The Handbook of Pidgin and Creole Studies (Blackwell Handbooks in Linguistics)

The Handbook of Pidgin and Creole Studies

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Blackwell Handbooks in Linguistics This outstanding multi-volume series covers all the major subdisciplines within linguistics today and, when complete, will offer a comprehensive survey of linguistics as a whole. Already published: The Handbook of Child Language Edited by Paul Fletcher and Brian MacWhinney The Handbook of Phonological Theory Edited by John A. Goldsmith The Handbook of Contemporary Semantic Theory Edited by Shalom Lappin The Handbook of Sociolinguistics Edited by Florian Coulmas The Handbook of Phonetic Sciences Edited by William J. Hardcastle and John Laver The Handbook of Morphology Edited by Andrew Spencer and Arnold Zwicky The Handbook of Japanese Linguistics Edited by Natsuko Tsujimura The Handbook of Linguistics Edited by Mark Aronoff and Janie Rees-Miller The Handbook of Contemporary Syntactic Theory Edited by Mark Baltin and Chris Collins The Handbook of Discourse Analysis Edited by Deborah Schiffrin, Deborah Tannen, and Heidi E. Hamilton The Handbook of Language Variation and Change Edited by J. K. Chambers, Peter Trudgill, and Natalie Schilling-Estes The Handbook of Historical Linguistics Edited by Brian D. Joseph and Richard D. Janda The Handbook of Language and Gender Edited by Janet Holmes and Miriam Meyerhoff The Handbook of Second Language Acquisition Edited by Catherine J. Doughty and Michael H. Long The Handbook of Bilingualism Edited by Tej K. Bhatia and William C. Ritchie The Handbook of Pragmatics Edited by Laurence R. Horn and Gregory Ward The Handbook of Applied Linguistics Edited by Alan Davies and Catherine Elder The Handbook of Speech Perception Edited by David B. Pisoni and Robert E. Remez The Blackwell Companion to Syntax, Volumes I–V Edited by Martin Everaert and Henk van Riemsdijk The Handbook of the History of English Edited by Ans van Kemenade and Bettelou Los The Handbook of English Linguistics Edited by Bas Aarts and April McMahon The Handbook of World Englishes Edited by Braj B. Kachru; Yamuna Kachru, and Cecil L. Nelson The Handbook of Educational Linguistics Edited by Bernard Spolsky and Francis M. Hult The Handbook of Clinical Linguistics Edited by Martin J. Ball, Michael R. Perkins, Nicole Müller, and Sara Howard

The Handbook of Pidgin and Creole Studies Edited by

Silvia Kouwenberg and John Victor Singler

A John Wiley & Sons, Ltd., Publication

This edition first published 2008 © 2008 Blackwell Publishing Ltd except for editorial material and organization © 2008 Silvia Kouwenberg and John Victor Singler Blackwell Publishing was acquired by John Wiley & Sons in February 2007. Blackwell’s publishing program has been merged with Wiley’s global Scientific, Technical, and Medical business to form Wiley-Blackwell. Registered Office John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom Editorial Offices 350 Main Street, Malden, MA 02148-5020, USA 9600 Garsington Road, Oxford, OX4 2DQ, UK The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK For details of our global editorial offices, for customer services, and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell. The right of Silvia Kouwenberg and John Victor Singler to be identified as the authors of the editorial material in this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Library of Congress Cataloging-in-Publication Data The handbook of Pidgin and Creole studies / edited by Silvia Kouwenberg & John Victor Singler. p. cm. — (Blackwell handbooks in linguistics) Includes bibliographical references and index. ISBN 978-0-631-22902-5 (hardcover : alk. paper) 1. Pidgin languages. 2. Creole dialects. I. Kouwenberg, Silvia, 1960– II. Singler, John Victor. PM7802.H36 2008 417′.22—dc22 2008010423 A catalogue record for this book is available from the British Library. Set in 10/12 point Palatino by Graphicraft Limited, Hong Kong Printed in Singapore by Fabulous Printers Pte Ltd 1 2008

Contents

Notes on Contributors List of Abbreviations 1 Introduction Silvia Kouwenberg and John Victor Singler Part I Properties of Pidgins and Creoles 2 Atlantic Creole Syntax Donald Winford 3 Forging Pacific Pidgin and Creole Syntax: Substrate, Discourse, and Inherent Variability Miriam Meyerhoff 4 Pidgin and Creole Morphology Terry Crowley 5 Creole Phonology Norval S. H. Smith 6 Pidgins versus Creoles and Pidgincreoles Peter Bakker 7 Non-Indo-European Pidgins and Creoles Kees Versteegh

vii xiv 1

17 19

48 74 98 130 158

Part II Perspectives on Pidgin/Creole Genesis

187

8 Pidgins/Creoles, and Second Language Acquisition Jeff Siegel 9 Creole Genesis: The Impact of the Language Bioprogram Hypothesis Tonjes Veenstra 10 Pidgins/Creoles and Historical Linguistics Sarah G. Thomason

189

219 242

vi Contents 11 12 13 14 15

Pidgins/Creoles and Contact Languages: An Overview Rajend Mesthrie Creole Studies and Multilingualism Pieter Muysken A Demographic Perspective on Creole Formation Jacques Arends The Sociohistorical Context of Creole Genesis John Victor Singler The Cultural in Pidgin Genesis Christine Jourdan

Part III 16 17 18 19

Grammaticalization in Pidgins and Creoles Adrienne Bruyn Creoles, Markedness, and Default Settings: An Appraisal Alain Kihm Semantic Evidence in Pidgin and Creole Genesis George Huttar Pidgins, Creoles, and Variation Peter L. Patrick

Part IV 20

21 22

Pidgins/Creoles and Linguistic Explanation

Pidgins/Creoles and Kindred Languages

The Case of Signed Languages in the Context of Pidgin and Creole Studies Judy Kegl Pidgins/Creoles and African American English Arthur K. Spears Spanish-Based Creoles in the Caribbean John M. Lipski

263 287 309 332 359 383 385 411 440 461 489 491 512 543

Part V Pidgins/Creoles in Society

565

23

567

24 25 26

Pidgins/Creoles and Discourse Geneviève Escure Pidgins/Creoles and Education Dennis Craig Language Planning in Pidgins and Creoles Hubert Devonish Literary Representations of Creole Languages: Cross-Linguistic Perspectives from the Caribbean Hélène Buzelin and Lise Winer

Author Index Language Index Subject Index

593 615

637

666 677 682

Contributors

Jacques Arends At the time of his death in 2005, Jacques Arends lectured at the University of Amsterdam. His main interests were the historical development of the Suriname creoles and the history and structure of Lingua Franca. His publications tracked syntactic change in Sranan, explored the demographic context of creole genesis in Suriname (e.g., The Early Stages of Creolization, Benjamins, 1995), and made early texts available to the community (e.g., Early Suriname Creole Documents, with Matthias Perl; Vervuert, 1994). He co-edited the Atlas of the Languages of Suriname (with Eithne Carlin; KITLV, 2002). An obituary and publication list appeared in the Journal of Pidgin and Creole Languages (by Adrienne Bruyn, 2007). Peter Bakker gained his PhD from the University of Amsterdam in 1992 and is a specialist in contact languages at Aarhus University, Denmark. He is especially interested in pidgins and in mixed and intertwined languages such as Michif, and languages that underwent significant contact-induced change such as Romani. Some publications which demonstrate this interest are A Language of Our Own: The Genesis of Michif, the Mixed Cree-French Language of the Canadian Métis (Oxford University Press, 1997), The Mixed Language Debate (co-edited with Yaron Matras; Mouton De Gruyter, 2003), and Bibliography of Modern Romani Linguistics (with Yaron Matras; Benjamins 2003). [email protected] Adrienne Bruyn currently participates in a project on the impact of contact on standard Dutch as it emerged between 1400 and 1700 (Radboud University, Nijmegen). She is also involved in the Suriname Creole Archive (SUCA), digitalizing eighteenth-century Sranan and Saramaccan materials. Her research on creoles focuses on historical developments (e.g., Grammaticalization in Creoles: The Development of Determiners and Relative Clauses in Sranan, IFOTT, 1995; St Kitts and the Atlantic Creoles: The Texts of Samuel Augustus Mathews in Perspective, University of Westminster Press, 1999, co-edited with Philip Baker).

viii

Contributors

She was president of the Society for Pidgin and Creole Linguistics in 2003– 2004. [email protected] Hélène Buzelin teaches translation practice and theory at the Université de Montréal. She specializes in the sociology and ethnography of translation. She is the author of Sur le terrain de la traduction: parcours traductologique au coeur du roman de Samuel Selvon, The Lonely Londoners (Éditions du Gref, 2005), has published in literary and translation studies journals such as Canadian Literature, JWIL, Target, Meta, TTR, The Translator, and contributed to a volume titled Constructing a Sociology of Translation (John Benjamins, 2007). She is also coeditor, with Deborah Folaron, of a special issue of Meta on Translation and Network Studies (2007). [email protected] Dennis Craig The late Dennis Craig retired from his last academic position (Vice-Chancellor, University of Guyana) in 1995, after a career spanning several decades at the University of the West Indies (Mona), which had made him Personal Professor of Language Education at that institution. He continued to be of service to educators in the West Indies as co-editor of the Journal of Education and Development in the Caribbean and as director of his institute, Education & Research Associates Ltd., up to his death in 2004. A tribute to Professor Craig and a partial list of his publications can be found at www.sclonline.net/drcraig.html. Terry Crowley At the time of his death in January 2005, Terry Crowley was Professor in Linguistics at the University of Waikato (New Zealand), having taught linguistics previously at the University of the South Pacific and the University of Papua New Guinea. His research interests and numerous publications centered around anglophone Pacific pidgins and creoles, as well as Oceanic languages. An obituary including an extensive biography and publication list by John Lynch appeared in The Journal of Oceanic Linguistics 44, 223–41 (2005). Hubert Devonish is a Guyanese linguist who is Professor of Linguistics at the University of the West Indies (Mona, Jamaica). His research interests cover the range from sociolinguistics and language variation, through language planning and language education policy in creole language situations, to the description of Caribbean creole languages, particularly their phonological and suprasegmental systems. Two works which demonstrate some of these interests are Language and Liberation: Creole Language Politics in the Caribbean (Karia Press, 1986) and Talking Rhythm, Stressing Tone: Prominence in Anglo-West African Creole Languages (Arawak Press, 2002). [email protected] Geneviève Escure is Professor of Linguistics in the English Department at the University of Minnesota (Minneapolis, USA). Her areas of research include language variation, language contact, gender and language, and endangered

Contributors

ix

languages. She has worked extensively with the English-based creoles spoken in Belize and Honduras. She has also documented oral variation in Chinese. Her current work involves the documentation of Garifuna, a mixed language now endangered in Belize and Honduras. Her publications include Creole and Dialect Continua: Standard Acquisition Processes in Belize and China (Benjamins, 1997) and Creoles, Contact and Language Change: Linguistic and Social Implications (Benjamins, 2004). [email protected] George Huttar began fieldwork in Suriname with Mary Huttar in 1968, focusing on Ndyuka. He has also conducted fieldwork on Aboriginal languages in Australia, and has taught linguistics there and in the US and Kenya. His research interests include semantics and cognition as well as creoles, particularly those of Suriname. His publications include Ndyuka (with Mary Huttar; 1994, Routledge) and a series of articles empirically investigating semantic organization of experience as reflected in creole languages. He now serves as an editor for SIL International. [email protected] Christine Jourdan is Professor in the Department of Sociology and Anthropology at Concordia University in Montreal where she teaches anthropology and ethnolinguistics. She has been doing fieldwork in the Solomon Islands since 1981, focusing on Solomon Island Pidgin (e.g., Pijin: A Dictionary of the Pidgin of the Solomon Islands, 2002: Parlons Pijin: Histoire sociolinguistics du pidgin des Iles Salomon, 2007), on urbanization in the Pacific, and on theories of culture and social change (e.g., Language, Culture, and Society, CUP, 2006, co-edited with Kevin Tuite). She has also published in journals such as the Annual Review of Anthropology and Language in Society. [email protected] Judy Kegl received her MA from Brown University and her PhD from MIT and is Professor of Linguistics and Director of the Signed Language Research Laboratory at the University of Southern Maine. She has held research grants from the National Science Foundation and the National Institutes of Health. She is perhaps best known for her discovery and documentation of the emergence of Nicaraguan Sign Language, which she has been studying since 1986. Her publications include The Syntax of American Sign Language (co-authored, MIT Press, 2000) and articles in journals such as Brain and Cognition, Sign Language Studies, and The Journal of Neurolinguistics. [email protected] Alain Kihm (CNRS, Paris) is a specialist in the morphology and syntax of creole languages. He published a description of the Portuguese-based creole of Guinea-Bissau (Kriyol Syntax: The Portuguese-based Creole Language of Guinea Bissau, John Benjamins, 1994) and several articles on Portuguese-based, Frenchbased, and Arabic-based creoles. He is interested in creolization per se and its relations to language change and language emergence (e.g., his chapter in Eckardt, Jäger, and Veenstra (eds.), Variation, Selection, Development: Probing the Evolutionary Model of Language Change; Mouton de Gruyter, 2007). [email protected]

x Contributors Silvia Kouwenberg lectures in linguistics at the University of the West Indies (Mona, Jamaica). Her research interests include substrate-related questions (in particular pertaining to Berbice Dutch and Jamaican), the historical context of creole genesis, and the grammar of creoles. Her publications include A Grammar of Berbice Dutch Creole (Mouton de Gruyter, 1994), Papiamentu (with Eric Murray; Lincom Europe, 1994), “Convergence and explanation in creole genesis” (2001), “L1 transfer and the cut off point for L2 acquisition processes in creole formation” (2006), and Twice as Meaningful: Reduplication in Pidgins, Creoles, and Other Contact Languages (editor, Battlebridge, 2003). [email protected] John M. Lipski is Professor of Spanish at the Pennsylvania State University. His research interests include Spanish phonology, Spanish and Portuguese dialectology and language variation, the linguistic aspects of bilingualism, and the African contribution to Spanish and Portuguese. His books include The Speech of the Negros Congos of Panama (John Benjamins, 1989), El español de América (Cátedra, 1996), A History of Afro-Hispanic Language Contact: 5 Centuries and 5 Continents (Cambridge University Press, 2005); Varieties of Spanish in the United States (Georgetown University Press, forthcoming). He is also editor of the journal Hispanic Linguistics. [email protected] Rajend Mesthrie is Professor of Linguistics in the Department of English at the University of Cape Town. He is currently President of the Linguistics Society of Southern Africa. Amongst his publications are the Concise Encyclopedia of Sociolinguistics (editor, Pergamon, 2001), and English in Language Shift: The History, Structure and Sociolinguistics of South African Indian English (Cambridge University Press, 1992). [email protected] Miriam Meyerhoff is Professor of Sociolinguistics at the University of Edinburgh. She has worked on Pacific creoles (especially Bislama) for more than a decade and has published papers on social and linguistic variation in a number of journals, as well as Constraints on Null Subjects in Bislama (Vanuatu) (2000, Pacific Linguistics), and is editor of the Benjamins Creole Language Library. Recently she has begun work on social and linguistic variation in the creole spoken on Bequia (St. Vincent & the Grenadines). She also works on language and gender, and has a number of publications in that field. [email protected] Pieter Muysken (BA Yale University, PhD University of Amsterdam) is Professor of Linguistics at Radboud University Nijmegen and Director of the Center for Language Studies (CLS) in Nijmegen. His research focuses on language contact, language change, and creole studies. Previously he was on the staff of the Universities of Amsterdam and Leiden, and he has supervised around 35 doctoral dissertations. He is a member of the Netherlands Royal Academy of Sciences and the Max-Planck-Gesellschaft. Muysken is the

Contributors

xi

recipient of the Prince Bernhard Prize, the Prix des Embassadeurs, and the Spinoza prize. He was awarded an Academy Chair of the Netherlands Royal Academy, starting in 2008. [email protected] Peter L. Patrick is Professor of Sociolinguistics at the University of Essex. He is the author of Urban Jamaican Creole: Variation in the Mesolect (Benjamins, 1999) and co-editor, with John Holm, of Comparative Creole Syntax: Parallel Outlines of 18 Creole Grammars (Battlebridge, 2007). Born in New York City and raised in Jamaica, his research focuses on Jamaican Patwa and African American diaspora languages from the perspectives of language variation and change and sociolinguistics. He is concerned with as well linguistic human rights generally. He has also given expert linguistic testimony in US and UK courts, and conducted discourse analysis of clinical interaction, both involving creoles. [email protected] Jeff Siegel is an Adjunct Professor of Linguistics at the University of New England (Australia). He is also an Associate Researcher at the University of Hawai[i, where he was formerly Foundation Director of the Charlene Sato Center for Pidgin, Creole, and Dialect Studies. His main areas of research are the origins of language contact varieties in the Pacific region and the use of such varieties in formal education. Publications include Pidgin Grammar: An Introduction to the Creole Language of Hawai[i (with Kent Sakoda; Bess Press, 2003) and The Emergence of Pidgin and Creole Languages (Oxford University Press, 2007). [email protected] John Victor Singler is Professor of Linguistics at New York University. He holds graduate degrees in African Area Studies (MA from SOAS, University of London) and linguistics (MA and PhD from UCLA). His areas of research within creole studies include the sociocultural issues surrounding creole genesis, tense-aspect, phonology, and West African pidgins and creoles. In addition, he has done research on the history of African American English, on Liberia’s Niger-Congo languages, and on Liberian literature. His publications include An Introduction to Liberian English (African Studies Center, 1981), Pidgin and Creole Tense-Mood-Aspect Systems (edited, John Benjamins, 1990). [email protected] Norval S. H. Smith holds degrees in Humanities (Latin and Greek, University of Glasgow) and in General Linguistics (School of Oriental and African Studies, University of London) and is an Associate Professor, specializing in Phonology and Creole Studies, in the Linguistics Department of the University of Amsterdam. In his creole work he concentrates on the Atlantic creole languages, with a particular interest in the creole languages of Suriname and their substrate languages. Recent publications include Creolization and Contact (co-edited with Tonjes Veenstra; Benjamins, 2001) and “Contact phonology” (in Phonology in Context, Martha C. Pennington (ed.), Palgrave, 2006). [email protected]

xii

Contributors

Arthur K. Spears is Professor and Chair in the Anthropology Department and Director of the Black Studies Program at The City College of The City University of New York, and Professor in the Linguistics and Anthropology Programs at its Graduate Center. He has published extensively on African American English, Romance-lexifier creoles, and language, race, and ideology. His most recent books are Black Linguistics: Language, Society, and Politics in Africa and the Americas (co-editor; Routledge, 2003) and Black Language in the English-Speaking Caribbean and United States: History, Structure, Use, and Education (editor; Lexington/Rowman & Littlefield, 2008). He served as President of the Society for Pidgin and Creole Linguistics 2007–9. [email protected] Sarah G. Thomason is the William J. Gedney Collegiate Professor of Linguistics at the University of Michigan. She is a historical linguist who specializes in contact-induced language change and related topics, including the genesis of pidgins, creoles, and bilingual mixed languages. Among her publications are “Chinook Jargon in areal and historical context” (1983), Language Contact, Creolization, and Genetic Linguistics (co-authored with Terrence Kaufman; University of California Press, 1988/1991), “Language mixture: ordinary processes, extraordinary results” (1995), Language Contact: An Introduction (Georgetown University Press, 2001), and “Language contact and deliberate change” (2007). [email protected] Tonjes Veenstra (PhD University of Amsterdam) is a research fellow at the Centre for General Linguistics, Typology and Universals Research (Berlin). His research interests include syntax, morphology, language acquisition, and language contact and evolution. Within creole linguistics, he has worked and published primarily on English-related creoles (e.g., Serial Verbs in Saramaccan: Predication and Creole Genesis, HAG, 1996), but his interests extend beyond (e.g., “The survival of inflectional morphology in French-related creoles,” with Angelika Becker, SSLA, 2003; Creolization and Contact, co-edited with Norval Smith, Benjamins, 2001; Variation, Selection, Development: Probing the Evolutionary Model of Language Change, co-edited; Mouton de Gruyter, 2007). [email protected] Kees Versteegh is Professor of Arabic and Islam at the University of Nijmegen. He graduated in Classical and Semitic languages and specializes in historical linguistics and the history of linguistics, focusing on processes of language change and language contact. His books include Pidginization and Creolization: The Case of Arabic (Benjamins, 1984), The Arabic Linguistic Tradition (Routledge, 1997), and The Arabic Language (Columbia University Press, 1997). He co-edited the Handbuch für die Geschichte der Sprach- und Kommunikationswissenschaft (Mouton de Gruyter, 2000–5) and is the editorin-chief of the Encyclopedia of Arabic Language and Linguistics (Brill, 2005–). [email protected]

Contributors xiii Lise Winer is Associate Professor in Second Language Education in the Faculty of Education at McGill University in Montreal. She has published extensively on the historical and social development of language in Trinidad & Tobago (a collection of this work appeared as Badjohns, Bhaaji and Banknote Blue: Essays on the Social History of Language in Trinidad & Tobago School of Continuing Studies, UWI St. Augustine, 2007), has re-published new editions of several nineteenth- and early twentieth-century novels from Trinidad (University of the West Indies Press), and has prepared the forthcoming Dictionary of the English/Creole of Trinidad & Tobago (McGill Queens University Press). [email protected] Donald Winford is Professor of Linguistics at the Ohio State University. His teaching and research interests are in creole linguistics, variationist sociolinguistics, contact linguistics, and African American English, and he has published widely in those areas. He is the author of Predication in Caribbean English Creoles (John Benjamins, 1993), and An Introduction to Contact Linguistics (Blackwell, 2003). He served as President of the Society for Caribbean Linguistics from 1998 to 2000, and has been editor of the Journal of Pidgin and Creole Languages since August 2001. [email protected]

List of Abbreviations

abil = ability abl = ablative case acc = accusative case aff = affirmative agr = subject agreement ant = anterior tense asp = aspect aux = auxiliary case = case caus = causative cis = cislocative case (towards the speaker) cl = classifier, (noun) class comp = complementizer compl = completive aspect conn = connective cont = continuative, continuous aspect cop = copula def = definite article dem = demonstrative des = desiderative modality det = determiner, article dim = diminutive dir = direction(al) do = direct object du = dual dur = durative aspect emph = emphatic excl = exclusive (with 1p)

exclam = exclamative f = feminine foc = focus fut = future modality gen = genitive case ger = gerundive hab = habitual aspect id = ideophone imm = immediate imp = imperfective aspect inc = inchoative incl = inclusive (with 1p) ind = indefinite article inf = infinitive infl = inflection inst = instrumental int = intensifier interj = interjection irr = irrealis modality loc = locative, location m = masculine mod = modality neg = negative, negator nom = nominalizer obj = object, objective case obv = obviative case om = object marker p = plural (pronoun) pass = passive morpheme past = past tense

List of Abbreviations xv pf = perfective, perfect aspect pl = plural pm = predicate marker poss = possessive post = postposition pot = potential ppunct = past punctual prep = preposition pres = present tense prob = probable prog = progressive aspect pron = pronoun prox = proximative aspect prt = particle q = question marker real = reality marker recip = reciprocal red = reduplicated refl = reflexive

rel = relative pronoun, relative clause introducer s = singular (pronoun) sd = sudden discovery sg = singular si = subject index sm = subject marker srp = subject referencing pronoun sub = subject, subjective case subo = subordination marker svc = serial verb construction tma = tense/mood/aspect tns = tense top = topic tr = transitive vb = verbalizer voc = vocative case 1 = first person 2 = second person 3 = third person

Introduction 1

1 Introduction SILVIA KOUWENBERG AND JOHN VICTOR SINGLER

1 Background That there is a reason – and a need – for The Handbook of Pidgin and Creole Studies in linguistics today comes from the recognition of the challenges that pidgin and creole languages pose for linguistics as a field. At the same time, an assessment of the state of pidgin and creole studies is timely for those who work within it. The questions that drive pidgin and creole studies today revolve around the nature and interaction of the forces that have shaped these languages. These questions are not new; they are what motivated the study of these languages in the first place. However, in recent years insights from other branches of linguistics as well as careful sociohistorical studies have moved the field forward. Creole studies has attained this point within linguistics, it can be argued, as a logical progression from three events that occurred half a century ago: • the publication in 1957 of Chomsky’s Syntactic Structures, • the publication four years earlier of Weinreich’s Languages in Contact, and • the convening in 1959 of the first international conference on creole studies at Mona, Jamaica.

1.1

The Chomskyan paradigm

The publication of Syntactic Structures – a defining moment in the history of modern linguistics – has, additionally, special relevance for creole studies. The Chomskyan paradigm emphasized the interaction of language and the mind. The structure of the brain for language provides the basis for universals – the properties that characterize and, indeed, define all human language. Further, linguists working within the Chomskyan paradigm sought to understand the nature of the acquisition of language, the way in which the intersection of universal principles with language-particular input resulted in children’s acquisition of language.

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

2 Silvia Kouwenberg and John Victor Singler Given this interest in the nature of acquisition, creoles posed a challenge. The best-known creoles had the lexicons of western European languages like English and French yet were most definitely not English or French; this raised intriguing questions about what had happened. Whatever the explanation, this was clearly not acquisition as usual. A further point about the intersection of generative linguistics with pidgin and creole studies is that, while definitions of pidgin and creole vary (and we address this below), a widely held view of the difference between the two was that a pidgin was no one’s native language, i.e., had no native speakers, while a creole was a nativized pidgin. Given the Chomskyan interest in acquisition, specifically first language (L1) acquisition, far more attention was paid to creoles than to pidgins.

1.2

Contact linguistics

After the achievements of historical linguists in the nineteenth and early twentieth centuries in working out genetic relationships among languages (as expressed by family trees), Uriel Weinreich’s Languages in Contact (1953) laid the groundwork for the study of the ways in which contact between languages has the power to change them. Despite its title, the book’s focus is overwhelmingly on bilingualism. Still, it moved linguists to consider external sources for change and has more recently given birth to the field of contact linguistics, which includes bilingualism but is not confined to it. In that sense, the book’s publication can fairly be said to have launched the field of contact linguistics. From the writings of Hugo Schuchardt (1882–8) onward, creole languages had been shown to be problematic for the Stammbaum model, i.e., the idea that each language was a direct descendant of some other language (in the way, for example, that Romance languages are seen as descending from Latin). Pidgins and creoles have more than one source language; indeed, Whinnom (1971) was later to suggest that pidginization (and creolization) could only occur when three or more languages were involved. Thus, pidgins and creoles are contact languages par excellence: by definition, a pidgin or creole cannot come into existence in a monolingual context. At first, attention to contact phenomena and attention to pidgins and creoles were seen as parallel yet distinct enterprises. In recent times a growing number of scholars have seen the two as part of the same field of inquiry. An illustration of this comes in recent books about contact linguistics by creolists (Thomason 2001, Winford 2003) and also books with a more narrow focus within creole studies that place contact linguistics in their titles (Migge 2003, Holm 2004).

1.3

The study of creole languages

The 1959 Mona Conference was the first to assemble those who studied creole languages, in particular – but not exclusively – those of the Caribbean. Certainly

Introduction 3 there had been work by individual scholars prior to this (Sylvain 1936 on Haitian Creole, Hall 1943 on Tok Pisin and Hall 1953 on Haitian Creole, and Taylor 1951 on Caribbean creoles among the most prominent), but this was the first time that scholars who worked on pidgins and creoles had convened in this way. In the years immediately following, with the emergence of a generation of scholars from the Caribbean, linguists more generally began to recognize creole languages as an appropriate object of intellectual interest. A part of the legacy of slavery in the Caribbean and elsewhere has been the stigmatization of the languages associated with slaves, i.e., creoles. Popular attitudes against pidgins and creoles were reflected in academic settings as well. Thus, up until this period linguists’ willingness to apply the concept of linguistic relativism – whereby every language is understood to be complete and valid – may have been extended to Hopi and Hausa, but it generally stopped short of being extended to pidgins and creoles. The value of the 1959 Mona Conference, then, was its scholarly attention to these languages, its recognition of them as speech systems. (See the conference papers in Le Page 1961.) The first conference at Mona was followed by a second one there nine years later. The growth in the field of creole studies in the years between the two conferences is reflected in the expanded format of the second conference and, especially, in the book that emerged from the conference, The Pidginization and Creolization of Languages, edited by Dell Hymes and published by Cambridge University Press in 1971. More than any other, this was the book that brought pidgins and creoles to the wider attention of linguists. The three events we have singled out contributed three strands to the study of pidgins and creoles that was to follow. The Mona conferences introduced a comparative component to creole studies and raised the question of how theories of language relate to pidgins and creoles, generativism asked how we as linguists relate what we know about acquisition to the emergence of creoles, and the Weinreich tradition asked similar questions from the perspective of language contact.

2 The Growth of a Field Subsequent developments within linguistics in the 1960s and 1970s were to inform the basic inquiry into pidgin and creole languages. Linguists as far back as Addison Van Name (1869–70) had posited a causal link between the social circumstances that produced creole languages and the particular properties that creole languages had in common. Now the emergence of the field of sociolinguistics (particularly variationist sociolinguistics, as led by William Labov) legitimized creolists’ attention to the link between social forces and their linguistic consequences. The observation within linguistics of a connection between Caribbean creoles and African languages dated at least as far back as Hugo Schuchardt (1882–8)

4 Silvia Kouwenberg and John Victor Singler (see Meijer & Muysken 1977). Beginning in the 1960s, much more research was being carried out on African languages. As a consequence, those interested in creoles with African substrates now had access to grammatical information about relevant African languages. The bonds between African languages and Caribbean creoles featured prominently in subsequent scholarship, particularly that of the new generation of Caribbean linguists (e.g., Alleyne 1980). Within creole studies itself, although most of the attention thus far had been devoted to Caribbean creoles and secondarily to Pacific pidgins, attention was now extended more generally to creoles and pidgins elsewhere, e.g., the Indian Ocean, West Africa, and the Pacific Northwest of North America.

2.1

Universals vs. Substrata

Across the history of the study of pidgin and creole languages – from scholars like Van Name, Schuchardt, and Hesseling (see Holm 1988) through the events we have referred to from the 1950s to the present day – the overriding question has been that of creole genesis. How do creoles come into being? What is to explain their character? Or, phrased another way, to the extent that a creole is distinct from the language from which it draws its lexicon (the lexifier language), what is the source of those differences? By the early 1980s, how to account for creole genesis had become the only question in creole studies. A fundamental difference in opinion pitted “universalists” and “substratists” against each other. The primary author of the universalist position was Derek Bickerton. In a series of works culminating in Roots of Language (1981) and “The Language Bioprogram Hypothesis” (1984), Bickerton argued that creole languages result when children, forced to acquire a first language on the basis of unprocessable input, create language by the application of hard-wired linguistic universals. He claims that this is the situation that arises in the multilingual context of plantation societies where the common language is a macaronic, defective pidgin. For many other creolists, the scenario for creole genesis that the Language Bioprogram Hypothesis requires is crucially at odds with established facts about the history of the colonies where the creoles arose. And various creolists presented detailed accounts (not all of them, to be sure, fully rigorous) that directly linked African language phenomena to creoles. The relevance of the African data was this: if the people who created Caribbean creoles originally were themselves either slaves brought from Africa or their locally born children, then their languages (the substrate languages) were likely to have played a role in shaping the resultant creole. No creolist denied the relevance of linguistic universals to creole genesis; indeed, we are hard pressed to understand what that would entail. Rather, the divergence of opinion lay in the strength of the substrate contribution to creole genesis, ranging from outright denial of its role (Bickerton 1986) to the assertion

Introduction 5 that in some cases at least a creole language is a relexified version of a substrate language (Lefebvre 1986, 1998, 2004). Despite the attention that Bickerton’s position attracted and the appeal of his work outside creole studies (see Veenstra, this volume), the anti-substratism of his view isolated him. When “The Language Bioprogram Hypothesis” appeared in Brain and Behavioral Sciences (1984), it was accompanied by commentaries from a number of creolists who voiced their skepticism about so absolute a universalist hypothesis. At the 1985 Amsterdam conference on “Substrata versus Universals in Creole Genesis,” most participants espoused positions which incorporated substrate contributions (see the papers of the conference in the volume of the same name edited by Muysken & Smith 1986).

2.2

Pidgin and creole studies today

Two decades after the publication of Substrata versus Universals, the debate continues. Virtually no one within creole studies denies a role either to the substrate or to (first) language acquisition. Rather, the questions that engage the field today involve the nature of the interaction of substrate, lexifier, and universal forces. During these two decades there has been far more attention to the cultural matrix (Alleyne 1971) of creole genesis; to the demography of the setting in which creole languages arose; to establishing links between individual creoles and the specific African languages that would have contributed to them as well as the variety/varieties of the lexifier language that would have been present; and to the social setting in which creole languages emerged. At the same time, creolists have increasingly framed their understanding of creoles in current theories of language, including theories of language change and language contact. Insights from the study of pidgins – in particular those of the Melanesian archipelago – have likewise contributed to the field, raising important questions about the relationship between morphosyntactic expansion and nativization as well as the role of substrate languages. Finally, there has been increased attention to the language-particular properties of individual pidgins and creoles.

3 The Organization of the Handbook This handbook sets out to represent the state of the art in creole studies by reflecting what creolists have learned, what the topics are that they are wrestling with, how pidgin and creole languages bear on larger questions about the nature of language, and how different developments in linguistics outside of the field bear on it. In preparing this volume, we have not sought to provide a catalogue of all pidgin and creole languages in the world. We refer the reader in search of this to Holm (1989) and Smith (1995).

6 Silvia Kouwenberg and John Victor Singler Several chapters in Part I of the book, “Properties of Pidgins and Creoles,” tackle the formal characteristics of pidgins and creoles, while others survey the inclusiveness of these labels. Most of what gets written about pidgins and creoles is about syntax. To do justice to the breadth of the syntax literature, we have included two chapters, Winford’s on Atlantic creole syntax and Meyerhoff’s on Pacific pidgin and creole syntax, alongside Crowley’s chapter on morphology and Smith’s on phonology. Within creole studies, the relationship of pidgins to creoles remains vexed. Bakker’s chapter aims to define different categories of language within the pidgin-creole ambit. While the preponderance of research on pidgins and creoles is devoted to varieties whose lexifier language is a Western European one and which emerged in colonial contexts, Versteegh’s chapter examines pidgins and creoles whose lexifier is non-Indo-European, and which arose in a wide array of contexts. It gives a feel of the immense variety awaiting further research efforts, while also pointing to common patterns of development. Part II, “Perspectives on Pidgin and Creole Genesis,” contains chapters that approach the central question in the field, that of genesis, from a diverse range of perspectives. These begin with second language acquisition (Siegel) and first language acquisition/the Language Bioprogram Hypothesis (Veenstra). Thomason presents the perspective of historical linguistics, Mesthrie that of contact linguistics generally, and Muysken that of bi- and multi-lingualism in particular. Arends, Singler, and Jourdan approach genesis from related fields – demography, social history, and anthropology, respectively. “Pidgins/Creoles and Linguistic Explanation,” Part III, considers pidgin and creole phenomena in the context of grammaticalization theory (Bruyn), markedness (Kihm), lexical semantics (Huttar), and variation studies (Patrick). In Part IV, “The Case of Signed Languages in the Context of Pidgin and Creole Studies,” Kegl explores the emergence of signed languages in relation to creolization, while Spears and Lipski assess African American Vernacular English and Bozal Spanish respectively from the point of view of their putatively creole characteristics. In Part V, “Pidgins/Creoles in Society,” Escure argues for the centrality of the study of discourse in these languages. Devonish focuses on the politics of language planning where pidgins/creoles are involved, while Craig considers these languages’ role in education. Buzelin and Winer analyze the different literary traditions involving French- and English-lexifier creoles of the Caribbean.

4

Themes

In the handbook’s coverage of the field of pidgin and creole studies, several themes recur. In the subsections that follow, we present some of them, not as an effort to provide a comprehensive summary of the handbook’s contents but rather to give a broad sense of consensus among many authors as to the nature of pidgin and/or creole languages and their genesis.

Introduction 7

4.1

On monocausal solutions

In addressing the debate regarding pidgin/creole genesis, we chose not to divide and organize the handbook by competing theories. With the exception of Veenstra’s “Creole Genesis: The Impact of the Language Bioprogram Hypothesis,” which assesses the contribution made to the field by Bickerton’s work, there are no chapters devoted exclusively to any one of the genesis theories. Certainly, the genesis question figures in many of the chapters, with respect to both pidgins and creoles (in the present discussion, we use “pidgin” to refer to “extended pidgin” – see section 4.2.1 for discussion of the terminology). Thus, Meyerhoff, in discussing developments in Pacific pidgins and creoles, identifies the semantics of the lexifier, synchronic functions and use in discourse, and speakers’ substrate models as having played mutually reinforcing roles. In her view, while substrate factors are seen as important, they should not necessarily be taken as the starting point for all grammaticalized structures in Pacific pidgins and creoles. Implied in this is the position that no single explanation for pidgin and creole genesis is sufficient. This is very much in line with positions taken by other authors, and we may be justified in postulating broad agreement among many of the contributors on the view that no single mechanism fully accounts for pidgin and creole genesis – indeed, that if the influence of any of these diverse factors were taken out of the equation, we would not expect to see the development of pidgins and creoles. Whenever specific properties of these languages are under scrutiny, authors identify lexifier-related properties, substrate-related properties, and properties that developed independently. This is true, for instance, of Crowley’s discussion of morphology, of Bruyn’s discussion of grammaticalized forms, of Smith’s discussion of marked phonemes, and of Huttar’s discussion of semantics. By and large, the idea of “ordinary processes, extraordinary results” (Thomason 1995) is applicable to the positions taken on pidgin and creole genesis in many of the chapters. Thus, Muysken adopts the Uniformitarian Hypothesis (Labov 1972), which means that the sociolinguistic and psycholinguistic processes operant in the genesis of pidgins and creoles are the same as those operant in contemporary multi-lingual settings. In this vein, Singler discusses the relevance of covert prestige in the emergence of creole languages – an insight gained from modern sociolinguistic research. Nevertheless, the “ordinary processes” that produce pidgin and creole languages need to be better understood. Chapters such as those by Siegel on second language acquisition, Muysken on multi-lingualism, and Patrick on variation address the nature of some of these processes. At the same time, Huttar identifies a glaring contradiction in the way many creolists have treated superstrate-related properties. The focus on structures that appear to be different from those of the respective superstrates implies that structures resembling those of the superstrates do not require explanation; it also implies that creole structures in general derive from superstrate sources

8 Silvia Kouwenberg and John Victor Singler – except for those structures studied because they are different. Recently, there has been a surge in interest in superstrate-related genesis scenarios and hence in superstrate-related properties (DeGraff 2001; Mufwene 2001). In this volume, Bruyn shows that some apparent grammaticalizations in pidgins and creoles are in fact extensions of processes already operative in the lexifier, thus pointing to the validity of “restructuring” as a process involved in pidgin and creole genesis (but she also identifies substrate-related grammaticalizations). Kihm points to parallelism between Romance-lexifier creoles and their lexifiers in the placement of negation and adverbs, contra DeGraff (1997) and Roberts (1999).

4.2 4.2.1

Nativization and terminology The life cycle model and nativization

An assumption that has been widespread in creole studies over the years and indeed dates to Schuchardt (1909, p. 443, cited in Meijer & Muysken 1977, p. 30) is that a creole arises from a pidgin. In this view, known as the “life cycle model,” the pidgin is seen as having emerged as a medium of communication among people who lacked a language in common and becomes a creole upon nativization, undergoing morphosyntactic expansion in the process. The life cycle model sees creoles as crucially different from prior varieties by virtue of their communicative expansion and greater regularization and holds, further, that it is nativization – when the speech variety in question becomes the first language of children born in the community – that effects these changes. The term “pidgin” in this scenario has meant widely divergent things to different people. For Bickerton, the term referred to an irregular, even chaotic, speech variety, inadequate for full communication. For others, any contact variety with few or no native speakers was designated a pidgin, no matter how fully adequate a system of communication it might be and no matter how extensive the grammatical system. Terminology arose to distinguish among non-native varieties, with Mühlhäusler (1997, p. 6) dividing what had hitherto been subsumed under “pidgin “ into “jargon,” “stable pidgin,” and “expanded pidgin.” In his chapter in this volume, Bakker distinguishes among “jargon,” “pidgin,” and “pidgincreole.” (Accordingly, what Bickerton calls a pidgin is what others would call a jargon.) Briefly, where such distinctions are made, there is assumed to be a line of progression in terms of morphosyntactic elaboration along a cline, from jargon initially to creole ultimately. The central position of the Caribbean in the history of pidgin and creole scholarship has had as a consequence that the varieties most studied have all been their speakers’ first languages and have long ago undergone nativization. Hence, the life cycle model and, especially, the role – or non-role – of nativization in creole genesis could not be tested on the basis of Caribbean language data. Rather, it is contact varieties of the Pacific that provide the most apt testing ground for it. In particular, varieties of Melanesian Pidgin are at various stages of nativization, yet the evidence from them generally argues

Introduction 9 against the idea that there is a direct and incontrovertible link between their grammatical elaboration and their nativization. This point is discussed at length in Meyerhoff’s and Bakker’s chapters, but others make similar points: Crowley points to mismatches between native/ non-native status and morphological elaboration, where non-native varieties sometimes display properties one would expect to see in native varieties instead. Bruyn’s discussion of the development of the Melanesian Pidgin future marker baimbai/bai illustrates that innovative developments are not confined to creole varieties but take place as well in extended pidgins. Escure points out that there is little merit in differentiating pidgin discourse from creole discourse. Versteegh points to cases where a speech community contains both first and second language speakers of the same variety. Veenstra, in elaborating the social context in which Hawaiian Creole English was formed, shows that the grammatical features ascribed to nativization in Bickerton’s work on this language emerged first in the non-native speech of a generation of bilinguals (Roberts 2000). Several authors point to extra-linguistic factors which bring about morphosyntactic elaboration. Bakker’s view is that the social trigger for the transformation of a non-native communication system into a full-fledged human language is the formation of a speech community, and that nativization plays only a minor role at most. Kegl, in her discussion of the emergence of signed languages, similarly assigns a pivotal role to the presence of a community of “speakers” – specifically a Deaf community – and argues that a critical mass has to be attained for language emergence to take place. She does, however, point out that the community must include a cohort of “languageready” children and argues that the gestural communication systems which arise outside of such a context do not constitute language. The latter include the gestural communication systems of mixed communities of hearing and deaf where the deaf are a minority, and the gestural system of the first generation of students in Nicaraguan schools for the deaf. In many ways, the views expressed in these and other chapters assign a pivotal role to language use as driving developments of expansion, whether in pidgins or in creoles. Thus, Meyerhoff’s view is that discourse is “both the starting point and the primary medium for the development of syntax.” Jourdan sees pidgins and creoles as the result of negotiation of linguistic meaning in the broader context of negotiation of culture, and identifies social interaction, power, the ideological dimension of culture, and human agency as constituting the “matrix” of pidgin and creole genesis. Siegel sees the adoption of L1 features into the contact variety occurring during a phase where its use is extended into different domains; speakers find themselves challenged to use a variety of which they have acquired too little, specifically to use it in new domains. If we recognize that the traditional distinction between pidgin and creole based on non-native vs. native speaker status is of little consequence in predicting either their social functions or the extent of their grammatical elaboration, the logical consequence is that the terms themselves become less useful as labels for particular kinds of languages. Indeed, where Pacific and West

10 Silvia Kouwenberg and John Victor Singler African pidgins and creoles are discussed, the terms are felt to be an encumbrance, to stand in the way of achieving proper insight in the developments that gave rise to these languages. Versteegh points to this problem when he posits that the continued use of the terms “pidgin” and “creole” with the implications for (non-)native status directs our attention toward what “a language can be” – that is, on issues of prototypicality – and away from the processes of restructuring which give rise to these languages and the potential range of their outcomes. Additionally, the life cycle model is based on assumptions that no longer stand up to scrutiny about the structure of the plantation societies where creole languages emerged; hence, its applicability even in the Caribbean context can be questioned, as argued by Singler. Different authors deal with the terminological problem in different ways. Thus, Jourdan consistently uses the term “pidgins” to refer to both non-native and native varieties – including those of the Caribbean, which are usually thought of as the prototypical “creoles.” In contrast, Meyerhoff and Devonish both prefer to speak of “creoles” for languages that are used for the full range of social functions, irrespective of native/non-native status – including the still largely non-natively spoken Melanesian pidgins. Thomason takes non-native/ native status to distinguish pidgins from creoles – as is traditionally done – but uses the term “pidgin” to refer to fully crystallized languages, excluding rudimentary pidgins from that label, although she allows that a pidgin which remains a secondary language used for quite limited purposes may be limited in lexicon and grammar. Alternative to all this is the viewpoint, associated with the work of authors such as Chaudenson and Mufwene, which holds that creole languages emerge from successive cycles of approximation of their lexifiers, and are their descendants; as pointed out in Singler’s chapter, this viewpoint entails that there is no prior pidginization.

4.2.2

Other terminological issues

In the previous section, we pointed to the difficulties surrounding the different ways in which authors have used “pidgin.” In this section we address other terminological points. Bickerton (1981) classifies creoles as fort creoles, plantation creoles, and maroon creoles. He bases his distinction on the social context in which the pertinent languages emerged. The distinction has become widely accepted, although the question of the formal typological implications remains unresolved. Fort creoles, the creoles which originated around European forts or settlements, often in mixed relationships (local mothers, European fathers), arose in situations that are presumed to have involved no displacement for the “substrate” population (although this claim is probably disputable, at least for some of the West African forts where the designation is commonly applied), whereas plantation creoles and maroon creoles arose in contexts of displacement. The plantation and maroon creoles are distinguished by access to

Introduction 11 superstrate speakers in the plantation context and isolation from the superstrate population in the maroon context. In fact, however, maroon communities subsisted on the fringes of plantation societies and their survival often depended on considerable contact with plantation populations. Some maroon communities did develop linguistic practices different from – although related to – those of the plantation societies, e.g., those in Suriname, Jamaica, and, in one instance, Colombia; however, for most known maroon communities, no special linguistic practices have been documented. Where relevant, the chapters in this book consider plantation creole data alongside maroon creole data; they do so without reflecting on the distinctions. Fort creoles are not considered here. The creolist community has appropriated the terms “superstrate” and “substrate” from historical linguistics but has altered their meanings in the process. For creolists, “superstrate” ordinarily refers to the language of the socially and economically dominant group. In colonial situations, this is usually the language of the colonial power. It typically provides the basis for the lexicon for the emergent pidgin or creole. When it does so, it is also referred to as the “lexifier language.” “Substrate” refers to the first languages of the socially and economically subordinated populations; in plantation situations the speakers of substrate languages usually comprise the labor force. Sometimes, as Bakker points out, there are pidgins whose lexical basis is a language that is not associated with socioeconomic or political power; however, the known instances of this are limited to unexpanded pidgins. Versteegh similarly points out that contact varieties may emerge in situations where the different linguistic groups are equal, but says that such situations are rare. The terms are generally used by the authors in this handbook with reference to situations where the socioeconomic conditions were such that substrate speakers were politically and economically subordinate to superstrate speakers. The term “lexifier” (in place of superstrate) has the advantage that it is devoid of implications with regard to socioeconomic status, but its adoption leaves a gap in that no straightforward alternative term is available for the notion of “substrate.” Apart from the terms substrate and superstrate, the term “adstrate” has had some relevance in the field, designating languages that have either had a peripheral presence in the contexts where pidgins and creoles emerged or came on the scene after pidgin and creole genesis. In either case, it is assumed that the pertinent languages were not in a position to make significant contributions to the grammar. In the Caribbean, indigenous languages – where they survived European onslaught – and late-arriving African and Asian languages are considered adstrates. Since adstrates are not considered to have relevance to pidgin and creole genesis, they do not figure in the present pages beyond passing mention. The “creole continuum” (originally the “post-creole continuum”) is a model posited by DeCamp (1971) to designate the situation that arose in, among

12 Silvia Kouwenberg and John Victor Singler other places, Guyana (Bickerton 1975, Rickford 1987) and Jamaica (Patrick 1999). In both these cases, the continued co-existence of creole and lexifier has resulted in intermediate varieties, such that the term “creole” is ambiguously used to refer to a range of varieties. In order to distinguish within the range, the terms “basilect,” “mesolect,” and “acrolect” are used to designate varieties farthest from, intermediate, and closest to the lexifier, with the possibility of further differentiation within, e.g., the basilect (cf., for example, Singler 1996). Although the competence of speakers within the continuum tends to correlate with the rural–urban divide and with socioeconomic status, continuum variation does not simply translate to geographical or social variation. Most speakers will be competent in more than one variety, and it is possible for individuals to have competence in varieties that are discontinuous on the continuum (Bickerton 1975). In some work by creolists, particularly those dealing with the Caribbean, the terms basilect, mesolect, and acrolect have been extended to designate what might be called a pan-creole continuum, with, for example, Bajan and Trinidadian Creole – because they are deemed closer to English than are most other Caribbean English Creoles – designated acrolectal and mesolectal, respectively. Still within this paradigm, it is sometimes the case that the Surinamese creoles, because they are much further from English than any of the other English-lexifier languages of the Caribbean, are given a designation like “radical” or “conservative.” In his chapter on Atlantic syntax, Winford designates them “radical.” Arends discusses some of the demographic factors which may have contributed to the status of creoles on this continuum. The terms “creolization” and “pidginization” have been subject to a range of definitions. As a general rule, pidginization is identified with simplification, and creolization with expansion, as in the following observation by Hymes (1971a, p. 84): “Pidginization is usually associated with simplification in outer form, creolization with complication in outer form.” For those who define a creole language as a nativized pidgin, the term creolization is often simply identified with nativization. As a result, except when explicitly defined, the domain of “creolization” is ambiguous. A further point in this regard is that, while creolists may tend to have positions regarding what they consider to be the most important contributing factors in pidginization and creolization (such as the substrate, universals of second language acquisition, and the like), they rarely have a clear position on the actual mechanisms involved in these processes. Exceptions to this statement include Bickerton’s Language Bioprogram Hypothesis (1984 and elsewhere) and Lefebvre’s Relexification Hypothesis (1998 and elsewhere), each of which postulates very specific mechanisms by which features enter creole languages. Both proposals have been subject to the criticism that no single mechanism is able to account for the full range of results, and both have required their authors to put forth auxiliary hypotheses to make up for their lack of explanatory power, thus coming much closer to compromise positions than their authors allow for.

Introduction 13

4.3

Pidgin/creole typology

As Mesthrie points out, from a synchronic point of view, creoles are “normal” languages (see Muysken 1988 and Singler, this volume, for the same point). It is only for certain linguistic inquiries, with either a technical historical focus or a typological focus, that their status as contact languages is at issue. Thus, Thomason, who approaches the subject from the point of view of historical linguistics, considers pidgins and creoles of interest as a class of languages because of their historically “mixed” nature. Nevertheless, the idea that we can identify certain characteristics with pidgin and creole languages has endured – although many authors now correlate these characteristics with social expansion rather than with nativization – alongside an increasing awareness that there is a great deal of structural diversity across pidgins and creoles. For instance, Bakker, who argues that distinctions between pidgins and creoles have primarily a social basis, still recognizes that the social categories appear to correlate with structural features. Winford identifies structures which are shared across many – if not all – Atlantic Creoles in the areas of word order, tense-aspect marking, movement processes (in passivization and focus constructions, etc.), serial verb constructions, and so on, while cautioning that we can find the full gamut of languages ranging from those that are quite close in structure to their superstrates to others that diverge quite significantly even within a single region and within a single lexifier group, for instance English-lexifier creoles of the Caribbean. Singler considers the proposition that at least some of the shared features in, for instance, Atlantic creoles, may be the result of diffusion, as strongly argued in the work of Baker (1999) and Baker and Huber (2001). While Singler supports diffusion in principle, he points to serious flaws in the methodology by which diffusion is purportedly established. Kihm addresses the question whether creole structures can be considered to instantiate “unmarkedness,” arguing that this term has to be interpreted in the Chomskyan sense of referring to core grammar, and more specifically that “unmarked” is to be interpreted as referring to parallelism in complexity between syntactic and LF representations, and that this represents the default option in grammar. One author who points to a possible difference between expanded pidgins and creoles is Huttar. In his chapter on semantic structure he identifies the auxiliary status of Tok Pisin as the reason that certain semantic domains in its lexicon suffer from lack of elaboration as compared to the substrate languages. A similar point is made by Jourdan, based on an evaluation of the system of kinship terms in Melanesian Pidgin. Thus, prolonged co-existence with ancestral languages may well mean that elaboration of culturally important semantic domains is held back – contrary to what one might expect. In contrast, rapid nativization of a creole language, which presumably goes hand in hand with equally rapidly diminished use of ancestral languages, means that the

14 Silvia Kouwenberg and John Victor Singler creole must be the vehicle for expression of culturally important concepts, which are expected to derive at least to some extent from the substrate language(s), as shown for Ndyuka in Huttar’s own work. Two chapters that look anew at the presence of creole features in varieties that are generally considered not to be creoles are those by Lipski and Spears, which consider this issue in regard to bozal Spanish and African American English, respectively.

