town: Balkema. van Jaarsveld, G.J., A.G. Jenkinson & A.S. de Wet. 2001. Die intertaalteorie en taalverskeidenheid. In Carstens, A. & H. Grebe (eds). Taallandskap. Huldigingsbundel vir Christo van Rensburg. Pretoria: Van Schaik. 148–156. von Wielligh, G.R. 1925. Ons geselstaal. ‘n Oorsig van gewes­te­like spraak soos Afrikaans gepraat word. Met ‘n Inlei­ding van Prof. J.J. Smith. Pretoria: Van Schaik. WAT: P.C. Schoonees et al. 1950–…. Woordeboek van die Afrikaanse taal. Pretoria: Staatsdrukker/ Stellenbosch: Buro van die WAT. Wolf, M. 1993. Hugo Schuchardt Nachlaß. Schlüssel zum Nachlaß des Linguisten und Romanisten Hugo Schuchardt (1842–1927). Graz: Leykam.

 Hans den Besten

2

Sources for CDP, CEP and CKh

Arnold, Chr. (ed.). 1672. Wahrhaftige Beschreibungen dreyer mächtigen Königreiche / Japan, Siam und Corea / [–]. Nürnberg: Michael and Joh. Friedrich Endters. Bastert, C. 1801. [letter to the Dutch Missionary Society] Berigten en brieven voorgelezen op de maandelijksche bedestonden van het Nederlandsch Zendeling-Genootschap 1801: 95–99. Rotterdam: N. Cornel. [Edition in den Besten, H. 2006b. Neerlandismen, pidginismen en Afrikaans in brieven van twee Khoekhoen uit 1800. Tydskrif vir Nederlands en Afrikaans 12. 25–42.] Bastert, J. 1801. [letter to the Dutch Missionary Society] Berigten en brieven voorgelezen op de maandelijksche bedestonden van het Nederlandsch Zendeling-Genootschap 1801: 99–102. Rotterdam: N. Cornel. [Edition in den Besten, H. 2006b. Neerlandismen, pidginismen en Afrikaans in brieven van twee Khoekhoen uit 1800. Tydskrif vir Nederlands en Afrikaans 12. 25–42.] Beeckman, D. 1718 [1924]. [fragment of] A Voyage to Borneo in 1714. In Botha, C.G. (ed.). Collectanea. Cape Town: Van Riebeeck Society. 112 –118. Bolling, F.A. 1678. Oost-Indiske Reisebog [...] Copenhagen. [Dutch translation by J. Visscher. In Bijdragen tot de Taal-, Land- en Volkenkunde van Neder­landsch-Indië 68, The Hague, 1916.] [Partial English translation in Raven-Hart, R. (ed.). 1971. Cape Good Hope 1652 – 1702. The first fifty years of Dutch colonisation as seen by callers. 2 vols. Cape Town: Balkema. 143– 154.] Boniface, C.E. 1830 [1971]. [Interview with Hendrik Kok in De Zuid-Afrikaan.] In Nienaber, G.S. (ed.). 1971. Afrikaans in die vroeër jare. Johannesburg: Voortrek­kerpers. 36–42. Bosman, D.B. & B. Thom (eds). 1952–1957. Daghregister gehouden by den oppercoopman Jan Anthonisz van Riebeeck, 3 vols. Cape Town: Balkema. Bövingh, J.G. 1712. Curieuse Beschreibung und Nachricht von den Hottentotten / […]. Gedruckt im Jahr 1712. Bövingh, J.G. 1714. Kurtze Nachricht von den Hottentotten – Bey dieser zweiten Auflage/ von den bey der ersten eingeschlichenen Fehlern gesaubert/ und mit eigener Feder ans Licht gebracht Von Johann Georg Bövingh, Missionario. Hamburg: Bey Caspar Jahkel. Büttner, J.D. 1716? [1970]. Waare Relation und Beschrijbung von / Cabo de Goede Hoop / und derselber natuurlicher Inwoonderen Natur- / Gebräuchen, Thun, und Wesen, […]. In Nienaber, G.S. & R. Raven-Hart (eds). Johan Daniel Buttner’s account of the Cape, Brief description of Natal, Journal Extracts on East Indies. Cape Town: Balkema. 13–119. de Flacourt, É. 1658. Chap. XXV: Langage des Sauvages de la Baye de Saldaigne an [sic: au] Cap de Bonne Esperance. In de Flacourt, É. Petit Recueil. Paris. 55–61. een ingesete van Stellebos. 1831 [1971]. [Fictitious letter to the editor of De Zuid-Afrikaan.] In Nienaber, G.S. (ed.). 1971. Afrikaans in die vroeër jare. Johan­nes­burg: Voortrek­kerpers. 54. Franken, J.L.M. 1953. Taalhistoriese bydraes. Cape Town: Bal­kema. Godée Molsbergen, E.C (ed.). 1916. Reizen in Zuid-Afrika in de Hol­landse tijd. Vol. 1. The Hague: Nijhoff. Gordon, R.J. 1779–1780 [1992]. Particularités relatives à quelques hordes Hottentottes. ms. [=] Gordon’s notes on the Khoikhoi / Gordons aantekeningen betreffende de Khoikhoi. In Smith, A. & R. Pheiffer (eds). 1992. Colonel Robert Jacob Gordon’s Notes on the Khoikhoi 1779–80. [=] Annals of the South African Cultural History Museum 5 (1). Cape Town: South African Cultural History Museum. 10–45. Janssen, L. & M. Proot. 1649. Remonstrantie. ms. [quoted via Nienaber (1963)]



Relexification and pidgin development 

Kolb, P. 1719. Caput Bonae Spei Hodiernum, das ist vollständige Beschreibung des Africa­nischen Vorgebürges. Nürnberg: Peter Conrad Monath. Kolb[e], P. 1727. Naaukeurige en Uitvoerige Be­schry­ving van De Kaap de Goede Hoop... waarby een Beschryving van den oor­sprong der Hottentotten [...]. 2 vols. Am­ster­dam: Balthasar Lakeman. Langhanß, Ch. 1715. Neue Ost-Indische Reise [...]. Leipzig: Michael Rohrlachs seel. Wittib und Erben. le Roux, T.H. (ed.). 1977. Die dagboek van Louis Trigardt. Uitgegee met Inleiding, Aantekeninge en Glossarium. 3rd ed. Pretoria: Van Schaik. Mentzel, O.F. 1787. Beschreibung des Vorgebirges der Guten Hoffnung. Vol. 2. Glogau: Chr.F. Günther. – quoted after the translation A geographical and topographical description of the Cape of Good Hope by G.V. Marais and J. Hoge, revised and edited by H.J. Mandelbrote. Part 3. Cape Town: Van Riebeeck Society, 1944. Muller, W. 1655 [1952]. [Report of an expedition by land.] In Bosman, D.B. & B. Thom (eds). 1952. Daghregister Gehouden by den Oppercoopman Jan Anthonisz van Riebeeck, 3 vols. Cape Town: Balkema. 416–421. Nienaber, G.S. 1963. Hottentots. Pretoria: Van Schaik. Nienaber, G.S. (ed.). 1971. Afrikaans in die Vroeër Jare. Johan­nes­burg: Voortrek­kerpers. Nienaber, G.S. 1989. Khoekhoense Stamname, ’n Voorlopige Verkenning. Prtetoria/ Kaapstad: Academica. Raven-Hart, R. (ed.). 1967. Before Van Riebeeck. Callers at South Africa from 1488 to 1652. Cape Town: Struik. Raven-Hart, R. (ed.). 1971. Cape Good Hope 1652 – 1702. The First Fifty Years of Dutch Colonisation as seen by Callers. 2 vols. Cape Town: Balkema. Smith, A. & R. Pheiffer (eds). 1992. Colonel Robert Jacob Gordon’s Notes on the Khoikhoi 1779– 80. [=] Annals of the South African Cultural History Museum 5 (1). Cape Town: South African Cultural History Museum. Tappe, D. 1704. Funffzehen-Jährige [...] Ost-Indische Reise-Beschrei­bung [...] Ha­n­no­ver and Wolfenbüttel: Gottfried Freytag. [Partial translation in Raven-Hart, R. (ed.). 1971. Cape Good Hope 1652 – 1702. The First Fifty Years of Dutch Colonisation as seen by Callers. 2 vols. Cape Town: Balkema. 236–242).] Teenstra, M.D. 1830 [1943]. De Vruchten Mijner Werk­zaam­heden, etc. First part [...] 1830. [...] Edited [...] by F.C.L. Bosman. Cape Town: Van Riebeeck Society. ten Rhyne, W. 1686 [1933]. Schediasma de Promontorio Bonae Spei; ejusve tractus incolis Hottentottis. Corrected and briefly annotated by Henr. Screta S. a Zavorziz. Schaffhausen: Joh. Mart. Oswald. – with a translation and a foreword by B. Farrington. In Schapera, I. (ed.). Early Cape Hottentots, Described in the Wri­tings of Olfert Dapper (1668), Willem ten Rhyne (1686) and Johannes Gulielmus de Grevenbroek (1695). Cape Town: Van Riebeeck Society. 78–153. Terry, E. 1622. A Voyage to East India […]. London. [Quoted via Raven-Hart, R. (ed.). 1967. Before Van Riebeeck. Callers at South Africa from 1488 to 1652. Cape Town: Struik.] Trigardt, L. 1836–1838 [1977]. Die dagboek van [–-]. In le Roux, T.H. (ed.). 1977. Die dagboek van Louis Trigardt. Uitgegee met Inleiding, Aantekeninge en Glossarium. 3rd ed. Pretoria: Van Schaik. van Meerhoff, P. 1661 [1957]. [Report of an expedition by land.] In Bosman, D.B. & B. Thom (eds). 1957. Daghregister gehouden by den oppercoopman Jan Anthonisz van Riebeeck, 3 vols. Cape Town: Balkema. 478–491.

 Hans den Besten van Overbeke, A. 1678 [1998]. Geestige en vermaeckelijcke reysbe­sc­hr­y­v­ing naer Oost-Indien. [Separate quire in van Overbeke’s Rym-wercken, publ. by J.C. ten Hoorn in Amsterdam.] In Barend-van Haeften, M. & A.J. Gelderblom (eds). Buyten Gaets. Twee Burleske Reisbrieven van Aernout van Over­beke. Hilversum: Verloren. 41–88. van Riebeeck, J. 1652–1662 [1952–1957]. Daghregister gehouden by den oppercoopman Jan Anthonisz van Riebeeck. In Bosman, D.B. & B. Thom (eds). 1952–1957. Daghregister Gehouden by den Oppercoopman Jan Anthonisz van Riebeeck, 3 vols. Cape Town: Balkema. [Treated as a sequence of eleven manuscripts] Wikar, H.J. 1779 [1935]. Berigt aan den Weleedelen Gestrengen Heer Mr. Joachim van Plettenberg […]. In Mossop, E.E. (ed.). 1935. The Journal of Hendrik Jacob Wikar (1779) [–] and The Journals of Jacobus Coetsé Jansz: (1760) and Willem van Reenen (1791) [–]. Cape Town: Van Riebeeck Society. 20–201.

Part 2

Sociohistorical contexts

Transmission and transfer Jeff Siegel University of New England (Australia) & University of Hawai‘i

1

Introduction

Whether or not there is a break in transmission of the lexifier language has joined the question of substrate influence in being among the most contentious issues in pidgin and creole studies. This chapter first examines the two main views regarding the transmission of the lexifier language in relation to creole genesis. It then looks at debates about the transmission of substrate features into creoles via language transfer. These analyses are followed by a discussion of the implications of the different views with regard to colonial and postcolonial ideologies, and current views of agency and identity in studies of second language acquisition.

2 Transmission of the lexifier 2.1

Break in transmission

The notion that creoles result from a ‘break in transmission’ of the lexifier has been around for a long time. According to this view, most if not all of the grammar of the lexifier is stripped away in pidginization, while a new grammar is reconstituted in expansion and creolization. In other words, creoles are seen as newly developed languages (see, for example, Hymes 1971; Kay & Sankoff 1974; Bickerton 1977). Thomason and Kaufman (1988: 9–11) define a ‘genetic relationship’ between languages according to historical linguistics. They present several assumptions regarding what they call ‘normal’ and ‘abnormal’ transmission. First, in normal transmission, a language is passed on from parents to children or from older to younger members of peer groups. This occurs “with relatively small degrees of change over the short run” (1988: 10). In other words, the entire language is transmitted – i.e. “a complex set of interrelated lexical, phonological, morphosyntactic and semantic structures” (1988: 11). As a language is passed on normally from generation to generation, cumulative gradual changes can eventually lead to a language that is so different that it is considered

 Jeff Siegel

a new language. So, for example, Language A can over time develop into Language B. But if the transmission has been normal, Languages A and B can be considered to have a genetic relationship. The linguistic evidence is that there are regular and systematic correspondences between the various subsytems of the languages. However, according to Thomason and Kaufman, transmission can also be “imperfect” (1988: 10), or “interrupted” (1988: 11). This may happen when “a whole population acquires a new language [i.e. a second language, L2] within possibly as little as a single lifetime, therefore necessarily other than by parental or peer-group enculturation” (1988: 10). As a result, a linguistic system may emerge with “massive interference” from the structures of the language or languages originally spoken by the population (L1 or L1s). In other words, not all aspects of the L2 have been transmitted. If this population is not integrated into the group that provided it with the L2 (e.g. Language A), their “deviant” way of speaking the L2 may crystallize into a new language (Language C). In this case, Languages A and C cannot be considered to have a genetic relationship because transmission has not been normal. The linguistic evidence is a lack of regular and systematic correspondences between the various subsystems of the languages. In fact, in Language C, it is “not possible to show that both the grammar and the lexicon derive from the same source” (1988: 11). While Thomason and Kaufman say that creoles arise from interrupted transmission, they argue against the notion that creoles necessarily arise from pre-existing pidgins (1988: 148–150). They hypothesize a process of “abrupt creolization” (1988: 150) in which an emerging contact language (or pre-pidgin) becomes the primary language of a community and is learned as a first language by children born into this community, thus rapidly expanding into a creole before first stabilizing as a pidgin. These “abrupt creoles” are basically the same as “radical creoles” (Bickerton 1981). Such creoles do not have genetic relationships with their lexifiers because of the break in transmission resulting from imperfect learning of the lexifier as a target language (TL) (1988: 152). Thomason and Kaufman do not see an abrupt creole as a changed form of the lexifier language, but rather as “an entirely new language – without genetic affiliation” (1988: 166). According to Thomason (2001: 177), the actual time required for abrupt contact language genesis is from a few years to 25 years. The terms ‘imperfect learning’ and ‘abnormal transmission’ have caused strong negative reactions in the field of pidgin/ creole studies, and more recently, Thomason (2002: 102) writes: Kaufman and I should have been more careful to spell out the precise meaning that the concept ‘genetic relationship of languages’ conveys to us and other historical linguists. The notion has nothing to do with voting rights or legitimacy, or with abnormality in the sense of wrongness (in retrospect, we shouldn’t have used an arguably pejorative term like ‘abnormal’)…

Therefore, the terms ‘break in transmission’ or ‘interrupted transmission’ will be used in the rest of this chapter.



Transmission and transfer 

Another scholar who has written in support of the notion of a break in transmission is McWhorter (1998, 2000, 2001, 2002). His view is that there are three traits which in combination define the prototypical creole: “(1) little or no inflectional affixation; (2) little or no use of tone to lexically contrast monosyllables or encode syntax; (3) semantically regular derivational affixation” (McWhorter 1998: 798). All three of these are the result of the simplification that occurred in pidginization at an earlier stage of development. In other words, creoles have these characteristics because they “emerged as radically reduced pidgins” (McWhorter 2000: 106), as the result of a break in transmission.

2.2

Normal transmission

A group of creolists working mainly on French-lexified creoles have a different point of view – especially, Chaudenson (1992, 2001, 2003), Mufwene (1996, 2000, 2001, 2004) and DeGraff (2001a, 2001b, 2003, 2005a, 2005b). These ‘superstratists’, as they are often called, believe that creoles developed gradually from their lexifiers without any significant break in transmission. For example, Chaudenson’s view is that the lexifier language was incrementally changed or ‘restructured’ to become the creole. This restructuring occurred when newly arrived slaves learned only ‘approximations’ of the colonial language from other slaves: “Creolization is thus a consequence, or the ultimate result, of approximations of approximations of the lexifier” (Chaudenson 2001: 305). Alleyne (2000) suggests that in French-lexified creoles, maximum restructuring occurred later in their historical development through cumulative divergent changes, a view that has become known as ‘gradual basilectalization’ (e.g. Mufwene 2001). The three main lines of argument for this view and against a break in transmission are: (1) that there is no evidence of a pre-existing ‘radically reduced pidgin’ for some creoles, (2) that morphology from the lexifier does exist in some creoles; and (3) that creole features result from normal (i.e. conventional) language change. 2.2.1 Lack of evidence of a pre-existing pidgin Chaudenson (1992, 2001) observes that in the early days of some plantation colonies, there were small farms or homesteads rather than large plantations, and often more indentured European workers than African slaves. Thus the slaves had access to the European language (the lexifier), and learned close approximations of it – with only minor reductions in morphology and overgeneralizations – rather than developing a radically reduced pidgin. It is clear, however, that this scenario does not apply to the early days of all creoles. For example, in Hawai‘i, there was no early period of small homesteads, and in Fiji, where there were such homesteads, the laborers learned pidginized Fijian rather than 1. DeGraff ’s view that a break in transmission is “anti-Uniformitarian”, and therefore racist, is discussed in section 4.

 Jeff Siegel

the language of the colonizers, English, and this continued to be used on the large plantations (Siegel 1987). Also, a radically reduced pidgin predecessor does exist for some creoles – for example, Hawai‘i Creole. Roberts (2004a) presents data on the early Hawai‘i Pidgin English (HPE) spoken in Honolulu from a 6,430-word corpus from 1870–1899, obtained mainly from newspapers, travelers’ accounts and court records. The features of HPE were as follows: a. no inflections on nouns or verbs:

(1) mi kamu hauki, mi papa hauki… ‘I came home to my father’s house…’ (Roberts 2004a: 168)

b. adverbs used to indicate temporal relations: Past before: (2) a.

Kiku my wife now, before Kiku and me make marry Japanese style, all same drink tea. ‘Kiku is my wife now. Kiku and I got married the Japanese way, so we drank tea (in the ceremony).’ (Roberts 2004a: 153)

b.

Garnie before eat too much Wahiawa pineapple, now get sore tooth. ‘Garnie ate so much Wahiawa pineapple that his teeth now hurt.’ (Roberts 2004a: 156)



Future by and by: (3) No got any, bimeby have some next steamer. ‘I don’t have any but I’ll have some when the next steamer comes.’ (Roberts 2004a: 153) Habitual all time:

(4) a.

What for Miss Willis laugh all time? Before Fraulein cry all time. ‘Why does Miss Willis often laugh? Fraulein used to always cry.’ (Roberts 2004a: 154)

b. Missionary alla time work, alla time say kanaka work. ‘Missionaries usually work and they tell the Hawaiians to work.’ (Roberts 2004a: 156)

2. Note that these data differ from Bickerton’s data (Bickerton & Odo 1976) from immigrants who arrived in Hawai‘i before 1935. According to Roberts (2004a: 147–148), their pidgin was typical of the plantation pidgin spoken in the 1920s and 1930s, not the HPE of Honolulu that was the forerunner of Hawai‘i Creole.



Transmission and transfer 

c. No copula

(5) a. Melican man he too much smart. ‘Americans are too smart.’ (Roberts 2004a: 149)

b.

Ae (yes), he only boy now, no got sense. By ’n by he man, he good. ‘Yes, he is just a boy now, lacking wisdom. When he is a man, he will be good.’ (Roberts 2004a: 149)

d. No existential marker

(6) Baby inside the hole, you go look, you come here. ‘There’s a baby inside the hole, come look, come here.’ (Roberts 2004a: 151)

e. Single preverbal negative marker

(7) a. I no give you slipper, baby slipper. ‘I didn’t give you that baby slipper.’ (Roberts 2004a: 150)

b. Mrs Thomas, you no tell me, me be good. ‘Mrs Thomas, you didn’t tell me to be good.’ (Roberts 2004a: 149) f. No complementizer (8)

Today go court house Ø buy license, go church Ø make marry, all same haole style. ‘Today I’m going to the court house to buy a license and going to a church to get married, just like whites do.’ (Roberts 2004a: 163)

In summary, early HPE exhibited absolute morphological simplicity – i.e. lacking any grammatical morphology (Siegel 2004a). In this way it is similar to other ‘restricted pidgins’ – i.e. pidgins that have not expanded grammatically and are used only for basic communication among people who do not share a common language. These include Chinese Pidgin English, Greenlandic Pidgin, the Hiri Trading Languages (Eleman and Koriki), Nauru Pidgin English, Ndyuka-Trio Pidgin, Pidgin Delaware, Pidgin Fijian, Pidgin Hawaiian, Pidgin Hindustani, Pidgin French of Vietnam and Russenorsk (for references, see Siegel 2006). Furthermore, in the Caribbean region, there is some indication that drastically reduced varieties were spoken early in the homestead period of different colonies – for example, the 1691 ‘merman’ text from Martinique and evidence from Suriname and Louisiana, described by McWhorter (1998: 800–803). On the other hand, there is still no evidence of a reduced pidgin predecessor for other creoles, such as Haitian Creole (DeGraff 2001a).

 Jeff Siegel

2.2.2 Existence in some creoles of morphology from the lexifier DeGraff (2001a, b) argues, contrary to McWhorter’s (1998) view, that Haitian Creole is actually rich in morphology – at least in derivational morphology, which is historically derived from French. He concludes that it must have been transmitted normally, rather than being acquired later through renewed contact with the lexifier (2001a: 84). In later work, DeGraff (2005b: 562) claims that “we find ample evidence for systematic lexical and morphosyntactic correspondences between ‘radical’ Creoles and their European lexifiers”. These issues with specific regard to Haitian Creole are debated between DeGraff (2001a, b) and McWhorter (2000, 2001). However, the waters are muddied by the fact that there is also some continuity of lexifier morphology in creoles that clearly did have a pidgin predecessor. For example, as shown above, Hawai‘i Creole was preceded by early Hawai‘i Pidgin English, which was characterized by morphological simplicity compared to its lexifier, English. The creole that emerged exhibits such simplicity in some grammatical areas – for example, in the absence of the comparative -er in the NP and of the past tense morpheme -ed in the VP. However, it also has some morphological features, not present in the preceding pidgin, that are derived from English in both form and function. These include plural marking with -s in the NP and progressive marking with -ing in the VP. While these are variable features, they are not modern innovations due to decreolization, but were present from the earliest days of the language, as shown by these examples of children’s language from the Hawaii Educational Review (1921: 11):

(9) He stay playing. ‘He is playing.’

(10) Us go push weeds? ‘Shall we go and pull weeds?’ So it seems clear that some creoles are not entirely derived directly from ‘radically reduced pidgins’ because even though they have some clear areas of morphological simplicity, they also have inherited morphological features from their lexifiers. A possible explanation for this is given below. 2.2.3 Conventional language change Along with Chaudenson, Mufwene (e.g. 2000, 2001) and DeGraff (e.g. 2001a, 2003, 2005a) argue that creoles should be treated as versions of their lexifiers that developed according to the usual processes of historical language evolution. For example, Mufwene (2004: 461) says that the genetic classification of creoles “should for all intents and purposes be Germanic or Romance”. Referring to the structural discrepancies or mismatches between two language varieties that are supposed to signal a non-genetic relation, DeGraff argues (2003: 399–400) that there are “no precise and operational structural litmus test, and no coherent theoretical tools, for deciding, on one hand where the ‘innovations’ of language change qua ‘normal transmission’ end and, on the



Transmission and transfer 

other hand, where the ‘drastic mismatches’ of creolization qua ‘abnormal transmission’ begin”. With regard to Haitian Creole (HC), he concludes (2005b: 562): In fact, core aspects in the development of HC grammar (with respect to, e.g., sound patterns, verb and object placement and inflectional morphology) fall within developmental patterns that are commonly manifested in Stammbaumtheorie-friendly instances of ‘regular’ language change (e.g., in the history of Romance and Germanic), except for the speed at which structural innovations spread within the corresponding speech communities… It could also be argued that along certain parameters – such as presence vs. absence of lexical case morphology and of movement-related properties like ‘free word order’ scrambling – French and HC are more similar to each other than French and Latin are.

Both Mufwene and DeGraff argue that the processes that led to the development of creoles are nothing out of the ordinary. Mufwene (2001: 9) says that with the large influx of slaves in the plantation stage, the “approximations of approximations” of the lexifier (as it was spoken in the colonies) occurred through typical “imperfect replication” (Lass 1997: 112) or “transmission error” (Deacon 1997: 114). This was intensified by larger proportions of “nonproficient speakers” and led to a process of gradual “basilectalization” (Mufwene 2001: 10), described in more detail as follows (ibid.: 51): After the creole populations [those born in the colony] became the minorities on the plantations, continually restructured varieties often became the models for some newcomers. This restructuring process led to the basilectalization of the colonial vernacular among its segregated users, i.e., the emergence of sociolects identified as basilectal.

But a closer look at the nature of this ‘restructuring’ reveals that both ‘imperfect replication’ or ‘transmission error’ refer to minor changes in normal first language acquisition that may gradually lead to conventional language change. Both Lass (1997: 112) and Deacon (1997: 115) say that such innovations or errors “creep into” the replication or transmission process. These are not on the scale or rapidity of changes normally referred to as restructuring in creole genesis. DeGraff ’s view (2001a, 2003) is that the morphological simplicity found in Haitian Creole is similar to that which results from language change in general, and falls within developmental patterns that are commonly attested in historical linguistics. He notes (2001a: 72) that it is “not surprising that we find that the ‘erosion’ of inflectional morphology and/ or the regularization of morphological distinctions recur in the history of any language…”, and concludes that the same mental processes are involved in the development of creoles and other situations involving language change. However, at the same time DeGraff refers to several factors that seem to exceptionalize the kind of language change found in creoles. First is the time factor. As can be seen in the quote above, he says that structural innovations may spread at greater speed in creole formation than in normal language change (see also DeGraff 2003:

 Jeff Siegel

399). Although he refers to arguments by Mufwene (2001: 130) that creoles do not develop more rapidly than other languages, it is clear that morphological simplicity in particular grammatical areas, such as verbal inflection, emerged rapidly and comprehensively in the history of creoles, rather than gradually spreading as in the structural changes usually described in historical linguistics. Second, with regard to simplification in creoles, DeGraff (2005a: 257) observes that it is not absolute: “What we are dealing with is gradient simplification with respect to the languages in contact and their respective complexity in particular domains of grammar” (emphasis in the original). Third, DeGraff (2003: 399) characterizes the kind of language change evidenced in creoles as “language change via language contact”. Furthermore, he points out a sociohistorical difference from other situations involving language change in that a larger number of language groups were in contact in the creole situation than in other situations (2003: 401). An important feature of contact-induced language is that it involves second language acquisition, and all of the proponents of the normal language evolution scenario for creole genesis appear to agree that second language acquisition plays a significant role. Although Mufwene refers to the first language acquisition processes of “imperfect replication” and “transmission error”, as mentioned above, he later says (2001: 60) that “the basilectalization process was more a by-product of imperfect acquisition of the target by second-language learners”. Chaudenson (2001, 2003) also emphasizes the importance of strategies involved in informal second language acquisition in the development of creoles. This was especially significant in the plantation stage when “approximations of approximations” of the lexifier were being made. One result of these strategies he refers to is the stripped down nature of early interlanguage – for example, indicating tense and aspect with temporal adverbs rather than verbal inflection – (see Valdman 2005: 455). In this regard, Chaudenson (2003: 190–192) describes an article by Klein and Perdue (1997) about the results of the European Science Foundation (ESF) project which took place in the 1980s (Perdue 1993). This was a 30-month longitudinal study of 40 adult immigrants with various first languages: Arabic, Italian, Finnish, Spanish and Turkish. The target languages were Dutch, English, French, German and Swedish. In their article, Klein and Perdue describe that approximately one third of the learners acquired only what they call the Basic Variety (BV). They summarize the structural features of this variety as follows: “Strikingly absent from the BV are … free or bound morphemes with purely grammatical function” (1997: 332). Features of the BV include use of temporal adverbs rather than grammatical TMA markers, no existential marker, a single word for negation, no overt complementizer, minimal pronouns, and only a few prepositions. 3. It has been suggested that structural changes could have spread more in creole communities because of their small size. However, the density and multiplexity of social networks in such small communities would have reinforced conservative norms, rather than accelerating change (Milroy 1987).



Transmission and transfer 

DeGraff (2005a: 316) also refers to the Klein and Perdue article, and in this and earlier work (1999, 2001b) he argues that second language acquisition (L2A) in the context of language contact is a crucial factor in language change. For example, he says (2005a: 316) that “the output of L2A by adults – under ‘duress’, in many cases – has a crucial role in language change, particularly in the context of language contact”. According to DeGraff, the simplification found in creoles is a result of adult second language learning, and he states: “What seems particularly affected in 2LA is the learning of inflectional paradigms…” (2005a: 316). Later, he mentions “the inflectional erosion that seem typical of language change (via L2A in contact situations)” (2005a: 335). DeGraff (2005a: 316–317) concludes: The important – if familiar, but often neglected – point here is that the nature of the PLD [primary linguistic data], obviously a key factor in language change and creation, is greatly influenced by the absence or presence of adult learners and by their cognitive and psychosocial limitations – for example, take the aforementioned morphological fossilization, which is a hallmark of adult learners’ early acquisition.

2.3

Discussion

Let us step back now and compare the positions of the two camps: interrupted transmission vs. normal transmission. Mufwene (2001: 10) argues: A break in transmission of the lexifier would have entailed no exposure to any form of the language and therefore nothing to restructure. This is quite different from the historical reality that the slaves who arrived during the plantation period were exposed to varieties more and more different from the languages brought from Europe or spoken in earlier colonial periods.

But no one has suggested that there was no exposure to any form of the lexifier language. Remember that Thomason and Kaufman (1988: 10) say that interrupted transmission occurs when “a whole population acquires a new language [i.e. a second language, L2] within possibly as little as a single lifetime, therefore necessarily other than by parental or peer-group enculturation”. And from the preceding discussion it is clear that both groups believe that processes of second language acquisition and exposure to second language versions of the lexifier are relevant to creole formation. The process discussed above, and recognized by the superstratists, is ‘simplification’, which leads to the development of second language varieties (or interlanguages) characterized by extreme formal simplicity – such as the absence of grammatical morphology described for the Basic Variety. It would seem that exposure to “varieties more and more different” from the European languages (Mufwene 2001: 10) – i.e. second language varieties characterized by extreme formal simplicity – is not the same as exposure to native varieties of the lexifier. Nevertheless, Mufwene says (2001: 76) that

 Jeff Siegel

there was “no time when the lexifier was not normally transmitted from one group to another, even as it was restructured in the process”. In summary, although the two sides seem to agree more than they disagree about what happens in creole genesis, they have differences that boil down to a matter of definition. One side believes that normal transmission entails only the thorough acquisition of a language as a mother tongue through enculturation. The other side believes that normal transmission can also entail the partial learning of a language as a second language. The first side believes that normal transmission involves only a small degree of language change, while the second believes that it can involve a larger degree. If there is a large degree of change (or restructuring) due to, for example, second language learning, the first side considers transmission to be interrupted, whereas the second side does not. In other words, if significant changes occur in the way a language is spoken due to adult second language acquisition, it is considered a break in transmission by one side but uninterrupted transmission by the other. Of course, there is no way of saying who is right and who is wrong, because there is no clear dividing line between what can be considered significant versus insignificant changes. The two camps are also closer than it may seem with regard to the issue of pidginization. For example, DeGraff (1999: 524) says that in gradual basilectalization, the L2 versions of the lexifier “became more and more pidginized via the interlanguages created in the context of successive waves of additional slave arrivals”. He says further that these “pidginized interlanguages” were a consequence of “less and less successive attempts” at second language acquisition, due to increased “social and psychological distance” between the slave population and the native speakers of the lexifier. From the discussion above, we can assume that these ‘pidginized interlanguages’ were characterized by morphological simplicity and a lack of lexifier morphology. In fact, it is hard to find a difference between the features of restricted pidgins and those of early interlanguage. For example, some readers may have noticed that the features described for early Hawai‘i Pidgin English are nearly identical to those that are typical of the Basic Variety (Klein & Perdue 1997), as described above. Thus, although DeGraff, Chaudenson, Mufwene and others may be correct in saying that there was no stable pidgin in the history of some creoles, there were clearly “pidginized interlanguages” (DeGraff 1999: 524) being used by the slaves. Again, this position is not that different from that of Thomason and Kaufman (1988: 148–150) who argue that creoles can emerge from a pidginized variety or pre-pidgin rather than necessarily from a pre-existing stable pidgin. If we accept, as both camps seem to do, that pidginized or pidgin-like varieties were the precursors of creoles, we are still left with the question of how some morphological features of the lexifier get into creoles along with the formal simplicity. One obvious possible answer is that no creole has evolved directly from a pre-pidgin, pidginized interlanguage or a stable restricted pidgin. In other words, at no stage did a group of people suddenly start speaking an unexpanded pidginized variety as their first language – even in rapid nativization. Rather, the pidginized variety was one of the many sources in the contact environment, and its simplified features entered the “pool



Transmission and transfer 

of variants” (Siegel 1997: 136) which provided the primary linguistic data (PLD) for language acquisition. There were also unsimplified features in the pool, including some morphology directly from the various varieties of the lexifier in use. As the creole emerged, some of these features were leveled out (Siegel 1997), but others were retained, and these included some simplified features of the pidginized predecessor as well as some unsimplified features from the lexifier. Thus, a creole is made up of a subset of the features that are used for communication in the language contact situation. These include features from pidginized L2 varieties of the lexifier and from the lexifier itself, and, as will be shown in the following section, features from other contact languages as well as from the substrate languages. While the two approaches may agree on the processes involved in creole genesis – in principle if not in name – they differ significantly in relation to what they concentrate on as being transmitted. The normal transmission camp focuses on the lexical forms from the superstrate language, and these clearly make up the majority of the lexicon of most if not all creoles. (Of course, that’s why the superstrate language is called the lexifier.) Other approaches focus more on grammar, and in contrast to the superstratists, they maintain that much of the grammar of creoles does not come from the lexifier. Two possible origins are usually postulated for grammatical properties of creoles that cannot be attributed to the lexifier: universal properties of language and features of the substrate languages. According to the universalist approach promoted by Bickerton (1981, 1984, 1999), there is no transmission as such, but rather the recreation of grammatical features by children according to their innate linguistic capacity, the ‘language bioprogram’. (For a critical evaluation of this approach, see Siegel 2007.) In contrast, the ‘substratists’ believe that some grammatical features of the substrate languages are transmitted to creoles. This transmission is via a process of ‘language transfer’, which is described in the following section.

3 Transmission of substrate features 3.1

Language transfer

Language transfer is a psycholinguistic process that occurs when people fall back on their first language to compensate for a lack of linguistic knowledge in a second language when they are trying to acquire or communicate in that language. In the field of second language acquisition (SLA), the term transfer refers to a form of cross-linguistic influence that involves “carrying over of mother tongue patterns into the target language” (Sharwood-Smith 1996: 71), or more accurately, into the interlanguage. Transfer in SLA may be phonological, morphosyntactic or semantic and either positive or negative. Here I consider only negative transfer – where the L1 (first language) pattern differs from that of the L2 (target or second language). I also concentrate on morphosyntactic transfer – where the grammar of the L1 is used when speaking the L2

 Jeff Siegel

– more specifically, what I call ‘functional transfer’, using L2 forms with L1 grammatical properties. Many creolists now believe that transfer of features in adult SLA provides the answer to the question of substrate features are transmitted to creoles (e.g. Mufwene 1990; Wekker 1996; Siegel 1997, 1999, 2003, 2004b, 2006; DeGraff 1999, 2001b). Other scholars have used different terms in accounting for the transmission of substrate features. For example, writing about Haitian Creole, Lefebvre (e.g. 1998) and Lumsden (e.g. 1999) use the term ‘relexification’. However, in recent work, they have shown its relationship to L1 transfer. Lefebvre (1998: 34) makes it clear that “the type of data claimed to be associated with the notion of transfer in creole genesis corresponds to the result of the process of relexification… That is, it is claimed that substratal features are transferred into the creole by means of relexification”. Lumsden (1999: 226) says that relexification “plays a significant role in second language acquisition in general” and uses the term ‘negative transfer error’ to refer to an example of the process. Migge (1998, 2000, 2003) concludes that substrate ‘retention’ was the key process in the formation of ‘Surinamese Plantation Creole’. However, she describes this process as consisting of the retention of the syntactic and semantic behavior of features of the primary substrate languages with relexification from the superstrate. This is basically the same process as transfer in SLA (see also Winford 2003: 210). Negative transfer is also referred to as ‘interference’, the term used by Thomason and Kaufman (1988: 11) in their definition of interrupted transmission. Also, Thomason (2001: 147), in her discussion of mechanisms of contact-induced language change, says that one strategy “is to maintain distinctions and other patterns from the learners’ native language (their L1) in constructing their version of TL [target language] grammar, by projecting L1 structures onto TL forms”. Again, this is another way of describing functional transfer. The usual method of trying to demonstrate the role of earlier transfer in the development of creole features is to illustrate parallels with corresponding features in one or more of the substrate languages. For example, Melanesian Pidgin (as represented by Bislama, the dialect spoken in Vanuatu) has subject referencing pronouns and transitive suffixes on verbs: (11) a. Bislama:

Man ya i still-im mane. man det 3sg -tr money ‘This man stole the money.’



Ol woman oli kat-em taro. pl woman 3pl cut-tr taro ‘The women cut the taro.’

b. Bislama:

The forms of the subject referencing pronouns are derived from English morphemes (he, all he), as are those of the transitive suffixes (him, them), but the functions are typical of the Eastern Oceanic substrate languages, such as Arosi and Kwaio:



Transmission and transfer 

(12) a. Arosi:

E noni a ome-sia i ruma art man 3sg see-tr.3sg art house ‘The man saw the house.’ (Lynch 1993: 143)

(12) b. Kwaio:

Ta’a geni la a’ari-a go’u people female 3pl carry-tr taro ‘The women carried taro.’ (Keesing 1988: 220)

There have been many criticisms of the notion of substrate influence due to earlier language transfer, most notably by Bickerton (1981, 1984), although these have been refuted in several studies (Siegel 2000, 2007). His first point is that illustrating superficial similarities between selected structures in creoles and corresponding structures in the substrate languages does not prove substrate influence because the creole rules often do not correspond exactly to the substrate rules or do not have the same distribution (Bickerton 1981: 48). For example, the transitive suffix in Melanesian Pidgin marks only transitivity in contrast to that in Arosi which marks both transitivity and the persons and number of the object. However, Thomason and Kaufman (1988: 161) argue that “point-by-point identity” with substrate features should not necessarily be expected, and scholars such as Boretzky (1993) have clearly shown that in many cases of language contact, speakers carry over only some aspects of complex rules from one language to another. Second, Bickerton says that no explanation has been given as to how substrate features are actually incorporated into creoles. He states: “In order to make a case, they [those who believe in substrate influence] have to describe exactly and explicitly how, in creolization, syntactic structures got from substratum languages into creole languages” (1992: 314, italics in the original). Granted, this is difficult to explain if one accepts Bickerton’s earlier view that creoles were created by children who were not exposed to input from the substrate languages. It is far easier to explain, however, according to Bickerton’s more recent position (1999: 55), which is that the children who created creoles, at least in Hawai‘i, did know the substrate languages. Therefore, transfer from these languages could have been possible, and transferred features could have joined the pool of variants in the contact environment. Bickerton (1999: 55) admits that transfer of parameter settings could have occurred in principle, but he says that in the case of Hawai‘i transfer did not actually occur. The evidence he gives is on the basis of the headedness parameter: Hawai‘i Creole is SVO despite the different word orders in some of the key substrate languages, namely Japanese and Filipino languages. Besides the fact that Bickerton considered only word order transfer and not functional transfer, there is another problem with his conclusion. Smith (1999: 252) refers to what he calls “Bickerton’s Edict” – that “substratophiles” need to provide evidence that “the right speakers were in the right places at the right times for features to be transmitted” (Bickerton 1984: 183). Chaudenson (2001, 2003) also insists that sociohistorical data be used to determine which substrate languages could have possibly influenced the developing creole. In the case of Hawai‘i Creole, Roberts’ detailed so-

 Jeff Siegel

ciohistocial research (1998, 1999, 2000, 2004a) clearly shows that among the immigrant population, the most important substrate languages when the creole was emerging were not Japanese or Filipino languages, but Cantonese and Portuguese. Both of these languages are SVO, like Hawai‘i Creole. Another criticism of the notion of substrate transfer is it that it does not explain why some features are transferred but others are not. Bickerton (1986: 38) states: “The gravest weakness of the substratophile position is that it is absolutely unconstrained”. And Mufwene comments: “The fact that no attempt has been made to suggest any principle regulating… a selection of substrate features is deplorable” (1990: 6). But since these criticisms were made, several studies have demonstrated ‘availability constraints’ which can predict the likelihood of a grammatical feature being transferred and ‘reinforcement principles’ which can predict the likelihood of a particular transferred feature being retained in the emerging contact language (Siegel 1997). The most important factors allowing transfer to occur appear to be perceptual salience and congruence (Siegel 1999; Aboh & Ansaldo this volume). For a substrate feature to be transferred, it must have “somewhere to transfer to” (Andersen 1983) – i.e. there must be a lexical item or morpheme in the L2 (the lexifier or expanding pidgin) that can be assigned the functions of a grammatical morpheme in the L1 (here, one of the substrate languages). This L2 form must have a function or meaning related to that of the corresponding L1 morpheme and have the same syntactic position. The absence of such a form in the L2 will constrain transfer, and thus the availability of the particular substrate feature. This constraint (along with the reinforcement principle of frequency) has been shown to account for the substrate features that are not found in Melanesian Pidgin (Siegel 1999), Tayo (Siegel, Sandeman & Corne 2000) and Australian Kriol (Munro 2004), as well as for the features that are found. Another example can be shown here with Hawai‘i Creole and one of its primary substrate languages, Cantonese. Cantonese has several aspect markers that follow the verb and indicate such categories as perfect, experiential, progressive and habitual. Hawai‘i Creole has one optional postverbal aspect marker awredi (already) that indicates perfect: (13) a. Dah buggah dead already. ‘The poor guy’s dead.’ (Pak 1998: 321)

b. Da tako no come in already Olowalu-side. ‘The octopus doesn’t come to the Olowalu area anymore.’ (Masuda 1998: 232)

In Cantonese perfect aspect (often referred to as ‘perfective’) is marked optionally with verbal suffix jó or postverbal particle dóu: (14) a. Ngóh ga chē waaih-jó. I clf car broken-pfv ‘My car’s broken down.’ (Matthews & Yip 1994: 204)



Transmission and transfer 

b. Ngóh ngām-ngām sāu dóu chin. I just-just receive v-prt money ‘I’ve just received the money.’ (Matthews & Yip 1994: 211)

It appears then that in Hawai‘i Creole, the functions of the Cantonese perfect marker have been transferred to the English-origin word already. Some supporting evidence comes from Singapore Colloquial English, which has the same perfect marker: (15) a. I only went there once or twice already. (Platt & Weber 1980: 66)

b. I work about four months already. (Bao 1995: 182)

Platt and Weber (1980: 66) show that this use of already is analogous to that of the perfect particle liaú in Hokkien, a major substrate language in Singapore: (16) Gún tháùke tńg chhù liaú. our boss return home already [prf] ‘Our boss has returned home.’ (Platt & Weber 1980: 66) Bao (1995: 185) also shows similarity with the particle le in Mandarin: (17) Ta qu niuyue le he go New York le [prf] ‘He went to New York.’ (Bao 1995: 185) The question then is why only one of the Cantonese postverbal aspect markers was transferred into the contact language that became Hawai‘i Creole. The availability constraint applied above provides the answer. In the lexifier, the adverb already shares some meaning with the Cantonese perfect marker, and comes after the verb as well; thus, there was somewhere to transfer to. But there is no common postverbal adverb in English (or in the preceding pidgin) that could be interpreted as indicating other aspectual categories such as progressive and habitual. For example, adverbs such as always and usually come before the verb. Thus there was nowhere to transfer to and these features were therefore not available to the developing creole. One of Chaudenson’s (e.g. 2001, 2003) main criticisms of the notion of transmission of substrate features via transfer is that while a particular feature of a creole may occur in a substrate language, it might also be a feature of a non-standard variety of the lexifier, or of innovations that could have occurred in the lexifier. This seems to be a valid criticism. For example, in the examples above illustrating transfer in Melanesian 4. Bao (2005: 258–263) also proposes two constraints on substrate influence, ‘system transfer’ and ‘lexifier filter’. While these can account for the transfer of the functions of the Chinese postverbal experiential aspect marker to preverbal ever in Singapore English, they cannot account for the lack of transfer of the other postverbal aspect markers into Hawai‘i Creole.

 Jeff Siegel

Pidgin, the subject indexing pronouns could have been modeled on the resumptive pronouns that are common in non-standard and indigenized varieties of English, such as Fiji English (Mugler & Tent 2004: 784): (18) a. My dad he works for FEA [Fiji Electricity Authority].

b. Some [i.e. teachers] they treat us badly.

Thus, features chosen to illustrate transfer must be clearly different from anything found in the lexifier language. Here is one such example from Hawai‘i Creole. A single lexical item expresses both negative possessives and negative existentials – nomo(a) (no more), which is related to the preverbal negator no: (19) a. How come I no more one real glove? ‘Why don’t I have a real glove?’ (Chock 1998: 29)

b. Nomo kaukau in da haus. ‘There’s no food in the house.’ (Sakoda & Siegel 2003: 83)

As Roberts (2004a: 256) points out, Cantonese has a functionally similar form móuh, derivationally related to the negator mh: (20) a. Móuh yàhn gaau ngóh Jūngmán neg.have person teach 1sg Chinese ‘There’s no one to teach me Chinese.’ (Matthews & Yip 1994: 138)

b. Ngóh móuh saai chin la wo. 1sg neg.have all money prt prt I’m out of money.’ [‘I don’t have any money.’] (Matthews & Yip 1994: 283)

Again, there is a strong possibility that the functions of Cantonese móuh have been transferred to the lexifier form no more, which has become the negative possessive and existential marker in Hawai‘i Creole. At the same time, one must be careful not to assume that if the creole uses forms of the lexifier, the grammar is the same. For example, Chaudenson (2001, 2003) argues that although French creole TMA markers have been attributed to substrate influence, they are actually derived from usages in popular French or regional dialects. For example, the preverbal imperfective marker ap is said to derive from après, as in Jean est après manger. ‘John is eating.’ (Lefebvre 1998: 113). However, scholars such as Baker (2001) and Mather (2001) criticize Chaudenson’s analysis of French creole TMA systems by pointing out that he merely examines the etymology of the markers rather than looking at the underlying syntax of the systems as a whole. As Lefebvre (1998: 114–133) describes, the semantic and structural properties of the Haitian Creole TMA markers are strikingly different from those of their French etyma, but similar to corresponding markers in the substrate languages. Not the least of these differences is that the creole TMA markers can be combined to form complex tenses, something that does not occur with these forms in French, but does occur in the substrate languages.



Transmission and transfer 

Another argument against creole features originating via transfer from the substrate languages comes again from Bickerton (1981, 1984). He maintains that certain syntactic similarities are found in creoles the world over, despite the fact that they have very different substrate languages – for example, African languages for the Atlantic creoles compared to Cantonese, Portuguese, Hawaiian, Japanese, etc. for Hawai‘i Creole. Of course, this argument says nothing about the potential transfer of features that are not shared with other creoles – for example, the negative possessive and existential marker just described for Hawai‘i Creole. However, in order to counter the ‘similar features, different substrates’ argument, one would have to show either that the different substrates have similar features (which has so far not been investigated thoroughly) or that with regard to a particular supposedly shared feature, a creole is actually more similar to a substrate language than to other creoles. This in fact has been done with Hawai‘i Creole and Portuguese with regard to the ‘nonpunctual’ marker stei (stay). According to Bickerton (1981: 58), the nonpunctual aspect particle in creoles expresses both progressive and durative when it occurs with nonstative verbs; when it occurs with adjectival stative verbs, it gives an ‘inchoative’ interpretation (1981: 69). However, in Hawai‘i Creole, stei (stay) is used to indicate progressive but not habitual, and it does not express inchoative with adjectives or stative verbs. Furthermore, unlike most other creoles, the nonpunctual particle stei (stay) has other functions – i.e. serving as a copula with adjectives and locatives and as a preverbal perfective marker. As shown in previous studies (Siegel 2000, 2003), this range of functions precisely matches that of the Portuguese verb estar, as shown in Table 1. Table 1.  Functions of Hawai‘i Creole stei compared to Portuguese estar and nonpunctual markers in most creoles

progressive marker habitual marker inchoative with statives copula with adjectives copula with locatives perfective marker

Hawai‘i Creole stei

Portuguese estar

most creoles nonpunctual

+ – – + + +

+ – – + + +

+ + + – – –

A final argument against the role of language transfer in transmitting substrate features is more problematic. Language transfer is supposed to be a process of second language acquisition (SLA). However, while substantial evidence exists about the transfer of word order in early SLA (e.g. Odlin 1990; Schwartz 1998), there is no clear evidence of the use of L2 morphemes with L1 grammatical properties – i.e. no evidence of functional transfer. For example, the European Science Foundation (ESF)

 Jeff Siegel

study mentioned earlier reported some limited cases of word order transfer but no transfer of L1 properties onto L2 forms. Chaudenson (2003: 192–193) uses this evidence to argue against the notion of the large scale transfer that is supposed to occur in relexification (e.g. Lefebvre 1998). Some studies have found similarities between features of a creole and those of learners’ interlanguage or L2 versions of the creole’s lexifier (e.g. Véronique 1994; Mather 2000), but these similarities can be attributed to simplification rather than functional transfer from the substrate. With regard to TMA markers, for example, Mather (2000: 258) refers to “the mystery” of TMA markers in French creoles being similar to those of the substrate languages while “there is very little evidence of TMA markers in any French or other European interlanguage variety”. In fact, Bardovi-Harlig (2000: 411) observes: “No significant L1 effect has been identified in the longitudinal studies of the acquisition of temporal expression”. The lack of evidence of this kind of transfer in the interlanguage of L2 learners would seem to argue against substrate features being transmitted by transfer. However, while functional transfer may not occur as the result of targeted learning of a second language, it does occur as a compensatory strategy in the use of a second language (Siegel 2003, 2004b, 2006). It is found especially when there is pressure to use the L2 to communicate more explicitly and about new topics. With regard to pidgin/ creole development, it would take place when a speaker starts extending the use of an existing pidgin into wider contexts. Here the speaker’s first language (a substrate language) is the L1, and the pidgin (not the lexifier) is the L2. While a pidgin with few grammatical resources may be fine for speakers to use for intermittent basic communication, when they need to use it more frequently for wider purposes, they find that compared to their first languages (L1), it lacks certain basic structures. Therefore they may subconsciously fall back on their L1 to expand or elaborate the grammar of the pidgin – in other words, they may transfer features from their L1 to the L2. While transfer is an individual psycholinguistic process, some of the transferred features may become accepted and used by the wider pidgin-speaking community. Thus, transfer by individuals in using a pidgin more extensively as an L2 may eventually lead to the grammatical expansion and elaboration that is typical of an expanded pidgin. In the case of Hawai‘i, for example, Roberts (2000, 2004a) demonstrates that the earlier Hawai‘i Pidgin English began to expand in the late 19th and early 20th century when it came to be used more widely by older children or adult speakers born in Hawai‘i. These speakers were bilingual in the pidgin and their first language – at that time primarily Cantonese, Portuguese or Hawaiian. If one considers the role of transfer in second language use, it is then no accident that many of the features that emerged in the expanded pidgin have striking parallels in these substrate languages (see Siegel 2005, 2006, 2007).

5. For a discussion of the differences between L2 acquisition or learning and L2 use, see Ellis (1994: 13); Kasper (1997: 310); Gass (1998: 84).



3.2

Transmission and transfer 

Substrate reinforcement

Substrate features can also be transmitted to a creole in a less direct manner than through transfer – namely, through substrate reinforcement (Siegel 1998). One way that this can occur concerns diffusion (e.g. Parkvall 1999; Siegel 2000). For example, when the use of a feature of an existing pidgin or creole spreads from one population to another, the existence of an analogous structure in a prominent substrate language in the recipient population can reinforce the use of this feature so that there is a greater chance of it remaining in the expanding pidgin and eventually ending up in the creole. One example may be the transitive suffix in Melanesian Pidgin, as illustrated in example (11) above. Baker (1993) shows that this was a feature of the Australian Aboriginal pidgin that spread to Melanesia. So it may be that this feature was reinforced by an analogous structure in the Eastern Oceanic substrate languages of Melanesia, rather than originating via transfer from these languages. In another example, like other creoles, Hawai‘i Creole uses a single lexical item get to indicate both possessives and existentials – for example: (21) get wan wahine shi get wan data ‘there is a woman who has a daughter’ (Bickerton 1981: 67) Cantonese similarly uses a single lexical item yáuh for both possessives and existentials (Siegel 2000: 213–214): (22) a. Kéuideih yáuh sāam go jái. 3pl have three clf son ‘They have three sons.’ (Matthews & Yip 1994: 279) b. Yáuh (yāt) ga chē jó-jyuh go chēut-háu. have (one) clf car block-cont clf exit-mouth ‘There’s a car blocking the exit.’ (Matthews & Yip 1994: 89) Thus, it may seem that this feature, like others described above, originated via transfer in the expanding Hawai‘i Pidgin English and ultimately ended up in the creole. However, both Chinese Pidgin English (CPE) and Pacific Pidgin English (PPE), which were brought to Hawai‘i via trading and Chinese immigration, already had a similar feature: the use of got (or alternatively in CPE, habgot) to indicate both possessive and existential. Therefore, a better explanation may be that the presence of this feature in Cantonese in Hawai‘i reinforced the use of this diffused feature. It may have been further reinforced by the other prominent substrate languages. Portuguese ter and haver are also used to indicate both possessive and existential (Siegel 2000: 214; Roberts 2004a: 254), as is Hawaiian loa‘a (Roberts 2004a: 255–256). Another way that substrate features can be indirectly transmitted to a creole is by reinforcing features that occur independently in language acquisition or language change. For example, according to Mather (2000) the use of the French postposed deictic là as a postnominal determiner that occurs in French-lexified creoles is also

 Jeff Siegel

found in the interlanguage of Ewe-speaking learners of French. Since postposed determiners are found in Ewe, this could be the result of transfer. However, Zobl (1982) shows that a similar use of là as a postnominal determiner occurs in one stage of the normal developmental sequence in the acquisition of French by speakers of languages in general that have postnominal deictic markers. Thus, the existence of a postnominal determiner in a substrate language may reinforce the continued use of feature that normally occurs in second language development but would usually be transient. With regard to language change, Chaudenson uses the term l’autorégulation (2003: 182) or ‘self-regulation’ (2001: 158) to refer to the language-internal processes, such as a move towards transparency or regularity, that diachronically lead to linguistic changes in a language. According to Chaudenson, such processes do not affect all areas of a language – only those that are sensitive to change. With respect to French he calls these areas français zéro “(F0)” (e.g. 2001: 170). His analysis of the origin of the postposed determiner la in French-lexified creoles is that this area of French grammar was subject to variation and linguistic change toward salience and invariability. The fact that Kwa languages, such as Ewe, had postnominal determiners is merely an example of a coincidental similarity between a substrate feature and the result of language change in French, and not necessarily of any direct influence. In fact, he says that “clearly there are almost no positive ‘transfers’ of obviously non-European linguistic features” (Chaudenson 2001: 148). But with regard to processes of change outside those normally found in the lexifier, Chaudenson does accept the possibility that the direction of the restructuring processes “could be partly determined by convergences with the learners’ original linguistic systems” (2001: 171). Furthermore, he says that in areas outside F0, “the slaves’ languages can be claimed to have influenced their appropriations of the target language” (2001: 183). In other words, features of the substrate languages could have reinforced particular features that emerged from both language change and the process of second language learning. Valdman (2005: 452) sums up Chaudenson’s position as follows: In the final analysis, Chaudenson considers the structure of substrate languages as serving as a filtering device that focused the attention of alloglots on certain features of the TL [target language]. For example, the fact that their L1s express verbal tense and aspect with preposed free forms led African alloglots to select French auxiliaries and modals rather than highly diverse inflected forms, which also occurred in the TL.

DeGraff (2005a: 327) appears to agree with this point of view: “The reanalysis of Fr verbal periphrases into the HC-extended VP system may have also been favored by substratal features via L2A by Kwa-speaking adults”. However, unlike Chaudenson, he appears to believe that direct transfer from the substrate languages occurs in the genesis of contact languages, including creoles:



Transmission and transfer 

As with the PLD in other situations of language contact, the PLD in the creation by Creole children of Creole I-languages would have been influenced by adult learners’ early interlanguages and the latter’s substrate-influenced innovations. (2005a: 338)

DeGraff (2001b: 258–259) says as well that “language contact also entails language transfer through second-language acquisition, which will inevitably carry over into the emerging contact language some of the complexity from the languages in contact”. In summary, the superstratists do not deny that some grammatical features from the substrate languages may be transmitted to a creole either via transfer, or less directly, by substrate reinforcement. However, they argue that the amount of transferred features is not great enough to make the creole anything else other than a version of the lexifier that has been influenced slightly by contact with other languages. As with the measurement of linguistic changes, there is no clear cutoff point regarding the quantity of transfer that would distinguish the superstratists from those who maintain that creoles are not versions of their lexifier. So again, the two approaches agree in general terms that there is substrate influence in creoles; their differences come down to a question of degree. However, despite the general agreement, the approaches are in stark contrast when examined in light of different ideologies.

4 Associated ideologies The question of transmission has been debated not only with regard to language but also with regard to wider aspects of culture – especially in the Caribbean region of the ‘New World’. The same issues are involved: whether the culture of the former colonial power has been transmitted to creole-speaking countries where it underwent changes or whether there was a break in transmission and a new culture developed with influences from various sources.

4.1

The development of post-colonial ideology in the ‘New World’

During colonial times, the paradigm of sociocultural change used to describe the history of the New World was that stronger (or ‘donor’) cultures, those of European colonial powers, dominated weaker (‘recipient’) cultures, those of indigenous people or African slaves. The result, according to this paradigm, was the obliteration of the weaker cultures and acculturation (or assimilation) to the dominant cultures. The subtext was, of course, the superiority of the dominant cultures. While it was accepted that sociocultural changes had occurred, the belief was that the European culture continued in the colonies, even though it may have been diluted to some extent. The direction of change was unidirectional – flowing from the European homeland. Permeating

 Jeff Siegel

this paradigm was also an ideology of racism: that Europeans were superior to native Americans and Africans, and that one reason European culture became diluted in the colonies was that the inferior colonized people were incapable of acquiring it fully. A new postcolonial ideology began to emerge more than 30 years ago among some Caribbean intellectuals and American historians. Most importantly it argued against the continuing domination of European culture in the New World. For example, Brathwaite (1971) emphasized that the colonial experience changed the cultures of both the colonials and their subjects, and that their cultures influenced each other. He argued that in Jamaica a new distinctive culture emerged – “neither purely British nor West African”. He called this a “creole culture” (1971: xiii), and referred to the social processes that led to this culture as “creolization” (1971: 296). Since then, the term creolization, borrowed from linguistics, has been adopted in cultural and historical studies as a term for the creation of cultures (or new aspects of culture) as a result of cultural contact. Mintz and Price (1992[1976]) also argued that there was a new creole cultural formation that drew on both European and African elements, and today there is general agreement that sociocultural creolization involves “evolution of a unique society” (Eckkrammer 2003: 105). Many scholars in the sociocultural areas assume that a new, unique language also evolves. For example, Nuttall and Michael (2000: 6) provide the following definition: Creolization has usually been understood as the process whereby individuals of different cultures, languages, and religions are thrown together and invent a new language, Creole, a new culture, and a new social organization.

In addition, there is an emphasis on the idea that a new kind of identity emerges – not merely the transformation of the European identity in the New World. For example, Arion (1998: 110) defines creolization as the “process of forging new human and cultural identities”. Rath (2000: 99) presents a similar point of view: “It is a way of forming a ‘native’ identity in a situation where there is no natal society”. The second part of Rath’s statement is crucial. Conventionally, a search for identity entails a search for roots going back to a common identity within a particular nation state. In modern ideology, this nation state is assumed to have a common standard language. Some Caribbean intellectuals, however, have reacted against this conventional view. For example, Glissant (1997[1990]) talks about “identité relationelle” – identity defined in terms of relationships between people (either individuals or groups) rather than identification with a particular nation state. Instead of referring to roots, he uses the image of a rhizome (network of roots that is always changing). In this view, identity is dynamic rather than static. He also sees identity in the Caribbean characterized not by monolingualism but by multilingualism, and not by the norms of a standard language but by the variability of a creole language (see Hofman 2003). Along these lines, Sheller(2003: 276) writes that sociocultural creolization involves an “achieved indigeneity–a new claim of belonging to a locale, but a belonging grounded



Transmission and transfer 

in movement, difference and transformation rather than stasis or permanence”. She continues (2003: 276): Becoming ‘Creole’ is a process of achieving indigeneity through the migration and recombination of diverse elements that have been loosed from previous attachments and have reattached themselves to a new place of belonging.

The notion of mixture of elements from different cultures is another important component of the concept of sociocultural creolization. For example, Trouillot (1998: 21) observes: Africans and their descendants had to create, so to speak, a new cultural world, with elements gathered from the many African cultures they came from and the European cultures of those who dominated them.

Thus, the postcolonial ideology rejects the notion that the cultures of colonized peoples were abandoned. Rather, these cultures are part of the mixture from which the new culture emerges. Some of the earliest work in sociocultural creolization followed the debates about the influence of African culture on African American culture. Frazier (e.g. 1948[1939]) argued that African cultural traditions had been obliterated by slavery; Herskovits (e.g. 1958[1941]) disagreed, describing many American practices with African origins. In this light, Brathwaite (1971: 231) argued that: the African influence remained, even if increasingly submerged, as an important element in the process of creolization. European adaptations or imitations could never be whole-hearted or complete. There might be apparent European forms, but the content would be different.

Mintz and Price (1992[1976]) maintained that while there are no direct continuities between African culture and the new African American culture, there are some similarities based on preserved underlying values and beliefs, which they compared to “unconscious ‘grammatical principles’” (1992: 9–10). Rath (2000: 103) comments: Mintz and Price contended that values and beliefs, because they existed at a deeper level, had a better chance of surviving the middle passage and planters’ culturestripping strategies than did material objects and surface expressions. Creolized cultures, they contended, were like creolized languages, in that African-derived deep structures survived even where the various lexicons and surface expressions might not. [italics in original]

Another important part of the post-colonial ideology is agency: that slaves and their descendants were not passive participants in history, compliantly assimilating to the

 Jeff Siegel

dominant culture, but that they had some agency – i.e. “the socioculturally mediated capacity to act” (Ahearn 2001: 112). For example, Berlin (1998: 2) points out that: slaveholders severely circumscribed the lives of enslaved peoples, but they never fully defined them. Slaves were neither extensions of their owners’ will nor products of the market’s demand. The slaves’ history – like all human history – was made not only by what was done to them but also by what they did for themselves.

Berlin continues (1998: 5): “Although much of slave life took shape beyond their masters’ eyes from sundown to sunup, slaves also created their own world under the owners’ noses from sunup to sundown.” This included the creation of a new “nationality” (1998: 105). Thus, Trouillot (1998: 9) points out that we need to give “due credit to the creativity that Africans and their descendants demonstrated right from the beginning of plantation slavery”. This creativity is also seen to apply to language. As mentioned earlier, the colonial view of history was that slaves and their descendants lost their language along with their culture, and adopted (or tried to adopt) that of the dominant Europeans. For example, Trouillot observes (1998: 10–11): “Up to the second part of this century, most observers and many speakers viewed the creole languages of the Caribbean as burlesque versions of European tongues”. Condé (1998: 102) similarly points out that: when Creole became widespread in each island, at its outset, it was not perceived as a unique linguistic creation, but rather as a distortion, a perversion of the model of the European colonizers’ language.

But the postcolonial view, as articulated by Caribbean scholars, sees creole languages not as distorted versions of the colonial language, but as new languages, created by their speakers – for example: [O]nce taken over by the slaves and their descendants, European languages did not remain the same. They acquired sounds, morphological and syntactic patterns unknown in Europe. More important, they were shaped to express joys, pains and reflections of hundreds of thousands of humans. In one word they were creolized. (Trouillot 1998: 11)

Caribbean scholars see agency manifested not only in creativity but also in resistance to the dominant culture of the colonial power. For example, Brathwaite (1977: 55) writes: [T]he idea of creolization as an ac/culturative, even interculturative process between ‘black’ and ‘white’, with the (subordinate) black absorbing ‘progressive’ ideas and technology from the white, has to be modified into a more complex vision in which appears the notion of negative or regressive creolization: a self-conscious refusal to borrow or be influenced by the Other, and a coincident desire to fall back upon, unearth, recognize elements in the maroon or ancestral culture that will preserve or apparently preserve the unique identity of the group. (quoted in Gundaker 1998: 18)



Transmission and transfer 

Some consider the resistance found in sociocultural creolization to be oppositional in nature – for example, “a process of differentiation from the metropole through the development of a specific view and way of life” (Hoffman 2003: 11). Others, however, see it as a struggle against colonial power – especially with regard to language. Hofman (2003: 282) describes Glissant’s (1997 [1990]) belief that “creole is not the result of restricted input, but the product of strategies of resistance”, and that the formation of a creole is “creative appropriation”. Balutansky and Sourieau (1998: 10) refer to “the subversive process of appropriating the inherited ‘master’ language and turning it into their own expressions of identity”. They also mention that the Haitian scholar Maryse Condé “reaffirms Creole languages as linguistic subversion as well as resistance to colonial oppression” (1998: 10). Thus, according to Eckkrammer (2003: 98), a creole language “constitutes the most manifest symbol of the phenomenon [of creolization] and suggests a unique opportunity for freeing cultures from colonial or postcolonial oppression by linguistic means”.

4.2

Discussion

It would appear that the uninterrupted transmission camp does not really fit in with the postcolonial ideology. Most significant is its view about the linear development of the lexifier (i.e. the colonial language) into a creole language by gradual restructuring. As we have seen, Chaudenson, Mufwene and DeGraff all consider French creoles to be varieties of French. Chaudenson (2001: 46) writes that in the time of slavery, “social integration occurred rapidly and, as dreadful as it may sound, efficiently”. With regard to ‘cultural systems’ in general, including language, he stresses that “one is struck by the role of the European model in all cases” (2001: 305), and with regard to the Frenchlexified creoles, he asserts that the “basic linguistic materials obviously originated from French” (2001: 145). Although Chaudenson might not agree, his work is referred to extensively by neocolonialists such as Métellus (1998), who says that “the creole spoken in the Antilles and the Indian Ocean is a French language” (1998: 123), and therefore rejects the connection between Creole and Haitian identity: Our creole truly stems from the French language and was first created by whites to simplify their own verbal communications before imposing it on the slaves. Creole does not enable any Haitian to legitimate his or her roots since it does not represent the Haitian’s indigenous culture. (1998: 122)

Chaudenson’s dismissal of the possibility of direct transmission of substrate features via transfer, as described above, also does not mesh with postcolonial ideology. For

 Jeff Siegel

one thing, he appears to use the view of unidirectional domination to reject the general notion of linguistic mixing in creoles – for example (2001: 305): The theory that views linguistic creolization as simply a ‘mix’ of coexistent linguistic systems is not consistent with the most common linguistic reality. The constant outcome of the contact of two languages in the same community is much more the domination of one by the other than a harmonious mix. This is even more so in the colonial societies where creoles developed.

In addition, Chaudenson (2001) expresses the view that the African language and culture of the slaves were erased by the conditions of slavery, referring to the “rapid disappearance of slaves’ ancestral languages” (2001: 309). Again, Métellus (1998: 119) follows suit by alluding to “the virtual oblivion of African languages” in Haiti and “the total and permanent erasure of identities”. Like Chaudenson, Mufwene (2000, 2004) is opposed to the view that creole languages are unique creations rather than versions of their lexifiers. He questions “linguists’ self-license to go around the world baptizing some vernaculars ‘creoles’, when in some cases their speakers do not even know the word creole, let alone its meaning in linguistics” (2000: 67). He calls this “the disfranchising act by which some vernaculars are marginalized from other normal, natural developments of their lexifiers”. While there is merit in saying that colonized people should not be stigmatized by linguistic labeling, these particular views would come as some surprise to the many Caribbean intellectuals mentioned above who have adopted the notion of creolization and made creole languages central to their characterizations of creole identity or créolité. Similarly, DeGraff (2003: 398) declares that ‘broken transmission’ is one of the “myths [that] turn creoles into languages with no historical past”. He calls for a “postcolonial creolistics” aiming to “fully deconstruct” such fallacies of “creole exceptionalism”. Yet, the idea of new languages and cultures without a clear link to the past was used by the Caribbean intellectuals as a “powerful tool for political critique of colonialism and imperialism” (Sheller 2003: 279). As just described, this post-colonial movement emphasized the creation of unique languages, cultures and identities, not pale versions of those from Europe supposedly transmitted (unbrokenly) to the colonies.

6. There appear to be contradictions in Chaudenson’s positions. While he asserts that the slaves’ languages rapidly disappeared, he also says, as mentioned earlier, that they “can be claimed to have influenced their appropriations of the target language” (2001: 183). Similarly, in his discussion of ‘cultural creolization’ (1992, 2001) (which surprisingly does not refer to any of the works in the sociocultural paradigm described above), Chaudenson takes the view that French culture and religion strongly dominated the development of folk medicine, oral literature and even voodoo. At the same time, he allows for mixing and non-European influence in music, creole oral literatures and cuisine. If such mixing can occur in these cultural areas, it is not clear why it could not occur in language.



Transmission and transfer 

DeGraff also asserts that the notion of broken transmission is part of a racist ideology that has permeated creole studies – for example (DeGraff 2005b: 298): Most contemporary linguists seem to have abandoned the explicitly racist claims of the colonial era. Yet one widely held dogma in historical linguistics still considers Creole languages to have emerged through ‘broken’, thus ‘abnormal’, transmission. The postulation of such extraordinary ‘break in transmission’ has traditionally forced Creole languages in an exceptional class – namely, the class of ‘non-genetic [i.e. parentless] languages’.

His accusations of racist dogma are based on assertions that the “affixless pidgin” scenario described here in section 1 is “anti-Uniformitarian”. This is because it “makes early ‘prototypical’ Creole speakers unlike any other speakers” in that they are “incapable of pattern matching and of detecting morphological and semantic relationships in the primary linguistic data” (DeGraff 2001a: 85). DeGraff goes on to juxtapose the “catastrophic genesis” of creoles with the racist belief that their genesis represents “failures on the part of ‘inferior’ beings to acquire ‘superior’ languages” (2001a: 90). He then tries to show that the simplification proposed to result from a “radical break in transmission” is extraordinary and unnatural, and therefore that its proponents are implying that Creole speakers “form a biological/ cognitive class of their own” (2001a: 92), thus contributing to the racist ideology that Creole speakers are intellectually inferior. However, it has been shown that pidgins do exist which are drastically simplified in comparison to their lexifiers – for example, early Hawai‘i Pidgin English. Modern creolists depict the origins of such pidgins, not as the result of any shortcoming in language learning capacity, but as the result of sociohistorical circumstances, such as lack of access to the lexifier or the urgent need for a medium of communication. Furthermore, since such pidgins are spoken only as an auxiliary language and their speakers already have a ‘normal’ first language, there is nothing about this situation that implies a lack of the basic human capacity for language. When scholars such as McWhorter (2000: 106) say “creoles are usually born as pidgins”, they do not mean that creole speakers acquire only a drastically reduced pidgin as their first language. By definition, creoles have expanded and more complex grammars, and while they still may not have the amount of bound morphology found in their lexifier, they use other grammatical means (often periphrastic constructions) to mark grammatical categories, for example in TMA. It seems ironic that some creolists in the uninterrupted transmission camp have come to view creolization in a negative way while some of their creole-speaking counterparts in other fields think of it so positively. For example, Mufwene (2004: 461) says that the term creolization is a misnomer that reflects “negative colonial biases toward non-European populations and their ways of speaking languages”, while Trouillot’s (1998: 8) view is that “creolization is a miracle begging for analysis”.

 Jeff Siegel

4.3

‘Imperfect’ learning

Before finishing this chapter, one prevailing viewpoint found in both accounts of transmission and transfer needs to be critically examined. This is the notion that unsuccessful or ‘imperfect’ learning of the lexifier as a second language is a crucial factor in creole genesis, and that this was generally the result of restricted access to the language as it was spoken by native speakers. Thus, creoles result from circumstances beyond the control of their originators, and not because they chose to act in certain ways. In other words, they are passive rather than active participants in the process. Again, this conflicts with the postcolonial ideas described above about agency and identity formation. However, some recent perspectives in the field of second language acquisition point out that learners often do have agency (e.g. Norton Peirce 1995; McKay & Wong 1996; Ibrahim 1999; Miller 2000; Norton 2000). Pavlenko (2002: 293) puts it this way: While various sociopsychological theories view L2 learners as members of homogeneous groups and as passive recipients of input and output, poststructuralist L2 users are portrayed as agents in learning: in many cases they may decide to learn the second, or additional language, only to the extent that it allows them to be proficient…

This was shown in the work of Klein and Perdue (1997), for example, where speakers of the ‘Basic Variety’ learned enough of the target languages to be able to communicate but went no further, despite opportunities for further exposure to the languages. As Rampton (1997: 294) observes: “People are not always concerned with improving their L2 interlanguage”. An ‘imperfect’ variety may be used to express a particular identity of the speaker, to show solidarity with a peer group or to indicate attitudes towards society. For example, research by Ohara (2001) describes female learners who, after four years of studying Japanese and at least one year in Japan, ‘failed’ to acquire the conventional use of high pitch to express politeness and femininity. Interviews revealed that this was the result of a “rational and conscious decision based on their own perceptions of the language habits of Japanese women and their greater social implications” (2001: 244). For example, one interviewee said: “I just don’t want to sound like them.” Firth and Wagner (1997: 292) also point out that non-native-like structures may be “deployed resourcefully and strategically to accomplish social and interactional ends”. Furthermore, the decision not to use native-like L2 forms or not to use the L2 at all may represent a form of resistance, which is a kind of communication strategy (Rampton 1991: 239). It follows, then, that in many situations native-like proficiency is not the target of language learning. Thus a deliberate decision, whether conscious or unconscious, may be responsible for ‘imperfect’ second language acquisition exhibited in pidgins – not the lack of success or failure either stated or implied in most creolist literature, as criticized by Baker (1994). As Sebba (1997: 79) remarks: “A more pragmatic view would be that pidgins



Transmission and transfer 

represent successful second language learning from the point of view of their learners – who learn just enough to communicate what they want to communicate and no more” (emphasis in original). So rather than ‘imperfect’ SLA, a preferable term would be one such as ‘suboptimal’ acquisition (Dahl 2004: 110) or ‘strategic SLA’ (Siegel 2005). Another aspect of this strategic SLA noted by Pavlenko (2002: 285) is that “L2 learners may opt for constructing new and mixed identities”. Transfer of L1 features is one way of indexing identity and differentiating speakers from the L2 group. For example, Færch and Kasper (1987: 124) point out: In certain types of IL [interlanguage] communication, ‘low-prestige language varieties’ may be interpreted as IL varieties with high transfer load. The clearest case for this phenomenon of transfer is ethnic minority groups marking their groupmembership by preserving features of the L1 when using the dominant L2.

The relationship between identity and creole formation has been considered by a few creolists – for example, Le Page and Tabouret-Keller (1985) and Roberts (2000, 2004a, 2004b), but it would seem to deserve more attention.

5 Conclusion This chapter has shown that there is general agreement among creolists that second language learning plays a crucial role in the development of creole languages. While differences clearly exist with regard to the notion of a break in transmission of the lexifier and the transfer of substrate features, they are basically a matter of degree and definition. Nevertheless, these relatively minor differences have major ideological implications. Finally, the agreed upon paradigm of second language learning used in creole studies needs to be re-evaluated so that learners are recognized as active participants in the creation of creole languages and new social identities.

References Ahearn, L. M. 2001. Language and agency. Annual Review of Anthropology 30. 109–137. Alleyne, M.C. 2000. Opposite processes in “creolization”. In Neumann-Holzschuh, I. & E.W. Schneider (eds). Degrees of Restructuring in Creole Languages. Amsterdam/ Philadelphia: John Benjamins. 125–133. Andersen, R.W. 1983. Transfer to somewhere. In Gass, S.M. & L. Selinker (eds). Language Transfer in Language Learning. Rowley, MA: Newbury House. 177–201. Arion, F.M. 1998. The victory of the concubines and the nannies. In Balutansky, K.M. & M.-A. Sourieau (eds). Caribbean Creolization: Reflections on the Cultural Dynamics of Language, Literature, and Identity. Gainsville: University Press of Florida. 110–117. Baker, P. 1993. Australian influence on Melanesian Pidgin English. Te Reo 36. 3–67.

 Jeff Siegel Baker, P. 1994. Creativity in creole genesis. In Adone, D. & I. Plag (eds). Creolization and Language Change. Tübingen: Max Niemeyer. 65–84. Baker, P. 2001. No creolisation without prior pidginisation? In Crowley, T. & J. Siegel (eds). Studies in creole linguistics in memory of Chris Corne 1942–1999. [Te Reo 44.] Auckland: Linguistic Society of New Zealand. Balutansky, K.M. & M.-A. Sourieau. 1998. Introduction. In Balutansky, K.M. & M.-A. Sourieau (eds). Caribbean Creolization: Reflections on the Cultural Dynamics of Language, Literature, and Identity. Gainsville: University Press of Florida. 1–11. Bao Z.-M. 1995. Already in Singapore English. World Englishes 14 (2). 181–188. Bao Z.-M. 2005. The aspectual system of Singapore English and the systemic substratist explanation. Journal of Linguistics 41. 237–267. Bardovi-Harlig, K. 2000. Tense and Aspect in Second Language Acquisition: Form, Meaning, and Use. Oxford: Blackwell. Berlin, I. 1998. Many Thousands Gone. The First Two Centuries of Slavery in North America. Cambridge, MA: Belknap Press of Harvard University Press. Bickerton, D. 1977. Pidginization and creolization: Acquisition and language universals. In Valdman, A. (ed.). Pidgin and Creole Linguistics. Bloomington, IN: Indiana University Press. 49–69. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Bickerton, D. 1984. The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7. 173–221. Bickerton, D. 1986. Creoles and West African languages: A case of mistaken identity. In Muysken, P. & N. Smith (eds). Substrata Versus Universals in Creole Genesis. Amsterdam: John Benjamins. 25–40. Bickerton, D. 1992. The sociohistorical matrix of creolization. Journal of Pidgin and Creole Languages 7 (2). 307–318. Bickerton, D. 1999. How to acquire language without positive evidence: What acquisitionists can learn from creoles. In DeGraff, M. (ed.). Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press. 49–74. Bickerton, D. & C. Odo. 1976. Change and Variation in Hawaiian English, Vol. 1: General Phonology and Pidgin Syntax. Honolulu: Social Sciences and Linguistics Institute, University of Hawaii. Boretzky, N. 1993. The concept of rule, rule borrowing, and substrate influence in creole languages. In Mufwene, S.S. (ed.). Africanisms in Afro-American Language Varieties. Athens: University of Georgia Press. 74–92. Brathwaite, E.K. 1971. The Development of Creole Society in Jamaica, 1770–1820. Oxford: Clarendon. Brathwaite, E.K. 1977. Nanny, Same Sharpe, and the Struggle for People’s Liberation. Kingston, Jamaica: API for the National Heritage Week Committee. Chaudenson, R. 1992. Des Îles, des Hommes, des Langues. Paris: L’Harmattan. Chaudenson, R. (revised in collaboration with S.S. Mufwene). 2001. Creolization of Language and Culture. London: Routledge. Chaudenson, R. 2003. La Créolisation: Théorie, Applications, Implications. Paris: L’Harmattan. Chock, E. 1998. Da glove. In Chock E., J.R. Harstad, D.H.Y. Lum & B. Teter (eds). Growing Up Local: An Anthology of Poetry and Prose from Hawai‘i. Honolulu: Bamboo Ridge. 28–29.