4.4

Applied issues

The recognition of the morphosyntactic independence of pidgin and creole languages from their lexifiers (e.g., Muysken 2004) has led to a recognition among linguists in pidgin- and creole-speaking societies that the institutions of these societies need to make use of the pidgin/creole in order to function efficiently. In particular, linguists have long argued that vernacular literacy is a prerequisite to educational success. This issue is taken up in Craig’s chapter. Devonish’s chapter addresses the role of the state in efforts to include or exclude these languages from public functions. Pidgin/creole-speaking societies usually have long-standing oral traditions which make use of the vernacular. This “orality” is often seen as a factor in the lack of interest both on the part of pidgin/creole-speaking populations and their governments to accord official status to these languages. Buzelin and Winer’s chapter considers the ways in which writers of the Caribbean have brought creole within the purview of a literary tradition and have struggled with questions of orality and literacy.

REFERENCES Alleyne, Mervyn (1971) Acculturation and the cultural matrix of creolization. In: Hymes (ed.), pp. 169–86. Alleyne, Mervyn (1980) Comparative Afro-American. Ann Arbor: Karoma. Baker, Philip (1999) Investigating the origin and diffusion of shared features among the Atlantic English Creoles. In: Philip Baker and Adrienne Bruyn (eds.) St Kitts and the Atlantic Creoles: The Texts of Samuel Augustus Mathews in Perspective. London: University of Westminster Press, pp. 315–64. Baker, Philip and Magnus Huber (2001) Atlantic, Pacific, and world-wide features in English-lexicon contact languages. English World-Wide 22, 157–208. Bickerton, Derek (1975) Dynamics of a Creole System. Cambridge: Cambridge University Press. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984) The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7, 173–221. Bickerton, Derek (1986) Beyond Roots: The five-year test. Journal of Pidgin and Creole Languages 1, 225–32.

Introduction 15 Chomsky, Noam (1957) Syntactic Structures. The Hague: Mouton. DeCamp, David (1971) Toward a generative analysis of a post-creole speech continuum. In: Hymes (ed.), pp. 349–70. DeGraff, Michel (1997) Verb syntax in, and beyond, creolization. In: Liliane Haegeman (ed.) The New Comparative Syntax. London: Longman, pp. 64–94. DeGraff, Michel (2001) On the origin of creoles: A Cartesian critique of neo-Darwinian linguistics [Commentary on McWhorter]. Linguistic Typology 5, 213–310. Hall, Robert, Jr. (1943) Melanesian Pidgin English: Grammar, Texts, Vocabulary. Baltimore: Linguistic Society of America. Hall, Robert, Jr. (1953) Haitian Creole: Grammar, Texts, Vocabulary. The American Anthropologist, vol. 43, Memoire no. 74. Holm, John (1988) Pidgins and Creoles, Vol. 1. Cambridge: Cambridge University Press. Holm, John (1989) Pidgins and Creoles, Vol II. Cambridge: Cambridge University Press. Holm, John (2004) Languages in Contact: The Partial Restructuring of Vernaculars. Cambridge: Cambridge University Press. Hymes, Dell (1971a) Introduction: General conceptions of process. In: Hymes (ed.), pp. 65–90. Hymes, Dell (ed.) (1971b) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. Labov, William (1972) Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press. Lefebvre, Claire (1986) Relexification in creole genesis revisited: The case of Haitian Creole. In: Muysken & Smith (eds.), pp. 279–300. Lefebvre, Claire (2004) Issues in the Study of Pidgin and Creole Languages. Amsterdam: John Benjamins. Lefebvre, Claire (1998) Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Amsterdam: John Benjamins. Le Page, Robert B. (ed.) (1961) Creole Language Studies II: Proceedings of the Conference on Creole Language Studies (University of the West Indies, Mona, 1959). London: Macmillan. Meijer, Guus and Pieter Muysken (1977) On the beginnings of pidgin and creole studies: Schuchardt and Hesseling. In: Albert Valdman (ed.) Pidgin and Creole Linguistics. Bloomington: Indiana University Press, pp. 21–48. Migge, Bettina (2003) Creole Formation as Language Contact: The Case of the Suriname Creoles. Amsterdam: John Benjamins. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mühlhäusler, Peter (1997) Pidgin and Creole Linguistics, 2nd edn. London: University of Westminster Press. Muysken, Pieter (1988) Are creoles a special type of language? In: Frederick J. Newmeyer (ed.) Linguistics: The Cambridge Survey, Vol. II. Linguistic Theory: Extensions and Implications. Cambridge: Cambridge University Press, pp. 285–301. Muysken, Pieter (2004) Pidginization, creolization, and language death. In: Geert Booij, Christian Lehmann, Joachim Mugdan, and Stavros Skopeteas (eds.) Morphologie/ Morphology: Ein internationales Handbuch zur Flexion und Wortbildung [Morphology: An International Handbook on Inflection and Word-Formation]. Berlin: Mouton de Gruyter, pp. 1653–60. Muysken, Pieter and Norval Smith (eds.) (1986) Substrata versus Universals in Creole Genesis. Amsterdam: John Benjamins.

16 Silvia Kouwenberg and John Victor Singler Patrick, Peter L. (1999) Urban Jamaican Creole: Variation in the Mesolect. Amsterdam: John Benjamins. Rickford, John R. (1987) Dimensions of a Creole Continuum: History, Texts, Linguistic Analysis of Guyanese Creole. Stanford: Stanford University Press. Roberts, Ian (1999) Verb movement and markedness. In: Michel DeGraff (ed.) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press, pp. 287–327. Roberts, Sarah (2000) Nativization and the genesis of Hawaiian Creole. In: John McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 247–300. Schuchardt, Hugo (1882–8) Kreolische Studien I–IX (Sitzungsberichte der phil.-hist. Classe der kais. Akademie der Wissenschaften). Vienna: Carl Gerold’s Sohn. Schuchardt, Hugo (1909) Die Lingua Franca. Zeitschrift für Romanische Philologie 33, 441–61. Singler, John Victor (1996) An OT account of pidgin phonology: Coda consonants in Vernacular Liberian English. In: Jan Johnson, Matthew L. Juge, and Jeri L. Moxley (eds.) Proceedings of the Twenty-Second Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society, pp. 375–86. Smith, Norval (1995) An annotated list of creoles, pidgins, and mixed languages. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 331–74. Sylvain, Suzanne (1936) Le créole haïtien: Morphologie et syntaxe. Port-au-Prince: Imprimerie de Meester. Taylor, Douglas (1951) Structural outline of Caribbean Creole. Word 7, 43–59. Thomason, Sarah G. (1995) Language mixture: Ordinary processes, extraordinary results. In: Carmen Silva-Corvalán (ed.) Spanish in Four Continents: Studies in Language Contact and Bilingualism. Washington, DC: Georgetown University Press, pp. 15–33. Thomason, Sarah G. (2001) Language Contact: An Introduction. Washington, DC: Georgetown University Press. Van Name, Addison (1869–70) Contributions to creole grammar. Transactions of the American Philological Association 1, 123–67. Weinreich, Uriel (1953) Languages in Contact. New York: Linguistic Circle of New York. (Reprinted 1968, The Hague: Mouton.) Whinnom, Keith (1971) Linguistic hybridization and the “special case” of pidgins and creoles. In: Hymes (ed.), pp. 91–115. Winford, Donald (2003) Contact Linguistics. Oxford: Blackwell.

Atlantic Creole Syntax 17

Part I Properties of Pidgins and Creoles

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Atlantic Creole Syntax 19

2

Atlantic Creole Syntax DONALD WINFORD

1 Introduction Providing an overview of Atlantic creole syntax and the issues arising from its study is a daunting task. For one thing, Atlantic creoles (ACs) comprise a large and diversified number of languages, spoken in Africa, the Caribbean, the South American mainland, and in North and Central America. They include creoles that are lexically affiliated to Dutch, English, French, Spanish, and Portuguese. The largest group, the English-lexicon creoles, includes languages spoken in over 25 countries, while languages belonging to the second largest group, the Caribbean French creoles (CFCs), are spoken in Haiti, St. Lucia, Dominica, and in French départements outre-mer, Guiana, Martinique, Guadeloupe (including St. Barth). There is a great deal of structural diversity within both these groups, particularly among the English-lexicon creoles, which range from languages quite close in structure to their superstrates, e.g., Bajan and Trinidadian Creole, to others that diverge quite significantly, e.g., the Surinamese creoles. The latter are sometimes referred to as “radical” creoles. The “basilectal” varieties spoken in Belize, Guyana, Jamaica, and other islands stand somewhere in between. Given this diversity, any account of Atlantic creole syntax would be hard pressed to describe the full variety of structural properties these languages display. The task is complicated by the fact that many of the languages have not been fully described. I will therefore focus primarily on the more “radical” English-lexicon creoles and the French-lexicon creoles. I will refer to other creoles where appropriate. My discussion will be restricted to some of the more salient syntactic phenomena found in these languages, which have been the main focus of studies of creole syntax. Henceforth, I shall distinguish the Surinamese creoles from other Caribbean English-lexicon creoles, reserving the designation Caribbean English creole (CEC) for the latter. The Surinamese creoles themselves comprise roughly seven closely related varieties, including Sranan Tongo (the lingua franca of

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

20 Donald Winford Suriname), the Eastern Maroon creoles (Aluku/Boni, Kwinti, Ndjuka, and Pamaka) and the Western Maroon creoles (Saamaka (Saramaccan) and Matawai). My examples will be drawn primarily from Sranan (SR), Pamaka (PM), and Saamaka (SM). For CEC, I will employ examples mainly from Belizean (BEL), Guyanese (GC), and Jamaican (JC) – languages that illustrate the extensive similarities, as well as the differences, within the CEC group. With regard to the Caribbean French-lexicon creoles (CFCs), my illustrations will be drawn mainly from Haitian (HC) and St. Lucian (SLC), the latter being representative of the Antillean group. The explanation of the differences and similarities across creoles remains perhaps the central challenge for creole studies. Indeed, it was this challenge, more than anything else, which first stimulated interest in creole syntax, particularly in the 1980s. Since then, many studies of syntactic phenomena in various creoles have appeared. This is why the present discussion must be eclectic in its choice of syntactic phenomena to cover. My concern is not with issues specific to one or another theoretical model of syntax, but rather with broader questions that are of interest to creolists of all persuasions.

2

The Background

The study of creole syntax began to blossom in the 1980s in response to claims that there was a class of “prototypical” creoles that could be identified on the basis of a number of shared (morpho-)syntactic properties. This view in fact goes back earlier, to the contention that relexification of a prior pidgin played a major role in the formation of Atlantic creoles (e.g., Taylor 1963, 1971) – a view that is now discredited (Den Besten, Muysken, & Smith 1995). The idea of a creole prototype gained currency once more when Bickerton (1981, 1984) proposed that a language bioprogram that guided children’s acquisition of a deficient pidgin was the source of the structural features “radical” creoles shared. Bickerton (1981) proposed a list of about 15 such properties, some morphosyntactic, others syntactic; they are listed in Figure 2.1. The core (morpho)syntactic features in figure 2.1 remain the chief focus of current studies, whether their goal is to uncover the sources of creole structure or simply to describe the inner workings of creole grammar. Most of the issues arising from earlier debates on creole formation are still very much alive today. Thus, there is continuing discussion of the role of universals of language acquisition in creole formation. Closely related to this is the question of which aspects of creole grammar were shaped by influence from superstrate or substrate sources, or arose through innovation and internal change. Exploration of such issues has added much to our understanding of creole syntax. Creoles are now being studied on their own terms, rather than through the filter of their lexifier or substrate languages. The study of creole syntax has increasingly become part of the mainstream of syntactic theory, and a testing ground for current models of syntax.

Atlantic Creole Syntax 21 Morphosyntactic features 1 A three-way organization of the TMA system into Anterior tense, Irrealis mood, and Non-punctual aspect, expressed by invariant preverbal markers occurring in that order. 2 Use of a single negator as well as negative concord in negative sentences. 3 A distinction between equative, attributive, and locative copulas. 4 A system of articles including a definite article for “specific-reference NPs known to the listener,” an indefinite article for “specific-reference NPs not known to the listener,” and zero for “non-specific NP.” 5 The expression of existential and possessive by means of a single lexical form. 6 The behavior of “predicate adjectives” as verbs. 7 The use of bimorphemic question words. Syntactic features 8 Invariant word order in statements and questions. 9 The movement of focused constituents to sentence-initial position. 10 The use of “passive equivalents” lacking an agentive phrase, as distinct from the periphrastic passive found in the European lexifiers. 11 Relativization with subject pronoun copying. 12 A distinction between realized (purposive) complements introduced by ‘go’ and unrealized complements introduced by ‘for’. 13 The finite status of complements to perception verbs. 14 The distinction in complementizer status between ‘say’ and ‘go’, on the one hand, and ‘for’, on the other. 15 The use of serial verb constructions in which serial verbs have a “casemarking” function, introducing directional, benefactive, dative, and instrumental arguments. Figure 2.1 Features of the creole prototype (after Bickerton 1981)

The following discussion of a selection of the core features listed above is organized into features relating to basic clause structure, and those relating to complex constructions. Structures are often shared across ACs, and many of the illustrations of specific structures from particular languages can be considered representative of ACs in general.

3 Basic Clause Structure 3.1

Word order

The three basic sentence types – declaratives, yes/no interrogatives, imperatives – all have SVO ordering in ACs, with interrogatives employing rising intonation, as distinct from the other two types which have falling intonation. Even in sentences with pronominal objects, Romance-lexicon creoles have SVO ordering, unlike their lexifiers.

22 Donald Winford (1) BEL

a.

Jan mi iit di mango. John past eat det mango ‘John ate the mango.’ b. Jan mi iit di mango? ‘Did John eat the mango?’ c. (Yu) iit di mango! 2s eat det mango ‘(You) eat the mango!’

CFCs employ an optional question marker ès(ke) derived from their lexifier, as in (2). SVO order is also found in embedded clauses and wh-questions, discussed later. (2) SLC

3.2

ès yo bat li? q 3p beat 3s ‘Did they beat him?’

The TMA system

With rare exceptions, ACs express functional categories of negation, tense, modality, and aspect by means of preverbal auxiliaries. The syntax and semantics of these auxiliaries have been topics of continuing debate, much of it triggered by Bickerton’s claim about the “prototypical” creole TMA system, described above. A classic illustration of the so-called prototype can be found in the Surinamese creoles, as shown in the following Sranan example (unless otherwise indicated, examples are from my own field data). (3) SR

Jan ben sa e sribi. John past mod imp sleep ‘John would have been sleeping.’

Various studies have shown that both the semantic and syntactic claims made by Bickerton are inaccurate in several respects, and that most creoles depart from this prototype in various ways (see contributions to Singler 1990). For instance, among CECs, only Belizean allows the TMA ordering seen in the Suriname creoles (4). Such a combination would be ungrammatical in Guyanese (5) and Jamaican Creole: (4) BEL

Jan mi wa di iit wen wi get de. John past fut prog eat when 1p get there ‘John would have been eating when we arrived there.’

(5) GC

*Jan ben go a iit John past fut prog eat

Atlantic Creole Syntax 23 Moreover, the inventories of TMA categories vary significantly across CECs and the Surinamese creoles. Still, they share a common core of categories, including a Relative Past (expressed by bin and variants), a predictive Future (expressed by go and variants), a Completive Perfect (expressed by don in CEC, kaba ‘finish’ in Suriname), and a Perfective aspect, expressed by the unmarked verb. The English-lexicon creoles of the Eastern Caribbean also share an Imperfective category, expressed by (d)e in Suriname and (d)a in the other creoles. There are striking similarities between these English-lexicon creoles and those of other lexical affiliations. Thus, all of the TMA categories of the former can also be found in French-lexicon creoles (Spears 1990). A notable difference is that, while the former in most cases employ TMA markers derived from English lexical items, the latter for the most part employ forms derived from cognates that have similar grammatical functions in their lexifiers (DeGraff 2007). The interaction between Aktionsart and TMA has also been vigorously debated in the literature. For instance, Bickerton’s claim that unmarked statives always have present time reference while unmarked non-statives have simple past reference has been shown to be false, since contextual factors can attach different temporal meanings to each. Similarly, Bickerton’s claim that statives and non-statives marked by the “anterior” (Relative Past) tense have simple past and past-before-past meanings respectively has been falsified (e.g., Jaganauth 1988; Winford 2000a). The question of how discourse or pragmatic factors influence the interpretation of these markers is still a matter of disagreement (Gooden 2000). With regard to the syntax of the verb complex, Veenstra (1996) argues that the preverbal markers are base-generated heads, each with its own functional projection. This seems fairly uncontroversial (see also Winford 1993, ch. 3). It has also been argued that the distinctive periphrastic structure of AC TMA systems, and in particular the fact that verbs lack morphological inflection, explains why ACs fail to exhibit properties such as verb movement, which some models propose for the European lexifier languages (Veenstra 1996). From a broader perspective, it has been argued that the structure of the verb complex in ACs includes all of the functional categories in the universal INFL complex, ordered as follows (see Veenstra 1996, p. 47; Déprez & Vinet 1997, p. 213): (6) AGR > NEG > TNS > MOD > ASP While the case for AGR is highly theory-dependent and hence subject to question, it seems clear that the rest of the hierarchy is valid for ACs in general. There are certain restrictions on the interaction between the negative marker and negative quantifiers as well as polarity items in these and other ACs, by

24 Donald Winford contrast with their lexifiers. For example, DeGraff (1993a, p. 67) contrasts sentences like the following: (7) French HC

Ce n’est pas rien. 3s ne+is neg nothing Sa pa anyen. that neg nothing ‘This is something.’ (Lit. ‘This is not nothing.’)

DeGraff (1993a, p. 68) argues that the French sentence is an instance of double negation while the Haitian sentence is a case of negative concord. In general, only the latter is found in ACs. (See DeGraff 1993a for discussion of what this implies for differences in scope of negation between Haitian Creole and French.) However, the canonical ordering shown in (6) is by no means the only one found, even in the most “radical” creoles. In Sranan, for example, Imperfective aspect can have scope over modality, as in the following example, where Imperfective e has scope over the modal verb musu: (8) SR A ben e musu e taki nanga unu. 3s past imp must imp talk with 1p ‘He usually had to be talking with us.’ Moreover, it is not the case that MOD always follows TNS in ACs. This holds only for root or deontic modals. Epistemic modals, by contrast, can have scope over NEG, TNS, and other categories, as in the following examples: (9) JC

Jan mosi no waan kom ya tunait. John prob neg want come here tonight ‘John probably doesn’t want to come here tonight.’

(10) HC Jan ka pa te vini. John mod neg past come ‘John might not have come.’ (DeGraff 1993b, p. 86) Interestingly, these patterns of ordering are in keeping with Cinque’s (1999, p. 85) claim that the orderings of functional heads cross-linguistically “are subsequences of a single universal sequence of functional heads present in all languages.” This universal sequence is as follows (some details omitted): (11) Epistemic moods > T (Past) > T (Future) > Deontic moods > Asp (Habitual) > Asp (Perfect) > Asp (Progressive) > Asp (Completive) > V

Atlantic Creole Syntax 25 In general, then, while the order of preverbal functional heads does not conform precisely to the TMA model that Bickerton posited, it can still be claimed that their order in creoles conforms to a universal pattern.

3.3

Variations on basic clause structure

Syntactic devices for rearranging the constituents of a clause, thereby achieving different kinds of information packaging, fall into two broad types: those that involve movement of core arguments within a single clause, e.g., passives, and those in which the moved constituent is external or peripheral to the clause from which it is moved. The latter include focus and wh-questions. Once more, we find a striking degree of similarity among ACs in these areas of grammar, though there are also marked differences in some cases.

3.3.1 Passivization Passivization in most ACs differs significantly from that found in their lexifiers, with few exceptions (e.g., Papiamentu; see Kouwenberg & Murray 1994). European languages typically employ a strategy exemplified by the following English structure: (12) NP2 be Vpass by NP1 Here, NP2 is the “undergoer” and NP1 is the agent, while the verb is morphologically marked for passive. Constructions of this type have traditionally been viewed as canonical passives, and used as a benchmark with which creole passives are compared, leading to a somewhat Eurocentric view of the latter. Unlike their lexifiers, passives in most ACs lack ‘be’ or a similar auxiliary, morphological marking on the verb, and an agentive prepositional ‘by’-phrase. The following illustrate this; sentence (16) comes from Dominican French Creole: (13) JC

Di raip planten dem don ruos yet? det ripe plantain pl pf roast yet? ‘Have the ripe plantains been roasted yet?’ (Bailey 1968, p. 217)

(14) SR Kande den suma disi ben kweki tra fasi perhaps det.pl person dem past raise other fashion ‘Perhaps these people were raised differently.’ (Winford 2000b, p. 95) (15) SLC

Kay–la kay pentiwé simen prochen house–det fut paint week next ‘The house will be painted next week.’ (Winford 1988, p. 281)

(16) DOM Pweson ka twape la. Fish imp catch there ‘Fish are caught there.’ (Amastae 1983, p. 63)

26 Donald Winford These facts have led some creolists (e.g., Roberts 1993) to claim that ACs generally lack the passive altogether or use it rarely. Other scholars, while acknowledging that “passive-like” structures exist in ACs, label them “agentless constructions” (Amastae 1983) or claim that they are “lexical passives of the adjectival sort” (DeGraff 2007). In fact, passivization in ACs is a highly productive process, though there are certain constraints on its operation. For instance, in AC passivization, oblique arguments such as indirect objects cannot be promoted to subject. Moreover, the class of verbs that allow passivization is restricted to activity verbs high on the scale of transitivity, that is, “verbs that express actions that are volitionally controlled by an agent and have a direct effect on the patient,” e.g., ‘wash’, ‘dry’, etc. (Winford 1993, p. 123). Verbs lower on the scale, for example verbs of perception and psychological states generally resist passivization (Craig 1980, p. 117; Allsopp 1983, p. 148). There are also restrictions on animate, especially human, NPs appearing as patient subjects in AC passives (Alleyne 1987, p. 74). Hence structures like (17), which has an animate passive subject, tend to be marginal; speakers tend to prefer active constructions, or ‘get’ passives (18): (17) CEC

Di pikni kot wid piis a glaas. det child cut with piece of glass ‘The child was cut by/with a piece of glass.’

(18) CEC

Di pikni get kot wid piis a glaas. det child get cut with piece of glass ‘The child got cut by/with a piece of glass.’

Various analyses have been proposed for the syntax of AC passives, each dependent on a particular syntactic theory. Within the Chomskyan generative tradition, scholars have attempted to derive AC passives via the same mechanisms proposed for European passives. For example, following Langacker and Munro (1975), Amastae (1983) argued that passives in Dominican French Creole contain an underlying BE which dominates the verb. More recently, LaCharité and Wellington (1999), employing the Government and Binding framework, argue that the passive in JC is triggered by a passive morpheme that is null but syntactically active. Winford (1993, p. 147) proposes a passive metarule for CEC passives based on the Generalized Phrase-Structure Grammar analysis of English passives. No analysis has so far succeeded in accounting for all the facts concerning these constructions, especially the restrictions mentioned above.

3.3.2

Cleft constructions

Cleft constructions manifest two different types of focus: presentational or information focus and identificational or contrastive focus. Presentational focus constructions present some new topic and usually involve the fronting of an NP, though other constituents may be moved, as in this Sranan example:

Atlantic Creole Syntax 27 (19) SR

Want na tu leisi mi nanga a man meki afsprak because foc two time 1s with det man make appointment kaba. already ‘Because it’s two times the man and I made an appointment already.’ (Winford 2000a, p. 433)

In identificational or contrastive focus, the fronted element may be any major constituent of the sentence, including [+verb] predicates. Since the latter raise a number of difficult questions, they are discussed separately below. The function of contrastive focus is generally to identify some participant, entity, place, or time, which is presumed to be unknown to the hearer, as the actual one involved in the situation described. For example, each constituent of sentence (20a) may be focused in this way, being introduced by a, the equative copula acting as focus marker: (20) CEC

a.

Jan fain di moni ina di yaad yeside. ‘John found the money in the yard yesterday.’ b. A Jan fain di moni ina di yaad yeside. ‘It’s JOHN who found the money in the yard yesterday.’ c. A di moni Jan fain ina di yaad yeside. ‘It’s THE MONEY that John found in the yard yesterday.’ d. A yeside Jan fain di moni ina di yaad. ‘It was YESTERDAY that John found the money in the yard.’ e. A ina di yaad Jan fain di moni yeside. ‘It’s IN THE YARD that John found the money yesterday.’

Various questions arise for a syntactic analysis of cleft constructions in ACs (see Veenstra & den Besten 1995). One concerns the categorial status of focus markers such as a. In most ACs, the focus marker is identical in phonetic shape to the equative copula. Exceptions include Saamaka (Smith 1996) and Berbice Dutch (Kouwenberg 1995), which employ postposed focus markers wE (Saamaka) and sa/so (Berbice Dutch), which appear to have been adopted directly from the respective substrates, Gbe for Saamaka and Ijo for Berbice Dutch (though Berbice Dutch sa/so combines with an initial copula da; Kouwenberg 1995, p. 235). If the copula-like focus markers have the same pronominal status that has been claimed for them in their equative function, this would suggest that they introduce an equative copula clause under which another clause is embedded in cleft constructions. Alternatively, Manfredi (1993) suggested that Haitian Creole cleft introducer se is always a verb preceded by a (possibly null) topic, and followed by the element it places in focus. Either of the above interpretations would mean that cleft constructions are bi-clausal, as has been argued (e.g., by Lumsden 1990). On the other hand, creolists like Byrne (1987) have suggested that the focused constituents in cleft constructions are preclausal operators similar to wh-elements.

28 Donald Winford Despite these disagreements, there is consensus that the type of movement involved in AC clefts is similar to the wh-type movement found in whquestions and relatives. In fact, wh-questions are often introduced by the same focus marker as cleft constructions (see 3.3.4). The connection among the three constructions is quite apparent in CFCs, where all are subject to the requirement that relativizer ki must fill the gap left by movement out of a subject position. Compare the following Haitian Creole sentences: (21) HC

a.

Se lougawou ki manje timoun nan foc werewolf rel eat child det ‘It’s a werewolf that ate the child.’ (Manfredi 1993, p. 32) b. Ki sa ki fè w pa vini yè swa? q that rel make 2s neg come last night ‘Why didn’t you come last night?’ (DeGraff, p.c., October 2002) c. Misie a pran tut lazhan ki te an kay la. man det take all money rel past in house det ‘The man took all the money that was in the house.’ (adapted from Carrington 1984, p. 145)

Clefting of locative and nominal predicates requires that a copula element surface in exposed position. In English-lexicon creoles, the locative copula de appears. In CFCs, which do not have a locative copula, the element ye must appear in exposed position. (See Manfredi 1993; DeGraff 1995; Déprez & Vinet 1997 for discussion of its status.) Where a (bare) nominal predicate is fronted, Haitian Creole also allows a copy of it to appear in situ (22b). Manfredi (1993) explains this as due to the fact that dokte has the status of a denominal predicate that behaves like other [+verb] predicates (see the discussion of predicate cleft in 3.3.3): (22) HC

3.3.3

a.

Se dokte Jak ye. foc doctor Jack ye b. Se dokte Jak dokte. foc doctor Jack doctor ‘Jack’s a DOCTOR.’ (Déprez & Vinet 1997, p. 230)

Predicate clefting

When [+verb] predicates are clefted, the focused element also appears in situ, as in the following examples: (23) HC

se bèl li bèl. foc pretty 3s pretty ‘She’s really pretty.’ (Lumsden 1990, p. 742)

(24) SLC Se ganye mwen te ganye i. foc buy 1s past buy 3s ‘I BOUGHT it.’ (Carrington 1984, p. 147)

Atlantic Creole Syntax 29 This type of construction, referred to (somewhat ambiguously) as “predicate clefting,” has generated a great deal of controversy, most of it revolving around the status of the fronted and in situ elements. Some approaches (e.g., Piou 1982; Koopman 1984), view the lower (in situ) element as a copy of the moved predicate. But this raises the question why the “copy” retains such properties as TMA marking, government, and case assignment. Manfredi (1993, p. 5) argues instead that “predicate cleft involves movement of a nominal argument which denotes the event of the verb” (the latter remaining in situ). In Englishlexicon creoles, support for this nominal analysis comes from the fact that the fronted element may be modified by the definite article, as in the following Krio example: (25) KRIO

Di álà wé ì bìn dè álà wék ólmàn. det shouting that 3s past prog shout wake everbody ‘His shouting woke everybody.’ (Williams 1977, p. 3)

Such structures are rare in the world’s languages, which prompts us to ask why they are found in ACs. The answer seems to lie in substrate influence. Manfredi suggests that the predicate cleft construction has models in various Kwa languages, some of which mark the fronted verb-like element morphologically as a nominalization, which others, e.g., Gbe, do not. The latter pattern appears to be the one adopted by ACs.

3.3.4

Wh-questions

Wh-questions in ACs display several interesting characteristics. Unlike their lexifiers, they do not allow auxiliary inversion, as the following illustrate (see also Belizean (1b), Haitian Creole (21b)): (26) GC

Wisaid Jan fain di pikni–dem? q-side Jan find det child-pl ‘Where did John find the children?’

(27) SLC

Ki maniè u ka fè farin? what manner 2s imp make farina ‘How do you make cassava flour?’ (Carrington 1984, p. 150)

Embedded wh-questions follow the same pattern. Some creoles, for example Haitian, allow the option of leaving the wh-expression in situ (De Graff 2007). Like clefts, wh-questions can be introduced (optionally) by the copula-like focus marker: (28) BEL

Da hu Mieri kom ya fu sii? foc who Mary come here for see ‘Whom did Mary come here to see?’

30 Donald Winford (29) HC Se ki moun ki ap vini? it’s which person rel prog come ‘Who’s coming?’ (Lumsden 1990, p. 742) This similarity among wh-questions, relatives, and clefts in ACs reflects a pattern found in Kwa languages, where all are “morphologically related through the presence of a copula-type morpheme” (Manfredi 1993, p. 24). One aspect of AC wh-questions that has attracted much attention is the fact that their wh-expressions are often compound forms corresponding to ‘what person’ (who), ‘what place’ (where), ‘what time’ (when), etc. CFCs employ a variety of such compounds (see table 2.1). The Surinamese creoles may appear at first to be exceptions to this pattern, since their wh-expressions are single forms like suma ‘who’, san ‘what’, pe ‘where’, etc. But these in fact derive from earlier analytic expressions like o suma ‘what person’, o sani ‘what thing’, o pe ‘what place’, etc. As Muysken and Smith (1990, p. 884) note, these analytic expressions can be represented abstractly as Question particle (Q) + questioned semantic unit. Other creoles depart from the pattern to differing extents, employing single wh-expressions of a more opaque character, drawn either from the lexifier or some substrate language. For instance, CEC has forms like huu ‘who’, wa(t) ‘what’, etc., all derived from English. Papiamentu has ken ‘who’ and other forms from Spanish/Portuguese, while most wh-forms in Berbice Dutch derive from Dutch. On the other hand, Saamaka has a few wh-forms that appear to derive from Gbe, such as ambE ‘who’ and andí ‘what’ (Smith 1987). Similarly, Berbice Dutch has the form wanga < *wa-anga [q-place] ‘where’, parallel to Kalabari to-anga [q-place] ‘where’ (Muysken & Smith 1990, p. 892). Table 2.1 illustrates wh-expressions in a few ACs. Each creole shown here is representative of others in its group (CECs, CFCs, and the Surinamese creoles). The emergence of analytic wh-expressions in creoles has been ascribed to universal principles of acquisition, in particular to the need for transparency (cf. Andersen’s 1984 “One to One Principle”). But, as Muysken and Smith (1990, p. 890) show, this pattern is also found in certain substrate languages, especially those of the Gbe family, and may also be due to their influence. Table 2.1 Some wh-interrogative forms in ACs Gloss

SLC

GC/JC

(Earlier) Sranan

‘who’ ‘what’ ‘when’ ‘where’ ‘how’ ‘why’

ki ki ki ki ki ki

huu wa(t) or we wa-taim/wen taim wi-said/we paat hou wa-mek

(o) suma (o) sani (o) ten (o) p(l)e(s) (o) fa(si) (fu) san ede/ san meki

mun sa tã kote maniè sa ki fè

Atlantic Creole Syntax 31

4 Complex Constructions This section surveys a number of multi-clause constructions in ACs, including cases of verb serialization (4.1), subordination (4.2), and relativization (4.3).

4.1

Serial verb constructions

ACs employ a variety of sentence structures containing two or more VPs linked together with no overt markers of coordination or subordination and no pause (or “comma intonation”) separating the VPs. Such serial verb constructions (SVCs) display a number of syntactic and semantic properties. Generally, the verbs involved have only one specification for TMA and polarity and share either a subject (30) or object (31). There has been some debate as to whether either subject or object sharing is a defining property of these constructions. Veenstra (1996, pp. 78–88) demonstrates that neither claim holds. (30) CEC

Di pikni-dem waak go a riva. det child-pl walk go loc river ‘The children walked to the river.’

(31) SR

Kofi naki Amba kiri. Kofi strike Amba kill ‘Kofi struck Amba, killing her.’

Veenstra (1996, p. 92) classifies SVCs into two broad categories, according to the degree of independence between the sub-events described. He makes a further distinction within each category according to the restrictions on the verbs that can appear in the “serial” slot:1 A: Hypotactic-like SVCs (less independence between sub-events) • Type 1: Lexically restricted, including directional, dative/benefactive, comparative, and “aspectual” types with verbs like VP-final kaba ‘finish’. • Type 2: Lexically free, e.g., resultative SVCs like (31) above. B: Paratactic-like SVCs (more independence between sub-events) • Type 3: Lexically restricted, but with looser interpretation of their event composition. Includes ‘take’ serials and causative constructions with ‘make’ (but see below). • Type 4: Lexically free. Various kinds of VP linkage similar to (32) below. The distinction between categories A and B corresponds roughly to that which Sebba (1987, p. 110) makes between “subordinating” and “coordinating” SVCs respectively. The latter express two propositions, and the selection of verbs is constrained only by semantic or pragmatic compatibility, as in the following example (my type 4):

32 Donald Winford (32) SR Amba go na wowoyo bai nyan. Amba go LOC market buy food “Amba went to the market and bought food.” SVCs in which ‘take’ is the first verb (V1), introducing arguments of various types, also belong in category B (my type 3). Since only ‘take’ appears in the V1 slot, this is a lexically restricted type. In the following examples, ‘take’ introduces instrumental, theme, and comitative arguments respectively: (33) JC

Im tek naif kot di kien. 3s take knife cut det cane ‘He used a knife to cut the cane.’

(34) SM Mi téi dí búku léi dí wómi. 1s take det book show det man ‘I showed the book to the man.’ (Veenstra 1996, p. 85) (35) SR

Kofi teki en sisa go bai krosi. Kofi take 3s.poss sister go buy clothes ‘Kofi took his sister to buy clothes.’

Veenstra (1996, p. 101) also includes ‘make’ causative constructions in this category, but argues later (p. 161) that mbei in SM is actually a complementizer. I will argue below that, at least in CECs and CFCs, ‘make’ is a verb that takes a sentential complement. The more hypotactic-like or “subordinating” SVCs involve greater cohesiveness between the verbs involved, and seem in fact to express a single proposition. The serial V2 in these cases has the function of adding to or modifying the meaning expressed by the V1. A brief overview of some SVCs of this type follows.

4.1.1

Directional SVCs

In directional SVCs, V1 is a verb of motion, and the V2 indicates the direction of motion, as in (30). Directional SVCs are most productive in the Surinamese creoles, which possess by far the widest range of serial verbs of this type, including komoto ‘come out’, fadon ‘(fall) down’, komopo ‘(come) up’, trowe ‘(throw) away’, lontu ‘(go) around’, etc. CECs have equivalents of only a few of these, such as go, come, and gaan. In other cases, they employ adverbials like awee ‘away’, or prepositions derived from ‘out of’, ‘around’, etc. to express the relevant direction. The inventory of directional serial verbs in CFCs appears to be restricted to just vin(i) ‘come’ and ale ‘go’. Like CECs, CFCs employ prepositions rather than serial verbs to express other kinds of direction (see Carrington 1984, p. 96), as well as verbs like monte ‘come up’ and sòti ‘come out’, which express both movement and direction.

Atlantic Creole Syntax 33 (36) SR

Kofi hari a pikin komoto na ini a olo. Kofi haul det child come-out loc in det hole ‘Kofi pulled the child out of the hole.’

(37) HC Bouki voye timoun yo ale (lekol). Bouki send children pl go (school) ‘Bouki sent the children away (to school).’ (DeGraff 1993b, p. 82) Also of interest are constructions in which the verbs kom, go and (in some of the CECs) gaan, acting as V2, take a VP complement and convey a purposive sense. We even find more complex serial strings like the following, where kom and go are used to express both directional (first occurrence of go) and purposive (second occurrence) meanings; the “realized” goals expressed in these cases contrast with the unrealized goals expressed by ‘for’ complements (discussed below): (38) CEC

4.1.2

Mieri kyari a pikni go a aspital go sii dakta. Mary carry det child go loc hospital go see doctor ‘Mary took the child to the hospital to see the doctor.’ (Winford 1993, p. 242)

Comparative SVCs

In CECs and CFCs generally, comparison is expressed by an SVC in which the V2 is a verb meaning ‘pass’. In addition, the Surinamese creoles employ a verb moro (< more) meaning ‘exceed’ in this function. (39) CEC

Jan smaat paas aal a dem. John smart pass all of 3p ‘John’s smarter than all of them.’

(40) SLC

Chèn li a gro pase sa la. chain 3s det thick pass dem det ‘His chain is thicker than that one.’ (adapted from Carrington 1984, p. 70)

(41) SR

a.

Amba tranga pasa Kofi. Amba strong pass Kofi ‘Amba’s stronger than Kofi.’ b. Amba tranga moro Kofi. Amba strong exceed Kofi ‘Amba’s stronger than Kofi.’

Among the Surinamese creoles, Sranan has also adopted more Dutch-like comparative structures (Winford 1997, p. 280). Increasingly the CECs are using more English-like comparative structures with moo(r) (< more), which has adverbial rather than verbal status, although it tends to appear after the predicate,

34 Donald Winford like moro in the Surinamese creoles, as shown in (42) for Jamaican Creole. CFCs also have more lexifier-like comparative structures employing both pli (< plus ‘more’) and pase. (42) JC Samwel wok haad muor dan Jan. Samuel work hard more than John ‘Samuel worked harder than John.’ (Bailey 1966, p. 127) (43) SLC

4.1.3

Bwa pen mwen pli bèl pase sa a u bread fruit 1s more beautiful pass dem det 2s ‘My breadfruit is lovelier than yours.’ (adapted from Carrington 1984, p. 72)

“Dative/Benefactive” SVCs

Another SVC typical of ACs is one in which a verb meaning ‘give’ functions as V2, and introduces a recipient or benefactive argument. The CFCs seem to restrict the use of serial ‘give’ to structures where V1 expresses some kind of transfer, for example, the sense of ‘bring’, ‘send’, ‘hand over’, ‘sell’, etc. CECs, however, also allow SVCs in which ‘give’ introduces a “benefactive” argument. In such cases, no direct transfer is involved. The Surinamese creoles employ ‘give’ to introduce several other types of argument, including “substitutive,” “experiencer,” and “source.” (See Migge 1998 for fuller discussion.) (44)

GC

(45) SR

Dem kliin di yaad gi di ool man. 3p clean det yard give det old man ‘They cleaned the yard for the old man.’ (Winford 1993, p. 278) Mi seri a oso gi en. 1s sell det house give 3s.obj ‘I sold the house to him/her.’

(46) HC Malis te pran yon flè bay Boukinèt. Malis past take det flower give Boukinèt ‘Malis gave a flower to Boukinèt.’ (DeGraff 1993b, p. 82) (47) SLC I te ka pòte on chay ti bonbon ba nu. 3s past imp bring det lot little sweet give 1p ‘He used to bring a lot of little sweets for us.’ (adapted from Carrington 1984, p. 124)

4.1.4

The syntax of SVCs

The variety of SVCs and the restrictions on verb selection in the various types have been problematic for attempts at a unified syntactic treatment. In particular, there has been disagreement concerning the nature of the concatenation involved. Sebba (1987) assigned different phrase structures to the paratacticlike versus the more hypotactic-like SVCs. He treated the former as cases of

Atlantic Creole Syntax 35 VP coordination, and the latter as cases in which the V1 was subcategorized for a VP complement. In support of this, Sebba argues that the coordinating type is subject to Ross’s (1967) Coordinate Structure Constraint, while the subordinating type is not. However, some paratactic-like SVCs, such as those with ‘take’ as V1, are not subject to Ross’s constraint: (48) SM a.

Andí a téi kóti dí beée? what 3s take cut det bread ‘What did he cut the bread with?’ b. Andí a téi dí fáka kóti? what 3s take det knife cut ‘What did he cut with the knife?’ (Veenstra 1996, p. 120)

Veenstra (1996, p. 122) suggests, contra Sebba, that SVCs in general are similar to cases of asymmetric coordination, that is, they show sequential effects, and are best treated as instances of adjunction. He argues for this position on the basis of various patterns of extraction out of the two VPs. The problems posed by SVCs raise interesting questions about the nature of the argument sharing involved, and the distinction between complements and adjuncts in creoles – questions that are still in need of answers.

4.2

Subordination

The subordinate clauses discussed in this section include those that function as complements and those that function as modifiers of nouns (relative clauses).

4.2.1 Complementation Many of the issues raised by Bickerton regarding the syntax of complementation in creoles have continued to be at the center of creolists’ attention. Perhaps the most central of these is whether creoles have only finite complements, which would mean they lack the kinds of infinitival and other reduced complements found in their lexifiers. Other (related) issues concern the distinction between factive (‘that’-type) and non-factive (‘for’-type) complements, as well as the status and sources of complementizers in different creoles. There is in fact evidence that ACs possess both finite sentential complements and others that are reduced in some way (e.g., lacking overt subjects, TMA marking, etc.). In general, factive complements in ACs are finite sentences, in the sense that they carry the same TMA marking found in independent clauses. The presence or acceptability of tense marking has traditionally been seen as a requirement for finite status in European languages. But it is debatable whether the same criterion is always appropriate for creoles, where sentences do not require tense marking to express a complete proposition. It is arguable that any AC sentential complement containing at least one marker of tense, aspect, or modality is finite. Complements containing unmarked verbs can arguably be treated as finite too.

36 Donald Winford

4.2.2

Factive complements

Factive complements in ACs include arguments of predicates like ‘seem’, complements of causative ‘make’, of assertion verbs (‘say’, ‘tell’, etc.), of psych state verbs (‘know’, ‘believe’, etc.), and of perception verbs (‘see’, ‘hear’, etc.). The predicate ‘seem’ and evaluative predicates like ‘true’ are always subcategorized for extraposed sentential complements, as in the following examples: (49) SLC

(50) JC

I sam yo kay dechire i 3s seem 3p fut destroy 3s ‘It seems that they will destroy it.’ (adapted from Carrington 1984, p. 113) A truu se Jan lov Mieri. 3s true comp John love Mary ‘It’s true that John loves Mary.’

ACs lack raising constructions such as those found in English where ‘seem’ takes an infinitive complement. It is clear that the complements in the ACs are finite sentences. Note that the English-lexicon creoles introduce such complements with a verb meaning ‘say’ (e.g., JC se), which is also used to introduce complements to verbs of saying, knowing, etc. In all these cases, CFCs employ a zero complementizer. Causative ‘make’ subcategorizes for sentential complements, and can have sentential subjects as well. (51) HC Se li bwa twop fè li tombe muri. it’s 3s drink too much make 3s fall die ‘[The fact that] he drank too much caused him to fall and die.’ (Arthur Spears, p.c., 2004) (52) SM Egbert bebé daán hía pói mbéi a fiká a wósu Egbert drink rum much spoil make 3s.sub stay loc house síki-síki. sick-sick ‘[The fact that] Egbert drank too much rum made him stay at home sick.’ (Veenstra 1996, p. 101) In these and other cases, ACs lack rules of subject-to-object raising (“Exceptional Case Marking”), in contrast to their lexifiers. Hence structures like the following containing an embedded subject in objective case are ruled out (but see the discussion of ‘for’ complements below): (53) GC Jan sliip aal maanin mek i/*am riich leet. John sleep all morning make 3s/3s.obj reach late ‘(The fact that) John slept all morning made him arrive late.’ (Winford 1993, p. 304)

Atlantic Creole Syntax 37 Veenstra (1996, p. 161) distinguishes between two types of ‘make’ sentential complements in Saamaka, a “fully blown (or finite) IP” as in (52), and a reduced type as in (54) (p. 158): (54) SM Dí hía bebé dáan u Egbert mbéi a fiká a wósu síki-síki. det much drink rum of Egbert make 3s stay loc house sick-sick ‘Egbert’s excessive drinking made him stay home sick.’ He argues that the complement in (54), unlike that in (52), does not have either a tense or negation projection, and that, in such cases, mbéi functions as a verb (p. 160). This analysis seems to extend to other ACs. On the other hand, he claims that IP-introducing mbéi in sentences like (52) has been grammaticalized into a complementizer meaning ‘so that’. Evidence for this includes the fact that mbéi cannot be marked for aspect or undergo predicate cleft, among other things (pp. 162–3). Whether this analysis extends to ACs in general remains to be investigated.

4.2.3

The case of complementizer ‘say’

The status of ‘say’ in factive complements in the English-lexicon creoles is problematic. Some have argued that ‘say’ is a complementizer equivalent to English that. Supporting this is the fact that ‘say’ is almost entirely bleached of its meaning, especially when it follows psych state verbs or appears in expressions like Jamaican Creole iivn if se ‘even if’. However, unlike typical complementizers, ‘say’ can be stranded under wh-movement, and extraction out of a ‘say’ complement is possible, unlike extraction out of a ‘that’ complement. (55) JC a.

A wa dem tel yu se? foc what 3p tell 2s say ‘What did they tell you?’ b. A huu dem tink se wi kom tunait? foc who 3p think say fut come tonight ‘Who do they think will come tonight?’

Moreover, the ‘say’ + complement unit cannot function as a sentential subject, unlike ‘that’-complements in English. For now, it seems best to interpret ‘say’ as an erstwhile serial verb that has been reanalyzed as a quasi-complementizer which still preserves its verbal character in some contexts (see discussion by Jaganauth 2001). It is generally accepted that this reanalysis was motivated by substrate models in which ‘say’ had similar functions, hence it is not a case of grammaticalization in the usual sense (see Bruyn 1996; this volume). At the same time, it seems that ‘say’ has extended its functions in some creoles, e.g., Sranan (Plag 1993, pp. 46–7).

4.2.4

Perception verb complements

Perception verbs take two types of complement in ACs, a finite type introduced by ‘say’ as well as a reduced (small clause) type without ‘say’. Veenstra

38 Donald Winford (1996, p. 57) demonstrates this distinction for Saamaka with examples like the following: (56) SM

a.

De sí táa dí ógi wómi bì disá dí kónde gó. 3p see say det bad man tns leave det village go ‘They saw that the wicked man had left the village.’ b. De sí dí ógi wómi disá dí kónde gó. 3p see det bad man leave det village go ‘They saw the wicked man leave the village.’

As Veenstra points out, the reduced type involves events that are simultaneous with the time of the matrix verb, while the finite type does not. This distinction appears to apply to ACs in general. Bickerton (1981, p. 117) argued for the finite status of the complement in cases like (56b) on the grounds that, at least for some speakers, there is no rule of subject-to-object raising in such complements, as in the following: (57) GC Mi sii i /*am a kom. 1s see 3s.sub/3s.obj prog come ‘I saw him coming.’ However, Veenstra (1996, pp. 58ff.) provides sound evidence that such complements are indeed non-finite, and are similar to the reduced complements of verbal ‘make’ discussed earlier. Note also that raising of subject to object in perception verb complements is allowed in some idiolects in CEC, providing clear evidence that they are in fact small clauses: (58) GC Mi hia am /i a baal. 1s hear 3s.obj /3s.sub prog bawl ‘I heard him/her/it crying.’ Moreover, ACs generally have small-clause complements to verbs like ‘find’ as in the following Saamaka example from Byrne (1987, p. 49): (59) SM

De fénde hen líbi-líbi. 3p find 3s.obj live-live ‘They found her/him/it alive.’

Such cases are clear exceptions to the claim that all AC complements are finite sentences. (See Veenstra 1996, pp. 63–8, for further discussion.)

4.2.5

‘For’ complements

Non-factive complements in ACs are introduced by an item that also functions as a preposition meaning ‘for’ and, in the English- and French-lexicon creoles,

Atlantic Creole Syntax 39 as a modal auxiliary. Thus we find fu/fi (< for) in the English-lexicon creoles, pou (< pour) in the French-lexicon creoles, and pa (< para) in Iberian-lexicon creoles. For convenience, I will refer to these collectively as ‘for’ complements. They express events that are hypothetical or unrealized in relation to the matrix event. They can appear as objects of aspectual (e.g., ‘start’, etc.), modality (‘have’, ‘able’, etc.) and desiderative (‘want’, etc.) predicates, among others. ‘For’ also introduces adverbial clauses of purpose and reduced relatives. These will not be considered. Subordinate clauses introduced by ‘for’ present a very complex picture across ACs. They have generated much debate concerning the finite/non-finite distinction in creoles, the nature of the control relationships involved, and the status of ‘for’ itself. The three issues are closely linked. ACs differ in the way they interpret ‘for’ clauses with null as opposed to overt subjects. In CECs, presence of an overt pronominal subject in ‘for’ clauses signals disjoint reference between the two subjects: (60) GC

Jani waan fu ij/*i go a maakit. John want for 3s go loc market ‘John wants him/her to go to the market.’