Transmission and transfer 

Condé, M. 1998. Créolité without the creole language? In Balutansky, K.M. & M.-A. Sourieau (eds). Caribbean Creolization: Reflections on the Cultural Dynamics of Language, Literature, and Identity. Gainsville: University Press of Florida. 101–109. Dahl, Ö. 2004. The Growth and Maintenance of Linguistic Complexity. Amsterdam/ Philadelphia: John Benjamins Deacon, T.W. 1997. The Symbolic Species: The Co-evolution of Language and the Brain. New York: Norton. DeGraff, M. 1999. Creolization, language change, and language acquisition: An epilogue. In DeGraff, M. (ed.). Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA/ London: The MIT Press. 473–543. DeGraff, M. 2001a. Morphology in Creole Genesis: Linguistics and Ideology. In Kenstowicz, M. (ed.). Ken Hale: A Life in Language. Cambridge, MA: MIT Press. 53–121. DeGraff, M. 2001b. On the origins of creoles: A Cartesian critique of Neo-Darwinian linguistics. Linguistic Typology 5 (2/3). 213–310. DeGraff, M. 2003. Against Creole exceptionalism. Language 79 (2). 391–410. DeGraff, M. 2005a. Linguists’ most dangerous myth: The fallacy of Creole Exceptionalism. Language in Society 34. 533–591. DeGraff, M. 2005b. Morphology and word order in “creolization” and beyond. In Cinque, G. & R.S. Kayne (eds). The Oxford Handbook on Comparative Syntax. Oxford: Oxford University Press. 293–372. Eckkrammer, E.M. 2003. On the perception of “Creole” language and identity in the Netherlands Antilles. In Collier, G. & U. Fleischmann (eds). A Pepper-Pot of Cultures. Aspects of Creolization in the Caribbean. [Matutu 27–28.] Amsterdam/ New York: Editions Rodopi, 85–108. Ellis, R. 1994. A theory of instructed second language acquisition. In Ellis, N.C. (ed.). Implicit and Explicit Learning of Languages. London: Academic Press. 79–114. Færch, C. & G. Kasper. 1987. Perspectives on language transfer. Applied Linguistics 8 (2). 111–136. Firth, A. & J. Wagner. 1997. On discourse, communication, and (some) fundamental concepts in SLA research. The Modern Language Journal 81 (3). 285–300. Frazier, E. F.. 1948[1939]. The Negro Family in the United States. Chicago: University of Chicago Press. Gass, S. 1998. Apples and oranges: Or, why apples are not orange and don’t need to be. A response to Firth and Wagner. The Modern Language Journal 82 (1). 83–90. Glissant, É. 1997[1990]. Poetics of Relation [Poétique de la Relation]. Translated by B. Wing. Ann Arbor: University of Michigan Press. Gundaker, G. 1998. Literacies, Creolization, and Vernacular Practice in African America. New York/ Oxford: Oxford University Press. Hawaii Educational Review. 1921. Herskovits, M.J. 1958[1941]. The Myth of the Negro Past. Boston: Beacon Press. Hoffman, L.-F. 2003. Creolization in Haiti and national identity. In Collier, G. & U. Fleischmann (eds). A Pepper-Pot of Cultures. Aspects of Creolization in the Caribbean. [Matutu 27–28.] Amsterdam/ New York: Editions Rodopi. 3–16. Hofman, S. 2003. Against monolingualism: Roots, rhizomes and the conceptualization of creole language. In Collier, G. & U. Fleischmann (eds). A Pepper-Pot of Cultures. Aspects of Creolization in the Caribbean. [Matutu 27–28.] Amsterdam/ New York: Editions Rodopi. 277–285. Hymes, D. 1971. Introduction [to Part III]. In Hymes, D. (ed.). Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. 65–90.

 Jeff Siegel Ibrahim, A. 1999. Becoming Black: Rap and hip-hop, race, gender, identity, and the politics of ESL learning. TESOL Quarterly 33 (3). 349–369. Kasper, G. 1997. “A” stands for acquisition: A response to Firth and Wagner. The Modern Language Journal 81 (3). 307–312. Kay, P. & G. Sankoff. 1974. A language universal approach to pidgins and creoles. In DeCamp, D. & I.F. Hancock (eds). Pidgins and Creoles: Current Trends and Prospects. Washington, DC: Georgetown University Press. 61–72. Klein, W. & C. Perdue. 1997. The basic variety (or: Couldn’t natural languages be much simpler?). Second Language Research 13 (4). 301–347. Lass, R. 1997. Historical Linguistics and Language Change. Cambridge: Cambridge University Press. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Le Page, R. B. & A. Tabouret-Keller. 1985. Acts of Identity: Creole-based Approaches to Language and Ethnicity. Cambridge: Cambridge University Press. Lumsden, J.S. 1999. Language acquisition and creolization. In DeGraff, M. (ed.). Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA/ London: The MIT Press. 129–157. Lynch, J. 1993. Pacific Languages: An Introduction. Suva, Fiji: University of the South Pacific. Masuda, B. 1998. No mo’ fish on Maui. In Chock, E., J.R. Harstad, D.H.Y. Lum & B. Teter (eds). Growing Up Local: An Anthology of Poetry and Prose from Hawai‘i. Honolulu: Bamboo Ridge. 232. Mather, P.-A. 2000. Creole genesis: Evidence from West African L2 French. In Gilbers, D.G., J. Nerbonne & J. Schaeken (eds). Languages in Contact. Amsterdam/ Atlanta: Rodopi. 247–261. Mather, P.-A. 2001. On the origin and linguistic status of Réunionnais. In Crowley, T. & J. Siegel (eds). Studies in Creole Linguistics in Memory of Chris Corne 1942–1999. [Te Reo 44.] Auckland: Linguistic Society of New Zealand. 85–109. Matthews, S. & V. Yip. 1994. Cantonese: A Comprehensive Grammar. London/ New York: Routledge. McKay, S.L. & S.-L.C. Wong. 1996. Multiple discourses, multiple identities: Investment and agency in second-language learning among Chinese adolescent immigrant students. Harvard Educational Review 66. 577–608. McWhorter, J.H. 1998. Identifying the creole prototype: Vindicating a typological class. Language 74. 788–818. McWhorter, J.H. 2000. Defining “creole” as a synchronic term. In Neumann-Holzschuh, I. & E. W. Schneider (eds). Degrees of Restructuring in Creole Languages. Amsterdam/ Philadelphia: John Benjamins. 85–123. McWhorter, J.H. 2001. The world’s simplest grammars are creole grammars. Linguistic Typology 5 (2/3). 125–166. McWhorter, J.H. 2002. The rest of the story: Restoring pidginization to creole genesis theory. Journal of Pidgin and Creole Languages 17 (1). 1–48. Métellus, J. 1998. The process of creolization in Haiti and the pitfalls of the graphic form. In Balutansky, K.M. & M.-A. Sourieau (eds). Caribbean Creolization: Reflections on the Cultural Dynamics of Language, Literature, and Identity. Gainsville: University Press of Florida. 118–128.



Transmission and transfer 

Migge, B. 1998. Substrate influence in creole formation: The origin of give-type serial verb constructions in the Surinamese Plantation Creole. Journal of Pidgin and Creole Languages 13 (2). 215–265. Migge, B. 2000. The origin of the syntax and semantics of property items in the Surinamese Plantation Creole. In McWhorter, J.H. (ed.). Language Change and Language Contact in Pidgins and Creoles. Amsterdam/ Philadelphia: John Benjamins. 201–234. Migge, B. 2003. Creole Formation as Language Contact. Amsterdam/ Philadelphia: John Benjamins. Miller, J. 2000. Language use, identity and social interaction: Migrant students in Australia. Research on Language and Social Interaction 33 (1). 69–100. Milroy, L. 1987. Language and Social Networks. 2nd ed. Oxford: Blackwell. Mintz, S. & R. Price. 1992[1976]. Birth of African-American Culture: An Anthropological Perspective. Boston: Beacon Press. Mufwene, S.S. 1990. Transfer and the Substrate Hypothesis in creolistics. Studies in Second Language Acquisition 12. 1–23. Mufwene, S.S. 1996. The Founder Principle in creole genesis. Diachronica 13 (1). 83–134. Mufwene, S.S. 2000. Creolization is a social, not a structural, process. In Neumann-Holzschuh, I. & E.W. Schneider (eds). Degrees of Restructuring in Creole Languages. Amsterdam/ Philadelphia: John Benjamins. 65–84. Mufwene, S.S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mufwene, S.S. 2004. Multilingualism in linguistic history: Creolization and indigenization. In Bhatia, T.K. & W.C. Ritchie (eds). The Handbook of Biulingualism. Oxford: Blackwell. 460–488. Mugler, F. & J. Tent. 2004. Fiji English: Morphology and syntax. In Kortmann, B. & E.W. Schneider (eds). A Handbook of Varieties of English, Volume 2: Morphology and Syntax. Berlin: Mouton de Gruyter. 770–788. Munro, J.M. 2004. Substrate Influence in Kriol: The Application of Transfer Constraints to Language Contact in Northern Australia. Ph.D. dissertation., University of New England, Australia. Norton, B. 2000. Identity and Language Learning: Gender, Ethnicity, and Educational Change. London: Longman. Norton Peirce, B. 1995. Social identity, investment, and language learning. TESOL Quarterly 29 (1). 9–31. Nuttall, S. & C.-A. Michael. 2000. Introduction: Imagining the present. In Nuttall, S. & C.-A. Michael (eds). Senses of Culture: South African Culture Studies. Oxford: Oxford University Press. 1–23. Odlin, T. 1990. Word-order transfer, metalinguistic awareness and constraints on foreign language learning. In Van Patten, B. & J.F. Lee (eds). Second Language Acquisition – Foreign Language Learning. Clevedon, Avon: Multilingual Matters. 95–117. Ohara, Y. 2001. Finding one’s voice in Japanese: A study of pitch levels of L2 learners. In Pavlenko, A., A. Blackledge, I. Piller & M. Teutsch-Dwyer (eds). Multilingualism, Second Language Learning, and Gender. Berlin: Mouton de Gruyter. 231–254. Pak, G. 1998. The valley of the dead air. In Chock, E., J.R. Harstad, D.H.Y. Lum & B. Teter (eds). Growing Up Local: An Anthology of Poetry and Prose from Hawai‘i. Honolulu: Bamboo Ridge. 319–327.

 Jeff Siegel Parkvall, M. 1999. Feature selection and genetic relationships among Atlantic creoles. In Huber, M. & M. Parkvall (eds). Spreading the Word: The Issue of Diffusion among Atlantic Creoles. [Westminster Creolistics Series 6.] London: University of Westminster. Pavlenko, A. 2002. Poststructuralist approaches to the study of social factors in second language learning and use. In Cook, V. (ed.). Portraits of the L2 User. Clevedon: Multilingual Matters. 277–302. Perdue, C. (ed.). 1993. Adult Language Acquisition: Cross-Linguistic Perspectives. Vol. II: The Results. Cambridge: Cambridge University Press. Platt, J. & H. Weber. 1980. English in Singapore and Malaysia: Status, Features, Functions. Kuala Lumpur: Oxford University Press. Rampton, B. 1997. A sociolinguistic perspective on L2 communication strategies. In Kasper, G. & E. Kellerman (eds). Communication Strategies: Psycholinguistic and Sociolinguistic Perspectives. London/ New York: Longman. 279–303. Rampton, M.B.H. 1991. Second language learners in a stratified multilingual setting. Applied Linguistics 12. 229–248. Rath, R.C. 2000. Drums and power: Ways of creolizing music in coastal South Carolina and Georgia, 1730–90. In Buisseret, D. & S.G. Reinhardt (eds). Creolization in the Americas. Arlington: University of Texas at Arlington. 99–130. Roberts, S.J. 1998. The role of diffusion in the genesis of Hawaiian Creole. Language 74. 1–39. Roberts, S.J. 1999. The TMA system of Hawaiian Creole and diffusion. In Rickford, J.R. & S. Romaine (eds). Creole Genesis, Attitudes and Discourse: Studies Celebrating Charlene J. Sato. Amsterdam/ Philadelphia: John Benjamins. 45–70. Roberts, S.J. 2000. Nativization and genesis of Hawaiian Creole. In McWhorter, J.H. (ed.), Language Change and Language Contact in Pidgins and Creoles. Amsterdam/ Philadelphia: John Benjamins. 257–300. Roberts, S.J. 2004a. The Emergence of Hawai‘i Creole English in the Early 20th Century: The Sociohistorical Context of Creole Genesis. Ph.D. dissertation, Stanford University. Roberts, S.J. 2004b. The role of style and identity in the development of Hawaiian Creole. In Escure, G. & A. Schwegler (eds). Creoles, Contact, and Language Change. Amsterdam/ Philadelphia: John Benjamins. 333–352. Sakoda, K. & J. Siegel. 2003. Pidgin Grammar: An Introduction to the Creole Language of Hawai‘i. Honolulu: Bess Press. Schwartz, B.D. 1998. The second language instinct. Lingua 106. 133–160. Sebba, M. 1997. Contact Languages: Pidgins and Creoles. New York: St Martin’s Press. Sharwood-Smith, M. 1996. Crosslinguistic influence with special reference to the acquisition of grammar. In Jordens, P. & J. Lalleman (eds). Investigating Second Language Acquisition. Berlin: Mouton de Gruyter. 71–83. Sheller, M. 2003. Creolization in discourses of global culture. In Ahmed, S., C. Castañeda, A.-M. Fortier & M. Sheller (eds). Uprootings/ Regroundings. Questions of Home and Migration. Oxford/ New York: Berg. 273–294. Siegel, J. 1987. Language Contact in a Plantation Environment: A Sociolinguistic History of Fiji. Cambridge: Cambridge University Press. Siegel, J. 1997. Mixing, leveling and pidgin/ creole development. In Spears, A.K. & D. Winford (eds). The Structure and Status of Pidgins and Creoles. Amsterdam/ Philadelphia: John Benjamins. 111–149.



Transmission and transfer 

Siegel, J. 1998. Dialectal differences and substrate reinforcement in Melanesian Pidgin. Journal of Sociolinguistics 2. 347–373. Siegel, J. 1999. Transfer constraints and substrate influence in Melanesian Pidgin. Journal of Pidgin and Creole Languages 14. 1–44. Siegel, J. 2000. Substrate influence in Hawai‘i Creole English. Language in Society 29. 197–236. Siegel, J. 2003. Substrate influence in creoles and the role of transfer in second language acquisition. Studies in Second Language Acquisition 25 (2). 185–209. Siegel, J. 2004a. Morphological simplicity in pidgins and creoles. Journal of Pidgin and Creole Languages 19 (1). 139–162. Siegel, J. 2004b. Morphological elaboration. Journal of Pidgin and Creole Languages 19 (2). 333–362. Siegel, J. 2005. Creolization outside creolistics. Journal of Pidgin and Creole Languages 20 (1). 141–166. Siegel, J. 2006. Links between SLA and creole studies: Past and present. In Lefebvre, C., L. White & C. Jourdan (eds). L2 Acquisition and Creole Genesis: Dialogues. Amsterdam/ Philadelphia: John Benjamins. 15–49. Siegel, J. 2007. Recent evidence against the Language Bioprogram Hypothesis: The pivotal case of Hawai‘i Creole. Studies in Language 31 (1). 51–88. Siegel, J., B. Sandeman & C. Corne. 2000. Predicting substrate influence: Tense-modality-aspect marking in Tayo. In Siegel, J. (ed.). Processes of Language Contact: Studies from Australia and the South Pacific. Montreal: Fides. 75–97. Smith, N. 1999. Pernambuco to Surinam 1654–1665? The Jewish slave controversy. In Huber, M. & M. Parkvall (eds). Spreading the Word: The Issue of Diffusion among Atlantic Creoles. London: University of Westminster Press. 251–298. Thomason, S.G. 2001. Language Contact: An Introduction. Washington, CD: Georgetown University Press. Thomason, S.G. 2002. Column: Creoles and genetic relationship. Journal of Pidgin and Creole Languages 17 (1). 101–109. Thomason, S.G. & T. Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Trouillot, M.-R. 1998. Culture on the edges: Creolization in the plantation context. Plantation Society in the Americas 5 (1). 8–28. Valdman, A. 2005. Review article: La Créolisation: Théorie, Applications, Implications by Robert Chaudenson. Studies in Second Language Acquisition 27 (3). 441–463. Véronique, D. 1994. Naturalistic adult acquisition of French as L2 and French-based creole genesis compared: Insights into creolization and language change. In Adone, D. & I. Plag (eds). Creolization and Language Change. Tübingen: Max Niemeyer. 117–137. Wekker, H. 1996. Creolization and the acquisition of English as a second language. In Wekker, H. (ed.). Creole Languages and Language Acquisition. Berlin: Mouton de Gruyter. 139– 149. Winford, D. 2003. An Introduction to Contact Linguistics. Malden, MA/ Oxford: Blackwell. Zobl, H. 1982. A direction for contrastive analysis: The comparative study of developmental sequences. TESOL Quarterly 16. 169–183.

The sociolinguistic history of the Peranakans What it tells us about ‘creolization’* Umberto Ansaldo*, Lisa Lim* & Salikoko S. Mufwene** *University of Amsterdam **University of Chicago

1

Creoles and the notion of ‘creolization’

There is considerable variation in the usage of the term creole (créole, criolou, or criollo in other languages) in reference to people. While it was generally used during the colonial period for locally-born populations of non-indigenous stock, it now serves to designate (partial) descendants of Africans in Mauritius, descendants of Europeans (the bekés) in Martinique (as opposed to the chabins ‘high yellow Mulattoes’), and Mestizo populations in parts of Latin America. On the other hand, it designates descendants of Africans or of the French from before the Louisiana Purchase (1803) in Louisiana, depending on the geographical location. For the purposes of this study, we stick to the less restrictive and race-neutral sense of locally-born person of (partly) non-indigenous descent adopted by Mufwene (1997), which remains relevant to the Founder Principle (Mufwene 1996, 2001), and how it applies to language evolution in a colony. Focusing on the development of plantation settlement colonies from the homestead to the plantation phases (Chaudenson 1992, 2001), we want to highlight similarities between, on the one hand, the significant influence exerted by Creole populations on language evolution in the American Caribbean region and the Indian Ocean and, on the other, the role of the Peranakans (explained below) in the indigenization of English in Southeast Asia. The Peranakans as (descendants of the) Chinese born locally during the colonial period, typically out of unions of Chinese men and Malayspeaking women, have been compared to Creoles in being (partly) of non-indigenous stock (e.g. Shellabear 1878/1913). They are also alike in being natives of the relevant territories, interestingly in a way consistent with Hall’s (1966) characterization of ‘cre* We wish to thank Bao Zhiming, James Collins, Anne Pakir and Wang Gungwu for brief discussions on the history of the Peranakans, as well as Anthea Fraser Gupta and three anonymous referees contacted by the Journal of Pidgin and Creole Languages for their feedback on an earlier version of this paper. The order in which the authors are listed is simply alphabetical and does not reflect the proportion of work contributed by each.

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

olization’ as ‘indigenization’, a process which, according to traditional views, not only transformed a pidgin into a creole but also made it a language indigenous to where this transformation putatively took place. As new ‘breeds’, both Creoles and the Peranakans are indigenous to the colonies that produced them, representing social categories unknown in the territories where their ancestors originated. Indeed as Mufwene (1997) inferred from Valkhoff (1966), the term creole applied to people, animals, and plants (before it applied to language varieties) seemed to have been associated with colonial peculiarities that were primarily unattested in the metropole. Caribbean creole vernaculars (henceforth creoles, without capitalization) have become our heuristic prototypes about the development of creoles in general (Muysken & Smith 1995; Mufwene 2000, 2001), despite the contribution that studies of Indian Ocean creoles (notably Baker & Corne 1982, 1986; Chaudenson 1992, 2001; Baker 1993, 1996; and Corne 1999) especially the work of Bickerton (e.g. 1981, 1984, 1999) have made to the subject matter. Neither the extension of such research to Pacific pidgins nor the controversial application of the terms creole and pidgin to contact-based varieties involving no European languages has made a difference. Researchers on these other language varieties have typically sought to prove that they have evolved in more or less the same manner as the Caribbean prototypes. Like Bickerton, they took no interest in going beyond the contact of the laborers and their employees and checking the extent to which the contact ecologies paralleled, or were different from, those of the Caribbean plantation colonies. A few studies such as Bailey and Maroldt (1977), Schlieben-Lange (1977), Chaudenson, Mougeon and Beniak (1993), Calvet and Chaudenson (1998) and Mufwene (2001, 2005, in press) have compared the development of creoles to show that the evolutionary trajectories are comparable, although the contact settings are quite different. The impact of this is that an increasing number of creolists, including Mufwene (2000, 2001, 2005), Ansaldo and Matthews (2001), Chaudenson (2003), DeGraff (2003, 2005), Aboh and Ansaldo (this volume), and Ansaldo (to appear) agree that, from a language evolution point of view, the development of creoles, pidgins, and other mixed language varieties differs from that of other language varieties considered unmixed, not by the restructuring processes involved but rather by the internal and external ecologies of the changes. In the present state of the art, with the terms creole and pidgin extended to varieties based on no European language (such as Sango, Lingala, and Bazaar Malay), two related questions still arise among many others: 1. Should the emergence of such contact varieties be associated with extraordinary social conditions of dominated, oppressed, or under-privileged populations? 2. Based on (1), is the typically-invoked extraordinary acquisitional scenario, based on notions of imperfect learning or break in the transmission of the lexifier (e.g. Thomason & Kaufman 1988; Thomason 2001), a necessary requirement to account for the formation of restructured vernaculars?



The sociolinguistic history of the Peranakans 

In this paper, we address these questions, taking into account the following facts: (i) the beginning of European colonization and subsequent economic domination of the world five centuries ago then foreshadowed what is identified today as economic globalization (Mufwene 2002, 2005); and (ii) communication by sea and rivers then became important routes by which European languages spread around the world (Chaudenson 2002). We argue, contra the dominant trend in the literature, (a) that slavery, oppression, or economic domination are not necessary conditions for the evolution of the hybrid or mixed language varieties that have concerned creolists and students of indigenized varieties; (b) that extraordinary acquisition is not always the dominant factor in the divergence process that has made the new varieties different from those that their ‘creators’ targeted; it is in fact not a necessary condition. To provide a broader context in which to address the above questions and observations, we focus in this chapter on Southeast Asia, in particular, on the contact settings of Malacca and Singapore and their impact on language evolution in the region. We highlight the role of the Peranakans first in the development of Baba Malay, and, subsequently, in the spread and indigenization of English in the region. We believe that the facts we bring to bear on language evolution here will help us better understand the ecological factors that generally bear on the evolution process, even about the development of creoles and pidgins. As noted above (see also section 2), the Peranakans, also known as Babas or Straits(-born) Chinese, are the Malay-speaking descendants of South Chinese immi1. While this position may not be overtly stated, the literature is replete with references to slavery and contract labor, making more allusion to the status disparity between native speakers of the ‘lexifier’ and the non-European laborers than to the actual patterns of social interaction (Mufwene 2005). References to segregation and to the break in the transmission of the lexifier that putatively ensued from this have focused so much on status differences that they have ignored the presence of Creoles of non-European descent who spoke the ‘lexifier’ natively. We return to this factor below. 2. It is useful to clarify at the outset that these terms are not entirely synonymous and have been used variably by different authors. Baba was originally used for male Peranakans, in opposition to the Nyonya, the females. While all people of Chinese ancestry born in the old Straits Settlements (namely Malacca, Penang and Singapore) are ‘Straits-born’, a distinction is made between “a first-generation Straits-born Chinese who lived and acted much like his immigrant Chinese parents” and “a third, fourth or fifth generation Baba whose customs, language and behaviour were Straits Chinese” (Png 1969: 98). For instance, only the latter group has spoken Malay as their only vernacular, while the former also spoke (some) Chinese (Png 1969: 99). The more inclusive usage serves a political ideology (Tan 1988a: 10; see also The Peranakan Association 2006). A different kind of usage restricts the term Baba to the locally-born descendants of a mixed Malay-Hokkien culture original to Malacca and later Singapore, applying Peranakan also to most individuals of Chinese origin living in the Malay/ Indonesian region. In the present chapter, we use the term restrictively but including both males and females. Our position is consistent with Song’s (1923: 69) distinction between the “Dutch Peranakans” (i.e. those of Indonesia) and the “Straits-born Chinese”, who would be associated with English.

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

grants (typically Hokkien or Teochew) and local Malay-speaking women. The Chinese had been an important maritime and trade power in Southeast Asia from the 12th to the 16th centuries (Pan 2000), and Chinese male immigrants continued to populate the important trade colonies of Penang and Malacca in Malaysia, as well as Singapore through to the 19th and 20th centuries, intermarrying with the local women. Out of these unions, populations of mixed Chinese and Malay/ Indonesian descent, comparable to Mulatto Creoles in the Caribbean and West Africa, emerged. The word Peranakan (pronounced [pranakán]), derived from the Malay root anak ‘child’, means a locally-born person. In this respect, the term is similar to earlier uses of the term Creole in Portuguese and Spanish colonies in reference to locally-born non-indigenous populations of (mixed) European and/ or African descent (Chaudenson 1992, 2001; Mufwene 1997, 2005; Berlin 1998, 2000). In fact, Skinner’s (1996) definition explicitly makes the parallel, where Peranakan means “creolized (Chinese) with Southeast Asian influences, particularly in regard to their language variety, Baba Malay”. We acknowledge that such populations of mixed ethnic descent are not unique to settlement colonies. There are undoubtedly others associated with trade colonies and the development of, for instance, ‘Guinea Coast Creole English’ (Hancock 1980) and ‘Indo-Portuguese’ (Clements 1991, 1996). We focus on the Peranakans because we are more familiar with their history and because the population evolved under ecological conditions independent of European colonization. It had emerged before the Europeans colonized the region, and the latter would in fact use them as auxiliaries to their enterprise. Their indigenous vernacular, Baba Malay (BM), has been treated as a pidgin or creole in the literature (see e.g. the classifications in Holm 1989; Smith 1995), because it arose in settings of multiple language contact leading to language shift. Regardless of whether or not BM must be identified as a creole/ pidgin, the emergence of both the language variety and its speakers is associated with combined trade and settlement colonization and offers useful comparative information in relation to ‘classic creoles’ of especially the American-Caribbean region and the Indian Ocean. We focus especially on the nature and ethnographic status of the target language, on the socioeconomic status of the appropriators of the language, and on patterns of human interactions.

2 The Peranakan population and the genesis of Baba Malay Although the emergence of the Peranakan community and culture is a consequence of migrations, these did not involve enslavement or oppression. Despite the fact that the Peranakans were minorities in Malaya, the old designation of the polities in which Malay has functioned as a vernacular or lingua franca, they were neither oppressed nor underprivileged. On the contrary, unlike Creoles around the Atlantic (Berlin 1998) and the Indian Ocean, their socioeconomic status would rise to the point where they became important powerbrokers during the European colonization of the Straits of Malacca in the 19th and 20th centuries. Because their language has often been classi-



The sociolinguistic history of the Peranakans 

fied as a creole, the comparison of the contact history that produced these new population, language, and culture with that of the creole people and vernaculars should be informative.

2.1

The non-traumatic birth of the Peranakans

As noted in section 1, the Chinese started trading in Southeast Asia as early as the 12th century, during the Southern Song dynasty. Their economic presence in the region would continue, with a few interruptions, until 1717, when an imperial decree by the Manchus banned all trade in the South China Sea. One of the most prominent figures in this Chinese maritime trade history was Zheng He, who commanded an extraordinary fleet and led expeditions to Southeast and South Asia over eight times in the early 15th century and apparently established small communities along the route (Pan 2000). It is indeed believed that the earliest Chinese stable trade colonies date back to this period, though it should also be noted that both He and his officers, as well as their translators, report finding Chinese settlers in Malacca during their expeditions (Tan 1988a: 28–29). A longer-lasting Chinese maritime empire would develop in the 16th and 17th centuries, led particularly by the Fujian-based traders Zheng Zhilong and Koxinga (Pan 2000: 49). Fujian traders, typically speakers of Min varieties (Norman 1988), had also been among the earliest to actually settle down in Southeast Asian ports and establish local communities. Led by the Portuguese, the Europeans started trade colonies in the region then identified as the ‘East Indies’ in the 16th century. (The region extended either from India or from Malaysia to New Guinea.) They found several Chinese-based communities already established, especially in Malacca, where they encountered the demographically most important Peranakan colonies. Within the Chinese diaspora, the Hokkiens were the dominant group. Quite typical of the time, the Chinese traders consisted of men. Women stayed behind to take care of their households and to observe filial piety and ancestor worship (Lim 1967: 63–65). Also the Manchu regime, which made emigration illegal throughout the greater part of the Qing dynasty, discouraged female emigration (Tan 1988a: 34). By the 19th century, an important proportion of Chinese women had been brought over to the Straits settlements for prostitution (Tan 1988a: 35). Nonetheless, Chinese men still greatly outnumbered Chinese women in Malaya, by 4 to 1 in 1881, an imbalance which led to frequent inter-ethnic marriages involving local women. While some marriages involved Muslim Malay women, only a subset of these are reported to have entailed conversion of the males to Islam. In general, even such conversion was nominal, as the Chinese generally continued observance of Chinese folk religion and brought up their children as Chinese (Skinner 1960: 96–97; Tan 1988a: 38f.). The maintenance of Chinese traditions was also facilitated by the fact that some of the Chinese settlers married non-Muslim women, often slaves of Balinese, Bugis or Javanese origin; such marriages required no conversion (Tan 1988a: 40–41; Pan 2000).

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

The male offspring would typically travel back to China to be educated there, while the women stayed in Malacca as shopkeepers and auxiliaries to their husbands’ or parents’ trade (cf. Song 1923 [1967]). In these emerging mixed communities, marriages of the Nyonyas with the locals were discouraged (Lim 1917: 876), and eventually between the Babas and the local women too. This endogamous practice became established especially after the Peranakans had reached a critical mass and the sexes had equalized. Marriages would occur predominantly between Peranakans or between them and new immigrants from China (Lim 1917: 867). In such a heterogeneous cultural environment, a new hybrid culture emerged with a unique set of cultural patterns. The first description of such a community seems to be by Crawfurd (1856: 96), who wrote that “the settlers, whenever it is in their power, form connections with the native women of the country; and hence has arisen a mixed race, numerous in the older settlements, known to the Malays under name of Paranakan China [i.e. cina ‘Chinese’]”. According to Tan (1988a: 42f.), the early intermarriages between Chinese and local women were an important factor in the acculturation of the Chinese and the subsequent emergence of the Peranakan community and culture. The Peranakans of Malacca have since been one of the most vivid examples of cultural ‘creolization’ in Southeast Asia: in all aspects of tradition, be they artistic, literary, religious or linguistic, the Peranakans developed unique traits that set them apart from other Chinese, the more indigenous local populations, and other ethnically mixed groups (Tan 1988b; Rudolph 1998). Non-linguistic examples include a mixed cuisine largely influenced by Malay traditions and the wearing of Malay/ Indonesian sarong and kebaya (instead of the Chinese dress) by the Nyonyas, which contrast with the retention of Chinese rituals, such as traditional wedding customs involving imperial era wedding costumes (Tan 1988b: 299). According to some observers, the Peranakans had “lost touch with China in every respect, except that they continued to uphold Chinese customs, and to practice, in variously modified forms, the social and religious practices of the forefathers” (Lim 1917). Nothing in the above suggests that the migrations of the Chinese men to the Straits Settlements were coercive; neither were the traders forced to remain in the region. While the early ones may have returned regularly to China to ship more goods, and the sons of the wealthier ones (a minority) to receive education (Lim 1917: 876), they always returned to the Straits. Furthermore, the Peranakans considered Malacca and Singapore their home. This is in contrast with the majority of the non-Peranakan Chinese in 19th-century Singapore who continued harboring plans to return to China: statistics from 1881 up to the 1960s show not only a continuous stream of Chinese immigration but also return emigration (Kwok 2000: 200). Until the 1950s, only the Straits Chinese could be considered ‘permanent’, ‘native’, or indigenized (Song 1923 3. This is supported by a study of emigrant communities in South China, according to which there were in the emigrant villages and especially in the schools supported by the emigrants in Southeast Asia “many boys of partly Malay blood” (Chen 1939: 143).



The sociolinguistic history of the Peranakans 

[1967]). And they were certainly not an enslaved nor oppressed population, as shown in section 2.3.

2.2

Multilingualism and the nature of transmission

In addition to all the aspects of culture that evolved within the emergent Peranakan community, a new language variety known as ‘Baba Malay’ (BM) developed as the mother tongue of the community. It is distinct from other Malay varieties due partly to the large amount of Chinese influence on it, particularly from Hokkien (see below), Teochew, and Cantonese. While most of the early Chinese settlers must have spoken Hokkien among themselves, there were many other southern Chinese among them, such as the Teochew and the Hakka (Ansaldo & Matthews 1999). This made allowance for the Chinese influence on BM not to be exclusively Hokkien. Marriages with women who did not speak Malay natively (e.g. the Balinese, the Bugis, or the Javanese) also favored other, non-Chinese ‘substrate’ influence. BM thus evolved in contact settings that were societally multilingual and involved competition of features not only between the substrate languages and the target language but also among the substrate languages themselves. Collectively these factors constituted a situation reminiscent of that in which ‘classic creoles’ emerged around the Atlantic and in the Indian Ocean. On the other hand, the situation is also reminiscent of that of Hawai‘i, where the contract laborers maintained their traditional ethnic identities (Mufwene 2004, 2005), in that the Malays, Hokkiens, and the other Indonesian and Sinitic populations generally maintained their individualities, unlike the slaves in the Atlantic and Indian Ocean colonies. An important difference is that the target language was indigenous in the 4. Note that BM developed where the new ethnic community not only reached enough critical mass to isolate itself as distinct but also had a certain amount of stability and permanence. This is why such developments have not been observed in many other Chinese trade colonies of today’s Malaysia and Indonesia. 5. Ansaldo & Matthews (1999) note several grammatical similarities between Hokkien and BM which can be attributed to Hokkien substrate influence or the congruence of the latter and nonstandard Malay varieties. Lexical contributions from Hokkien and other Chinese varieties, noted since Shellabear (1878), were unavoidable, as the Peranakans incorporated (albeit with adaptations) many traditional Chinese practices in their emergent culture. James Collins (p.c. 17 November 2001) points out that most of this vocabulary is attested in cultural domains that distinguish the Peranakans from the Malays, such as kinship and religion. (See also Pakir 1986; Tan 1988a.) 6. We use the term substrate rather reluctantly, only in keeping with the tradition in creolistics to refer to influence from languages other than the target. The languages spoken by the Chinese traders and other groups did not precede Malay in the contact setting, which is endogenous to the region. Moreover, as we argue in this paper, the Chinese traders were not socioeconomically inferior to the Malay-speakers, a situation that is different from what led Hall (1950) to misuse this term, borrowed from Romanistics, to creolistics (Goodman 1993).

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

Straits settlement colonies. As explained in section 2.1, it would take up to the 19th century before the Peranakans shifted to BM as their only vernacular. It is to this latter period that the following observation by Song (1923 [1967]: 3) applies: “In Malacca, however, where the Chinese had formed a continuous colony for about six centuries, the women-folk had entirely dropped the use of Chinese, while the Malacca-born Chinese males only acquired the Hokkien dialect colloquially for the purpose of trade”. This particular evolutionary difference from Hawai‘i can be attributed to the longer time during which the colony had evolved, the communicative dynamics of the interethnic marriages (especially during the founder period), and the fact that the Peranakans as a group have considered as their home the Straits of Malacca rather than China.

2.3

The Peranakans as privileged British subjects

Mining of gold and tin, the large-scale commercial agriculture business (in gambier, pepper, tapioca, and especially rubber), import-export business, and other economic enterprises had been drawing Chinese to Malacca, and gave rise to an influential class of Chinese capitalists, chiefly the Peranakans (Tan 1988a: 48). By the time of the European exploitation colonization of the region in the 19th century (i.e., subsequent to the trade colonization that preceded it), most Babas of Malacca had accumulated a lot of wealth and become prestigious subgroups in the region, distinguishing themselves from later Chinese immigrants and forming separate communities of their own. In Malacca, the well-off Baba were able to take over the houses of the great Dutch merchants in Heeren Street which then became “the fashionable and aristocratic resort of the Chinese” (Braddell 1853: 74). In Penang, it was also noted that the Chinese “who have long been settled in the place, and who have wedded native wives, dwell in large and elegant houses environed with fruit and flower-gardens” (Thomson 1875: 13). In Singapore, it is widely agreed (see e.g. Kwok 2000: 202) that the Babas were a class apart from the other ethnic groups. Although small in number (‘Malacca’ men comprised only 2.5% of the Chinese population in 1848, growing to just 9.5% in 1881), their social and economic influence was disproportionately strong in comparison, and they formed an important sector of the local elite (Kwok 2000: 202–204). They were seen to be the best educated, the wealthiest, and the most intelligent section of the Chi7. This difference should probably be conceived of in degrees of language loss or maintenance and not in dichotomous terms. While ethnic distinctions among Asians have survived in Hawai‘i, they are no longer conspicuous among descendants of contract laborers of European descent, perhaps with the exception of the ‘Portuguese’. On the other hand, there have been mixed marriages among people of the Asian stock too and offspring from these marriages consider Hawai‘i their home and speak only or primarily Creole (identified as ‘Pidgin’ by the locals) and English. 8. This is a situation reminiscent of Louisiana, where, as noted in section 1, descendants of Africans and of the French of the pre-Louisiana Purchase period identify themselves as Creoles, distinguishing their cultural heritage of the French colonial period from the descendants of Africans and the French who arrived later.



The sociolinguistic history of the Peranakans 

nese community (Nathan 1922: 77), as well as “more enlightened, and better merchants” (Earl 1837: 363). Their predominance in the commercial sectors, their knowledge of the local ways, and their closer contact with British administrators and merchants due to their command of the English language afforded them a significant role as intermediaries between Europeans, locals, and Asian newcomers (Tan 1988a; Kwok 2000). Indeed, since the early 19th century English became an increasingly important language in Southeast Asia, and the Peranakans were among the privileged few who acquired and benefited from it (Pan 2000: 172). They held a high regard for English-medium education and sent their children to English-medium schools. The establishment of four early educational institutions – the Anglo-Chinese College of Malacca (1818), Penang Free School (1816), Singapore Institution (1823; later, the Raffles Institution, 1868), and Malacca Free School (1826, Malacca High School since 1878) – was especially important to the development of the community (Tan 1988a: 52). By the mid19th century their ability to converse in this colonial language had strengthened their prominent socioeconomic position within other local communities, to the point where they were in fact sometimes referred to as “King’s Chinese” (Tan 1988a: 53). The growing importance of English among them as a lingua franca, as a vernacular, and as an identity marker can be clearly seen in the debates that arose during the mid-20th century. Though there is no doubt that most Peranakans identified themselves with BM (Tan 1988b), around this time more and more Babas were shifting to English as the only vernacular. While BM had competed fairly well with English as a lingua franca until the early 20th century, by the end of the 1960s English had almost completely prevailed de facto as the means of interethnic communication par excellence (Rudolph 1998: 335). The separate identity of the Peranakans was made distinct from the continuously increasing population of China-oriented immigrants not only by their Baba culture but also by their local (Malayan) orientation and their pro-British sentiments (Tan 1988a: 54f.). Identifying politically with the British (Kwok 2000: 205), they formed the Straits Chinese British Association (SCBA) in August 1900, with English as its exclusive language and with an admitted aim to promote trade with, and foster loyalty to, the British Empire (Song 1923 [1967]: 319). The Malacca branch of SCBA was formed in October of the same year (1900), and the Penang branch was founded later in 1920. The Association is still extremely active to date. We observe this inclination in the following passage from J.D. Vaughan (1879): One may see in Malacca Babas who can claim no connection with China for centuries, clad in long jackets, loose drawers, and black skull caps, the very counterparts of Chinese to be seen any day at Amoy, Chusan or under the walls of Nankin. Strange to say that although the Babas adhere so loyally to the customs of their progenitors they despise the real Chinaman and are exclusive fellows indeed; nothing they rejoice in more than being British subjects. The writer has seen Babas on being asked if they were Chinaman bristle up and say in an offended tone “I am not a Chinaman, I am a British subject”, an Orang putih literally, a white man; this term is invariably applied to an Englishman. They have social clubs of

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

their own to which they will admit no native of China. At these clubs they play billiards, bowls, and other European games, and drink brandy and soda ad libitum; yet they adhere strictly to the Chinese costume – the queues, thick soled shoes, mandarin dresses, and conical hats on state occasions, and the manners and customs of those people who otherwise they have no sympathies with. The Duke of Edinburgh when in Singapore in 1869, visited a Chinese Club and bowled with the Babas and expressed himself highly pleased with their pluck. The true Chinaman ridicules the idea of exercise in any shape.

The following quote referring to the period ca. 1850 clearly illustrates the position of the Babas in the community (Song 1923 [1967]: 69): Unlike the Dutch ‘peranakans’ [presumably Indonesian-Chinese] the Straits-born Chinese [i.e. the Babas] have during the last twenty-five years been trained to realize the relationship in which they, as a community, stand to the British Throne and Empire. The proofs of loyalty and patriotism and the service in numerous forms to the British Empire given by that community during the Great War should justify the British Imperial Government in putting an end to diplomatic uncertainty and claiming the right to protect, by the issue of unqualified British Passports, every Chinese born in the Colony, because he is a natural-born British subject.

It is clear from the preceding account that the Peranakans were neither an economically dominated nor an underprivileged group. In fact, they were quite the opposite, being prominent in Malacca and Singapore societies economically and socially. This suggests that the language restructuring and speciation processes identified in the literature as ‘creolization’ need not be associated with dramatic social conditions or emergency psycholinguistic dynamics. The explanation for the process must lie in the regular mechanisms of language appropriation and shift under ecology-specific conditions that allowed significant structural influence (often by congruence) from languages previously spoken by the appropriators onto the target language (Mufwene 2005: 54–58).

2.4

Baba Malay features

Thus far we have shown how the sociohistorical context of the emergence of the Peranakan population is quite distinct from that normally associated with the emergence of Creole people, creole languages, and creole cultures. We have questioned the ecological conditions typically associated with the development of these ‘mixed’ phenomena that have concerned especially creolists and students of indigenized varieties. In this section we turn to the varieties in the linguistic repertoire of the Peranakans, focusing on BM, to show how it can be seen as a hybrid or mixed variety that owes many of its structural 9. This is also reminiscent of the Mulattoes in South Carolina who, up to the time of the Abolition of slavery, formed their own caste, on the model of white planters, and afforded their own slaves (Joyner 1991).



The sociolinguistic history of the Peranakans 

features to the specific contact situation that produced it (Ansaldo & Matthews 1999). This is necessary because, at first sight, one may be tempted to compare the formation of BM with what is claimed to typify ‘admixture’ or ‘intertwining’ (e.g. Thomason 1997). There are several reasons why this parallel is not taken up in this chapter: i. It is unclear whether a class of ‘mixed languages’, different in genesis and in type from other (contact) languages, does indeed exist (Mufwene 2001, 2005, in press). In an evolutionary scenario of language change, the same mechanisms of formation are in place in all contact-induced change (e.g. Croft 2000; Aboh & Ansaldo this volume; Ansaldo to appear). ii. Pure bilingualism is hard to prove in most contact environments (e.g. Bakker & Mous 1994; Matras & Bakker 2003); and the case of the Peranakans surely shows a high degree of societal and individual multilingualism. What this suggests regarding the notion of transmission is that, in most contact settings, features transfer from the input varieties to the new grammar in stages of interlanguage, i.e. through various adaptations and not directly from the source to the product (see Siegel 1997). iii. BM cannot be captured as an even distribution of lexical versus grammatical features of two different (groups of) languages, viz., the lexifier and the substrate; nor can its grammar solely be accounted for in terms of either Hokkien or Malay features. The lexicon is mixed, and a combination of typological congruence and innovations can be found, similar to the type of restructuring we find in many creole [and non-creole] languages (Ansaldo & Matthews 1999). We must underscore the fact that Malacca had been a major trading center for East, South and Southeast Asia even before the Peranakans emerged as a distinct group. Trade Malay or ‘Pidgin Malay Derived’ varieties (PMD, Adelaar & Prentice 1996) flourished in the area. In referring to ‘Malay’ in what follows, we assume that PMD features would have been part of the repertoire of Chinese traders – even if they (the Chinese) had immediate exposure to colloquial Malay spoken by the local women they married – and local Malays alike. Though a colloquial form of Malay appears to have been the main lexifier of BM, influence of Sinitic origin, in particular Hokkien, can also be found in core areas. As already shown in Pakir (1986) and Tan (1988a), kin terms as well as words pertaining to salient cultural areas, such as clothing, cuisine, rituals, and religious and other abstract Chinese concepts, show a predominantly Hokkien origin.10 Moreover, an examination of the recently published BM dictionary (Gwee 2006) confirms the extent of Hokkien contributions to be very significant. Table 1 lists a few examples of the vast number of kin terms in BM (Tan 1988a: 210f.; Pakir 1986: 106–107) which demonstrate the pattern found throughout the system: the kin terms are of Hokkien origin; the corresponding Malay forms for the terms are clearly different. 10. See Pakir (1986, chapter IV).

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

Table 1.  Kinship terms in BM kin term

BM

Hokkien

Malay

elder brother/ similar-aged or older male elder sister/ similar-aged or older female mother-in-law maternal grandmother paternal grandfather father’s older brother father’s younger brother mother’s brother

hiã

toa hiã

tači

toa či

abang ‘brother’/ pacik ‘uncle’ kak ‘sister’/ macik ‘aunt’

nio gua ma/ mama lai kong m-peh ng-ček ng-ku

niu ‘lady’ gua ma lai kong a-peh a-ček a-ku

nenek nenek bapa saudara bapa saudara bapa saudara

The BM pronominal system involves a 1st and 2nd singular pronoun of Hokkien origin, with a 2nd and 3rd person plural forms in which the Hokkien lang ‘people’ appears as the Malay calque orang. The first person plural is of Malay origin (Lim 1981; Pakir 1986; Ansaldo & Matthews 1999), though it has lost the inclusive/ exclusive distinction used in both Hokkien and Malay, suggesting that the restructuring process is more complex, sometimes producing simplification where none would be expected. This pronominal system is typical of many PMD varieties of Southeast Asia pre-dating European colonial rule. Table 2.  Pronominal systems of BM, Hokkien and Malay Pronoun

BM

Hokkien

Malay

I you (sing.)

gua lu

gua lu

he/ she/ it we

dia kita

you (plu.)

lu-orang

i gun-lang (incl.)/ lan-lang (excl.) lin-lang

they

dia-orang

in-lang

saya/ aku kamu/ awak/ engkau dia kita (incl.)/ kami (excl.) kamu/ awak/ engkau mereka

Sinitic influence is attested at the level of syntax as well, as illustrated by the following examples.11 11. Ansaldo & Matthews (1999) warn about the danger of assuming exclusive Hokkien influence, as Hokkien shares features with other Chinese varieties that were spoken in the Straits, as noted above.



The sociolinguistic history of the Peranakans 

Possessive constructions with the Malay morpheme punya (or mia) follow a Sinitic pattern (Ansaldo & Matthews 1999), e.g.:

(1) BM

dia punya bilik 3P POSS room ‘His/ her room’



(2) Malay

bilik dia/ bilik nya room 3P ‘His/ her room’



(3) Hokkien

i e pang-keng 3P POSS room ‘His/ her room’

The demonstrative precedes the Noun rather than follow it, showing Sinitic, not Malay word-order e.g.:

(4) BM

itu buk that book



(5) Malay

buku itu book that



(6) Hokkien

cît kiêng this building

Passive constructions marked by Malay kena ‘get’ (Pakir 1986) and kasi ‘give’, as in example (7), superficially resemble Malay (8); in fact they reveal Sinitic syntax, as is evident from the presence of the dummy agent orang in preverbal position (Ansaldo & Matthews 1999) in the typical Hokkien structure [SUBJ ho lang V] seen in (9), where ho lang (‘give people’) has been reanalyzed as a passive marker:

(7) BM

Nkoh gua kena orang tipu brother 1SG PASS people cheat ‘My brother was cheated.’ (Ansaldo & Matthews 1999: 54)



(8) Malay

Orang yang tak biasa senang kena tipu di pasar people who not used easy PASS cheat at market ‘People who aren’t used to it easily get cheated in the market.’ (Mintz 1994: 317)



(9) Hokkien

Guâ ē ā-hiã hŏ-lāng phiăn 1SG POSS brother by-person cheat ‘My brother was cheated.’

We also find what is likewise typological congruence of Malay and Sinitic language varieties in, among others, existential constructions based on the verb ‘to have’:

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

(10) BM

Ada manyak temput lagi have much time more ‘There is plenty of time.’ (Ansaldo & Matthews 1999: 57)

(11) Malay

Ada koran di meja have newspaper on table ‘There is a newspaper on the table.’

(12) Hokkien

Ĕ-mńg, ŭ lû-tiăm Amoy have hotels ‘There are hotels in Amoy.’

On this point, Tan (1988b) notes that the Babas, just like Chinese, make more frequent use of existential constructions than the Malay (Ansaldo & Matthews 1999: 57). More structural admixture can be found in the TMA system, summarized in Table 3 (for a full description see Ansaldo & Matthews 1999). The three pre-verbal markers suda, mo/ nanti and lagi, are clearly all derived from Malay and follow Malay word order, as in (13) and (14). However, as is evident from (15), the strategy is basically the same in Hokkien, although its marker of perfect follows the main verb. Table 3.  TMA markers in BM Function

Marker

Notes

(a) Perfective (b) Future (c) Progressive comparative

suda mo/ nanti lagi

– Immediate-intentional/ generic [lagi X lagi Y]

In (13) to (15) we see how the lexical item as well as the structure for the BM perfect is that derived from Malay. (13) BM

Gua suda makan I PERF eat ‘I have (already) eaten.’

(14) Malay

Saya sudah makan I PERF eat ‘I have (already) eaten.’

(15) Hokkien

Guà ciăq liâu-āu/ la I eat PERF ‘I have eaten.’

Where future is concerned, Ansaldo & Matthews (1999: 59) suggest that while nanti functions as a general future marker, mo (from Malay mau ‘want’) seems to be specifically indicating immediate/ intentional future, illustrated in (16) and (17).



The sociolinguistic history of the Peranakans 

(16) BM

Nanti satu haru li pun nanti charek dia later one day you also FUT seek her ‘You too will need her one day.’

(17) BM

Lu pun mo charek dia you also FUT seek her ‘You are also going to need her.’

Pakir (1986: 197–198) also notes that mo often yields a reading expressing immediate expectation and attributes this to the Hokkien aspectual marker teq. This is corroborated by Bodman (1987: 222), who observes that in Hokkien the aspectual marker běq indicates future (18), while the combination tèq běq usually signals immediacy (19): (18) Hokkien

î běq laí la 3P FUT come PRT ‘He is coming.’

(19) Hokkien

î tèq běq laí la 3P FUT come PRT ‘He is on his way now.’

In Malay we find two basic markers: mau, indicating intention and akan, a generic future marker, but no specialized marker of immediate expectation. It seems thus that BM has selected Malay lexical items with their basic functional properties and extended these with at least one Hokkien-derived semantic feature. The progressive comparative construction in BM, shown in (20) and (21), is particularly interesting as it is structurally a comparative construction with a progressive reading (Ansaldo & Matthews 1999). In Malay, lagi can mean ‘more’ or ‘again’ and is used in comparison of degree and time-phrases (Mintz 1994). In BM however, congruence results in a calque of the pan-Sinitic construction [lu Adj lu Adj] as shown in the Hokkien example in (20). Note that, in BM, unlike in Sinitic, we find instances of verbs (21) and nouns (22) modified by lú ‘more’: (20) Hokkien

Lú tua lú suì more old more pretty ‘(Getting) prettier the older she gets.’

(21) BM

Lagi tengok lagi manis more look more sweet ‘Sweeter with each look.’ (Ansaldo & Matthews 1999)

(22) BM

Lu lagi hari lagi tua 2P more day more old ‘You are getting older and older.’

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

2.5

Summary and reflections

Above, we have shown that in the sociohistorical setting in which the Peranakans have emerged as a group, a new language, BM, has also evolved, characterized by a mixed grammar to which both Malay and Sinitic language varieties appear to have contributed significantly in terms of lexical and grammatical features. BM evolved in a multilingual setting characterized crucially by the absence of the following traits traditionally associated with creole vernaculars: (i) traumatic geographical displacement of their creators; (ii) unequal social relations between speakers of the target language and its learners/ appropriators, with the latter being underprivileged; (iii) an evolution characterized by simplification; (iv) break in the transmission of the lexifier. From a structural perspective, BM seems to have selected and adapted its structures from sources typologically as diverse as those of classic creoles. On the other hand, this study shows that no particular extraordinary kind of contact need be invoked to account for emergence of BM. There was certainly no break in the transmission of the lexifier, as the creators of BM were always minorities surrounded by, and in many cases cohabiting with, speakers of colloquial Malay, despite the influence of Bazaar Malay they encountered during their actual trade contacts. The structures of BM can thus be accounted for by invoking processes of competition and selection of features triggered by a specific contact ecology at a particular sociohistorical juncture that allowed for a normal admixture of linguistic materials from different sources: this made the Peranakans different from the indigenous and other local populations. These observations may indeed prompt some readers to argue that BM falls in the category of mixed languages, if it is not a creole, and therefore its emergence calls for a special account in terms of externally-motivated change. As argued in Mufwene (2001, 2005, in press), all linguistic changes are externally-motivated, being triggered by changes in the direct or indirect ecology in which a language evolves, which affects the balance of power among the variants, largely by introducing some new variants into the feature pool or removing some. In addition, all modern languages are mixed in some way or another, depending on how the target being learned by their ‘creators’ was influenced by other languages it came in contact with and what specific ‘xenolectal’ features were incorporated in the emergent systems; mixed languages, are really quite normal (Muysken 2000). Heine & Kuteva (2005) illustrate this well in the case of Western European languages. In comparison with traditional ‘creole’ communities, one particularly interesting point here is the role played by the founder population in having Malay, the language of their socialization with the local population, prevail as the vernacular of the Peranakan community. The application of the Founder Principle (Mufwene 1996, 2001) is also evident from the particular features selected into BM, reflecting largely the kinds of settings in which the early Chinese traders learned Malay. These include the pronominal system typical of early contact-Malay varieties of the area, as well as in a number of lexical items. As noted above, the Hokkien-derived pronouns are a well-



The sociolinguistic history of the Peranakans 

known feature of contact-Malay varieties predating Western colonization, such as Bazaar Malay; these varieties would have been spoken in and around Malacca before the establishment of the Peranakan community and have left their traces in BM to this day. In a nutshell, the first Chinese traders were instrumental in the emergence of BM in their mixed households; the later immigrants and their children adopted it, rather than any of the surrounding native varieties, undoubtedly with minor modifications, as their communal vernacular. The social, cultural and economic distinctiveness of the emergent Peranakan community among other local groups seems to have contributed to the retention of the grammatical features that have fostered the divergence of BM from the more indigenous colloquial variety. Otherwise, one might have expected the ecology in which speakers of traditional colloquial Malay are the overwhelming majority to have fostered the convergence of BM with the mainstream variety, in a way similar to the alleged debasilectalization of Caribbean English creoles. Although it does not have exactly the same social significance, the identification of the Peranakans as a distinct social group may be compared to the racial segregation in New World and Indian Ocean plantation settlement colonies in the sense that the latter too fostered the basilectalization qua divergence of the Black Creoles’ vernaculars. A case in point is the role of race segregation in the divergence of African American vernacular English as a distinct ethnolect, in a larger population where African Americans are minorities. What matters in this particular case is not the institution of segregation per se but the fact that it produced a population structure in which African Americans would socialize more among themselves than with members of other groups. However, such social isolation does not entail break in the transmission of the lexifier. It simply means that the dynamics of social interaction in such ‘segregated’/ isolated communities tend to favor the selection of some features that are disfavored in the larger, mainstream community. Because of the role that the Peranakans have played in the local business world, BM has sometimes been confused with Bazaar Malay, as it functioned, at least up until the early 20th century, as the business Malay variety. Shellabear (1913: 51) notes that “Baba Malay is now, and is likely to be for an indefinite period, the business language of Singapore, Penang, and the Federated Malay states”. Nathan’s (1922: 77) comments in his 1921 census report, which reflects the official colonial view of the community, is also in the same spirit, suggesting that BM had gained social prestige in the Malay world: “Not only is the Baba Malay the language of the Straits-born Chinese, the best educated and wealthiest and most intelligent section of the Chinese community, but it is the language of market and shop and counting house throughout the Straits and Federated Malay States”. This observation is corroborated by Tan (1988a: 121–122), who comments that BM was the variety of commerce as well as the lingua franca in the Straits Settlements, and adds that it was in fact regarded as the language of the refined and the wealthy. Lim (1887: i) also reports that even the more urban Malays would use BM in their social interactions:

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

Those Malay who live in town and all those who constantly have intercourse with the public speak the Baba’s Malay when communicating with the people in town […] It should be understood that the Baba’s Malay is not as some think a ‘vulgar tongue’ […] the highest class of Malay when holding conversations with the Baba always speak as the latter do.

The evolution of BM is thus an interesting illustration of the long-standing assumption in linguistics that the prestige of a language variety does not stem from its structural features but rather from the socioeconomic status of its users. It also illustrates that creole-like, ‘mixed’ varieties need not be ranked low in a society. That creoles are ranked ethnographically low in especially the Caribbean, the Indian Ocean, and Hawai‘i has to do not with their origins in multilingual contacts nor with their being ‘mixed’ language varieties, but rather with the fact that the overwhelming majorities of their speakers have remained socioeconomically underprivileged to date. In this respect BM parts company with creole vernaculars, despite the emergent trend in several creole-speaking polities to promote them as national languages and as part of their indigenous heritage.

3 Final remarks There is more to be learned from this study, especially if we compare the Peranakans with some additional populations, as we do briefly in this part of the present section. It is not clear from the literature whether the term ‘Peranakan’ was selected by the relevant population itself or cast on them by outsiders. The Peranakans claim a separate ethnic identity from the more traditional Chinese groups such as Hokkien, Cantonese, and Hakka, and they are obviously distinct from the traditional Malay population. This identity is strong within the community and recognized by the society at large. Their situation finds parallels with at least three other peoples: (1) the Afrikaners in South Africa and (2) the White Creoles in Louisiana and the Békés of Martinique, both of whom distinguish themselves from the post-Emancipation French immigrants, although the Békés also have French citizenship. The term ‘Afrikaner’ is used to identify those descendants of Europeans, particularly of Dutch, German, or Huguenot descent (A dictionary of South African English on historical principles, 1996) who consider themselves indigenous to Africa, as opposed to those other European populations whose ancestors were not Boers and who presumably migrated to South Africa later than them, from the late 18th century onwards and were associated with the British rule. Likewise, Creoleness applied to Blacks, in contradistinction from the Bozal slaves or the 19th-century contract laborers, articulates a claim of indigenity that other peoples who are not Native Americans in the case of the New World do not have. It is a claim of being native to the land and being entitled to more rights than the others.



The sociolinguistic history of the Peranakans 

The mixed linguistic repertoire of the Peranakans (earlier in Chinese and Malay, and later in Malay and English), combined with their knowledge and understanding of local cultures, put them in an advantageous position as traders before the European colonization and as power brokers afterwards. In particular the Malacca-born developed very close ties with the Europeans and, according to Earl (1837), they learned English reasonably well. Many of them also accumulated considerable wealth. Such successful individuals often lent capital for the development of plantations and / or occupied important positions in the colonial administration and economic systems. Many of them worked for the Dutch and British East India Companies (Tan 1988a: 51f.). Eventually, the Peranakans would largely shift from Baba Malay to English as their (only) vernacular and then stood as models to other (Southeast) Asian populations, determining the path of the indigenization of this colonial language. All this happened in more or less the same way that Creole slaves stood as linguistic and cultural models to the Bozal slaves before the abolition of the slave trade and later to 19th-century contract laborers. During the British rule of South Africa, the Afrikaners too clung to the legitimacy of their own Afrikaans, next to English, as the language of ethnic identity. Like the Peranakans, they also transformed it into a powerful alternative business language, at least until recently. Although in this particular case there was no language shift on the part of the Dutch immigrants, this variety indigenous to South Africa reflects much of the contact ecology in which it evolved (den Besten 2003; Roberge 2003). What is significant about all three groups is the role of the Founder Principle in influencing the evolutionary trajectory of a modified form of a traditional language that would be appropriated as a vernacular or as a lingua franca by other populations, such as the Colored people in South Africa. Those who continue to define creole vernaculars as nativized pidgins should recall here that Hall (1966: xiii-xiv) also suggests a connection between nativization and indigenization, i.e., the emergence of a distinct variety that can be considered ‘indigenous’ or ‘native’ to the relevant European colony. Moreover, he explains that nativization need not apply to creoles only and can be extended to indigenized varieties such as Indian and Philippine Englishes which are used non-natively by most of their speakers but are ‘indigenous’ or ‘native’ to the polities where they are spoken. His position appears in fact to have been heeded by students of indigenized Englishes, who also identify them with the alternative term ‘nativized Englishes’. We believe Hall was correct in establishing these parallelisms which are more extensively explored in Mufwene (2001, 2005).  Hall (1966: xiii) also addresses the question of whether a creole need be associated with “conditions of slavery or economic oppression”. His answer is that “[i]t can develop whenever multilingual groups come together on their own free will, as has happened in a number of villages in present-day New Guinea”. The case of the Peranakans clearly shows that Hall is right in allowing for alternative scenarios for the evolution of new, hybrid varieties. He also establishes evolutionary similarities between, on the one hand, the development of creoles and, on the other, that of the Romance languages and the like. The comparison, taken up by Schlieben-Lange (1977), is discussed exten-

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene

sively in Mufwene (2005, in press). However, we should beware of the danger of associating the process traditionally identified as ‘creolization’ with subordinate populations under conditions of slavery or contract labor, an association that has been too often invoked in the literature on the development of creoles in the Caribbean and Indian Ocean. As noted above, and elsewhere in this volume, there are no restructuring processes that are peculiar to creoles nor used particularly by members of the socioeconomically subordinate social classes. We must now add that there are no particular kinds of population contact that are specific to creoles, except under the historical definition of the term (Mufwene 1997, 2003, 2005; DeGraff 2003; Ansaldo to appear). Our comparison of the Peranakan and Creole populations is not concerned with whether or not BM should be treated as a creole. Either way, it was certainly not developed by an oppressed or subordinate group in a precarious social environment. We can safely consider treating slavery in the development of Atlantic and Indian Ocean creoles as the factor – itself created by a special socioeconomic practice – that produced the contact conditions in which the new vernaculars developed (Mufwene 2005). We can thus conclude that the patterns of population growth (highlighted, for instance, by Baker 1993, similar to those identified in Chaudenson 1992, 2001, and Mufwene 1996, 2001) and population structure (including patterns of interethnic interaction, Mufwene 2005) at the critical phase of the colonies where the relevant language varieties developed are more relevant factors than slavery, oppression, and being underprivileged. The question of the proportion of those who developed the new variety relative to that of native or fluent speakers of the language they targeted cannot be overlooked here. It has been a central and recurring one in the literature about the development of creoles and pidgins, despite the fact that the demographics of communities have clearly shown that, if it were all a matter of numbers, some vernaculars treated as creoles or creoloids would not have diverged from the models spoken by colonial descendants of Europeans (Mufwene 2001). Réunionnais Creole would not have developed if one had to depend on the 20%:80% disproportion stipulated by Bickerton (1984) in favor of the slave population. African American Vernacular English would not have diverged from American White Southern English either, because (descendants of) Africans were and have generally remained in the minority, relative to the overall populations of the American former tobacco and plantation colonies (Rickford 1997; Mufwene 1999). The study of the development of BM among the Peranakans clearly suggests that population structure (including social isolation/ segregation) can be a more significant factor than population ratio. It seems that creoles diverged during the plantation phases of the histories of their communities from their kindred varieties spoken by European colonists not so much because of the demographic majority of their speakers but because of the segregation that the particular population structures of the plantations as socioeconomic institutions imposed on these speakers (Mufwene 2005). In sum, the present study of the Peranakans in Southeast Asia adds substance to the following principal observations:



The sociolinguistic history of the Peranakans 

1. New, hybrid cultures and vernaculars do indeed emerge in different settings of population contact, which produce novel ecologies for the appropriation of traditional languages targeted by new speakers. 2. The contact settings need not be considered as exceptional, although each one of them is unique in some ways, because contact has played some catalyst role in the speciation of almost any modern language today.

References A Dictionary of South African English on Historical Principles. 1996. Managing editor: Penny Silva. New York: Oxford University Press. Adelaar, K.A. & D.J. Prentice. 1996. Malay: Its history, role and spread. In Wurm, S.A., P. Mühlhäusler & D.T. Tryon (eds). Atlas of Languages of Intercultural Communication in the Pacific, Asia and the Americas. Berlin: Mouton de Gruyter. 673–693. Ansaldo, U. in press. Revisiting Sri Lanka Malay: Genesis and classification. In Dwyer, A., D. Harrison & D. Rood (eds). A World of Many Voices: Lessons from Documenting Endangered Languages. [Studies in Language.] Amsterdam/ Philadelphia: John Benjamins. Ansaldo, U. & S.J. Matthews. 1999. The Minnan substrate and creolization in Baba Malay. Journal of Chinese Linguistics 27 (1). 38–68. Ansaldo, U. & S.J. Matthews. 2001. Typical creoles and simple languages. Linguistic Typology 5 (2/3). 311–324. Bailey, C.-J.N. & Maroldt, K. 1977. The French lineage of English. In Meisel, J.M. (ed.). Pidgins – Creoles – Languages in Contact. Tübingen: Narr. 21–53. Baker, P. 1993. Assessing the African contribution to French-based creoles. In Mufwene, S.S. (ed.). Africanisms in Afro-American Language Varieties. Athens: University of Georgia Press. 123–155. Baker, P. 1996. Pidginization, creolization and français approximatif. [Review article of R. Chaudenson (1992), Des Îles, des Hommes, des Langues.] Journal of Pidgin and Creole Languages 11. 95–120. Baker, P. & C. Corne. 1982. Isle de France Creole: Affinities and Origins. Ann Arbor: Karoma. Baker, P. & C. Corne. 1986. Universals, substrata and the Indian Ocean creoles. In Muysken, P. & N. Smith (eds). Universals versus Substrata in Creole Genesis. Amsterdam/ Philadelphia: John Benjamins. 163–183. Bakker, P. & M. Mous (eds). 1994. Mixed Languages. 15 Case Studies in Language Intertwining. Amsterdam: IFOTT. Berlin, I. 1998. Many Thousands Gone: The First Two Centuries of Slavery in North America. Cambridge, MA: Harvard University Press. Berlin, I. 2000. Atlantic Creoles: The charter generations. In Appiah, K.A. & H.L. Gates, Jr. (eds). Microsoft Encarta Africana. CD-ROM. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Bickerton, D. 1984. The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7. 173–221.

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene Bickerton, D. 1999. How to acquire language without positive evidence: What acquisitionists can learn from Creoles. In DeGraff, M. (ed.). Language Creation and Language Change: Creolization, Diachrony and Development. Cambridge, MA: MIT Press. 49–74. Bodman, N.C. 1987. Spoken Amoy Hokkien. Vol 1. NY: Spoken Language Services Inc. Braddell, T. 1853. Notes of a trip to the interior from Malacca. Journal of the Indian Archipelago and Eastern Asia 7. 73–104. Calvet, L.-J. & R. Chaudenson. 1998. Saint-Barthélemy: Une Énigme Linguistique. Aix-en-Provence: CIRELFA – Agence de la Francophonie. Chaudenson, R. 1992. Des Îles, des Hommes, des Langues. Paris: L’Harmattan. Chaudenson, R. (revised in collaboration with S.S. Mufwene). 2001. Creolization of Language and Culture. London: Routledge. Chaudenson, R. 2002. L’eau et les langues. ms. Chaudenson, R. 2003. La Créolisation: Théorie, Applications, Implications. Paris: L’Harmattan. Chaudenson, R., R. Mougeon & E. Beniak. 1993. Vers une Approche Panlectale de la Variation du Français. Aix-en-Provence: Institut d’Etudes Créoles et Francophones. Chen, T. 1939. Emigrant Communities in South China: A Study of Overseas Migration and its Influence on Standards of Living and Social Change. Shanghai: Kelly & Walsh Ltd. Clements, C. 1991. The Indo-Portuguese creoles: Languages in transition. Hispania 74. 327–346. Clements, C. 1996. The Genesis of a Language: The Formation and Development of Korlai Portuguese. Amsterdam/ Philadelphia: John Benjamins. Corne, C. 1999. From French to Creole: The Development of New Vernaculars in the French Colonial World. London: University of Westminster Press. Crawfurd, J. 1856. A Descriptive Dictionary of the Indian Island and Adjacent Countries. London: Bradbury & Evans. Croft, W. 2000. Explaining Language Change. An Evolutionary Approach. Edinburgh: Longman. DeGraff, M. 2003. Against Creole exceptionalism. Discussion note. Language 79. 391–410. DeGraff, M. 2005. Linguists’ most dangerous myth: The fallacy of Creole exceptionalism. Language in Society 34. 533–591. den Besten, H. 2003. Khoekhoe syntax and implications for L2 acquisition of Dutch and Afrikaans. Journal of Germanic Linguistics 14 (1). 3–56. Earl, G. 1837. The Eastern Seas or Voyages and Adventures in the Indian Archipelago in 1832–33– 34. London: Allen & Co. Goodman, M. 1993. African substratum: Some cautionary words. In Mufwene, S.S. (ed.). Africanisms in Afro-American Language Varieties. Athens: University of Georgia Press. 64–73. Gwee, T.H.W. 2006. A Baba Malay Dictionary. Singapore: Periplus Publishing. Hall, R.A. 1950. The African substratum in Negro English. American Speech 25. 51–54. Hall, R.A. 1966. Pidgin and Creole Languages. Ithaca: Cornell University Press. Hancock, I.F. 1980. Gullah and Barbadian: Origins and relationships. American Speech 55. 17–35. Heine, B. & T. Kuteva. 2005. Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Holm, J. 1989. Pidgins and Creoles, Volume 2: Reference Survey. Cambridge: Cambridge University Press. Joyner, C. 1991. One people: Creating an integrated culture in a segregated society, 1526–1990. In Chesnut, D.R. & C.N. Wilson (eds). The Meaning of South Carolina History: Essays in Honor of George C. Rogers, Jr. Columbia: University of South Carolina Press. 214–244.



The sociolinguistic history of the Peranakans 

Kwok, K.W. 2000. Singapore. In Pan, L. (ed.). The Encyclopedia of the Chinese Overseas. Singapore: Chinese Heritage Centre; Cambridge, Mass.: Harvard University Press. 200–217. Lim, B.K. 1917. The Chinese in Malaya. In Feldwisk, W. (ed.). Present Days’ Impressions of the Far East and Prominent and Progressive Chinese at Home and Abroad: The History, People, Commerce, Industries and Resources of China, Hong Kong, Indo-China, Malaya and Netherlands India. London: Globe Encyclopedia Co. Lim, H.S. 1887. A Manual of the Malay Colloquial such as is Spoken by All Nationalities in the Colonies of the Straits Settlements, and Designed for Domestic and Business Purposes. Singapore: Koh Yeu Hean Press. Lim, J.H. 1967. Chinese female immigration into the Straits Settlements 1860–1901. Journal of the South Seas Society 22. 58–110. Lim, S. 1981. Baba Malay: The Language of the “Straits-born” Chinese. M.A. dissertation, Monash University. Matras, Y. & P. Bakker (eds). 2003. The Mixed Language Debate: Theoretical and Empirical Advances. Berlin: Mouton de Gruyter. Mintz, M. 1994. A Student’s Grammar of Malay and Indonesian. Singapore: SNP Publishing. Mufwene, S.S. 1996. The founder principle in creole genesis. Diachronica 13. 83–134. Mufwene, S.S. 1997. Jargons, pidgins, creoles, and koinés: What are they? In Spears, A. & D. Winford (eds). The Structure and Status of Pidgins and Creoles. Amsterdam/ Philadelphia: John Benjamins, 35–70. Mufwene, S.S. 1999. Some sociohistorical inferences about the development of African-American English. In Poplack, S. (ed.). The English History of African American English. Oxford: Blackwell. 233–263. Mufwene, S.S. 2000. Creolization is a social, not a structural, process. In Neumann-Holzschuh, I. & E.W. Schneider (eds). Degrees of Restructuring in Creole Languages. Amsterdam/ Philadelphia: John Benjamins. 65–84. Mufwene, S.S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mufwene, S.S. 2002. Colonization, globalization, and the future of languages in the twenty-first century. MOST Journal on Multicultural Societies 4 (2). 162–193. (Electronic) Mufwene, S.S. 2003. Genetic linguistics and genetic creolistics. Short note. Journal of Pidgin and Creole Languages 18. 273–288. Mufwene, S.S. 2004. Multilingualism in linguistic history: Creolization and indigenization. In Bhatia, T. & W. Ritchie (eds). Handbook of Bilingualism. Malden, MA: Blackwell. 460–488. Mufwene, S.S. 2005. Créoles, Écologie Sociale, Évolution Linguistique. Paris: L’Harmattan. Mufwene, S.S. in press. Population movements and contacts in language evolution. Journal of Language Contact 1. Muysken, P. 2000. Bilingual Speech: A Typology of Code-mixing. Cambridge: Cambridge University Press. Muysken, P. & N. Smith. 1995. The study of pidgin and creole languages. In Arends, J., P. Muysken & N. Smith (eds). Pidgins and Creoles: An Introduction. Amsterdam/ Philadelphia: John Benjamins. 3–14. Nathan, J. 1922. The Census of British Malaysia, 1921. London: Waterloo & Sons. Norman, J. 1988. Chinese. Cambridge: Cambridge University Press. Pakir, A. 1986. A Linguistic Investigation of Baba Malay. Ph.D. dissertation, University of Hawai‘i. Pan, L. (ed.). 2000. The Encyclopedia of the Chinese Overseas. Singapore: Chinese Heritage Centre; Cambridge, Mass.: Harvard University Press.

 Umberto Ansaldo, Lisa Lim & Salikoko S. Mufwene Peranakan Association Singapore. 2006. Peranakan History. The Peranakan Association. Singapore. Accessed 11 March 2007. http://peranakan.org.sg Png, P.-S. 1969. The Straits Chinese in Singapore: A case of local identity and socio-cultural accommodation. Journal of Southeast Asia History 10 (1). 95–114. Rickford, J.R. 1997. Prior creolization of AAVE? Sociohistorical and textual evidence from the 17th and 18th centuries. Journal of Sociolinguistics 1. 315–336. Roberge, P. 2003. Convergence and the formation of Afrikaans. Journal of Germanic Linguistics 14 (1). 57–93. Rudolph, J. 1998. Reconstructing Identities. A Social History of the Babas in Singapore. Aldershot/ Vermont: Ashgate Publishing Co. Schlieben-Lange, B. 1977. L’origine des langues romanes: Un cas de créolisation? In Meisel, J. (ed.). Langues en Contact – Pidgins – Creoles – Languages in Contact. Tübingen: Gunter Narr. 81–101. Shellabear, Rev. W.G. 1878. Baba Malay: An introduction to the language of Straits-born Chinese. Journal of the Straits Branch of the Royal Asiatic Society 2. 168–174. Shellabear, Rev. W.G. 1913. Baba Malay: An introduction to the language of the Straits-born Chinese. Journal of the Straits Branch of the Royal Asiatic Society 65. 49–63. Siegel, J. 1997. Mixing, levelling, and pidgin/ creole development. In Spears, A.K. & D. Winford (eds). The Structure and Status of Pidgins and Creoles. [Creole Language Library 19.] Amsterdam/ Philadelphia: John Benjamins. 111–150. Skinner, G. 1960. Change and persistence in Chinese culture overseas: A comparison of Thailand and Java. Journal of the South Seas Society 16 (1/2). 86–100. Skinner, G. 1996. Creolized Chinese societies in Southeast Asia. In Reid, A. (ed.). Sojourners and Settlers: Histories of Southeast Asia and the Chinese. St. Leonards: Allen & Unwin. Smith, N. 1995. An annotated list of pidgins, creoles, and mixed languages. In Arends, J., P. Muysken & N. Smith (eds). Pidgins and Creoles: An Introduction. Amsterdam/ Philadelphia: John Benjamins. 331–374. Song, O.S. 1923 [1967]. One Hundred Years’ History of the Chinese in Singapore. London: John Murray. Reprinted 1967, Singapore: University of Malaya Press. Tan, C.B. 1988a. The Baba of Malacca. Culture and Identity of a Peranakan Community in Malaysia. Petaling Jaya/ Selangor: Pelanduk Publications. Tan, C.B. 1988b. Structure and change. Cultural identity of the Baba of Melaka. Bijdragen tot de Taal, Land- en Volkenkunde 144. 297–314. Thomason, S.G. 1997. A typology of contact languages. In Spears, A.K. & D. Winford (eds). The Structure and Status of Pidgins and Creoles. [Creole Language Library 19.] Amsterdam/ Philadelphia: John Benjamins. 71–88. Thomason, S.G. 2001. Language Contact: An Introduction. Washington, DC: Georgetown University Press. Thomason, S. & T. Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Thomson, J.T. 1875. The Straits of Malacca, Indo-China and China, or Ten Years’ Travels, Adventures and Residence Abroad. London: Sampson Low; Marston: Low & Searle. Valkhoff, M. 1966. Studies in Portuguese and Creole, with Special Reference to South Africa. Johannesburg: Witwatersrand University Press. Vaughan, J.D. 1879. Manners and Customs of the Chinese in the Straits Settlements. Singapore: The Mission Press. Reprinted 1971, 1992, Oxford University Press.

The complexity that really matters: The role of political economy in creole genesis Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz University of Puerto Rico, Río Piedras

1

Introduction

This chapter considers some of the linguistic consequences of the political economies of the different European colonial regimes in the Americas with the goal of underscoring the considerable impact that social, political, and economic context has on the results of language contact in general, as well as on the results of creolization in particular. In order to do this, we adopt the Matrix of Creolization framework originally proposed by Alleyne (1971) to argue that the social and economic factors which constituted key parts of the matrix within which power relations were realized in the colonies (that is, the political economy of each colonial regime) played a significant role in Creole genesis. Focusing on the Caribbean in the seventeenth and eighteenth centuries, we contend that the political economy which prevailed within the sphere of influence of each major European colonial power differed sufficiently from the political economy of the other colonizing nations to have had palpable linguistic consequences, which are the results of discrete and practices and processes (Woolard 1985: 745), especially in terms of the development of Creole languages. During the first centuries of colonization of the Caribbean, the political economy of the Spanish colonial system differed from that of the British, the Dutch, and (to a lesser extent) the French and the Portuguese. It is our hypothesis that differences in political economy can do much to account for the fact that Spanish-lexifier Creoles are understood as being relatively rare in the Americas (but perhaps not completely absent, see Lipski 2005, 2006). Beyond the specific issue of the ‘missing Spanish Creoles’ (McWhorter 2000), however, we also argue that the differences between the French and English-lexifier Creoles themselves cannot be satisfactorily explained without taking into account the political economies that typified each colonial regime and the links between these manifestations of Empire and current efforts to construct ‘expert’ knowledge about the speech communities distributed within them.

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

Arguments about language genesis are among the most tendentious discussions in contemporary Creole Studies and have been at the center of debate since the inception of the field. Over the last two decades, research on Creole genesis has generally fallen into two areas (Arends 1995): the first is considered ‘linguistic’ and includes the study of early Creole documents; the second area tends toward the ‘non-linguistic’, the investigation of sociohistorical and demographic factors contributing to Creole genesis. This latter area of research underscores the importance of ‘extralinguistic’ phenomena, recognizing elements such as social inequality, politics, and culture as useful in determining and describing the nature of language contact. In this chapter, we look at these two approaches to Creole Studies as interconnected in order to offer an account of how and why strikingly different linguistic scenarios emerged under the different experiments in colonialism carried out by European powers in the Caribbean. This approach differs markedly from other accounts of genesis. However, we are less concerned with presenting our own arguments as singularly authoritative than with grappling with fundamental questions regarding the relationship among various positions that linguists have put forward to account for the origins of Creole languages. In recent work by a number of linguists (e.g. Alleyne 1996; McWhorter 2000; Chaudenson 2001; Aceto 2003), there seems to exist a widening gap between the empiricist and often microanalytical approach with which specific varieties are analyzed and categorized (mainly in terms of superstrate and substrate influences) and the desire to take into consideration relevant (socio)linguistic phenomena from the Caribbean region as a whole. Our efforts to find our bearings amidst these and other positions demand first that we look at the relationship between theory and data from individual languages and then ask if it is possible that the generalizations made about Creoles (e.g. Chaudenson’s 2001 view of Caribbean Creoles as the outcome of one language dominating another) are actually more appropriate if we apply them to a more restricted group of languages, e.g. the French-lexifier Creoles only. Studying these smaller groups of languages one by one rather than considering Creole origins categorically can reveal that seemingly contradictory assertions are less opposed to one another than has often been assumed (Le Page & Tabouret-Keller 1985: 23). One of the most important publications in this area is Mintz (1971). Mintz’s approach squarely addresses broad political and historical issues, yet at the same time it can be said to manifest some degree of reluctance to provide a close analysis of questions that touch on issues such as power relations, control, and marginalization. Surprisingly, this line of investigation has received limited attention from creolists in the hundreds of inquiries into language genesis completed since the first conference of creolists in Mona in 1968. We maintain that this hiatus between empirical discussions of linguistic forms and functions and the critical analysis of constructions such as race, class, nation, and even ‘Creole’ is due on the one hand to the need for more detailed attention to political economy, and on the other to what Foucault (1980) called the dominant ‘regime of truth’, i.e., the embedding of academic discourses in extant proc-



The complexity that really matters 

esses of investing authority in particular methods of producing, distributing, and receiving social scientific knowledge.