In other ACs, however, an overt third person pronominal may be co-indexed either with the matrix subject or some other party, leading to ambiguity: (61) HC

pou lii/j vini. Jaki vle Jak wants for 3s come ‘Jak wants (him/her) to come.’ (Manfredi 1993, p. 30)

a wosu faai/j skifi di lete. (62) SM Kofii ko Kofi come loc house for-3s write det letter ‘Kofi came to the house (for him/her) to write the letter.’ (Caskey 1990, p. 700) The same applies in the Spanish-lexicon creole Papiamentu (Kouwenberg 1990, p. 42). It is clear then that coreference relationships depend on many factors, some of which are not uniform across ACs. In certain cases where the complement subject must be co-indexed with a matrix object, the French- and Portuguese-lexicon creoles allow free variation between null and overt subjects, as shown here for Haitian: (63) HC

Jani di Marij pu lij/Øj ale. John tell Mary for 3s/Ø go ‘John told Mary to go.’ (Caskey 1990, p. 698)

Alternation between null and overt subjects can lead to different interpretations in other cases. For instance, in the Surinamese creoles, ‘for’ complements

40 Donald Winford to desiderative verbs are interpreted quite differently when they have null subjects as opposed to pronominal ones: di wenke. (64) SM a. Di mujeei hakisi da di womij fu PROi go a det woman ask give det man for go loc det store ‘The woman asked the man [permission] to go to the store.’ di wenke. b. Di mujeei hakisi da di womij faaj go a det woman ask give det man for-3s go loc det store ‘The woman asked the man to go to the store [requested that he go].’ (Caskey 1990, p. 701) In general, then, variation between null and pronominal subjects in ‘for’ complements is freer in the Romance-lexicon creoles than in the English-lexicon creoles, but the alternation does not follow similar patterns across all ACs. Caskey (1990, p. 694) argued that such relationships depend in large part on the inherent meaning of matrix predicates, but we still do not fully understand the complexity of control relationships in such complements across ACs. Kouwenberg (1990, p. 49) argues for Papiamentu that ‘for’ complements with an overt subject are finite, since they may allow tensing. Subjectless complements do not have this option, and are therefore non-finite. We can assume that this applies to all ACs. This raises the question whether ‘for’ has the same status in both types of complement. The general consensus is that ‘for’, when it precedes a finite complement, has the status of a complementizer, without the case-marking properties associated with the English prepositional complementizer ‘for’. In CECs, support for this comes from the fact that pronominal complement subjects following ‘for’ are assigned nominative case, as shown here for Providence Island Creole (PIC): (65) PIC

Im spred bredskromz fi dey/(*dem) fala di trak. 3s spread breadcrumbs for 3p.sub/(*3p.obj) follow det track ‘He spread bread crumbs for them to follow the trail.’

The real issue, then, is the status of ‘for’ in non-finite clauses, where it immediately precedes the verb. The problem is complicated by the fact that ‘for’ can function as a modal auxiliary in the English- and French-lexicon creoles, and can be tensed. Is it therefore a modal or a complementizer in a structure like the following? (66) PIC

Ai ekspek me fi go tu Jan haus yeside. 1s expect past for go to John house yesterday ‘I expected to go to John’s house yesterday.’

Atlantic Creole Syntax 41 Such structures led Bickerton (1980) to conclude that fu/fi in these cases is a modal auxiliary – an analysis that may extend to other ‘for’ complements. It is not clear, then, whether a unified analysis of ‘for’ complements in CEC is possible. It may well be the case that these creoles employ two distinct strategies for this type of complement.

4.3

Relative clauses

Restrictive relative clauses in ACs include reduced and infinitival relatives. Here, I confine my attention to full relative clauses. As noted earlier, there is consensus that the derivation of such clauses involves wh-movement. But there are still some unresolved questions about their syntax. One concerns the status of the relativizers that introduce these clauses, and their variation both within and across creoles. With the exception of place relatives, CFCs and the Surinamese creoles have only one overt relativizer, ki in the former, di in the latter. CFCs employ ki only when a subject NP is relativized, using Ø otherwise. It has been argued variously that ki is a pronominal, like its French cognate qui (Law 1995), or a complementizer (Koopman 1982; DeGraff 1992). DeGraff provides various kinds of evidence against the former position. For instance, ki cannot substitute for pronominals like ki moun [q person] ‘who’ that introduce wh-questions; it lacks semantic content, unlike ki moun and other proforms, and cannot be used as a resumptive pronoun. Unlike ki, the Surinamese Creole relativizer di can represent relativized objects, as in (67), indirect objects, and objects of prepositions, in addition to subjects. (67) SR

Mi e wani go libi nanga a sma di mi o sori yu. 1s imp want go live with det person rel 1s fut show 2s ‘I want to go and live with the person that I’ll show you.’ (Winford 2000b, p. 103)

The same diagnostics that identify ki as a non-pronominal apply to di as well, suggesting that the latter is also a complementizer. (See Bruyn’s 1995 discussion of the evolution of di from the demonstrative pronoun disi ‘this’, which persisted in the function of relative clause introducer into the early twentieth century.) Interestingly, Sranan is increasingly employing its interrogative pronouns, particularly suma ‘who’ and san ‘what’, as relativizers, perhaps on the model of pe ‘where’, which is long established for place relatives. CECs also employ complementizers like dat (< English ‘that’) and (nonlocative) we (< English ‘where’) as relativizers. But they also use wh-proforms such as huu ‘who’, wat ‘what’, wen ‘when’, etc. on the model of English. Budhai (2002) found that we relatives were by far the most common in her Jamaican Creole data, suggesting that CECs fall in line with other ACs in preferring a relativizer distinct from their interrogative pronouns.

42 Donald Winford Another interesting aspect of relativization in ACs concerns the kinds of noun functions that are accessible to relativization. In general, functions such as subject and object, which are high on the Noun Phrase Accessibility Hierarchy (NPAH) (Keenan & Comrie 1977), lend themselves more readily to the type of relativization that leaves a gap at the site of the relativized noun. When nouns with functions lower on the scale are relativized (e.g., objects of prepositions, possessives, objects of comparison), most ACs require that a resumptive pronoun occupy the position of the relativized NP. The consensus seems to be that no wh-movement takes place in these cases, and that the relativizers in question are base-generated. The following illustrate: (68) SR

(69) HC

Dan a man di mi nanga en e taki . . . then det man rel 1s with 3s imp talk . . . ‘Then the man who I was talking with . . .’ Te gen yon fi m pa shonzhe non-li. past have a girl 1s neg remember name-3s ‘There was a girl whose name I don’t remember.’ (adapted from Sylvain 1938, p. 269)

Note that CECs generally allow relativization of nouns like those in (68) and (69) without requiring a resumptive pronoun as in the following examples: (70) CEC

Dem fain di pikni we mi bina taak wit. 3p find det child rel 1s past-prog talk with ‘They found the child that I was talking with.’

(71) GC/JC Di pikni huu/huufa buk mi fain kom ya laas nait. det child who/whose book 1s find come here last night ‘The child whose book I found came here last night.’ In general, use of resumptive pronouns in CECs is restricted to cases involving nouns low on the NPAH, particularly those where the relativized noun is moved out of an embedded clause, for example in sentences like the following (which can also be found in English – McCawley 1988, p. 444): (72) GC

i gat wan boi we awii no noo if i waan kom. 3s have one boy rel 1p neg know if 3s want come ‘There’s a boy who we don’t know if he wants to come.’

The resumptive pronoun strategy is linked to constraints on phenomena like preposition stranding and pied-piping, which operate in different degrees across ACs. Thus, neither CFCs nor the Surinamese creoles allow relatives like the following:

Atlantic Creole Syntax 43 (73) SR

a.

*Unu sabi a man di mi ben e taki nanga. 2p know det man rel 1s past imp talk with Attempted reading: ‘You know the man that I was talking to.’ b. *Unu sabi a man nanga di mi ben e taki. 2p know det man with rel 1s past imp talk Attempted reading: ‘You know the man with whom I was talking.’

One apparent exception to this is the Surinamese Creole form gi ‘give’, an erstwhile serial verb that now functions to all intents and purposes like a preposition ‘for’. It accepts both stranding and (marginal) pied-piping. Such facts have implications for the criteria by which prepositions might be identified in these creoles. Note that Sranan does allow pied-piping (but not preposition stranding) with its more recent wh-relativizers, a further indication of change in this area of its grammar. According to Dreyfuss (1977, p. 65), some Haitian Creole speakers allow pied-piping, but not stranding of ak ‘with’. CECs are once again different. They freely allow preposition stranding, irrespective of the choice of relative clause introducer (complementizer-like dat and we or wh-relativizers huu ‘who’ etc.). They also allow pied-piping, but only with their wh-relativizers. Jamaican Creole even uses a possessive relativizer, huufa ‘whose’. To summarize, despite a number of significant differences, the derivation of relative clauses across ACs involves wh-movement, and shows a preference for a single complementizer-like relativizer. Also noticeable is a shared prohibition against zero relativizers when subject NPs are relativized, and the use of resumptive pronouns when nouns with functions lower on the NPAH are relativized.

5 Conclusion This brief overview provides only a partial picture of the syntactic phenomena of Atlantic creoles. Many of these phenomena remain relatively underresearched in particular creoles, and there are many unanswered questions about their internal structure and their sources. Perhaps the most vital of these continues to be the issue of origin. It has become clear that no single explanation of the sources of creole grammar is adequate on its own. Rather, we must weigh the contributions of both superstrate and substrate languages, and assess the role played by universal principles of acquisition, as well as internally motivated changes. The interaction among these factors explains many of the similarities, as well as the differences, found among creoles of different lexical affiliation. One thing that has become clear is that there are no syntactic phenomena peculiar to creoles as a type of language. Most of them appear to have models in one or another of the input languages, though creole creators reshaped

44 Donald Winford them in various ways. It has also become clear that the processes that gave rise to creoles are the same as those that have affected the evolution of all languages. Consequently, creoles do not constitute a “special” or unique type, either in terms of their origins or in terms of their synchronic structure. These facts have various implications for the ways in which the study of creole syntax can contribute to linguistic theory in general. In the first place, exploration of the processes of restructuring that shaped creole grammar promises to enrich our understanding of creole formation and its relationship to first and second language acquisition. The interaction between superstrate and substrate inputs can tell us much about the nature of contactinduced change in general, and the principles that regulate its outcomes. The internal evolution of creole grammar also provides a rich testing ground for theories of grammaticalization, as work on phenomena such as relativization, complementation, and so on has shown. None of these avenues of inquiry can be pursued without a solid understanding of contemporary creole syntax. More generally, the creation and design of creole grammar raises important questions concerning the design of human language in general. One issue that has not attracted the attention it deserves is whether creoles can clarify the distinction between the unmarked core and the more complex periphery of syntactic structure (see Muysken 1988). Does the architecture of creole syntax reflect what would be predicted by theories of markedness or by claims about the workings of Universal Grammar? The research of the last decades has provided a strong foundation for pursuit of these questions, by bringing the study of creole syntax more into the mainstream, and face to face with the central issues of syntactic theory. The next decade promises to increase our understanding of these issues and provide deeper insight into the workings of creole grammar.

NOTES I wish to express my gratitude to Michel DeGraff and Arthur Spears and his students for providing me with data from Haitian Creole, as well as comments on several points. Thanks also to the volume editors for their helpful suggestions. 1 I have modified his labels and classification somewhat: my types 1-2-3-4 correspond to this types 1-3-2-4.

REFERENCES Alleyne, Mervyn C. (1987) Predicate structures in Saramaccan. In: Mervyn C. Alleyne (ed.) Studies in Saramaccan Language Structure (Caribbean Culture Studies 2).

Atlantic Creole Syntax 45 Amsterdam/Mona, Jamaica: University of Amsterdam/University of the West Indies, pp. 71–87. Allsopp, Richard (1983) The creole treatment of passivity. In: Lawrence Carrington (ed.) Studies in Caribbean Language. St Augustine, Trinidad: Society for Caribbean Linguistics, pp. 142–54. Amastae, Jon (1983) Agentless constructions in Dominican Creole. Lingua 59, 47–75. Andersen, Roger W. (1984) The One to One Principle of interlanguage construction. Language Learning 34, 77–95. Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) (1995) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins. Bailey, Beryl (1966) Jamaican Creole Syntax: A Transformational Approach. Cambridge: Cambridge University Press. Bailey, Beryl (1968) Jamaican Creole Language Course. US Peace Corps. Contract No. PC-25-1501. Bickerton, Derek (1980) Decreolization and the creole continuum. In: Albert Valdman and Arnold Highfield (eds.) Theoretical Orientations in Creole Studies. New York: Academic Press, pp. 109–28. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984) The language bioprogram hypothesis. The Behavioral and Brain Sciences 7, 173–88. Bruyn, Adrienne (1995) Grammaticalization in Creoles: The Development of Determiners and Relative Clauses in Sranan. Amsterdam: Institute for Functional Research into Language and Language Use (IFOTT). Bruyn, Adrienne (1996) On identifying instances of grammaticalization in creole languages. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Relating to Grammaticalization in Contact Languages. London: University of Westminster Press, pp. 29–46. Budhai, Emmogene (2002) Relative Clause Formation in Jamaican Creole. MPhil thesis, University of the West Indies, Jamaica. Byrne, Frank (1987) Grammatical Relations in a Radical Creole. Amsterdam: John Benjamins. Carrington, Lawrence (1984) St. Lucian Creole: A Descriptive Analysis of its Phonology and Morpho-Syntax. Hamburg: Helmut Buske Verlag. Caskey, Alexander F. (1990) Controlling into purpose clauses the creole way. Linguistics 28, 689–712. Cinque, Gugliemo (1999) Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York: Oxford University Press. Craig, Dennis (1980) A Creole English continuum and the theory of grammar. In: Richard Day (ed.) Issues in English Creoles: Papers from the 1975 Hawaii Conference. Heidelberg: Julius Groos Verlag, pp. 111–31. DeGraff, Michel (1992) The syntax of predication in Haitian. In: Proceedings of the TwentySecond Meeting of the North-Eastern Linguistics Society. University of Massachusetts, Amherst: Graduate Linguistic Student Association, pp. 103–17. DeGraff, Michel (1993a) A riddle on negation in Haitian. Probus 5, 63–93. DeGraff, Michel (1993b) Is Haitian Creole a pro-drop language? In: Frank Byrne and John Holm (eds.) Atlantic Meets Pacific: A Global View of Pidginization and Creolization. Amsterdam: John Benjamins, pp. 71–90. DeGraff, Michel (1995) Predicate movement, quantification, events and properties. In: Victor Manfredi and Karl Reynolds (eds.) Niger-Congo Syntax and Semantics 6. Boston: Boston University African Studies Center, pp. 69–82.

46 Donald Winford DeGraff, Michel (2007) Kreyòl Ayisyen, or Haitian Creole (‘Creole French’). In: John Holm and Peter Patrick (eds.) Comparative Creole Syntax: Parallel Outlines of 18 Creole Grammars. London: Battlebridge. Den Besten, Hans, Pieter Muysken, and Norval Smith (1995) Theories focusing on the European input. In: Arends, Muysken, & Smith (eds.), pp. 87–98. Déprez, Viviane and Marie-Thérèse Vinet (1997) Predicative constructions and functional categories in Haitian Creole. Journal of Pidgin and Creole Languages 12, 203–35. Dreyfuss, Gail. R. (1977) Relative Clause Structure in Four Creole Languages. PhD dissertation, University of Michigan. Gooden, Shelome (2000) Stativity and past marking in Belizean Creole. Pre-Conference Bulletin. The Thirteenth Biennial Conference of the Society for Caribbean Linguistics. Mona, Jamaica: Society for Caribbean Linguistics, pp. 171–81. Jaganauth, Dhanis (1988) Time Reference in Two Creoles: The Non-Referential Component. SCL Occasional Paper 26. St. Augustine, Trinidad: Society for Caribbean Linguistics. Jaganauth, Dhanis (2001) The use of se in Jamaican. In: Pauline Christie (ed.) Due Respect: Essays on English and English-Related Creoles in the Caribbean in Honour of Professor Robert Le Page. Mona, Jamaica: University of the West Indies Press, pp. 135–54. Keenan, Edward L. and Bernard Comrie (1977) Noun phrase accessibility and universal grammar. Linguistic Inquiry 8, 63–99. Koopman, Hilda (1982) Les constructions relatives. In: Claire Lefebvre, Hélène MagloireHolly, and Nanie Piou (eds.) Syntaxe de l’haïtien. Ann Arbor: Karoma, pp. 167–203. Koopman, Hilda (1984) The Syntax of Verbs: From Verb Movement Rules in the Kru Languages to Universal Grammar. Dordrecht: Foris. Kouwenberg, Silvia (1990) Complementizer PA, the finiteness of its complements and some remarks on empty categories in Papiamento. Journal of Pidgin and Creole Languages 5, 39–51. Kouwenberg, Silvia (1995) Berbice Dutch. In: Arends, Muysken & Smith (eds.), pp. 233–43. Kouwenberg, Silvia and Eric Murray (1994) Papiamentu (Languages of the World/ Materials 68). Munich: Lincom Europa. LaCharité, Darlene and Jean Wellington (1999) Passive in Jamaican Creole: Phonetically empty but syntactically active. Journal of Pidgin and Creole Languages 14, 259–83. Langacker, Ronald W. and Pamela Munro (1975) Passives and their meaning. Language 51, 789–830. Law, Paul (1995) L’extraction du sujet à l’interrogatif en créole haïtien. Linguistique Africaine 14, 47–59. Lumsden, John (1990) The biclausal structure of Haitian clefts. Linguistics 28, 741–59. Manfredi, Victor (1993) Verb focus in the typology of Kwa/Kru and Haitian. In: Frank Byrne and Donald Winford (eds.) Focus and Grammatical Relations in Creole Languages. Amsterdam: John Benjamins, pp. 3–51. McCawley, James D. (1988) The Syntactic Phenomena of English, vols 1 and 2. Chicago: The University of Chicago Press. Migge, Bettina (1998) Substrate influence in creole formation: The origin of give-type serial verb constructions in the Surinamese Plantation Creole. Journal of Pidgin and Creole Languages 13, 215–65. Muysken, Pieter (1988) Are creoles a special type of language? In: Frederick J. Newmeyer (ed.) Linguistics: The Cambridge Survey, vol. 2. Cambridge: Cambridge University Press, pp. 285–301.

Atlantic Creole Syntax 47 Muysken, Pieter and Norval Smith (1990) Question words in pidgin and creole languages. Linguistics 28, 883–903. Piou, Nanie (1982) Le redoublement verbal. In: Claire Lefebvre, Hélène Magloire-Holly, and Nanie Piou (eds.) Syntaxe de l’haïtien. Ann Arbor: Karoma, pp. 152–66. Plag, Ingo (1993) Sentential Complementation in Sranan. Tübingen: Max Niemeyer. Roberts, Peter A. (1993) West Indians and their Language. Cambridge: Cambridge University Press. Ross, John Robert (1967) Constraints on Variables in Syntax. PhD dissertation, MIT. Published in 1986 as Infinite Syntax, Hillsdale, NJ: Erlbaum. Sebba, Mark (1987) The Syntax of Serial Verbs. Amsterdam: John Benjamins. Singler, John Victor (ed.) (1990) Pidgin and Creole Tense-Mood-Aspect Systems. Amsterdam: John Benjamins. Smith, Norval S. H. (1987) The Genesis of the Creole Languages of Suriname. PhD dissertation, University of Amsterdam. Smith, Norval S. H. (1996) WE-focus in Saramaccan: Substrate feature or grammaticalization? In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Relating to Grammaticalization in Contact Languages. London: University of Westminster Press, pp. 113–28. Spears, Arthur K. (1990) Tense, mood and aspect in the Haitian Creole preverbal marker system. In: Singler (ed.), pp. 119–42. Sylvain, Suzanne (1938) Creole tales from Haiti. Journal of American Folklore 51, 219–346. Taylor, Douglas (1963) The origin of West Indian creole languages: Evidence from grammatical categories. American Anthropologist 64, 800–14. Taylor, Douglas (1971) Grammatical and lexical affinities of creoles. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 293–6. Veenstra, Tonjes (1996) Serial Verbs in Saramaccan: Predication and Creole Genesis. Dordrecht: ICG Printing. Veenstra, Tonjes and Hans den Besten (1995) Fronting. In: Arends, Muysken & Smith (eds.), pp. 303–15. Williams, Wayne R. (1977) The so-called relativized and cleft predicates in Krio: One step closer to an understanding of creolization. In: Paul Kotey and Haig DerHoussikian (eds.) Language and Linguistic Problems in Africa. Columbia, SC: Hornbeam Press, pp. 467–78. Winford, Donald (1988) Stativity and other aspects of the creole passive. Lingua 76, 271–97. Winford, Donald (1993) Predication in Caribbean English Creoles. Amsterdam: John Benjamins. Winford, Donald (1997) Property items and predication in Sranan. Journal of Pidgin and Creole Languages 12, 237–301. Winford, Donald (2000a) Tense and aspect in Sranan and the creole prototype. In: John McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 383–442. Winford, Donald (2000b) Irrealis in Sranan: Mood and modality in a radical creole. Journal of Pidgin and Creole Languages 15, 63–125.

48 Miriam Meyerhoff

3

Forging Pacific Pidgin and Creole Syntax: Substrate, Discourse, and Inherent Variability MIRIAM MEYERHOFF

1

Introduction

Two processes are fundamental to a discussion of the syntax of Pacific pidgins and creoles. The first is the interplay between the substrate languages of the region and the discursive and communicative needs of speakers, which helps to shape the contact languages of the Pacific. The second is the role that systematic variation plays in shaping the grammar. Over the years, the first process – especially the influence of the substrate – has been emphasized by a number of researchers working on Pacific pidgins and creoles (henceforth, Pacific P/Cs). However, the second process is less thoroughly integrated into research in the region than it is, for example, in the study of Atlantic creoles. This chapter will try to explore how aspects of Pacific P/C syntax are usefully informed by these two factors. The chapter has the following structure. First, I present an overview of the contact languages that have been recorded in the Pacific, giving references to primary literature since there are far too many languages to discuss each in detail in this chapter (section 2). In section 3, I focus on examples where the synchronic structure of Pacific P/C syntax clearly illustrates the importance of analyzing these languages in terms of lengthy or ongoing contact with substrate languages; I further undertake a critical evaluation of some of the strongest claims for substrate influence on Pacific P/C syntax and consider the proposition that creoles (not just pidgins) can be said to be somehow “simpler” syntactically than their input languages. I suggest that this is a matter of perspective and does not hold up if the entire syntactic system of a Pacific P/C is evaluated

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Forging Pacific Pidgin and Creole Syntax 49 and if probabilistic features of the syntax are taken into account (section 4). In section 5, I present several case studies which argue that a synthesis of factors – substrate, discourse, and more – provides analyses that are most accountable to the data, giving the fullest account of the patterns observed. Finally, I offer some concluding remarks drawing the preceding sections together. In the course of this discussion, some of the major issues that have been debated in Pacific P/C syntax will surface: the nature of so-called “predicate marking” or subject agreement in Melanesian Englishes, focus marking, pronominal systems and the marking of inclusive/exclusive distinctions, and the semantics of verbal auxiliaries.

2 Language Contact in the Pacific Despite the absence of hard records, we must suppose, because of the extensive mobility and high linguistic diversity of the Pacific, that there has been language contact there as long as there have been languages. We may also suppose that contact then, as now, resulted in community-wide or individual bi-/multilingualism or perhaps in the creation of distinct contact varieties. We know, for instance, that Motu, Dobu, Siassi, and Gawa (Kawa) all served as vehicular languages in the trading circuits of Papua New Guinea (PNG) before the colonial period (see Sankoff 1977a). Vehicular Motu and Siassi are clearly identified as distinct trading varieties by their local names, namely tok Siassi haphap ‘broken Siassi language’ and Hiri Motu possibly meaning ‘trading voyage Motu’ (Sankoff 1977a, p. 271, quoting Harding 1967 – though this etymology has been queried; see references in Smith 2002). Vanuatu, too, had well-established internal trading circuits before European contact (Bonnemaison 1986), so it is possible that vehicular varieties of certain languages emerged there as they did in PNG, although of course people may have acquired multiple languages rather than creating contact varieties of local languages; Crowley (1990, pp. 48–9) notes that high levels of multilingualism can still be found in Vanuatu today. Certainly, language contact has had major effects on the syntax of individual languages in the region, even if a pidgin or creole stage is not postulated. Grant (2001) notes that Takia (spoken on Karkar Island, PNG) has structural traits typical of both Austronesian and non-Austronesian languages. He suggests that this resulted from a prior period when bilingualism in Takia (an Austronesian language) and Waskia (non-Austronesian) was widespread. Written records of language contact in the Pacific are only available after European contact, starting in the late eighteenth century (Drechsel 1999), and increase in their detail thereafter. The number of contact languages documented as being in use in the Pacific and peri-Pacific area can be startling. The following provides some idea of the richness we are talking about. (References are given here for languages that are not discussed in more detail below.)

50 Miriam Meyerhoff Obviously, there are P/Cs lexified by colonial languages. For example, English lexified Bislama in Vanuatu, Pijin in the Solomon Islands, Tok Pisin in Papua New Guinea, Ngatik Men’s Creole used on Ngatik (Sapuahfik) Atoll in the Federated States of Micronesia (SIL International 2002), Broken and Kriol in Australia (Harris 1991 and references therein), Pitkern (Pitcairn Island, Källgård 1993; Mühlhäusler 1998), Nauru (Siegel 1990), and Pidgin in Hawai[i. French lexified Tayo in New Caledonia, German lexified Unserdeutsch (Mühlhäusler 1984) in PNG, and Spanish lexified varieties of Chabacano in the Philippines. (Reinecke 1975, p. 338, discusses why there were no Dutch-lexified P/Cs in Indonesia.) Long (2002) provides the only discussion I am aware of in a European language of linguistic varieties emerging from contact between Japanese and other East Asian languages. In addition, there are historical records of P/Cs in the Pacific with the following lexifiers: Hawaiian, Maori, Dehu (New Caledonia, Drechsel 1999, pp. 86–7), Pohnpeian (Drechsel 1999, p. 87), Fijian and Hindi (Siegel 1987), Aleut (Copper Island/Mednyj Aleut, Thomason 1997), Chinook (Silverstein 1972a, 1972b; Thomason 1983), Vietnamese (Tây Bôi, Reinecke 1971), various non-Austronesian languages in PNG (especially Yimas, Williams 1993, 2000) and Tetun (Tetum Prasa or Tetum Dili, East Timor, SIL International 2002). In the peri-Pacific area, SIL International’s Ethnologue (2002) records a number of vehicular varieties of Indonesian and Malay as well as Betawi (spoken in Java and Bali), Portuguese-lexified Macanese in Macau and Hong Kong, and Papia Kristang (spoken in peninsular Malaysia and Singapore). In some cases, there is dispute about the stability of the codes and hence their status as jargons or P/Cs. These terms may mean slightly different things in the hands of different researchers. Clark (1990), for example, differentiates a jargon from a pidgin principally in terms of who uses the code and when. For him, a jargon is relatively stable but used between linguistic groups, while a pidgin is essentially a jargon that has come to be used within a non-native speaking community, though he concedes that this presupposes a pidgin having “a new degree of coherence and stability” (1990, p. 106). Drechsel (1999) on the other hand seems to distinguish jargon from pidgin primarily in terms of the relative regularity of the latter, speaking of a jargon as “highly variable” (1999, p. 91). Trudgill’s “pre-pidgins” (2002, p. 69) seem to correspond to what many people might call jargons, though there is a teleological dimension to Trudgill’s terminology that isn’t necessarily there in Clark’s or Drechsel’s. Clearly, the distinction between a jargon and a pidgin is hard to pin down on linguistic grounds, and it is not actually crucial to this chapter (see discussion in Bakker, this volume). The principal distinction between pidgin and creole relevant to this chapter is the social elaboration of the codes. I take pidgins to have limited social functions, while I will refer to languages that are used for the full range of social functions as creoles. It is also important to note that the languages referred to as Pidgin (Hawai[i), Pijin (Solomon Islands), and (Tok) Pisin (PNG) clearly have the social functions and structural stabilization definitional of creoles. I follow local usage, however, in referring to them by these names.

Forging Pacific Pidgin and Creole Syntax 51 Many of the Pacific contact languages have independent histories. Some, though, are clearly linked, and for others there is debate about their independence from each other. There is strong evidence that there was an Englishbased form of communication in widespread use throughout the Pacific in the nineteenth century (Clark 1979) and that this and especially the contact English spoken in Eastern Australia (Baker 1993) helped shape all the English-lexified Pacific P/Cs (but cf. Goodman 1985, 1999, and references therein; Holm 1988; and discussion in Veenstra, this volume). In the latter part of the nineteenth century, the labor trade (especially that centered in Queensland) had a significant effect in shaping the P/Cs now spoken in Melanesia.

3 The Theoretical Importance of Ongoing Contact with the Substrate Let me define what I mean by the “substrate.” For the purposes of this discussion, the substrate can be most usefully seen in opposition to the lexifier language. There is generally one language that has contributed the bulk of the lexicon (hence “lexifier”) to the languages we are about to examine more closely, but there are often numerous languages that were (and continue to be) acquired as the first language (L1) of many if not all speakers of the P/C speech community. These languages can be considered the substrate to the P/C on both temporal and conceptual grounds: they are generally the language spoken first; moreover, their influence on the structure of the Pacific P/Cs may be less transparent than the influence of the lexifier. There are some striking similarities between the syntax of Pacific P/Cs and their substrate languages. Indeed, they can be so striking that some researchers have made quite strong claims for relexification (Keesing 1988; Camden 1979; Charpentier 1979; Kihm 1995; Corne 2000). The sociolinguistic histories of Pacific P/Cs, which have involved a long period in which P/Cs were neither the first nor the only language of most speakers, mean that continued contact with the substrate is a major factor to deal with. Derek Bickerton argued (e.g., 1981) that varieties of Melanesian English could not be used as tests of the effects of an innate language bioprogram for this reason. To Bickerton, the syntax of the Bislamic languages (Pijin, Bislama, and Tok Pisin) was irrelevant to the bioprogram because Eastern Oceanic languages have been spoken for so long in conjunction with these Pacific P/Cs that the conditions of abrupt language shift required to trigger the bioprogram are not met. In fact, Bickerton’s theory about the role of universals and theories that emphasize the influence of the substrate are not as diametrically opposed as some of the key players have cast them. Aside from the excellent collection in Muysken and Smith (1986), it is worth noting that a lot of recent research in second language acquisition deals directly with the question of how an L1 grammar interacts with Universal Grammar when people learn or acquire a

52 Miriam Meyerhoff second language (e.g., Archibald 2000; see also Siegel, this volume; Mesthrie, this volume). In section 3.1, I review some examples of Pacific P/Cs where it seems that the effect of the substrate has been considerable. In 3.2, I then go on to evaluate how much parallelism in form and function is necessary in order to claim substrate influence.

3.1

Reflexes of substrate languages on Pacific P/C syntax

The first example I will look at is Zamboagueño spoken in the Philippines. Zamboagueño is one variety of what is sometimes collectively known as Chabacano or Chavacano (< Spanish chabacano ‘cheap’, see Lorenzino 1993, p. 400, hence a name not favored by all speakers of these varieties). The language derives much of its lexicon from Spanish, but it shows some very clear syntactic evidence of its Austronesian substrate. Thus, Forman (1972, 1993) observes that, like Tagalog and the Visayan languages, Zamboagueño tends toward VSO word order, not SVO as in Spanish (though note that VS is certainly possible in Spanish). This is clearly seen in (1)–(3): (1) Ya-keda kami el mucaca. asp-become 1p.excl do maid ‘We became the maid [i.e., took over her duties].’ (Forman 1993, p. 179) (2) Ay-lyeba bende si tóng el asúkar. asp-carry sell sbj.marker Tong do sugar ‘Tong will take the sugar to sell.’ (after Forman 1993, p. 169) (3) Bali-bali kel kuhida di ayer. very.good that catch prep yesterday ‘Yesterday’s catch was quite valuable.’ (Molony 1977, p. 145) In addition, (2) shows that when the post-verbal subject is a full NP it is preceded by the morpheme si in Zamboagueño. This is identical in function and even form with the subject marker in Tagalog (Forman 1993, pp. 166–7). Another interesting and clear transfer from the Austronesian substrate is the distinction between first person plural inclusive and exclusive. As (1) shows, Zamboagueño not only draws the semantic distinction from the substrate, but even the form, kami. Lipski (1987, p. 97) notes that Spanish-derived ustedes is now being introduced as the [+respect] form, while Austronesian kamó is becoming the second person plural [–respect] form. Keesing’s (1988) study of the structure of Solomons Pijin makes one of the strongest claims for the role of speakers’ substrate languages in shaping the synchronic structure of a Pacific P/C. Keesing argues that the following are all most economically accounted for by attributing them to transfer from the

Forging Pacific Pidgin and Creole Syntax 53 substrate: the use of a third person pronoun as a plural marker in NPs; the use of a suffix to transitivize a verb; the semantic distinctions marked in the pronoun inventory; the use of (what Keesing analyzes as) a resumptive pronominal at the left edge of a predicate; the semantics and form of overt tense, aspect, and mood markers; and even tendencies with respect to word order. So, for example, Keesing saw parallels between constituents in Kwaio (a language of Malaita in the central Solomons) and both the semantics and word order position of Pijin bae(bae), signaling a future event, and tes/des, which I interpret from Keesing’s discussion to be a proximative temporal marker.1 Keesing argues the crucial data are in (4), which shows that Pijin is not using des/tes like the English temporal adverb ‘just’, but rather using it as a verbal auxiliary like Kwaio bi’i, (5): (4) Pijin Baebae iumi des luk-im. fut srp(1p.incl) just see-tr (5) Kwaio Ta-goru bi’i aga-si-a. fut-srp(1p.incl) just see-tr-3s both: ‘We’ll see it in a while.’ (Keesing 1988, pp. 214–15) Keesing is by no means alone in seeing substrate languages as being important to the structure of Pacific P/Cs. Camden (1979) and much of Charpentier (1979, esp. p. 361) explore the substrate influences pervading the grammar of Bislama. Verb serialization is another good candidate for substrate influence in Pacific P/Cs. Examples of what most people would recognize as serial verb constructions (SVCs) are readily found in Oceanic languages (Durie 1988) and similarly prototypical SVCs are also found in Pacific P/Cs. The SVCs found in Bislama, for example, very closely match the patterns found in the Eastern Oceanic languages of Vanuatu in two ways. First, the kinds of semantic relations that are marked in a Bislama SVC are the same as in most of the substrate languages (viz., ‘stay’, ‘go’, ‘come’). Second, the subject agreement marking on the second verb in the series in Bislama is cross-linguistically rather unusual, but mirrors the agreement marking found in a subset of the substrate languages (Meyerhoff 2001). The semantic parallelism on its own is weak evidence for substrate transfer. Thus, Early (2000) notes that in many languages, not just Austronesian ones, ‘stay’ and basic posture verbs occur as auxiliaries and as serial verbs. This means that the semantic parallels between the relations expressed by Bislama SVCs and the substrate languages may arise from general facts about the way human languages encode human experience (see also Romaine 1999). It does, however, seem eminently plausible and natural to suppose that multilingual speakers will actively draw on the syntactic resources of languages they are familiar with in developing a P/C (Faraclas 2002), and the formal

54 Miriam Meyerhoff parallelism in the use of the agreement marker strengthens the claim for substrate transfer.

3.2

Weighting potential substrate influence: Examples from Tayo

In this section, I will focus on Tayo, the French-lexified creole spoken in New Caledonia which has become the pet example for quite strong arguments in favor of the substrate as the primary determinant of Pacific P/C syntax. Yet, as we will see, the arguments adduced raise some interesting questions about how much weight to give to substrate parallels, especially when there are limited matches of substrate form and function. Tayo emerged following substantial in-migration to the area around the Marist mission at St-Louis, close to Nouméa (New Caledonia). It stabilized around 1855–80 as the medium of communication among speakers of a number of New Caledonian languages (principally those classified as Centre North and Far South). In addition, a number of French convicts, Indians from Réunion, and later Japanese, Javanese and Vietnamese laborers, also moved into the area (Corne 1990, pp. 5–6; Ehrhart 1993). Corne’s thorough pursuit of the substratist position leads him to propose some interesting and quite specific hypotheses about the linguistic effects of different social and demographic factors in situations of language contact (Corne 1994, 1995). His work on Tayo is worth looking at in some detail because he tries to form a coherent picture of substrate influence on the grammar as a whole (cf. Camden 1996; DeGraff 2001). This provides a rich sense of the processes of language contact, but it also raises some difficult questions about how we should weight aspects of the substrate grammar(s) that do and do not seem to transfer. I will illustrate this problem with two examples, Corne’s discussion of relative clauses and Kihm’s discussion of possession and TMA marking. Corne concludes that in Tayo “there can be no doubt that relativization . . . [is a] diluted reflex of Melanesian patterns” (1994, p. 293) and he supports this with data from Cèmuhî and Nráa Drùbea2 because their speakers were respectively the first, and the most plentiful early settlers in St-Louis (cf. Mufwene’s 1996 Founder Principle). Yet, if we examine the Cèmuhî and Nráa Drùbea data closely, Corne’s use of the word “diluted” becomes crucial. The parallels between Tayo and the substrate languages are strong, but so are parallels with the lexifier. In Tayo, when the relativized constituent is the subject of the relative clause, there is an obligatory subject index (si) in the relative (a si can be thought of as a kind of agreement morpheme or clitic pronoun), e.g., (6): (6) Si nu a di kekechos sa le move. if 1p irr say something rel si bad ‘If we said things which were bad.’ (Corne 1994, p. 287)

Forging Pacific Pidgin and Creole Syntax 55 Corne notes a comparable si with a subject head of a relative clause in Cèmuhî but not in Nráa Drùbea (1994, p. 292). But even in Cèmuhî, the rule appears to apply less categorically than in Tayo: “subject relativization allows a following subject indexing pronoun” (1994, p. 294, my emphasis). Corne claims that such subject indexes are not found in either “popular or standard” French (p. 292), and therefore the only plausible source for the Tayo pattern is the substrate. However, the facts about vernacular French are not so clear-cut. Nonstandard spoken French does allow a clitic pronoun with an extracted subject. The example below, accepted by native speakers of both European and Canadian French, shows this: (7) Si nous avons dit des choses qu’elles sont mauvaises. ‘If we said things which (they) were bad.’ Thus, it may be a little hasty to ascribe the Tayo relativization patterns to substrate transfer. Because the substrate models invoked seem to have a degree of variability not found in Tayo, Tayo looks no more like them than it looks like nonstandard forms of the lexifier. Another example illustrating the difficulty of weighting substrate influence is that of possession marking in Tayo, as discussed in Kihm (1995). Three different prepositions are used, pu (< French pour) when the complement is animate, de (< French des) when it is inanimate and a (< French en) when it denotes materials (Kihm 1995, pp. 244–5). The animate/inanimate distinction seems a likely feature to have transferred from a Melanesian substrate. Yet if it is transfer, it is subtle and complex. Like Tayo, Nráa Drùbea has three different ways of modifying a head noun with another noun, but in Nráa Drùbea they do not all involve a preposition, nor do the strategies encode the same semantic distinctions as in Tayo. Kihm also makes an important methodological point, namely, the need to analyze both the unmarked and the marked. For instance, Kihm’s discussion of the Tayo TMA system first considers the many meanings associated with bare, unmarked verbs. This analysis then becomes the backdrop to an evaluation of overt TMA markers, i.e., the completive fini (< French fini, ‘finished’), the future va (also wa, a < French va(s), ‘go’), and a progressive atra-de (< French en train de, ‘in the middle of’) (Kihm 1995, p. 238). Kihm convincingly shows that what emerges in Tayo’s TMA system does not correspond to the semantic categories supposed to emerge from the Language Bioprogram Hypothesis (see Veenstra, this volume). For example, Kihm shows that Nráa Drùbea marks very similar semantic categories in the verb phrase and, like Tayo, it does so immediately before the main verb. However, he cautions that there may be differences in the principal functions of the corresponding lexical items in both languages (1995, p. 246). Furthermore, his data show that not all Tayo sentences translated with a progressive are marked with atra-de (8); nor are all utterances translated with a future marked with va (9):

56 Miriam Meyerhoff (8) Ta ekri kwa? 2s write what ‘What are you writing?’ (after Kihm 1995, p. 237) (9) Tu ale u taler? 2s go where later ‘Where will you go later?’ (after Kihm 1995, p. 237) Again, as with the relative clauses, we have the problem of weighting the significance of matches and mismatches with the different substrate grammars and the possible effects of the communicative context. There may simply be problems with the translations in cases like these, but they do suggest that there may be more variability here than meets the eye. In short, when we are tempted to make claims of substrate transfer we need to be sure that we are actually comparing like elements, that is, those that are alike in their meaning when used and those that are alike in their distribution (cf. Kouwenberg 2001, who points out that similarity is a metric for which analysts have no agreed, objective measures). Tayo seems to invite comparison with its substrate languages on some formal levels, but the semantic functions of the Tayo forms are not direct calques of those in the substrate. The lesson perhaps to be learned from Tayo is that we should be cautious before attributing a P/C form unqualifiedly to substrate transfer. And as Kihm shows, it is important that we consider those aspects of the syntax of Pacific P/Cs that are not overtly marked as well as those that are.

3.3

“Simpler” than the substrate and lexifier?

Arguments that P/Cs constitute a distinct typological class of human languages often fall back on the rhetorical theme of simplicity.3 This holds that P/Cs are structurally “simpler” than non-P/C languages. Much of this argumentation refers to morphology and so lies beyond the scope of this chapter (see Crowley, this volume), but certain comments can be made about this argument with respect to syntactic features of Pacific P/Cs. DeGraff (2001, esp. pp. 250–74) argues that “simplicity” is a dangerously under-specified notion on which to base linguistic description and theory. The metrics by which any language or any structure within a language can be argued to be “simpler” than comparable languages or structures remain poorly defined and tend to appeal to vague or intuitive (and hence possibly cultureand language-dependent) notions of whether something is more or less simple (cf. Muysken 1988). Moreover, even widely agreed upon intuitive notions of simplicity fail to consistently pick out as a class those languages that are considered on sociohistorical grounds to be P/Cs. At the same time, instances of apparent P/C-like simplification can be found in languages that have not had the same sociohistorical background as P/Cs. There is no space to rehearse arguments about simplicity here; my own belief is that when we are talking

Forging Pacific Pidgin and Creole Syntax 57 about pidgins it is probably a useful notion because pidgins are clearly restricted in social function. However, it is a less useful notion when talking about the functionally expanded codes I would call creoles. The following examples from the syntax of Pacific P/Cs illustrate some of the problems with the notion that Pacific creoles are structurally “simpler” than their substrates or lexifiers. The first example concerns the marking of (in)alienable possession in Tok Pisin and Bislama. The substrate languages in Melanesia systematically distinguish between alienable and inalienable referents and generally mark the distinction in the way in which they grammaticalize possession. Examples (10)–(11) show the difference in Tamambo, the language spoken on Malo in northern Vanuatu. ‘Mother’ is considered an inalienable possession in Tamambo; ‘broom’ is considered alienable and, unlike ‘mother’, appears in a periphrastic possessive: (10) tina-na mother-3s.poss ‘his/her mother’ (11) no-na itevi cl-3s broom ‘his/her broom’ (Jauncey 1997, p. 226) At first blush it would appear that the distinction between inalienable and alienable possession is not made in the Pacific P/Cs spoken in the region. Dutton and Brown specifically say this for Hiri Motu (1977, pp. 773–4). However, Sankoff and Mazzie (1991) show that the inalienable/alienable distinction emerges in subtle and unexpected ways in Tok Pisin. Their analysis of noun phrases in Tok Pisin found a statistically significant difference in the form in which inalienably possessed entities are realized. Inalienably possessed referents were more likely to occur with an overt phrase marking possession, e.g., brata bilong en [brother of him] ‘his brother’, than alienable referents were. Meyerhoff (2002) shows an even more evanescent grammaticalization of (in)alienability. In theory, any object can be phonetically null or overt in Bislama, but in practice inalienably possessed referents are significantly more likely to be overt object NPs. So, for example, suppose you had the sentences in (12) and (13): (12) Bubu blong mi i no save draeva. grandparent poss 1s agr neg abil drive ‘My grandmother/father doesn’t know how to drive.’ (13) Tija blong Klas Faev i no save draeva. teacher poss class five agr neg abil drive ‘The teacher of Class 5 doesn’t know how to drive.’

58 Miriam Meyerhoff The speaker might continue: (14) Bae mi tijim (hem/Ø). irr 1s teach 3s/Ø ‘I’m going to teach him/her.’ In theory, the pronoun and the null argument are both options whether you are referring to your bubu or to the teacher. But if (14) follows (12) and the object in (14) refers to your bubu, it is statistically more likely to be overt, e.g., the pronoun (hem), than it is to be null. If, on the other hand, you are talking about the tija, then the object in (14) is no more or less likely to be null than it is to be overt. These findings suggest two things. First, that there are a number of different levels on which we need to investigate the structure of a Pacific P/C before we can claim with certainty that a particular semantic feature is or is not grammaticalized (cf. Camden’s (1996) point about analyzing these languages as whole systems). Second, they suggest that we should be pretty tough with ourselves in defining what we mean by “simpler.” Bislama may not mark inalienable possession with bound morphemes on nouns, but if it marks it as a probability in discourse, I for one would not want to argue that Bislama or the substrates are more or less simple than the other. Another problem is that arguments about the simplicity of P/Cs generally rest only on observations about the leveling of linguistic forms, and they do not weigh up the sociolinguistic and social psychological complexity of negotiating leveled forms through interaction. Dutton and Brown (1977, p. 763), for instance, suggest that an analytic post-verbal negator occurs in Hiri Motu instead of the synthetic negatives used in Motu because analytic forms are used in Toaripi and Kiwai, the languages most frequently encountered on the hiri (the trade voyage). That is, they presuppose a role for interspeaker accommodation in shaping the syntax of Pacific P/Cs (on accommodation theory, see Thakerar, Giles, & Cheshire 1982; Giles & Coupland 1991). But in general, accommodation has not been well explored in accounts of creolization (some exceptions are Le Page & Tabouret-Keller’s 1985 classic work; Kouwenberg 1992; Trudgill 2002; Meyerhoff & Niedzielski 1994; S. Roberts 2004). Again, it is not clear to me how the social and psychological skills required to negotiate a novel, regularized medium of communication through processes of interspeaker accommodation are a priori “simpler” than the skills required for a speaker to attempt bilingual mastery of the socially dominant code. That is, we might have a very different view of the syntax of Pacific P/Cs if we took social motives and linguistic variation as a starting point instead of surface structure. Consider too questions of word order. One can argue that SVO word order is inherently simpler than VSO in that it is structurally unmarked (I. Roberts, 1999). Hence SVO word order in pidgin forms of Hawaiian and Maori might be seen as simplifications compared to the VSO order of Hawaiian (15) and Maori (16):

Forging Pacific Pidgin and Creole Syntax 59 (15) a.

Pidgin Hawaiian: Awa, oe kuai lama kela pake? Awa you sell liquor that Chinese b. Hawaiian: E Awa ua ku’ai aku ‘oe i ka lama i kela pake? voc A. pf sell dir you om the liquor om that Chinese both: ‘Ah Wa, did you sell liquor to that Chinaman?’ ( J. Roberts, 1995, p. 39)

(16) a.

Pidgin Maori (Clark, 1990, p. 103, quoting Savage 1807): Tungata nue nue kikie. man big big eat b. Maori (Mary Boyce, p.c.): He nui nga kai a tena tangata. pm big det food poss that man both: ‘That man has plenty to eat.’

But SVO is not necessarily unmarked typologically, since there are more languages in the world with SOV word order than with SVO. Whether or not SVO should be considered the simplest word order depends on the measures used. However, it is clear from these examples that pidgin forms of Hawaiian and Maori lack the case markers and the auxiliaries that are so important in the non-vehicular varieties. Clark notes that there is a general absence of functional elements in the pidgin English used in New Zealand at that time, except that ‘the’ is reported in the English spoken by the Maori. He speculates that this may be because ‘the’ is similar in both form and function to the Maori definite article te (1990, p. 111). (Note that this seems to suggest that the nonstandard nineteenth-century pronunciation of interdental fricatives was a voiceless stop, not the voiced stop heard in twentieth-century Maori English.) Finally, even if pidgins are a natural class of simple languages, this simplicity is quickly left behind once a Pacific P/C has stabilized enough to become a community language (whether the sole community language or not). At that point, there is the potential for many aspectual distinctions to emerge through familiar processes of language change and grammaticalization. For example, Shnukal (1988) gives a wide range of aspect and mood markers in Broken (spoken on the islands in Torres Strait between Australia and PNG). They are a structurally heterogeneous set. For instance, mas ‘must’ (deontic in the affirmative and epistemic in the negative), kin ‘be able’, spostu ‘ought’, and sud ‘should’ seem to only occur immediately after the subject and before tense and negation. On the other hand, kan ‘not able’ and kasa ‘just’ appear to occur immediately before the lexical verb (Shnukal 1988, pp. 43–8). Also, some of what Shnukal (pp. 48–52) classifies as aspect markers vary in their placement. Compare oltaim in (17)–(18) and pinis in (19)–(20) (from Shnukal 1988, p. 49):

60 Miriam Meyerhoff (17) Em i oltaim diskain. 3s agr hab like.this ‘He tends to be like this.’ (18) Oltaim ai bi plante watamelen ya. hab 1s past plant.tr watermelon here ‘I used to plant watermelons here.’ (19) Ai pinis pute suka lo ti blo yu. 1s compl put.tr sugar prep tea of 2s ‘I’ve already put sugar in your tea.’ (20) Wulp bi kaikai ol sip blo em pinis. wolf past eat pl sheep of 3s compl ‘The wolf had eaten his sheep.’ This suggests that oltaim and pinis are still adverbs en route to grammaticalization as auxiliaries. Either way, it is clear that Broken now has ample, and fairly complex, means of marking mood and aspect. The syntax of Broken is also characterized by some basic subject agreement, or predicate marking (see discussion of Bislama below). Shnukal’s texts suggest that i is obligatory in presentational sentences such as the following: (21) I bin gad wan ol man. agr? past have one old man ‘There was an old man.’ (Shnukal 1988, p. 323) The i also occurs at the left edge of the verb phrase with some third person subjects. There is a good deal of variability within and between speakers: in Shnukal’s texts the range is from a high of 88% in one text and 75% in two others, to a low of 7% in a fourth. Within one of the longer texts, Da Ol Man ‘The Old Man’, there is a tendency for i to be absent when any one of the following precede the lexical verb: no ‘negative’, bin ‘past’, the adverb/aspect marker oltaim. Clearly, work building on Shnukal’s important foundational study of Broken is much needed.

4

What Kind of Universals? Another Look at Hawai[i Pidgin

I now turn to a consideration of the role that language universals – or perhaps, better stated, universal tendencies – play in shaping the syntax of Pacific P/Cs. In the next sections we’ll see that there are some rather different notions of universals we can make reference to: Universal Grammar-like structural universals, typological universals, and communicative universals. I will begin

Forging Pacific Pidgin and Creole Syntax 61 with a look at the Pacific P/C most strongly associated with universalist claims, the language known in Hawai[i as Pidgin. Pidgin (“Hawai[i Creole English” in linguistic circles) is spoken natively by most locals in Hawai[i. It holds a special place in the creolists’ canon because it was the keystone of Derek Bickerton’s Language Bioprogram Hypothesis (the LBH; see Bickerton 1981, 1984; Veenstra, this volume). Bickerton argued that child learners were obliged to fall back on the innate syntax of the language bioprogram because the structural input which they heard was extremely variable and unsystematic. This meant that Pidgin creolized very abruptly in Hawai[i. He concluded that the bioprogram provides all humans with a very basic grammar consisting of only nouns and verbs with verbal arguments entering into the most informationally transparent relationship with the verb, i.e., SVO. This simplest syntax mapped onto some basic semantic categories in a way that Bickerton hypothesized was also innate. For instance, the need to distinguish anterior from unmarked tense, irrealis from realis mode, and nonpunctual from punctual aspect meant that TMA markers marking these features would all emerge in abrupt creolization. Bickerton’s provocative thesis and his unflinching argumentation for it over the years have been a lightning rod for many other scholars. His arguments served the very useful purpose of giving creolistics a position and vocabulary with which to engage with formal, generative linguistics, but the validity of his claims, even for Hawai[i Pidgin, have been challenged. Sato (1978) pointed out that there were problem cases in his own texts for Bickerton’s claims that go(n) marked irrealis. More recently, S. Roberts (1998) notes that, historically, parts of the Pidgin TMA system that are supposedly generated by the LBH actually emerged in the speech of both native speakers and non-native speakers at about the same time, raising questions as to how specifically the imperative to grammaticalize these structures is tied to L1 status. In addition, she found that other features of the LBH TMA system emerged in the speech of non-native Pidgin speakers first, raising questions about the key role that highly unsystematic input was supposed to play in forcing speakers to regularize these features. In a more fundamental challenge to Bickerton, Siegel (2000) claims that the substrate languages spoken on Hawaiian plantations are plausible sources for the semantic categories grammaticalized in Pidgin’s TMA system, and for some other features in Bickerton’s creole prototype. However, Siegel concedes that in many cases it is hard to tell whether the outcomes reflect direct substrate transfer or a more fluid system in which many possible sources for a feature combine to facilitate its survival. Similarly, Day’s (1973, p. 131) discussion of the copula in Pidgin notes that the use of ste ‘stay’ as a copula is consistent with the influence of Portuguese estar both phonologically and in the tendency for ste to occur least often with NPs. However, Day rightly notes that the distribution of ste is also consistent with influence from standard English ‘be’, and this can be seen in the spread of ste to the same linguistic domains where ‘be’ is used in standard English.

62 Miriam Meyerhoff Velupillai (2003) provides the most comprehensive recent evaluation of the TMA system of Pidgin. She bases her analysis on a spoken and written corpus of over 269,000 words gathered in all five islands in 1999–2000 – most of which is clearly sourced, publicly available and supplemented by video. Consequently, her conclusions are the most reliable ones for Pidgin (and perhaps any Pacific P/C) to date. Her approach is to consider the function of TMA markers in their discourse context. She confirms that there is minimal overlap in the kinds of TMA categories grammaticalized in Pidgin and the categories grammaticalized in standard English. I cannot review all her findings here, but recommend the study to readers. For example, her data confirm the widespread observation that bare forms of the verb (uninflected for tense as in standard English) are frequent in Pidgin narratives. However, she notes that when a third person singular subject does trigger -s inflection on the verb in Pidgin, this tends to imply either that the event time and speaking time are simultaneous, or to assert a generic truth, as in: (22) ji no w hana minz . . . hana minz wrk. you know what hana means hana means work ‘You know what hana means? . . . Hana means “work”.’ (adapted from Velupillai 2003, pp. 50–1) Velupillai also shows that the distribution of -ing forms in narratives signals a much richer set of aspectual distinctions than are signaled by the standard English progressive. All V-ing forms are progressive, but stei V-ing indicates a much narrower focus on the contextual “now” of a narrative. Compare (23) (without stei), where the utterance expresses a more generic state of affairs, with (24), where there is an immediate focus on the narrative moment. The two forms, she stresses, are not interchangeable. (Parentheses signal that a segment is perceptually unclear.) (23) pipo(l) kra(j)in dae dei laig dei neishin baeg. people shout that they des their nation back ‘People are shouting that they want their nation back.’ (after Velupillai 2003, p. 90). (24) ho ji kam hom dha fon stei ringing tu fawDi. exclam you come home the phone imm ring two forty ‘Ho, you come home and the phone is ringing like crazy.’ (after Velupillai 2003, p. 93). Pidgin also has the option of stei V (without -ing) which Velupillai shows overlaps with both standard English continuous and habitual. Notice that under Velupillai’s account the source of the lexical items realizing these TMA features is a completely separate question. Substrate or lexifier models may have a more or less direct input, but the kinds of semantic

Forging Pacific Pidgin and Creole Syntax 63 distinctions drawn fit in with cross-linguistic tendencies and are independent of the specific linguistic history of Pidgin. The really important question is whether the Pidgin TMA system looks in any sense peculiarly P/C-like, given its critical role in Bickerton’s theory. Velupillai’s study is particularly valuable since she approaches the study of Pidgin as part of a larger, systematic cross-linguistic study of TMA systems in natural languages. She shows that the features found in Pidgin are also found in the majority of languages sampled in a typological study conducted at the Max Planck Institute for Evolutionary Anthropology (Leipzig). She concludes, therefore, that Pidgin provides poor evidence for a structurally unified creole prototype. Of course, we still need to ask how common is the combination of the factors found in Pidgin in languages that would (on sociohistorical grounds) be classified as P/Cs, as compared to languages outside this set, because this is the question that directly addresses the LBH claims. As typological work is increasing in its detail, the answer to this question may soon be forthcoming.

5 Convergent Influences on the Syntax of Pacific P/Cs Perhaps the most promising trend in recent work on Pacific P/Cs is a renewed openness among researchers to try and see syntactic structures as reflecting a convergence of many factors – some linguistic, some instrumental, and some interpersonal. Gillian Sankoff has been pointing out for decades that the most comprehensive and also the most economical account of the structural development of Tok Pisin combines the influence of the substrate, the lexifier, and communicative universals. She argues that “a gradual process of grammaticalization from discourse through syntax is not an alternative to, but is indeed quite compatible with, an initial source involving calquing on a substrate pattern” (Sankoff 1994, p. 298). This has not been a mainstream approach in Pacific P/C linguistics, though Crowley (1990) takes a similarly synthetic approach for Bislama and Masuda (1999) applies this approach to discourse structure in Pidgin. Sankoff has made her case in work on many aspects of Tok Pisin syntax: the development of syntactic strategies demarcating relative clauses (Sankoff & Brown 1976), changes in subject agreement marking in the main verb phrase (Sankoff 1984, 1994), the development of NP focus strategies in Tok Pisin (Sankoff 1993), and the difference between focus patterns in Tok Pisin and in Solomon Islands Pijin and Bislama (Sankoff 2002). Sankoff (1996) is a good introduction to her approach as it discusses several aspects of Tok Pisin grammar. Sankoff and Brown (1976) established an approach in which aspects of the syntax of the substrate, the lexifier, and processing and interpersonal constraints on verbal interaction all feed into the analysis. This approach is

64 Miriam Meyerhoff rightly becoming more influential in the study of Pacific P/Cs, though some scholars emphasize one aspect of the synthesis over others. For example, Romaine (1992) concluded that the effect of the substrate is less important than general processing constraints on the structure of Tok Pisin relative clauses, and Siegel’s recent work (e.g., 2000) seems to me to place greater emphasis on the substrate than on any other factors. However, they share the basic orientation of Sankoff and Brown’s ground-breaking (1976) study. Because of that paper’s importance, it is worth reviewing how it argues in favor of a balance between multiple sources for the way Tok Pisin constructs relative clauses.

5.1

Tok Pisin focus marking and relative clauses

The discussion in Sankoff and Brown (1976) begins with an analysis of the distribution of Tok Pisin ia (< English here), which functions as a focus marker. Evidence is then presented that in some groups of speakers it is grammaticalizing as a means of delimiting the boundaries of a relative clause, thereby separating it from the main clause. Sankoff and Brown argue that the synchronic use of ia as a relativizer derives from the meaning of ia when it was first introduced from English. They point out that here has inherent properties of discourse focus because it is a deictic. They argue that this focus meaning was generalized and that gradually ia began to be used more and more often as a way of focusing hearers’ attention on either “(a) an item that has been mentioned and characterized earlier in the same conversation; or (b) an item that is being mentioned for the first time . . . but that hearers can uniquely identify” (1976; see 1980 reprint, p. 242). They explore in detail the constraints on the use of ia in a corpus of Tok Pisin conversations. The relevant constraints in this case are to do with information packaging, i.e., principles we might like to consider cognitive or functional universals. However, the article goes on to argue that the generalization of ia depended on more than just the semantics of here and principles of information packaging. These factors consolidated because they were reinforced by similar structures found in substrate languages. For example, Buang (an Austronesian language spoken in PNG) is typical of Tok Pisin substrate languages in using a single deictic as a place adverbial, a demonstrative, and a means of introducing relative clauses (p. 254). Thus, three factors – the semantics of the lexifier, synchronic functions and use in discourse, and speakers’ substrate models – are all deemed to mutually reinforce each other, creating an environment that fosters the growth and development of this particular syntactic structure.

5.2

Subject agreement and null anaphora

The importance of the relationship between syntax and discourse for the development of syntactic structure in Pacific P/Cs can also be seen clearly in the way some of the English-lexified creoles have grammaticalized subject agreement and null anaphora. Sankoff (1977b, 1984, 1994) shows that the

Forging Pacific Pidgin and Creole Syntax 65 development of i (variously known as a subject marker, subject agreement, predicate marker, or subject referencing pronoun; see Sankoff 1994, p. 295 and Crowley, this volume, for references) seems to have been greatly constrained by interlocutors’ needs to effectively track referents in discourse. In early stages of Tok Pisin, i often functions like a marker of switch reference, occurring with a full NP when the subject of the clause is different from the subject of the immediately preceding one (Sankoff 1994, p. 297). Although many Papuan and some Austronesian languages have fully grammaticalized morphosyntactic switch-reference systems, the switch-reference effects in Tok Pisin emerge only as a probabilistic tendency, not as a categorical rule, and only for a short period in the history of the language. Moreover, there is a similar tendency for NP + pronoun sequences to be used in this way in vernacular English, as first noted by Hall (1966, p. 83). So, as with ia, it seems that the effect of lexifier and substrate models as well as cognitive factors provides the best account (cf. Baker 1990, pp. 256–7). There are interesting things to be said, too, about the variation between subject pronouns and null subjects in some of the Pacific P/Cs, and there are some reasons to think that this variable may not be unrelated to the way subject agreement has developed. Hawai[i Pidgin, Bislama, and Tok Pisin all have null expletive subjects, and all three also allow null referential subjects to varying degrees (contra I. Roberts’s 1999 claim that because creoles lack verb movement they may have null expletive but not null referential subjects). The overall frequency of null referential subjects appears to be highest in Bislama, where 44 percent of all referential subjects in a corpus of mainly conversational speech were null (Meyerhoff 2000, p. 134). The frequency of null subjects is 39 percent for third person singular subjects in Tok Pisin but we do not know what percentage of this total was non-referential, i.e., expletive, subjects; Sankoff and Laberge’s data (1973) suggest that there are no null subjects in first and second person and in third person plural. The frequency is lowest, 8 percent, in Hawai[i Pidgin, where the null variants are always third person singular, usually existential or expletive subjects, but a few referential subjects as well. (These indicative figures for Pidgin were calculated from a small sample of 202 clauses of written and spoken Pidgin.) I am inclined to attribute the variation among the languages to discoursefunctional factors. Of the three varieties, Bislama has developed the richest paradigm of subject agreement. In Bislama, third peson singular and plural subjects, whether overtly realized or not, are marked with predicate proclitics i and oli respectively, while both first and second person subject agreement is null in the singular and i in the plural. Hence, the third person paradigm in Bislama can be considered the most informative one. It is, therefore, perhaps unsurprising to find that null subjects are more likely to occur in the third person, while pronouns are more likely to be used with first and second person, because the agreement marker does more “work” in third person than it does in first and second. The paradigm in Tok Pisin is less informative, as third person singular and plural both occur with i.

66 Miriam Meyerhoff But semantic transparency of the agreement is not the whole story; other factors interact as well. So the next most important factor for Bislama speakers (after the informativeness of agreement) is the information status of the referent itself. A highly topical subject, one for which the referent was also the subject of the preceding clause, has a greater chance of being null than of being pronominal or an NP. In addition, if the referent had already been realized as a null subject it is more likely to be null than any other subjectsubject continuity. This mirrors the kinds of constraints on discourse structure found in a diverse range of languages (see Gundel, Hedberg, & Zacharski 1993) and I would suggest that it is less a fact to do with the development of syntactic structure in Bislama than it is to do with cross-linguistic principles governing the realization of given vs. new information. In Hawai[i Pidgin, on the other hand, there is no overt subject agreement at all. The lack of anything in the verb phrase which can identify the subject referent may contribute to the very low levels of subject deletion in Pidgin (cf. subject deletion rates of 1–2 percent in spoken English; Meyerhoff 2000, p. 133). Presumably, only discourse salience will play a role in Pidgin, and we can expect the effects of this to be muted in the absence of semantically helpful morphosyntax. Interesting similarities with subjects show up in the realization of direct objects in Bislama. With objects, the question of informative verbal agreement is not relevant. The transitive suffix -im/-em/-um on verbs derives historically from the English third person singular and plural pronouns him, them, but not all speakers make use of this referential potential to identify a third person object (as, for instance, the i and oli subject agreement prefixes continue to identify arguments related to their source lexemes he and all he). The discourse status of the referent is the main constraint: if a referent has been a null subject or object in the last clause, it is more likely to be a null direct object.

6

Directions for the Future

The importance of taking discourse, as Sankoff does, as both the starting point and the primary medium for the development of syntax has important consequences. First, while substrate factors are seen as important, they are not necessarily taken as the starting point for all grammaticalized structures in Pacific P/Cs (contra, e.g., Keesing; Faraclas; Corne). Second, cognitive factors, such as the way information processing is contextually embedded, are given a lot of synchronic and historical weight (contra, e.g., Siegel’s emphasis on perceived similarity between lexifier and substrate syntax and semantics). Third, the variability inherent in a linguistic system is foregrounded. This approach is essentially sympathetic to the principles of speaker agency and creativity that are fundamental to accommodation theory. It means that the interplay between speakers and their ongoing negotiation of meanings and forms is presented as not just linguistically but also psychologically

Forging Pacific Pidgin and Creole Syntax 67 functional. Hence, there are obvious links with Keesing’s and Faraclas’s analyses that highlight speaker agentiveness. The need to conduct research that is sympathetic to speakers’ agency is not simply a matter of politics or taste. Faraclas’s and Sankoff’s work show that, by taking this perspective, it is possible to resolve outstanding puzzles about why some substrate features but not others are found in Pacific P/Cs. Other approaches may provide a reasonable account for why a feature transfers, but they won’t necessarily explain why others do not (see, too, Siegel 1998). A good example is the inclusive/exclusive pronominal distinction. This is marked in the Bislamic languages (Tok Pisin, Pijin, and Bislama) just like it is marked in their Melanesian substrate. But presence in the substrate alone does not account for its presence in these three Pacific P/Cs. Features don’t just transfer because they’re there. Research on Pacific P/Cs has tended to focus on structural factors, but this ignores the importance of socio-cultural factors. We should perhaps consider the possibility that transfer happens when transfer matters. That is, it is only effected when some significant subgroup of speakers assign social or cultural meaning to some variable that is present in the substrate. Along these lines, Faraclas (2002) argues that the inclusive/exclusive distinction transfers into Melanesian Englishes because – along with (in)alienable possession – this distinction is intimately bound up with key Melanesian notions of identity. Other linguistic features that lack this socio-cultural significance do not transfer even though the substrate models are just as readily available to the same speakers. For instance, there is an overwhelming tendency for languages in northern Vanuatu4 to mark the verb overtly in the realis mood whereas irrealis is the unmarked form (Jauncey 1997, pp. 483ff.). Nevertheless, realis does not seem to have transferred into Bislama as being in any way the marked verb form (perhaps on the contrary; cf. Meyerhoff 1996). It seems reasonable to suppose that the marking of realis on verbs has not transferred into Bislama because it does not express any significant social or cultural values, certainly nothing along the lines of the notions of identity that are so closely bound up with inclusivity and inalienability. I believe that the analysis of the syntax of Pacific P/Cs cannot be undertaken properly unless we forge an account out of the influences of the substrate, the lexifier, and the internal pressures exerted on a language in use. As I have tried to show, if we omit any one of these factors, we can provide only a partial view of the sense that lies beneath the syntactic structures of Pacific P/Cs. Sometimes debates over Pacific P/C syntax have been muddied by an understandable tendency to think in terms of the alternation and competition between different, homogeneous linguistic systems coming into contact, with a single grammar emerging (like Minerva) from the headache that ensues. I realize that it is possible that my own discussion of Pacific P/C syntax may seem to operate with such idealized objects, since that is often an unfortunate artefact of a survey article. However, I would certainly stress the reciprocal influence of apparently independent forces on the syntax of Pacific P/Cs, and

68 Miriam Meyerhoff the importance of interpersonal negotiation in the task of forging coherent norms from ambiguous or variable input. If we abstract away from the variability inherent in the languages in contact, it is easy to ignore the variability inherent in the outcome. This variability is constrained by linguistic principles, to be sure, and typological principles too, but we have to remember that ultimately those constraints are hammered out between speakers and in speech communities. If forty years of research in variationist sociolinguistics has told us anything, it must surely be this: inherent variability in a language is precisely that which provides the resources to be transformed and interpreted as the engine of further change. This holds true in P/Cs no less than in any other language. It seems to me that, although this is an important observation from the point of view of linguistic theory, it is of more than simply academic interest. We might consider the extent to which this notion of inherent variability in grammars could inform or interact with the application of our descriptive work on Pacific P/Cs. We might, for instance, come to rather different conclusions about what the most logical ends to policy and planning initiatives are, or the best means by which to achieve those ends, depending on whether we assume the primacy of substrate transfer, or primacy of the lexifier, or some innate structure. On the other hand, if we accept the idea of speakers’ dynamic use of inherent variation as being characteristic of any medium of communication, this avoids procrustean gestures which may intentionally or unintentionally convey messages about what is the most valued or prestigious language variety in a multilingual community. By admitting inherent variability into our analysis of the syntax of Pacific P/Cs, we open the door to incorporating social and linguistic variables into our theories of syntactic development and change. It is to be hoped that this will help forge a lasting research program that is as close to the hearts and concerns of the speakers of these languages as it is to the hearts of creolists.

NOTES I am extremely grateful to a number of people for discussions during the preparation of this chapter. First and foremost Dave Britain, who provided an extensive critique of an early draft, and the editors for comments on a later one. Special thanks, too, go to Alison Tonaki who gave me permission to use data based on field recordings she conducted in Pidgin in May 2000. Thanks also to Nik Gisborne, Greg Guy, Caroline Heycock, and Peter Patrick for general discussions; Laurie Reid and Bob Blust for explaining the subtleties of ma- in Philippine languages; Mike Forman for commentary on Zamboagueño; Corine Atesano, Pierrette Thibault, and Julie Auger for judgments on spoken, vernacular French and their informative commentaries thereon; and Mary Boyce for discussion of standard Maori. Despite their labors, there will be faults in this chapter; they are all mine. 1 Keesing transcribes one of his speakers with tes and another with des.

Forging Pacific Pidgin and Creole Syntax 69 2 In some of his work (Corne 1994, 2000), he adds Xârâcùù, but his chronology of settlement shows that Xârâcùù speakers arrived considerably later than speakers of Cèmuhî and Nráa Drùbea and in “small accretions” (1994, p. 284). In addition, he specifically says that whether migrants from the South region were speakers of Xârâcùù “is not known” (1990, p. 6). Since Cèmuhî and Nráa Drùbea are reliably attested as early input to Tayo through significant speaker numbers, I’ll only comment on data from them. 3 Since 2002, the debate over creole simplicity has expanded significantly. 4 Northern Vanuatu was the principal early source of Ni-Vanuatu labor to Queensland, where it is agreed that the language that developed into Bislama stabilized (Price & Baker 1976).

REFERENCES Archibald, John (ed.) (2000) Second Language Acquisition and Linguistic Theory. Oxford: Blackwell. Baker, Philip (1990) Pacific reorientations. Journal of Pidgin and Creole Languages 5, 253–69. Baker, Philip (1993) Australian influence on Melanesian Pidgin English. Te Reo 36, 3–67. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984) The language bioprogram hypothesis. Behavioral and Brain Sciences 7, 173–88. Bonnemaison, Joël (1986) L’Abre et le Pirogue (Les fondements d’une identité territoire, histoire et société dans l’archipel de Vanuatu (Mélanésie), livre 1. Paris: Éditions de l’Orstom. Camden, Bill (1996) The transitive verbs using long in Bislama. In: John Lynch and Fa’afo Pat (eds.) Oceanic Studies: Proceedings of the First International Conference on Oceanic Linguistics (Pacific Linguistics, Series C-133). Canberra: The Australian National University, pp. 319–51. Camden, William G. (1979) Parallels in the structure of lexicon and syntax between New Hebrides Bislama and the south Santo language as spoken at Tangoa. In: Papers in Pidgin and Creole Linguistics, no. 2 (Pacific Linguistics, Series A-57). Canberra: The Australian National University, pp. 51–117. Charpentier, Jean-Michel (1979) Le pidgin Bislam(n) et le multilinguisme aux NouvellesHébrides. Paris: Société d’Etudes Linguistiques et Anthropologiques de France. Clark, Ross (1979) In search of Beach-la-Mar: Towards a history of Pacific Pidgin English. Te Reo 22/23, 3–66. Clark, Ross (1990) Pidgin English and Pidgin Maori in New Zealand. In: Allan Bell and Janet Holmes (eds.) New Zealand Ways of Speaking English. Wellington: Victoria University Press, pp. 97–114. Corne, Chris (1990) Tayo pronouns: A sketch of the pronominal system of a Frenchlexicon creole language of the South Pacific. Te Reo 33, 3–24. Corne, Chris (1994) Relativization and thematization in Tayo and the implications for creole genesis. Journal of Pidgin and Creole Languages 9, 283–304. Corne, Chris (1995) Nana k nana, nana k napa: The paratactic and hypotactic relative clauses of Réunion Creole. Journal of Pidgin and Creole Languages 10, 57–76.

70 Miriam Meyerhoff Corne, Chris (2000) Na pa kekan, na person: The evolution of Tayo negatives. In: Jeff Siegel (ed.) Processes of Language Contact: Studies from Australia and the South Pacific. Saint-Laurent, Quebec: Champs Linguistiques, pp. 293–317. Crowley, Terry (1990) From Beach-la-Mar to Bislama: The Emergence of a National Language. Oxford: Oxford University Press. Day, Richard R. (1973) Patterns of variation in copula and tense in the Hawaiian post-creole continuum. University of Hawai[i Department of Linguistics Working Papers 5, 2. DeGraff, Michel (2001) On the origins of creoles: A Cartesian critique of NeoDarwinian linguistics. Linguistic Typology 5, 213–310. Drechsel, Emanuel J. (1999) Language contact in the early colonial Pacific: Evidence for a maritime Polynesian jargon or pidgin. In: John Rickford and Suzanne Romaine (eds.) Creole Genesis, Attitudes and Discourse. Amsterdam: John Benjamins, pp. 71–96. Durie, Mark (1988) Verb serialization and “verbal prepositions” in Oceanic languages. Oceanic Linguistics 27, 1–23. Dutton, Thomas E. and Herbert A. Brown (1977) Hiri Motu: The language itself. In: Stephen A. Wurm (ed.) New Guinea Area Languages and Language Study, vol. 3: Language, Culture, Society, and the Modern World. (Pacific Linguistics, Series C-40). Canberra: The Australian National University, pp. 759–93. Early, Robert (2000) Sit, stand, lie: Posture verbs and imperfectives. In: Bill Palmer and Paul Geraghty (eds.) SICOL: Proceedings of the Second International Conference on Oceanic Linguistics, vol. 2: Historical and Descriptive Studies. Canberra: Pacific Linguistics, pp. 81–103. Ehrhart, Sabine (1993) Le créole français de St-Louis (le tayo) en Nouvelle-Calédonie. Hamburg: Buske. Faraclas, Nicholas G. (2002) From Old Guinea to Papua New Guinea: A comparative study of Nigerian Pidgin and Tok Pisin. Paper presented at the Society for Pidgin and Creole Linguistics, San Francisco, CA. Forman, Michael L. (1972) Zamboagueño Texts with Grammatical Analysis: A Study of Philippine Creole Spanish. Unpublished PhD dissertation, Cornell University. Forman, Michael L. (1993) Verb serialization, word order typology, and Zamboagueño: A comparative approach. Oceanic Linguistics 32, 163–82. Giles, Howard and Nikolas Coupland (1991) Language: Contexts and Consequences. Pacific Grove, CA: Brooks/Cole. Goodman, Morris (1985) Review of Roots of Language, by Derek Bickerton. International Journal of American Linguistics 51, 109–37. Goodman, Morris (1999) On the origins of Hawaiian Creole English: A rejoinder to Roberts. Language 75, 347–9. Grant, Anthony P. (2001) Language intertwining: Its depiction in recent literature and its implications for theories of creolization. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 81–111. Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski (1993) Cognitive status and the form of referring expressions in discourse. Language 69, 274–307. Hall, Robert A., Jr. (1966) Pidgin and Creole Languages. Ithaca, NY: Cornell University Press. Harding, Thomas G. (1967) Voyagers of the Vitiaz Strait: A Study of a New Guinea Trade System. Seattle: University of Washington Press. Harris, John W. (1991) Kriol: The creation of a new language. In: Suzanne Romaine (ed.) Language in Australia. Cambridge: Cambridge University Press, pp. 195–203. Holm, John (1988) Pidgins and Creoles. Cambridge: Cambridge University Press.

Forging Pacific Pidgin and Creole Syntax 71 Jauncey, Dorothy (1997) A Grammar of Tamambo, the Language of Western Malo. Unpublished PhD dissertation, Australian National University, Canberra. Källgård, Anders (1993) Present-day Pitcairnese. English World-Wide 14, 71–114. Keesing, Roger M. (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford: Stanford University Press. Kihm, Alain (1995) Tayo, the strange bird from New Caledonia: Determiners and tenseaspect in Tayo and their implications for creolization theories. Journal of Pidgin and Creole Languages 10, 225–52. Kouwenberg, Silvia (1992) From OV to VO: Linguistic negotiation in the development of Berbice Dutch Creole. Lingua 88, 263–99. Kouwenberg, Silvia (2001) Convergence and explanations in creole genesis. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 219–47. Le Page, Robert B. and Andrée Tabouret-Keller (1985) Acts of Identity: Creole-Based Approaches to Language and Ethnicity. Cambridge: Cambridge University Press. Lipski, John M. (1987) Modern Spanish once-removed in Philippine Creole Spanish: The case of Zamboagueno. Language in Society 16, 91–108. Long, Daniel (2002) Contact varieties of Japanese. Paper presented at Methods XI, University of Joensuu, Finland, 5–9 August. Lorenzino, Gerardo A. (1993) African vs. Austronesian substrate influence on the Spanish-based creoles. In: Francis Byrne and John Holm (eds.) Atlantic Meets Pacific: A Global View of Pidginization and Creolization (Selected Papers from the Society of Pidgin and Creole Linguistics). Amsterdam: John Benjamins, pp. 399–408. Masuda, Hirokuni (1999) Exploration of the trinary components in creole discourse: Universals, substrata, and superstrata. In: John Rickford and Suzanne Romaine (eds.) Creole Genesis, Attitudes and Discourse. Amsterdam: John Benjamins, pp. 373–89. Meyerhoff, Miriam (1996) Transitive marking in contact Englishes. Australian Journal of Lingustics 16, 57–80. Meyerhoff, Miriam (2000) Constraints on Null Subjects in Bislama (Vanuatu). Canberra: Pacific Linguistics. Meyerhoff, Miriam (2001) Another look at the typology of serial verb constructions: The grammaticalization of temporal relations in Bislama (Vanuatu). Oceanic Linguistics 40, 247–68. Meyerhoff, Miriam (2002) Formal and cultural constraints on optional objects in Bislama. Language Variation and Change 14, 323–46. Meyerhoff, Miriam and Nancy Niedzielski (1994) Resistance to creolization: An interpersonal and intergroup account. Language and Communication 14, 313–30. Molony, Carol H. (1977) Recent relexification processes in Philippine Creole Spanish. In: Ben G. Blount and Mary Sanches (eds.) Sociocultural Dimensions of Language Change. New York: Academic Press, pp. 131–59. Mufwene, Salikoko S. (1996) The founder principle in creole genesis. Diachronica 13, 83–134. Mühlhäusler, Peter (1984) Tracing the roots of pidgin German. Language and Communication 4, 27–57. Mühlhäusler, Peter (1998) How creoloid can you get? Journal of Pidgin and Creole Languages 13, 355–71. Muysken, Pieter (1988) Are creoles a special type of language? In: Frederick G. Newmeyer (ed.) Linguistics: The Cambridge Survey, vol 2. Cambridge: Cambridge University Press, pp. 285–301.

72 Miriam Meyerhoff Muysken, Pieter and Norval Smith (eds.) (1986) Substrata versus Universals in Creole Genesis: Papers from the Amsterdam Creole Workshop, April 1985. Amsterdam: John Benjamins. Price, Charles and Elizabeth Baker (1976) Origins of Pacific Island Labourers in Queensland, 1863–1904: A research note. Journal of Pacific History 11, 106–21. Reinecke, John E. (1971) Tây Bôi: Notes on the pidgin French spoken in Vietnam. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 47–56. Reinecke, John E. (ed.) (1975) A Bibliography of Pidgin and Creole Languages (Oceanic Linguistics Special Publication no. 14). Honolulu: The University Press of Hawai[i. Roberts, Ian (1999) Verb movement and markedness. In: Michel DeGraff (ed.) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press, pp. 287–327. Roberts, Julian M. (1995). Pidgin Hawaiian: A sociohistorical study. Journal of Pidgin and Creole Languages 10, 1–55. Roberts, Sarah J. (1998) The role of diffusion in the genesis of Hawaiian Creole. Language 74, 1–39. Roberts, Sarah J. (2004) The role of identity and style in the development of Hawaiian Creole. In: Genevieve Escure and Armin Schwegler (eds.) Creoles, Contact, and Language Change: Linguistics and Social Implications. John Benjamins: Amsterdam, pp. 331–50. Romaine, Suzanne (1992) The evolution of complexity in a creole language: Acquisition of relative clauses in Tok Pisin. Studies in Language 16, 139–82. Romaine, Suzanne (1999) The grammaticalization of the proximative in Tok Pisin. Language 75, 322–46. Sankoff, Gillian (1977a) Multilingualism in Papua New Guinea. In: Stephen A. Wurm (ed.) New Guinea Area Languages and Language Study, vol. 3: Language, Culture, Society, and the Modern World. (Pacific Linguistics, Series C-40). Canberra: The Australian National University, pp. 265–307. Sankoff, Gillian (1977b) Variability and explanation in language and culture: Cliticization in New Guinea Tok Pisin. In: Muriel Saville-Troike (ed.) Georgetown University Round Table on Languages and Linguistics. Washington, DC: Georgetown University Press, pp. 59–73. (Reprinted in Sankoff (1980), pp. 257–70.) Sankoff, Gillian (1980) The Social Life of Language. Philadelphia: University of Pennsylvania Press. Sankoff, Gillian (1984) Substrate and universals in the Tok Pisin verb phrase. In: Deborah Schiffrin (ed.) Meaning, Form, and Use in Context: Linguistic Applications. Washington DC: Georgetown University Press, pp. 104–19. Sankoff, Gillian (1993) Focus in Tok Pisin. In: Francis Byrne and Donald Winford (eds.) Focus and Grammatical Relations in Creole Languages. Amsterdam: John Benjamins, pp. 117–40. Sankoff, Gillian (1994) An historical and evolutionary approach to variation in the Tok Pisin verb phrase. Parasession on Variation and Linguistic Theory. Chicago: Chicago Linguistic Society, pp. 293–320. Sankoff, Gillian (1996) The Oceanic substrate in Melanesian pidgin/creole revisited: A tribute to Roger Keesing. In: John Lynch and Fa’afo Pat (eds.) Oceanic Studies: Proceedings of the First International Conference on Oceanic Linguistics (Pacific Linguistics, Series C-133). Canberra: The Australian National University, pp. 421–50.

Forging Pacific Pidgin and Creole Syntax 73 Sankoff, Gillian (2002) Divergence, drift, and substrate: The evolution of focus in three Pacific creoles. Paper presented at the Annual Meeting of the Society for Pidgin and Creole Linguistics, San Francisco, January. Sankoff, Gillian and Penelope Brown (1976) The origins of syntax in discourse: A case study of Tok Pisin relatives. Language 52, 631–66. (Reprinted in Sankoff 1980, pp. 211–55.) Sankoff, Gillian and Suzanne Laberge (1973) On the acquisition of native speakers by a language. Kivung 6, 32–47. (Reprinted in Sankoff 1980, pp. 195–209.) Sankoff, Gillian and Claudia Mazzie (1991) Determining noun phrases in Tok Pisin. Journal of Pidgin and Creole Languages 6, 1–24. Sato, Charlene Junko (1978) Variation in Hawaiian Pidgin and creole English: go + verb constructions. Unpublished MA dissertation, University of Hawai[i at Mänoa. Shnukal, Anna (1988) Broken: An Introduction to the Creole Language of Torres Strait. Canberra: Pacific Linguistics. Siegel, Jeff (1987) Language Contact in a Plantation Environment. Cambridge: Cambridge University Press. Siegel, Jeff (1990) Pidgin English in Nauru. Journal of Pidgin and Creole Languages 5, 157–86. Siegel, Jeff (1998) Substrate reinforcement and dialectal differences in Melanesian Pidgin. Journal of Sociolinguistics 2/3, 347–73. Siegel, Jeff (2000) Substrate influence in Hawai[i Creole English. Language in Society 29, 197–236. SIL International (2002) Ethnologue: Languages of the World, 14th edition. www.ethnologue.com. Silverstein, Michael (1972a, b) Chinook Jargon: Language contact and the problem of multilevel generative systems. Language 48, 378–406 (Part I), 596–625 (Part II). Smith, Geoff P. (2002) Growing Up with Tok Pisin: Contact, Creolization and Change in Papua New Guinea’s National Language. London: Battlebridge. Thakerar, Jitendra N., Howard Giles and Jenny Cheshire (1982) Psychological and linguistic parameters of accommodation theory. In: Colin Fraser and Klaus R. Scherer (eds.) Advances in the Social Psychology of Language. Cambridge: Cambridge University Press, pp. 205–55. Thomason, Sarah G. (1983) Chinook Jargon in areal and historical context. Language 59, 820–70. Thomason, Sarah G. (1997) Mednyj Aleut. In: Sarah G. Thomason (ed.) Contact Languages: A Wider Perspective. Amsterdam: John Benjamins, pp. 449–68. Trudgill, Peter (2002) Sociolinguistic Variation and Change. Edinburgh/Washington, DC: University of Edinburgh Press/Georgetown University Press. Velupillai, Viveka (2003) Hawai[i Creole English: The Tense-Mood-Aspect System. London: Palgrave. Williams, Jeffrey P. (1993) Documenting the Papuan-based pidgins of Insular New Guinea. In: Francis Byrne and John Holm (eds.) Atlantic Meets Pacific: A Global View of Pidginization and Creolization (Selected Papers from the Society for Pidgin and Creole Linguistics). Amsterdam: John Benjamins, pp. 355–67. Williams, Jeffrey P. (2000) Yimas-Alamblak Tainim Tok: An indigenous trade pidgin of Papua New Guinea. Journal of Pidgin and Creole Linguistics 15, 37–62.

74 Terry Crowley

4

Pidgin and Creole Morphology TERRY CROWLEY

1

Introduction

Pidgins and creoles are typically contrasted with their superstrate or lexifier languages in terms of their hallmark morphological simplicity. As a result, many surveys of pidgins and creoles provide little discussion of morphological topics, preferring to comment on phonology, lexicon, and syntax, as these areas are taken to be where the real “action” is. Holm (2000), for example, includes chapters on lexicosemantics, phonology, and syntax, with only a short subsection on morphology, while Alleyne (1980) devotes only slightly more than one page to morphology. Bakker’s overview of pidgin languages comments that “pidgins have reduced inflectional and derivational morphology as compared to the source languages” (Bakker 1995, p. 31), adding that this appears to be the only valid structural generalization that we can make about all such languages. Nevertheless, this chapter aims to describe the morphological features that typically survive in the development of pidgin and creole languages, and also to describe the kinds of morphological features that subsequently develop as such languages evolve independent of the original language contact. In seeking to generalize about pidgins and creoles, of course, we immediately come up against the question whether we can reasonably expect to comment on all pidgins and all creoles with a single set of statements. With regard to just creoles, there are quite divergent views as to how such languages should be characterized. The traditional view that a creole is a pidgin that has become the first language of a speech community (Hall 1966, p. 1) tells us very little, particularly given that there are cases where there is little evident difference between the way that native (creole-)speakers and non-native (pidgin-)speakers speak (see Bakker, this volume). Bickerton (1981) redefines creole such that only a small subset of “traditional” creoles fall into this category. Mufwene (1986) and DeGraff (2001a) go in the opposite direction, effectively arguing that creoles as a definable linguistic type cannot be distinguished from other

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Pidgin and Creole Morphology 75 languages with any definition referring only to sociohistorical factors surrounding their origins. The term “pidgin” is similarly fraught. Languages designated as pidgins range from extremely rudimentary short-term contact languages used in only a very narrow range of contexts to structurally and lexically far more expanded varieties which have been in use over an extended period and for a much broader range of functions, even though in each case there may be a shared lack of native speakers. It is probably the case that far more is known about expanded pidgins than more rudimentary ones, given that they have been around long enough for linguists to be able to devote their attentions to them. The expanded Tok Pisin of today, for example, is far better known than its more rudimentary precursor of 125 years ago simply because professional linguists are in a position to describe how people speak today. Our sources for Tok Pisin in its earliest stages are far more limited both in their scope and their reliability. In this chapter, I use the term “pidgin” to refer both to rudimentary varieties as well as structurally expanded varieties such as Tok Pisin, which are referred to by Bakker (this volume) as pidgincreoles. I will begin by considering the question of morphological simplicity and how it may be manifested in pidgins and creoles (section 2). In section 3 on inflectional morphology, I will look at both loss and retention of inflection from the lexifier. In section 4, on derivational morphology, retention from the lexifier as well as the independent development of derivational affixes will be considered. Other types of morphology (compounding, reduplication, clitics) as well as the reanalysis of polymorphemic forms from the lexifier and the existence of allomorphy – the latter unexpected from the point of view of the short history of these languages – are the subject of section 5. This is followed by a discussion of the significance of the morphological life cycle (section 6) and a final outlook on the work that awaits us as pidgin and creole specialists (section 7).

2 Morphological Simplicity There is often little explicit discussion of what constitutes morphological “simplicity,” perhaps because many of us operate with a fairly clear set of intuitions as to what it entails. Whaley (1997, pp. 128–36) points out that languages can be ranged along two axes in terms of their morphological typology. The axis of synthesis involves the amount of affixation, while the axis of fusion involves the extent to which morphemes are individually segmentable. There is a continuum of synthesis between isolating languages with just a single morpheme per word and polysynthetic languages with large numbers of morphemes packed into a single word, as well as a continuum of fusion between agglutinating languages with clear boundaries between monofunctional morphemes and fusional languages in which morpheme boundaries are typically obscured. These four types of languages can be ranged along a single cline in

76 Terry Crowley terms of level of morphological complexity, with isolating languages being morphologically the least complex and polysynthetic fusional languages being the most complex. Thurston (1987) makes a useful distinction between esoteric and exoteric languages, with the latter typically being used for communication between speakers of different languages and the former being used exclusively for intra-group communcation. Exoteric languages, he argues, more closely approximate a one-to-one correspondence between form and meaning, with relatively few bound morphemes, and the few bound morphemes exhibiting little allomorphic variation or suppletion. Esoteric languages, by way of contrast, are characterized by more bound morphology, a greater number of portmanteau morphemes and suppletion, along with a greater amount of morphophonemic derivation. Esoteric languages are often regarded by members of multilingual linguistic ecologies as being “difficult” to learn, while exoteric languages are often regarded as “easy.” While there are problems in assuming a direct correspondence between ease of learning and exoteric language use (Crowley 2000a), these folk judgments seem to correspond fairly neatly to real differences in morphological typology. While pidgins and creoles are by no means exclusively isolating in their morphologies, they nevertheless constitute a sizable proportion of the total number of isolating languages in the world, and are generally morphologically simpler than any of the languages which constituted part of the linguistic mix from which they developed in the first place. This suggests that a tendency toward isolating morphology represents a major outcome of multilingual contact in an imperfect learning situation. However, we must be careful that our expectations of morphological simplicity not blind us to complexity when we are faced with it. It is only recently, for example, that we have become aware that the agentive nominalizing suffix -ma (< English man) in Saramaccan is substantially more complex in its behavior than is the agentive suffix ‘-er’ in English (Bakker, Smith, & Veenstra 1995, p. 173). In addition to being added to a simple verb, e.g., sabi-ma ‘expert’ (< sabi ‘know’ + ma), this suffix can also be added to a sequence of verb plus object, e.g., bebe-daan-ma ‘drunkard’ (< bebe ‘drink’ + daan ‘rum’ + ma), and even to verbs with objects and accompanying prepositional phrases, e.g., tja-boto-go-a-wosu-ma ‘pilot’ (< tja ‘carry’ + boto ‘boat’ + go ‘go’ + a ‘to’ + wosu ‘house’ + ma). In this case, the suffixal nature of -ma was presumably not apparent to earlier observers who perhaps assumed that since -ma derived from English man, this constituted a phrase-level rather than a word-level construction. Pidgins and creoles have largely (or possibly exclusively) arisen in situations of language contact in which a number of non-isolating languages have been predominantly involved, i.e., pidgins and creoles do not seem to have been attested as evolving when only purely isolating languages come into contact. Of course, English is very nearly an isolating language – at least as far as inflectional morphology is concerned – yet it is still possible to find evidence of morphological reduction in the case of English-lexifier pidgins and creoles.

Pidgin and Creole Morphology 77 Morphological simplicity in pidgins and creoles can generally be taken to imply the following: • • • •

a near-total absence of inflectional morphology the near-total absence of redundantly encoded morphological information little or no derivational morphology close approximation toward one form, one meaning in the sense that there is a complete absence of portmanteau morphemes (while recognizing that multifunctional forms may still be frequent) • the near-total absence of morphological irregularity and suppletion • a minimal amount of allomorphic variation Below, we will consider the extent to which these features are manifested in pidgins and creoles. Other less commonly cited features, which are widely – if not universally – distributed among pidgins and creoles, include the following: • a reliance on contiguous morphology over discontiguous morphology (hence the lack of infixation, simulfixation, or root modification) • an avoidance of sequences of inflectional morphemes or derivational morphemes, i.e., a preference for a maximum of only one inflectional category and/or derivational category within a single word A preference for contiguous and for suffixing morphology as set out above represents, in fact, a strong statistical tendency across all categories of languages, though it may be that there is a stronger preference in these directions among pidgins and creoles than in other languages.

3 Inflectional Morphology Inflectional morphemes tend to be phonologically less salient than derivational morphemes. The elimination of inflectionally expressed information is, in fact, what we typically find in pidgins and creoles, and Sebba (1997, p. 44) comments: “It is inflectional morphology which typically fares worst in a pidgin. Many pidgins have none at all . . .” Holm (2000, p. 127) similarly notes that “[p]idgins and creoles are sometimes claimed to be languages without any inflectional morphology whatsoever,” but points to a small number of cases where some inflectional morphology has possibly survived the restructuring process. However, there is room for some doubt about the genuinely inflectional status of the examples that he cites. McWhorter (1998, p. 792) notes that inflectional categories in creoles and expanded pidgins are likely to be only “weakly obligatory,” though optionality is not a feature that we would normally expect with inflectional categories, which are typically added to any root of a given lexical class with the same meaning. On the other hand, Adamson

78 Terry Crowley and Smith (1995, p. 225) refer to debate about the status of obligatory preverbal markers in Sranan, suggesting that they might well be best treated as inflectional morphemes. Moreover, inflectional marking in Berbice Dutch for aspect on verbs and number on nouns is clearly not optional (see Kouwenberg 1994). I now turn to a consideration of specific cases of loss (section 3.1) and retention (section 3.2) of inflectional categories, followed by more general discussion (section 3.3).

3.1

Loss

In pidgins and creoles, an inflection in the lexifier language distinguishing between singular and plural on nouns is typically eliminated completely from the morphology. In most cases, the resultant word is derived from the morphologically unmarked singular form. Thus, English ‘dog’ and ‘dogs’ are both expressed simply as dog in Bislama, with the meaning of plural being expressed by preposed ol (< English all) (all Bislama examples from my own field experience). In some cases, however, the pragmatically unmarked form is that which is morphologically marked, which may then be taken as the basis for a reanalyzed root. “Matches,” for example, are more likely to be encountered in multiples, so the Bislama word is based on the English plural as masis rather than being derived from singular match. Anis ‘ant’ is similarly derived from the plural ants rather than from singular ant, while sus ‘shoe’ derives from plural shoes rather than singular shoe. In addition, traoses ‘trousers’ derives from a grammatically plural form in English, yet there is nothing grammatically plural about the resultant noun in Bislama. In some cases, however, it is difficult to explain why a morphologically marked form becomes the basis for morphological reanalysis in terms of pragmatic salience. In contrast to the form sus in Bislama, the form but ‘boot(s)’ is based on the singular form in English. This inconsistency may be explained by the consonant-final form but which, if it were to end in -s, would result in an illicit word-final cluster in Bislama. However, other forms appear to be fairly arbitrarily based on morphologically marked forms. Most English-lexifier verbs in Bislama are reflected by their English unmarked present tense forms, e.g., fesem ‘play against (in sport)’ (< face + the Bislama transitive suffix -em). However, a handful of verbs are based unpredictably on regular past tense roots, e.g., prestem ‘press’ (< pressed + -em), and occasionally irregular past tense forms, e.g., brokem ‘break’ (< broke + -em). The elimination or reduction of inflectional morphology in a pidgin or creole normally results in the loss of redundantly encoded information. The category of plural, for example, is multiply encoded in an English sentence such as “These two girls are singing,” given the contrast with singular “This girl is singing.” In Bislama, the loss of plural marking on nouns, as well as on demonstratives and verbs, does not result in any loss of information in sentences such as the following:

Pidgin and Creole Morphology 79 (1) a.

Tufala gel ia i stap singsing. two girl this pm cont sing ‘These two girls are singing.’ b. Gel ia i stap singsing. girl this pm cont sing ‘This girl is singing.’

The meaning of plurality is encoded only in the lexical item tufala ‘two’, while gel ‘girl(s)’, ia ‘this/these’, etc., do not call for separate forms simply because the plural meaning is lexically encoded. The loss of subject prefixation on verbs in pidginized Motu, usually referred to as Hiri Motu, can be seen as an instance of the same tendency. Both the subject prefixes on Motu verbs (na- 1s, o- 2s/p, e- 3s/p, ta- 1p.incl, a- 1p.excl) and the object suffixes (-gu 1s, -mu 2s, -(i)a 3s, -da 1p.incl, -mai 1p.excl, -mui 2p, -dia 3p) are lost in pidginized Motu (Dutton 1985, p. 10). Information expressed morphologically in Motu is expressed by free pronouns in pidginized Motu, as in (2). Given that Motu na-ita-mu ‘I see you’ can alternatively appear with a redundant free subject pronoun, i.e., (lau) na-ita-mu ‘I see you’, where the subject prefix encodes redundant information, the isolating pidgin strategy can be seen as reducing redundancy. (2) a.

Motu:

(lau) na-ita-mu (1s) 1s-see-2s b. Hiri Motu: oi lau itaia 2s 1s see both: ‘I see you.’

Morphological irregularity in inflectional morphology also seldom survives in pidgins and creoles. Thus, English verbs with tense-related irregularities such as catch/caught become regular when tense is expressed syntactically rather than morphologically in Bislama, e.g., kasem ‘catch’ and bin kasem ‘caught’. Dutton (1985, pp. 22–35) shows how extensive patterns of suppletion in the verbal morphology of the languages of the Papuan Gulf were regularized in the Koriki Pidgin and Elema Pidgin used in pre-colonial trading expeditions, and Foley (1988, p. 169) indicates that extensive patterns of suppletion in vernacular Yimas were also completely regularized in Pidgin Yimas.

3.2

Retention

Bakker (1995, pp. 31–3) notes that pidgins derived from highly inflected sources may preserve systems of inflection that are substantially more complex than we find in most other pidgins and creoles lexified from less highly inflected languages, albeit considerably reduced vis-à-vis their sources. He cites Pidgin Ojibwe o-kot-aan niin ‘he fears me’, where o- marks the third person singular

80 Terry Crowley animate subject of kot ‘fear’. However, while -aan is glossed as third person singular inanimate, Bakker (1995, p. 32) notes that in Pidgin Ojibwe inanimate verbs are always used, even for animate actors. This raises the suspicion that the verb root here might in fact be kotaan with the original inanimate suffix -aan having been reanalyzed as part of the root, particularly as there is no participant with third person singular inanimate reference involved in this event and the free pronoun niin expresses the first person singular object (see section 5.4 on morphological reanalysis). However, even if -aan may no longer be a productive suffix, o- does appear to have been retained as an inflectional prefix in Pidgin Ojibwe. Other pidgins and creoles show clear evidence of having retained – or subsequently developed – greater amounts of inflectional morphology, particularly where the languages in contact are typologically relatively uniform. Dutton (1985, p. 9) reports that in the variety of Hiri Motu used in the Central Province of Papua New Guinea – where there are more speakers of languages closely related to Motu than in other areas where Hiri Motu is used – the original pronominal object suffixes are retained rather than being expressed as free forms. Compare the non-Central Province and the Central Province varieties in (3): (3) a.

Motu: e-ita-gu / ia e-ita-gu 3s-see-1s 3s 3s-see-1s b. Hiri Motu (non-Central Province): lau ia itaia 1s 3s see c. Hiri Motu (Central Province): ia ita-gu 3s see-1s all: ‘(S)he saw me.’

Perhaps also favoring the retention of morphologically marked objects in this pidgin is the fact that, in contrast to the subject prefixes, pronominal objects are not optionally redundantly marked by means of free form pronouns. Thus, Motu e-ita-gu can be expressed alternately as ia e-ita-gu, but the pronominal object cannot be expressed as a free form. Hinnebusch (1979, p. 210) reports that while the complex noun classes and subject and object cross-reference system on verbs of standard Swahili are lost in Pidgin Swahili, in some varieties of the pidgin the inflectional distinction between singular and plural on nouns is retained with forms having human reference, e.g., m-levi ‘drunkard’, wa-levi ‘drunkards’. This retention can possibly also be attributed to typological similarity between the substrate Bantu languages, though the pragmatic salience of nouns with human reference is possibly a major part of the reason why the number distinction was retained with this particular subset. This may also have contributed to the (limited) retention of pronominal affixes in some varieties of Hiri Motu, as pronominal categories are likely to involve pragmatically highly salient referents.

Pidgin and Creole Morphology 81 It is tempting to generalize that if pidgins and creoles do retain inflectional morphology, the system will be morphotactically simplified in that only a single category will be marked prefixally and a single category marked suffixally on any word. This statement holds for all cases of which I am aware other than Sudan Pidgin Arabic, where a verb can attract both the future prefix bi- and the progressive prefix gi- in that order (Owens 1997, p. 149). If these forms are in fact both prefixes (rather than, for the sake of argument, preverbal particles) this may well be the only pidgin or creole with any kind of agglutinating inflectional morphology.