1.1

Purpose

In our discussion we will attempt to show that the perceived existence or absence of Creole languages in specific sociohistorical and political-economic situations in the Caribbean region can be explained with considerable precision by examining the ideological and socioeconomic underpinnings of its diverse societies. Moreover, if we position these trajectories within certain sets of macroscopic social boundaries, then we can account for variation across island societies and across historical periods. Our analysis differs from others that focus on questions about the beginnings of Creoles in that they typically bring together what was happening during specific periods to show continuity, rather than to acknowledge and account for difference across sociohistorical situations.

2 Interaction, not simply ‘access’ McWhorter (2000) defends the hypothesis that Creoles were formed in Africa as pidgins and then brought to the Americas as fledgling languages. He calls this process of transfer ‘Afrogenesis’, a term which was coined much earlier by Richard Allsopp (1976) to refer not only to the case of genesis but to the preponderant role of the African substratum. The central argument of the text is its opposition to what McWhorter calls the ‘limited access model,’ or ‘superstratist argument’. This is the claim that “…plantation Creoles of the New World and the Indian Ocean developed as a result of African slaves having had limited access to the lexifier language spoken on plantations, due to the disproportion of blacks to whites in such settings” (McWhorter 2000: 1). The author admits that this is merely one concept that can be extracted from the literature on Creole genesis. Nevertheless, he seems reluctant to accept the fact that no account of genesis by any linguist can be simplistically reduced to this single concept. At the same time, it seems impossible for a theory of language contact that sets out to account for the outcomes of such contact to avoid the ‘access’ factor. We prefer to think of McWhorter’s ‘access factor’ as the ‘interaction factor’, considered not only in terms of density, mode, and access to institutions, but also in terms of daily experiences, practices, and social relationships mediated by language. Under all of these parameters, every case of European-African contact whether in settler or slave societies, or in Spanish, Portuguese, French, English, Dutch or Danish colonies, qualifies as ‘limited’, to some extent, in terms of participation in institutions mediated by language (e.g. school, religion, courts, plantations and other venues for social and economic interaction) and in terms of speakers’ reification of processes and norms that

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

characterize intercultural communication. Yet access is certainly not a categorical case of ‘limited’ or ‘unlimited’, but one of degrees of limitation.

2.1

Correlating colonization and types of interaction

The correlation between the nationality of the colonizing European power (e.g. Spanish vs. British) and the degrees of richness/ poverty characteristic of interaction, itself apparently corresponds with the extent of pervasiveness of African elements in Creoles. Imagine a map of Europe with a linguistic map of the Caribbean superimposed on it. As we move from the Mediterranean (Spain) toward the North (through Portugal, France, England, and into Holland), we also move from relatively few easily identified African elements in the Creole languages generated by these systems of colonization (e.g. the Caribbean dialects of Spanish) to quite substantial African elements (e.g. Berbice Dutch, Saramaccan). This should set us thinking about the effects, not just of ‘access’, but of all the other factors which distinguish one system of colonization from another (e.g. characteristics that distinguish Spanish colonial policy from British colonial policy). In accordance with the critique put forward by McWhorter (2000: 197), we are not talking about who were “the kinder, gentler, slavemasters” but rather about identifiable, systemic interactional patterns. As suggested by Mintz (1961), slavery did not exist as a universal social phenomenon, but was shaped deeply by the different strategies of economic development in the countries and colonies that instituted it. Thus if we want to understand the emergence of the varieties that developed in the case of the colonization of the Greater Antilles, then the relevant question cannot shift to the one McWhorter centers – “why did Spanish Creoles not develop?” This question obscures the diversity of situations within and outside the Spanish Caribbean and pushes aside as irrelevant important sociohistorical circumstances and ideological frames. To counter this act of erasure, evidence such as the following needs to be considered: i. When slavery became important in Puerto Rico and Cuba (but not in what is now the Dominican Republic) in the late 1700s, there were already large numbers of Spanish speakers of ‘mixed’ racial descent on these islands (Mintz 1971: 483). ii. British colonization of the Caribbean was characterized not only by the presence of forts in West Africa, but also by the absence of: a firm Creole culture, a numerically significant intermediate group, and institutions that could unify the colonial population (ibid.: 488). iii. Though the outcomes of linguistic creolization (the English-lexifier Creole languages) within the British Empire were comparatively similar, the paths leading to the existence of these languages were shaped by economies that differ significantly one from the other (see Higman 1995). These three points serve as a reminder that even research on historical events relies on metadiscursive practices by which researchers and scholarship are situated in social, political, and historical terms. The paragraphs below focus on describing the sets of



The complexity that really matters 

conditions that in certain societies led Africans and their descendants to create linguistic systems that lend themselves to be interpreted as Africanized European languages (e.g. the Spanish Caribbean dialects) and in others to create varieties that lend themselves to be interpreted as Europeanized African languages (e.g. the ‘non-decreolized’ English-lexifier Creoles).

3 Beyond correlation: The descriptive and explanatory power of the Matrix of Creolization in relation to key debates in creolistics The Matrix of Creolization that is described and utilized in this chapter, using a framework originally proposed by Alleyne (1971), with its emphasis on interaction, aims to take sociohistorical analysis beyond the quantitative model that serves as the hallmark of much contemporary sociolinguistic work on Creole languages (e.g. the traditionally Labovian paradigm). It rejects the idea that an analysis of genesis should be built on the simplistic idea that ‘language reflects society’. We see, among other things, that language is socially constituted and that language contact in the Caribbean during the time of the Atlantic Slave Trade was influenced by a number of factors, some ideological (e.g. religion, the social construction of race) and others more material (e.g. geography, economy, historical links). Several different frameworks put forward by a number of critics (i.e. Harris 1981; Romaine 1984; Cameron 1997) call for precisely such an approach in contemporary studies. We see their critiques of sociolinguistic methodology as relevant for work that intends to describe the contexts in which Creole languages were formed. With this in mind we see our project (i.e. to better understand the relationship between political economic systems and language genesis) in two parts: our first goal is two-fold, to start with the description of prototypical European systems of colonization (British, French, and Spanish) alongside West African sociocultural systems and then to describe these in terms of patterns of interaction. In doing so, superstrate and substrate influences are each described emphasizing three different sets of phenomena: i. economy; ii. ideology, culture, and linguistics; iii. politics. This schema allows for the contextualization of linguistic interaction within (as well as between and among) individual colonies culturally and historically by showing that language genesis occurred within systems shaped by broader economic, political, and social forces (e.g. patterns of production and consumption, tensions between colony and metropole, traditions that assigned divergent roles to individuals on the basis of factors such as gender, origin, religion, and status). The options/ choices/ paths available to individual speakers and speech communities determined and were determined by these systems. Moreover, speakers and communities are recognized as sustaining

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

and creating anew socioeconomic and cultural structures that rewarded, penalized, demanded, and made unnecessary specific types of linguistic behavior. Focusing on historical situations characterized by rapid social change and the constant renegotiation of the social contract in the Caribbean colonies, our final goal is perhaps the most challenging: to respond to earlier calls for positioning language and sociolinguistic phenomena in terms of matrices, thus our discussion of the Matrix of Creolization. The meta-critiques offered by the aforementioned critics of sociolinguistic theory and practice offer a set of principles that can guide the interpretation of this Matrix so that language can be related to “rationality, intentionality, and the function of social agents and human actors” (Romaine 1984: 26). Such concerns require that we reject the assumption that the structures characteristic of plantation economies existed before the languages spoken in the colonies emerged (Bickerton 1981; Chaudenson 2001). In fact, they remind us that African and European languages are among the ingredients that shaped interactions among and between the races, in the forts of West Africa, on the ships to and from the Americas, in the fields of plantations, and in the transatlantic region’s many struggles for freedom. This view calls for the study of linguistic phenomena in sociohistorical context and releases hierarchies of race, class, and gender (all characteristics of European colonialism which shape any scholarly understanding of language) from their banishment to the category of non-linguistic phenomena. Recognizing these and other manifestations of hierarchy as important to theories of Creole origins, we attempt to create a space in which each individual speech variety is considered “as it emerged, as part of the social, interacting with other modes of behavior and just as important as any of them” (Cameron 1997: 82, original emphasis) and to see the study of language as an ongoing product of academic ‘regimes of truth’ (Foucault 1978, 1980, 1991). For these reasons, we find it necessary to insist in the most emphatic manner possible that the various configurations of African, indigenous Caribbean, and European languages that emerged in the Caribbean Creole space had a significant influence over the economic, political, and ideological factors that in turn played a role in shaping Caribbean Creole languages and cultures. In our view, this complex and dialectical relationship among language, means of production, power relations, and ideology constitutes the central dynamic, not only in the development of individual varieties and characteristically Caribbean forms of language, but also in the processes of language variation and change at work in all of the languages on the planet. However, considering that the historical contexts in which knowledge about processes of genesis and variation has been produced are a key component in this relationship, this discussion is not limited to the roles that speakers played in creating language. This is because even the most ‘objective’ of linguists’ descriptions impose systems of classification that are “doubly determined”, linked to “overt political taxonomies” and “the struggle over the monopoly of legitimate ideological production” (Bourdieu 1991: 168–169). A key debate in creolistics which the Matrix of Creolization can help us to address is that surrounding the notion of Creole Exceptionalism (see DeGraff 2001, 2003, 2004,



The complexity that really matters 

2005; Bickerton 2004; DeGraff & Walicek 2005). The position against Creole Exceptionalism has been most vociferously and rigorously advanced by Michel DeGraff (2003). Like some of the other discourses gaining currency among contemporary creolists, DeGraff ’s key argument resuscitates and elaborates a position that received some attention in the 1970s which suggested that: (1) Creole genesis is not a special phenomenon requiring special theories and processes; (2) genesis can be accommodated within general principles of language contact and language change; and (3) within this set of general principles, there is no subset of linguistic principles that applies specifically and exclusively to the development of Creoles as a group. Immediately called into question is the need for a single special genesis hypothesis to account commonly for all Creoles, as well as linguists’ desire to treat these languages categorically as a distinct typological or phylogenetic class. This set of assertions has far-reaching implications for future work in creolistics, since apart from the issue of genesis, Creole Studies could be argued to be just like any other linguistics. As Mufwene (2001: 75) explains: There is really no particular reason why the developments of creoles should not be treated as consequences of normal linguistic interaction in specific ecological conditions of linguistic contacts involving not only speakers (as in any monolingual speech community) but also different language varieties. Creoles should prompt us to rethink some established assumptions about language change and the role of ecology.

These and related concerns have lead some to question why we have a special sub-discipline called creolistics (e.g. Muysken 1988; Mühlhäusler 1993; Lefebvre 2000). The main arguments put forward in this chapter attempt to show that it is impossible to describe and analyze Creole languages in any comprehensive way without taking into account such ‘non-linguistic’ or ‘extralinguistic’ factors as political economy. If we are able to prove this hypothesis and if Creoles are not exceptional, then we must also conclude that not one of the world’s languages can be described or analyzed without taking into account the relations of power in the community or communities where it is spoken. In this way, the sharp distinction between ‘linguistic’ and ‘non-linguistic’ mentioned in section 1 above must be abandoned, not just in the study of Creoles but in the study of any human language. Moving in this direction, we take issue with Chaudenson’s (2001: 314) warning that “it is hazardous to extend data and theories concerning creolization to the analysis of Creole cultural systems”. Ironically, DeGraff ’s anti-exceptionalist position and his emphasis on the language-ideology interface, when taken to its logical conclusion, in one way actually supports McWhorter’s contention that creolistics has played and will continue to play a very special role in linguistic science. This is because among all human languages, Creoles are often the ones that force us most to question such fundamental but highly dubious linguistic notions as langue vs. parole, competence vs. performance, I- vs. Elanguage, internal vs. external change, L1 vs. L2 acquisition, simplification vs. complexity, the genetic classification of languages, the idealized monoglossic speaker in a

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

homogeneous language community, language as a computational system, etc. Thus Creoles, precisely because they are not exceptional have the potential to play an exceptional role in redefining what is ‘normal’ both in terms of the objects of linguistic study and in terms of the conceptual frameworks that we use to study them.

4 Toward a typology of colonization and creolization: Political economy and the continua, matrix, and space of Afro-Caribbean creolization In this work, we use the term ‘political economy’ to refer to a constellation of political, economic, and ideological/ cultural/ linguistic parameters that define a typology of colonization and creolization created by two of the principal agents in the Afro-Caribbean creolization process from the end of the 15th century to the end of the 18th century: (i) the African and Afro-Caribbean working classes (substrates) and (ii) the European and Creole ruling classes (superstrates), i.e. the two migrant groupings [to the Caribbean] – the masters and the slaves – neither of which was able to transfer more than a portion of its cultural traditions to the islands (Mintz 1971: 484). Although we recognize the important contributions of other populations in this process, such as indigenous Caribbean peoples, indentured laborers, the Irish, and other European working classes, the scholarly attention that they so desperately deserve is beyond the scope of this chapter. Following the model in Table 1, we will refer to each of the continua of possible settings in relation to each of the aforementioned parameters as an Afro-Caribbean Creolization Continuum. When considered together, the entire set of Afro-Caribbean Creolization Continua defines the Afro-Caribbean Creolization Matrix schematized in Figure 1, following a model suggested by Alleyne (1971). The Afro-Caribbean Creolization Matrix, in turn, can be used to describe the Afro-Caribbean Creolization Space that typifies a particular Caribbean island society at a particular time in its history. The Afro-Caribbean Creolization Space changes shape as the parametric settings change along one or more of the continua that defines its Matrix. Because the substrate and superstrate economies, politics, and ideologies differ in time and space, the AfroCaribbean Creole Space differs from one period of time to another, from one island society to the next, and within a given island society.



The complexity that really matters 

Table 1.  Typology of colonization and creolization English colonies

French colonies

Spanish colonies

Nature of colonial era creolization Less interaction between Africans & others Largely restricted to Afro-Caribbean population African influences concentrated but narrow Creoles as markers of difference/ resistance Europeanized African languages

More interaction between Africans & others Encompasses entire society Dispersed but broad Creolization of entire culture Africanized European languages

Superstrates (European and Creole ruling classes) Superstrate economies Sugar monoculture

Mixed sugar/ non-sugar economy

Big sugar established well before 18thC Early switch to private corporate financing More labor-intensive agriculture More capital-intensive agriculture Large plantations Crops require local industrial processing Sugar production encouraged No early sugar production in metropole Owners remote or absentee Inhospitable climatic conditions

Non-sugar economy

Big sugar established just before 19thC Late switch to corporate financing Less labor-intensive agriculture Less capital-intensive agriculture Smaller plantations, individual blocks/ plots Little or no local processing required Sugar production discouraged Early sugar production in metropole Personal relations between owner and slave Wider range of climatic conditions

Superstrate ideologies/ cultures/ linguistics Colonization = purely economic enterprise Maximizing profits/ primitive accumulation No recent experience of ‘ethnic cleansing’ Stress on apartheid Modern capitalist notion of slavery Rigid concept of slavery/ less mobility Rigid/ binary concept of race Strong tendency to avoid interracial contact Less powerful church Less aggressive evangelism

Colonization = ‘civilizing’ mission Integration into metropole Extension of Reconquista/ Inquisición Stress on spread of metropolitan culture, language Medieval Mediterranean notion of slavery Fuzzy concept of slavery/ more mobility Fuzzy/ graded concept of race Interracial contact common More powerful church More aggressive evangelism

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

Protestantism/ Calvinism/ notion of ‘God’s elite/ elect’ Church supports planters Limited prior contact with Africa No prior creolization with Africans Remote from Mediterranean contact zone

Catholicism/ universalism Church opposes planters Significant prior contact with Africa Prior creolization with Africans Part of Mediterranean contact zone

Superstrate politics Colony-based/ decentralized/ rural Slave colonies Planters form conscious class from 17th C Planters vs. slaves Government by and for planter class Elites retain European identity Bourgeois civil law New bourgeois slave laws

Metropole-based/ centralized/ urban Settler colonies No conscious planter class until late 18th C Metropole vs. colonial elite Metropole often opposes planters Elites adopt local Creole identity Feudal Crown law Old feudal slave codes

Substrates (African and Afro-Caribbean peoples) Substrate economies/ demographics Reluctant involvement in formal economy Struggle for subsistence plots Marginalized subsistence economies Greater ratio Africans: others Few or no peasants Afro-Caribbean population replaced by import Higher ratio slaves: emancipated Higher sex ratio (M:F) among slaves Low longevity and birthrate among slaves Most slaves field hands on large plantations Mutual survival bonds  community apart

Creation of full subsistence economies Many Afro-Caribbean people become peasants Creation of colony-wide subsistence economies Lower ratio Africans: others Large peasant sector A-C population replaced by natural increase Lower ratio slaves: emancipated Normal sex ratio among slaves Normal longevity and birthrate Most slaves domestics, craftspeople Integration into colony-wide community



The complexity that really matters 

Substrate ideologies/ culture/ linguistics Heteroglossia among Afro-Caribbean peoples Hybridization among Afro-Caribbean population Defensive/ offensive reinforcing of color lines Fewer interracial families Creation of ‘Black’ language (‘creoles’) Creation of cultures of identity and resistance

Colony-wide syncretic heteroglossia Colony-wide hybridization Breaking color lines More interracial families Africanization of colony-wide language Africanization of colony-wide culture

Substrate politics Politics of identity and opposition Counter-identification/ disidentification Afro-Caribbeans form communities apart Creation of marginal political spaces Strikes, go-slows, ‘quota’ system, rebellions Runaways, maroon republics Long struggle for emancipation and land

Politics of transformation Identification/ disidentification Alliances with others for subsistence Africanization of colony-wide political space Establishment of subsistence economies Massive subsistence ‘squatter’ peasantry Early manumission and access to land

From 1500 to 1800, the substrate and superstrate agents under one colonial regime (e.g. Spanish, British, French) typically situated themselves differently along each continuum than did their counterparts under another regime. In the vast majority of cases, the political-economic regimes established by the Northern Europeans (primarily the Dutch and British) define one pole of each continuum (hereafter the Northern pole), while those of the Southern Europeans (the Spanish and to a lesser extent the Portuguese) define the other pole (hereafter the Southern pole), with the French beginning their colonial trajectory in the Caribbean somewhere in between, but moving decisively toward the Northern pole by the end of the 17th century (Chaudenson 2001: 102, 113). For example, while the British generally attempted to establish sugar monoculture plantation economies in their Caribbean territories, the Spanish established more diversified small-holder settler economies in their possessions (Stinchcombe 1995: 95). Similarly, the Afro-Caribbean working classes on the Spanish islands (apart from their experience with slavery, which up until the turn of the 19th century, was often limited to a single generation) participated as agents in the formal economy, playing a relatively active role in shaping that economy as subsistence producers involved in horticulture and/or fishing, traders, and craftspeople, whereas their counterparts on the British islands involved themselves more reluctantly in the formal economy, reserving their highest levels of energy and enthusiasm for their participation in the informal economies that they themselves created in the interstices of slave society, such as household plots, provisions plots, and Saturday or Sunday markets (Tomich 2000).

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

SUPERSTRATE ECONOMIES

SUPERSTRATE IDEOLOGIES

SUPERSTRATE POLITICS

CREOLE SPACE

SUBSTRATE POLITICS

SUBSTRATE IDEOLOGIES

SUBSTRATE ECONOMIES Figure 1.  The Afro-Caribbean Creolization Matrix, following Alleyne (1971)

It should be borne in mind, however, that even under a single colonial regime, each Caribbean island, each speech community, and each speaker of a Caribbean Creole represents a unique Creolization Space. For example, while the early success of sugar in the lowlands of Jamaica under the British made them a paradigmatic example of the prototypically Northern Creolization Space that came to characterize the British Caribbean, the early failure of sugar and most other plantation crops on British Anguilla,



The complexity that really matters 

the subsequent departure of the majority of the planter class from the island, and the relatively equitable redistribution of land among the Afro-Caribbean population that remained made the Creolization Space on Anguilla very different from that inhabited by most of the populations of the British colonies in the Caribbean. Moreover, the Creolization Space on a particular island could and did change radically at different points during its colonial history. An obvious example is Jamaica, where the shift from Spanish to British colonial rule in 1655 brought about the eventual complete reorientation of the Creolization Space there from a typically Southern to a typically Northern one. Even under the same colonial regime, however, the Creolization Space did change significantly over time, as was the case in the Spanish islands toward the end of the 18th century, when the Bourbon Reforms were reshaping the political economy of Spanish colonialism to conform more closely to the Northern model (Stinchcombe 1995: 15). Following Table 1, we will now consider where the political economies of the different colonial regime situate themselves along the various Afro-Caribbean Creolization Continua during the period from 1500 to 1800.

4.1

Superstrate economies

In the pre-19th century colonial Caribbean, the economic continua created by the European and Creole ruling classes are defined primarily by a capital intensive plantation slave economy based on sugar at the Northern pole and a diversified economy of small holdings, mines, ranching, limited and largely unsuccessful plantation agriculture, etc. at the Southern pole (Mintz 1971: 484–486). Craton (1997: 8–31) traces the history of sugar cultivation on the Iberian peninsula as well as in the Caribbean. Sugar had been cultivated in Morocco (perhaps using African slaves) since the 9th century and the Moors made Iberia a center for sugar production. The sugar industry expanded in both Portugal and Spain immediately after the Reconquista, first using moriscos (Muslims who had been forced to convert to Christianity) as serfs and later by utilizing African slaves. Developed in the Canary Islands in the early 16th century, the colono system, whereby cane was grown and processed in small units by a mixture of share-cropping peasants from Spain and African slaves became the model for sugar production in much of the Spanish Caribbean. A similar system typifies early sugar cultivation in Brazil. Under this system, the Portuguese were able to make Brazilian sugar an essential component of the vibrant transatlantic trade involving Portugal, West Africa, and Brazil which was in place by the end of the 16th century. When they conquered northeastern Brazil in 1630, the Dutch completely revamped the sugar industry there, by increasing both the size of production units and the inflow of African slaves from their newly conquered forts in West Africa, by making credit available for consolidation of holdings, machinery, provisions, etc. by introducing the most advanced techniques for growing and processing cane, by laying the

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

groundwork for an extremely rigid and racially based slave code, and by consolidating and expanding the transatlantic trade first established by the Portuguese through the incorporation of Brazilian sugar production into their vast refining and distribution networks. The Dutch also tapped the commercial and cane production expertise of Jewish refugees from the Portuguese Inquisition in Brazil, given that the Jews had played a key role over the preceding four centuries in proto-capitalist sugar plantation experiments in Palestine, Cyprus, Crete, Sicily, and Iberia itself. When the Dutch and their Jewish allies were expelled from northeastern Brazil some twenty years later, they transferred this powerful mercantile-agro-industrial complex to their colonies in the Guianas, to the French colonies of Guadeloupe and Martinique, to British Barbados, and to St. Croix (Chaudenson 2001: 136). Within a few decades of adopting the Dutch ‘method of Pernambuco’, Barbados had become the first Caribbean territory to qualitatively transform sugar production in such a way as to make it the cornerstone for the establishment of capitalism as a world system. By the end of the 17th century, the British had wrested control of the seas from the Dutch. In the process, they completely appropriated the Dutch model and switched from an earlier preference for tobacco and other crops that could be cultivated on smallholdings to focus on large-scale sugar monoculture. But the Dutch emphasis on control over trade eventually shifted with the British to an emphasis on control over industrial production. In this way, the transatlantic trade became the primary engine for the development of British manufacturing capacity in the early 18th century and the major force behind the Industrial Revolution that followed (Williams 1944; Stinchcombe 1995). Meanwhile, Brazil reverted for the next two centuries to the pre-capitalist mode of sugar production that the Dutch sought to replace there (Craton 1997: 29). Thus, the two major poles of the superstrate economic continuum were established in the Caribbean. At the Northern pole, the British favored sugar monoculture on large, densely populated, labor- and capital-intensive plantations situated in low lying, wet, disease ridden environments, which relied heavily on private corporate finance, slave labor, a significant degree of local industrial processing, metropolitan refining and distribution networks, and absentee plutocratic European landowners who used a group of European or Euro-Caribbean managers to operate their holdings. At the southern pole, the Spanish actually discouraged large-scale sugar production through high taxes and monopolistic shipping practices because it competed with their own domestic sugar industries and diverted scarce labor resources from more appealing and immediately profitable economic activities such as mining and ranching (Batie 1976: 212). Spanish agriculture was more settler- and smallholder-based, could be practiced in more less disease-prone climatic regions than sugar, and if slaves were involved, they often worked alongside their owners or Euro-Caribbean and mestizo peasants (Mintz 1971: 486; Stinchcombe 1995: 8, 108, 132). The French switched definitively from experimentation with a variety of crops such as tobacco and indigo to sugar after the English did so, with Martinique and Guadeloupe remaining sparsely populated and French colonial shipping levels ‘medio-



The complexity that really matters 

cre’ until the end of the 17th century (Batie 1976: 11; Butel 2000: 198–199; Chaudenson 2001: 102; Parkvall 2000: 157). The Dutch played an even more prominent role in the reorientation of French colonial production to sugar than they had in the case of the British (Batie 1976: 219–220). Not only were Dutch planters, merchants, and creditors an integral element in the initial forays into sugar cultivation on Martinique and Guadeloupe, but in France as well the Netherlanders dominated large portions of the processing and trade of colonial products, to the extent that the French bureau for colonial trade was first located in the Low Countries (Antwerp) rather than in France itself (Stinchcombe 1995: 73; Butel 2000: 196–197). With the spectacular rise of sugar production in St. Domingue (Haïti) during the 18th century, French colonial trade began to outstrip that of Britain itself. But while Britain was attempting to extract itself from the web of monopolistic colonial mercantilism, minimalizing the influence of outsiders over its economy, and focusing on using the transatlantic trade for the expansion of its own domestic industrial sector, French commerce remained based on the exploitation of privilege, colonial exclusivity (monopoly), dependence on others (especially the Low Countries and Germany), and the re-export of colonial produce to the North of Europe and the Levant, rather than industrial production (Butel 2000: 202–204). Stinchcombe (1995) demonstrates how the involvement of first the Dutch, then the British, and then the French in sugar monoculture parallels the general corporate reorganization of colonial commerce at the northern end of the superstrate economic continua, which occurred first in the Netherlands toward the end of the 17th century, then in England in the early 18th century, and finally in France toward the middle of the 18th century (ibid.: 57–59). The gradual establishment of a complex network of contractual agreements governing privately owned and controlled corporate finance, insurance, and marketing networks for long distance shipping led to a decisive shift from a feudal mode of production to a mercantile capitalist one in the home ports of the colonial trade in Northwestern Europe. The Spanish, however, continued to “manage their colonial commerce in corporatist government convoys … insurance, inspection, route of travel, the conditions of commerce, and the like were strongly regulated by the central government” (ibid.: 59). The long distance trading ship, with its gear, cargo, finance, and insurance in the hands of several different private shareholder companies and with its closely supervised crew members expected to work eighty-two hours per week for a wage with maximum care and cooperation with co-workers to ensure mutual survival against rough seas and hostile attacks, is seen by many as the prototype for the organization of production under later stages of capitalism, such as the industrial factory (Linebaugh 1992) and the sugar plantation itself: The origin of capitalism, in the sense of extraction of surplus value from free labor by finance capital through corporate or ‘bureaucratic’ supervision of that labor, was first found in port cities and on the high seas….Capitalism was first a world

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

system, and only later penetrated national economies….Plantations were also ‘bourgeois’ or ‘capitalist’ institutions, oriented to production in a market, involving investment of other people’s money, but they did not employ much proletarian wage labor (Stinchcombe 1995: 58).

Craton (1998: 155) cites a number of studies that characterize sugar plantation slaves as part-proletarian or proto-proletarian, especially in relation to their ability to organize themselves to improve their conditions and defend these gains over hundreds of years. He also traces the origin of the plantation work gang to the organization of ships’ crews (1997: 53–54). Some of the smaller islands of the Caribbean, like long distance trading ships, provided the colonial powers with near ideal experimental conditions for amassing a body of knowledge and experience about the extraction of maximum value from human labor in general, from both the visible and invisible work performed by women in particular (but also men), as well as from nature and the land (Mintz 1971: 483; Moitt 2000).

4.2

Superstrate ideologies, cultures, and linguistics

After the very initial stages of the colonial period, two clear conceptions of colonization emerged for the dominant classes. For the British, colonization was chiefly a capitalist economic enterprise, while for the Spanish colonization was not only about accumulating wealth, but also about a mission to integrate the peoples whom they conquered into Spanish and Catholic ‘civilization’. The differences between the Northern and Southern poles of the ideological and cultural continua created by the European and Euro-Caribbean ruling classes are often related to the Protestant-Catholic divide among them. In this scenario, the Spanish with their integrationist feudal ideology of ‘civilization’ fall squarely on the Catholic side, the Dutch and the British with their segregationist capitalist ideology of wealth as a blessing of God upon an divinely designated elect/ elite group fall clearly on the Protestant side, while the French (with a largely Catholic ruling class and a largely Protestant commercial class) and the Portuguese (with a largely Catholic ruling class and a largely Jewish commercial class) fall somewhere in between (see, for example Goveia 2000 and Le Page & Tabouret-Keller 1985). All of the principal European transatlantic powers underwent major religious upheavals at the beginning of their colonial endeavors. The Christian Reconquista of the Iberian peninsula from the Muslims immediately preceded the first Portuguese voyages to Africa and the first Spanish voyages to the Caribbean, while the initial period of both Portuguese and Spanish colonialism coincided with the Inquisición, which resulted in the nearly complete ethno-religious ‘cleansing’ of those two countries by the beginning of the 17th century. In many ways, the ‘civilizing’ ideology of the Reconquista and Inquisición came to define the Spanish colonial project, in which the state forged an intimate alliance with the church designed to spread the Spanish language, Spanish culture, and the Catholic faith to all of the inhabitants of the empire. In this



The complexity that really matters 

alliance the priests played a particularly influential part in determining the policy of an increasingly centralizing and interventionist bureaucracy (Mörner 1967: 35, 47–48). The Protestant Dutch fought a bitter war against their Catholic rulers in Spain which not only led to the independence of the northern Netherlands in 1581, but shaped the entire initial period of Dutch colonialism as a militaristic, patriotic, Protestant, and proto-capitalistic ‘crusade’ against the Catholic kings. During the first half of the 17th century, the Dutch consistently put their financial resources, trading networks, and commercial and militarily expertise at the disposition of their British and the French allies in order to counter their Spanish and Portuguese enemies. By the end of the 17th century, however, first the British and then the French had so effectively mastered and perfected Dutch-style mercantile and agro-industrial capitalism, that the Netherlanders were thereafter forced into a subsidiary position in the Caribbean. Nevertheless, the groundbreaking role played by the Dutch in the worldwide transition to capitalism left an indelible mark on Northern cultures, with such key institutions as the nuclear family and bourgeois ‘home’ first consolidated on a society-wide basis in the Netherlands before spreading to the rest of northern Europe (Rybczynsky 1986: 52–54). In Britain, the struggle between the feudal-leaning Anglican Stuart monarchy and the capitalistic Puritans framed the establishment of the first English transatlantic colonies, all of which failed or were on the brink of failure until the Civil Wars brought the Puritan Cromwell to power. Cromwell and his successors from the Dutch House of Orange succeeded in making the transition to capitalism inevitable in Britain, in no small measure by reshaping the country’s colonial policy in the image of that of the Netherlands. In contrast to the Spanish, the Dutch and the British showed little interest in spreading their language, culture, and religion to the people they colonized, especially in the initial period of transatlantic expansion (Craton 1997: 154). In general, intimate and frequent interaction between Euro-Caribbean and Afro-Caribbean populations was much more limited in the British and Dutch colonies than in the Spanish and Portuguese colonies, with the French (especially during the initial period of colonization) once again situating themselves somewhere between the Northern and Southern poles of this continuum. France’s first transatlantic voyages came at a time when that country was engulfed in a bloody confrontation between Catholics and Protestant Huguenots. Although the Huguenots led the initial French efforts to challenge the supremacy of Spain in the Americas, none succeeded until the Huguenots were politically defeated and peace was made with the Spanish at the end of the 16th century (Craton 1997: 37). The French Protestants, however, could be said to have lost the religious battle but to have won the war between feudalism and emergent capitalism, and thereby continued to play a leading and often predominant part in colonial commerce. The French were ambivalent, if not schizophrenic, in their conceptualization of their colonial mission. While the Protestant commercial class showed little or no interest in the spread of French language or culture, and were largely prohibited from either owning slaves or from carrying out proselytizing activities in the colonies, the Catholic ruling class put some emphasis on the spread of

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

French ‘civilization’ and the conversion of all of their subjects, including the enslaved, to Catholicism (Goveia 2000: 590). It is no accident, then, that the first colonies to eventually be officially integrated into metropolitan political structures, with parliamentary representation were French: first St. Domingue for a brief period at the end of the 18th century and finally Martinique and Guadeloupe toward the end of the 19th. Dominant attitudes toward race and slavery also differed significantly between the Southern and Northern colonial powers. In distinction to the Dutch and the British, the Spanish and Portuguese were no strangers to African people, civilizations, or slavery as they entered the colonial era (Oliver & Atmore 2001; Roberts 2004: 18). As an integral part of the Mediterranean Contact Zone, most of the Iberian peninsula had actually been ruled by Africans and could be said to have undergone more than five centuries of creolization with an African Islamic superstrate and a Catholic Romanic substrate, just prior to the first Portuguese and Spanish intercontinental voyages. Sugar plantations, worked first by Iberian and then by African slaves were an important part of the local economy before, during, and after the Reconquista. Both the Spanish and the Portuguese carried to the Caribbean a set of ideologies and attitudes toward race and slavery that had characterized the Mediterranean world for millennia. Given the fact that the Mediterranean is an area in which Africa, Europe, and Asia meet, the rich variety of skin shades, facial features, body shapes, etc. found among its peoples make racial categorization a graded, flexible, and sometimes even imprecise process. Because most Mediterranean peoples had both enslaved other people and been enslaved by others under a plethora of pre-capitalist modes of production over the years leading up to the colonial period, often their ideas about slavery were not bound to a particular mode of production, a particular ‘level’ of civilization/ humanity, or a particular race of people. For the Dutch, British, and perhaps a lesser degree the French, African people, civilizations, and slavery were relatively unknown until the advent of colonial plantation slavery, especially in its most capitalistic form adapted to the production of sugar. Capitalist sugar production in the Caribbean as developed by the Dutch and subsequently perfected by the British and French required a type of slavery which was unprecedented in history in terms of its rigidity, lack of possibility of manumission, association with a particular ‘level’ of civilization/ humanity, and dependence on a binary, inflexible, and precise determination of race. Because sugar cultivation and the trading activities that supported it eventually became the dominant economic paradigm in the Dutch, British, and French Caribbean, the concepts of race and slavery necessitated by capitalist sugar production predominated in the regimes that they imposed. Particularly during the initial years of colonization, many more European males than females migrated to the Caribbean. The sex ratio was particularly skewed in this direction among the Spanish and Portuguese, as well as for a briefer period among the French (Chaudenson 2001: 99, 102). While children of European or Euro-Caribbean fathers and African or Afro-Caribbean mothers appeared everywhere in the Caribbean, under the British and French regimes, these children were usually not recog-



The complexity that really matters 

nized by their fathers and were considered to be Black and eligible to be enslaved. In the Spanish Caribbean, however, things were very different. The majority of the nonindigenous population of all of the Spanish islands was of mixed European, Indian, and/ or African blood from the first Creole generation onward (Stinchcombe 1995: 169). Children of European or Euro-Caribbean fathers and non-European mothers were normally claimed by their fathers and raised practicing their father’s culture and religion as well as speaking their father’s language. Neither a binary Black/ White nor even a ternary Black/ Colored/ White distinction could handle such a situation. Instead both de jure and de facto racial categorization was much more nuanced, with people being classified into dozens of imprecisely defined castas, whose legal and social boundaries were fluid to the point that one could purchase cedulas de gracias a sacar in order to remove oneself from one racial category and become an ‘honorable’ (but still sometimes vulnerable) member of another (Mörner 1967: 45). Besides marriage and birth to a European or Euro-Caribbean father, there were many more avenues open to slaves to obtain a greater degree of freedom in the Spanish and Portuguese Caribbean than in the sugar colonies of the Dutch, the British, and to a slightly lesser extent, the French. Rates of manumission indicate that the French (even in St. Domingue) were more likely than the English to free their slaves (Mintz 1971: 481; Pérez 1988: 63–64; Stinchcombe 1995: 95, 116). In fact, giving attention to the varying nature and frequency of processes through which the enslaved gained freedom may do more to explain the widely recognized differences in race relations across the Caribbean than do narrow assertions about the character of slavery in each case of European colonization.

4.3

Superstrate politics

Two very distinct legal frameworks help define the two poles of the political continua created by the European and Euro-Caribbean ruling classes: the feudal laws of the Southern European monarchies which prevailed in the Spanish Caribbean and the bourgeois civil law of the northern European port cities which held sway first in the Dutch and then in the British territories, with significant elements of British colonial law and law enforcement mechanisms modeled directly on those of the Netherlands (Linebaugh & Rediker 2000: 33). Commenting on this Northern model of governance and its proto-capitalist precedents in Italy and Catalonia during the late Middle Ages, Stinchcombe (1995: 85) observes: The early colonies of the Netherlands, France, and England were…governed dominantly by ‘civil law’ basically the law of the port cities. Civil law is distinguished from criminal and administrative law mainly by providing state power to force compliance with contracts voluntarily entered into…. It was contracts, together with the organization of corporations…that constituted the central government forms of early colonies….Colonies in their very first beginnings had

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

constitutions rather like ‘ships that happen to be on land.’ Later on they came to have a structure more like corporate subsidiaries of the great colonial merchant companies….then without changing much…they became territorial governments of the colonies of [the British, French, and Dutch] empires.

Colonial rule in the Spanish Caribbean was highly centralized and tightly controlled by the monarchy through an urban-based bureaucracy. The Spanish established wellplanned cities in their Caribbean territories, with all of the infrastructure needed to project metropolitan power, including impressive buildings to house government officials and their staffs; military installations (barracks, armories, etc.), and centers for the propagation of ideology such as cathedrals and universities. British colonial administration in most of its Caribbean colonies was decentralized, with assemblies of local planters enjoying a considerable degree of autonomy from the metropole in matters of local governance. Rule under the British was generally plantation-focused and governmental infrastructure was the minimum necessary to ensure a profitable return for investors. Politics in the British Caribbean was that typical of Slave Colonies and therefore was defined on the whole by the relationship between planters and their slaves, whereas in the Spanish Caribbean the key political relationship was that typical of Settler Colonies and therefore was defined by the relationship between the metropolitan administration and the locally born Creole elites (most of whom were not sugar planters before the 19th century). Wishing to avoid the commercial failure of the Spanish colonial model, the French attempted to replicate the plantocratic system practiced by the British, to the extent allowed by the extremely centralized nature of French governance (Mintz 1971: 487; Stinchcombe 1995: 74; Craton 1997). While laws were made by local councils and assemblies in the British Caribbean, the laws in the French territories were made by the French monarchy and its officials as well as by local councils (Goveia 2000: 589). The degree to which sugar plantation slavery dominated the economy and politics of each colonial regime is reflected in the laws put into place under each metropolitan power to regulate slavery as an institution. Up until the end of the 18th century, slavery in the Spanish Caribbean was governed according to the general principles laid down in the 13th century Siete Partidas, the legal code which formed the framework for common law in the Spanish colonies. Under Spanish law, slaves were guaranteed the right to marry, to the services of priests, and to buy themselves out of slavery (Stinchcombe 1995: 171). Because slavery never completely dominated the economies of the Spanish Caribbean during this period, Goveia (2000: 581–583) states that: the Spanish slave laws were less completely adapted to the will of the slave-owning ‘planter’ than was the case elsewhere [in the West Indies]….the principle of slave law was, on the whole, a principle friendly to the protection of the slave and to his claims for freedom. For the Partidas envisaged the slave as ‘a persona’ and not as ‘mere property’….In the Partidas slavery is undoubtedly accepted as legal. It is not accepted as good…[S]lavery is looked upon as a misfortune, from the conse-



The complexity that really matters 

quences of which slaves should be protected…because they are men [sic], and because man [sic] is a noble animal not meant for servitude….The truth is that this ‘medieval’ slave code was probably the most humane in its principles ever to be introduced in the West Indies….The relative despotism of the Spanish government acted as a check on the local oligarchies, which did not necessarily share the view of slavery expressed in the Siete Partidas.