3.3

Discussion

Given that at least some inflectional morphology clearly can survive – or evolve – in pidgins and creoles, it would be interesting to see if any sorts of statements can be made about the nature of the categories that these inflectional markers may express. A survey of pidgin and creole languages of a wide variety of superstrate and substrate sources reveals the following reliably attested inflectional survivals and developments: • Bold (1974, p. 10) notes that Fanakalo has morphologically marked number on nouns referring to animals, though the pattern of number marking is substantially reduced vis-à-vis the complex system of morphological markers found in Zulu languages. Given that number inflection on nouns in Pidgin Swahili is most likely to survive with human nouns, it seems that high animacy status may favor the retention of this inflectional feature. • Pronominal marking on verbs for subjects or objects is sometimes retained, as illustrated by Ojibwe Pidgin subject prefixation or Central Province Hiri Motu object suffixation. Again, the high animacy of pronouns may favor this retention. • In Sudan Pidgin Arabic, plural imperatives are marked by means of a verbal suffix (Owens 1997, p. 149). Again, high animacy value is a possible factor in this retention. • Foley (1988, p. 169) indicates that a two-way tense distinction is retained with Pidgin Yimas verbs in contrast to the near universal tendency not to mark TMA categories inflectionally in pidgins and creoles. However, this two-way contrast represents a drastic reduction of the seven-way inflectional distinction in vernacular Yimas. Perhaps just as interesting as what kinds of inflectional morphology survive in pidgins and creoles is the question of what appears not to survive. The following general observations seem to hold: • Noun class or gender systems on nouns appear to be systematically lost. We see this most dramatically with Pidgin Swahili with the loss of the rich system of noun classes in standard Swahili (Hinnebusch, 1979). Note,

82 Terry Crowley

•

•

•

•

•

•

however, that De Rooij (1995, p. 185) indicates that Shaba Swahili retains rather more of the original noun class system than was reported for the variety of Pidgin Swahili documented by Hinnebusch (1979), though there is still substantial reduction vis-à-vis standard Swahili. We also encounter complete loss of noun classes in Pidgin Yimas (Foley 1988, p. 170), and the loss of the masculine/feminine gender distinction in Romance-lexifier pidgins and creoles is also consistent with this generalization. Affixed case morphology on nouns appears to be lost. Relatively few languages with elaborate systems of case inflections have been taken as the lexical sources of pidgins and creoles, though Russenorsk points to this kind of outcome. Sebba (1997, p. 44), however, cites Nagamese Pidgin from India as having an inflectional distinction on nouns between locative and dative case (and presumably also a zero-marked nominative). Pronouns, however – especially third person pronouns – are more likely to survive in pidgins and creoles with different forms for subject, object, and possessor. Constructions expressing different kinds of possessive relationships such as alienable vs. inalienable possession, or different kinds of alienable possession such as the distinction between owning something that is for eating as against something that is for drinking (commonly distinguished morphologically in Oceanic languages), are not retained, in for instance Hiri Motu and Pidgin Fijian. Pronominal cross-reference for both subject and object categories on verbs seems not to survive intact (cf. the discussion of retained object suffixes in some pidgin Motu varieties in section 3.2). Negation, if it survives as an inflectional category at all, is likely to be redundantly expressed in a pidgin or creole also by means of a free negative marker, as suggested by the redundant negative marking in Shaba Swahili (De Rooij 1995, p. 189) and Fa d’Ambu (Post, 1995, p. 197). Interestingly, negation shows up cross-linguistically as often being discontiguously marked (see contributions to Kahrel & van den Bergh 1993). Polysynthetic structures do not survive, with incorporated arguments expressed as free forms, as evidenced by Eskimo Pidgin (Van der Voort, 1995, p. 145) and Pidgin Yimas (Foley 1988). Passive morphology typically fares badly in pidgins and creoles. Papiamentu, however, constitutes a partial exception in that there are passive participle formations (Kouwenberg & Muysken 1995, p. 211).

It is possible that examples of languages with substantially richer inflectional systems do not represent legitimate pidgins or creoles. Bhattacharjya’s (1994) description of Nagamese, for example, is subtitled “Pidgin, creole or creoloid?”, which is suggestive of the possibility that this does not represent a pidgin or creole at all. De Rooij (1995, p. 180) also indicates that there is some dispute as to whether or not Shaba Swahili represents a genuine creole.

Pidgin and Creole Morphology 83 Of course, this leads to a problem of circularity of definition which is likely to bedevil attempts to generalize about any structural feature of pidgins and creoles. When we encounter a variety that in some respect does not meet our expectations, it is very tempting then to discount it by arguing that it is not a genuine pidgin or creole after all. The terms “creoloid” and “semi-creole” (Holm 2000, p. 10) seem to be particularly frequently invoked as labels for any variety that we think ought to be called a creole on the basis of its social history and where there is some simplification in its structural features – though not as much as usual in pidgins and creoles.

4 Derivational Affixing Pidgins and creoles also tend to be less complex with respect to their derivational affixing morphologies than many other languages. However, there is much more evidence for derivational affixes in a wide range of pidgins and creoles than is the case for inflectional morphology. This is consistent with the morphological universal that a language may have either derivational morphology without inflections, or both derivational and inflectional morphology, but not inflectional morphology alone. In the following, I distinguish between derivational affixing retained from the lexifier (section 4.1) and developed independently (section 4.2), followed by a brief note on zero derivation (section 4.3).

4.1

Retention

Interestingly, while inflectional prefixing morphology often does not survive into pidgins and creoles, derivational prefixes seem to fare rather better. In Bislama, for example, there is just a single prefix and that is derivational eks(< English ex-), which is productively used with proper nouns to indicate what something used to be called formerly (often prior to political independence, especially where a more up-to-date post-independence name has not been devised, or where such a name is not yet widely known), e.g., eks-Franis jel ‘the former French jail’. Given that a substantial proportion of the lexicon of any pidgin or creole is directly incorporated from a superordinate lexifier language, there are likely to be partial similarities in form and function between a substantial number of words in the pidgin or creole which correspond to productive or semiproductive patterns of derivational morphology in the lexical source. The question is whether or not these affixes represent productive patterns of derivation in the resultant pidgin or creole. Dijkhoff (1993), for example, discusses this question in detail with regard to Papiamentu, which has inherited many morphologically complex forms from its principal lexifier Spanish. Lefebvre (1998, pp. 303–17) discusses this question in relation to the derivational morphology of Haitian Creole. DeGraff (2001b) adds to her discussion, arguing

84 Terry Crowley that there is much more morphological complexity and productivity in Haitian Creole derivational morphology than has previously been acknowledged. The pairing of Haitian forms such as vivasyon ‘conviviality’ and viv ‘live’, where vivation does not appear at all in French, suggests that -syon has acquired productive status in Haitian as a nominalizer, especially given that the vast majority of Haitian Creole speakers do not speak standard French. Similar observations can be made for rural Bislama (speakers of which seldom have a high level of competence in English), with the pairing of forms such as flot ‘to float’ and flota ‘flotation device on fishing net’. I am not aware of any speakers of English who use the word ‘floater’ to express this meaning, and it appears that speakers of Bislama have adopted the productive agentive/ instrumental suffix -er in English as -a. In Bislama, however, this affix is still only marginal, as any number of other logically possible nouns derived on the same pattern are simply not attested. Speakers who have been highly educated in English may use jaj as a verb meaning ‘to judge’ and the corresponding noun jajmen ‘judgment’. The form -men should not be treated as a genuine nominalizing suffix given that it again does not have productive distribution in Bislama. Nor is the fact that the same educated speaker may use the verb riplae ‘reply’ – rather than the better established form ansa (< answer) – along with the corresponding nominalization riplaemen an indication that -men has acquired productive derivational status in Bislama. Rather, such a speaker would simply be engaging in ad hoc borrowing of the nonstandard form “replyment” in local English, as this is where the suffix appears to be acquiring greater productivity. The question of what constitutes a productive pattern and what does not is an issue that plagues not just the study of pidgins and creoles. How productive, after all, is -ment as a nominalizing suffix in standard English, given the logical possibility for forms such as *replyment? Also, a form such as -mit has some kind of morphological status in English in emit, remit, submit, and permit in terms of the stress placement rules, and it has an identifiable origin as a distinct morpheme in Latin. However, it is clearly not a “full” morpheme in modern English in that we can no longer associate it with any identifiable meaning, nor can we readily create new words based on this “root.” In constrast to submit, for example, it is hard to imagine any speaker of English successfully coining a word such as *supramit (cf. Aronoff 1976).

4.2

Language-internal development

A significant number of derivational affixes in pidgins and creoles appear to represent internal developments rather than coming directly from derivational affixes in the lexifier language, having evolved typically from free forms that have been morphologized. Thus, for example, the original pronominal object him with transitive verbs in English has been reanalyzed in Tok Pisin as a fairly productive causative transitivizing suffix on verbs, e.g., sanap ‘stand’ > sanapim ‘put upright’, slip ‘sleep’ > slipim ‘put in a prone position’. Most of the

Pidgin and Creole Morphology 85 suffixed markers in Haitian, including possessive clitics as in tèt-li ‘his head’ (where -li derives from French lui ‘him’), are similarly descended from nonbound forms in French; these appear to represent post-genesis reductions of free forms, which evolved gradually during a lengthy period of use as the primary language of the community. In Fa d’Ambu, free forms such as mina ‘child’, xatán ‘captain’, and syiolo ‘gentleman’ have been reanalyzed as prefixes expressing respectively the diminutive, e.g., mina-palma ‘small palmtree’, augmentative, e.g., xatán-dadalán ‘real crook’, and ameliorative, e.g., syiolo-palma ‘very good palmtree’ (Post 1995, p. 195). In Sranan, the noun man ‘man’ has been morphologized as a human agentive suffix on verbs, e.g., taki-man ‘speaker’ (cf. taki ‘speak’), or – when added to nouns or statives – to indicate a human characterized by something denoted by the stem, e.g., bere-man ‘pregnant woman’ (cf. bere < belly), siki-man ‘sick person’ (cf. siki ‘ill, sick’) (Adamson & Smith 1995, p. 222; see also the discussion of Saramaccan -ma in section 2 above). A problem which we occasionally need to face in this context is the question of appropriate linguistic analysis of particular linguistic phenomena. McWhorter (1998, p. 797), for example, cites Tok Pisin -pasin as a derivational suffix in a form such as gutpasin ‘virtue’. His original source (Mühlhäusler 1985, p. 625), however, clearly indicates that this is a compound based on the roots gut (< good) and pasin (< fashion).

4.3

Zero derivation

While perhaps not strictly speaking a derivational process, mention should also be made of the fact that pidgins and creoles often make fairly extensive use of zero derivation, i.e., the use of a single form belonging to quite different word classes (Voorhoeve 1980). (Of course, this is by no means a feature unique to pidgin and creole languages.) For instance, in Bislama, the noun Krismas ‘Christmas’ can also be used as a verb meaning ‘spend Christmas’. Both simplex and complex forms submit to functional shift in this way. For instance, the Bislama complex adjective longfala (long + adjectival suffix -fala) can be used as a nominal in longfala blong hem [long poss 3s] ‘its length’. Mühlhäusler (1997, p. 196) notes that corresponding to intransitive kuk ‘cook’ and derived transitive kuk-im is the use of the latter with zero derivation in an unambiguously nominal context such as kukim blong saksak [cook-tr poss sago] ‘the (way of) cooking of sago’. For the most part, such constructions appear only in written texts and seem to represent conscious calques on abstract nouns in English, often in translated written material. However, such patterns do occasionally crop up in spontaneous speech and in Bislama, a Verb + Object sequence (wasem maot) was recently observed in an unambiguously nominal context: (4) Bae yumi gat wanem kaen wasem maot? fut 1p have what kind wash mouth ‘What kind of snacks will we have?’

86 Terry Crowley

5 5.1

Other Morphology Compounding

While derivational affixation is generally not highly productive in pidgins and creoles, it appears by way of contrast that compounding is often very productive. Compounding, of course, often lies on the boundary between morphology and syntax and it is perhaps the fact that such patterns can often be viewed as phrase-level rather than word-level processes which accounts for their greater use. Rather than simply incorporating established lexical items from lexifier languages, pidgins and creoles are quite likely to utilize their existing lexical resources to create new compounds. In Tok Pisin, for example, rather than borrowing English volcano, speakers have made use of the pre-existing roots maunten ‘mountain’ and paia ‘fire’ to give maunten paia ‘volcano’. The structure Head + Modifier is clearly unlike English, which would be more likely to have produced a Modifier + Head construction *paia maunten (i.e., ‘fire-mountain’) instead. In this case, the model appears to be the typical substrate pattern of Head + Modifier, in keeping with a general tendency for substrate patterns to be retained in pidgin and creole patterns of compounding (Brousseau 1989). With greater exposure to English, however, new compounds in Tok Pisin are now frequently generated on the Modifier + Head pattern, as illustrated by gris-meri ‘female chatterbox’ (< gris ‘chatter’ + meri ‘woman’). Compounding can be seen in the derivation of a range of vocabulary referring to items for which lexifier languages typically do not have widely used names, or where there are no lexicalized expressions at all. For instance, speakers of pidgins and creoles frequently generate compounds to refer to local flora and fauna. We therefore find examples in Bislama such as stingwud ‘Dysoxylum gaudichaudianum’ (< ‘stink’ + ‘wood,’ because this word refers to a tree that has an offensive smell) and nilfis ‘Arothron hispidus’ aka ‘porcupine fish’ (< ‘nail’ + ‘fish,’ because this fish blows itself up into a ball with spikes sticking out of its skin when it is disturbed). As indicated above, the boundary between morphological compounding and lexicalized phrasal constructions is often somewhat obscure. Forms such as stingwud and nilfis in Bislama clearly stress as single phonological words. On the other hand, items such as sotfala han ‘short-sleeved shirt’ (< sotfala ‘short’ + han ‘hand/arm/sleeve’, where -fala is a suffix deriving from English fellow which is commonly added to monosyllabic attributive adjectives) are stressed as two phonological words. However, it is much more difficult to decide the status of forms such as aelandres (or aelan dres?) ‘Mother Hubbard dress’ (< aelan ‘island’ + dres ‘dress’). The stress pattern seems to be less clear, so it is not certain whether such a form should be treated as part of the morphology or as part of the syntax of Bislama. However, this is again a problem that is not unique to pidgins and creoles.

Pidgin and Creole Morphology 87

5.2

Reduplication

Reduplication also seems to have some kind of special status in the morphologies of creoles and expanded pidgins, including European-lexifier varieties, although it is typically not a productive feature in the morphological systems of IndoEuropean languages. Bakker (1995, 2003) claims, by way of contrast, that reduplication is rare in more rudimentary pidgins. Crowley (1990, pp. 307–8) points out that morphologically productive reduplication did not emerge in Bislama until the language had been developing for well over half a century. However, the fact that forms such as sipsip ‘sheep’ (< English sheep) and pispis ‘urinate’ (< English piss) are found in the mutually intelligible present-day varieties Tok Pisin and Bislama points to the probable use of lexical reduplication in the adaptation of English words into the earliest pidgin from which these two varieties subsequently developed. Cross-linguistically, reduplication appears to be particularly iconic, being commonly used in the expression of meanings such as plurality, variety, attenuatedness, reciprocal action, distributed activity, randomness, durativity, or iterativity. Having made these generalizations, of course, it is possible for reduplication to express additional functions that go beyond any notion of iconicity (see Kouwenberg & LaCharité 2001, 2003). The kinds of functions associated with reduplication cross-linguistically are the very ones that we most frequently encounter when forms are reduplicated in pidgins and creoles. In Fa d’Ambu, for example, reduplication expresses intensity (5a), iteration (5b), and distribution (5c) (Post 1995, p. 196): (5) a. kitsyi ‘small’ > kitsyi-kitsyi ‘very small’ b. nda ‘walk’ > nda-nda ‘stroll’ c. dosy ‘two’ > do-dosy ‘both’ In Bislama, it can also express reciprocal action (6a), random action (6b), and, for noun modifiers, plurality (6c) and variety (6d) (Crowley 1990, pp. 311–20): (6) a. b. c. d.

faetem ‘punch’ > faet-faetem ‘punch one another’ sakem ‘throw’ > sak-sakem ‘throw all over the place’ longfala plang ‘long plank’ > long-longfala plang ‘long planks’ difren kastom ‘different tradition’ > dif-difren kastom ‘all different kinds of traditions’

However, as pointed out, pidgins and creoles also reduplicate in ways that go beyond iconicity of this sort. Sometimes, for example, reduplication is used to signal a change of word class, such as when Saramaccan verbs are used as adjectives, e.g., fátu ‘be fat’ > fátu-fátu ‘fat’ (Bakker, Smith, & Veenstra 1995, p. 171), or when Bislama transitive verbs are used intransitively, e.g., kat-em ‘cut-tr’ > kat-kat ‘cut’.

88 Terry Crowley Although reduplication appears in many European-lexifier pidgin and creole morphologies, and despite the fact that reduplication is not a feature of the superstrate languages, it is often not entirely straightforward to posit a direct morphological transfer from the substrate, even in cases where the substrate languages exhibit patterns that are very similar to what we find in resultant pidgins and creoles. The reason for this is the fact that reduplication has such a narrow range of seemingly iconic functions cross-linguistically that a certain amount of structural parallelism could simply reflect this near-universal iconicity (see discussion in Kouwenberg & LaCharité 2004). However, while the functions of reduplication may well reflect the iconicity of the construction, its specific forms may more directly reflect specific substrate patterns. As an example, Bislama reduplication can be complete (e.g., dringdring ‘drink habitually’ < dring ‘drink’) or partial (e.g., dri-dring, with the same habitual meaning); partial reduplication may copy only certain elements from a sequence of segments, as in Solomons Pijin di-dring. The patterns of partial copying in Bislama and Solomons Pijin reflect features of the local substrates which differ in this respect. Thus, where we find a correspondence between the distribution of different specific patterns such as these and substrate patterns, we are on much safer ground in arguing for direct substrate transfer.

5.3

Clitic morphology

Clitics deserve to be accorded special discussion in an overview of pidgin and creole morphology, particularly given that they have so often been ignored, both in general treatments of pidgins and creoles, as well as in descriptions of individual languages. Clitics are forms which are in some way phonologically bound to other material, yet which function syntactically at a level higher than that of the word to which they are attached. Anderson (1992, p. 201) defines a clitic as “a lexical item whose phonological form does not include assignment to a prosodic unit at the level of ‘word’.” This means that clitics are essentially “phrasal affixes,” which we can illustrate with forms such as the English bound possessive marker -’s. While this looks like an ordinary nominal suffix in an example such as “Jeff’s dog,” its phrase-level rather than purely word-level status is indicated by examples such as “the Prime Minister of New Zealand’s ballgown,” where “New Zealand” is clearly not the owner of the ballgown, and “the guy who I was just talking about’s dog” where the possessive marker is attached to “about,” which is not even a noun. One could almost be misled by the lack of extensive reference to clitics in descriptions of pidgins and creoles into thinking that cliticization is not a major feature of these languages. However, some accounts of pidgins and creoles clearly do recognize the existence of clitic morphology, e.g., Veenstra’s (1996) discussion of Saramaccan TMA markers and its so-called weak pronouns. Part of the problem seems to be the failure of linguistic descriptions to rigorously address the issue of word boundaries. Adamson and Smith (1995) refer to a number of preverbal markers in Sranan which are often treated as

Pidgin and Creole Morphology 89 separate particles, though some have argued that these should be treated instead as verbal prefixes. A clearer statement of how phonological word boundaries should be recognized in Sranan would enable us to better place the discussion of these forms in the grammar. Mention must also be made here of so-called “predicate marking” in Melanesian Pidgin. There is an extensive literature on the functions and behavior of the form i in Tok Pisin, Solomons Pijin, and Bislama which it would be out of place to comment on in any detail here, though see Mühlhäusler (1987), Keesing (1988), Crowley (2000b), and Meyerhoff (2000) for a range of observations. Although i is usually written as an independent word in pre-VP position (e.g., Tom i bin slip ‘Tom slept’), it is clitic-like in that it never receives stress as an independent word. Sebba (1997, p. 110), in fact, states that Tok Pisin i- is an inflectional verbal prefix. However, since this form attaches to both verbs and preverbal elements of any kind, it would clearly be more appropriate to regard it as a clitic than an inflectional prefix. In Bislama, a small number of additional free forms are currently acquiring clitic status, though these forms still alternate with their full-word equivalents. The possessive/purposive preposition blong (‘of, for’) and the general oblique preposition long (‘in, on, at, by means of’) are more frequently attested as blo and lo respectively where the following word begins with a consonant. When the following word begins with a vowel, these often undergo further reduction to bl= and l= respectively (= marks a clitic boundary). These are clearly clitics rather than case prefixes, as they attach to whatever happens to be the initial element of a noun phrase: (7) a. lo(ng) nekis aftenun prep next afternoon

b.

‘in the next afternoon’

l=aftenun prep=afternoon ‘in the afternoon’

c.

l=evri aftenun prep=every afternoon ‘in every afternoon’

Cliticization is not limited to just these prepositional constructions in Bislama. The same two forms can also be used to introduce complement clauses and the same process of optional cliticization applies: (8) Mi kam bl=askem kwestin. 1s come for=ask-tr question ‘I came to ask a question.’ A number of other forms also appear to be currently undergoing phonological reduction to clitic status. In nineteenth-century Bislama, the English adverbial by and by was incorporated as a future tense marker baembae, which underwent phonological reduction to bambae, which subsequently reduced even further to bae, which is the most commonly used future marker today (Crowley 1990, p. 208). This is now undergoing further reduction and has acquired the proclitic variant of the shape b= before a vowel-initial form. Thus, compare:

90 Terry Crowley (9) a.

5.4

bae mi go. fut 1s go ‘I will go.’

b. b=oli go. fut=3p go ‘They will go.’

Morphological reanalysis

Reanalysis of morphologically complex words as monomorphemic in pidgins and creoles is quite common. The Hiri Motu verb itaia ‘see’, for example, is derived from the third person singular suffixed form ita-ia ‘see him/her/it’ in Motu. This type of reanalysis which takes a third person singular inflection as the basis for a new root in a pidgin or creole represents a common theme, in line with a universal tendency in language change for third person singular forms to be taken as the basis for the creation of new paradigms. Interestingly, however, the English verbal third person inflection -(e)s has apparently never been taken as the basis for a new verb root, presumably given the overwhelming presence of zero inflections elsewhere in the paradigm. We find a similar tendency in the reanalysis of morphologically marked possessive constructions on nouns. Inalienably possessed nouns in Motu are suffixed for different pronominal possessors, e.g., tama-gu ‘my father’, tama-mu ‘your father’, tama-na ‘his/her father’. In Hiri Motu, the latter form has been reanalyzed as a morphologically indivisible root. Hiri Motu uses possessive pronouns preposed to the noun instead, e.g., lauegu tamana ‘my father’, oiemu tamana ‘your father’, iena tamana ‘his/her father’. In a historically unconnected situation resulting in the development of Pidgin Fijian out of vernacular Fijian – which is genetically related to vernacular Motu and structurally similar to Motu in many respects – an almost exactly parallel development has taken place, the only difference being that the resulting free pronouns are postposed to the reanalyzed inalienably possessed noun rather than preposed. Thus, compare vernacular Fijian tama-mu ‘your father’ and tama-na ‘his/her father’ with Pidgin Fijian tamana koiko ‘your father’ (Siegel 1987, p. 110). Morphological reanalysis extends also to cliticized constituents at the syntactic level, with lexical items and associated clitics sometimes being reanalyzed as morphologically simple words. The best-known example is that of nouns with proclitic definite articles in French, which are commonly reanalyzed in French-lexifier pidgins and creoles as inseparable parts, e.g., Mauritian Creole lera ‘rat’ (< French le rat ‘the rat’) (Baker 1984; Grant 1995). In Bislama, for which the vocabulary is predominantly of English source, there is nevertheless a substantial subcomponent of the lexicon which derives from French. Many of these French-derived nouns also incorporate the original definite article in this way, e.g., lapul ‘light bulb’ (< French l’ampoule ‘the light bulb’); lafet ‘festival’ (< French la fête ‘the celebration’); lelu ‘children’s game of tag’ (< French le loup ‘the wolf’). Where the pragmaticallly unmarked form of a noun is grammatically plural, this reanalysis may involve instead the plural form of the article, e.g., sora(e) ‘ears’ (< French les oreilles ‘the ears’). Similarly, Bislama items derived instead from Oceanic substrate languages in which there is a widely

Pidgin and Creole Morphology 91 distributed proclitic (or preposed) noun phrase marker na have incorporated this as an unanalyzable part of the root: nakamal ‘meeting house’ < na kamal; namele ‘cycad palm’ > na mele. In English, third person pronominal objects to verbs are typically reduced and unstressed in the spoken language and are arguably clitics in that they are phonologically reduced with the only vowel being a schwa, e.g., give them > give ’em, show him > show ’im. In English-lexifier pidgins and creoles in the Pacific, transitive verbs typically incorporate these pronominal clitics as an inseparable part of the verb root. In Bislama, for example, we now find givim ‘give’. In early New South Wales Pidgin, it was the cliticized form of the neuter pronoun which underwent the same kind of reanalysis, resulting instead in gibit ‘give’.

5.5

Allomorphic variation

Almost as frequent in the literature as observations about the lack of inflectional morphology in pidgins and creoles are comments about the lack of allomorphic variation among their relatively few bound morphemes, whether derivational, inflectional, or clitic. If there is any allomorphic variation at all, it is likely to be encountered with only a single exceptional affix, with most other morphemes being invariant in shape. However, it is certainly not true that allomorphic variation is systematically lacking in pidgins and creoles. There is clear evidence in Haitian Creole of a number of bound morphemes with phonologically conditioned variants, one of which is the enclitic marking definiteness on nouns which varies in shape between -la/-a/-nan/-an (e.g., Nikiema 1999). Third person singular pronominal objects are also noted as alternating in shape between -l and -li (e.g., Muysken & Veenstra 1995, p. 154). The infrequency of allomorphic variation in pidgins or creoles with shallower time-depths than the much older Haitian Creole opens up the possibility that such a feature may in fact be systematically absent in those pidgins or creoles which have only been in existence for a short time. However, the history of the transitive marker in Bislama with the canonical shape -Vm suggests that even this generalization cannot be sustained. Tok Pisin is well known as marking transitive verbs with invariant -im, which is derived from the cliticized form of the object pronoun him in English (see above), yielding transitive examples such as mi rit-im buk [1s read-tr book] ‘I read the book’. While a suffix of the same function is equally widely distributed in Bislama, the historical evidence indicates that by the 1890s – less than 50 years after the language first began to develop – the transitive suffix was beginning to acquire phonologically conditioned allomorphic variants. The final vowel began assimilating to a high vowel in the preceding syllable (Crowley 1990, pp. 298–302) with -em representing the default value of the suffix, yielding variants in modern Bislama such as kil-im ‘kill’, huk-um ‘hook’, and kar-em ‘carry’.

92 Terry Crowley

6

A Morphological Life Cycle?

This discussion has for the most part treated pidgins and creoles as a single category of languages in the search for generalizations about their morphological characteristics. While pidgins and creoles are clearly morphologically reduced in a variety of ways vis-à-vis their superstrate lexifier languages and very often also their various substrate languages, we should ask ourselves if there are any morphological correlates to the traditional distinction between pidgins and creoles. In many early discussions of creolization, it was argued that a structurally minimal pidgin – spoken by definition as a second language by all – would expand lexically and structurally as the pidgin replaced the substrate vernaculars and came to be the dominant (or exclusive) language of the community. Of course, as pointed out in the introduction, assigning a language unambiguously to the category of pidgin or creole is often far from unproblematic. Crowley (1990, pp. 381–90), for example, discusses the impossibility of describing Bislama today either as a pidgin or a creole, and the same has quite possibly been true for much of the period in which the language has been spoken. The fact is that while there are people for whom Bislama is undeniably their native language, such people probably constitute no more than about 10 percent of the total number of speakers of the language, most of whom have acquired a knowledge of Bislama only after first learning one of the Oceanic vernaculars. There are no shibboleths by which we can distinguish a first-language speaker from a second-language speaker of Bislama, and the same has been demonstrated for Tok Pisin (Sankoff & Laberge 1973). Clearly, then, the traditional distinction between pidgin and creole, while it might be useful for discussing aspects of the socio-political contexts, is of extremely limited use when discussing changes in linguistic structures. Mühlhäusler (1997) emphasizes the fact that pidgin and creole languages involve dynamically evolving systems, which implies that we should not expect to find major structural discontinuities at particular stages in the evolution of a newly developed contact language; nor should we expect to find ideal “paradigm cases” of pidgins or creoles. For various stretches of this developmental continuum he uses terminology that many will by now be familiar with. Structurally unstable and lexically impoverished jargons may stabilize to become pidgins. A variety of subsequent changes may see such a language then become an expanded pidgin and, in the appropriate social circumstances, a creole. Yet other social circumstances may see a pidgin or a creole become part of a post-pidgin or post-creole continuum, which may subsequently decreolize (though we need to recognize that a continuum between basilectal and acrolectal features may well be present from the earliest stages of development) (see also Bakker, this volume). Mühlhäusler associates each of these developmental steps with particular kinds of morphological change (as well as other kinds of linguistic changes).

Pidgin and Creole Morphology 93 The jargon stage is associated with minimal fixed grammar and extensive holophrastic reorganization of entire sentences, so we can expect all morphology to be stripped, and for morphologically complex forms to be reanalyzed as morphologically simple. At the stable pidgin stage, we should not expect to see the development of a great deal of morphology, though we sometimes find evidence for the gradual development of a limited number of morphological markers. Mühlhäusler (1997) claims that we begin to see the development of inflectional morphology at the expansion stage, which is the point in the life cycle where present-day Tok Pisin is located. At the creolization stage, he argues that we can expect to find a gradual and ongoing further development of the morphological developments already begun at the expanded pidgin stage, with much greater use of derivational possibilities already evident in occasional patterns at earlier stages. At the post-pidgin and post-creole stage, as expected, we find increasing amounts of morphology being incorporated from the patterns of the dominant language. Some pidgins and creoles appear to have more extensive patterns of morphology than most others. This appears to be true of Saramaccan and Haitian, for instance. However, as I have already indicated in the discussion of allomorphic variation, there are features which are supposedly associated with long periods of linguistic evolution that can be found even in recently emergent pidgins. It is also difficult to find any explanation in terms of a pidgin-creole “life cycle” to account for the contrast between Bislama and Tok Pisin with regard to the early development of phonologically conditioned allomorphs in the former, and the failure of any similar development to take root in the latter, where the transitive suffix is invariant -im. Arguably, both variants of Melanesian Pidgin were – and still are – learned and used by their speakers in broadly similar sorts of social contexts. More recently, the idea of any kind of a morphological life cycle has become the subject of heated debate, and this is a debate which brings the notion of “creole” as any kind of linguistically definable category of languages into question. According to the “superstratist” model, creoles do not arise as a result of the acquisition of native speakers by a structurally minimal antecedent pidgin or jargon; rather, they are seen as beginning as close approximations by second-language speakers of the dominant language in a multilingual contact situation where the disempowered group has rather more access to the superordinate language than has often been assumed (see Alleyne 1971). As slave numbers gradually increased, large numbers of people ultimately acquired approximations of approximations of approximations to (often nonstandard and not infrequently less complex) regional forms of the superordinate language, resulting eventually in the development of what now appear as structurally reduced creoles. Chaudenson (1979) argues strongly for this interpretation of the development of French-lexifier creoles, which Mufwene (1996) extends to creoles in general with his notion of the “founder principle.” However, my own familiarity with

94 Terry Crowley historical sources relating to the development of Melanesian Pidgin in the nineteenth century suggests that this view almost certainly does not apply to all situations. For example, the early years of the sandalwood stations in southern Vanuatu, and the plantations that were subsequently established, offered little opportunity for access to substantial numbers of native speakers of English, as the numbers of Europeans involved were always very low. This leaves open the possibility that any pidgin-creole life cycle should probably still be seen as involving sudden drastic morphological reduction along the lines that have been set out in the body of this chapter. At best, then, the notion of a morphological life cycle involves an increasing openness to particular kinds of morphological developments taking place, but if a particular development can take place at almost any point along the developmental continuum, then this life cycle is more of a rough track that pidgins or creoles can travel along, rather than being a well-made highway from which they may not deviate.

7

Outlook

There is much that pidgin and creole specialists already know about the morphological systems of these languages, and there is much that we can all agree on. The notion of morphological simplicity is certainly fairly central in everybody’s understanding of how these languages should be characterized, at least with respect to their superstrate languages. However, there is also plenty for us to work on in the future. As our understanding of morphological typology in general progresses, additional questions will no doubt present themselves to scholars with interests in pidgin and creole morphology. Clearly, then, pidgin and creole scholars must not ignore developments in morphological typology and morphological theory in general. However, there are some more mundane aims that we should also embrace in attempting to further the study of pidgin and creole morphology. One task that awaits us is the production of more reliable and carefully argued descriptions of individual languages. Crowley (2000b), for example, shows how a number of previous accounts of “predicate marking” in Bislama – including his own – have oversimplified some aspects of what is going on in the language. As noted above, pidgin and creole specialists need to pay closer attention to word boundary issues in order to distinguish affixation from other kinds of morphological processes such as compounding. There also needs to be greater care in distinguishing between inflectional and derivational affixes on the one hand and clitics on the other. The widely held view that pidgins and creoles are of little interest to morphologists because they have “no morphology” is clearly false. In fact, these languages have a considerable amount of fascinating data to offer the morphologist, though we have some way to go before we can safely say that we have adequately laid all of this data out for public analysis.

Pidgin and Creole Morphology 95

NOTE Many thanks to Jeff Siegel and Claire Lefebvre for discussion and comments about issues relating to the content of this chapter. This acknowledgment should not be taken as endorsement of the ideas expressed, for which the author bears final responsibility.

REFERENCES Adamson, Lilian and Norval Smith (1995) Sranan. In: Arends, Muysken, & Smith (eds.), pp. 219–32. Alleyne, Mervyn C. (1971) Acculturation and the cultural matrix of creolization. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 169–86. Alleyne, Mervyn C. (1980) Comparative Afro-American. Ann Arbor: Karoma. Anderson, Stephen R. (1992) A-Morphous Morphology. Cambridge: Cambridge University Press. Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) (1995) Pidgins and Creoles: An Introduction (Creole Language Library 15). Amsterdam: John Benjamins. Aronoff, Mark (1976) Word Formation in Generative Grammar. Cambridge, MA: MIT Press. Baker, Philip (1984) Agglutinated French articles in creole French: Their evolutionary significance. Te Reo 27, 89–129. Bakker, Peter (1995) Pidgins. In: Arends, Muysken, & Smith (eds.), pp. 25–39. Bakker, Peter (2003) The absence of reduplication in pidgins. In: Silvia Kouwenberg (ed.) Twice as Meaningful: Reduplication in Pidgins, Creoles and Other Contact Languages (Westminster Creolistics Series 8). London: Battlebridge, pp. 37–64. Bakker, Peter, Norval Smith, and Tonjes Veenstra (1995) Saramaccan. In: Arends, Muysken, & Smith (eds.), pp. 165–78. Bhattacharjya, Dwijen (1994) Nagamese: Pidgin, creole or creoloid? California Linguistic Notes 24, 34–50. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bold, J. D. (1974) Fanagalo Phrase Book, Grammar and Dictionary. Johnannesburg: Hugh Keartland. Brousseau, Anne-Marie (1989) De “nù-fló” à “po-bouche”: hypothèse sur l’origine des composés en haïtien. Revue canadienne de linguistique 34 (special issue, Claire Lefebvre and John S. Lumsden (eds.) La créolisation), 285–312. Chaudenson, Robert (1979) Les créoles français. Evreux: Nathan. Crowley, Terry (1990) Beach-la-mar to Bislama: The Emergence of a National Language in Vanuatu. Oxford: Clarendon Press. Crowley, Terry (2000a) Simplicity, complexity, emblematicity and grammatical change. In: Jeff Siegel (ed.) Processes of Language Contact: Studies from Australia and the South Pacific. Montréal: Fides, pp. 175–93. Crowley, Terry (2000b) “Predicate marking” in Bislama. In: Jeff Siegel (ed.) Processes of Language Contact: Studies from Australia and the South Pacific. Montréal: Fides, pp. 47–74.

96 Terry Crowley DeGraff, Michel (2001a) On the origin of creoles: A Cartesian critique of neo-Darwinian linguistics. Linguistic Typology 5, 213–310. DeGraff, Michel (2001b) Morphology in creole genesis: Linguistics and ideology. In: Michael Kenstowicz (ed.) Ken Hale: A Life in Language. Cambridge, MA: MIT Press, pp. 53–121. De Rooij, Vincent (1995) Shaba Swahili. In: Arends, Muysken, & Smith (eds.), pp. 179–90. Dijkhoff, Marta B. (1993) Papiamentu Word Formation. PhD dissertation, University of Amsterdam. Dutton, Tom (1985) Police Motu: Iena Sivarai (Its Story). Port Moresby: The University of Papua New Guinea Press. Foley, William A. (1988) Language birth: The process of pidginization and creolization. In: Frederick J. Newmeyer (ed.) Linguistics: The Cambridge Survey, vol. IV, Language: The Socio-cultural Context. Cambridge: Cambridge University Press, pp. 162–83. Grant, Anthony (1995) Article agglutination in Creole French: A wider perspective. In: Philip Baker (ed.) From Contact to Creole and Beyond. London: University of Westminster Press, pp. 149–76. Hall, Robert A., Jr. (1966) Pidgin and Creole Languages. Ithaca: Cornell University Press. Hinnebusch, Thomas J. (1979) Swahili. In: Timothy Shopen (ed.) Languages and Their Status. Cambridge, MA: Winthrop Publishers, pp. 209–93. Holm, John (2000) An Introduction to Pidgins and Creoles. Cambridge: Cambridge University Press. Kahrel, Peter and René van den Bergh (eds.) (1993) Typological Studies in Negation. Amsterdam: John Benjamins. Keesing, Roger M. (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford, CA: Stanford University Press. Kouwenberg, Silvia (1994) A Grammar of Berbice Dutch Creole. Berlin: Mouton de Gruyter. Kouwenberg, Silvia and Darlene LaCharité (2001) The iconic interpretations of reduplication: Issues in the study of reduplication in Caribbean Creole languages. European Journal of English Studies 5 (special issue, Olga Fischer and Max Nänny (eds.) Iconicity in Language and Literature), 59–80. Kouwenberg, Silvia and Darlene LaCharité (2003) The meanings of “more of the same”: Iconicity in reduplication and the evidence for substrate transfer in the genesis of Caribbean Creole languages. In: Silvia Kouwenberg (ed.) Twice as Meaningful: Reduplication in Pidgins, Creoles and Other Contact Languages (Westminster Creolistics Series 8). London: Battlebridge, pp. 7–18. Kouwenberg, Silvia and Darlene LaCharité (2004) Echoes of Africa: Reduplication in Caribbean Creole and Niger-Congo languages. Journal of Pidgin and Creole Languages 19, 285–331. Kouwenberg, Silvia and Pieter Muysken (1995) Papiamento. In: Arends, Muysken, & Smith (eds.), pp. 205–18. Lefebvre, Claire (1998) Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. McWhorter, John H. (1998) Identifying the creole prototype: Vindicating a typological class. Language 74, 788–818. Meyerhoff, Miriam (2000) Constraints on Null Subjects in Bislama (Vanuatu): Social and Linguistic Factors. Canberra: Pacific Linguistics. Mufwene, Salikoko S. (1986) Les langues créoles peuvent-elles être définies sans allusions à leurs histoires? [Can creole languages be defined without reference to their history?] Études Créoles 9, 135–50.

Pidgin and Creole Morphology 97 Mufwene, Salikoko S. (1996) The founder principle in creole genesis. Diachronica 13, 83–134. Mühlhäusler, Peter (1985) The scientific study of Tok Pisin: Language planning and the Tok Pisin lexicon. In: Stephen A. Wurm and Peter Mühlhäusler (eds.) Handbook of Tok Pisin (New Guinea Pidgin) (Pacific Linguistics, Series C, Number 70). Canberra: The Australian National University, pp. 595–664. Mühlhäusler, Peter (1987) Tracing predicate markers in Pacific Pidgin English. English World-Wide 8, 97–121. Mühlhäusler, Peter (1997) Pidgin and Creole Linguistics, expanded and revised edn. (Westminister Creolistics Series 3). London: University of Westminster Press. Muysken, Pieter and Tonjes Veenstra (1995) Haitian. In: Arends, Muysken, & Smith (eds.), pp. 153–64. Nikiema, Emmanuel (1999) De la variation morphophonologique du déterminant /la/ dans les créoles haïtien et st-lucien [On morphophonological variation in the determiner /la/ in Haitian and St. Lucian Creole]. Lingua 107, 69–93. Owens, Jonathan (1997) Arabic-based pidgins and creoles. In: Sarah G. Thomason (ed.) Contact Languages: A Wider Perspective (Creole Language Library 17). Amsterdam: John Benjamins, pp. 125–72. Post, Marike (1995) Fa d’Ambu. In: Arends, Muysken, and Smith (eds.), pp. 191–204. Sankoff, Gillian and Suzanne Laberge (1973) On the acquisition of native speakers by a language. Kivung: Journal of the Linguistic Society of Papua New Guinea 6, 32–47. Sebba, Mark (1997) Contact Languages: Pidgins and Creoles. Houndmills, UK: Macmillan Press. Siegel, Jeff (1987) Language Contact in a Plantation Environment: A Sociolinguistic History of Fiji. Cambridge: Cambridge University Press. Thurston, William R. (1987) How exoteric languages build a lexicon: Esoterogeny in West New Britain. In: Ray Harlow and Robin Hooper (eds.) VICAL 1: Oceanic Languages (Papers from the Fifth International Conference on Austronesian Languages). Auckland: Linguistic Society of New Zealand, pp. 555–79. Van der Voort, Hein (1995) Eskimo Pidgin. In: Arends, Muysken, & Smith (eds.), pp. 137–51. Veenstra, Tonjes (1996) Serial Verbs in Saramaccan: Predication and Creole Genesis (HIL Dissertations 17). The Hague: Holland Academic Graphics. Voorhoeve, Jan (1980) Multifunctionality as a derivational problem. In: Pieter Muysken (ed.) Generative Studies on Creole Languages. Dordrecht: Foris, pp. 25–34. Whaley, Lindsay J. (1997) Introduction to Typology: The Unity and Diversity of Language. Thousand Oaks: Sage Publications.

98 Norval S. H. Smith

5

Creole Phonology NORVAL S. H. SMITH

1

Introduction

The study of “Creole Phonology” could be described as a neglected field. In the sense that relatively little work of any depth has been done on the phonologies of most creole languages, this might be said to be true. The question is whether this is a field at all in any strict sense. I do not intend to spill much ink on the question of whether creole languages are typologically identifiable by reason of their lack of irregularities, or anything else, because I do not regard this as a question of much interest. Rather, it is a matter of common sense that “younger languages” will tend to be less irregular than “elder languages” – that is if one is an adherent of the idea that creoles are indeed younger languages, i.e., languages that have only been created at some roughly identifiable point since the onset of human language. I will proceed on the assumption that this is so. And the very fact that creole languages are more regular in various ways than some other languages is one of the reasons for assuming the correctness of this view. I regard creolization as involving a combination of universal, substrate, and superstrate features, no doubt varying according to the particular situation in which creolization took place. To the extent that universals play a role, then there will clearly be certain unmarked (and regular) morphosyntactic and phonological features present following creolization. The grammar will later tend to become more marked because of the effects of normal historical and contact-induced change. The degree to which universals play a role in creolization will, as I have just said, be no doubt variable, according to the sociohistorical circumstances pertaining to the particular creolization event. McWhorter’s sole phonological universal (2001) that “creole languages do not make use of distinctive lexical tone” is unprovable, since we have no access to the immediate post-creolization phase of most recognized creole languages. It does not seem inherently likely, however, since a number of undisputed creoles in fact make extensive use of tone contrasts, such as

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Creole Phonology 99 Saramaccan, Ndyuka, São Tomense, and Angolar, although it is certainly the case that the study of tone, and prosody in general, is something that requires much more work in the creole field. One important hedge that we must also make concerns the set of languages that have so far been identified as creoles. It is highly conceivable that our views on what creole languages are like are colored to a significant degree by the fact that the best-known creole languages are spoken in the general Caribbean area. These have West European superstrates, such as English and French, which are poor in verbal and nominal inflection, and West African substrates, in particular languages of the Gbe group, which are similarly poorly endowed. It is then perhaps unsurprising that the resultant creoles are also poorly endowed with inflectional morphology. Note that this “poor inflectional endowment” does not imply any qualitative judgment on the complexity of these Caribbean creoles, or of the West African or West European languages involved. It purely means that in the swings and roundabouts of morphosyntax the syntax has more work to perform in these cases, and the morphology less. Now whether creole languages arising from highly complex morphological superstrates and/or substrates would differ significantly in morphological complexity from Caribbean creoles is something we do not really know yet. This is an issue of some importance, however, and it will also have ramifications at the phonological level. In the rest of this chapter I will concentrate on two main aspects of the phonologies of creoles. Firstly, I will deal with various aspects of synchronic phonology. Some will be best explained from the substrate language(s), others from the superstrate language, while universals may be responsible for some developments. Some will be due to the normal phonological processes of change – which may of course be shared by whole groups of creoles, and may also be due to universal aspects of grammar in any case. Others may in fact be due to unrecognized substrate influences. It is difficult to make a clear case for the introduction of true universal phonological features in the creoles of the Atlantic zone, as a number of claimed unmarked aspects – of syllable structure, for example – are also present in potential or known West African substrate languages. Secondly, I will examine the profit that can be achieved by subjecting creole languages to the normal discipline of comparative historical linguistics. To anyone approaching creoles from a universalist point of view this might not seem to make much sense. Some creolists of this persuasion proceed from the assumption that each creole language is created ab ovo novo, as it were. However, an examination of the forms that (sets of) lexical items display suggests clear evidence for groupings into families and subfamilies, in terms of the well-known techniques of historical phonology. The classifications arising from the application of these techniques to the phonological forms of words confirm some classifications made on other grounds, as well as providing evidence for new groupings. It must be emphasized that such features may also be present by reason of later (adstrate) contact. Some “African” features of Gullah (Sea Island Creole/

100

Norval S. H. Smith

SIC), for example, are probably to be ascribed to nineteenth-century African influences, as they are absent in Gullah’s sister language Afro-Seminole Creole (ASC): Lexically and phonologically, SIC and ASC share common forms in excess of 90 percent, with the significant difference that the large number of items from Sierra Leonean languages such as Mende and Vai, as well as a number of features found in Krio (e.g. the marker of habitual action blan(t), and the negated copula nOto do not exist in ASC. The latter creole also lacks the African palatal and labiovelar phonemes found in SIC . . . Sierra Leonean influence upon Sea Island Creole is well-documented for the period 1800–1860. (Hancock 1987, p. 270)

The Afro-Seminole speakers – best known under the term Black Seminole – descend from runaways from the English Carolina colony to Spanish Florida. They were effectively removed from English influence until the Seminole wars of the 1830s–1840s, following which they were moved to the Seminole reservation in Oklahoma. But already in 1848 many had left for Mexico. Some returned in 1870 to work in the US army. No difference has been reported in the Afro-Seminole spoken by the three present historical groups – those still in Mexico, those that returned in 1870 (Texas), and those that did not go to Mexico in 1848 (Oklahoma) (Hancock 1987). This point has been ignored by most creolists working on Gullah. So we must clearly also take account of “African” features entering creoles through later contact – i.e., as adstrate rather than substrate phenomena. And, as is likely in the case of Gullah, even features from other creoles, as in the Krio/Gullah example quoted above. Superstrate phenomena also play a significant role, although as in the case of substrate phenomena, some aspects of superstrate influence could conceivably better be regarded as cases of adstrate influence. I am thinking here of cases where the present “basilectal” form of speech appears to have been modified by reason of contact with the superstrate. The whole question of adstratal influence has pretty much been ignored in creole studies. For non-basilectal creole varieties the general assumption has been that if the superstrate language remains spoken, then a continuum of varieties will arise, with mesolectal varieties taking up an intermediate position between the basilect and the local variety of the superstrate. In fact, the dynamics are very much more complex. Stable “mesolects” may arise at different points in the history of a creole language. One example of this might be illustrated by the example given in Smith and Van de Vate (2006) of the development of diphthongal /ie, uo/ in Jamaican, Kittitian, and other creoles under the influence of southwestern dialects of English. In this sense Jamaican Creole, even in its most basilectal form, might be regarded as “mesolectal,” as compared with the basilectal former Jamaican Maroon Creole (Harris 1994) which now survives as the ritual Maroon Spirit Language (Bilby 1983, 1994). This may also be relevant to some of those varieties actually termed mesolectal, which appear to be more than just mere decreolized varieties of a basilectal

Creole Phonology 101 creole, in that they contain features present neither in the basilect nor in the standard language. This may well be the case with so-called mesolectal forms of Jamaican. Clearly, more informed studies are required of both the creolization and earlier post-creolization phases of many creole languages.

2 The Synchronic Phonology of Creole Languages The synchronic phonologies of creole languages vary widely. Some languages mirror fairly closely the phonologies of their superstrate languages, although the lexical distribution, or “incidence,” of phonemes or suprasegmental phenomena in individual lexical items may differ from the superstrate language in varying degrees; others seem to reflect the phonologies of their substrate languages to a greater degree; while a third category might be seen as doing neither. So Saramaccan displays various features of the Gbe languages spoken by many seventeenth-century slaves. A case of a creole displaying greater than average superstrate influence in its phonology might be Trinidadian Creole English. The now virtually extinct Berbice Dutch of Guyana displays a phonology that in many ways is neither particularly close to either of its two main component languages – (Zeelandic) Dutch and Eastern Ijo of the Niger Delta. Obviously no claim that the phonology of a particular creole language reflects substrate influence can be made without actual linguistic evidence. So general claims of African, West African, or Kwa family influence on particular creole languages are on the whole fairly meaningless. Undoubtedly there has been influence from West or Central African languages on various creole languages, but we have to be more specific about the sources of such influence. In this regard, Bickerton (1981) rightly criticized the “cafeteria” approach whereby some creolists in the past (and present!) have simply looked at any grammars of African languages they had to hand, and assumed that any resemblance they saw, or thought they saw, was proof of substrate influence. He proposed the sensible adage that there had to be some evidence that the right people had been in the right place at the right time to take such proposals beyond the level of mere speculation. This applies in particular to resemblances seen by some creolists in the forms of words in creole languages and what were claimed to be the relevant substrate languages. Demographic research into origins will clearly be revolutionized by the introduction of DNA techniques. Vague references to Kwa language family influence are tempting to make, because such identifications give the impression of greater precision, with regard to creole morphosyntactic patterns. These are however also best avoided. Firstly, while the Kwa languages are typologically fairly similar, they are certainly not identical or very closely related by any means, which gives scope for a considerable amount of fudging. Secondly, as I have already mentioned,

102

Norval S. H. Smith

many aspects of Kwa languages are shared by two of the most important superstrate languages – English and French: a tendency to short or monosyllabic morphemes, a tendency to a lack of inflectional morphology, the importance of aspect in the verbal system, etc. In essence, selecting suitable different characteristics from random Kwa languages is virtually as methodologically unsound as selecting different characteristics from random African languages. I refer to this then as the “Kwafeteria” approach. However, having the right people in the right place at the right time, while a necessary criterion, is not a sufficient one. It is quite possible that a number of slaves were imported with a particular ethnic background without this having had any linguistic consequences. They may have had too low a status, been too few in number, arrived too late in the day, etc. It is obviously wrong to make the assumption that any group of slaves necessarily has some linguistic effect. The relevant sociolinguistic factors have to be applicable before any linguistic effect will result. To demonstrate clear substrate influence, we also need linguistic evidence. A body of lexical items incontrovertibly assignable to a particular African language, or evidence of some uncommon grammatical pattern, and preferably more than one, or some grammatical pattern in combination with a particular grammatical marker from an African language, will go a long way toward demonstrating substrate influence. However, the presence in a creole language of one or two lexical items from an African language is not necessarily proof of direct substrate influence from that language on that particular creole. For instance, a frequent feature of the English-lexifier creoles of the Atlantic area is the presence of the word nyam or something directly derivable from that, such as nyan in Sranan. This word means ‘eat’, and, considering its phonology and semantics, appears to best reflect the form of that word in the Wolof language of Senegal. This certainly should not be taken to imply that Wolof is a substrate language of the Englishlexifier creoles of the Atlantic area in general, or even at all. There was no ascertainable direct substrate influence from Wolof on any of the Atlantic English-lexifier creoles. In fact this word is one of a group of forms of disparate African linguistic origin shared by most of the Atlantic English-lexifier creoles. This group of African items is one I have referred to under the name “Ingredient X,” and which I have claimed to have its origins in an early macaronic pidginized English used in the earliest English colonies in the Caribbean in the period directly following 1625 (Smith 1997). As is known, jargon pidgins (cf. the definition given by Mühlhäusler 1997) take words from a great variety of sources, hence the term macaronic pidgins. If we ignore such isolated forms, fewer languages remain to be regarded as serious candidates for substrate languages. So we require larger numbers of lexical items from a single language to demonstrate direct substrate influence from that language on a creole’s vocabulary. Identifying African items in creole languages is a skilled job, and not one for the uninitiated. It requires expertise both in phonology and in historical

Creole Phonology 103 phonology. The basic techniques date from the nineteenth century and were established by the so-called Junggrammatiker (neogrammarians). Mere casual inspection is not sufficient, for the simple reason that resemblances can be deceptive. Just going by mere resemblances has led to such pre-neogrammarian howlers as the famous wrong relationship between English day and Latin die-s with the same meaning. We know now that such an equivalence is completely nonsensical in Indo-European terms. Day is from Proto-Germanic *daƒaz, ultimately from IE *dhegh- ‘be hot’, while dies is from IE *dyew-s ‘heaven’. Examples of this type that amateur substratist phonologists have produced are legion. Of course Bickerton’s adage regarding the need for evidence that the right people had been in the right place at the right time in order to make substrate influence conceivable concerns only the less directly visible forms of substrate influence, regarding morphosyntactic structure or questions of semantics. As soon as we see evidence of actual forms we are on a stronger footing. So my discovery of a whole series of lexical items and several suffixes of Ijo origin in Berbice Dutch Creole (Smith 1999; Smith, Robertson, & Williamson 1987) provides irrefutable evidence that slaves speaking Eastern Ijo were brought to Berbice (Guyana). While there is no direct record of any slave trafficking from the Niger Delta to Berbice, the linguistic evidence is so strong that it itself provides evidence that at least one ship carried slaves on this route.

2.1

Segmental phonology

I will restrict my discussion to one of segmental inventories and incidence here. By “inventory” I refer to the set of segmental phonemes in a particular language. “Incidence” is a modern term used in English dialect phonology to refer to sets of lexical items, and the occurrence in these of particular phonemes from the inventory of a particular dialect or accent. This term is also useful in the discussion of the phonemes used in the expression of particular (sets of) words in the various creole languages. There is little point in indulging in a general discussion of the phonological systems of creole languages, as they do not form any kind of unique type. While some creoles have fairly unmarked phonological systems, so do numerous other languages, which are not known to be creoles, although the possibility cannot be discounted that some of them are. Thus any discussion of the sound systems of such languages in general would speedily turn into a discussion of unmarked phonological systems in general, which is not what this chapter is supposed to be about. The vowel systems of creole languages tend to resemble those of their superstrates more than their consonant systems do. Sometimes this can be ascribed to coincidence. So Haitian has French as its superstrate language, and probably Fon as its main substrate language. This conclusion is based on Aboh and Smith (forthcoming) where it is argued that while the slave exports from the Gbe areas came largely from what is now southern Benin, in other words,

104

Norval S. H. Smith

from a variety of Gbe-speaking groups, the dominant Gbe language was already Fon. Both French and Fon have nasalized vowels. It is true that Gbe has a larger system than French, but the fact that both had such vowels presumably influenced the fact of Haitian having them. The Suriname creole languages also have nasalized vowels, as does their main substrate language, Fon, while the superstrate, the Early Modern English of southern England obviously does not. The substrates of most other English-lexifier creoles have not been so clearly identified. It is fairly clear that the vowel systems of these other English-lexifier creoles resemble that of English to a much greater extent. For example the tense/lax contrast of English could be seen to have been replaced by a long/short contrast in most of these creoles, whereas the Suriname creoles do not have a reflex of the tense/lax contrast (see section 3.2 for a detailed discussion). Note that neither the tense/lax distinction of most forms of modern English nor its putative long/short correspondent in most Englishlexifier creoles has been properly subjected to phonetic analysis. Until this is done the precise details of the phonological oppositions involved will perforce remain hazy. I will restrict myself in this section to a discussion of marked consonant types occurring in creole languages, but will return to vowel systems in section 3. There are probably few creoles which have identical consonant systems to their lexifiers. The departures are of two types. Firstly we have systems that lack various consonant phonemes possessed by their lexifier languages. So the creole languages deriving their lexica principally from European languages tend in general to lack a number of marked consonants of those languages such as /θ, D, Á, x, ƒ/. Not only are these sounds marked, they do not occur to any great extent in West Africa. West African languages, in turn, also have many marked consonants. Most of the Gbe languages have uvular fricatives /χ, ‰, χw, ‰w/, for instance – including rounded uvular fricatives, which are cross-linguistically rare. A second type of difference is where some creoles display marked sounds or sequences typical of West African languages, such as implosive stops /∫, ∂/ and labial-velar coarticulated stops /kp, gb/, or typical of Central African languages, such as initial nasal-stop combinations /mb, nd/. Most creoles tend to represent these sounds by less marked ones, however, and are thereby at least closer to their lexifiers.

2.1.1

Marked sounds of European languages

French /Á/ is a highly marked segment, but I will not consider it here; I will examine its complex treatment in some French-lexifier creoles in a later section on palatalization (3.3). 2.1.1.1 Dental fricatives /θ, D/ English-lexifier creoles, for instance, generally lack /θ/ and /D/, replacing them with /t, f/ and /d/ respectively. In fact creoles are not alone in avoiding

Creole Phonology 105 such sounds. All other Germanic languages replaced them long ago except for Icelandic and a few now obsolete Frisian languages/dialects. Obviously /t, f/ and /d/ are less marked than /θ/ and /D/. However, we still want to know whether these were the only pronunciations available to the first generation of creole speakers, and whether speakers of African languages were wholly responsible for the substitutions found. It turns out that /t, f/ for /θ/ is widespread in nonstandard London dialect (Cockney). Sivertsen (1960) gives the example of /fáwz@nd ~ táwz@nd/ for /θáwz@nd/. Both variants are also evidenced for seventeenth-century Cockney (Matthews 1938), and in the written logbooks of seventeenth-century seamen (Matthews 1935). It seems then that the contribution of the Africans who were responsible for developing the various English-lexifier creoles may well have been limited to choosing the least marked among these variants. Note that /t, f/ are the only substitutes used, whereas other languages use different substitutions, e.g., Amrum North Frisian has /s/ in inherited lexical items, while French speakers employ /s/ in loans from English. The Suriname creoles and the Anglo-Caribbean creoles differ in the option they choose, interestingly enough. The Suriname creoles select /t/ initially, and /f/ non-initially, although there are admittedly not so many examples, while the Anglo-Caribbean creoles select /t/ in all positions. Cf. Jamaican /tiit/ and Sranan /tífi/ ‘tooth’, both derived from the plural form ‘teeth’. English /D/ is treated differently. A subcategory deserving separate treatment is the group of pronominal/adverbial elements in English with initial /D/: there, then, that, this, though, etc. These are pronounced with /d/ or some derivative in all Atlantic English-lexifier creoles. (“Derivative” are those pronunciations resulting from later developments, such as /l, r/ in some Suriname creoles; cf. brother which is /baála/ in Ndyuka). This possibly reflects the widespread nonstandard/dialect representation of this initial /D/ as /d/ in southeast England. Lexical items with intervocalic /D/ also have largely /d/ in the creoles, avoiding here the well-known Cockney variant /v/. In the word tother ‘other’ (a practically universal British dialect form) an isolated development to /r/ (or developments from that) is widespread. This, or its reflexes, is found in Suriname (Sranan /tra/, Ndyuka /táa/, Aluku /tá(l)a/), Jamaican /tara/, and Gullah /tvQ@/. 2.1.1.2 Velar fricatives /x, ƒ/ Dutch /x/ is yet another marked sound. Not only have the Dutch-lexifier creoles of Guyana and the Virgin Islands had to cope with this sound, but also the English-lexifier creoles of Suriname, containing as they do a large number of (adstratal) Dutch lexical items. Here /x/ becomes /k/. Compare Dutch dag /dAx/ ‘day’ > Saramaccan /∂áka/, Dutch brug /brYx/ ‘bridge’ > Sranan /bróki/. A similar treatment is found in Berbice Dutch, as is illustrated in table 5.1. The forms with /g/ in Berbice Dutch are presumbly to be explained as developments from Dutch /ƒ/; in the case of /hogo/ this would require the form to be derived from the Dutch inflected form hoge.

106

Norval S. H. Smith

Table 5.1 Dutch /x/ > Berbice Dutch Dutch

Phonemic

Berbice Dutch = phonemic

Gloss

acht bocht dag vroeg hoog jagen

Axt bOxt dAx frux/vrux ho:x ja:ƒ@/ja:x@

akti boktu daka fruku hogo jagi/jaki

‘eight’ ‘river bend’ ‘day’ ‘early’ ‘high’ ‘to hunt’

Table 5.2 Dutch /ƒ/ > Sranan Dutch orthography

Dutch phonemic

Sranan orthography

Sranan phonemic

Gloss

negen eigen hagel kogel dreigen

ne:ƒ@/neiƒ@ Eiƒ@ ha:ƒ@l ko:ƒ@l/kouƒ@l drEiƒ@

neygi eygi agra kugru dreygi

néigi éigi ágra kúgru dréigi

‘nine’ ‘own’ ‘lead shot’ ‘bullet’ ‘pester’ (v.)

Dutch /ƒ/ is a very marked sound. We are presumably justified in assuming that the intervocalic development to /g/ in many Dutch items in the Suriname creoles reflects an original voiced /ƒ/ still used in non-Western Standard Dutch. Examples are given in table 5.2.

2.1.2

Marked sounds of African languages

Some creole languages display what are felt to be African marked sounds, such as implosive stops /∫, ∂/, or labial-velar coarticulated stops /kp, gb/, or initial nasal clusters /mb, nd/. Implosive stops, or more strictly, voiced stops produced with the help of an ingressive airstream mechanism, are a feature of various West African languages. An example of a language with an opposition between ordinary voiced /b/ and the implosive /∫/ is Igbo (/∫/ is written gb in this language). Simultaneously articulated labial-velar stops /kp/ and /gb/ are also a feature of the West African linguistic area. In some West African languages /kp/ seems to fill the place that /p/ would otherwise occupy. This occurs for example in Yoruba. In most Gbe languages /p/ is only evidenced in a few onomatopoetic words. Here the sound was apparently lacking in the Proto-Gbe system which was then similar to neighbouring Yoruba in having only /kp/. The opposition /b/ – /gb/ is more frequent.

Creole Phonology 107 Nasal-stop combinations such as /mb, nd, ng/ etc. occur frequently in Central African (i.e., Bantu) languages. Here they are assumed to syllabify word-internally as onsets. Whatever the syllabification of these sounds word-internally in creole languages – sometimes the clusters are claimed to form onsets, and sometimes to be a coda + onset cluster – I will assume for the purposes of this chapter that the sounds concerned are clusters, and not single segments. Note that these sounds, where they occur in creole languages, are not restricted to words of African origin. European-derived items may display them too. Most creole languages simplify words of African origin containing such sounds, however. I will give examples of implosive stops, labial-velar coarticulated stops, and nasal-stop combinations in creole languages. 2.1.2.1 Implosive stops /∫, ∂/ These occur at least in Angolar, Ndyuka, and Saramaccan. Their occurrence in Saramaccan was only reported very recently (Haabo 2002; Smith & Haabo 2004), so that the orthography developed by the SIL in connection with their literacy programs does not reflect them. Their incidence is on the face of it rather odd, as they seem frequently to occur in words of European origin. It is a striking fact, and one which requires explanation, that none of the three substrate languages known to have had any significant degree of influence on Saramaccan – Gbe (principally Eastern Gbe languages), Kikongo, and Akan – now has implosives. In any case /∫/ seems to be much more frequent than /b/ except in a nasal cluster. Kalabari, the Eastern Ijo dialect that has had large-scale influence on the Dutch-lexifier creole Berbice Dutch, has a contrast of /b, d/ with /∫, ∂/, but Berbice Dutch itself does not preserve this, representing both types by plain /b, d/ (but see now Smith & Haabo 2007). Table 5.3 provides some illustrations from Haabo (2002) of lexical items in Saramaccan containing /∫/ (Kikongo etyma here and in the following from Daeleman 1972).1 Initial /b/ is very rare by comparison. Some Portuguesederived items like /baziá/ ‘descend’ from baixhar do not have an implosive stop. This is also true of some English-derived words, such as /bigí/ ‘begin’, and /bígi/ ‘(be) big’. Table 5.4 contains some illustrations of Saramaccan items containing /∂/. Words displaying /d/ in Saramaccan are more frequent than those with /b/ (table 5.5). The reason for this difference in the treatment of English and Dutch lexical items in /b/ and /d/ is not yet clear. In Ndyuka all b’s and d’s are implosive, as also in Angolar (Maurer 1995). 2.1.2.2 Labial-velar coarticulated stops /kp, gb/ These occur in Saramaccan, Ndyuka, and Principense at least. In Saramaccan the occurrence of these sounds is subject to (at least) two dialectal differences. In the Upriver dialect (Upper Suriname River), West African words with /kp, gb/ preserve these sounds, and words of disparate origin containing original /kw, gw/ also display /kp, gb/. In general there is also a degree of allophonic

108

Norval S. H. Smith

Table 5.3 Saramaccan /∫/ Source orthography

Source phonemic

Saramaccan orthography

Saramaccan phonemic

Gloss

bOt(@)l hEvK

bata hebi

∫áta hé∫i

‘bottle’ ‘(be) heavy’

bAt@rEi

batulèi

∫atulùi

bOrst@l

bòsò

∫csO

‘electric torch’ ‘brush’

vistír åburisér

bisi buuse

∫isí ∫uusé

‘to dress’ ‘hate’

mboongo mbuungi

bòngò bundji

∫OngO ∫undZi

‘relations’ ‘mould’

Source phonemic

Saramaccan orthography

Saramaccan phonemic

Gloss

DEm DKs

de disi

∂é ∂ísi

‘they, them’ ‘this’

dAx

daka

∂áka

‘day’

dar di díå

da didia

∂á ∂i∂íya

‘give’ ‘in the daytime’

English bottle heavy Dutch batterij ‘battery’ borstel ‘brush’ Portuguese vestir ‘to dress’ aborrecer ‘to hate’ Kikongo mboongo ‘family’ mbuungi ‘mildew’

Table 5.4 Saramaccan /∂/ Source orthography English them this Dutch dag ‘day’ Portuguese dar ‘give’ de dia ‘in the daytime’

variation between [kp ~ kw] and [gb ~ gw]. In the Downriver dialect (Lower Suriname River), the latter class preserve /kw, gw/ (cf. Smith & Haabo 2004). This correlates with the distribution found in the eighteenth-century recordings (Schumann 1778). To begin, table 5.6 shows some Saramaccan words with original /kp, gb/. And table 5.7 shows some examples of the dialect variation in Saramaccan in words with original /kw, gw/ (upriver = UR, downriver = DR). In the majority of creole languages, words derived from words with original

Creole Phonology 109 Table 5.5 Saramaccan /d/ Source orthography English doctor devil Dutch drie ‘three’ dreigen ‘threaten’

Source phonemic

Saramaccan orthography

Saramaccan phonemic

Gloss

dOkt@r dEv(@)l/dKv(@)l

data didibi

dáta didí∫i

‘doctor’ ‘devil’

dri drEiƒ@

dii delegi

dií delégi

‘three’ ‘pester’

Table 5.6 Saramaccan /kp, gb/ Saramaccan phonemic

Gloss

Fon

Gloss

agbán dagbé dEkpE

‘earthenware pot’ ‘sp. snake’ ‘dagger’

agbÜn dangbé dEnkpè

‘dish, pot’ ‘sp. python, snake god’ ‘cutlass’

Table 5.7 Saramaccan /kw, gw/ UR Saramaccan

DR Saramaccan

Gloss

< Language

Form

Gloss

kpéfa kpéi

kwéfa kwéi

‘baby’s bonnet’ ‘to square off the end of a log’

Portuguese English

coifa square

‘coif’

/kp, gb/ in African languages are pronounced with simple /p, b/. This even applies in the case of languages long removed from the influence of their European lexifiers (table 5.8). Sometimes the simplification is to /kw, gw/, as in table 5.9. 2.1.2.3 Initial nasal-stop sequences /mb, nd/ Once again these (combinations of) sounds only occur in the creoles exhibiting a more African type of sound system. And once again in some cases words from other languages that did not originally possess these sounds have them too. These sounds occur in Saramaccan, Ndyuka, Angolar, and Principense at least. Some English-derived items in Saramaccan and Portuguese-derived items in Principense display them. Their realization varies from language to language. In Saramaccan initial #NC-combinations such as /mb, nd/ are simple onset

110

Norval S. H. Smith

Table 5.8 /kp, gb/> /p, b/ Creole

Form

Gloss

Source

Form

Gloss

Sranan

gbOxEùn

‘asthma’

Fon

kpédZerekún

Sranan

léba

Fon

lágba

Berbice Dutch Berbice Dutch Berbice Dutch Berbice Dutch

pEmbE

‘cough’ ‘consumption’ ‘Xylopia species’ ‘guardian spirit’ ‘hatch (eggs)’

Gun

Sranan

boen(boen) older bohen pegreku

Kalabari kpEmgbE

bErE

‘story’

Kalabari egberi

‘Xylopia Aethiopica’ ‘guardian spirit’ ‘remove crust’ ‘story’

bain

‘cover’

Kalabari gbáKn

‘cover’

potE

‘old’

Kalabari kpc-tE

‘be old (pf)’

Table 5.9 /kp, gb/> /kw,gw/ Creole

Form

Gloss

Source

Form

Gloss

Sranan

dag(u)wé

‘sp. snake’

Fon

dangbé

‘sp. snake’

clusters, while in Ndyuka they are syllabic nasals followed by ordinary consonants – in other words an extra syllable is involved. The number of possibilities is then also, unsurprisingly, greater. For instance, Ndyuka has initial sequences with voiceless stops, like mpiye ‘toucan’, and also with fricatives like nzaú ‘elephant’, while such combinations are forbidden in Saramaccan. Compare the Saramaccan equivalent of Ndyuka nzaú, which is zaun. First table 5.10 presents some cases of initial NC-structures derived from African (i.e., Bantu) languages. In Saramaccan, a variant of Meinhoff’s law (forbidding two nasal clusters in succession) seems to apply to NCV.NCV words (from Kikongo). These lose the first nasal element (table 5.11). This variant of Meinhoff’s law is unusual. Still, it is presumably an African rather than a Surinamese phenomenon. If so, it may aid us in localizing a form of Kikongo which has been especially influential in Suriname. More frequent is the following variant of Meinhoff’s law encountered in Angolar. Here the second nasal cluster tends to have its nasal component removed, as compared to Kimbundu, the important Bantu language of Angola, which is massively represented in the Angolar lexicon (see table 5.12).

Creole Phonology 111 Table 5.10 NC-structures from Bantu sources Creole

Form

Gloss

Source

Form

Gloss

Saramaccan, Ndyuka Saramaccan

mbalu

‘woodchips’

mbalu

‘woodchips’

mbeka

‘steep place’

mbeka

‘cliff’

Saramaccan

ndOki-má

ndoki

‘witch’

Ndyuka

mbukóko

‘black magician’ ‘ibis, flamingo’

Kikongo (Yaka) Kikongo (Kishikongo) Kikongo (Kintandu) Kikongo

mbulukoóko

Ndyuka Ndyuka Principense

mpiye nzaú mbasá

‘toucan’ ‘elephant’ ‘rib’

‘blue plantain eater’ ‘sp. bird’ ‘elephant’ ‘rib, side’

Kikongo Kikongo Kikongo (Kintandu)

mpiya nzawu mbaansya

Table 5.11 Meinhoff’s law in Saramaccan Saramaccan

Gloss

Kikongo-variety

Form

Gloss

bámbi

‘sp. large lizard’ ‘side, wall’ ‘family’ ‘cayman’ ‘mildew’

Kintandu

mbáambi

‘sp. large lizard’

Kintandu Kintandu Kintandu Kintandu

mbaansya mboongo mbóongo mbuungi

‘rib, side’ ‘family’ ‘large wood-lizard’ ‘mildew’

Yombe

ngáanda

‘clearing in village’

Kintandu

ngaandú

‘crocodile’

Kintandu

nyuundú

‘otter’

bandja bOngO boongó bungi/ bundji gandá gandu jundú

‘village, clearing’ ‘cayman (taboo)’ ‘otter’

The examples in table 5.13 are cases where the creole form has nasal-stop clusters that were absent in the corresponding forms in the original languages. The only case where we can give a straightforward substrate explanation is the case of initial /ng/ in Angolar, where /g/ is disallowed word-initially. This is a reflection of the same situation in Kimbundu. All the other cases are somewhat mysterious. The eighteenth-century sources for Saramaccan, for instance, suggest that the English words had plain nasals at that period. On the other hand, the Fon words, where we have a medial cluster, display this in the eighteenth-century sources.

112

Norval S. H. Smith

Table 5.12 Meinhoff’s law in Angolar Angolar

Kimbundu/Kikongo

Gloss

mbándji ~ mbátxi mbita mbogo ndatxi ngàndù ~ ngàtù ngebu

ki-mbandji (Kimbundu) mbinda (Kimbundu) mbòngo (Kikongo) ndanji (Kimbundu) ngandu (Kimbundu) ngembu (Kikongo)

‘side (of body)’ ‘calabash’ ‘descendant’ ‘root’ ‘crocodile’ ‘bat’

Table 5.13 Innovated nasal-stop clusters Creole language

Gloss

Saramaccan mbéti ‘animal, meat’ mbéi ‘make’ ndéfi ‘knife’ andí ‘what?’ ambù ‘who?’ Sranan kindí ‘knee’ Angolar m(b)ezi ‘month’ ngagu ‘stutterer’

Source

Orthographic Phonological form form

Gloss

EME*

meat

/me:t/

‘meat’

EME EME Fon Fon

make knife

/me:k/ /n@if/ /aní/ /mw/

‘make’ ‘knife’ ‘what?’ ‘who?’

/kni:/

‘knee’

English/Dutch knee/knie Portuguese Portuguese

mes gago

‘month’ ‘stuttering’

* EME = Early Modern English

2.2

Suprasegmental phonology

It turns out that the Atlantic creoles nicely illustrate a cross-section of the possible systems occurring on the world scene. In other words, there is nothing particularly creole about any particular suprasegmental type. Basically, the languages of the world can be divided into three major types as far as word-prosodic systems are concerned: 1 2 3

Stress-accent languages Pitch-accent languages Tone languages

Creole Phonology 113 Stress-accent languages mark one syllable in every (non-clitic) word as prominent. The realization of this prominence may vary according to circumstances. It can be longer, exhibit tone movement, and/or be louder. Which of these are used depends on the location of the word in the Intonational Phrase. I will not go into further detail here, as there is an abundant literature on this subject. The important factor here is that there is no fixed association between any particular syllable and any particular tone. Pitch-accent languages come in all shapes and sizes, but have in common that the lexical entries of words in these languages (may) bear tonal information. The type of pitch-accent language illustrated by the creoles that have this can be defined as metrical pitch-accent (Clark 1988), with a H(igh) tone associated with that syllable of the word which is marked as prominent. Tone languages in the strict sense of the term have every syllable of the word associated with at least one tone. There are obviously other types, including mixed cases. We will see that by and large all the creole languages spoken in the Atlantic area fall into one of these three categories.

2.2.1

Stress accent

To give an example of a language that is a simple case of a stress-accent language we can take Berbice Dutch (Kouwenberg 1994), which has a penultimate stress rule. This means that words are (normally) stressed on their secondlast syllables: (1) pámba hulwátri kujáha kàkaláka

‘wing’ ‘wave’ ‘spirit’ ‘cockroach’

In the last example we can see that Berbice Dutch has a secondary stress rule accenting every second syllable leftwards. An exceptional group of words has initial stress in trisyllabic words. (2) dáliki kóroma pótoko

‘just now’ ‘lay eggs’ ‘mud’

When the first of the intervocalic consonants is a liquid, the vowel immediately following it frequently goes unrealized, with the resulting form no longer exceptional: (3) dálki ~ dáliki kórma ~ kóroma Berbice Dutch is a language which makes use of a number of suffixes and many clitic elements. When they are present, the deletion of unstressed vowels

114

Norval S. H. Smith

or the change of vowels to glides frequently serves to achieve the normal penultimate stress pattern: (4) kúmu-tE ními-tE fùndE-tE tóko-apu kùnE-apu bífi-arE bu-arE

> > > > > > >

kómtE níntE fùntE tókap kùnap bífja bwárE

‘come-pf’ ‘know-pf’ ‘find-pf’ ‘child-pl’ ‘person-pf’ ‘say-imp’ ‘drink-imp’

Sranan is also frequently described as a stress-accent language. In general this is true. However it does seem to involve lexical high (H) tone with some ideophones. One case would concern the two ideophones pi(n) and pi meaning respectively ‘quiet’ and ‘black’ (cf. Smith & Adamson 2007). (5) a. a píín 3s quietly (id) ‘He is quiet.’ b. a tan so píí 3s stay so quietly (id) ‘He stayed so still.’ c. a blaka so píí 3s black so very black (id) ‘It’s so black.’ These forms stand in contrast to other ideophones which don’t involve H tones. (6) a.

a krim a oso fia 3s clean det house completely (id) ‘He cleaned the house completely.’ b. a todo djompo so tjubum det toad jump thus ploppingly (id) ‘The toad jumped (into the water) with a plop.’

Ideophones often involve phonological elements not otherwise employed in a given language. For instance, in Ndyuka, the normal open syllable structure may be violated by the occurrence of coda consonants in ideophones (Huttar & Huttar 1994).

2.2.2

Pitch accent

Papiamentu is an example of a pitch-accent language. The most prominent syllable in nearly all words is associated with a H tone. It has been claimed by

Creole Phonology 115 some to be a tone language, but this is inaccurate, since the tone contrasts it displays are minimal (c.250 lexical items according to Joubert 1991) and only appear in disyllabic words. Tone as such functions marginally in the morphology, but there are a variety of tonal polarity effects involving non-accented syllables and function words (Römer 1991). According to Kouwenberg and Murray (1994) the placement of H tone and main stress can be predicted from categorial class membership and syllable weight with few exceptions. The stress is on the final or penultimate syllable. Pre-accentual syllables are either all L or alternating HL – this varies between speakers. In general, too, there is a clear relation between the position of the stress and H tone in a Papiamentu word and the position of the stress in the European model. The vast majority of Papiamentu words are based on such European models. Most of the exceptional disyllabic words referred to above are verbs of Spanish or Portuguese origin. We can take the verb duna ‘give’ as an example. The stress is on the penultimate syllable dúna. However, in such disyllabic verbs the H tone is not realized on the stressed syllable but on the final syllable (an acute marks stress): (7)

H | dúna ‘give(s)’

Historically speaking we can see that the stress pattern has been changed in this group. For instance a clitic (indirect) object pronoun results in the following pattern: (8)

H | dunámi ‘give(s) me’

Trisyllabic verbs of the same origin do not show this stress shift: (9)

H | ataká ‘attack(s)’

H | atakámi ‘attack(s) me’

By reason of this historical shift of stress, a distinction has been introduced between the basic verb form and the participial form in the disyllabic cases: (10)

H | dúna ‘give(s)’

H | duná ‘given’

116

Norval S. H. Smith

This contrast is absent in trisyllabic verbs, however: (11)

H | ataká ‘attack(s)’

H | ataká ‘attacked’

With disyllabic verbs a third form exists in the imperative; once again, trisyllabic verbs do not show this contrast: (12)

H | dúna ‘give!’

H | ataká ‘attack!’

This gives us the contrasts: (13)

H | dúna imperative

H | dúna verb

H | duná participle

Only in the second case is there a dislocation between stress/prominence and the H tone. This is a comparatively minor violation of the usual rule in Papiamentu that primary-stressed syllables bear a H tone, especially when we take into account that there are only about 250 such words in the language. It has been noted by Hubert Devonish in various work, brought together in Devonish (2002a), that various Atlantic creoles such as Guyanese may be regarded as pitch-accent languages, with two contrasting types of pitch-pattern in English-derived words.

2.2.3

Tone languages

Angolar and Santomense both appear to be tone languages (Maurer 1995, 2005). Disyllabic words in Angolar appear to display all four logical possibilities of H (′) and L (`) tones. (14) áwá ‘water’

àntc ‘then’

ándà ‘again’

àndà ‘walk’

The tone situation as regards monosyllables is not yet clear. Clear cases involving lexical as opposed to grammatical words are as yet hard to find. (15) bí ‘come’

fà ‘say’

ì ‘and’

yó ‘island’

mc ‘hand’

dc, rc ‘two’

The tone languages falling in the creole category have yet to be studied thoroughly. Saramaccan, as discussed at greatest length in recent work by Good (2004), is in fact a mixture between a tone language and a pitch-accent

Creole Phonology 117 language. The African vocabulary consists largely of items where every mora is specified for tone, while the European vocabulary consists largely of items where only the mora corresponding to the accented syllable or mora bears a H tone. This H-toned syllable corresponding to the stressed syllables of European words bears other characteristics of stressed syllables, such as a potential greater length under emphasis. The H-toned syllables of African words show no characteristics of stress at all.

3 Diachronic Phonology 3.1

Methodology

The application of traditional historical linguistic methods to creole language data has only taken place on a relatively small scale. Partly this is to be explained by the fact that these methods have been best worked out in the field of phonology, which is precisely the field where work on creole languages has been most sparse. Two examples come to mind of such traditional historical phonological work. One is Devonish (2002b) on “pitch accent” in Atlantic English-lexifier creoles. The other is Smith (1987) on the segmental reflexes of English and Portuguese in the creoles of Suriname. There I attempted to derive new evidence bearing on the genesis of the Suriname creoles by means of the application of traditional historical phonological principles to the recorded data from these creoles from the eighteenth to the twentieth centuries. I will utilize this case to illustrate the usefulness of these traditional principles in the field of creole languages. The method used differs in one crucial respect from the practices of the neogrammarians. In their case they were reconstructing an unknown older historical stage on the basis of later evidence. Cases in point would be the reconstruction of Proto-Indo-European on the basis of Sanskrit, Ancient Greek, Latin, and other Indo-European languages, and the reconstruction of ProtoGermanic on the basis of Old English, Old Frisian, Old Saxon, Old Norse, Gothic, and others. In the case of the Suriname creoles we know that the (principal) lexifier language was English, so we can reasonably take some form of seventeenth-century English as our starting point. Note, however, that I am not at all assuming that the creoles of Suriname are in any way to be regarded as dialects of English. It is clear that from their inception these must be regarded as different language(s) from English – incomprehensible to speakers of any kind of English to be found on the face of this earth! While the form of many (but by no means all) words is clearly derived from English, although considerably altered, other aspects of lexical entries differ considerably, such as the semantics, which show strong influences from African languages (see Huttar, this volume). Other aspects of grammar, like prosodic phonology and syllable structure, morphology, syntax, and sentence semantics are much less like English.

118

Norval S. H. Smith

To return to the Suriname creole–English connection, the strategy I adopted in Smith (1987) was as follows. I took as my basic hypothesis that the primary form of English that was relevant was some kind of London English, primarily Standard Early Modern English (EME). As is well known – although possibly not to creolists – the English that had developed in the general London area is ancestral to all forms of English that have developed external to the British Isles, barring a very few remote and geographically isolated exceptions like parts of Newfoundland. The relevant variety is usually some form of Standard London English, while in the case of Australia, nonstandard London English has had greater influence. This is not to say that Standard English was the first speech form of most migrants to the colonies – this was most certainly not the case. The large majority of migrants from the British Isles spoke a local dialect of English or Scots as their first language, and there were also numerous firstlanguage speakers of Celtic languages, Irish, Scottish Gaelic, Welsh, Manx, and Cornish. However, some form and some degree of Standard (London) English was by the seventeenth century known to most English (and Scots) speakers, whether as an active second dialect, or as a passive second dialect. By this time, the ancestor of Standard English had already won the status war in the British Isles. Many Celtic language speakers at this period were still monolingual, but they were much outnumbered by the speakers of English and Scots. And so in the various colonial environments, where the native speakers of various English and Scots dialects came together, there was no need for the development of a new koiné – some form of Standard English was already waiting in the wings. Colonial English does not exhibit the great variation in phonemic incidence in word forms that dialects in the British Isles do. Anyone doubting this merely has to examine the various dialect studies produced during the last hundred years or so. I quote one example here (modified to IPA). It concerns the modern reflexes of Southumbrian Middle English (SME) /o:/ (/o/).2 (By “Southumbrian” Middle English I refer to forms of Middle English spoken south of the River Humber in England; Northumbrian Middle English (NME), including for our purposes Middle Scots, differed radically from SME especially in regard to the vowel system.) First, in table 5.14 I give examples of three words with a vowel corresponding to SME /o:/ as pronounced in a selection of English and Scots local dialects. A preliminary conclusion might be that the vowels of some of these dialects would be difficult to derive from SME /o:/. This is a correct conclusion as all the dialects except those of Windhill and Hilton qualify as NME, where the vowel was /y:/. Let us in table 5.15 compare this with the pronunciation of the same three words in Early Modern English (i.e., the standard variants used in London in Early Modern English times), a variety of modern Standard Englishes and Cockney, as well as Krio, Sranan and Saramaccan, in cases where the lexical item is used in these languages (– indicates that it is not). The (standard southern seventeenth-century) EME forms seem to bear the clearest relation to the creole forms (the Suriname creole languages arose during the period 1650–1700

Creole Phonology 119 Table 5.14 Modern reflexes/cognates of Southumbrian Middle English /o:/ (adapted from Mather & Speitel 1986) English

Byers Green Co. Durham (Orton 1933)

Dentdale Yorkshire (Hedevind 1967)

Windhill Yorkshire (Wright 1892)

Hilton Dorset (Widén 1949)

Meikle Wartle Aberdeenshire (LAS 3* 1986)

Fair Isle Orkney (LAS 3 1986)

Pirnmill Ayrshire (LAS 3 1986)

Canonbie Dumfriesshire (LAS 3 1986)

hook moon spoon

j@k mI@n spI@n

iuk miun spiun

UIk mUIn spUIn

huk mu:@n spu:n

hjük min spin

hÅk møn spøn

h√k mxn spxn

høk møn

spøn

* LAS = Mather and Speitel (1986)

Table 5.15 EME, standard, and creole reflexes of Southumbrian Middle English /o:/ English

Early Modern English

Southern English Standard

Western Scottish Standard

General American Standard

Cockney

Krio

Sranan

Saramaccan

hook moon spoon

hu(:)k mu:n spu:n

hUk mUun spUun

h¨k m¨n sp¨n

hUk mu:n spu:n

Uk m@¨n sp@¨n

uk mun (s)pun

uku mun spun

huku – –

according to a number of accounts). The modern Standard English varieties are fairly close to the creoles, in particular General American. The forms displayed by the English and Scots local dialects in table 5.14 are mostly very different from the Standard forms and the Suriname creoles, however. The southwestern Dorset dialect (Hilton) is the best fit here, with more or less the same vocalic values as EME. It does display voicing of initial fricatives preceding vowels, however. This is something not found in Caribbean English-based creole languages. Similar points can be made for most vowels in the English system. Some vowels involve an even greater disparity than the one illustrated here. I will return to the issue of the source of the English element in the creole forms in the following section. Let us turn to an examination of three brief case studies illustrating ways in which historical linguistic techniques can be used to extract evidence about the development and history of creole languages.

3.2

Case study 1: Early Modern English vowels > Saramaccan vowels

Here I will briefly examine a creole vowel system that appears most different from the basic Standard English system. The triangular seven-vowel system of

120

Norval S. H. Smith

Saramaccan, being a fairly unmarked system as vowel systems go, is very different from the Standard English system most common Saramaccan words are derived from: i

u e

o E

O a

The Saramaccan vowels may occur as short, long, or extra-long. Further, all types may be nasalized. In addition, a limited number of vowel combinations occur. The relationship between the developments of EME vowels in Saramaccan and in Standard English is so complex that I will not attempt to represent this with any degree of completeness. I will just give the usual EME source, and provide the most typical Saramaccan equivalent. One important difference between Saramaccan and a number of other creoles on the one hand and Standard English on the other hand is the total loss of the tense/lax distinction. In that this contrast appears to be lacking in most languages of the world, it could be seen to be a marked opposition. Some phonologists equate it with the ATR contrast present in many West African languages, but this seems inherently unlikely as the classical tense/lax distinction present in English – and in other Germanic languages like Dutch and German for instance – normally involves both length and quality differences, whereas the length differences are completely lacking in the ATR distinctions in West African languages. However, the distinctions in the mid-vowels of Saramaccan may require to be analyzed in terms of tongue-root position. Saramaccan has a limited kind of stem harmony forbidding sequences of high and low mid-vowels. Whether you analyze the original tense/lax distinction in terms of vowel length or vowel quality doesn’t matter; the result in Saramaccan is nearly always a short monophthong. Only the “real” diphthongs of Standard EME, /ai, au, oi/, survive as such, and then only when they are in word-final position. There are only one or two exceptions to this principle. Vowel length in words of English origin in present-day Saramaccan is always the result of the loss of intervocalic consonants – usually liquids.

3.2.1

Reflexes of English lax vowels

The lax vowels of English appear in Saramaccan as illustrated in table 5.16. The two notable features are the replacement of the difficult English vowel /√/ by /o/ – a treatment occurring frequently in Caribbean creoles – and the appearance of an unrounded correspondent of /O/, /a/. Instead of six lax vowels we end up with five vowels.

3.2.2

Reflexes of English tense vowels

The tense vowels of English appear as in table 5.17. Here we see unrounding of the tense English vowel congener of /O/, /O:/.

Creole Phonology 121 Table 5.16 English lax vowels > Saramaccan Early Modern English

Saramaccan

Examples

Gloss

I E a √ O U

i e a o a u

líbi bédi báka djógu gádu údu

‘live’ ‘bed’ ‘back’ ‘jug’ ‘god’ ‘wood’

Table 5.17 English tense vowels > Saramaccan Early Modern English

Saramaccan

Example

Gloss

i: e: a: O: o: u:

i e (E) a a o (O) u

míti fési láfu wáka boóko fútu

‘meet’ ‘face’ ‘laugh’ ‘walk’ ‘broke’ ‘foot’

Once again we see a reduction in the number of six vowels to five, ignoring the variation in mid-vowels (where both open and close values occur, with the open mid-vowels being in the minority), which is partly explicable in terms of other factors. So we end up with the same vowels that appear as equivalents of the lax vowels. This is not a change that occurs in any historical dialect of English. The lax/tense opposition – however realized – is present in all local English dialects of the British Isles and England’s former colonies, with the possible exception of Scottish forms of English. What can we learn from the above facts? First, that Saramaccan (and the other Suriname creoles) shares with other creoles spoken in the Caribbean area a number of developments in the vowel system, supporting the kinds of dispersal patterns of (Proto) English Creole which I argued for in Smith (1998). Second, that, in general, we can establish rules relating the Saramaccan outcome to the EME input. All the Saramaccan forms, barring a couple of exceptions, are derivable from forms that were used in EME. This direct relatability of the Suriname Creole and EME systems (demonstrated in more detail in Smith 1987) has obvious consequences for how we view the creolization process in the circum-Caribbean area. It means that English dialect input was insignificant in the formation of the Suriname creoles.

122

Norval S. H. Smith

3.3 Case study 2: Palatalization of velars in French Caribbean creoles An example of the historical development of elements derived from a lexifier language concerns the palatalization of various segments in Trinidad Creole French and other French-lexifier creoles in the Caribbean area. First, palatalization can be observed where French velar stops are followed by front vowels. However, it is notable that the vowels concerned are all mid-vowels, and that this process does not seem to be illustrated preceding the high front vowel /i/. The former fact is illustrated in table 5.18 (– indicates that it is not known whether the form is used). There are not many cases illustrating the sequences /ki, gi/, but Creole French items with this sequence seem mostly to be exempt from palatalization, as shown in table 5.19, where Table 5.18 French velar stops followed by /e, E/ (Trinidad French Creole data from Thomas 1969 [1869]; Dominica French Creole data from KEK 1991) French

Trinidad French Creole

Dominica French Creole

Gloss

quinze marquer (em)barquer quelque quelqu’un baguette guèpe gai guère guérir

tSúz mAtSe bAtSe tSEk tSetSú badZEt dZEp – dZEr –

tSúz mAtSe bAtSe tSEk tSEz{ badZEt dZEp dZe dZE dZewi

‘fifteen’ ‘mark’ ‘embark’ ‘some’ ‘someone’ ‘baguette’ ‘wasp’ ‘merry’ ‘hardly’ ‘cure’

Table 5.19 French velar stops followed by /i/ (Trinidad French Creole data from Thomas 1969 [1869]; Dominica French Creole data from KEK 1991) French

Trinidad French Creole

Dominica French Creole

Gloss

qui quitter crier gris grille guide en guise de

ki kite kriye gri

ki kite kiye/kwiye gwi gwi gid ÑdZiz

‘who, what?’ ‘leave’ ‘call’ ‘grey’ ‘grill’ ‘guide’ ‘instead of’

ÑdZiz

Creole Phonology 123 only the form derived from French “en guise de” shows palatalization. For a number of these cases it could be claimed that the /w/ in Dominican or /r/ in Trinidadian blocks the palatalization. For instance, Sranan, where /k, g/ are palatalized when followed by /i/, displays a blocking of palatalization by an intervening /r, w/; hence, no palatalization is observed where /kr, gr, kw/ are followed by /i/. However this will not explain those cases in Frenchlexifier creoles where this blocking factor is lacking, but where palatalization still does not take place, as in the creole reflexes of French qui, quitter, guide. When we examine the front rounded vocoids of French, such as /Ái, y, ø, œ/ we see a different picture. Here we do not have to take regard of vowel height at all. In every case an initial velar is palatalized. There are two different developments. In the first the fronting appears to be directly transferred from the vowel to the preceding consonant, palatalizing it. To illustrate this, first consider the cases involving the French creole diphthong /Ái/ (table 5.20), then the front rounded vowels /y, œ/ (table 5.21). This process is different from what happens with the set of words in table 5.22, where the front vowels remain, although the preceding velar stop is still palatalized. The rounding of the vowel is however lost in these cases. When we look at these forms, we have to ask ourselves why palatalization was able to complete its work in the cases where the French creole vowel is /i/. We cannot speak of transfer of a palatal feature in these cases, as we saw above Table 5.20 French velars followed by /Ái/ (Trinidad French Creole data from Thomas 1969 [1869]; Dominica French Creole data from KEK 1991) French

Trinidad French Creole

Dominica French Creole

Gloss

cuite cuillèr cuisse cuisine

tSwit

tSwit tSwiyE/tSuyE tSwis/twis tSwizin/twizin

‘cooked’ ‘spoon’ ‘thigh’ ‘kitchen’

Table 5.21 French velars followed by /y, œ/ (Trinidad French Creole data from Thomas 1969 [1869]; Dominica French Creole data from KEK 1991) French

gueule gueuler (re)culer cueillir

[œ] [œ] [y] [œ]

Trinidad French Creole

Dominica French Creole

Gloss

dZOl

– dZole

‘mouth of animal’ ‘howl’ ‘recede’ ‘pick’

tSule tSuyi

124

Norval S. H. Smith

Table 5.22 More French velars followed by /y, œ/ (Trinidad French Creole data from Thomas 1969 [1869]; Dominica French Creole data from KEK 1991) French

cœur écume figure culotte rancunière

[œ] [y] [y] [y] [y]

Trinidad French Creole

Dominica French Creole

Gloss

tSEr

tSE tSim fidZi tSilOt

‘heart’ ‘foam’ ‘figure’ ‘trousers’ ‘malice-bearing’

fidZi tSilOt rÑtSinEz

that palatalization is not triggered by high vowels in the case of unrounded front vowels, but is by mid-vowels. If it is just assimilation, the front high vowels should always participate. What is the lesson to be learned here? Here we find a constellation of facts that seems to lack logic in some cases. At this point, more information is required before we can tease out the sequence of historical changes involved. A larger number of forms needs to be examined, and a larger number of “dialects,” including the closely related St. Lucian, Grenadan, Martinican, and Guadeloupean, and possibly others. Note, however, that it is precisely because of their lack of a simple explanation in terms of natural phonetic developments that such aberrant patterns can be useful in historical phonological studies. We have mentioned the similar but not identical palatalization facts in Sranan. The search for an explanation, or combination of explanations, for the French creole facts is likely to involve both the superstrate and the substrate. We should not forget in this connection that palatalization of velar stops before front vowels occurs also generally in the modern Gbe lects (Capo 1991).

3.4

Case study 3: Reflexes of Middle English /E:/ in the Suriname creoles

In modern English the Middle English (ME) vowel /E:/ has three reflexes. The true inherited reflex is /e:/, but dialect mixture in the London area led to the widespread dominance of an originally nonstandard reflex /i:/. In addition, a third reflex had already arisen by shortening of the ME vowel to /E/. Some examples of these reflexes in modern English are given below in (16). Items that occur in the Suriname creoles are included in bold type. (16) a.

examples of the /e:/-reflex great, break, steak, yea, wear, tear, bear, there, pear

Creole Phonology 125 b.

c.

examples of the /i:/-reflex speak, leave, leaf, clean, heap, meat, dream, pea(se), beam, cleave, beat, eat, ear, hear, beard, weary, near, dear, here (?), rear, shear, clear examples of the shortened /E/-reflex dead, head, already, heavy, lead, spread, bread

The items in (a) and (c) that occur in the Suriname creoles exhibit a vowel reflex that followed the Standard English development. However, some of those in (b) display irregularity (as far as Standard English is concerned). Examples of the regular reflexes would be as in (17). Since Sranan does not have an open/close distinction in the mid-vowels, the vowels in both (17a) and (17c) have /e/ as their Sranan reflex. The forms in (17b) that exhibit the same development as Standard English show up with /i/. (17) Sranan:

a.

brekten bréki wéri de b. píki líbi krin c. déde éde bréde

‘noontide’ (< break-time) ‘breakfast’ (1783) ‘wear’ ‘there’ ‘answer’ (< speak) ‘leave’ ‘clean’ ‘dead’ ‘head’ ‘bread’

However, there are several forms in (16b) which display a different development, reflecting the inherited /e:/-reflex, rather than the Standard English /i:/-reflex (see table 5.23). For our purposes we can ignore the last two forms (‘ears’ and ‘hear’), as these have similar reflexes in all Anglo-creoles in the Caribbean area, and therefore must receive a different explanation.

Table 5.23 Forms containing an irregularly inherited /e/

meat dream beam weary pease ears hear

Sranan

Ndyuka

Saramaccan

Schumann 1778

méti dren – wéri pési jési jére

méti deén – wéli pési jési jée

mbéti – bén wéi pési jési jéi

meti – bem wêri (pisis) jessi

126

Norval S. H. Smith

However, we can now consider the meaning of the odd fact that the various Suriname creoles illustrate six forms between them displaying the original Standard English reflex, while Standard English itself has adopted another set of forms from some nonstandard dialect. Two conclusions can be drawn from this. First, that as far as the development of this vowel is concerned, the Suriname creoles reflect the ME dialect ancestral to Standard English better than Standard English itself. Secondly, while the Suriname creoles also reflect the process of replacing the /e:/-reflex with the /i:/-reflex, as we can see from the forms in (17a) above, this was less advanced in the form of English spoken in the Caribbean area between 1625 and 1667. Dobson (1957), the standard work on EME pronunciation, says: “As the seventeenth century advances, the evidence for [i:] from a ME variant e [see n.1] progressively increases, i.e., the [i:] variant is recorded in more and more words (especially in vulgar sources).” In other words, the English reflected in the Suriname creoles dates from a period before the absorption of nonstandard forms from the /i:/-dialects had reached the stage that it eventually did later on in the seventeenth century, when it finally ceased altogether. No other Atlantic area English-lexifier creoles are known which share this feature. This allows us to date the English items in the Suriname creoles as “earlier” than in other English-lexifier Caribbean creoles. Note that we can be fairly certain that it is a form of Standard English that was involved here, because of the cases in (16c) that undergo shortening. The examples are not numerous, but English dialects vary to a fair extent in the words that undergo this shortening, and where Standard English shows shortening, we find a mid-vowel in the Suriname creole reflexes. So what useful conclusion can we draw from this relevant to the question of the grouping of the Atlantic creoles in general? In view of the numerous points of resemblance between the English creoles of Suriname and the Caribbean, we have to conclude that they share some common ancestor. Whether this was a stable pidgin, or some kind of Proto-creole is a point requiring more attention. However this difference in the reflexes of ME /E:/ means that the influence of Standard English continued to exert itself on the Caribbean creoles, but not on the Suriname creoles. For a similar conclusion on other grounds, see Smith and Van de Vate (2006).

4

Conclusion

I started out with the premise that creole phonology was not really a field at all. In purely phonological terms it is not. However, the value of studying the synchronic and historical phonologies of creole languages does not lie in what we can learn about phonology, for nothing novel is to be learnt here. The value to be gained from the incorporation of phonological arguments and insights into the general creolist debate is vast, however. Arguments on groupings and

Creole Phonology 127 subgroupings may be based on phonological criteria. And the non-European vocabulary of the Atlantic creoles is a goldmine for phonologically informed researchers interested in African sources. Together with the increasing knowledge about the details of the trans-Atlantic slave trade, early colonial history, the increasing number of older textual sources that are becoming available, and the promise of future information from the DNA of Afro-American populations, phonology will help make creole studies more dynamic for the foreseeable future.

NOTES 1 Tables 5.3–5.13 make use of data from Haabo (2002). 2 /o/ and /e/ are the standard Anglicist way of referring to long high-mid vowels in Middle English, i.e., /o:/ and /e:/.

REFERENCES Aboh, Enoch O. and Norval S. H. Smith (forthcoming) Migrations and geo-linguistics in the eastern Aja-Tado cultural area. In Pieter C. Musken and Norval S. H. Smith (eds.) The Transatlantic sprachbund: Benin and Surinam. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma Press. Bilby, Kenneth (1983) How the “older heads” talk: A Jamaican Maroon language, spirit possession and its relationship to the creoles of Suriname and Sierra Leone. New West Indian Guide 57, 37–88. Bilby, Kenneth (1994) Maroon culture as a distinct variant of Jamaican culture. In: E. Kofi Agorsah (ed.) Maroon Heritage: Archaeological Ethnographic and Historical Perspectives. Kingston, Jamaica: Canoe Press, 72–85. Capo, Hounkpati B. C. (1991) A Comparative Phonology of Gbe. Berlin/Garome, Benin: Foris Publications/Labo Gbe. Clark, Mary (1988) An accentual analysis of the Zulu noun. In: Harry Van der Hulst and Norval Smith (eds.) Autosegmental Studies on Pitch Accent. Dordrecht: Foris, pp. 51–79. Daeleman, Jan (1972) Kongo elements in Saramacca Tongo. Journal of African Languages 11, 1–46. Devonish, Hubert (2002a) Talking Rhythm Stressing Tone. Kingston, Jamaica: Arawak Publications. Devonish, Hubert (2002b) On the Sierra Leone–Caribbean connection: Hot on the trail of the “toneshifted” items in Anglo-West African varieties. In: Devonish (2002a), pp. 165–80. Dobson, Eric J. (1957) English Pronunciation, 1500–1700. Oxford: Clarendon Press. Focke, Hendrik C. (1855) Neger-Engelsch Woordenboek. Leiden: P. H. van den Heuvel. Good, Jeff (2004) Tone and accent in Saramaccan: Charting a deep split in the phonology of a language. Lingua 114, 575–619.