Because the British sugar islands had become fully-fledged slave societies under local control of a conscious class of planters whose very existence depended on slavery (Stinchcombe 1995: 53), Goveia (2000: 583–585) asserts that the planter-designed slave laws that held sway in the British Caribbean: left the power of the master over his [sic] property, the slave, virtually unlimited….The basic conception of the English law in relation to the slave was not, as with the Spaniards, that he [sic] was an inferior kind of subject. It was rather that he [sic] was a special kind of property….[P]rotective enactments were relatively few and sometimes rather ambiguous. Police regulations occupied the most ample proportion of the attention of British West India legislators. In the…British slave laws…the dominant tendency was to recognize the slave as ‘a persona’ in a sphere far more limited than that allowed him [sic] in either the Spanish or French slave law. English slave law almost totally neglects the slave as a subject for religious instruction, as a member of a family, or as a member of society, possessing some rights, however inferior. Insofar as the slave is allowed personality before the law, he [sic] is regarded chiefly, almost solely, as a potential criminal.

The Code Noir, which was formulated in 1685 to regulate slavery in the French Caribbean, was similar to the Spanish Siete Partidas in that it was a compilation of laws that was imposed fairly uniformly on all colonies by a centralized metropolitan administration. But the Code Noir was in many respects much closer to the British slave laws than to the Siete Partidas in its actual purpose and intent. In this connection, Goveia (2000: 588–591) states that: The Code Noir bears some resemblance to the Siete Partidas because they both were influenced to some extent, by the concepts of Roman and canon law. Nevertheless, it more fundamentally resembles the slave laws of the British West Indies by reason of its intention and function….[B]efore the Code Noir was instituted, the French colonies already possessed a fairly comprehensive series of slave laws, and…the Code Noir really may be regarded as an extended codification of these laws….Some of these laws were made by [royal] officials and some by the [local] Council; and perhaps it is significant that the Council appears to have concerned itself mainly with police laws. It is notable, however, that the Council, as a court, heard cases arising from the cruelty of masters and…had made judgments punishing cruelty. This point is probably significant of a contrast in attitudes in the British and French islands arising from a contrast in their political traditions.

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

In British law, the tendency was to limit the sphere of the influence of the crown, and to foster, in particular, a respect for the rights of private property. In France and its colonies, because the power of the crown was less limited, the sphere of interference, even with private property, was commonly accepted to be much wider. The slave, by being private property, did not cease to be in his [sic] person a matter of public concern, and public interference in the management of slaves was more taken for granted at this stage in the French West Indies than it was in the British islands at the same time….[T]he Code Noir was…based on a wider conception of the slave as a ‘person’ and on a different conception of public order [than in the British slave laws]….[Slaves] were to be baptized and instructed as Catholics and their overseers could be of no other religious persuasion. They were to observe Sundays and the holidays of the church, to be married, and if baptized, buried on holy ground….[A slave was] allowed to make complaints to the….attorney general [who] was thus given a status as protector of the slaves….Manumissions were made easy for all masters who had attained their legal majority.

The Code Noir also fixed minimum allowances of food and clothing for slaves and prohibited concubinage between masters and slaves. Stinchcombe (1995: 141) echoes Goveia’s assessment, ranking the British as the least likely and the Spanish as the most likely to treat their Caribbean slaves as they treated free people, with the French somewhere in between these two poles. But, significantly, he also points out that under the Spanish, ‘free’ people often were subjected to forms of coerced labor (ibid.: 100). In any case, with the success of capitalist agro-industrial sugar production in the French Caribbean in the 18th century and in the Spanish Caribbean in the 19th, the Code Noir and the Siete Partidas, as well as the ideologies surrounding their formulation and interpretation, gradually lost ground in favor of much more repressive legislation and practice (Goveia 2000: 583, 591).

4.4

Substrate economies

Most of the Africans taken to the Caribbean as slaves from 1500 to 1800 would have come from communities along the West African coast and its hinterland. While we recognize that tendencies to idealize and romanticize subsistence economies are to be avoided at all costs, we cannot ignore that inalienable land tenure, alongside a very successful complex of economic activities including agriculture, horticulture, animal husbandry, hunting, fishing, food processing, trading, metalworking, and a host of other crafts and manufacturing pursuits had guaranteed West Africans a life of relative abundance for thousands of years (Nzewunwa 1980; Faraclas 2005). Despite the inequalities that existed in a number of the many different social configurations found in the region, it would be difficult to deny that the food-, housing-, land-, employment-, and social-security generally enjoyed by the peoples along the coast of West Africa contrasted sharply with the plight of the majority of the inhabitants of Western Europe



The complexity that really matters 

at this time, most of whose autochthonous power over land had been gradually lost since the time of the Roman Empire (Carney 2001). As Britain was establishing its colonial empire, even the few rights that the bulk of its rural population had managed to retain over land during the feudal period were being systematically extinguished through the Enclosures of Common and Estate lands, which caused a massive displacement of dispossessed peasants to the cities. British cities were growing far more quickly than the employment opportunities that they generated and when work was available in urban centers, the pay was usually not sufficient to meet basic survival needs. This resulted in a rapidly expanding number of marginalized town dwellers, forced to engage in a myriad of semi-legal and illegal activities, just to stay alive (Linebaugh 1992). What was happening to the working people of Africa and the Caribbean was inextricably bound to what was happening to their European counterparts. Sugar plantation slavery and the transatlantic trade that it gave rise to had made it more profitable for the British landowning class to produce sheep for the manufacture of cloth for the colonies than to continue to employ peasants to produce food. Many of the peasants who were uprooted in this process eventually ended up as the transported convicts, indentured laborers, privateers, etc. who became the ‘shock troops’ for the British transatlantic colonial enterprise, which in turn allowed the further expansion of plantation slavery in the British Caribbean. In many ways, this was a period of unprecedented experimentation in just how much labor and profit could be extracted from human beings while minimalizing the costs of their maintenance and reproduction (Mintz 1971: 483–485), with the techniques for labor extraction developed on the agro-industrial plantations of the Caribbean being almost immediately deployed in the burgeoning industrial manufacturing sector in Britain. In Spain, feudalism remained largely intact until the end of the 18th century, with the rural population subjected to many different types of forced labor by the landowning classes. While relatively few Northern Europeans emigrated to the Caribbean with the intention of becoming peasants, most of those who arrived in Hispaniola, Puerto Rico, and Cuba from Spain wanted land upon which they could establish an independent subsistence economy; but when they arrived, they were once again subject to forced labor (Batie 1976: 210; Stinchcombe 1995: 104, 108). The Spanish monarchy actually attempted to limit African slavery in its transatlantic territories during the first centuries of colonization, preferring to mobilize its own peasantry as well as the considerable numbers of people who had been displaced by the Reconquista and Inquisición to relieve shortages of exploitable labor in the colonies and using Amerindian slaves when these sources, like many of their counterparts in the Caribbean, proved inadequate (Mörner 1967: 15–16; Goveia 2000: 591). The main defining feature of the substrate economic continua as they are conceptualized here is the deep attachment of the African and Afro-Caribbean populations in the Antilles to a community-based subsistence economy and their desire to re-establish that economy in the Caribbean. The poles of these continua are defined by the

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

encounter of West African subsistence producers with agro-industrial capitalism and an uprooted European working class whose relationship to land and subsistence had been severely disrupted at the Northern pole, and the encounter of West African producers with a mass of European and mestizo peasants and indigenous peoples who shared their desire to establish or re-establish a subsistence economy in the Caribbean at the Southern pole. Successive waves of forced migration from Africa to the Caribbean would have continually reinforced and enriched the tradition of subsistence economics among Afro-Caribbean populations. The great majority of the enslaved African and Afro-Caribbean population of the British Antilles was incorporated into the gangs of field/ mill workers on large sugar plantations, where many were completely alienated from the land they worked, the labor they performed, and the fruits of their labors: the antithesis of subsistence (Stinchcombe 1995: 149; Moitt 2000: 1019–1029). But even in this situation, slaves ceaselessly struggled for the establishment of their own subsistence economies on the margins and in the interstices of the dominant agro-industrial sugar complex. Despite the resistance of planters, slaves won the right to cultivate their own subsistence plots, to consume and sell the produce of their plots, to organize their own weekend markets, and to conduct a number of other subsistence activities. Even though the spectacular success of sugar in the Antilles can be understood as having depended almost entirely on African and Afro-Caribbean science, knowledge, expertise, and labor (Moitt 2000: 1018–1023; Chaudenson 2001: 119, 227, 248; Havisser ms), there is substantial evidence that slaves engaged much more enthusiastically in marginalized subsistence activities than in the cultivation and processing of cane. Although subsistence plots were typically located on marginal land, could only be worked by slaves in their extremely limited free time, and were expected to provide slaves with most of their dietary needs across colonial regimes, the surplus from these plots often became the main source of food for the entire population of a given sugar island, including all of the town dwellers and the planters themselves (Tomich 2000; Carney 2001: 1). Where Afro-Caribbean populations were to some extent left by the British to their own devices (as in Anguilla and other marginal colonies), where Maroons established their own sovereignty (as in the highlands of Jamaica), and where land was redistributed after emancipation (as in Tortola), slaves or former slaves quickly established relatively viable community-based subsistence economies. Because capitalist agro-industrial plantation slavery constituted neither the basic foundation nor the main source of labor for the economies of the Spanish Caribbean until the 19th century, a great number of slaves there, if not the majority, could expect freedom within their lifetime by various means, including purchase, manumission, marriage, and maroonage (Pérez 1988: 64; Lipski 2005: 126–127). Manumission rates as well as the ratio of freed slaves to slaves were much higher in the Spanish Antilles than in the British Caribbean, with the numbers for the French territories falling somewhere in between these two extremes. African and Afro-Caribbean slaves came to constitute the



The complexity that really matters 

preponderant majority of the population of first the British and later the French Caribbean, but they never did so in the Spanish Antilles (Mintz 1971: 481–482). Slaves in first the British and later the French colonies mainly worked on sugar plantations; in the Spanish colonies, however, most worked as domestics or craftspeople. This trend, together with the fact that plantation size was largest under the British, slightly smaller under the French, and significantly smaller under the Spanish, meant that the great majority of African and Afro-Caribbean slaves would have had little or very limited contact with non-slaves from a very early period in the English territories, slightly more contact early on but dramatically less as time went on in the French Antilles, and significantly more contact in the period before the turn of the 19th century in the Spanish Caribbean (Goveia 2000: 571). The economics of subsistence in the extremely hostile environment of agro-industrial capitalism forced African and Afro-Caribbean peoples in the British Caribbean to forge strong bonds among themselves for mutual survival, yielding tightly-knit microeconomies and communities occupying exclusive niches in the margins of plantocratic society. In the Spanish Antilles, however, the African and Afro-Caribbean working classes found natural allies among their indigenous Caribbean, mestizo and European counterparts, populations which for a variety of reasons frequently shared their goal of the establishment of subsistence economies. This co-incidence of economic projects among all of the working classes in the Spanish Caribbean fostered the formation of colony-wide subsistence economies and subsistence communities, in which Africans and Afro-Caribbean peoples, with their extensive knowledge and mastery of tropical agriculture, animal husbandry, fisheries, architecture, metalworking, etc, would have played a significant role alongside indigenous Caribbean peoples (Carney 2001).

4.5

Substrate ideologies, cultures, and linguistics

West Africa is one of the most culturally and linguistically diverse regions on the planet, with well over one thousand distinct but highly interactive ethno-linguistic communities, each practicing multiculturalism in its own creative and unpretentious way. This situation graphically exemplifies Bakhtin’s (1934) notion of heteroglossia, a term which he uses to describe the linguistic landscapes that have typified the overwhelming majority of human cultures for the greater part of human history, in stark contrast to relatively recent consolidation of languages of domination in places like Western Europe. The imposition of ‘unitary’ standardized languages began in earnest during the initial period of European expansion into the Americas and continues to this day, but, as Bakhtin observes, the centrifugal forces of heteroglossia continually frustrate the centripetal tendencies of hegemonic language and culture. Cultural exchange, trade, and intermarriage between ethnic groups have always been the rule rather than the exception in West Africa (as they were in the indigenous Caribbean), so that each individual actor in society is expected to strike a dynamic balance between a strong sense of ethnic identity and a fluent command of many different

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

cultural, linguistic, and religious codes. It is this firm and secure sense of cultural identity often matched by both an acceptance, even a preference for inclusiveness, adaptability, and linguistic and cultural diversity, as well as an openness to heteroglossia, cultural exchange and hybridity which defines the cultural continua that the African and Afro-Caribbean working classes created in the Caribbean (for examples from Latin America see Urban 1991: 307). As is generally the case with subsistence producers worldwide (including the indigenous inhabitants of the Caribbean and speakers of Creoles in other regions of the world) the philosophies and ideologies that characterize West African societies are by and large centered around consciousness of one’s belonging to nature and society as holistic systems and on the establishment of harmonious relationships with and within these systems. Post-Platonic and especially post-Enlightenment European science relies on analytic reasoning with its emphasis on abstract formal logic, the conceptual dissection of reality into binary, unambiguous, independent categories and the belief that breaking down nature and society into component parts leads one to control them. In contrast, the sciences practiced by subsistence societies put a heavier emphasis on synthetic reasoning, experiential logic, the cultivation of ambiguity, relationships, and unity, and the idea that one’s personal power depends on the dynamic alignment of oneself with nature and society (Bennholdt-Thomsen & Mies 1999; Von Werlhof 2001; Havisser ms). The ideological continua that the African and Afro-Caribbean working classes brought to the Antilles are defined by a conscious cultivation and celebration of the heterodox and heteroglossic blurring and dissolution of boundaries, enclosures, divisions, and categorical distinctions of all kinds, from those used to separate and exclude human beings to those that are used to demarcate one linguistic category from another. Such practices had a number of dimensions that were rational and strategic, as well as others that were mediated by the importance and value that speakers attributed to the imaginary, the artful, and the symbolic. This multidimensionality serves as a reminder that language is intimately connected to consciousness, “the work of making meanings in social life” (Woolard 1985: 742). West Africans arrived in the Caribbean with a secure sense of identity that withstood all of the ravages of slavery and eventually thrived in its new environment. This sense of grounding and security in their own cultures and ideologies has historically given West Africans the confidence and openness to interact creatively and syncretically with cultures not their own, whether these other cultures are West African, indigenous Caribbean, or European, and whether contact with these other cultures occurs in their own village, in a regional market, in a slave trading fort along the West African coast, or on a plantation on the other side of the Atlantic Ocean. Successive waves of forced migration from Africa to the Caribbean continually reinforced and enriched these cultural traditions among Afro-Caribbean peoples. Because neither African nor indigenous Caribbean peoples had undergone the processes of cultural and linguistic domination which culminated in the Enlightenment-inspired imposition of standardized ‘unitary’ constructions of European lan-



The complexity that really matters 

guages (Bakhtin 1934), their view of the nexus between language and power was fundamentally different from that of their European counterparts (especially those from northern Europe, particularly after the middle of the 17th century). Unitary notions about language and culture are closely related to Enlightenment concepts of knowledge. As noted by Briggs (1993: 403–404), one of the fundamental tenets of modern science from the 17th century onwards has been that nature does not speak for itself – it can only be understood scientifically through the use of conceptual models devised by scientists. The powerful union of the rhetorics of authenticity, nationalism, nature, and preservation with the rhetoric of science was crucial here in that this hybrid rhetorical complex reserves, to scholars, textual authority over language, folklore, and the culture of Others.

In contrast, African and Afro-Caribbean peoples brought a decentered, heteroglossic sense of personal authority over language and of personal power through language into the Creole Space. Rather than viewing language and power as located ‘elsewhere’, that is, among the cultural and political elites (an attitude that would come to typify European and Euro-Caribbean peoples), Africans and Afro-Caribbean peoples retained their traditional sense of personal and community control over their languages and their lives, effectively undermining and in some cases toppling, dominant regimes of truth. As suggested by Woolard (1985: 741), the legitimacy of this alternative conceptualization of language and power is strengthened to the extent that a population that does not control a particular linguistic variety refuses to acknowledge and endorse its authority, its correctness, its power to convince, and its right to be obeyed. This type of heteroglossia has transformed the Creole Space and the Creole languages which emerged from it, not merely into languages/ acts of identity (Le Page & Tabouret-Keller 1985), but also into potent linguistic forces which have exerted and continue to exert considerable influence over the economic, political, and ideological parameter settings of the Creolization Matrix itself. Moreover, given that even an individual speaker exists as a site of differences, heteroglossia complicates the notion that a group or individual simply ‘has’ or ‘performs’ a specific identity. Linguistic identities are fluid formations constituted within webs of power relations (Laclau & Mouffe 1985; Moore 1994). Ironically, when it is mentioned at all, many scholarly accounts of language genesis have tended to describe this tension and the alternative ways of speaking and belonging that it embodies as the result of nation-state formation or as a future problem which the nation-state should solve (e.g. through language standardization, ‘development,’ or even ‘empowerment’). The reluctance to recognize cultural phenomena and social groupings that crosscut, undermine, hybridize, or exist outside dominant categories obscures the role that scholarship plays in creating, sustaining, and reproducing inequality based on race, language, gender, social class, and nationality. The poles of the substrate ideological and cultural continua created by the African and Afro-Caribbean working classes are defined by the forging of a separate Afro-Caribbean identity and separate Afro-Caribbean communities in the face of racism and

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

apartheid at the northern end, and an aggressive and creative engagement with colony-wide society in order to transform it, not just to accommodate but also to embrace African and Afro-Caribbean culture, at the Southern end. Attention to these differences assists in reformulating theories of genesis so that they build not just on a few select points of articulation between select locales in Africa, the Caribbean, and/ or Europe, but also nuance ahistorical, homogenizing accounts which disregard the concrete ways in which language transformed and was transformed by political-economic, ideological, and social relationships. The first Afro-Iberians and Africans to arrive in the Caribbean at the beginning of the 16th century were male, and they created intimate relationships and families with indigenous Caribbean women just as enthusiastically (if not more so) as their EuroIberian counterparts at the time. While women of African and indigenous Caribbean descent were systematically exploited and subjected to rape, concubinage, and other forms of violence, this fundamental and unabashed disregard by African and AfroCaribbean peoples for distinctions between race and casta in the Spanish Caribbean has in many instances continued unabated up until the present day, despite considerable efforts made by the ruling classes to combat it (Mörner 1967: 40). African and Afro-Caribbean peoples together with their subsistence-oriented allies in the indigenous Caribbean, mestizo, and Euro-Caribbean working classes were so successful in this cultivation and celebration of difference in fraternity, love, and familial unity that ever since the earliest years of colonization, most of the Dominicans, and many of the Cubans and Puerto Ricans who themselves have no obvious phenotypical manifestations of West African ancestry can readily identify at least one member of their proximal extended family who does (Martínez Cruzado 2002). The pervasive infusion of West African blood into the Spanish Caribbean is matched by an equally all-encompassing penetration of West African culture and values into nearly every sphere of Spanish Caribbean life, including but not restricted to religion, language, childrearing practices, interpersonal dynamics, architecture, literature, music, the visual arts, dance, cuisine, and attitudes toward work, play, and time. The universalizing and ‘civilizing’ mission of Spanish colonialism constituted the major constraint on the propagation of West African culture in the Spanish Antilles. This generally meant that, at least in the public sphere, West African cultural values, meanings, and expression had to be channeled through cultural forms or structures that could be accepted, appreciated, understood, and practiced by the entire Spanish world, such as Catholicism (Catholic ritual intertwined with Afro-Caribbean beliefs), the Spanish language (the Caribbean ‘dialects’ of Spanish), and Mediterranean cuisine (the simmering of pounded root crops or plantains in olive oil and garlic to make mofongo and mangú). This process was facilitated by considerable previous North and West African influence over Mediterranean culture in general and Iberian culture in particular under the Sahelian Moors and the close cohabitation between indigenous Caribbean, mestizo, Euro-Caribbean, and Afro-Caribbean populations in the Spanish Antilles. Indeed, the insistence on outward conformity to Spanish culture and ideology



The complexity that really matters 

probably encouraged the extremely widespread and incredibly inventive phenomenon of syncretistic subversion of European symbols by vigorous and resilient West African systems of discovering, understanding, and creating meaning in the world (Mazrui & Mazrui 1998; Mufwene 2001: 182–184). In the British Caribbean, the situation was radically different. The cultural and ideological propensities of the African and Afro-Caribbean population toward unity, exchange, syncretism, inclusion, and the blurring of conceptual and social boundaries were systematically frustrated by the lack of meaningful and sustained contact between the Euro-Caribbean and the Afro-Caribbean populations, the absence of previous African influence on Northern European culture, the segregationist and exclusivist doctrine of Calvinistic Protestantism, the detailed definition and strict enforcement of racial divisions, and the alienated and dehumanized nature of capitalist labor extraction. The result was the diversion of these unifying tendencies from the colonywide society to Afro-Caribbean communities themselves. Instead of the significantly more widespread syncretic fusion of West African cultures and values with European cultures and values that occurred in the Spanish Caribbean, a distinctive and self-conscious Afro-Caribbean cultural complex was forged from the West African ethnic mix found in the British territories. In this way, two very different cultures emerged in the British colonies along the fault lines of slavery and race: firstly a Black culture that developed its own church, language, music, cuisine, and other expressive traditions from the synthetic fusion of a number of distinct West African traditions (all of which shared some areal and typological similarities) and secondly, a White culture that identified principally with the metropole (Mintz 1971: 487). Black culture emerged defensively as a positive valorization and affirmation of Afro-Caribbean identity within a system that sought consistently to negate it and offensively as a means of resistance to that system. Given the similarities between Spanish and French imperial enterprises in the earliest period of French colonialism, a certain degree of colony-wide syncretistic fusion between West African and French values and cultures must have occurred, but this trend was quickly and dramatically reversed as the French moved swiftly and decisively toward the agro-industrial production of sugar that formed the basis of the British colonial project in the Caribbean.

4.6

Substrate politics

When considering substrate politics, it should always be kept in mind that even though West Africa was the first part of the globe outside of Europe and the Mediterranean to be reached by the Portuguese navigators and merchants of the 15th century, its peoples successfully resisted colonization by the Europeans until the latter half of the 19th century. West Africans came into the transatlantic trading system with a considerable body of knowledge and with experiences of resistance to domination by the European colonial powers (Oliver & Atmore 2001).

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

As subsistence producers, West Africans’ social and cultural traditions supported ‘bottom-up’ community-based politics that deal with the identification, analysis, and resolution of community problems in a practical way, over the polemical, abstract politics that typify many European societies. Stemming from an equally complex West African epistemology, a ‘grassroots’ politics that emphasizes transforming ‘the facts on the ground’ characterizes the substrate political continua created by the African and Afro-Caribbean working classes in the Caribbean. Successive waves of forced migration from Africa to the Caribbean continually reinforced and enriched these political tendencies among Afro-Caribbean peoples. At the Southern pole of these continua we find African and Afro-Caribbean populations busy transforming and re-creating all aspects of colonial society in their own image. At the Northern pole, we find African and Afro-Caribbean peoples forging a common Afro-Caribbean identity politics with which to defend themselves in a more hostile system, assembling a diversified arsenal of creative and resourceful ways to offensively subvert and reshape that system, and creating their own autonomous micro-political spaces on the margins and in the interstices of that system. It would be very difficult to find a more graphic and effective example of the concrete nature of a West African politics of ‘changing the facts on the ground’ than the rapid and complete subversion of all the founding racial and casta categories in the Spanish Caribbean through the demographic transformation effected by the defiant and energetic establishment of interracial sexual relationships and families by African and Afro-Caribbean peoples. Within a few generations, Spanish Antillean society became to a significant degree genetically and culturally West African, rendering all efforts to impose a British-type racialist regime totally ineffective, even after the implementation of the Northern-style Bourbon reforms designed to promote the growth of agro-industrial capitalism there during the 19th century and the importation of EuroIberians in a concerted effort to ‘whiten’ the Greater Antilles. Even peninsulares and lighter-skinned individuals who may have made use of a caste system to preserve their interests from intrusion by non-whites resented the ways in which it grew to systematically limit their aspirations and freedom (Chinea 2002: 172). The alliance for subsistence created between the Afro-Caribbean, indigenous Caribbean, mestizo, and EuroCaribbean working classes was thus literally ‘incarnated’ in an eminently successful and irreversible way (Martínez Cruzado 2002). With this set of acts alone, the African and Afro-Caribbean peoples of the Spanish Caribbean moved beyond any politics of defensive or offensive opposition to the dominant system by dismantling the binary categorical opposition between ‘Black’ and ‘White’ upon which that entire system rested. While racism and racial hierarchy have not by any means been eradicated from the Spanish Antilles, these conditions put them permanently and decisively on the defensive. This type of transformation and eventual progress toward transcending an oppressive system proved to be a much more complex and time-consuming endeavor in the British and French Caribbean. The African and Afro-Caribbean populations of the British territories had to adopt a politics of opposition within the binarisms that un-



The complexity that really matters 

derpinned the dominant system by both articulating a defensive identity politics and an offensive politics of resistance while simultaneously putting into practice a politics aimed at supplanting that system entirely. Identity politics ascribed a positive value to all aspects of Black culture, converting cultural forms such as Black language, Black music, Black religion, into signifiers of a distinct Black identity and intra-group solidarity. Resistance politics attached additional meaning to Black culture in opposition to White culture. In this way, the use of Black language became a vehicle for expressing one’s rejection of White language and culture and at times an exclusive and secretive means for the African and Afro-Caribbean populations of the British territories to communicate with one another without being understood or spied upon by the White community. Because French colonialism, especially in its earliest stages, resembled the Spanish system more closely than that of the British in some fundamental ways, the tendency toward the development of identity politics was less pronounced in the French Antilles. For example, because all slaves in the French territories were expected to become practicing Catholics, a distinctive Black church never emerged there before the 19th century. The African and Afro-Caribbean inhabitants of the British and French colonies resisted slavery in every manner imaginable, from deliberately slowing the pace of their day-to-day work all the way to organizing a sustained series of full-scale insurrections (Chaudenson 2001: 245). Whenever possible, however, a transformational subsistence politics of ‘changing the facts on the ground’ was practiced. Because of the comprehensive and all-sided nature of the repressive apparatus, first under the British and later under the French as well, this type of politics was only possible at the extreme margins of the system. But no margin or interstice was left unexploited by the slaves as a space from which a successful community-based ‘bottom-up’ politics of subsistence could be articulated and put into practice. Provision plots and weekend markets became powerful venues for the redirection of African and Afro-Caribbean labor and ingenuity from the production of sugar to subsistence. These spaces were so successfully cultivated by slaves and the abundance that they produced was so dramatically different from the scarcity created by agro-industrial monocropping, that they became crucial for the survival of both the Black and the White communities and therefore could not be suppressed. They often thrived despite the fact that the weekend markets became lively venues for the expression and celebration of Black cultures and politics of identity, resistance, and transformation. Another case in point is female slaves’ refusal to bear children despite the severe punishments meted out to those who did so (Stinchcombe 1995: 34; Moitt 2000: 1020– 1021). A significant number of African and Afro-Caribbean peoples were able to extricate themselves completely from the dominant system by running away into remote areas or to safe havens in the Spanish Antilles such as Puerto Rico’s San Mateo de Cangrejos (Chinea 2002: 175). In some cases, these slaves managed to create their own sovereign subsistence Maroon communities. Whatever their particular circumstances, however, the consistent political demands of African and Afro-Caribbean peoples re-

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

mained essentially the same; full emancipation and land, i.e. the two necessary and sufficient conditions for the creation of a community-based subsistence economy (Williams 1944: 202–205).

5 Conclusion: The linguistic outcomes As suggested above, we propose that there is a strong correlation between political economy (i.e. the political, economic, and ideological framework within which power relations were manifested in each colony) and the extent and nature of influences from African substrate languages on the speech varieties that developed under colonialism in each Caribbean society. The theory of language/ cultural contact presented here predicts that the English-lexifier Creoles, the French-lexifier Creoles, and the Caribbean dialects of Spanish differ in the way in which substrate influences play themselves out in each case, in terms of the volume of substrate influence, the scope of that influence, and in terms of the historical dynamics of each speech variety. Furthermore, this framework can account for differences within any of the three groups. These differences, like exceptions that can be made to the more macroscopic generalizations made above, cannot be understood independently of the cultural reworkings brought on by the experience of dislocation; the constant reworking of speakers’ subjectivities and language ideologies in response to considerations of geography, class, gender, ethnicity, and race; and the interplay between control and regulation on the one hand, and resistance and resilience on the other. When all the generally accepted and convincingly substantiated African substrate influences are considered (see Alleyne 1980; Faraclas 1990; Parkvall 2000), the full range will appear in English-lexifier Creoles, whereas in the French-lexifier Creoles some of these features will be totally absent and others very marginal (highly marked). English-lexifier Saramaccan could be said to provide the base or measuring stick from which all the other Creole languages may be evaluated in terms of the number of African substrate features they contain. Phonologically, Saramaccan has co-articulated stops, pre-nasalised stops, nasal high vowels, all-pervasive vowel-final syllables, vowel harmony, and distinctive lexical tone. Morphologically, it has vestiges of a Niger-Congo noun class system and some functional morphemes taken from African languages. In syntax, Saramaccan has the full array of serial verb structures, reflexivization using a word meaning ‘body’ or ‘skin’, predicate clefting, postpositions, and reduplication to derive adjectives from verbs. In the area of lexico-semantics, Saramaccan has the largest number of African language cognates in its basic vocabulary as well as a significant number of calques. Whereas African substrate features are clearly established in French-lexifier Creoles, no one of these languages will show a range comparable with Saramaccan, and, taken as a whole, these languages show fewer substrate features than the English-lexifier Creoles (Parkvall 2000: 155). For example, the only clear-cut phonological feature



The complexity that really matters 

attributable to the African substratum is the existence of nasal high vowels in Haitian. This feature occurs mainly in words of African language origin, but appears also in a few French-derived words. Even where a syntactic feature of African language origin occurs in French-lexifier Creoles, it may not exhibit the same range as in the Englishlexifier Creoles. For example, French-lexifier Creoles have the dative/ benefactive, directional, purposive, and comparative serials; but not the instrumental and complementizer (‘say’) serials (except as highly marked structures of dubious authenticity). In addition, the lexical class Adjective is more common than in the English-lexifier Creoles. For example, whereas sik(i) ‘sick’ is generally used as a verb with the progressive aspect marker in English-lexifier Creoles, malad + Progressive Aspect is not grammatical in some French-lexifier Creoles and in those where it does occur (St. Lucian for example), some speakers reject it as ungrammatical (Alleyne 1996: 175–185). The oldest layer of English-lexifier Creoles, best represented by Saramaccan, but also reconstructable for Sranan, Jamaican and others, has the largest volume of African substratum features, reflecting not only the fact that Saramaccan was cut off from contact with English at an early date but also the fact that the socioeconomic conditions of slavery from its inception favored such relatively heavy African influence. By contrast, the oldest layer of French-lexifier Creole varieties shows a form closer to French than later forms. This reflects the comparably favorable conditions for superstrate influence of the société d’habitation (Southern pole of the Creolization continua) typical of the initial period of French colonization. Later, when plantation slavery became firmly established and the number of enslaved Africans grew in the French colonies (moving them toward the Northern pole of the Creolization continua), African substratum features would have begun to proliferate, moving the languages further away from French and closer to West African languages. For example, Haitian shows signs of the initial development instrumental and complementizer serials, but these have yet to become firmly established in that language. Some of the grammatical features that have been attested to varying degrees in the Caribbean dialects of Spanish which can be attributed to influence from African languages include the following (Otheguy 1974; Lipski 2005): 1. Use of nominal and adjectival words without inflectional suffixes 2. Use of nominals without articles 3. Postposing of demonstratives 4. Use of third person pronouns which are invariant for gender 5. Frequent use of pronominal subjects 6. Use of verb roots without inflectional suffixes 7. Use of preverbal aspect and modality markers such as ta, ya, va 8. Fusion of copular forms ser and estar into composite form (e)sar 9. Use of equative, locational, and other types of clauses without copulas 10. Use of the generic preposition na/ ne 11. Certain case relationships are indicated without the use of prepositions 12. Use of subordinate clauses without complementizers

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz

13. Double negation 14. Questions without inversion 15. The use of the adverb más used before negatives While some of these features are restricted to Afro-Caribbean dialects, others are used by populations of all ethnic origins in the Spanish Caribbean. Differences among Iberian Creoles may stem, in part, from shifts in the West African populations that contributed to their formation (Ferraz 1979; Maurer 1987; Mufwene 1994). A number of scholars have suggested that Caribbean Spanish has undergone processes of pidginization, creolization, or semi-creolization (Granda 1968, 1971; Otheguy 1974). Whatever names are given to the processes involved, the linguistic outcomes of these processes seem at least to some extent to be different in the Spanish Caribbean from those attested in the British Caribbean, with the French Caribbean falling somewhere between these two poles. Although it would be very difficult to try to derive the more substrate-influenced varieties of English-lexifier Caribbean Creoles from English, it seems to be slightly less difficult to derive the French-lexifier Caribbean Creoles from French and much, much easier to derive the African-influenced speech varieties that are spoken in most parts of the Caribbean from Spanish. In this work, we have attempted not only to situate Creole languages within their economic, political, and ideological context, but we have also tried to articulate a view of creolization as a polycentric system that can include varieties of Caribbean Spanish. We recognize and point out that the main ideas presented above build on discursive tropes, each with a genealogy and a history of having been cast and recast in cultural and political terms (at times by speakers of the languages mentioned, but more often than not by linguists) that frustrate efforts to describe and analyze Creole languages on their own terms. Examples of such tropes include the highly problematic notions of ‘absence’ and ‘deletion’ (as found in the list above) which are commonly used to describe ‘deviant’ or ‘prototypical’ Creole language features in terms of deficiencies in relation to European-lexifier languages that result from some process of simplification. As Mufwene explains, creolists’ reliance on such concepts contributes to a ‘pernicious’ tradition of downplaying the role of contact in the study of ‘native varieties’ and invoking it in the case of Creole and ‘indigenized’ varieties, leading to notions of ‘legitimate’ and ‘illegitimate’ offspring of European varieties (2001: 107). The former, among which he would probably include arguments that Caribbean Spanish ‘is just a dialect’, are attributed to ‘normal evolution’. The latter, as discussed above, are defined as exceptional. Rather than being ‘innocent’ concepts that provide the currency for ‘neutral’ and/ or ‘scientific’ discourse, these notions are implicated in acts of symbolic violence that give form and substance to the very identities, cultural practices, and languages that linguists seek to describe (Bourdieu 1991). Keeping these reservations in mind, we hold that this configuration of ideas offers an open-ended yet empirical way of describing similarities and differences among the linguistic varieties that flourished in the Caribbean as result of the Atlantic Slave Trade,



The complexity that really matters 

helping to explain why languages that emerged from British sugar slavery in the Caribbean are generally considered by linguists to be autonomous varieties, and not dialects of English at all. Those speech varieties that developed under French Caribbean sugar slavery, perhaps because they are in a sense ‘caught in the middle’, have become the subject of intense debate among linguists as to whether they should be classified as dialects of French or not. Finally, there are the African-influenced speech forms that emerged from Spanish colonialism in the Caribbean which are normally classified by linguists as dialects of Spanish (except for Palenquero, some of the language varieties associated with sugar slavery in the Spanish Antilles in the 19th century, and Papiamentu, which could be considered to be a Portuguese-lexifier Creole that has been significantly relexified in the direction of Spanish). As we have suggested, for political and economic reasons Caribbean varieties of Spanish do not usually enter into dominant narratives of linguistic creolization per se, at least not in the anti-exceptionalist terms that demand a more extensive interrogation of context and reveal parallels and contrasts that both link them to and distinguish them from their English- and Frenchlexifier counterparts.

References Aceto, M. 2003. What are Creole languages? An alternative approach to the Anglophone Atlantic World with special emphasis on Barbudan Creole English. In Aceto, M. & J.P. Williams (eds). Contact Englishes of the Eastern Caribbean. Amsterdam/ Philadelphia: John Benjamins. 121–140. Alleyne, M.C. 1971. The cultural matrix of creolization. In Hymes, D. (ed.). Pidginization and Creolization of Languages. New York: Cambridge University Press. 169–186. Alleyne, M.C. 1980. Comparative Afro-American. Ann Arbor: Karoma Press. Alleyne, M.C. 1996. Syntaxe Historique Créole. Paris: Karthala. Allsopp, R. 1976. The case for Afrogenesis. Paper presented at the Society for Caribbean Linguistics conference, University of Guyana. 11–14 August 1976. Arends, J. (ed.). 1995. The Early Stages of Creolization. Amsterdam/ Philadelphia: John Benjamins. Bakhtin, M. 1934. Unitary language. In Emerson, C. & M. Holquist (eds). 1981. The Dialogic Imagination. Austin: University of Texas Press. 269–295. Batie, R. 1976. Why sugar? Economic cycles and the changing of staples in the English and French Antilles 1624–1654. In Shepherd V. & H. Beckles (eds). Caribbean Slavery and the Atlantic World. Kingston: Ian Randle. 207–225. Bennholdt-Thomsen, V. & M. Mies. 1999. The Subsistence Perspective: Beyond the Globalized Economy. London: Zed Books. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma Press. Bickerton, D. 2004. Reconsidering Creole exceptionalism. Language 80 (4). 828–833. Bourdieu, P. 1991. Language and Symbolic Power. Cambridge: Harvard University Press. Briggs, C. 1993. Metadiscursive practices and scholarly authority in folkloristics. Journal of American Folklore 106 (422). 387–434.

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz Butel, P. 2000. France, the Antilles, and European the seventeenth and eighteenth centuries: Renewals of foreign trade. In Shepherd, V. & H. Beckles (eds). Caribbean Slavery and the Atlantic World. Kingston: Ian Randle. 194–206. Cameron, D. 1997. Demythologizing sociolinguistics. In Coupland, N. & T.A. Jaworski (eds). Sociolinguistics. London: Macmillan Press. 55–67. Carney, J.A. 2001. Black Rice, The African Origins of Rice Cultivation in the Americas. Cambridge: Cambridge University Press. Chaudenson, R. 2001. Creolization of Language and Culture. London: Routledge. Chinea, J.L. 2002. Fissures in El Primer Piso: Racial politics in Spanish colonial Puerto Rico during its pre-plantation era, c. 1700–1800. Caribbean Studies 30. 1168–1204. Craton, M. 1997. Empire, Enslavement, and Freedom in the Caribbean. Kingston: Ian Randle. DeGraff, M. 2001. On the origins of Creoles: A Cartesian critique of Neo-Darwinian linguistics. Linguistic Typology 5 (2). 213–310. DeGraff, M. 2003. Against Creole exceptionalism. Language 79 (2). 391–410. DeGraff, M. 2004. Against Creole exceptionalism (redux). Language 80 (4). 834–839. DeGraff, M. 2005. Linguists’ most dangerous myth: The fallacy of Creole Exceptionalism. Language in Society 34. 533–591. DeGraff, M. & D. Walicek. 2005. Creole exceptionalism and accidents of history: A conversation with Michel DeGraff. In Walicek, D. (ed.). Sargasso: Creolistics and Caribbean Languages 2004–2005, I. 1–34. Faraclas, N. 1990. From Old Guinea to Papua New Guinea: A comparative study of Nigerian Pidgin and Tok Pisin. In Verhaar, J. (ed.). Melanesian Pidgin and Tok Pisin. [Studies in Language 20.] Amsterdam/ Philadelphia: John Benjamins. 91–169. Faraclas, N. 2005. Globalization and the future of Creole languages. Language and Politics 4 (2). 331–365. Ferraz, L.I. 1979. The Creole of São Tomé. Johannesburg. Witwatersrand University Press. Foucault, M. 1978. The History of Sexuality, Volume I: An Introduction. New York: Vintage. Foucault, M. 1980. Power/ Knowledge: Selected Interviews and Other Writings, 1972–1977. Gordon, C. (ed. & trans.). New York: Pantheon. Foucault, M. 1991. Governmentality. In Burchell, G., C. Gordon & P. Miller (eds). The Foucault Effect: Studies in Governmentality. Chicago: University of Chicago Press. 87–104. Goveia, E. 2000. The West Indian slave laws of the eighteenth century. In Shepherd, V. & H. Beckles (eds). Caribbean Slavery and the Atlantic World. Kingston: Ian Randle. 580–596. Granda, G. de. 1968. La tipología criolla de dos hablas del área lingüística hispánica. Boletín del Instituto Caro y Cuervo 23. 193–205. Granda, G. de. 1971. Algunos datos sobre la pervivencia de criollo en Cuba. Boletín de la Real Academia Española 51. 481–491. Harris, R. 1981. The Language Myth. New York: St. Martin’s Press. Havisser, J. ms. Cognitive issues related to interpreting the African Caribbean. Higman, B.W. 1995. Slave Populations of the British Caribbean, 1807–1834. Kingston: University of the West Indies. Laclau, E. & C. Mouffe. 1985. Hegemony and Socialist Strategy: Towards a Radical Democratic Politics. London: Verso. Lefebvre, C. 2000. What do Creole Studies have to offer mainstream linguistics? Journal of Pidgin and Creole Languages 15 (1). 127–153.