128

Norval S. H. Smith

Günther, Wilfried (1973) Das portugiesische Kreolisch der Jlha do Principe. Marburg: The author. Haabo, Vinije (2002) Grammatica en fonetiek van het Saramaccans. Unpublished manuscript. Hancock, Ian F. (1987) A preliminary classification of the anglophone Atlantic creoles, with syntactic data from thirty-three representative dialects. In: Glenn Gilbert (ed.) Pidgin and Creole Languages: Essays in Memory of John E. Reinecke. Honolulu: University of Hawai[i Press, pp. 264–334. Harris, Col. C. L. G. (1994) The true traditions of my ancestors. In: E. Kofi Agorsah (ed.) Maroon Heritage: Archaeological Ethnographic and Historical Perspectives. Kingston, Jamaica: Canoe Press, pp. 36–63. Hedevind, Bertil (1967) The dialect of Dentdale in the West Riding of Yorkshire. Uppsala: University of Uppsala. Huttar, George L. and Mary L. Huttar (1994) Ndyuka. London: Routledge. Joubert, Sidney M. (1991) Dikshonario (Handwoordenboek) Papiamentu-Hulandes. Willemstad, Curaçao: Cromotip. KEK (1991) Dominica’s diksyonnè Kwéyòl–Annglé; English–Creole Dictionary. Roseau, Dominica/Cave Hill, Barbados: Konmite Pou Etid Kweyol/The Department of Use of English and Linguistics, University of the West Indies. Kouwenberg, Silvia (1994) A Grammar of Berbice Dutch Creole. Berlin: Mouton de Gruyter. Kouwenberg, Silvia and Eric Murray (1994) Papiamentu (Languages of the World/ Materials 68). Munich: Lincom Europa. Mather, J. Y. and H.-H. Speitel (1986) The Linguistic Atlas of Scotland, Scots Section, vol. 3: Phonology. Beckenham: Croom Helm. Matthews, William (1935) Sailors’ pronunciation in the second half of the seventeenth century. Anglia 59, 192–251. Matthews, William (1938) Cockney, Past and Present: A Short History of the Dialect of London. London: George Routledge & Sons. Maurer, Philippe (1995) L’angolar: un créole afro-portugais parlé à São Tomé. Hamburg: Helmut Buske Verlag. Maurer, Philippe (2005) The tonal behaviour of disyllabic nouns in Santomense: A preliminary study. Paper presented at the Conference on Creole Language Structure between Substrates and Superstrates, Max Planck Institute for Evolutionary Anthropology, Leipzig. McWhorter, John (2001) The world’s simplest grammars are creole grammars. Linguistic Typology 5, 180–2. Mühlhäusler, Peter (1997) Pidgin and Creole Linguistics. London: University of Westminster Press, pp. 128–38. Orton, Harold (1933) The Phonology of A South Durham Dialect. London: Kegan Paul, Trench & Co. Römer, Raúl G. (1991) Studies in Papiamentu Phonology. Amsterdam/Kingston, Jamaica: University of Amsterdam/University of the West Indies. Schumann, Christian L. (1778) Saramakkanisch-Deutsches Wörterbuch. Unpublished manuscript, Bambey, Surinam. Sivertsen, Eva (1960) Cockney Phonology. Oslo: Oslo University Press. Smith, Norval S. H. (1987) The Genesis of the Creole Languages of Surinam. PhD dissertation, University of Amsterdam. Smith, Norval S. H. (1997) Ingredient X: The common core of African words in the Atlantic creoles. Paper delivered to the Society of Pidgin and Creole Linguistics Conference, University of Westminster, London, June.

Creole Phonology 129 Smith, Norval S. H. (1998) The vowel system of 18th century St Kitts Creole: Evidence for the history of the English creoles? In: Philip Baker and Adrienne Bruyn (eds.) St Kitts and the Atlantic Creoles: The Texts of Samuel Augustus Mathews in Perspective. London: University of Westminster Press, pp. 145–72. Smith, Norval S. H. (1999) Pernambuco to Surinam 1654–1665? The Jewish slave controversy. In: Magnus Huber and Mikael Parkvall (eds.) Spreading the Word: Papers on the Issue of Diffusion of Atlantic Creoles. London: University of Westminster Press, pp. 251–98. Smith, Norval S. H. and Lilian Adamson (2007) Tonal phenomena in Sranan. In: Parth Bhatt and Ingo Plag (eds.) Stress, Tone and Intonation in Creoles and Contact Languages (special issue of Sprachtypologie und Universalienforschung/Language Typology and Universals). STUF 59, 211–18. Smith, Norval S. H. and Vinije Haabo (2004) Suriname creoles: Phonology. In: Bernd Kortmann and Edgar W. Schneider (eds.) A Handbook of Varieties of English, vol. 1: Phonology. Berlin: Mouton de Gruyter, pp. 525–64. Smith, Norval S. H. and Vinije Haabo (2007) The Saramaccan implosives: Tools for linguistic archaeology? In: Bettina Migge and Norval S. H. Smith (eds.) Substrate Influence in Creole Formation (special issue of Journal of Pidgin and Creole Languages, 22), 101–22. Smith, Norval S. H., Ian E. Robertson, and Kay Williamson (1987) The Ijo element in Berbice Dutch. Language in Society 16, 49–90. Smith, Norval S. H. and Marleen Van de Vate (2006) Population movements, colonial control and vowel systems. In: Parth Bhatt and Ingo Plag (eds.) The Structure of Creole Words: Segmental, Syllabic and Morphological Aspects. Tübingen: Niemeyer, pp. 59–82. Thomas, John J. (1969 [1869]) The Theory and Practice of Creole Grammar. London: New Beacon Books. Widén, Bertil (1949) Studies on the Dorset Dialect. London: Williams & Norgate. Wright, Joseph (1892) A Grammar of the Dialect of Windhill, in the West Riding of Yorkshire. London: Kegan Paul, Trench, Trübner & Co.

130

6

Peter Bakker

Pidgins versus Creoles and Pidgincreoles PETER BAKKER

1

Introduction

The question “are pidgins a separate class of languages?” can be answered in different ways. First, one can address the possibility of a special status of pidgins, by considering whether pidgins are in any way different from the rest of the languages of the world. Second, one can discuss whether pidgins are different from other contact languages, notably creoles. This paper focuses on the second question. In many publications in the field of creolistics, one encounters generalizations about linguistic and sociolinguistic properties that are supposed to be valid for both pidgins and creoles. In fact, on closer scrutiny many of these properties appear not to be true for pidgins. This paper will consider the differences between pidgins on the one hand, and creoles and a third class, pidgincreoles, on the other hand. Obviously pidgins differ from other natural languages in at least one important social respect: pidgins are by definition nobody’s mother tongue. If it is the case that structural differences between non-native pidgins on the one hand and mother tongues on the other can be shown to exist, then one could perhaps attribute some of these differences to nativization, i.e., the social process through which speech varieties that are not first languages transform into mother tongues. One could interpret the structural changes as a consequence of this process. In other words, pidgins may lack properties found in all native languages, and therefore be different from other languages. In this paper it is argued that the role of nativization is overestimated as a factor in explaining the structural differences between pidgins and creoles. Other social circumstances, in particular the circumstance of languages becoming default languages of daily communication, will be argued to have much stronger effects. Before being able to contrast pidgins with creoles, it will be necessary to define these languages, and other types of contact languages that display

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Pidgins versus Creoles and Pidgincreoles 131 considerable reduction attributable to contact. This takes place in section 2. There I will argue for the necessity of distinguishing not two, but four classes of contact languages that have come about through processes of reduction of other languages and (in some cases) subsequent expansion. On social grounds I distinguish jargons, pidgins, creoles, and an intermediate class, pidgincreoles. These are all commonly used terms, except for pidgincreole, which was suggested by Philip Baker (p.c.) as an alternative to the ambiguous term expanded/ extended pidgin. A pidgincreole is a former pidgin that has become the main language of a speech community and/or a mother tongue for some of its speakers. Pidgincreoles are not creoles, because they are not the language of an ethnic or political group or the mother tongue for the whole community. In a quest for structural features of pidgins, pidgincreoles should not be taken into consideration – even though this has often been done. Social differences among these four types of languages will be considered in section 3. There are two social parameters that distinguish jargons and pidgins from creoles and pidgincreoles on social grounds: nativization and urbanization. Section 4 illustrates structural differences between the social categories of pidgins, pidgincreoles, and creoles. In section 5 I will assess the structural impact of these parameters, contrasting pidgins with pidgincreoles and creoles in this regard.

2 The Life Cycle Model of Pidgins and Creoles A classic view of pidgins and creoles is that of the life cycle, in which jargons (unsystematic and variable forms of a second language used in interethnic communication) can become pidgins (normative systems of communication that have to be acquired, but which are not yet mother tongues), and pidgins in their turn can become creoles, i.e., mother tongues and main community languages. It was even commonly assumed that all creole languages had descended from pidgins. This model of the life cycle of pidgins and creoles goes back to the nineteenth century, but was first formulated explicitly by Robert A. Hall, Jr. in the 1950s (cf. Hall 1962). The life cycle model is especially prevalent in the literature about Pacific contact languages (Romaine 1988; Mühlhäusler 1997). The life cycle illustrated below is based on Mühlhäusler’s (1997, p. 12) version. It relates to reduced and restructured forms of English.

Jargon

Stable pidgin

Expanded pidgin

Post-creole continuum Post-pidgin continuum

Lexifier (English)

Creole

Lexifier (English)

132

Peter Bakker

A pidgin (a “stable pidgin” in Mühlhäusler’s terms) may evolve into an expanded pidgin (our “pidgincreole”) if it is used in a range of functions (e.g. not only trade, but also art). On the other hand, if it is heavily influenced by the lexifier, it may lead to a “Post-pidgin continuum,” with forms ranging from the basic pidgin to the standard language (here English). The pidgincreole (“expanded pidgin”) may become a mother tongue (“creole”) and/or change into a continuum with the lexifier at one extreme end. Not all of these developments are historically documented. For instance, not all creolists think that creoles were necessarily preceded by a pidgin and/or pidgincreole (“expanded pidgin”) stage. I will leave these questions aside for now. Readers should note that the labels used by linguists for these languages are often different from the labels used by speakers. In particular, speakers’ use of the terms “jargon” and “pidgin” rarely coincide with the technical terms used by creolists (cf. Mufwene 1997). One should therefore not rely on the language labels themselves in order to determine whether a language is a jargon, a pidgin, or a creole. As with many categories used in the humanities, there are no clear-cut boundaries between jargons, pidgins, pidgincreoles, and creoles. For example, pidgins may become mother tongues for only a limited number of people: are they then to be considered pidgins or creoles? In this chapter a number of social criteria will be formulated that distinguish jargons, pidgins, pidgincreoles, and creoles; and these criteria will be shown to correlate with structural properties. In order to clarify my use of the terms, I contrast the terminology used here with other terms used in creolistics: My terms

Other terms used by linguists

jargon

unstable pidgin, early pidgin, pre-pidgin, pre-pidgin continuum, primitive pidgin, incipient pidgin, rudimentary pidgin stable pidgin extended pidgin, expanded pidgin, stable pidgin, creole creole

pidgin pidgincreole creole

Let us consider some examples of jargons, pidgins, pidgincreoles, and creoles. The brief extracts below are of course insufficient to show the real character of the four types of language, but they give at least an impression. The further we work down the list of the four types, the more the languages appear to have fixed structural properties and grammaticalized features. Note, however, that these four types do not necessarily represent developmental stages.

2.1 Jargons Jargons are characterized structurally by a lot of variation, interference from the mother tongues of the speakers, the experimental use of words from several

Pidgins versus Creoles and Pidgincreoles 133 languages, severe simplification of structure, very short and simple phrases, and a lack of structural norms. Here are two examples of a jargon spoken in Hawai[i from at least the last quarter of the eighteenth century: (1) Pehea you KAITAI, you hanahana all same lili more why 2s bastard 2s make-work all same little more me hanamake you! 1s make-dead 2s ‘Why, you bastard, if you do that again I’ll kill you!’ (1898, Chinese plantation worker, cited in Bickerton 1995, p. 52; 1999, p. 35; Hawaiian words bold, Cantonese in capitals) (2) Take care. By and by you dead. Tiana too many men. ‘Take care. You will soon be dead, because Tiana has many men.’ (late eighteenth-century, Bickerton 1999, p. 33; my gloss)

2.2

Pidgins

A jargon may evolve into a pidgin. A pidgin displays structural norms and less variation than a jargon. There is less experimentation on the part of the speakers. Pidgins are morphologically less elaborated than their lexifiers, and categories like tense, aspect, and plurality are seldom obligatory in pidgins. The Icelandic Pidgin speakers in (3) use a reduced compromise between several Scandinavian languages, but the pronouns are English or West Flemish, and they are reflexes of accusative case forms rather than nominative. Only one generic locative preposition (po) is used. In the Chinese Pidgin English in (4), the forms of some of the English words are changed, and there are several structural divergences from English, e.g., the use of the pronoun my as an object pronoun and the noun classifier piecee. (3) [Two sailors meet and converse:] A: Ju haber runet po den langi? you have rowed loc the land? B: Jú, jú, mi vildir ekki blíve po den kug Yes, yes, me want not stay loc the ship ‘A: Did you row to the land? B: Yes, yes, I did not want to stay on the ship.’ (Icelandic Pidgin “Scandinavian”; Bakker 1989, p. 129) (4) Boy! Makee pay my that two piecee book ‘Give me those two books, boy!’ (Chinese Pidgin English, as recorded in 1860; Baker & Mühlhäusler 1990, p. 99) Many pidgins draw their vocabulary mostly from one language, as is true of Icelandic Pidgin and Chinese Pidgin English in (3) and (4), but some draw from two languages, e.g. Russenorsk:

134

Peter Bakker

(5) Kak ju wil skaffom ja drikke te, what you will eat-vb and drink tea davaj på sjib tvoja ligge ne jes på slipom please on ship you lay down yes at sleep-vb ‘If you want to eat and drink tea, then come on board and lie down to sleep.’ (Russenorsk, a twentieth-century pidgin used between Norwegian fishermen and Russians around the North Cape; Russian elements italicized; Broch & Jahr 1984, p. 38; my gloss)

2.3

Pidgincreoles

Under certain social circumstances (to be discussed below), pidgins may expand structurally to become pidgincreoles. This is the case, for example, in Solomons Islands Pijin and Sango, spoken in the Central African Republic. The Pijin sentence in (6) displays the grammatical marking of transitivity with -em as well as the presence of the clitic i (see Crowley, this volume), while the Sango example in (7) shows the use of a particle, tí, to signal subordination as well as verb serialization as a strategy for rendering factive verbs nonfactive. (6) kat-emu googo finis hem-i safen-em lelebeti bodi cut-tr aux be.finished srp(3s) shave-tr slightly body bulong hem mek-em hem-i fiti insaeti longo baa bulong raefolo poss 3s caus-tr srp(3s) fit inside poss barrel poss rifle ‘He cut it and then he shaved down the casing slightly, so that it would fit inside the rifle barrel.’ (Solomons Islands Pijin; Kwaio first language; Keesing 1988, p. 233) (7) Kosi à-balà-híngà à-tznE baba tí lò à-kírì àwè. Kosi sm-see/know sm-say father subo 3s sm-return already ‘Kosi thought/was sure that his father had returned.’ (Sango pidgincreole, Central African Republic; Pasch 1997, p. 235)

2.4

Creoles

All three examples of Caribbean creoles below show Subject-Verb-Object (SVO) word order and preverbal particles expressing tense, mood, and aspect. Example (8) also illustrates predicate clefting (a verb focus construction), which is very common in Atlantic creoles. In all three examples of creoles (as in the pidgincreole in (6)), we see some examples of clitics: object pronouns in (8) and (10), and the possessive pronoun and the preverbal particles in (9). (8) ta duna m’a duna-bu e buki foc give 1s-pf give-2s.acc the book ‘I GAVE you the book.’ (Papiamentu Spanish/Portuguese Creole, Caribbean; Kouwenberg & Muysken 1995, p. 212)

Pidgins versus Creoles and Pidgincreoles 135 (9) granpapa-m t-ap-kondi yon lame grandfather-1s past-prog-lead ind army ‘My grandfather was leading an army.’ (Haitian French Creole, Caribbean; Hall 1953, p. 203)

(10) mi bi paamísi én taa mi ó go fom’én 1s ant promise 3s.acc comp 1s irr go beat-3s.acc ‘I threatened him to beat him up.’ (Saramaccan, De Groot 1977, p. 25) In addition, creoles generally show no morphological case marking and no gender marking, but some may distinguish nominative/accusative in (part of) their pronoun systems (e.g. Sranan). Most creoles have articles that mark specificity and knownness rather than definiteness, which is what articles express in their European lexifiers, but Cape Verde Creole and Guinea Bissau Creole have no definite articles. This illustrates some of the structural properties of the four types of languages. We now turn to their distinction on social grounds (section 3). The social groupings are then tested against a number of structural properties, whereby pidgins appear to differ systematically from pidgincreoles and creoles (section 4).

3 Social Differences The distinguishing criterion between pidgins and creoles that is usually advanced in linguistic literature is nativization: if a restructured language is a mother tongue, it is considered to be a creole. Even though linguistics textbooks tend to claim this, many creolists consider this criterion problematic (see Hancock 1980; Arends 1989; Singler 1992; and also Arends & Bruyn 1995). As I have indicated, in my view it is better to distinguish between four types – jargons, pidgins, pidgincreoles, and creoles – as is usually done in Pacific creolistics (Romaine 1988; Mühlhäusler 1997), but hardly ever in the Atlantic area. Both classifications are confronted with borderline cases, so the four classes remain gradient. If, however, one seeks to isolate structural properties of pidgins as against creoles, it is better to place pidgincreoles in a separate class. Pidgincreoles share some sociolinguistic properties with creoles and should not, therefore, be taken as representative of pidgins. Jargons have also been taken as examples of pidgins, producing an inaccurate impression of the social and structural properties of pidgins. In this section I will discuss distinctive social features of these four types of languages. A jargon is a set of individual ways of communicating used by people without a common language where there is a necessity of communication, for

136

Peter Bakker

example in a trade situation. Such a situation leads to some basic gestural and verbal contact. In a jargon, there are no norms with regard to what is acceptable. When these individual solutions become more systematic, one can call them a pidgin. Holm defines a pidgin and a jargon as follows: A pidgin is a reduced language that results from extended contact between groups of people with no language in common; it evolves when they need some means of verbal communication, perhaps for trade, but no group learns the native language of any other group for social reasons that may include lack of trust or of close contact . . . By definition the resulting pidgin is restricted to a very limited domain such as trade, and it is no one’s native language . . . Although individuals can simplify and reduce their language on an ad hoc basis (for example, New Yorkers buying sunglasses in Lisbon), this results not in a pidgin but a jargon with no fixed norms. (Holm 1988–9, pp. 4–5)

There are many alternative definitions of a pidgin. This is especially significant when creolists discuss the pidgins that are believed to have preceded creoles. Bickerton for instance uses jargon data and calls these “pidgin” when he discusses the language situation in early twentieth-century Hawai[i and illustrates the widely variable ways of speaking of immigrants who came to Hawai[i as young adults, and whose speech was recorded by Bickerton and his associates when they were old people, apparently assuming that their language use had remained unchanged during their lifetime. According to Bickerton, this highly variable speech developed into a systematic language when it became a mother tongue (see also section 5). Whereas Bickerton considers this highly unstable state characteristic of pidgins, in our definition pidgins (in contrast to jargons) have structural norms and display much more stability. The type of “broken English” recorded by Bickerton and documented by Roberts from written sources from the second quarter of the nineteenth century (cf. S. Roberts 2004) would rather fall under the label of a jargon. One can distinguish different kinds of pidgins with regard to the situations in which they are used. All pidgins are used in interethnic contacts, ranging from trade to religious activities and diplomacy. Trade pidgins are used in bartering and selling between different groups. Maritime pidgins are used on shipboard among multilingual crews, and also in the contacts of such sailors with other crews and people they meet on the mainland. Workforce pidgins may arise in situations where the labor force is multilingual, for example in mines or plantations. Some military pidgins are a special kind of workforce pidgin, used among soldiers; others are used by soldiers and officers in their contact with the local population. Some authors also mention what one could call tourism pidgins, whose users would be guides, hotel owners, and the like, with their regular customers. These have not been studied in detail (see Mehrotra 1997 for a possible example). Obviously almost all pidgins are used in a range of social circumstances, and have more than one function. Most pidgins emerge to facilitate communication, e.g. Russenorsk in Northern Europe, where users were sometimes unaware that they were using a reduced

Pidgins versus Creoles and Pidgincreoles 137 compromise language (Broch & Jahr 1981, 1984). Other people use pidgins in order to protect their native language from being understood by outsiders (cf. Thomason & Kaufman 1988). Thomason (2001, p. 173) quotes a missionary who claimed that the Delaware used Pidgin Delaware in order to keep their own language secret, and the same has been observed elsewhere (see Versteegh, this volume). Pidgins can also be characterized on the basis of the source of their lexicon. Most pidgins derive almost all of their vocabulary from one language, and therefore terms like “Pidgin English” are used. There are also pidgins with a mixed lexicon deriving from two languages, such as Russenorsk, Trio-Ndyuka Pidgin, and some others that came into being in specific circumstances (see below). These can be called “50–50 pidgins” (my term). Such mixed lexicons appear to be more common in jargons than in pidgins (cf. Bickerton 1999). A pidgin as defined here has the following social characteristics: (a) A pidgin is not learned as a first language. (b) A pidgin is not the main or default language of an ethnic, social, or political group. (c) A pidgin has structural norms and therefore has to be learned as such. A pidgin may grow into a pidgincreole under the following circumstances: (d) When the use of the pidgin is extended to become the main language of a speech community and/or when the pidgin acquires mother tongue speakers. Structural expansion usually takes place as a consequence of both types of social extension. As a correlate of the social characteristics (a–c), pidgins (but not exclusively pidgins) have the following structural properties: (e) A pidgin is reduced when compared to its source language(s). (f) A pidgin is therefore relatively poor in morphology. (g) A pidgin has a reduced lexicon deriving from the lexicon or lexicons of the “dominant” group(s). The structural features (e–g) only pertain to reduction; some non-reduction pidgin features will be discussed below. Linguistic reduction in itself is found in other social situations as well (e.g., language obsolescence, koineization). Severe reduction (e.g., a lexicon of 100–1,000 words, few grammatical elements) should be taken as diagnostic of a pidgin. Property (a) distinguishes pidgins and jargons from creoles and in fact from all other natural languages. Property (b) excludes varieties that have become the mother tongue for some speakers, or the dominant language in a speech community, even though speakers as well as linguists may continue to call them “pidgins.” Property (c) excludes spontaneous forms of second-language

138

Peter Bakker

speech, like the speech of migrant workers who try to cope with the language of their new country. Varieties like “Gastarbeiterdeutsch” have sometimes been called “pidgins” (e.g. Clyne 1968; Heidelberger Forschungsprojekt “PidginDeutsch” 1977), but they are not pidgins according to this definition since these varieties are very individual, despite some cross-speaker regularities. Furthermore, these varieties are not the goal of language learning, i.e., they do not have norms: one cannot be a good or bad speaker of “Gastarbeiterdeutsch.” Only if such an immigrant variety became a target of language learning might it be considered a “pidgin,” but to our knowledge this has not been documented. Property (c) also excludes the type of speech used by people who simplify their own language when talking to foreigners (Foreigner Talk). Property (d) specifies when a variety is no longer a pidgin on social grounds. When pidgins become mother tongues, and especially when they become community languages, they can no longer be taken to be representative of the class of pidgins. This brings us to (e–g), being some general structural properties of pidgins. (Section 4, in which pidgins are contrasted with creoles, contains further discussion of the structural properties of pidgins.) Property (e) excludes artificial languages. Pidgins are new languages, non-native, reduced, and restructured forms of another language: many semantic and grammatical distinctions (gender, tense, aspect) from the lexifier do not find their way into the pidgin, analytic structures are preferred to synthetic structures, and morphological markers tend to be dropped or reduced. The term “simplification” is avoided here because a language does not necessarily become simpler if it is reduced in this manner. If an English sentence “She hits him” is reduced to “him hit him,” then it is simplified as to case marking, tense marking, and the expression of the gender of the participants, but for the hearer it may actually be more complex to process because there is more ambiguity as to the gender and role (actor, patient) of the people involved and possibly also with regard to the time the action takes place (cf. also Mühlhäusler 2001). Thus, reduction and simplification are two different processes. Property (f ) is a consequence of a pidgin’s reduction. Even though all pidgins are reduced, some pidgins remain reasonably complex in their morphological system, albeit always reduced in comparison with their source or lexifier languages (see section 4.3 for examples). With regard to property (g), many creolists have claimed that pidgins and creoles are always based on the language of the (politically and economically) dominant group. However, for pidgins, as often as not, a local language which is not dominant in the political or economic sense can provide the lexical basis, as is the case for Carib Pidgin, Eskimo Pidgin, Delaware Pidgin, and Chinook Wawa in the Americas, which all have non-European lexifiers; for some of these, it has been argued that their genesis postdates European contact (see discussion in Versteegh, this volume). Local knowledge and survival skills gave these populations advantages, and thus a form of power. Thus, “dominant” may refer to an economically or politically dominant group (e.g., an invading group with superior weapons), or it may refer to a socially or

Pidgins versus Creoles and Pidgincreoles 139 numerically dominant group (often the local people). When a pidgin develops between equal partners, for example in certain trading situations, the lexicon may also be divided more or less equally between the two languages of the trading partners (50–50 pidgins). Pidgincreoles constitute a class in between pidgins and creoles. A pidgincreole is a restructured language which is the primary language of a speech community, or which has become the native language for only some of its speakers. In practice, it is often hard to draw a line between pidgins and pidgincreoles, or between pidgincreoles and creoles. Pidgincreoles may have a few native speakers, for example the children of parents who shared no language but the pidgin. Also, from a diachronic point of view, it is difficult to decide when a pidgin becomes a pidgincreole, even in a small community of speakers. A pidgin formerly used only in interethnic contacts may slowly gain in use in a community and become the default language to be used outside or even inside the immediate family. This gradience does not mean that the social distinctions between these varieties are useless. On the contrary, it may be revealing to see to what extent pidgincreoles are structurally more similar to pidgins and to what extent they are more similar to creoles. Therefore, we separate pidgincreoles from both creoles and pidgins in this study. The most conspicuous pidgincreoles are the restructured varieties of English spoken in the Pacific and West Africa. The pidgincreoles of the Pacific region are known locally under such names as Tok Pisin (Papua New Guinea), Pijin (Solomons Islands Pidgincreole), and Bislama (Vanuatu) (Keesing 1988; Baker 1993). Shared features between these languages and Australian pidgins and creoles (Tryon & Charpentier 2004) suggest the stability of some structural features at an early date. This makes it very difficult to find data in the region for varieties that are clearly pidgins rather than pidgincreoles. The contemporary varieties of Aboriginal English such as Kriol of Australia (Sandefur 1979) and Broken of the Torres Straits Islands (Shnukal 1988) are creoles rather than pidgins or pidgincreoles. One Pacific variety which never expanded or creolized is Cantonese Pidgin English of the China coast (Hall 1944). In West Africa there is a wide range of restructured forms of English from the Gambia south to Cameroon (a stretch of some 3,500 kilometers), including Liberia, Ghana, Nigeria, and Equatorial Guinea. They have different local names like Pidgin, Broken, or Kamtok, and almost all qualify as pidgincreoles. However, Krio of Sierra Leone and related Aku/Oku of Gambia would qualify as creoles as these are native languages, and also the main language of their respective political unities or ethnic groups. In Liberia, the term “Liberia English” actually refers to two different varieties, one a pidgincreole, the other not: the restructured variety widely spoken by indigenous Liberians is Vernacular Liberian English, a pidgincreole, but the language of Liberia’s small Settler minority is a descendant of nineteenth-century African American Vernacular English, hence not a pidgincreole (Singler 1997). There are also pidgincreoles with non-European lexifiers (see Versteegh, this volume). In Africa there are Ngbandi-based Sango in the Central African Republic, Bantu-based Kituba (and Lingala) of Congo. In Asia there is Naga

140

Peter Bakker

Pidgin or Pidgin Assamese (Sreedhar 1974). Hiri Motu in the southern part of Papua New Guinea is also a pidgincreole, with an Austronesian lexicon. Creoles differ from pidgins on the social grounds that creoles are mother tongues, and the main language of a speech community. A speech community may be defined on ethnic grounds, e.g., the Saramaccans, or on political grounds, e.g., the inhabitants of Mauritius or Suriname. In sum, a reduced contact language which has no native speakers and is not the main or primary language of a speech community is a jargon or a pidgin. A restructured language that has become a community language and/or the native language of an ethnic or political group belongs to the class of pidgincreoles or creoles.

4

Pidgins versus Pidgincreoles and Creoles: Structural Differences

There are a number of structural differences between the class of pidgins on the one hand and the class of creoles and pidgincreoles on the other. In some respects, creoles may be more homogeneous structurally than pidgins. Seuren (1998, pp. 292–3) characterizes creoles as follows: “If a language has a Creole origin it is SVO, has TMA [tense-mood-aspect] particles, has virtually no morphology.” In this section, I will show that none of these properties (or others) of creoles of the colonial expansion era are valid for pidgins. Pidgincreoles are structurally like creoles, and will be mentioned only in passing. Jargons are not discussed here, since they are rather variable and not subject to structural norms. However, to the extent that they have structural features, jargons resemble pidgins.

4.1

Word order

The only creoles with word orders which diverge from the basic SVO order are some creoles in Asia and Grand Ronde Chinook Wawa in the northwestern United States. Sri Lanka Portuguese and Sri Lanka Malay both have verbfinal word order. This, however, is undoubtedly a recent phenomenon, under pressure of local Sinhalese and Tamil, as earlier sources of these creoles show the creole SVO word order (see Bakker 2000, 2006 for a brief comparison, based on data in Smith 1979, 1984, Adelaar 1991, Hussainmiya 1987 and other sources). Recent developments in some Philippine creole varieties of Spanish may similarly be responsible for VSO ordering, presumably under pressure of local languages like Ilocano, Cebuano, and Hiligaynon (see, e.g., Grant 2001). Word order in the creole Grand Ronde Chinook Wawa is free (both OV and VO). Subject markers precede the verb, and object markers follow the verb. (This variety is discussed below in section 5.1.) Apart from these exceptions, even if the lexifier has a different word order, creoles turn out as SVO. Berbice Dutch Creole has SVO word order

Pidgins versus Creoles and Pidgincreoles 141 (Kouwenberg 1992), even though both Dutch and Eastern Ijo, which is the African language cluster involved in the genesis of the creole, have SOV as their basic word order. Arabic is a verb-initial language, but Nubi Creole Arabic of Kenya and Juba Creole of Sudan have SVO as their basic word order (Heine 1982; Watson & Ola 1985; Wellens 2005). In contrast with the clear creole preference for SVO order, pidgins show the word orders we find in the languages of the world in roughly comparable proportions, i.e., mostly verb-medial or verb-final word order. If the word order of a pidgin is not or not only SVO, it appears that the word order it displays is also present in one of the contact languages, as in Pidgin Ojibwe, an Amerindian pidgin of the Great Lakes area, which has the variable word order of its lexifier (John Nichols 1995). Similarly, Herschel Island Eskimo Pidgin has free word order, with SOV statistically prevalent, like its lexifier (Stefánsson 1909; Van der Voort 1997). In other cases the pidgins display the word order of one of the “substrate” groups. For instance, Ussuri Pidgin Russian has SOV as its basic order, like the native Tungusic language of its speakers (Johanna Nichols 1986), and so have Kjachta Pidgin Russian (Wurm 1992) and Hiri Motu of Papua New Guinea (Holm 1988–9, p. 586). Yokohama Pidgin Japanese shows verb-final word order, like Japanese (Holm 1988–9, p. 594). In Papua New Guinea, the neighboring but linguistically distinct Papuan languages Yimas and Arafundi both have free word orders, but the lexically mixed pidgin used between these groups has SOV word order (Foley 1988, 2006). Mobilian, which was spoken in the American southeast, and which had the Amerindian Muskogean language Choctaw as its main lexifier, had OSV and SOV word order (Drechsel 1997). It seems, on the other hand, that SVO has developed in some pidgins where the lexifier did not have SVO order. For instance, even though Hawaiian is VSO, Pidgin Hawaiian has SVO as its basic word order ( J. Roberts 1995a, b). In some cases there is an evolutionary development, as in Pidgin Fijian, which has SVO as its basic word order, but had VSO and later both VSO and SVO in earlier phases (Siegel 1992). This represents a development from a highly unusual word order to SVO, the universal order for creoles and possibly the most common one for pidgins. As an explanation for the near universality of SVO order in creoles, one could argue that this order is the clearest and most transparent way to distinguish the grammatical functions of subject and object. On the other hand, most pidgins lack morphological case marking to distinguish subject and object, but still many pidgins do not display SVO word order. There is a clear difference between creoles, which consistently show SVO word order, and pidgins, in which a variety of word orders can be found.

4.2

The expression of tense, mood, and aspect

All creoles make use of a system of preverbal particles to express tense, mood, and aspect (TMA) (Bickerton 1981, 1984; Givón 1982; Bakker, Post, & Van der

142

Peter Bakker

Voort 1995). According to Bickerton (1981, 1984), the core of these creole systems expresses anterior tense, irrealis mood, and non-punctual aspect, in this linear order, before the verb. The systems show some striking similarities across creoles of different lexifiers, but there are also important differences among individual creoles (see, for example, the articles in Singler 1990). Pidgins, on the other hand, do not have these preverbal particles. Preverbal particles are not the only way of expressing TMA in creoles. Some creoles inherited TMA forms from their lexifiers. A number of Portuguese creoles display several inherited forms and categories, e.g., Korlai Portuguese (Clements 1996, p. 27) which has four different forms of the verb: (11) base form gerund simple past perfect

-V -Vn -o/-Vw -Vd

kat-a kat-an kat-o kat-ad

‘sing’ ‘singing’ ‘sang’ ‘sung’

Portuguese Portuguese Portuguese Portuguese

cant-ar cant-ando cant-ou cant-ado

Other TMA categories in this language are expressed using preverbal markers, which is the more usual creole way. Pidgins show much less uniform marking of TMA, and no pidgin displays anything similar to the TMA system found in creoles. This may be the most significant structural difference between creoles and pidgins. Pidgins express these categories almost always adverbially, using free morphemes such as “tomorrow” or “long-time-ago” or “possible,” and their expression is not obligatory. Some pidgins use inherited morphological marking, such as -ato for past in the Italian-lexifier pidgin known as Ethiopian Simplified Italian (Marcos 1976) and the Mediterranean Lingua Franca (Schuchardt 1909), or nga- for past in the Bantu-based Pidgin A-70 (Alexandre 1963) (see 4.3 for more examples). Other pidgins such as Fanakalo and Pidgin Ojibwe have a more elaborate set of forms (Bold 1958; John Nichols 1995). Aspect is prominent in creoles, but expression of durativity, habituality or perfectivity and other aspectual categories is rarely found in pidgins. In short, TMA is expressed very differently in pidgins and creoles.

4.3

Morphology I: Inflection

Contact languages may have different sources for their inflectional markers: inflectional morphemes can be inherited from the lexifier, they can be borrowed from another language in the contact situation, they can be spontaneous, language-internal formations, developed through grammaticalization, or they can be calques on structures in a different language (for instance, a preposition used as a case ending because some dominant language has a case system). In Bakker (2003c) I give several examples of all four of these possibilities, showing that inflectional morphology is in fact more frequent in pidgins than in creoles. There are several examples of inherited TMA markers additional to those in the preceding section. South African Fanakalo (Pidgin Nguni), for

Pidgins versus Creoles and Pidgincreoles 143 example, has several inflectional and derivational suffixes, e.g., -ile “past” (< Zulu “perfect”), -a “present,” -wa “present passive,” -iwe “past passive,” zo- “future” (< Zulu “immediate future”), -isa “causative” (Bold 1958, pp. 54–5). Turku (Chad Pidgin Arabic) displays some number inflection in verbs (Tosco & Owens 1993, pp. 217, 220). Arafundi-Enga pidgin and Pidgin Ojibwe show person agreement on the verb (Williams 1995). Most of these morphemes are inherited, but sometimes they do not derive from the lexifier. For example the predominantly Malay-based Broome Lugger Pidgin of northwestern Australia borrowed several bound morphemes from Japanese (Hosokawa 1987), with a change in meaning: -ya “predicate formation” (Japanese -da/-ja “copula”) and -ka “question, conditional” (Japanese -ka “interrogative, indefinite marker”). Bakker (2003c) lists examples from eight more pidgins in a non-exhaustive overview. The higher incidence of inflection in pidgins than in creoles may be the result of a simple accident of history: many of the documented creoles seem to have originated in contact situations involving isolating languages and inflectional languages with restricted paradigms (Western European languages, Malay, Arabic – the last with rich morphological fusion but few affixes), whereas known pidgins have often arisen out of contact involving agglutinative, polysynthetic, and more heavily inflectional languages (e.g., Bantu, Amerindian). Also, pidgin speakers are often familiar with other morphologically complex languages, some of them related to the lexifier, which may have contributed to the inheritance of morphological features in the pidgin. Speakers of morphologically complex languages such as Bantu languages and West Atlantic languages were involved in the formation of some Caribbean creoles, but this has apparently had no effect on these creoles. Nevertheless, it is not clear whether the higher frequency of inflection in pidgins than in creoles is due to different processes of genesis, or a historical accident.

4.4

Morphology II: Reduplication

The morphological process of reduplication is almost universal in creoles but virtually absent in pidgins. Here, reduplication is taken to be a productive process with regard to at least one word class, in which the partial or complete doubling of a form conveys a meaning different from that of the simplex forms. Thus, via a productive process of reduplication in Sranan Creole, jere means ‘to hear’ and jerejere means ‘to hear at times’, or ‘to hear a bit’. Most pidgins have a number of “doubled” forms, but these do not contrast with simplex forms. Either the meaning change is not predictably patterned, or related unreduplicated forms do not exist. Bakker (2003a) reports on a sample of almost 30 pidgins. Productive reduplication appears to be absent in all but four pidgins; for two of these the evidence is doubtful, leaving only two pidgins with productive reduplication: Ethiopian Simplified Italian, which uses distributive reduplication (Marcos 1976), and the Ussuri dialect of Manchuria Chinese Pidgin Russian, where quantifiers can be converted into distributives

144

Peter Bakker

(odin ‘one’, odin-odin ‘one each’) and interrogatives into indefinite pronouns (cego ‘what’, cego-cego ‘something’) – as in the first language of these pidgin speakers, the Tungusic language Nanai (Johanna Nichols 1986). To this, Bakker (2003b) adds Malaysian Bazaar Malay, where reduplicated nouns express plurals of diversity (bungabunga ‘all kinds of flowers’). In summary, reduplication is rare in pidgins. In contrast, out of almost a hundred creoles surveyed, only a handful show no evidence of some process of productive reduplication (Parkvall 2003). Over the range of creoles, nominal and verbal reduplications display a variety of semantic functions (Bakker & Parkvall 2005).

4.5

Ideophones

Ideophones form a word class that is not common in European languages. They usually function as adverbials, are often imitative or onomatopoetic, and are not fully integrated into the sentence structure. In this creole example the ideophones are italicized: (12) a bi-djomboo viiin te a wata djuubu 3s.nom ant-jump quickly till the water splash ‘He jumped quickly, splash! into the water.’ (Saramaccan, Suriname; Bakker, Smith, & Veenstra 1995, p. 174) Ideophones are common in African languages; we need not be surprised that we find them in African-influenced Atlantic creoles, therefore (see Bartens 2000 for an inventory and analysis). For instance, Faraclas (1988, pp. 190–2) shows that most Nigerian languages, including Nigerian Pidgin (a pidgincreole), have ideophones. According to Parkvall (2000, p. 140) their frequency in Atlantic creoles and their relative absence in pidgins and creoles elsewhere (see Bartens 1996, p. 132) makes it plausible that they are substrate-induced. He further points out that ideophones are more numerous in Atlantic creoles still in contact with African languages, i.e., the West African English creoles and the Guinea-Bissau Portuguese creoles. Ideophones appear to be absent from pidgins and pidgincreoles, including those based on African languages, even if the lexifiers do have a category of ideophones. There are no ideophones in the pidgincreole Fanakalo, at least not in the material we have been able to consult. For two other African pidgins/ pidgincreoles, linguists explicitly discuss the absence of ideophones. Thus, there are no ideophones in Sango (Pidgincreole Ngbandi) and Bilkiire (Pidgin Fula; Noss 1979). Pasch (1997, p. 259) explains their absence in Sango as follows: “The loss of ideophones can surely be interpreted as the result of imperfect language acquisition and, in the example given, as a result of pidginization.” Noss (1979, p. 179, n. 7) attributes the loss of ideophones in both Bilkiire and Sango to pidginization and the limited use of pidginized language. He points out that ideophones are used with greatest creativity and freedom in oral

Pidgins versus Creoles and Pidgincreoles 145 performances, with only a limited number being used in everyday speech. “Since oral performances are not normally a part of pidgin usage, they tend to be lost except for the most common that become formalized as adverbs or in idioms.” In other words, ideophones are relatively frequent in Atlantic creoles but not in creoles elsewhere, strongly suggesting African influence.

4.6

Other differences: Question words, prepositions, word class markers

There are a number of additional structural differences between creoles and pidgins. Although none of these constitute absolute differences, there are remarkable proportional or statistical differences between the two classes of languages. For instance, it seems that pidgins make use of fewer distinct question words than creoles. Pidgin question words tend to be taken directly from one of the source languages, and are frequently monomorphemic, whereas creole question words are more often bimorphemic, the grammaticalized combination of a question element and a noun (Muysken & Smith 1990; Clements & Mahboob 2000). Further, pidgins seem to rely on a smaller set of prepositions than creoles (recall the Icelandic Pidgin example in (3)). No quantitative data are available, however. Finally, there are some pidgins that use formal markers of word classes. For example, Pidgin Basque nouns and adjectives end in -a whereas verbs can end in a range of vowels and consonants (Bakker 1989; Bakker et al. 1991). Clearly these categories are based on the citation form in Basque, which ends in -a. Turku verbs end often in -u and nouns in -a (Tosco & Owens 1993; Bakker 2003c). The verbal suffix -u is probably derived from an inflectional suffix for third person plural or third masculine singular object (Tosco & Owens 1993, p. 241). Admittedly, examples of this kind are few, but it is quite remarkable since formal markers for word classes are cross-linguistically uncommon. In short, there are few absolute distinctions between pidgin and creole structures, but there are clear statistical differences. Pidgincreoles, like creoles but unlike pidgins, have SVO word order, preverbal TMA marking, and some morphology not inherited from the lexifier language (see Bakker 2003c).

5 Nativization Nativization, the process whereby a language becomes a mother tongue, is often cited as the social factor that sets pidgins apart from creoles. Here we will consider the question whether the structural differences between pidgins and creoles can also be explained by nativization. This appears difficult to prove historically, since there is very little documentation that allows us to distinguish clearly between pidgin and creole stages. Almost all documented examples of nativization of reduced contact languages pertain to the nativization

146

Peter Bakker

of pidgincreoles. Considering those studies, it appears that nativization has only minimal structural impact. Studies of the structural effects of the evolution of a pidgin to a pidgincreole suggest that social expansion rather than nativization leads to structural changes. Nativization is sometimes called “creolization,” but the term creolization is used both for a social process (a language becoming a mother tongue) and a structural process (changes taking place in languages that make the language look like other languages identified as creoles). In order to avoid the ambiguity of the term “creolization” and the circularity of a structural definition, I will use the term “nativization” when discussing the social process. In this section, I will first deal with the contrast between pidgins and creoles, and then with other situations in which we find a contrast between native and non-native languages.

5.1

Nativization of pidgins

One of the few empirical studies of the role of nativization is of Nigerian Pidgin English (NigPE) in an urbanizing area. The researchers only published a preliminary report (Shnukal & Marchese 1983), in which the conclusion was nevertheless clear: adults rather than first-language speakers are responsible for most of the structural changes, and young and first-language speakers take these changes further. Certain phenomena that appear indicative of nativization in NigPE include an “increase in tempo and fluency among younger speakers, as well as a certain amount of phonological reduction” and a “change in tone and intonation between older and younger speakers of NigPE” (1983, p. 22). Shnukal and Marchese noted more significant effects where the pidgin had become the community lingua franca of urban adults, than where it had become the first language of speakers who grew up in the city. According to them these results “raised the question of the validity of first-language speakers as a criterion for creolization” (p. 22). In the Pacific, a similar study was conducted about which more results were published. Christine Jourdan recorded rural and urban speakers of Solomon Islands Pidgincreole English. Some spoke it as their second language, others as their mother tongue. Her conclusions, as reported in Jourdan (1991) and Jourdan and Keesing (1997) were that the development from jargon to pidgin to pidgincreole to creole is not necessarily linear: creoles are not necessarily preceded by a pidgin, creolization can take place at any point of the “life cycle” and pidgincreoles do not necessarily nativize. Furthermore, creolization does not only occur on plantations; it sometimes takes place in urban settings, and children play only a limited role in structural change. Adults have a creative impact on the language, in expanding the already rich syntactic resources and lexicon; whereas the children have a regularizing impact, particularly as they streamline and condense phonology and generalize grammatical patterns. ( Jourdan 1991, p. 195).

Pidgins versus Creoles and Pidgincreoles 147 Sankoff and Laberge (1974) reached similar conclusions with regard to the effects of nativization on the structural characteristics of Tok Pisin Pidgincreole English of New Guinea. These large scale surveys from urbanizing communities suggest that it is not so much nativization that is responsible for most structural changes in the pidgin but more the fact that adults had to speak pidgin most of the time in the multilingual city. Therefore Jourdan suggests it is not nativization but “the linguistic praxis of [pidgin] speakers” that distinguishes pidgins from creoles (Jourdan 1991, p. 195). In other words, the process of urbanization can have structural effects on pidgins similar to what we find in nativized creoles. On the other hand, at least two cases have been reported where mother tongue speakers and second-language speakers of a pidginized language did use a structurally different kind of language. The first is that of Chinook Wawa Pidgin, a pidgin now only remembered by a few elderly speakers along the Northwest American coast, based mostly on Amerindian languages. It is possible to distinguish between data from Grand Ronde (where nativization and community status establish it as a creole) and elsewhere (where it is neither nativized nor a community language, hence a pidgin), and there appear to be considerable differences between the pidgin and the creolized variety. Chinook Wawa was used both by Europeans and Amerindians in interethnic contacts. It has been documented as a pidgin in locations spread over a huge area, and covering a significant time period (see Grant 1996a). It was also one of the only common languages in interethnic marriages in the Grand Ronde community in Oregon, USA, and formerly the principal medium of communication on this multilingual reservation, where attempts have now been undertaken to revitalize it, with some success. Texts in Chinook Wawa from Grand Ronde were collected in the 1930s by Melville Jacobs (e.g., Jacobs 1932) and in the 1970s by Henry Zenk (e.g., Hymes & Zenk 1987). These Grand Ronde texts display a different kind of language from the texts recorded from non-native speakers elsewhere. Boas (1933) points out that Jacobs’s (nativized) data were very different from Boas’s data collected over the years from no fewer than nine different native Amerindian language groups. He described Jacobs’s data as “certainly not the jargon [i.e., pidgin Chinook Wawa] that has been used for years along the coast” and pointed to its exceptionality by calling his own data “the usual form of the jargon” (Boas 1933, pp. 208–9). Zenk (1988) and Grant (1996b) noted the following properties of Grand Ronde Chinook Wawa that are not found in the pidgin variety: person agreement on the verb, a continuative-frequentative prefix, a negative prefix, aspectual reduplication of the verb, and a modal preverbal particle. These differences between Grand Ronde creole varieties and other, non-native, varieties can be the result of nativization, but there seem to be other factors involved, most significantly the fact that Chinook Wawa had become the main language of the community in Grand Ronde (cf. Zenk 1988). Interestingly, except for a preverbal modal particle, none of the structural properties noted as peculiar to the Grand Ronde variety of Chinook Wawa

148

Peter Bakker

exist in the colonial era creoles of the Atlantic and elsewhere. In fact, Grand Ronde Chinook Wawa’s grammar is quite atypical for a creole. An explanation for its divergent nature could be that the structure of the other languages spoken in Grand Ronde stimulated the development of more complex verbal morphology. The second case where differences have been documented between native and non-native varieties is that of Hawai[i. Here the nature of the evidence is different – and controversial. Bickerton (1981) points to significant differences in the English-lexified speech used by people born on the island versus those who had come to Hawai[i as young adults, both recorded in the 1970s. The first group of recordings would represent creole speech, the second group “pidgin.” Bickerton attributes the structural differences to the nativization process (e.g., 1981, 1984). He argues that the speech of the second-language speakers is variable, unsystematic, influenced by the mother tongue, and mixed with other languages such as Hawaiian and Japanese (cf. Bickerton 1999), whereas the speech of the locally born is systematic, consistent, and homogeneous. Postulating the Language Bioprogram Hypothesis (see Veenstra, this volume), Bickerton attributes the presumed differences between the pidgin and creole varieties to Universal Grammar, thus connecting the creole properties with the nativization process. Bickerton’s hypothesis led to a fierce debate in creole studies, and it has been criticized from many different angles. One of the points of criticism is the following. It appears that some of the properties he attributes to nativization are found also in some contact languages that have not become mother tongues. For instance, the pidgincreole Tok Pisin, excluded from consideration by Bickerton because its genesis did not fulfil the social conditions which would have allowed the bioprogram to surface, nevertheless shares a number of his bioprogram features, and did so before the language nativized. This is even more clearly so for Ghanaian Pidgincreole English, which, as the default language of certain social groups in Ghana, counts in our classification as a pidgincreole (criterion (b) in section 3). Huber (1999) compares the structuralsemantic properties of the bioprogram with those of this clearly non-native pidgincreole and finds striking matches: the non-native pidgincreole follows most bioprogram features closely, even though these are supposed to be introduced into the pidgin by nativization. Hence, processes other than nativization have led to quite similar structural properties. More devastating evidence was presented in recent years by Sarah J. Roberts (2000, 2004). She collected texts from a wide range of sources from Hawai[i dating from first contact with English speakers in the late eighteenth century to the twentieth century: court records, ethnic newspapers, travel accounts, diaries, school papers, questionnaires, etc., from a wide range of ethnic groups. It appears that some of the creole’s properties are indeed connected with the mother tongue speakers in the early decades of the twentieth century (Roberts 1998). However, it is mostly in the speech of the first generation of urban speakers, bilingual in different ancestral languages and a form of pidgin

Pidgins versus Creoles and Pidgincreoles 149 English, that the creole-like features surfaced. The creolization process did not take place on the plantations, as surmised by Bickerton, but in urban settings. Summarizing, the evidence regarding nativization as the decisive factor in the structural transition from pidgins and pidgincreoles to creoles is slight and rather contradictory. In some cases, no significant differences were observed between pidgincreoles which were spoken as a second language in an urban context and the same pidgincreoles which had nativized (Nigeria, Solomon Islands, Hawai[i). In those cases, the differences between urban speech and rural speech appeared much more significant. Urbanization had turned a rural pidgin into the main vehicle of communication in the cities, and this appeared to have had a far more important impact in the form of structural expansion than did the impact of nativization. This leaves only the contrast between pidgin Chinook Wawa and nativized Chinook Wawa – and that language is structurally very much unlike prototypical creoles. Hence, Bickerton’s scenario of sudden nativization with far-reaching linguistic consequences must be repudiated.

5.2 Nativization of sign languages and twins’ autonomous languages Given the debate over the role of nativization as a motor for changes in structure, it may be useful to briefly survey a number of other cases of nativization outside the pidgin-creole context. The first to be considered is sign languages of the deaf. Close to 90 percent of deaf people have hearing parents. This means that often deaf children have to develop a sign language at least in part by themselves (Goldin-Meadow 2003). In comparisons of the structure of sign languages and creoles, linguists have concluded that there are quite a few grammatical similarities between the two types of language (e.g., Fischer 1978; Edwards & Ladd 1983; Gee & Goodheart 1985). These similarities have been linked to the nativization process. The best documented case of a language that was developed recently by children who did not have a language in common is the genesis of Nicaraguan Sign Language. This has been described in several publications, such as Kegl and McWhorter (1997) and Kegl, Senghas, and Coppola (1999). After deaf children of varying ages were brought together in Nicaraguan schools for the deaf, they developed a sign language over a very short time. Two factors appear to have played a role in the attainment of fluency in the sign language: age of entry into the schools and year of entry. The younger the speakers were, the more fluent they became. The longer the schools had existed when the children entered them, the more fluency they would acquire. Only children older than 10 did not show an effect of year of entry. Overall, the establishment and growth of a speech community appears to have been at least as important as the nativization factor. It has sometimes been observed that twins or other close siblings speak a language among themselves, a so-called “autonomous language,” which

150

Peter Bakker

functions well between them, but which is unintelligible to the outside world. In a comparative study based on a literature survey of these languages, Bakker (1987) concludes that the lexicon of autonomous languages is largely based on phonologically distorted words from the family language, and that the grammatical system is divergent both from the family language and from creoles, even though most children studied were relatively advanced in age (i.e., were old enough to show full maturation of grammar). These languages should be considered native languages, being (often) the only languages spoken by these children, but they appear to be hugely different from creoles in their structure. These languages are also typologically divergent from natural languages, and seem “deficient” in several respects: they remain grammatically incomplete, perhaps comparable with jargons or pidgins, despite the fact that they are first languages. For example the Danish sentence in (13a) is reduced in a Danishbased twins’ language to (13b), where bap < Mandse, ep < hest, and dop is probably onomatopoetic: (13) a.