The complexity that really matters 

Le Page, R. & A. Tabouret-Keller. 1985. Acts of Identity. Creole-based Approaches to Language and Ethnicity. New York: Cambridge University Press. Linebaugh, P. 1992. The London Hanged. New York: Cambridge University Press. Linebaugh, P. & M. Rediker. 2000. The Many Headed Hydra: Sailors, Slaves, Commoners, and the Hidden History of the Revolutionary Atlantic. Boston: Beacon Press. Lipski, J. 2005. A History of Afro-Hispanic Language; Five Centuries and Five Continents. Cambridge: Cambridge University Press. Lipski, J. 2006. Afro-Bolivian Spanish: The survival of a true Creole prototype. Paper delivered to the Meeting of the Society for Pidgin and Creole Linguistics, Albuquerque, USA. January 2006. Martínez Cruzado, J.C. 2002. The use of mitochondrial DNA to discover pre-Columbian migrations to the Caribbean: Results for Puerto Rico and expectations for the Dominican Republic. KACIKE: The Journal of Caribbean Amerindian History and Anthropology [On-line Journal], Special Issue, Lynne Guitar, ed. Available at: http://www.kacike.org/MartinezEnglish.pdf. Accessed 20 December 2005. Maurer, P. 1987. La Comparison des morphèmes temporels du papiamento et du palenquero: Arguments contre la théorie monogénétique de la genèse des langues créoles. In Maurer, P. & T. Stolz (eds). Varia Creolica. Bochum: Brockmeyer. 27–70. Mazrui, A. & A. Mazrui. 1998. The Power of Babel: Language in the African Experience. Chicago: University of Chicago Press. McWhorter, J.H. 2000. The Missing Spanish Creoles. Recovering the Birth of Contact Languages. Berkeley: University of California Press. Mintz, S.W. 1961. Review of S.M. Elkins, 1959, Slavery, University of Chicago Press. American Anthropologist 63 (3). 579–587. Mintz, S.W. 1971. The socio-historical background to pidginization and creolization. In Hymes, D. (ed.). Pidginization and Creolization of Languages. New York: Cambridge University Press. 481–496. Mörner, M. 1967. Race Mixture in the History of Latin America. Boston: Little, Brown, and Co. Moitt, B. 2000. Women, work, and resistance in the French Caribbean during slavery, 1700– 1848. In Shepherd, V. & H. Beckles (eds). Caribbean Slavery and the Atlantic World. Kingston: Ian Randle. 1017–1029. Moore, H. 1994. A Passion for Difference. Bloomington: Indiana University Press. Mufwene, S. 1994. Restructuring, feature selection, and markedness: From Kimanyanga to Kituba. In Moore, K., D. Peterson & C. Wentum (eds). Historical Issues in African Linguistics. Berkeley Linguistics Society. 67–90. Mufwene, S.S. 2001. The Ecology of Language Evolution. New York: Cambridge University Press. Muysken, P. 1988. Are creoles a special type of language? In Newmeyer. F.J. (ed.). Linguistic Theory: Extensions and Implications. Cambridge: Cambridge University Press. 285–301. Mühlhäusler, P. 1993. What is the use of studying pidgin and creole languages? Language Sciences 14. 309–317. Nzewunwa, N. 1980. The Niger Delta: Aspects of its prehistoric economy and culture. Cambridge Monographs in African Archaeology 1. [Bar International Series 75.] Oxford: B.A.R. Oliver, R. & A. Atmore. 2001. Medieval Africa, 1250–1800. New York: Cambridge University Press.

 Nicholas Faraclas, Don E. Walicek, Mervyn Alleyne, Wilfredo Geigel & Luis Ortiz Otheguy, R. 1974. The Spanish Caribbean: A creole perspective. In Bailey, C. & R. Shuy (eds). New Ways of Analyzing Variation in English. Washington, DC: Georgetown University Press. 323–399. Parkvall, M. 2000. Out of Africa. London: Battlebridge. Pérez, L.A. 1988. Cuba, Between Reform and Revolution. New York: Oxford University Press. Roberts, P.A. 2004. Language, Race and Ecology: The Shaping of Identity in the Caribbean. St. Thomas: Streborp Press. Romaine, S. 1984. The status of sociological methods and categories in explaining linguistic variation. Linguistische Berichte 90. 25–38. Rybczynsky, W. 1986. Home: A Short History of an Idea. New York: Viking. Stinchcombe, A.L. 1995. Sugar Island Slavery in the Age of Enlightenment, the Political Economy of the Caribbean World. Princeton: Princeton University Press. Tomich, D. 2000. The other face of slave labor: Provision grounds and internal marketing in Martinique. In Shepherd, V. & H. Beckles (eds). Caribbean Slavery and the Atlantic World. Kingston: Ian Randle. 194–206. Urban, G. 1991. The semiotics of State-Indian linguistic relationships: Peru, Paraguay, and Brazil. In Urban, G. & J. Sherzer (eds). Nation-States and Indians in Latin America. Austin: University of Texas Press. 307–330. Von Werlhof, C. 2001. Losing faith in progress: Capitalist patriarchy as an ‘alchemical system’. In Bennholdt-Thomsen, V., N. Farclas & C. von Werlhof (eds). There is an Alternative: Subsistence and Worldwide Resistance to Corporate Globalization. London: Zed. Williams, E. 1944. Capitalism and Slavery. Chapel Hill: University of North Carolina Press. Woolard, K.A. 1985. Language variation and cultural hegemony: Toward an integration of sociolinguistics and social theory. American Ethnologist 12 (4). 738–748.

Creole metaphors in cultural analysis* Roxy Harris & Ben Rampton King’s College London

1

Introduction

In their attempt to model cultural processes within globalization and transnational flow, anthropologists have wondered whether linguistic research on creoles and creolization might provide a helpful set of metaphors. In a series of influential discussions (1987, 1989, 1996), Hannerz suggests that working as ‘a root metaphor’, linguistic creolization might help to promote a research agenda which involved: a. a decentering of anthropological description, freeing it from totalizing assumptions about integrated cultural systems and homogenous communities; b. a distributive view of culture as systems of meaning, which also reckons with the reflexivity of human agents, and with the ways “in which people develop a certain awareness of, and familiarity with, cultural forms which are not primarily theirs, at least not at the given moment; even if these forms are out of reach … or not actually well understood” (1987: 549); c. an engagement with popular culture and with the ability of media technology “to create new social relations and contexts” (1987: 555); d. a recognition of the ways in which “the merger of quite different streams can create a particular intensity in cultural processes” (ibid.); e. an analysis that addresses global inequalities but that allows “small facts [to] get in the way of large issues, [that isn’t] too sure that the dominant is totally dominant, [and that is] concerned with what the peripheries do both for themselves and to the centre” (ibid.: 556).

* An earlier version of this paper was published in Critique of Anthropology 22 (1): 31-52, and we would like to thank Steve Vertovec and Charles Stewart, organizers of the seminar series where our first draft was presented (‘Considering Creolisation’, Department of Anthropology, University of Oxford, Autumn Term 1999 [part of the ESRC Research Programme on Transnational Communities]). We are also indebted to two anonymous CoA reviewers for their helpful comments. 1. See Harris (2006) for an attempt to work through some of these kinds of conceptualization with empirical support in contemporary conditions in the UK.

 Roxy Harris & Ben Rampton

This chapter very much agrees with Hannerz on the significance of these issues, and to the extent that ‘creolization’ has in fact already served as a useful heuristic in the development of this kind of ‘macro-anthropology’, we welcome it as a valuable example of cross-disciplinary borrowing and adaptation. At the same time, though, we are concerned that the underlying view of creole language study might be just a little too kind to the linguistic traditions from which it emerged. In a number of key respects, creole linguistics belongs to the very same ‘episteme’ that Hannerz seeks to transcend, and it displays several major features that are potentially antipathetic to the interests that he expresses. In what follows, we will (i) offer a characterization of some of the main debates that have shaped creole language study, at least in the Atlantic and Caribbean, arguing that it has been heavily influenced both by ideologies of ‘race’ and by emphatically modernist preoccupations with total explanation, system and autonomy. After that, we will (ii) try to situate creole linguistics within linguistics more generally, pointing to a major paradigm shift currently in progress, and then (iii) illustrate the recent development of alternative modes of sociolinguistic analysis which, we think, are likely to provide much sharper insight into the issues that Hannerz is concerned with. It is worth beginning with an outline of some of the central ideological strands in creole language study.

2 Ideologies in creole linguistics There is general agreement that pidgins and creoles developed as a result of interaction in different parts of the world between mainly European traders and colonizers on the one hand, and mainly colonized, mostly black or brown people on the other. One of the most geographically extensive and also most historically abrupt examples of this was the forced collision associated with the Atlantic slave trade and its aftermath, and debates within creole linguistics reflect the continuing resonance of this. Right up until the mid-20th century, debates were dominated by a crude view advanced by white European and North American linguists that Atlantic pidgins and creoles were defective corruptions of ‘higher’ European languages and that they were not really worthy of serious study (cf. Holm 1988: 13–70). According to Holm, “[t]he ‘thick lips and thick minds’ theory of the origins of black speech varieties has a long history, even in the works of creolists who were enlightened for their times” (1988: 23). Creole speakers were considered uncivilized and incompetent learners of standard European languages, and the origins of pidgins and creoles were, for example, sometimes explained in terms of ‘baby-talk’, a theory that Holm (1988: 33) attributes to Schuchardt at the start of the 20th century but that can also be found in Bloomfield:

2. Holm credits Dillard (1972: 11) with this characterization of creole theory.



Creole metaphors in cultural analysis 

Speakers of a lower language may make so little progress in learning the dominant speech, that the masters, in communicating with them resort to ‘baby-talk’. This ‘baby talk’ is the masters’ imitation of the subjects’ incorrect speech. There is reason to believe that it is by no means an exact imitation, and that some of its features are based not upon the subjects’ mistakes but upon grammatical relations that exist within the upper language itself. The subjects, in turn, deprived of the correct model, can do no better now than to acquire the simplified ‘baby-talk’ version of the upper language. (1935: 472)

This interest in the origins and development of creole languages points to a recurrent sense of their ‘Otherness’, of their typological distinctiveness and of the ways in which they differ from ‘normal’ European languages. But it also carries a major methodological problem. Analysis of the early history of creoles is considerably impeded by the limited availability of written historical records on the original creole speaking populations. The subordinated position of these populations generally compelled them to develop and rely on largely oral and covert traditions (Scott 1990), and as McWhorter (1999a) noted in oral debate. [even b]efore I was born, we [creole linguists] noticed that these Creoles tended to be spoken in places where there used to be plantations, and we decided well, it must be the plantation demographics that created these languages, but we must remember that it wasn’t filmed, it wasn’t written down, nobody was there, that’s just a hypothesis, we don’t know. (1999a: 315)

McWhorter points to the necessarily very speculative character of research on the origins and development of creole languages, and in fact, one of the most common phrases appearing in the work of creole linguists as they attempt their reconstructions of the past is ‘must have’. For example: In any case, it seems clear that an English-based contact language with a fairly stable structure must have been spoken in St. Kitts before slaves made up a large majority of the total population, which is something that conflicts with traditional creologenetic theory. (Parkvall 1998: 71)

or [b]ut, as the population [of slave plantation societies] soared, these early immigrants too were reduced to a small percentage of the population. Increasingly, throughout the growth period, newcomers must have acquired whatever knowledge of the superstrate they could gather, not from native speakers, nor even from those who had learned from native speakers, but from those who had learned from non-native speakers, speakers who in turn may have had very limited exposure to non-native varieties of the superstrate. In other words, the growth phase must normally have been characterised by a progressive dilution of the superstrate. (Bickerton 1989: 18)

 Roxy Harris & Ben Rampton

Or again [t]he existence of a Pidgin English at Kormantin, expanded beyond the jargon stage, has been shown to be highly unlikely. While a restructured variety must certainly have existed, this very probably was used for communication with whites only and therefore there would have been no motivation for using it in [McWhorter’s] plantation diglossia. (Huber 1999: 103)

This relative lack of empirical evidence not only creates a wide space for theoretical speculation. It also provides open ground for ideological contention, and this can be seen, for example, in the dispute between universal, monogenetic, superstrate and substrate theories of the origins, development and classification of creole languages. The Danish linguist Otto Jespersen, for example, attributed their origins to universal processes of language acquisition, judging that the minds of their speakers were “just as innocent of grammar as those of very small babies” (1922: 228; Holm 1988: 36). In the interactions where pidgin and creole languages emerged, subordinated populations were credited with little creative influence, and little significance was attached to the source languages that they had brought with them (for example, Yoruba, Ewe or Akan among creole and pidgin speakers from West Africa). The influence of African source languages – the ‘substrate’ – became an especially significant issue, however, in the wake of the movements in the 1960s for independence in Africa and the Caribbean and for black power and civil rights in the U.S. Here, the emergence of creoles was construed as spectacular collective survival and creativity, accomplished by people living in conditions of extreme subordination, and Alleyne’s comments point to the ideology that motivated this view: In the early sixties when the work began to take shape in my own mind, my interest in creole language studies was overwhelmingly academic and objective... The study of so-called “creole” languages and of the language of Afro-Americans in general has since taken on new dimensions and a new significance, having become involved in the social, cultural, and political conflicts of our times... “the Black revolution” threw Afro-American studies into a new focus and gave them a new urgency... In the Caribbean, the reevaluation and regeneration of non-European-derived forms of behaviour is seen as a necessary factor in the development of the area. (1980: 1; see also e.g. Dalphinis 1985; Warner-Lewis 1996)

In line with this, there was an intense interest in those aspects of the social history of Africans in the Americas that led to African cultural retentions, and language was treated alongside music, dance, religion, art, folk-tales, proverbs, etc. “Language is similar to other aspects of culture in that it is learned in the socialization process. It undergoes change and development in relation to social interactional processes” (Alleyne 1993: 167). Between these two perspectives – the universalist and the substrate – there has in fact been a heated argument that has continued in a number of different guises to the



Creole metaphors in cultural analysis 

present, with one side implicitly seeing subordinated creole-speaking peoples as relatively passive vessels for the operation of general linguistic and psychological processes, and the other treating language as the product of people struggling to retain cultural roots in spite of repression. But among other things, many of the sides in the creole origins debate have been united in a search for total explanation licenced by the paucity of the historical record. This applies, for instance, to so-called superstratists (Chaudenson 2001; Mufwene), so-called monogeneticists (McWhorter), and so-called substratists (Lefebvre 1998). Chaudenson, in a chapter which approvingly cites Mufwene, states: The mistake of many creolists has lain in their constant inability to understand how, in societies where francophone Whites seem to have been a small minority, creoles could have developed whose basic linguistic materials obviously originated from French... A historical sociolinguistic analysis makes it possible to show some essential facts, the chief of which is the development of a system of approximations of French. (2001: 145–146)

McWhorter, meanwhile, writes in defense of little-acknowledged comparative and sociohistorical evidence which suggest that all plantation Creoles of a given lexical base shared single Pidgin ancestors, which themselves emerged not on plantations, but in trade settlements on the West African coast. (1999b: 122)

Finally, Lefebvre, while allowing for some superstrate influence, nevertheless concludes firmly that the creators of a creole language, adult native speakers of the substratum languages, use the properties of their native lexicons, the parametric values and semantic interpretation rules of their native grammars in creating the Creole. Creole lexical entries are mainly created by the process of relexification. (1998: 9)

Admittedly, all three of these authors present more nuanced statements of position elsewhere, but even so, when referring to each other’s work, they tend to draw rhetorically fierce and firm boundaries and lines of distinction, each in their different ways seeking totalized explanation of Creole genesis. In fact, Mühlhäusler – himself a creolist – identifies a number of other features that draw, for example, substratists and universalists together in spite of their disagreements. Their positions diverge to the extent that [u]niversalists believe that human agency in grammar making is just superficial tinkering and biologically-based forces will determine the more basic and lasting aspects of Pidgin and Creole grammar. Substratists, on the other hand, regard Pidgin and Creole as new containers for old culturally determined knowledge. (1995: 244)

 Roxy Harris & Ben Rampton

They share, however, a common metaphorical frame which comprises: 1. the reification metaphor that languages are things which can be looked at objectively outside their temporal and social context; 2. the conduit metaphor which assigns a central role to the grammatical code – the device which transforms messages into signals and vice versa; 3. a kind of container metaphor, that Pidgins and Creoles are containers which can be filled with either substratum or biologically founded grammatical rules. (1995: 244) Since this metaphoric frame is in fact characteristic not just of creolistics but also of a much wider field of linguistic research, at this point it is worth turning to the position of creole language study within linguistics more generally.

3 Creole language study and the shift in linguistics So far, our characterization of creole linguistics has pointed to a history of ideologically charged arguments about the pedigree of creoles (their origins, development and classification), to the limitations of the data available for such analysis, and to the totalizing ambitions of creole language theory. Finally, we cited Mühlhäusler on a number of problems in the conceptualization of language itself, and it is this that the current section elaborates on. Within the linguistics academy, there are a number of ways in which the study of creole languages has actually been quite radical. Both Sebba (1997: 288) and Mühlhäusler (1992) talk about it as a ‘theory-buster’, challenging gradualist (‘family tree’) models of language evolution and the synchronic bias of structural linguistics, and there can be no doubt that in some quarters, it has involved a very profound rethinking of the basis of linguistics – Le Page and Tabouret-Keller (1985), for example, argue that creoles provide a model for a great deal of language generally, and they use this to develop a fully dialogical – indeed rather Bakhtinian – view of language. Even so, an enormous amount of the day to day empirical work appears to accept the assumptions that have dominated linguistics at large, namely: 1. that language study is centrally concerned with systematicity in grammar (and coherence in discourse), 2. that people learn to talk grammatically (and coherently) from extensive early socialization in families and fairly stable local social networks. In line with this, grammatical systematicity has, as Mühlhäusler intimates, been something of the creole linguist’s holy grail (= 1), and the question of whether or not there are infants and children involved has been central to differentiation of a creole from a pidgin (= 2). In fact, the tenacity of assumptions 1 and 2 extends well beyond creole studies, and can even be found in those branches of language study where one might least expect them. Sociolinguistics has long contested Chomsky’s prioritization of grammar and



Creole metaphors in cultural analysis 

his idealization about homogeneous speech communities (cf. Mühlhäusler’s ‘reification metaphor’), and it has made variation and diversity its raison d’ être from the 1960s onwards. In spite of this, however, the belief has been that variation and diversity are describably structured, and whenever they have encountered variation, sociolinguists’ strongest instinct has been to unearth what they supposed was an orderliness and uniformity beneath the surface, an orderliness laid down in the early years of socialization. More recently, however, a number of linguists have been attending to debates about late modernity in sociology and cultural theory, and to the kinds of issue that Hannerz foregrounds (cf. (a) to (e) in section 1 above). There is a growing resonance when Zygmunt Bauman wonders whether “the reality to be modelled is... much more fluid, heterogeneous and under-patterned than anything sociologists have tried to grasp intellectually in the past” (1992: 65), and this kind of reconceptualization is carried over into creole and sociolinguistics by papers such as Le Page’s ‘Projection, focussing and diffusion’ (1980), or Pratt’s ‘Linguistic utopias’ (1987). To get a glimpse of the shift in progress, it is worth briefly reviewing some of the ways in which notions such as ‘speech community’ and ‘grammatical regularity’ are being reassessed. Historically, a speech community has been conceptualized as an empirically identifiable ‘real’ thing, a body of people who interact regularly, who have attitudes and/ or pragmatic rules in common, and who constitute the largest unit that one can generalize about in any given study (cf. Gumperz 1968; Labov 1989). That view, however, is now breaking down and the concept is splitting in two (actually compatible) directions (cf. Rampton 1998 for full discussion). In one direction, the concept translates into an intensive focus on the lived texture of situated experience in ‘communities of practice’ – community here being face-to-face interaction in well-established settings and social relationships like workshops, classrooms, marriages etc (cf. Eckert & McConnell-Ginet 1992). In the other direction, following scholars such as Anderson (1983), ‘community’ itself is analyzed as a totalizing concept, as an ideological sign that is used to constitute groups and to link languages with peoples (Gal & Irvine 1995). These reformulations substantially complicate the traditional idea that a creole is the language of a ‘new speech community’. The fragmentation of community entailed in the first approach synchronizes well with late-modern uncertainty about grand tantalizations, and it invites us to conceptualize any given creole as just one element in a much larger local multilingual repertoire, with different varieties operating in different communities-of-practice. In the second perspective – ‘community’ seen as an ideological sign – attention turns to the social and political mechanisms and interests involved in the definition, description and evaluation of creole-using groups, thereby 3. This can be seen in the variationist’s quest for the vernacular; in research on code-mixing and code-switching, where the emphasis was on systematic patterns established within relatively stable bilingual ingroups; and even in work on cross-cultural conflict and misunderstanding, where the problem was attributed to the gap between integrated cultural and linguistic systems.

 Roxy Harris & Ben Rampton

providing us with analytic insight into exactly the same kinds of issue that were addressed in section 2 (stereotyping, othering, total explanation). Turning to the issue of grammatical regularity, another set of problems arises. For most of the last 30 years, linguists working on natural speech data have given primacy to language use that is relaxed, informal and unselfconscious, judging that this is where grammatical system and regularity can best be found. Recently, however, there has been a very substantial growth of interest in artful performance, where “there’s heightened awareness of both the act of expression and the performer” (Bauman 1986: 3, see Bauman & Briggs 1990), as well as in intertextuality and double-voicing, where there is an uneasiness in speech produced by its penetration by other people’s talk (see below and note 12). Developments like these start to challenge the emphasis that mainstream variationist sociolinguistics has given to the idea of an unconscious core vernacular, making it look rather Fordist, and they are underpinned by a larger shift in our sense of the origins of meaning, a shift which moves the locus of meaning from system to situated practice (cf. Verschueren 1999). Instead of seeing language simply as system output, language-as-a-set-of-conventions-or-mental-structures gets pushed down to being just one among a number of semiotic resources available for local text production and interpretation. And instead of system itself being viewed as the main carrier of meaning – cf. Mühlhäusler on the ‘conduit metaphor’ – meaning gets analyzed as a process of here-and-now inferencing that ranges across all kinds of locally contingent percept, sign and knowledge, and that is carried out by agents with often very different sense-making resources at their disposal. All of these developments now make it much easier for linguistics to engage in the nonce, the anomalous, and the spectacular (cf. (b), (c) and (d) in section 1 above). As the quotation from Mühlhäusler itself testifies, a number of creolists are responding seriously to developments like these, and according to Jourdan, there is now “a new conception of social formations that puts the individual at the center of social relations, thus stressing the dialogic, contextual, and fluid nature of individual and collective praxis and agency” (1991: 189, and Sebba 1993 for an excellent example). Even so, it is important not to underestimate the extent to which these reorientations carry consequences for creole studies as a whole. The traditional idea that “[c]reoles... are... pidgins which have become the first language of a new generation of speakers” (Mühlhäusler 1997: 186) falls into disarray when the study of contemporary pidgins identifies processes of grammatical development that “result... from [their] use in new media (radio, print) and [their] use as a form of art” (Mühlhäusler 1997: 184, also Jourdan 1991; Sebba 1997, also (c) in section 1 above), and if it is likely that the dynamics of interaction and artful performance are important for the development of grammatical complexity – if the maturation of infant brains is no longer adequate as an explanation – then decontextualized texts, filleted ready for structural analysis, are no longer adequate as data, and lexical or grammatical analysis are no longer sufficient as analytic tools. Invoking social context ad hoc as an auxiliary factor when grammatical analysis ran aground may have been acceptable in the past (Hymes 1971: 9; Le Page 1994), but



Creole metaphors in cultural analysis 

this can hardly be an option if the situated use of a language might actually play quite a fundamental role in determining its grammatical shape. All in all, any approach which treats language as a stable, delimitable object of study – the ‘reification metaphor’ – becomes highly problematic. Since neither of us are committed creole linguists, we are not in a position to assess the extent to which creole studies has successfully adapted to the growing influence of these alternative perspectives, though in passing it is perhaps worth noting that they produce some tangible uneasiness in the concluding chapters of the textbooks on creoles and pidgins by both Mülhäusler (1997) and Sebba (1997). Even so, there do seem to be good grounds for doubting the value of traditional creole language study as a ground-breaking model or template for the analysis of cultural contact. Although there are some outstanding exceptions (e.g. Le Page & Tabouret-Keller 1985), on the whole it would be safer and more accurate to regard creole linguistics as the relatively recondite product of a subsidiary special interest group rather than as a beacon for the investigation of intercultural processes. In fact, if one wants to see some of the fundamental ways in which people process cultural mixing, translation and difference, then it would be much better to go directly to dialogic interaction, and it is this that we will try to illustrate in the next section.

4 Interaction as a site of ‘transcultural’ encounter In this section, we shall look closely at a short stretch of face-to-face interaction between a white monolingual Anglo adult and several adolescents, one of them Anglo and two of second-generation migrant descent. In it, both of the boys of Pakistani descent switch into a strongly accented Indian English, one of them uses (a version of) Jamaican Creole, and the Anglo boy produces some playground Punjabi. This kind of switching into other people’s ethnic languages – or ‘crossing’ as it is called in Rampton (1995) – is actually fairly common, both in Britain and elsewhere (see also Hewitt 1986; Back 1995; Rampton 1999), and in the data-set that the extract comes from, there were about 68 episodes in which white and African Caribbean adolescents used Punjabi, about 120 exchanges involving stylized Asian English, and more than 250 episodes in which a Jamaican creole influence was clearly detectable in 4. The extract comes from an ESRC-funded project entitled Language Use in the Multiracial Adolescent Peer Group, in which the ethnography of communication (Hymes 1972) and interactional sociolinguistics (Gumperz 1982) provide the main methodological tools. The project involved two years of fieldwork with 23 eleven to thirteen year olds of Indian, Pakistani, African Caribbean and Anglo descent in 1984, and approximately 64 fourteen to sixteen year olds in 1987. Data-collection focused mainly on a youth club and on lunch and breaktime recreation at school, and it included radio-microphone recording (approximately 145 hours), participant observation, interviewing and retrospective participant commentary on extracts of recorded interaction. The research is written up in full in Rampton (1995).

 Roxy Harris & Ben Rampton

the speech of white or Asian youngsters. Given the dearth of empirical data on interactional processes leading to the emergence of Atlantic and Caribbean creole languages, we certainly cannot claim that the social and interactional relations evidenced here should be of special interest to creole linguists. But we are confident that in many parts of the world, broadly comparable processes can be found in situations of domination, inequality and ethnic difference (cf. Rampton 1999), and that generalised speculation about the way in which groups are or were likely to respond to such conditions provides very little purchase on the intricate patterning of actual linguistic behaviour. The analysis that follows differs from traditional creole linguistics in paying relatively little attention either to grammatical processes or to the reconstruction of linguistic history. In their place, however, it points to a framework for sociolinguistic analysis that – offers a much richer set of tools for addressing cultural processes within globalization and transnational flow (cf. (a) to (e) in section 1), – retains an ability to engage with historical processes, – is well-tuned to the reconceptualizations of language practice outlined in section 3, and – also affords productive connection with traditional anthropological notions like ritual and liminality. In the 30 seconds or so of talk that the data extract covers, institutional power differences, ethnic inheritance, racism and histories of imperialism are all at issue. They are transposed, however, to a level of local interactional negotiation where one can observe human agents reflexively reworking the baggage and boundaries of ethnic descent, tentatively constructing new solidarities through language switching. Here is the extract:

Extract 1 Participants: Asif (15 yrs old, male, Pakistani descent), Kazim (15, male, Pakistani descent), Alan (15, male, Anglo descent), Ben (the researcher/author, 30+, male, Anglo descent). Setting: 1987. Having recorded these three friends with radio-microphones during their informal recreation, Ben is trying to get some feedback on extracts from the recordings. But the boys are in high spirits, Asif and Alan have just been talking playground Punjabi into the microphone from close up, and Ben is now trying to reestablish their commitment to the listening activity. 1 2 3 4 5 6 7

Ben: right shall I- shall we shall we stop there Kazim: no Alan: no come [ on carry on Asif: [ do another extract Ben: le- lets have (.) [ then you have to give me more= Alan: [ carry on Ben: =attention gents



Creole metaphors in cultural analysis 

8 Asif: ((quieter)) yeh [ alright 9 Alan: ((quieter)) [  alright 10 Asif: ((quieter)) [   yeh 11 Ben: I need more attention 12 Kazim ((in Indian English)): I AM VERY SORRY BEN JAAD [ 13 Asif ((in Indian English)): ATTENTION BENJAMIN  14 : [ ((laughter)) 15 Ben: [ right well you can- we cn16 Alan: [ BENJAADEMIN 17 Ben: we can continue but we er must concentrate a bit 18 [ more 19 Asif: [ yeh 20 Alan: alright [ (go on) then 21 Asif((in Indian English)): [ concentrating very hard [] 22 Ben: okay right 23 : ((giggles dying down)) 24 Kazim((in Indian English)): what a stupid ( )  25 Ben ((returning the microphone to what he considers to be a better position to catch all the speakers)): concentrate a little bit26 Alan: alright then 27 Kazim ((in Creole)): stop movin dat ting aroun [] 28 Ben: WELL YOU stop moving it around and then I’ll won’t 29 need to (.) r[ight 30 Kazim ((in Creole)): [ stop moving dat ting aroun  31 Ben: right okay [ 32 Kazim: [ BEN JAAD 33 Alan: ((laughs)) 34 Ben: what are you doing 35 Alan: ben jaa[ad 36 Ben: [ well leave ( ) alone 37 Kazim: IT’S HIM that ben jaad over there 38 Ben: right ((Ben continues his efforts to reinstitute the listening activity))

 Roxy Harris & Ben Rampton

We will return to more general transcultural issues in due course, but before that, it is worth taking a little time to attend to two aspects of the fine grain in the interaction here: i. the precise siting of the switches into different speech varieties, particularly stylized Asian English and Caribbean creole (taken here as an emic local dialect rather than as an analytic problem space); and ii. the processes of symbolic evocation involved in these switches away from ordinary vernacular English.

4.1

Interactional siting: Ritual and remedial interchanges

It is fairly obvious that the extract involves a struggle between two different definitions of the situation – very approximately, Ben’s research-oriented ‘retrospective-participant-commentary-on-extracts-of-recorded-data’ vs. their ‘havin’-a-good-time-listening-to-Ben’s-tapes’. But within this higher level indeterminacy, the precise moments when the boys actually switch codes are also significant. In lines 5 & 7 of the extract, Ben lays down the conditions for carrying on with the listening activity, and he also implies that the boys have made it pretty difficult hitherto and that it will be their own fault if the event is terminated. Asif and Alan appear to accept the conditions, and then a small sequence of ritual remediation begins in which the boys use stylized Asian English: Kazim apologizes in line 12; Asif declares his allegiance to what Ben wanted in lines 13 and 21; and Kazim seems to take Ben’s perspective in line 24’s muttered disapproval. But of course none of this can be taken at face value. According to Goffman (1971), in apologies people split themselves into two parts – the self that was guilty in the past, and now the new self that recognizes the offence and disavows the self of old. And so normally, you would expect people apologizing for noisy disorder to signal the split by switching into relatively quiet, serious, sincere voices. Not so here. In this episode the boys apologize for messing around by moving into a conspicuously false accent, which is accompanied with an equally contradictory loudness and hilarity. In fact a moment later, just as Ben seems to be signaling ‘back-to-business’ by repositioning the microphone, the boot moves to the other foot, Kazim switches into creole in line 27 and himself directs a ‘prime’ at Ben, this time constructing Ben’s activity as an impropriety. Rather than a remedial sequence, this leads to a short ‘run-in’ in which Ben accounts for his action by laying the offence with Kazim, a move which Kazim ignores by simply repeating his directive. Ben does not then take issue with this, but instead continues his efforts to reinstate the listening activity, using some optimistic boundary markers (“right, okay, right” – lines 29, 31 & 38). The boys respond with 5. What follows reproduces the data-analysis contained in Rampton (1999), where it is set within a discussion of how interactional sociolinguistic research can contribute to discussions of liminality in cultural studies.



Creole metaphors in cultural analysis 

“ben jaad”, a nickname in multi-ethnic peer-group Punjabi, which is later discovered to be an accidental invention of Alan’s, falling ambiguously between [], meaning ‘Ben, friend’, and [], ‘sister fucker’. The main structural point to make is that the boys switch into Asian English, Creole and Punjabi just at the moments when transgression and impropriety are made the focal issues. As we have already said, our sense of the common moral order of everyday life is temporarily jeopardized when infractions arise, and when this happens, we do not simply seek to repair whatever has been damaged or disrupted. What we mainly look for are signs of the culprit’s more general respect and regard for social rules and the order we approve (Goffman 1971: 98). What the boys provide, of course, is something rather different. It is not just that they withhold support for the norms and decorum that Ben’s appealing to. They actually switch into language varieties which symbolically activate domains of meaning where a white man’s judgment loses a lot of its legitimacy. The boys switch into Asian English in a sequence where they bow to Ben’s calls to order, and in doing so, they conjure a stereotype of Asian ‘babu’ deference which is historically ensconced in white British racism and which can be depended on to embarrass a white liberal conscience. The switch indexes race stratification as a potentially relevant issue in our encounter, and this strategic racialization was carried further in the switch to Creole, a code associated with the rejection of illegitimate white power.  Summing up so far, our first point is that there are points of indeterminacy – of liminality – in the interaction order that become showcase moments for the symbolic display of one’s social allegiance, and second, that people use these for all sorts of social play and experimentation, some of it contesting dominant orders, or seeking to redefine them. We can now move on to a third point, which has to do with the way the boys seem to position themselves around the symbolic voices they evoke, and the kind of historical consciousness that this evidences.

6. Especially for boys, (versions of) Jamaican creole had strong symbolic connotations of tough­ ness, vitality, excellence in youth culture and opposition to authority (cf. Hewitt 1986; Sebba 1993; Rampton 1995). 7. The switch into multiracial playground Punjabi worked on a slightly different tack. One of its effects could be to evoke a world of jocular peer group recreation in which the best role a monolingual adult could hope for would be the role of a benign but gullible onlooker. Another could be to maintain the ties with Alan, who was white like Ben but who was also a regular participant in multiracial playground Punjabi. 8. For a full discussion of liminality in interaction, cf. Rampton (1999).

 Roxy Harris & Ben Rampton

4.2

Processes of symbolic evocation: Historical consciousness in situated code-switching

When the boys used stylized Asian English, there was a fairly clear break between the deferential words uttered through the ‘babu’ persona on the one hand, and on the other, their obvious commitment to enjoyment on their own terms. In contrast, in the case of Kazim’s creole, it is not at all clear that he does not mean what he says: there are not any obvious accompanying cues to suggest he is joking, and the switch starts a sequence in which dispute is much more explicit than before. This difference fits with a very general pattern in the data-set: when adolescents used Asian English, there was nearly always a wide gap between self and voice, whether they were subverting the control of adults, criticizing age-mates, or pretending to be the referee in games. In contrast, when they crossed into Creole, the gap substantially diminished, and language features that youngsters tended to identify as black often worked their way into routine vernacular language use.10 Both of these patterns connected with local adolescent views of the different social worlds indexed by each code. It was clear from interviews and other data that Asian English was not only associated with the babu stereotype: it was also associated (a) with adults who had come to England from India and Pakistan, and (b) with recently arrived Bangladeshi peers.11 Youngsters generally expressed solidarity with their family relations, while there was intense hostility toward Bangladeshis, but in one way or another, Asian English consistently symbolized distance from the main currents of adolescent life, and in all of its connotations, it seemed to stand for a stage of historical transition that most adolescents now felt they were leaving behind. 9. When Asian English was used to criticize a peer, either seriously or in joking, it was used as a ‘say-for’, a voice not being claimed as part of the speaker’s own identity but as one that was relevant to the person being targeted. In fact, these peer criticisms seemed to achieve their effect as a negative sanction by threatening the recipient with regression, symbolically isolating them on a path of historical development now abandoned by adolescents who had arrived at an endpoint they now took for granted. In games, Asian English was used as a language of commentary, it featured in praise and encouragement (‘very good shot’), and it served to announce or keep the score. It still connoted a certain remoteness from the main currents of multi-ethnic adolescent youth culture, but whereas before this was something to be scorned, in games the detachment becomes much more altruistic and authoritative, helping to place the speaker at some remove from the competitors’ concentrated struggle for advantage. See Rampton (1995: Ch 6) for a full elaboration of these patterns. 10. The invariant question tag “innit” was, for example, analyzed as being originally black, and young people most often attributed Creole roots to new words in the local English vernacular. Other elements of unselfconscious local speech that could be identified as have Creole origins were: the omission of ‘s’ in the possessive, in the plural and third person singular present tense; zero realization of the indefinite article; absent past tense and participle markers; and copular and auxiliary omission. See Rampton (1995: 127–128) for further discussion. 11. See Rampton (1995: Ch 2) for a full account of this.



Creole metaphors in cultural analysis 

In contrast to the retrospective time frame conjured by Asian English, Creole stood for an excitement and excellence in vernacular and mass-mediated youth culture which many kids aspired to, and it was even described as ‘future language’. In line with this, when it was used in interaction, Creole tended to lend emphasis to evaluations that synchronized with the identities that speakers maintained in their ordinary speech. The use of Creole lent power to the speaker, and indeed when it was directed towards deviance, it often expressed approval.12 It is sometimes assumed that if one opts for relatively fine-grained interactional micro-analysis, one loses a view of history. Here, however, the different self-voice dynamics evidenced in the interactional use of these two varieties look very much like the small-scale refraction of much larger, longer processes of historical migration and transnational flow, with adolescents using language code selection to position themselves both in relation to their family’s past in the Indian subcontinent, and in relation to their futures in a multi-ethnic UK. In fact, if we turn back to Extract 1, we can provide a more acute characterization of how the broader symbolic resonance of these two speech varieties could fit into the ongoing interactive negotiation of social relations. In Extract 1, switching into Asian English and Creole was closely associated with moments of disagreement and tension, and in their different ways, both language varieties were counterposed to the dominant order of the interview, one subverting it through mock subservience, the other contesting it more overtly. The boys’ symbolic 12. These processes can in fact be articulated in the terms of Bakhtin’s theory of double-voicing: the self-voice opposition running through the uses of stylized Asian English resembles Bakhtin’s ‘vari-directional double-voicing’, while the much closer self-voice identification in Creole conforms to his account of ‘uni-directional double-voicing’. When they engage in double‑voicing, speakers use someone else’s discourse (or language) for their own purposes, “inserting a new semantic intention into a discourse which already has... an intention of its own. Such a discourse... must be seen as belonging to someone else. In one discourse, two semantic intentions appear, two voices” (Bakhtin 1984: 189). Bakhtin describes several kinds of double‑voicing, and one of these is described as ‘uni‑directional’. With uni‑directional double‑voicing, the speaker uses someone else’s discourse “in the direction of its own particular intentions” (1984: 193). Speakers themselves go along with the momentum of the second voice, though it generally retains an element of otherness which makes the appropriation conditional and introduces some reservation into the speaker’s use of it. But at the same time, the boundary between the speaker and the voice they are adopting can diminish, to the extent that there is a “fusion of voices”. When that happens, discourse ceases to be double‑voiced, and instead becomes ‘direct, unmediated discourse’ (1984: 199). The opposite of uni‑directional double‑voicing is vari-directional double‑voicing, in which the speaker “again speaks in someone else’s discourse, but... introduces into that discourse a semantic intention directly opposed to the original one”. In vari‑directional double‑voicing, the two voices are much more clearly demarcated, and they are not only distant but also opposed (1984: 193). On Bakhtin’s notion of double-voicing in sociolinguistics, see Hill and Hill (1986), Cazden (1989), Fairclough (1992), and Rampton (1995: Chs 8.5 and 11.1). In cultural studies, see e.g. Mercer (1994: 62f.) and Bhabha (1996: 57).

 Roxy Harris & Ben Rampton

display of historical consciousness was activated, in other words, in quite a sharp micro-political encounter. At the same time, however, it is vital to recognize the lack of fit between ‘map’ and ‘territory’, between the symbolic imagery of race relations and the here-and-now actuality of who was participating on whose side in the event. Even though stylized Asian English and Creole might conjure images of white domination and black resistance, the interaction itself did not break down into white/ Anglo vs. black/ Asian. Alan was also present, he was white, and as his playground Punjabi ‘ben jaad’ made clear, he was with Asif and Kazim. In sum, it is not only that the use of these linguistic resources displayed a sense of historical trajectory in which Asian youngsters were developing an alignment with young people of African Caribbean descent (through Creole as ‘future language’). In the situated dynamics of the encounter itself, white friends in the same subordinated position could also be included. Language, history and changing group relations have of course been defining concerns of creole linguistics, and so it is perhaps worth clarifying the way in which the approach illustrated in the analysis of Extract 1 is different. Unlike much of creole language study, our analysis makes no attempt to show how the economic and political relations between groups lead to the historical development of particular types of language or lexico-grammatical structure. Though we certainly accept that material conditions are linked to ideologies of ethnicity and ‘race’, which are linked to situated interaction, which is linked to grammar, for us anyway, both the multiple connections and the partial autonomy of these ‘levels’ makes the analysis of how history impacts on language a dauntingly complex task. But this inability to run a line of causation from, say, population movements down into the details of language structure does not entail a return to the static, ahistorical analysis that creole linguists originally broke away from (Hymes 1971: 7; Bickerton 1975: 166), motivating them to construct the dynamic models of creole system and structure that anthropologists have found attractive (e.g. Drummond 1980). Indeed, although our analysis might lack the grand empirical sweep of the creole linguist’s effort to reconstruct the historical origins of a particular pidgin or creole, it lays a great deal more emphasis on agency, and it brings us far closer to the kind of practical consciousness that new social formations can emerge from (Williams 1977; Gal 1988). What we can see in Extract 1 is a felt sense of historical positioning, inscribed in practical consciousness, articulated through symbolic language use in situations that lead, cumulatively, to the development of new solidarities that cut across the ‘ethnic absolutism’ of dominant ideologies (Gilroy 1987; Hall 1988). This is close to what Williams calls ‘creative practice’ – “not casting off an ideology, or learning phrases about it, but confronting a hegemony in the fibres of the self and in the hard practical substance of effective continuing relationships” (1977: 212).13

13. Volosinov’s (1973) notion of ‘behavioural ideology’ is also relevant here; see Rampton (2001). For a full discussion of Williams’ notions of ‘hegemony’ and ‘creative practice’ in relation to language and social class, see Rampton (2006: Part III).



Creole metaphors in cultural analysis 

This line of interpretation, we suspect, is likely to be quite compatible with Hannerz’s general approach to cultural processes. But the apparatus lying at the centre of our interaction analysis owes little to creole linguistics – instead, it is an eclectic mixture of Hymes, Gumperz, Goffman, Bakhtin and conversation analysis, broadly comparable to what one can find in Duranti’s (1997) textbook on Linguistic Anthropology. In conclusion, we would briefly like to review the kind of contribution that sociolinguistic discourse analysis can make to Hannerz’s agenda for research on globalization and transnational flow.

5 Conclusion Although there are limits to what a single extract can reveal, there are a number of ways in which the analysis of interaction in section 4 connects to Hannerz’s agenda. In terms of the substantive issues he raises, there is only a glimpse of half-understanding and unevenly distributed linguistic resources in Alan’s ‘ben jaad’ in line 36 of Extract 1 (compare (b) in section 1 above), but phenomena like this were constitutive features in the language switching and crossing evidenced more widely (Rampton 1995). They were often associated with the kind of interactional intensity displayed in the laughter and dispute in lines 14–30 ((d) in section 1), and popular media also had a major influence, conspicuously affecting young people’s rights and access to the interactional use of another language (section 1 (c); Rampton 1995: Part IV). Methodologically, sociolinguistic micro-ethnography provides rigorous constraints on the totalizing explanation and the free-wheeling analytic speculation that we associated with the ‘must have’ tradition in creole linguistics in section 2 (see also (a) in section 1). Conversation analysis, for example, operates with an analytic ‘aesthetic of slowness’ (Silverman 1999: 415) which insists that one tries to track participants’ sense making procedures from one moment to the next, and this provides a validity check on notions like ‘contradiction’, ‘liminality’ and ‘ambivalence’, which in macro-studies sometimes seem more like analyst attributions than participant experiences. Alternatively, because one is usually working with dozens (even hundreds) of examples, “small facts continually get in the way of large issues” (section 1 (e)), and it is blindingly obvious that there are lots of different things going on with, say, a set of acts that one might broadly call ‘resistance’. Indeed, with any single one, there is usually a lot of data on the context, and so if you are interested in political analysis, you can look at a particular act as a micro-political intervention in specific social relations there-and-then. All of this makes it quite hard to fall head long into romantic Davidand-Goliath interpretations, and in the project described above, for example, while one could sometimes see Asian English as a subversive probe directed at white adults, elsewhere it was used in the crudest forms of racism. Our overall position, then, is that it would be a great shame if anyone thought that creole language studies represented the sum total of what (socio)linguistics has to offer

 Roxy Harris & Ben Rampton

to the understanding of transnational processes. We are obviously not speaking here as disgruntled disciplinary purists, and if in the past, the linguistic modeling of creoles has helped anthropologists like Drummond to escape the hold of static synchronic structuralism,14 we would welcome this a valuable interdisciplinary process, in the same way that we appreciate the inspiration that ‘creolization’ has provided in Hannerz’s analyses of cultural complexity (1992, 1996). But at the same time, it is important not to mistake the metaphorical resonance of linguistic concepts for the weight they have in the field where they find their centre of gravity. If you do, you risk losing sight of the fact that for certain dimensions of social life, interactional and ethnographic sociolinguistics really do provide the most accurate analytic tools that we have. Quite how and how far language, discourse, and interaction articulate with other levels of cultural organization and flow, and exactly how easily reconciled the different theoretical idioms used in their analysis might be, we are not sure. But from our point of view anyway, that is one of the most important things for an interdisciplinary research program to address, and we hope we have managed to provide just a glimpse of what non-creolist sociolinguistics can offer.

References Alleyne, M. 1980. Comparative Afro-American. Ann Arbor: Karoma Press. Alleyne, M. 1993. Continuity versus creativity in Afro-American language and culture. In Mufwene, S (ed.). Africanisms in Afro-American Language Varieties. Athens, Georgia: University of Georgia Press. 167–181. Anderson, B. 1983. Imagined Communities. London: Verso. Back, L. 1995. X amount of Sat Siri Akal!: Apache Indian, reggae music and intermezzo culture. In Ålund, A. & R. Granqvist (eds). Negotiating Identities: Essays on Immigration and Culture in Present-Day Europe. Amsterdam: Rodopi. 139–166 Bakhtin, M. 1984. Problems in Dostoevsky’s Poetics. Minneapolis: University of Minnesota Press. Bauman, R. 1986. Story, Performance and Event. Cambridge: Cambridge University Press. 14. Following Bickerton’s (1975) attempt to describe the variation in Guyanese creole language as a single grammatical continuum, Drummond (1980) suggested that the use of ethnic categories in Guyana could also be theorised as a continuum system. There is nothing controversial in Drummond’s emphasis on ethnicity as a set of relational identities which a lot of different people share some knowledge of, but more than 20 years later, we would have to say (a) that the emphasis on ethnicity as a fundamental principle driving local social activity (e.g. 1980: 363, 368) now looks very reductive; (b) though the conceptualization of system itself might be reworked in Drummond’s paper, the attempt to force local cultural knowledge into a single account resembles the same kind of totalization that we criticized in Section 2; and (c) any appropriation of the continuum model would now have to confront a number of serious theoretical and empirical problems that linguists have subsequently identified in Bickerton’s approach (see Sebba 1997: Ch 3.5).



Creole metaphors in cultural analysis 

Bauman, R. & C. Briggs. 1990. Poetics and performance as critical perspectives on language and social life. Annual Review of Anthropology 19. 59–88. Bauman, Z. 1992. Intimations of Postmodernity. London: Routledge. Bhabha, H. 1996. Culture’s inbetween. In Hall, S. & P. du Gay (eds). Questions of Cultural Identity. London: Sage. 53–60. Bickerton, D. 1975. Dynamics of Creole System. Cambridge: Cambridge University Press. Bickerton, D. 1989. The lexical learning hypothesis and the Pidgin-Creole cycle. In Pütz, M. & R. Dirven (eds). Wheels within Wheels: Papers of the Duisburg Symposium on Pidgin and Creole Languages. Frankfurt am Main: Verlag Peter Lang. 11–31. Bloomfield, L. 1935. Language. London: George Allen & Unwin. Cazden, C. 1989. Contributions of the Bakhtin circle to ‘communicative competence’. Applied Linguistics 10 (2). 116–128. Chaudenson, R. (revised in collaboration with S.S. Mufwene). 2001. Creolization of Language and Culture. London/ New York: Routledge Dalphinis, M. 1985. Caribbean and African Languages: Social History, Language, Literature and Education. London: Karia Press Dillard, J.L. 1972. Black English: Its History and Usage in the United States. New York: Random House. Drummond, L. 1980. The cultural continuum: A theory of intersystems. Man 15. 352–374. Duranti, A. 1997. Linguistic Anthropology. Cambridge: Cambridge University Press. Eckert, P. & S. McConnell-Ginet. 1992. Think practically and look locally: Language and gender as community-based practice. Annual Review of Anthropology 21. 461–90. Fairclough, N. 1992. Discourse and Social Change. Oxford Polity. Gal, S. 1988. The political economy of code-choice. In Heller, M. (ed.). Codeswitching: Anthropological and Sociolinguistic Perspectives. The Hague: Mouton. 245–264. Gal, S. & J. Irvine. 1995. The boundaries of languages and disciplines: How ideologies construct difference. Social Research 62 (4). 967–1001. Gilroy, P. 1987. There Ain’t No Black in the Union Jack. London: Hutchinson. Goffman, E. 1971. Relations in Public. London: Allen Lane. Gumperz, J. 1968. The speech community. In International Encyclopedia of the Social Sciences. London: Macmillan. 381–386. Gumperz, J. 1982. Discourse Strategies. Cambridge: Cambridge University Press. Hall, S. 1988. New ethnicities. ICA Documents 7. 27–31. Hannerz, U. 1987. The world in creolisation. Africa 57 (5). 546–559 Hannerz, U. 1989. Culture between center and periphery: Toward a macroanthropology. Ethnos 54 (3/4). 200–216. Hannerz, U. 1992. Cultural Complexity: Studies in the Social Organisation of Meaning. New York: Columbia University Press. Hannerz, U. 1996. Transnational Connections. London: Routledge. Harris, R. 2006. New Ethnicities & Language Use. Basingstoke: Palgrave Macmillan. Hewitt, R. 1986. White Talk Black Talk. Cambridge: Cambridge Universtiy Press. Hill, J. & K. Hill. 1986. Speaking Mexicano. Tucson: Arizona University Press. Holm, J. 1988. Pidgins and Creoles, Volume 1: Theory and Structure. Cambridge: Cambridge University Press. Huber, M. 1999. Atlantic English Creoles and the Lower Guinea Coast: A case against Afrogenesis. In Huber, M. & M. Parkvall (eds). Spreading the Word: The Issue of Diffusion among the Atlantic Creoles. London: University of Westminster Press. 81–110

 Roxy Harris & Ben Rampton Hymes, D. (ed.). 1971. Pidginisation and Creolisation of Languages. Cambridge: Cambridge University Press. Hymes, D. 1972. Models of the interaction of language and social life. In Gumperz, J. & D. Hymes (eds). Directions in Sociolinguistics. Oxford: Blackwell. 35–71. Jespersen, O. 1922. Language: Its nature, development and origin. London: Allen and Unwin. Jourdan, C. 1991. Pidgins and Creoles. Annual Review of Anthropology 20. 187–209. Labov, W. 1989. The exact description of the speech community: Short a in Philadelphia. In Fasold, R. & D. Schiffrin (ed.). Language Change and Variation. Amsterdam/ Philadelphia: John Benjamins. 1–57. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Le Page, R. 1980. Projection, focussing and diffusion. York Papers in Linguistics 9. 7–32. Le Page, R. 1994. The notion of ‘linguistic system’ revisited. International Journal of the Sociology of Language 109. 109–120. Le Page, R. & A. Tabouret-Keller. 1985. Acts of Identity: Creole-based Approaches to Language and Ethnicity. Cambridge: Cambridge University Press. McWhorter, J. 1999a. A debate on Creole origins. In Huber, M. & M. Parkvall (eds). Spreading the Word: The Issue of Diffusion among the Atlantic Creoles. [Westminster Creolistics Series 6.] London: University of Westminster Press. 305–318 McWhorter, J.H. 1999b. The Afrogenesis hypothesis of plantation Creole origin. In Huber, M. & M. Parkvall (eds). Spreading the Word: The Issue of Diffusion among the Atlantic Creoles. [Westminster Creolistics Series 6.] London: University of Westminster Press. 111–152. Mercer, K. 1994. Welcome to the Jungle. London: Routledge. Mühlhäusler, P. 1992. What is the use of studying pidgin and creole languages? Language Sciences 14 (3). 309–316. Mühlhäusler, P. 1995. Pidgins, creoles and linguistic ecologies. In Baker, P (ed.). From Contact to Creole and Beyond. London: University of Westminster Press. 235–250. Mühlhäusler, P. 1997. Pidgin and Creole Linguistics. London: University of Westminster Press. Parkvall, M. 1998. A short note on the peopling of English St Kitts. In Baker, P. & A. Bruyn. (eds). St. Kitts and the Atlantic Creoles: The Texts of Samuel Augustus Mathews in Perspective. [Westminster Creolistics Series 4.] London: University of Westminster Press. 63–74. Pratt, M.L. 1987. Linguistic utopias. In Fabb, N., D. Attridge, A. Durant & C. MacCabe (eds). The Linguistics of Writing: Arguments between Language and Literature. Manchester: Manchester University Press. 48–66. Rampton, B. 1995. Crossing: Language and Ethnicity among Adolescents. London: Longman. (2nd ed., 2005. Manchester: St Jerome Press.) Rampton, B. 1998. Speech community. In Verschueren J., J.-O Östman, J. Blommaert & C. Bulcaen (eds). Handbook of Pragmatics. Amsterdam/ Philadelphia: John Benjamins. 1–34. Rampton B. 1999. Sociolinguistics and Cultural Studies: New ethnicities, liminality and interaction. Social Semiotics 9 (3). 355–373. Rampton, B. (ed). 1999. Styling the Other. Special issue of Journal of Sociolinguistics 3 (4). Rampton, B. 2001. Critique in interaction. Critique of Anthropology 21 (1). 83–107. Rampton, B. 2006. Language in Late Modernity: Interaction in an Urban School. Cambridge: Cambridge University Press. Scott, J. 1990. Domination and the Arts of Resistance. New Haven: Yale University Press. Sebba, M. 1993. London Jamaican. London: Longman.



Creole metaphors in cultural analysis 

Sebba, M. 1997. Contact Languages: Pidgins and Creoles. Basingstoke: Macmillan. Silverman, D. 1999. Warriors or collaborators: Reworking methodological controversies in the study of institutional interaction. In Sarangi, S. & C. Roberts (eds). Talk, Work and Institutional Order. Berlin: Mouton de Gruyter. 401–425. Verschueren, J. 1999. Understanding Pragmatics. London: Edward Arnold. Voloshinov, V. 1973. Marxism and the Philosophy of Language. Harvard: Seminar Press. Warner-Lewis, M. 1996. Trinidad Yoruba: From Mother Tongue to Memory. Tuscaloosa: University of Alabama Press. Williams, R. 1977. Marxism and Literature. Oxford: Oxford University Press.

Transcription conventions [ ] (.) (1.5) [ [ CAPITALS ( ) (text) ((text:)) bold

IPA transcription, revised to 1979 pause of less than a second approximate length of pause in seconds overlapping turns loud speech inaudible speech hard to discern, analyst’s guess ‘stage directions’ stretches of speech conspicuously influenced by Indian English and Jamaican Creole

Index A accusative  54–55, 59–61 Acehnese  112–115, 118–120, 124, 128–132, 134–138 adaptation  128, 151, 153–156, 160 admixture  5, 7, 14, 44, 63, 109–110, 126, 129, 136, 213, 216, 218 adpositions  60, 120, 155–157 affixation(al)  22–24, 29, 30–31, 33–34, 132, 169 African American Vernacular English  219, 222 African cultural retentions  268 Afrikaans  6, 141, 144, 147, 153–154, 157, 221, Afrikaners  220–221 Afrogenesis  229 agency  5, 167, 189–190, 194–195, 269, 272, 280 agreement  31–34, 41, 48, 73–74, 81 Angolar  31, 110, 129, 134–136 Annobón  31, 109 anthropology  266 antigroup  111 Arabic  22, 77, 114, 120, 174 artful performance  272 articulation index  92–93 Asian Creoles  9 Asian English  273, 275–281 Aslian languages  115–116 Aspect  29–30, 33, 42, 74–76, 84–85, 92–93, 133, 150, 174, 180–184, 259 association operator  72–75, 80–81 associational semantics  71, 75, 89 Atlantic Creoles  22, 27, 29, 40, 48, 90, 183, 206, 209, 222, 266, 274 Atlantic slave trade  231, 252, 260 Austronesian  125–129, 131–133

B Baba Malay (BM)  69, 204, 206, 209–220, 222 Babas  205, 208, 210, 212, 216 baby-talk  266–267 Bahnar(ic)  112, 114–115, 120, 125–127, 131, 133–137 Balinese  207, 209 Basic Variety  174–176, 194, basilectalization  134, 167, 169, 173–174, 176, 219 Bazaar Malay  62, 129, 204, 218–219 Berbice Dutch  23–24, 29, 31, 110, 230 bilingualism  91, 137, 213 Bislama  71, 84, 87, 89–94, 102, 178 Bozal slaves  220–221 British  188, 210–212, 220–221, 230–231, 237–251, 255–257, 260–261, 277 C Cantonese  180–185, 209–220 Cape Dutch Pidgin (CDP)  141– 160 Cape English Pidgin (CEP)  142, 144–145, 151 Caribbean  171, 187–188, 203, 206, 220, 222, 227–234, 237–240, 242–261, 266, 268, 273–274, 276, 280 Caribbean (plantation) society  135, 204 Caribbean intellectuals  188, 190, 192 Caribbean Creoles  8, 9, 43, 123, 219 case  41, 43, 54–63, 145–146 case syncretism  59, 62 Cham(ic)  109–116, 118–139

Chinese  26, 33, 76, 95–96, 112–114, 119–120, 125–126, 132, 171, 182, 203, 205–213, 216, 218–221 Hainan Chinese  132 Minnan Chinese  116 Chinese Pidgin English  185 Chru  26, 112, 115, 120, 130 classifiers  116, 119, 124, 127, 133 code-switching  278 colonial  8–9, 12, 14, 187–193, 203–204, 211, 214, 219, 221, 222, 227, 230, 237–246, 249–250, 255–256 colonization  205–206, 210, 219, 221, 227, 230–231, 234–235, 242–245, 249, 254–255, 259 competition  40–49, 56–57, 63–64, 209, 218 complexity  6–7, 31, 56, 68–71, 75–76, 79, 90–94, 174, 233, 282 compositional semantics  71–75, 80, 89–91, 93 compositionally articulated languages  74, 79, 88, 92 compositionally associational languages  74, 76, 79, 92 compounding/ compounds  22, 25–27, 34, 155, 157 congruence  40, 43, 45, 56–57, 61–62, 180, 213, 215, 217 contact-induced change  43, 132, 136, 213 continuative  30–31 control  59–60 conversion  26, 34 creole creole exceptionalism  4–5, 13–14, 192, 232–233 creole myths  12–14 creole origins  8, 228, 232, 269, 278 creole paradigm  1–2, 4, 10, 12, 14 creole phenotype  50

 Deconstructing Creole Creole Prototype (CP)  6–12, 14, 109–112, 115, 119, 121, 126, 130, 133, 136 gradual restucturing (into creole)  191 primordial creole  110 prototypical creole  109–111, 126, 136, 169, 193, 260 creolization  3, 5, 8–12, 15, 39, 46, 63, 129, 134–136, 167–169, 172–173, 179, 188–193, 204, 208, 212, 222, 229, 230–236, 238–239, 249, 253, 259–261, 265–266, 282 creolization continua  234, 239, 259 sociocultural creolization  188–189, 191 D dative  43, 59–61, 259 deconstruction  1, 2, 5, 12 decreolization  9, 13, 172 determiner  53, 185–186 double-voicing  272 Dravidian  57, 157 Dutch  6, 8, 26, 91, 174 Dutch East India Company  221 E East Indies  8, 207 ecology  11, 14, 212, 218–219, 221, 233 economy  49 political economy  227, 231, 234, 237, 239, 241, 244, 246, 249, 250, 258 elaboration  41, 75, 93, 184 E-language  14, 33, 233 English  8, 14, 26, 40, 43, 70–72, 74–79, 84, 87–91, 93, 95, 129, 145, 170, 174, 178, 181, 211, 221, 258 indigenization of English  203, 205 English-based/-lexifier Creoles  27, 42, 227, 230–231, 260–261 evolution  5–7, 11–12, 40, 58, 63, 73–74, 135, 188, 210, 213, 218, 220–221, 260 language evolution  14, 172, 174, 203–205, 270

exceptional(ist)  4–5, 12–14, 39, 57, 63, 193, 223, 260–261 exceptional diachrony  13, 233–234 F Feature Pool (FP)  39, 44–47, 53, 55, 57, 62–63 Fiji English  182 Fongbe  71, 84, 87, 89–90, 104 Founder Principle  63, 203, 218, 221 frequency  14, 45, 49, 57, 60, 62–63 Fujian  207 G Gbe  43, 47–56 genetic relationship  167–168 globalization  205, 265, 274, 281 grammaticalization  24, 30 Gungbe  47, 48, 50–52, 64 H Hakka  209, 220 Haroi  112, 115, 130 Hawai‘i  9, 10, 169–170, 172, 179, 180, 184–185, 209–210, 220 Hawai‘i Creole  10, 25, 121, 170, 172, 179–185 Hawai‘i Pidgin English (HPE)  170–172, 176, 184–185, 193 Hebrew  71, 84, 87–99 heteroglossia  251–253 Hlai  115–116, 125, 132 Hokkien  129, 181, 206, 209–210, 213–218, 220 Hrê  115, 133 I Iban  113 iconicity  24–25, 71 identity  167, 188, 190–192, 194–195, 211, 220, 221, 251–257 ideology  2, 187–193, 231–233, 242, 246, 254, 268, 280 I-language  14, 187 imperfect learning   168, 204 imperialism  192, 274 Indian Ocean  191, 203–204, 206, 209, 219–220, 222, 229 indigenization  199, 203–205, 221 Indo-European languages  10 Indo-Portuguese  206

Indonesian  58, 69–70, 75, 130, 206, 208–209, 212 inflectional  28, 40 interference  168, 178 interlanguage  13, 174, 176–177, 184, 186, 194–195, 213 intertwining  52, 63, 213 isolating languages   22, 45–46, 57, 80, 84, 88–89, 94 J Jakarta Malay  129 Jarai  112, 115, 117, 119, 120 Javanese  207, 209 K Kéo  94, 110 Kerinci  113 Khmer  114–115, 120, 124–129, 133–135 Kirinda Java  57, 60–61 Khoekhoe  149–151, 153–160 Khoekhoe Afrikaans  6, 141, 144–147 Kimbundu  110 L Language Bioprogram Hypothesis (LBH)   2, 9–10, 42, 121, 134 language change  4, 6, 11, 13–14, 32, 39, 44, 114, 130, 167, 169, 172–176, 185–186, 213, 233 language creation  1, 3, 6–7, 10–11, 14, 39, 42–43, 48, 63 lexifier  8, 13, 40, 43, 46, 48, 53, 56–57, 90–91, 121, 129, 167–169, 172–177, 180–182, 184, 186–187, 191, 193–195, 204, 213, 218–219, 227–231, 258–259 linguistic feature  44, 50, 64, 68, 186 Louisiana  171, 203, 220 M Malacca  205–208, 210–213, 219, 221, 224, 226 Malay  9, 57–58, 60–62, 64, 69, 112–116, 118–120, 124, 128–129, 131, 133–134, 205–209, 212- 221 Malay-based Creoles   69 Malay/ Indonesian   69–70 Malaya/ Malaysia  115, 200, 206–207

Malayic  41, 43, 58, 111–114, 119, 124–125, 127–134 Malayo-Polynesian  70, 110–111, 113, 119, 126, 128–129, 131, 134–137 Malayo-Portuguese  8 Mandarin  33, 70, 95, 115, 181, 212 markedness  15, 44–45, 62–63, 65 Martinique  8, 171, 203, 220, 240–241, 244 Melanesian pidgin  178–181, 185 metaphor(s)  7, 265 conduit metaphor  270, 272 container metaphor  270 reification metaphor  270– 271, 273 metatypy  125 Min  207 Minangkabau  70–71, 80, 84, 87–90, 93–96, 113 Mindanao Chabacano  110 mixed language(s)  5, 57, 110, 204–205, 213, 218, 220 Mon-Khmer  111, 113–115, 117–121, 124–130, 133, 135–136 Mulatto Creoles  206 multilingual  5, 44, 63, 209, 218, 220–221, 271 multilingualism  5, 14, 188, 213 N nativized/ nativization  176, 221 Nyonyas  208 O oppression  191, 205–206, 221–222 P Pacific  8, 90, 201 Pacific Pidgins  204 Pacific Pidgin English  185 Papiamentu  26–27, 31, 68, 71, 84, 87–94, 100, 261 paradigm  1, 2, 4–5, 10, 12, 14, 187–188, 195, 231, 244, 266 inflectional paradigm  40 nominal paradigm  40 verbal paradigm  40 sociocultural paradigm  187– 188, 192 Penang  206, 210–211, 219 Peranakans  203–208, 210–213, 218–222

Index  person-gender-number (pgn) marker  151 pidgin  8–11, 13, 27, 33, 109, 129, 141, 143–144, 153–155, 167–172, 176, 182, 184, 193, 204, 206, 213, 268–270, 280 pidginization  9, 13, 27, 40, 109, 132, 136, 167, 169, 176, 260 Pidgin Malay Derived (PMD)  58, 213–214 plantation colonies  169, 204, 222 politics  228, 231, 234, 237–238, 245–246, 255–257 Portuguese  31, 41, 46, 66, 127, 129, 148, 180, 183–185, 255, 261 Portuguese colonies  206–207, 227, 229, 242–245 Portuguese navigators  255 Portuguese-lexifier Creole  124, 148, 261 possessives  53–54, 152–153, 182, 185 postcolonial  5, 167, 194 postcolonial ideology  188– 191 pragmatics  75–79 prefix(ation)/ prefixing  13, 22–23, 31, 33 Principense  135 Pro-drop  76, 147–149, 160 Proto-Austronesian   112, 118, 131 Proto-Chamic  112–113, 119–120, 124, 126–133, 135 Proto-Malayic  113 Proto-Malayo-Polynesian  113, 131, 134 R racism  188, 253, 256, 274, 277, 281 Rade  112, 115, 120, 130 reduplication  22, 24–25, 28–31, 34, 258 relexification  6, 13, 90, 114, 126, 128–129, 141–142, 144–147, 149–160, 178, 184, 269 replication  44, 57, 173, 174 resistance  190–191, 194, 250, 255, 257–258, 263–264, 280–281, 284

restructuring  9, 39, 41, 50, 63, 69, 94–97, 169, 173, 176, 186, 191, 195, 198–199, 204, 212–214, 222 Riau Indonesian  33, 36, 69–71, 74–80, 94–96, 110–111, 113, 129 Roglai  112, 115, 120, 124, 130 S Sanskrit  8, 113–114, 120, 127, 135 Saotomense  110, 135 Saramaccan  25, 28, 31, 47–48, 52, 69, 71, 94, 109–110, 120, 127, 129, 134–136, 230, 258–259 second language acquisition  5– 6, 9, 40, 167, 174–178, 183, 194 selection  6, 11, 40, 42–47, 49, 52–53, 56–57, 63–64, 95, 180, 218–219, 279 semantic indeterminacy  74–75 semantic transparency  27, 44, 62, 133 settlement colonies   203, 206, 210, 219 Siak Malay 69–70, 94 sign language  33–34 simplicity  2, 4, 6–7, 10, 12–13, 21, 39, 56, 69–70, 76, 79, 94–97, 171–176 simplification  6–7, 10, 13, 40, 46, 70, 95–96, 169, 174–175, 184, 193, 214, 218, 233, 260 Singapore  205–206, 208, 210–212, 219 Singapore Colloquial English  181 Sinhala  41, 43, 45, 57–62 Sinitic  5, 70, 121, 213–215, 217–218 slavery  10, 12, 189–192, 205, 221–222, 230, 237, 244–252, 255, 257, 259, 261 social interaction  219, 268 South Africa  220–221 Southeast Asia  5, 89, 203, 205–208, 211, 213–214, 222 SOV  45, 141–144, 150, 155, 160 speech community   32, 233, 238, 271 Sranan  8, 24–27, 30, 41, 43, 47–48, 51–52, 71, 84, 87–94, 101, 110, 136, 259 Sri Lanka Malay (SLM)  32, 40–41, 43, 45–46, 54, 56–63

 Deconstructing Creole Straits Settlements   208, 219 Straits-born Chinese   212, 219 stripping  142, 144–153, 155–158, 160 subsistence  236–237, 248–252, 254, 256–258 substrate(s)  6, 8–10, 17, 27, 40, 43, 46, 53, 56, 90, 145, 156–157, 167, 177–187, 191, 195, 209, 213, 228, 231, 234, 236–238, 244, 248–249, 251, 253, 255–256, 258, 260, 268 suffix(ation)/ suffixing  22–24, 29, 31, 43, 48, 53, 62, 84, 143–144, 146, 151, 155–156, 158, 178–180, 185, 259 Sundanese  70, 80, 84, 87–90, 93–96, 108 superstrate(s)  6, 46, 56, 91, 126, 177–178, 228, 231, 234–242, 244–245, 259, 267–269

theory  1–4, 10, 15, 228–229, 232, 258, 266–267, 270–271 theory of genesis  43, 46 topicalization  155, 158 trade colonies  206–207 transfer  41, 54, 91, 126, 177–187, 189, 191, 194–195 transmission  10, 13–14, 44, 70, 167–169, 172–178, 181, 187, 191–195, 204, 209, 213, 218–219 Tsat  112, 114–115, 119–120, 125, 130, 132–133 Twi  71, 84, 87–89, 103 typology  5, 39, 41, 42, 46, 59, 70, 74, 127 association typology  79–80, 93 Chamic typology  121 Lankan typology  60 typology of colonization  234–235

T Tamil  41, 43, 45, 57–62, 65, 115 temporal anchoring  149, 160 Teochew  206, 209

U Universal Grammar (UG)  2, 10–12, 40, 49 universal processes  268

V vagueness  78 vernacular  173, 206, 210–211, 218–219, 221, 272, 276, 278–279 Vietic  120 Vietnamese  71, 84, 87–89, 106, 112–113, 115–116, 120, 124, 126, 133 W West Africa  206, 230, 232, 239, 248, 250–25, 256, 260, 268 West African culture  188, 231, 245–255 West African languages  9, 12, 27, 89–90, 110, 259 White Creoles  220 word-formation  22–26, 34 Y Yoruba  33, 71, 84, 87, 89–90, 105, 268 Z zero-derivation  22, 26–27

Typological Studies in Language A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 73 Ansaldo, Umberto, Stephen Matthews and Lisa Lim (eds.): Deconstructing Creole. 2007. xi, 290 pp. 72 Næss, Åshild: Prototypical Transitivity. ix, 231 pp. + index. Expected July 2007 71 Nedjalkov, Vladimir P. (ed.): Reciprocal Constructions. With the assistance of Emma Š. Geniušienė and Zlatka Guentchéva. xxviii, 2083 pp. + index (4 vols.). Expected July 2007 70 Zúñiga, Fernando: Deixis and Alignment. Inverse systems in indigenous languages of the Americas. 2006. xii, 309 pp. 69 Aranovich, Raúl (ed.): Split Auxiliary Systems. A cross-linguistic perspective. 2007. vii, 277 pp. 68 Abraham, Werner and Larisa Leisiö (eds.): Passivization and Typology. Form and function. 2006. x, 553 pp. 67 Veselinova, Ljuba N.: Suppletion in Verb Paradigms. Bits and pieces of the puzzle. 2006. xviii, 236 pp. 66 Hickmann, Maya and Stéphane Robert (eds.): Space in Languages. Linguistic Systems and Cognitive Categories. 2006. x, 362 pp. 65 Tsunoda, Tasaku and Taro Kageyama (eds.): Voice and Grammatical Relations. In Honor of Masayoshi Shibatani. 2006. xviii, 342 pp. 64 Voeltz, F. K. Erhard (ed.): Studies in African Linguistic Typology. 2006. xiv, 426 pp. 63 Filimonova, Elena (ed.): Clusivity. Typology and case studies of the inclusive–exclusive distinction. 2005. xii, 436 pp. 62 Couper-Kuhlen, Elizabeth and Cecilia E. Ford (eds.): Sound Patterns in Interaction. Crosslinguistic studies from conversation. 2004. viii, 406 pp. 61 Bhaskararao, Peri and Karumuri Venkata Subbarao (eds.): Non-nominative Subjects. Volume 2. 2004. xii, 319 pp. 60 Bhaskararao, Peri and Karumuri Venkata Subbarao (eds.): Non-nominative Subjects. Volume 1. 2004. xii, 325 pp. 59 Fischer, Olga, Muriel Norde and Harry Perridon (eds.): Up and down the Cline – The Nature of Grammaticalization. 2004. viii, 406 pp. 58 Haspelmath, Martin (ed.): Coordinating Constructions. 2004. xcv, 578 pp. 57 Mattissen, Johanna: Dependent-Head Synthesis in Nivkh. A contribution to a typology of polysynthesis. 2003. x, 350 pp. 56 Shay, Erin and Uwe Seibert (eds.): Motion, Direction and Location in Languages. In honor of Zygmunt Frajzyngier. 2003. xvi, 305 pp. 55 Frajzyngier, Zygmunt and Erin Shay: Explaining Language Structure through Systems Interaction. 2003. xviii, 309 pp. 54 Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.): Studies in Evidentiality. 2003. xiv, 349 pp. 53 Givón, T. and Bertram F. Malle (eds.): The Evolution of Language out of Pre-language. 2002. x, 394 pp. 52 Güldemann, Tom and Manfred von Roncador (eds.): Reported Discourse. A meeting ground for different linguistic domains. 2002. xii, 425 pp. 51 Newman, John (ed.): The Linguistics of Sitting, Standing and Lying. 2002. xii, 409 pp. 50 Feigenbaum, Susanne and Dennis Kurzon (eds.): Prepositions in their Syntactic, Semantic and Pragmatic Context. 2002. vi, 304 pp. 49 Wischer, Ilse and Gabriele Diewald (eds.): New Reflections on Grammaticalization. 2002. xiv, 437 pp. 48 Shibatani, Masayoshi (ed.): The Grammar of Causation and Interpersonal Manipulation. 2002. xviii, 551 pp. 47 Baron, Irène, Michael Herslund and Finn Sørensen (eds.): Dimensions of Possession. 2001. vi, 337 pp. 46 Aikhenvald, Alexandra Y., R.M.W. Dixon and Masayuki Onishi (eds.): Non-canonical Marking of Subjects and Objects. 2001. xii, 364 pp. 45 Bybee, Joan and Paul J. Hopper (eds.): Frequency and the Emergence of Linguistic Structure. 2001. vii, 492 pp. 44 Voeltz, F. K. Erhard and Christa Kilian-Hatz (eds.): Ideophones. 2001. x, 436 pp. 43 Gildea, Spike (ed.): Reconstructing Grammar. Comparative Linguistics and Grammaticalization. 2000. xiv, 269 pp. 42 Diessel, Holger: Demonstratives. Form, function and grammaticalization. 1999. xii, 205 pp.

41 Frajzyngier, Zygmunt and Traci S. Curl (eds.): Reciprocals. Forms and functions. Volume 2. 2000. xii, 201 pp. 40 Frajzyngier, Zygmunt and Traci S. Curl (eds.): Reflexives. Forms and functions. Volume 1. 2000. xiv, 286 pp. 39 Payne, Doris L. and Immanuel Barshi (eds.): External Possession. 1999. ix, 573 pp. 38 Siewierska, Anna and Jae Jung Song (eds.): Case, Typology and Grammar. In honor of Barry J. Blake. 1998. 395 pp. 37 Giacalone-Ramat, Anna and Paul J. Hopper (eds.): The Limits of Grammaticalization. 1998. vi, 307 pp. 36 Newman, John (ed.): The Linguistics of Giving. 1998. xv, 373 pp. 35 Givón, T. (ed.): Grammatical Relations. A functionalist perspective. 1997. viii, 350 pp. 34 Givón, T. (ed.): Conversation. Cognitive, communicative and social perspectives. 1997. viii, 302 pp. 33 Fox, Barbara A. (ed.): Studies in Anaphora. 1996. xii, 518 pp. 32 Bybee, Joan and Suzanne Fleischman (eds.): Modality in Grammar and Discourse. 1995. viii, 575 pp. 31 Gernsbacher, Morton Ann and T. Givón (eds.): Coherence in Spontaneous Text. 1995. x, 267 pp. 30 Downing, Pamela A. and Michael Noonan (eds.): Word Order in Discourse. 1995. x, 595 pp. 29 Kahrel, Peter (PJK) and René van den Berg (eds.): Typological Studies in Negation. 1994. x, 385 pp. 28 Givón, T. (ed.): Voice and Inversion. 1994. viii, 402 pp. 27 Fox, Barbara A. and Paul J. Hopper (eds.): Voice: Form and Function. 1994. xiii, 377 pp. 26 Lord, Carol: Historical Change in Serial Verb Constructions. 1993. x, 273 pp. 25 Svorou, Soteria: The Grammar of Space. 1994. xiv, 290 pp. 24 Perkins, Revere D.: Deixis, Grammar, and Culture. 1992. x, 245 pp. 23 Kemmer, Suzanne: The Middle Voice. 1993. xii, 300 pp. 22 Payne, Doris L. (ed.): Pragmatics of Word Order Flexibility. 1992. viii, 320 pp. 21 Downing, Pamela A., Susan D. Lima and Michael Noonan (eds.): The Linguistics of Literacy. 1992. xx, 334 pp. 20 Croft, William, Suzanne Kemmer and Keith Denning (eds.): Studies in Typology and Diachrony. Papers presented to Joseph H. Greenberg on his 75th birthday. 1990. xxxiv, 243 pp. 19:2 Traugott, Elizabeth and Bernd Heine (eds.): Approaches to Grammaticalization. Volume II. Types of grammatical markers. 1991. xii, 558 pp. 19:1 Traugott, Elizabeth and Bernd Heine (eds.): Approaches to Grammaticalization. Volume I. Theoretical and methodological issues. 1991. xii, 360 pp. 18 Haiman, John and Sandra A. Thompson (eds.): Clause Combining in Grammar and Discourse. 1988. xiii, 428 pp. 17 Hammond, Michael, Edith Moravcsik and Jessica Wirth (eds.): Studies in Syntactic Typology. 1988. xiv, 380 pp. 16 Shibatani, Masayoshi (ed.): Passive and Voice. 1988. xi, 706 pp. 15 Austin, Peter (ed.): Complex Sentence Constructions in Australian Languages. 1988. vii, 289 pp. 14 Hinds, John, Shoichi Iwasaki and Senko K. Maynard (eds.): Perspectives on Topicalization. The case of Japanese WA. 1987. xi, 307 pp. 13 Never published. 12 Nedjalkov, Vladimir P. (ed.): Typology of Resultative Constructions. Translated from the original Russian edition (1983). Translation edited by Bernard Comrie. 1988. xx, 573 pp. 11 Tomlin, Russell S.: Coherence and Grounding in Discourse. Outcome of a Symposium, Eugene, Oregon, June 1984. 1987. viii, 512 pp. 10 Ransom, Evelyn N.: Complementation: its Meaning and Forms. 1986. xii, 226 pp. 9 Bybee, Joan: Morphology. A Study of the Relation between Meaning and Form. 1985. xii, 235 pp. 8 Slobin, Dan I. and Karl Zimmer (eds.): Studies in Turkish Linguistics. 1986. vi, 294 pp. 7 Craig, Colette G. (ed.): Noun Classes and Categorization. Proceedings of a symposium on categorization and noun classification, Eugene, Oregon, October 1983. 1986. vii, 481 pp. 6 Haiman, John (ed.): Iconicity in Syntax. Proceedings of a symposium on iconicity in syntax, Stanford, June 24–26, 1983. 1985. vi, 402 pp. 5 Rutherford, William E. (ed.): Language Universals and Second Language Acquisition. 1984. ix, 264 pp.