Mandse har slået hesten itu (Danish) Mandse has beaten the-horse in-two b. bap ep dop. (Danish Autonomous Language) ‘Mandse has damaged the horse.’

One can note here, for instance, severe phonological simplification, deletion of the auxiliary har and the cliticized article -en in hesten, and SOV rather than Danish SVO word order. The following example from an Estonian autonomous language displays only fossilized grammatical morphemes, in the form of the adessive marker -l that has become part of the locative postposition: (14) a.

poiss istu-b tooli pea-l (Estonian) boy sit-3s chair top-adessive b. poiss istu tool bäl (Estonian Autonomous Language, Saareste 1936) boy sit chair on ‘The boy is sitting on the chair.’

This brief survey suggests that the presence of a speech community is much more important than nativization for the structural differences between pidgins and creoles. Similarly, in the Hawaiian English case and the Grand Ronde Chinook Wawa Pidgin case, the development of speech communities may have been more important than the nativization factor.

6

Conclusions

The four categories of restructured contact languages discussed here are distinguished on social grounds. To some extent, the social categories appear

Pidgins versus Creoles and Pidgincreoles 151 to correlate with structural features. Jargons are unsystematic, normless, and sometimes lexically diverse, with a lack of grammatical processes and functional categories. Pidgins do display a rudimentary grammatical system, evidenced in their word order properties and sometimes in the presence of inflectional morphology. Pidgincreoles and creoles are even more regularized, generally displaying SVO word order, and with TMA distinctions expressed predominantly by preverbal markers. Pidgins display a wider typological diversity than creoles whereas, structurally, the category of pidgincreoles resembles that of creoles. Therefore, pidgincreoles cannot be taken as representative of the category of pidgins, despite the fact that they may share their non-native status. Social triggers are needed, such as a speech community, to transform a non-native communication system into a full-fledged human language. Nativization plays only a minor role at most.

FURTHER READING There are few comparative studies of pidgins; Bakker (2003c) and Roberts and Bresnan (in press) are comparative papers on pidgin (and creole) morphology. Bakker (1995) provides a general overview, Heine (1973) is a comprehensive study of Bantu-based pidgins, and Bresnan (2001) focuses on pidgin pronouns. Thomason and Kaufman (1988) is a highly readable study of contact-induced change. Thorough studies of individual pidgins include Schuchardt (1909), Cifoletti (1989), and Do Couto (2001) on the Mediterranean Lingua Franca; Broch and Jahr (1981, 1984) on Russenorsk; Kaltenbrunner (1996) on Fanakalo; Drechsel (1997) on Mobilian; Roberts (1995a, b) on Pidgin Hawaiian; Tosco and Owens (1993) on Turku Pidgin Arabic; Long (2007) on Bonin Island English. Collections on pidgins include Bakker et al. (1991), devoted to Basque pidgins, and Jahr and Broch (1996) devoted to the Arctic. Monographs on pidgincreoles include Sreedhar (1974) and Boruah (1993) on Naga Pidgin; Mühlhäusler (1974, 1979) on Melanesian Pidgin/Tok Pisin; Crowley (1990) on Bislama; Tryon and Charpentier (2004) on Pacific Englishes; Faraclas (1996) on Nigerian Pidgin English; Huber (1999) on Ghanaian Pidgin English. Textbooks with information about pidginization include Romaine (1988), Mühlhäusler (1997), Sebba (1997), and Thomason (2001).

REFERENCES Adelaar, K. Alexander (1991) Some notes on Sri Lanka Malay. In: Hein Steinhauer (ed.) Papers in Austronesian Linguistics 1 (Pacific Linguistics, Series A-81). Canberra: Research School of Pacific Studies, pp. 23–37. Alexandre, Pierre (1963) Aperçu sommaire sur le pidgin A-70 du Cameroun [Brief overview of Pidgin A-70]. Cahiers d’Études Africaines 12, 577–82. Arends, Jacques (1989) Syntactic Developments in Sranan. Unpublished PhD dissertation, University of Nijmegen.

152

Peter Bakker

Arends, Jacques and Adrienne Bruyn (1995) Gradualist and developmental hypotheses. In: Arends, Muysken, & Smith (eds.), pp. 111–20. Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) (1995) Pidgins and Creoles: An Introduction (Creole Language Library 15). Amsterdam: John Benjamins. Baker, Philip (1993) Australian influence on Melanesian Pidgin English. Te Reo 36, 3–67. Baker, Philip and Peter Mühlhäusler (1990) From business to pidgin. Journal of Asian Pacific Communication 1, 87–115. Bakker, Peter (1987) Autonomous Languages: Signed and Spoken Languages Created by Children in the Light of Bickerton’s Bioprogram Hypothesis (Publications from the Institute for General Linguistics 53). University of Amsterdam, Institute for General Linguistics, Amsterdam. Bakker, Peter (1989) A French-Icelandic nautical pidgin. Journal of Pidgin and Creole Languages 4, 129–32. Bakker, Peter (1995) Pidgins. In: Arends, Muysken, & Smith (eds.), pp. 25–39. Bakker, Peter (2000) Rapid language change: Creolization, intertwining, convergence. In: Colin Renfrew, April McMahon, and Larry Trask (eds.) Time Depth in Historical Linguistics, vol. 2. Cambridge: McDonald Institute for Archaeological Research, pp. 585–620. Bakker, Peter (2003a) The absence of reduplication in pidgins. In: Silvia Kouwenberg (ed.) Twice as Meaningful: Reduplication in Pidgins, Creoles and Other Contact Languages (Westminster Creolistics Series 8). London: Battlebridge, pp. 37–64. Bakker, Peter (2003b) Bazaar Malay. In: Seventh International Symposium on Malay/ Indonesian Linguistics (ISMIL 7) book of papers. 19 pp. Bakker, Peter (2003c) Pidgin inflectional morphology and its implications for creole morphology. In: Ingo Plag (ed.) Yearbook of Morphology: Special Section on Pidgins and Creoles. Dordrecht: Kluwer, pp. 3–33. Bakker, Peter (2006) The Sri Lanka Sprachbund: The newcomers Portuguese and Malay. In: Yaron Matras, April McMahon, and Nigel Vincent (eds.) Linguistic Areas: Convergence in Historical and Typological Perspective. Houndmills: Palgrave MacMillan, 135–59. Bakker, Peter and Mikael Parkvall (2005) Reduplication in pidgins and creoles. In: Bernhard Hurch in collaboration with Veronika Mattes (eds.) Studies on Reduplication (Empirical Approaches to Language Typology 28). Berlin: Mouton De Gruyter, pp. 511–31. Bakker, Peter, Gidor Bilbao, Nicolaas G. H. Deen, and José Ignacio Hualde (1991) Basque Pidgins in Iceland and Canada. Donostia/San Sebastian: Gipuzkoako Foru Aldundia/Diputación Foral de Gipuzkoa. Bakker, Peter, Marike Post, and Hein van der Voort (1995) TMA particles and auxiliaries. In: Arends, Muysken, & Smith (eds.), pp. 247–58. Bakker, Peter, Norval Smith, and Tonjes Veenstra (1995) Saramaccan. In: Arends, Muysken, & Smith (eds.), pp. 165–78. Bartens, Angela (1996) Der Kreolische Raum: Geschichte und Gegenwart [The space of creole studies: History and present]. Helsinki: Suomalainen Tiedeakatemia. Bartens, Angela (2000) Ideophones and Sound Symbolism in Atlantic Creoles. Helsinki: The Finnish Academy of Science and Letters. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984) The language bioprogram hypothesis. Behavioral and Brain Sciences Studies 7, 173–88.

Pidgins versus Creoles and Pidgincreoles 153 Bickerton, Derek (1995) Language and Human Behavior. Seattle: University of Washington Press. Bickerton, Derek (1999) Pidgins and language mixture. In: John R. Rickford and Suzanne Romaine (eds.) Creole Genesis, Attitudes, and Discourse: Studies Celebrating Charlene J. Sato (Creole Language Library 20). Amsterdam: John Benjamins, pp. 31–43. Boas, Franz (1933) Note on the Chinook Jargon. Language 9, 208–13. Bold, J. D. (1958) Dictionary, Grammar and Phrase-book of Fanagalo (Kitchen Kafir), 5th revised edn. South Africa: Central News Agency Ltd. Boruah, B. K. (1993) Nagamese: The Language of Nagaland. New Delhi: Mittal Publications. Bresnan, Joan (2001) Pidgin genesis and optimality theory. In: Jeff Siegel (ed.) Processes of Language Contact: Case Studies from Australia and the Pacific. Montreal: Les Éditions Fides, pp. 145–73. Broch, Ingvild and Ernst Håkon Jahr (1981) Russenorsk: et Pidginspråk i Norge [Russnorsk: A pidgin language in Norway] (Tromsø Studier i Språkvitenskap 3). Oslo: Novus. (2nd, revised edition 1984.) Broch, Ingvild and Ernst Håkon Jahr (1984) Russenorsk: A new look at the RussoNorwegian Pidgin in northern Norway. In: P. Sture Ureland and Iain Clarkson (eds.) Scandinavian Language Contacts. Cambridge: Cambridge University Press, pp. 21–65. Cifoletti, Guido (1989) La lingua franca mediterranea [The Lingua Franca of the Mediterranean]. Padova: Unipress. Clements, J. Clancy (1996) The Genesis of a Language: The Formation and Development of Korlai Portuguese (Creole Language Library 16). Amsterdam: John Benjamins. Clements, J. Clancy and Ahmar Mahboob (2000) WH-words and question formation in pidgin/creole languages. In: John McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles (Creole Language Library 21). Amsterdam: John Benjamins, pp. 459–97. Clyne, Michael (1968) Zum Pidgin-Deusch der Gastarbeiter [About guest workers’ Pidgin German]. Zeitschrift für Mundatrforschung 35, 130–9. Crowley, Terry (1990) Beach-La-Mar to Bislama: The Emergence of a National Language in Vanuatu. Oxford: Clarendon Press. De Groot, A. (1977) Woordregister Nederlands-Saramakkaans [Vocabulary DutchSaramaccan]. Paramaribo: VACO. Do Couto, Hildo Honório (2001) A Língua Franca Mediterrânea: Histórico, Textos e Interpretação. Brasil: Editora Plano. Drechsel, Emanuel (1997) Mobilian Jargon: Linguistic and Sociohistorical aspects of a Native American Pidgin. Oxford: Oxford University Press. Edwards, Vivian and Paddy Ladd (1983) British Sign Language and West Indian Creole. In: Jim Kyle and Bencie Woll (eds.) Language in Sign: An International Perspective on Sign Language. London: Croom Helm, pp. 147–58. (Also published 1982 in Sign Language Studies 35, 101–26.) Faraclas, Nicholas G. (1988) Nigerian Pidgin and the languages of Southern Nigeria. Journal of Pidgin and Creole Languages 3, 177–97. Faraclas, Nicholas G. (1996) Nigerian Pidgin English. London: Routledge. Fischer, Susan D. (1978) Sign language and creoles. In: Patricia Siple (ed.) Understanding Language through Sign Language Research. London: Academic Press, pp. 309–31. Foley, William A. (1988) Language birth: The processes of pidginization and creolization. In: Frederick Newmeyer (ed.) Linguistics: The Cambridge Survey, vol. 2. Cambridge: Cambridge University Press, pp. 162–83.

154

Peter Bakker

Foley, William A. (2006) Universal constraints and local conditions in Pidginization: Case studies from New Guinea. Journal of Pidgin and Creole Languages 21, 1–44. Gee, James Paul and Wendy Goodhart (1985) Nativization, linguistic theory, and deaf language acquisition. Sign Language Studies 49, 291–342. Givón, Talmy (1982) Tense-aspect-modality: The creole prototype and beyond. In: Paul J. Hopper (ed.) Tense-Aspect: Between Semantics and Pragmatics. Amsterdam: John Benjamins, pp. 115–63. Goldin-Meadow, Susan (2003) The Resilience of Language. New York: Psychology Press. Grant, Anthony P. (1996a) Chinook Jargon and its distribution in the Pacific Northwest and beyond. In: Stephen Wurm, Peter Mühlhäusler, and Darrell Tryon (eds.) Atlas of Languages of Intercultural Communication in the Pacific, Asia, and the Americas. Berlin: Mouton De Gruyter, pp. 1184–208. Grant, Anthony P. (1996b) The evolution of functional categories in Grand Ronde Chinook Jargon: Ethnolinguistic and grammatical considerations. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions . . . Papers Relating to Grammaticalization in Contact Languages (Westminster Creolistics Series 2). London: University of Westminster, pp. 225–42. Grant, Anthony P. (2001) El Chabacano Zamboangueño: Una lengua mezclada [Chabacano of Zamboango: A mixed language]. Papia 12, 7–40. Hall, Robert A., Jr. (1944) Chinese Pidgin English Grammar and Texts. Journal of the American Oriental Society 64, 95–113. Hall, Robert A., Jr. (1953) Haitian Creole: Grammar . . . Texts – Vocabulary (The American Anthropologist 55, no. 2 part 2, Memoir no. 74). Philadelphia: American Folklore Society/American Anthropologist. Hall, Robert A., Jr. (1962) The life cycle of pidgin languages. Lingua 11, 151–6. Hancock, Ian F. (1980) Lexical expansion in creole languages. In: Albert Valdman and Arnold Highfield (eds.) Theoretical Orientation in Creole Studies. New York: Academic Press, pp. 63–88. Heidelberger Forschungsprojekt “Pidgin-Deutsch” (1977) Heidelberger Forschungsprojekt “Pidgin-Deutsch spanischer und italienischer Arbeiter in der Bundesrepublik”: Die ungesteuerte Erlernung des Deutschen durch spanische und italienische Arbeiter. Eine soziologische Untersuchung [The Pidgin Research Project of Heidelberg “Pidgin German of Spanish and Italian workers in the Federal Republic of Germany”: The unregulated learning of German by Spanish and Italian workers. A sociological study] (Osnabrücker Beiträge zur Sprachtheorie, Beiheft 2). Heine, Bernd (1973) Pidgin-Sprachen im Bantu-Bereich [Pidgin languages in the Bantu realm]. Berlin: Dietrich Reimer Verlag. Heine, Bernd (1982) The Nubi Language of Kibera, an Arabic Creole (Language and Dialect Atlas of Kenya 3). Berlin: Dietrich Reimer Verlag. Holm, John (1988–89) Pidgins and Creoles, 2 vols. Cambridge: Cambridge University Press. Hosokawa, Komei (1987) Malay talk on boat: An account of Broome pearling lugger pidgin. In: Donald C. Laycock and Werner Winter (eds.) A World of Language: Papers Presented to Professor S. A. Wurm on his 65th Birthday (Pacific Linguistics A-100). Canberra: Australian National University, pp. 287–96. Huber, Magnus (1999) Ghanaian Pidgin English in its West African Context. Amsterdam: John Benjamins. Hussainmiya, B. A. (1987) “Melayu Bahasa”: Some preliminary observations on the Malay creole of Sri Lanka. In: Lost Cousins: The Malays of Sri Lanka. Kebangsaan, Malaysia: Universiti Kebangsaan, pp. 153–72.

Pidgins versus Creoles and Pidgincreoles 155 Hymes, Dell and Henry Zenk (1987) Narrative structure in Chinook Jargon. In: Glenn Gilbert (ed.) Pidgin and Creole Languages: Essays in Memory of John E. Reinecke. Honolulu: University of Hawai[i Press, pp. 445–65. Jacobs, Melville (1932) Notes on the structure of Chinook Jargon. Language 8, 27–50. Jahr, Ernst Håkon and Ingvild Broch (1996) Language Contact in the Arctic: Northern Pidgins and Contact Languages. Berlin: Mouton de Gruyter. Jourdan, Christine (1991) Pidgins and creoles: The blurring of categories. Annual Review of Anthropology 20, 187–209. Jourdan, Christine and Roger M. Keesing (1997) From Fisin to Pidgin: Creolization in process in the Solomon Islands. Language in Society 26, 401–20. Kaltenbrunner, Stefan (1996) Fanakalo: Dokumentation einer Pidginsprache [Fanakalo: Documentation of a Pidgin language] (Beiträge zur Afrikanistik 53). Vienna: Veröffentlichungen der Institute für Afrikanistik und Ägyptologie der Universität Wien. Keesing, Roger M. (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford: Stanford University Press. Kegl, Judy and John McWhorter (1997) Perspectives on an emerging language. In: Eve Clark (ed.), Proceedings of the Twenth-Eighth Annual Child Language Research Forum. Stanford: CSLI, pp. 15–38. Kegl, Judy, Ann Senghas, and Marie Coppola (1999) Creation through contact: Sign language emergence and sign language change in Nicaragua. In: Michel DeGraff (ed.) Language Creation and Language Change. Cambridge, MA: MIT Press, pp. 179– 237. Kouwenberg, Silvia (1992) From OV to VO: Linguistic negotiation in the development of Berbice Dutch creole. Lingua 88, 263–99. Kouwenberg, Silvia and Pieter Muysken (1995) Papiamento. In: Arends, Muysken, and Smith (eds.), pp. 205–18. Long, Daniel (2007) English on the Bonin (Ogasawara) Islands. Durham, NC: Duke University Press for the American Dialect Society. Marcos, Habte-Mariam (1976) Italian. In: M. Lionel Bender, J. Donald Bowen, Robert L. Cooper, and Charles A. Ferguson (eds.) Language in Ethiopia. London: Oxford University Press, pp. 170–80. Mehrotra, Raja Ram (1997) Reduplication in Indian Pidgin English. English Today 13, 45–9. Mufwene, Salikoko S. (1997) Jargons, pidgins, creoles, and koines: What are they? In: Arnold K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles (Creole Language Library 19). Amsterdam: John Benjamins, pp. 35–70. Mühlhäusler, Peter (1974) Pidginization and Simplification of Language (Pacific Linguistics B-26). Canberra: Australian National University. Mühlhäusler, Peter (1979) Growth and Structure of the Lexicon of New Guinea Pidgin. Canberra: Department of Linguistics, Research School of Pacific Studies. Mühlhäusler, Peter (1997) Pidgin and Creole Linguistics, revised edn. (Westminster Creolistics Series 3). London: University of Westminster Press. Mühlhäusler, Peter (2001) Typology and universals of pidginization. In: Martin Haspelmath, Ekkehard König, Wulf Oestereicher, and Wolfgang Raible (eds.) Language Typology and Linguistic Universals: An International Handbook. Berlin: De Gruyter, pp. 1648–55. Muysken, Pieter and Norval Smith (1990) Question words in pidgin and creole languages. Linguistics 28, 883–903.

156

Peter Bakker

Nichols, Johanna (1986) The bottom line: Chinese Pidgin Russian. In: Wallace Chafe and Johanna Nichols (eds.) Evidentiality: The Linguistic Coding of Epistemology. Norwood, NJ: Ablex, pp. 238–57. Nichols, John D. (1995) The Ojibwe verb in “Broken Oghibbeway.” Amsterdam Creole Studies 12, 1–18. Noss, Philip A. (1979) Fula: A language of change. In: Ian F. Hancock (ed.) Readings in Creole Studies. Ghent: Story Scientia, pp. 173–88. Parkvall, Mikael (2000) Out of Africa: African Influences in Atlantic Creoles. London: Battlebridge. Parkvall, Mikael (2003) Reduplication in the Atlantic Creoles and other contact languages. In: Silvia Kouwenberg (ed.) Twice as Meaningful: Reduplication in Pidgins, Creoles and Other Contact Languages (Westminster Creolistics Series 8). London: Battlebridge Publications, pp. 19–36. Pasch, Helma (1997) Sango. In: Sarah G. Thomason (ed.) Contact Languages: A Wider Perspective (Creole Language Library 17). Amsterdam: John Benjamins, pp. 209–70. Roberts, Julian (1995a) Pidgin Hawaiian: A sociohistorical study. Journal of Pidgin and Creole Languages 10, 1–56. Roberts, Julian (1995b) A structural sketch of Pidgin Hawaiian. Amsterdam Creole Studies 12, 97–126. Roberts, Sarah J. (1998) The role of diffusion in the genesis of Hawaiian Creole. Language 74, 1–39. Roberts, Sarah J. (2000) Nativization and the genesis of Hawai[i Creole. In: John McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles (Creole Language Library 21). Amsterdam: John Benjamins, pp. 257–300. Roberts, Sarah J. (2004) The Emergence of Hawai’i Creole English in the Early 20th Century: The Sociohistorical Context of Creole Genesis. PhD dissertation, Stanford University. Roberts, Sarah J. and Joan Bresnan (in press) Retained inflectional morphology in pidgins: A typological study. Linguistic Typology. Romaine, Suzanne (1988) Pidgin and Creole Languages. London: Longman. Saareste, A. (1936) Language enfantin conservé par trois jeunes estoniens de onze, neuf et huit ans [Child language preserved by three youngsters aged eleven, nine and eight]. Tartu: Publications des archives de la langue estonienne. Sandefur, John R. (1979) An Australian Creole in the Northern Territory: A Description of Ngukurr-Bamyili Dialects. Darwin: Summer Institute of Linguistics, Australian Aborigines Branch. Sankoff, Gillian and Suzanne Laberge (1974) On the acquisition of native speakers by a language. In: David DeCamp and Ian Hancock (eds.) Pidgins and Creoles: Current Trends and Prospects. Washington, DC: Georgetown University Press, pp. 73–84. Schuchardt, Hugo (1909) Die Lingua Franca. Zeitschrift für romanische Philologie 33, 441– 61. (Translated as (1987) The Lingua Franca. In: Glenn Gilbert (ed.) Pidgin and Creole Languages: Selected Essays by Hugo Schuchardt. Cambridge: Cambridge University Press, pp. 65–88. Also published (1979) as The Ethnography of Variation: Selected Writings on Pidgins and Creoles. Translated by Thomas L. Markey. Ann Arbor: Karoma.) Sebba, Mark (1997) Contact Languages: Pidgins and Creoles. New York: St. Martin’s Press. Seuren, Pieter (1998) Western Linguistics: An Historical Introduction. Oxford: Blackwell.

Pidgins versus Creoles and Pidgincreoles 157 Shnukal, Anna (1988) Broken: An Introduction to the Creole Language of Torres Straits (Pacific Linguistics C-107). Canberra: Research School of Pacific Linguistics, Australian National University. Shnukal, Anna and Lynell Marchese (1983) Creolization of Nigerian Pidgin English: A progress report. English World-Wide 4, 17–26. Siegel, Jeff (1992) The transformation and spread of Pidgin Fijian. Language Sciences 14, 287–308. Singler, John Victor (ed.) (1990) Pidgin and Creole Tense-Mood-Aspect Systems (Creole Language Library 6). Amsterdam: John Benjamins. Singler, John Victor (1992) Nativization and pidgin/creole genesis: A reply to Bickerton. 7, 319–33. Singler, John Victor (1997) The configuration of Liberia’s Englishes. World Englishes 16, 205–31. Smith, Ian R. (1979) Convergence in South Asia: A Creole example. Lingua 48, 193–222. Smith, Ian R. (1984) The development of morphosyntax in Sri Lanka Creole Portuguese. In: Mark Sebba and Loreto Todd (eds.) Papers from the Creole Conference, York University. York: York University, Department of Language, pp. 291–301. Sreedhar, M. V. (1974) Naga Pidgin: A Sociolinguistic Study on Interlingual Communication Patterns in Nagaland. Mysore: Central Institute of Indian languages. Stefánsson, Vilhalmjur (1909) The Eskimo trade jargon of Herschel Island. American Anthropologist 11, 217–32. Thomason, Sarah G. (2001) Language Contact: An Introduction. Edinburgh: Edinburgh University Press. Thomason, Sarah G. and Terence Kaufman (1988) Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Tosco, Mauro and Jonathan Owens (1993) Turku: A descriptive and comparative study. Sprache und Geschichte in Afrika 14, 177–267. Tryon, Darrell T. and Jean-Michel Charpentier (2004) Pacific Pidgins and Creoles: Origins, Growth and Development (Trends in Linguistics. Studies and Monographs 132). Berlin: Mouton de Gruyter. Van der Voort, Hein (1997) New light on Eskimo Pidgins. In: Arthur K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles (Creole Language Library 19). Amsterdam: John Benjamins, pp. 373–94. Watson, Richard L. and Louis Biajo Ola (1985) Juba Arabic for Beginners. Nairobi: Summer Institute of Linguistics. Wellens, Inek. (2005) The Nubi language of Uganda: An Arabic Creole in Africa. Leiden: Brill Academic Publishers. Williams, Jeffrey P. (1995) A note on the pronominal system of Arafundi-Enga pidgin. Journal of Pidgin and Creole Languages 10, 171–5. Wurm, Stephen A. (1992) Some contact languages and pidgins and creoles in the Siberian region. Language Sciences 14, 249–85. Zenk, Henry (1988) Chinook Jargon in the speech economy of Grand Ronde Reservation, Oregon: An ethnography of speaking approach to an historical case of creolization in process. International Journal of the Sociology of Language 71, 107–24.

158

7

Kees Versteegh

Non-Indo-European Pidgins and Creoles KEES VERSTEEGH

1

Talking to Strangers

In 1874 the Reverend W. G. Lawes from the London Missionary Society came to Port Moresby in New Guinea, together with his family. The reverend soon learned the local language Motu and started to translate the Bible into it, until his son Frank, who had learned the language while playing with the local boys of the village, drew his attention to the fact that what he had been learning was just a simplified form of the language, which differed markedly from the “true” Motu he himself spoke (Dutton 1985, p. 36). According to contemporary sources Reverend Lawes found it much more difficult to learn this real Motu than the form he had previously learnt. The story of Frank Lawes is somewhat reminiscent of that of Dorothy Chambers, daughter of the Reverend Chambers who around 1900 had settled in Livingston in eastern Texas. Her parents had allowed her to associate closely with the Alabama-Coushatta Indians who lived there and were under the impression that she knew their language well. Yet, when they met a trader who claimed to be able to speak Alabama, she was unable to communicate with him. It turned out that the man knew only the simplified Mobilian Jargon that the Indians used in their contacts with outsiders. It was not until then that Chambers realized that there were actually two forms of Alabama, one of them a reduced register, the other the “true” language of the Alabama (Drechsel 1997, p. 241). Dorothy Chambers and Frank Lawes were privileged in being children rather than linguists and thus being able to learn Motu and Alabama the natural way, as well as being allowed by the native speakers to hear the true language. Both peoples were in the habit of keeping their own language away from outsiders, addressing them instead in a reduced register. Presumably, not only European missionaries were unaware of the fact that in contacts with outsiders these peoples used a simplified language. Apparently, the Motu themselves believed that on their expeditions to trade with other groups they

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Non-Indo-European Pidgins and Creoles 159 had learnt the actual languages of their trade partners, whereas in fact these too were more or less conventional simplified registers of the languages involved (Dutton 1985, pp. 21ff.). Examples of the aversion to using one’s own language with foreigners abound. The Yakoma who lived along the Ubangi river in Central Africa deliberately used a simplified form of their language, Ngbandi, in outside contacts. Even before the arrival of the French and the Belgians to this region, this variety had become a widespread lingua franca along the Ubangi river, the basis for what was later to become Sango (Pasch 1997, p. 215). Some peoples even go so far as to use a different language altogether in their contacts with foreigners. This may have been the case with Lingala, a language whose lexifier is the Bantu language Bobangi, spoken on the northern, mid course of the Congo River. Lingala was used by the local population in northwest Congo (Kinshasa) instead of their own language, Lomóngo-Lonkundó, in contacts with foreigners, so that these were under the impression that it was in fact the local language (Mufwene 1997, p. 189). Likewise, the indigenous population of the North West Cape of Australia used a reduced variety of a neighboring language, Ngarluma, when dealing with foreigners, as we know from the travel account of a shipwrecked crew (Dench 1998, p. 45). The motive for the use of a reduced language is doubtless that many communities do not wish others to know their “real” language and therefore keep the knowledge of this language to themselves. In other cases, such as that of the Inuit (Stefánsson 1909; Van der Voort 1996a), foreigners were not actually kept from speaking the “real” language, but it became a source of ridicule if they tried to do so. The form of the reduced register used with foreigners may vary according to their linguistic background. An interesting example of such a difference in foreigner talk is provided by the Eskimo Jargon of Herschel Island (Stefánsson 1909): in their contacts with the whalers the Inuit probably used the first person singular form of the verb as verb stem, but with the Gwich’in Athabascans they used a morphologically richer variety. In New Guinea, the Motu varied their register according to the language of their interlocutors (Dutton 1985, p. 38). Thus, two varieties of foreigner Motu exist: the central dialect used with peoples speaking an Austronesian language like the Motu themselves; and the dialect known as Hiri Motu (formerly Police Motu), used with foreigners speaking Papuan or European languages. The central variety contains a number of traits absent from Hiri Motu but similar to those found in the related Austronesian languages, for instance the expression of the pronominal object by a suffix on the verb; in Hiri Motu this is expressed with an independent pronoun. Similar differences are reported about Swahili native speakers in Kenya who have various styles at their disposal according to the level of Swahili spoken by their interlocutors (Wald 1981). The common factor in these situations is that the visiting foreigners, often traders, are expected to use the language of their hosts and readily accept this general rule. Obviously, when you’re trading, it’s good practice to use the language of those from whom you hope to obtain something. The hosts do not

160

Kees Versteegh

use their “real” language but a reduced version of it that is easier to understand. This rule seems to apply everywhere in the world where trade takes place. Thus, the Chinese traders who came to the Russian town of Kyakhta north of Lake Baikal (Neumann 1966; Belikov 1996, p. 1027) used a trade language whose lexicon was based almost completely on Russian. Likewise, on their trading trips the Motu adopted the foreigner versions of the nonAustronesian languages of their hosts, the Toaripi and Korili, but when they themselves were visited by neighboring groups, they used a reduced version of their own language. In those cases where trading contacts take place on more or less neutral ground and where there exists a balance of power, each group can choose those lexical items that are most familiar, so that a lexical mixture is used, as in the Ndyuka-Trio trade language in east-central Suriname (Huttar and Velantie 1997, p. 114) or in the North American jargons such as Mobilian. Such jargons did not have the low prestige that is usually attributed to simplified languages (Drechsel 1997, p. 35): they constituted an acknowledged way of communicating between heterolingual communities. Therefore, they could also function as a buffer between these communities, preserving the “true” languages and keeping them relatively free from outside interference. Apparently, then, in interethnic contact the choice of language is less dependent on relations of prestige than is commonly thought. According to Kropp Dakubu (1997, pp. 166–7), in West Africa language choice in interethnic contacts is not governed by rules of relative prestige, but by a patron–client or host–guest relationship, in which the clients or guests are supposed to use the language of their patron or host. What does seem to be important is a societal setting in which region-wide or long-distance trade is facilitated by the presence of a state or chiefdom (Croft 2003: 22). When European traders established contact with indigenous populations, they acted as guests or visitors and started to learn the language of their trading partners rather than the other way round. The Portuguese, for instance, at first trained young captives as interpreters because they felt obliged to use the indigenous languages as a means of communication rather than imposing a variety of Portuguese (Naro 1978). In Mexico, the Spanish adopted Nahuatl as the principal means of communication (Lipski 1994, pp. 79–80). Likewise, in Latin America in their contacts with the native population missionaries and traders adopted those forms of the Tupi-Guaraní languages that were already in use in the region as lingua francas (Rodrigues 2002). In Central Africa, the local lingua franca Sango, which was used for outside relations by ethnic groups in the Ubangi region, was taken over by the French and eventually became the general means of communication in the entire region. In China, the first European traders attempted to communicate in Chinese (Bolton 2003). Something similar occurred in Southeast Asia where Malay was in use as a trade language and continued to be used thus even after the advent of the age of colonialism. Obviously, there are also instances where a foreign language is imposed by force, as in large-scale conquests. When the Romans conquered Europe, or the Arabs North

Non-Indo-European Pidgins and Creoles 161 Africa and the Middle East, or the Aryans South Asia, their language or a restructured version of it became the new administrative language and, in time, the vernacular language of the peoples that had been subjugated.

2 The Labels “Pidgin” and “Creole” In spite of the abundance of examples of restructured registers all over the world, the linguistic literature focuses almost exclusively on the restructured registers of colonial languages such as English, Dutch, French, Spanish, and Portuguese. After the initial stages of contact, restructured varieties of the colonial languages often became the accepted means of communication, not only between Europeans and indigenous populations, but also between peoples speaking different indigenous languages. In terms of the model proposed by Kropp Dakubu, the Europeans had changed from visitors into settlers and became the new “patrons/hosts.” This is what happened in Oceania (Tok Pisin), Asia (from Chinese Pidgin English to Tây Bôi), and Africa (Kriol). For the restructured – and eventually nativized – varieties of the colonial languages, the terms “pidgin” and “creole” have become current. The vast majority of studies on pidgins and creoles focus on an even more specific sociolinguistic situation, that of the colonial languages as spoken by the indigenous peoples after their transportation as slaves to the Caribbean region and the Americas. When nativized, these varieties became the creoles par excellence, and it is hardly surprising that the definitions in the linguistic literature reflect precisely this situation. Linguists rarely take into account the kind of linguistic interaction that took place elsewhere, even though they are fully aware of the fact that restructuring as the result of language contact is a common phenomenon all over the world. In this respect the recent shift toward contact linguistics as a more general treatment of such phenomena is an improvement (Thomason and Kaufman 1988; Winford 1997). The familiar Atlantic pidgins and creoles all derive from a few IndoEuropean lexifiers. Cases outside this group do not fit into the general framework and their status has been the topic of mostly futile discussions about the question of whether or not variety X or Y is a pidgin/creole. Inevitably, a host of intermediate notions, such as proto-pidgin, pre-pidgin, pseudo-pidgin, semipidgin, creoloid, secondary hybrid, non-pidginized creole, have been introduced, mainly in order to avoid assigning true pidginhood or creole status to cases outside the restricted corpus of Atlantic creoles. Some researchers, such as Mufwene (2001), even opt to use the term “creole” in a way that is stripped of any linguistic connotations, reserving it for specific instances of restructuring of Indo-European lexifiers in the Caribbean and the Indian Ocean. Others recognize the existence of pidginized varieties elsewhere but maintain that the emergence of reduced registers of non-Indo-European languages could only have taken place under the influence of Europeans, as Samarin asserts of Mobilian Jargon (1986) and of Sango (1982).

162

Kees Versteegh

The main problem with such approaches is their essentialist character. They reify the notions of “pidgin” and “creole” and regard them as something a language can be, rather than concentrating on the process of restructuring and its possible outcome (Jourdan 1991). The apparent typological unity of the most frequently studied pidgins and creoles has contributed to this tendency. Since they all descend from Indo-European lexifiers and emerged under similar circumstances, it is not surprising that the outcome of the restructuring process is similar, too. There is no reason to assume that this particular outcome is the only one possible in contact situations. All instances of restructuring mentioned thus far start with communication between native speakers and second-language (L2) learners. Usually, a distinction is made between those situations in which the target of the learners is a reduced version of the vernacular, and those in which the target is the vernacular itself. The question arises, however, whether this distinction has any basis in reality. There seems to be a consensus that when the target is a reduced variety this represents a break in the transmission of linguistic habits, which leads to a certain amount of restructuring. But as Andersen (1983, p. 13) asserts, it is hard to see why there should be a principled distinction between learning an L2 in general and learning a reduced version. In all cases of L2 acquisition a series of internal grammars has to be created by the learner, which to some degree all differ from the target. The degree of restructuring in each communication process involving L2 learners depends on the input of both sides. At one extreme is the learning process of adult sophisticated learners of a foreign language in a scholastic environment, whose explicit aim is the adoption of vernacular-like features. At the other extreme are the cases of complete restructuring in situations of restricted communication. Since in most contact situations the target offered by the native speakers is a reduced version of the vernacular, most of the resulting varieties exhibit a high degree of restructuring. Only in some cases is there enough exposure to the full version of the vernacular to give the learners an opportunity to upgrade their proficiency. But even then many learners may prefer to stay at their own level. Apparently, Kenyan speakers of L2 Swahili are content to stick to their own variety and they are reinforced in this attitude because speakers of Standard Swahili use a similar variety when speaking to them (Wald 1981, p. 9). The interaction between both sides in the accommodation process may vary widely. Dioula véhiculaire, a restructured variety of ethnic Dyula, is the L2 of many people in the Ivory Coast. Its origin probably lies in the language of children with a non-Dyula mother. Partmann (1979) comments that this variety of Dyula is learned by the children from their mother, but with some degree of monitoring on the part of the father. In this variety there is a fair amount of restructuring in the form of syntactic reduction (loss of distinction between alienable/inalienable possession and between present/future tense, lack of derivation, tendency toward analytic surface representation) and especially of stylistic reduction (Gingiss 1979), but compared to other instances of

Non-Indo-European Pidgins and Creoles 163 restructuring the degree of reduction could be called moderate. Rather than starting a debate about the “pidgin status” of Dioula véhiculaire, as Gingiss does, it seems preferable to concentrate on the surface changes of such varieties to see in what way they correlate with the special circumstances of their acquisition. Theoretically, the approach advocated here implies that the notion of “pidginization” may be used for all cases of L2 acquisition; in practice it is reserved for those cases in which there is a high degree of restructuring. The notion of “creolization” is used for those cases of L1 acquisition in which the input is restricted and monitoring by native speakers is lacking, which results in an output that is structurally different from the input. There is a certain amount of controversy regarding the criterion of nativization – in the sense of “acquiring native speakers” – as a necessary condition in the process of creolization. According to Mufwene (1997, pp. 195, 197) a point of critical mass must be reached before one can speak of “nativization.” But this applies only if one insists on putting a label on a certain linguistic variety. In that case one needs to know at which point there are enough native speakers to regard the language as a creole. The point is, however, that we are dealing with a process, which means that the acquisition of the language by even one individual speaker suffices to set the mechanisms in motion, regardless of whether or not this new way of speaking spreads to a substantial portion of the community, which is a different matter altogether. Put differently, it is very well possible for a speech community to contain both L1 and L2 speakers of a certain variety, at which point the question of whether the language may be called a “pidgin” or a “creole” becomes moot (see Bakker, this volume). The input for a process of creolization can consist of a pidginized variety, but also of other kinds of non-native input, for instance an artificial language like Esperanto (Versteegh 1993a), when the parents are not native speakers themselves so that they are unable to provide the necessary monitoring. This should be distinguished from first language (L1) acquisition in situations of language attrition, in which the parents use a reduced register with their children. Even though the parents’ output in such a situation is generally grammatical, the resulting acquisition process is imperfect because of the lack of exposure to the “full” language. Situations like these have been described for the acquisition of Moroccan Arabic by emigrant children in the Netherlands (El Aissati 1996), for the acquisition of Inuktitut in Baffin Island (Dorais 1993, 1996, pp. 135–43), and for Modern Tiwi in Australia (McConvell 2001). Reduced varieties like Modern Tiwi in their turn may become the point of departure for a process of koineization, as in the case of Dhuwaya, a variety of the Australian language Yolngu Matha (North East Arnhemsland), which may have its origin in a caretaker talk register (Amery 1993) and now serves as a regional lingua franca for some 4,000 people. Because of the presence of a dominant language in the community, children speaking such varieties often use a large amount of code-switching. In some cases this may lead to the emergence of a so-called “mixed language.” The emergence of “mixed languages” may involve imperfect L2 learning; at least, this is implied by the scenario

164

Kees Versteegh

sketched by Mous (2003) for Ma’a/Mbugu (Tanzania). This Bantu language has two lexical registers, one Bantu and one Cushitic. Originally, the speakers of Ma’a probably spoke a Cushitic language; the second register they created consists largely of material from their original language. There is some evidence that in shifting to Bantu they introduced the kind of reduction that is typical of L2 learners. Most instances of “mixed languages” mentioned in the literature seem to represent, however, cases of borrowing rather than shifting; for instance, Media Lengua (Muysken 1997), Mednyj Aleut (Thomason 1997), and Michif Cree (Bakker 1997; Bakker and Papen 1997). They all involve a group wishing to set themselves apart from the rest of the community, often because they are of mixed descent. In such a situation the introduction of a large amount of foreign lexical items may serve to underline the identity of the group. Such cases differ from pidginization because there are no L2 learners to introduce structural restructuring. There is no break in transmission and people continue to speak what they regard as their own language (Thomason 1997, p. 462). Such cases are therefore excluded from the present survey. Sign languages, however, sometimes fall into the category of interrupted transmission. Some deaf children grow up in isolated communities where they do not have access to normal education in sign language. Such children may develop a new system of communication relying on spontaneous gestures by their parents, but developing these into a consistent structure. Even when hearing parents consciously try to present them with a gestural model, the children surpass them by far in the amount of ordering and structural patterns, introducing new combinations and markers along the line (GoldinMeadow and Mylander 1990; Singleton, Morford, and Goldin-Meadow 1993). This is similar to children creolizing a restricted input such as a pidginized variety. On the other hand, deaf children who are introduced to sign language at a later stage when they have already been educated in a hearing language become acceptable signers at best, but their gestures never reach the complexity of those of native signers. It may be somewhat unusual to regard this as an instance of pidginization of a non-Indo-European language; nevertheless, it does seem to be a perfect analogue to some of the situations sketched here, and in fact the strategies used by these children in their signing are similar to those used in pidginization.

3

Early Contact and Later Expansion

All over the world, contacts between heterolingual groups, usually for purposes of trade, have necessitated linguistic accommodation. A regional survey of the most important examples will be given below; here we are concerned with the general features of such contacts. In many cases there is a clear distinction between clients/visitors and hosts. The speakers of the host group sometimes intentionally disguise their own language in order to facilitate acquisition by the visiting group, who are led to believe or assume that what

Non-Indo-European Pidgins and Creoles 165 they are learning is really the language of their hosts. After the initial contact several scenarios can be distinguished. When the visitors learn a reduced version of the language of their hosts, they may for a long time use it in their trading relations with them, as the Motu did on their expeditions. The arrival of European traders does not necessarily change this situation, since they may acquire and retain the reduced variety and, once they have established themselves as the new power, use it as a master–servant talk. In the first stages of the labor trade in the Pacific, European plantation owners used Pidgin Fijian as the main language of communication with the workers (Siegel 1987). In some cases, for instance Sango in the Central African Republic, this function of the reduced variety may be expanded into the direction of a regional lingua franca. In time, social or political developments may lead to the adoption of the lingua franca at a national level, in which case it is likely to become the L1 of some children. In a second scenario the European “visitors” eventually replace the existing means of communication by their own language, which in a restructured form becomes what is usually called an endogenous pidgin, such as the pidgin Englishes of West Africa. These are used at first in contacts between Europeans and indigenous populations but may end up as a general means of communication between indigenous heterolingual groups among themselves. If this lingua franca becomes the language of mixed households, children may start using it as their native language, i.e., the language is creolized. The third scenario is the classic tale of pidginization/creolization, in which the European colonial powers transport the indigenous population as slaves, in which case they are more or less forced to adopt a restructured version of the colonial language. This leads to the emergence of an exogenous pidgin and eventually to an exogenous creole. A very different situation is that of a military conquest in which the language of the conquerors is imposed upon the indigenous population. In such a case the pre-existing languages usually cease to be spoken because the population shifts to the new language, in the process introducing restructuring changes to their new language, as may have happened in the case of Arabic (Versteegh 1984, but see Baker 1996 for a critical view of this idea) and Sanskrit (Houben 1996). In some rare cases of linguistic contact there is no distinction between a host or a guest group, but the groups are more or less equal. Russenorsk, the language used between Russian and Norwegian traders, seems to have been an example of such a non-hierarchical contact in which both groups contributed to the lexicon (Fox 1983). Some of the North American jargons may constitute other examples: in the formation and use of Chinook Jargon, for instance, no single ethnic group seems to have been dominant, so that each group was free to contribute to the lexicon. There is every reason to assume that each of the different scenarios of language acquisition sketched here – and the list is by no means exhaustive – has its own concomitant linguistic effects, varying according to the role of the

166

Kees Versteegh

speakers and the learners, and also varying according to the different groups of new speakers, as in the case of the two varieties of reduced Motu. They all share one thing: the involvement of new speakers who are learning the language of their hosts as an L2 or who learn this variety as their L1. Since Whinnom’s article on secondary and tertiary hybridization (1971), it has been customary to distinguish various stages of pidgins. With the earliest stages of linguistic contact, the notion of “jargon” is sometimes used (Dench 1998, p. 52, referring to Mühlhäusler 1986, pp. 135, 147) for an “individual solution[s] to the problem of linguistic communication” as against a pidgin as a stable, conventionalized solution. There is no denying that there is a difference between the kind of reduced variety that is used in early trade contacts with foreigners, and second-language use of a reduced variety within an entire society. Such a distinction might be useful in discussing the effects on linguistic structure. But it should be pointed out that in the context of North American contact languages, common usage has adopted the term of “jargon” even for those cases where non-native speakers of the lexifier used it as a means of communication also among themselves. In Whinnom’s model this would be called “tertiary hybridization”; in Mühlhäusler’s terminology this would be a “real” pidgin. Some authors introduce the notion of a creole without a pre-existing pidgin to cover those cases in which a reduced variety has gone through an expansion of function and structure while remaining an L2 for most speakers, a process called by Mufwene (2001) “vernacularization.” Jourdan (1991, p. 195) states that this occurs when the reduced variety becomes the main language of the community even while remaining an L2. She adduces examples from Solomons Pijin that show L2 speaking parents expanding and L1 speaking children regularizing the structure. In the non-Indo-European cases surveyed here, creolization in the sense of nativization is rare, precisely because most of these varieties were and still are used as contact languages that are not supposed to be transmitted to one’s own children. In the case of Mobilian Jargon, for instance, even in bilingual marriages both partners insisted on teaching their children the full languages, even though among themselves they communicated in the jargon (Drechsel 1997, pp. 263, 336). Expansion or vernacularization of pidginized varieties is a quite common phenomenon. Linguistically, such an expansion results in a reduction of variation, the introduction of obligatory markers, and an increase in complexity through grammaticalization. In this process the lexifier language, if present, may be an important source. In Sango, for instance, the increased use of the verbal subject marker à- instead of a free pronoun may be attributed to Ngbandi influence (Samarin and Walker 1994). At this stage of the process, influence from the substratal languages, i.e., the L1s of the speakers, may become more marked, too. Keesing (1991), for instance, explains some of the grammatical elements in Melanesian pidgins (e.g., the transitivizer -em, the use of prepositional verbs, the grammaticalization of the motional verbs kam and go, and the grammaticalization of preverbal markers of tense, mode, and aspect) as the

Non-Indo-European Pidgins and Creoles 167 result of substratal influence from various Oceanic languages. This contrasts with the earlier situation in Tok Pisin when elements from Tolai were largely absent (Mosel 1980). In those communities in which L2 speakers live together with new L1 speakers, feedback from the latter may of course also result in the introduction of obligatory morphological and syntactic elements. In communities speaking related languages, a similar development may take place. When the speakers of such languages are forced to interact, usually in an urban environment, the result is a process of koineization, in which the differences tend to disappear. In some cases this may lead to a regularization of patterns, the disappearance of marked forms and other phenomena that are also characteristic of restructured registers. This is what happened, for instance, in the case of the Hindi varieties that sprang up in the diaspora (Siegel 1987; Mesthrie 1993) and also with some of the African languages that were used as a lingua franca, for instance Town Bemba and perhaps Lingala. This situation can become quite complicated when the same variety is also used as an L2 by speakers of other languages. As a result it is sometimes difficult to distinguish between changes connected with koineization by L1 speakers and those connected with the kind of restructuring that takes place in pidginization. Again, the question of whether or not the language in question is a pidgin is rather uninteresting.

4 Reduction and Expansion Two mechanisms are involved in the restructuring process. The first mechanism is the structural reduction that always takes place in L2 acquisition, but that is particularly active when the native speakers deliberately reduce the input in order to facilitate comprehension or to withhold knowledge about the “true” language from the new learners. The second mechanism is the restructuring that occurs when a reduced variety acquires native speakers, or when its function and structure are expanded because it becomes the main language in a community (Jourdan 1991, p. 195). This involves grammaticalization on a large scale and the disappearance of free variation. It is impossible to list all the effects of these mechanisms in non-IndoEuropean pidgins and creoles. One particular problem is that there is insufficient information about the effects of reduction in languages outside the group of familiar Indo-European languages, for instance, in languages with an ergative structure. It is sometimes asserted that in pidginization there is a general tendency towards SVO (Holm 2000, p. 233). Some of the non-Indo-European pidgins, however, present a clear preference for (S)OV word order, or even OSV with pronominal subjects (see also Bakker, this volume). A possible explanation might be that this occurs in ergative languages with cross-reference of subject and object in the verb. Transitive sentences with both overt subject and object seem to be rare in pidginized varieties (Dench 1998, pp. 34–5; Huttar and Velantie 1997, p. 105). Therefore, an absolutive noun followed by a verb

168

Kees Versteegh

may be reinterpreted as an object. Since in restructuring cross-reference endings disappear and pronominal subjects are often expressed by overt pronouns, the result is a sentence with an OSV order, e.g.: (1) Eskimo Pidgin nuna elekta awoña picuktu land depart 1s want ‘I want to go ashore.’ (Stefánsson 1909, p. 228) (2) Mobilian Jargon neta eno hoyo bear 1s hunt ‘I hunt bear.’ (Drechsel 1997, p. 120) (3) Ndyuka-Trio trade language kaikushi pumi wani wa dog 1s want neg ‘I don’t want a dog.’ (Huttar and Velantie 1997, p. 105) Reinterpretation may also play a role in the shift from VSO to SVO in Pidgin Hawaiian (Roberts 1995, p. 6). Here, the topic, usually the subject, is focused on by fronting and then reinterpreted as the subject of the sentence. In the absence of sufficient information on the restructuring of non-Indo-European languages, it is difficult to tell whether this reinterpretation derives from the foreigner talk of the native speaker or from the L2 learner. Strategies used in foreigner talk have been described extensively in the literature (see for instance Ferguson 1981; Ferguson and DeBose 1977; Ellis 1994, pp. 251–67), but they are rarely connected with the kinds of change in nonIndo-European pidgins given by Dutton (1985) for Hiri Motu, by Roberts (1995) for Pidgin Hawaiian, by Heine (1983) for Nubi, and by Heine (1979) and Manessy (1977) for African pidginized languages. A few examples will illustrate these changes: • reduction of phonological distinctions, resulting in the disappearance of a number of marked phonemes, for instance the velarized and laryngeal consonants of Arabic in Nubi or the implosives of Fula in Pidgin Fula. In Fanakalo the Zulu clicks do not appear in the speech of those speakers whose L1 does not contain these phonemes; • disappearance of case system when this exists in the lexifier: (4) Pidgin Ngarluma thatharruka watharri turtle look.for.fut ‘We’ll look for turtle.’ (Ngarluma thatharruka-ku watharri; -ku accusative suffix) (Dench 1998, p. 33)

Non-Indo-European Pidgins and Creoles 169 (5) Eskimo Jargon tuktu tautuk picuktu awoña caribou watch want 1s ‘I want to see (hunt) caribou.’ (Inuktitut tuktu-mik tautu-juma-junga; -mik instrumental suffix, -juma ‘want’, -junga 1s ending) (Stefánsson 1909, p. 223) • disappearance of a noun class or grammatical gender system (Heine 1979, p. 93); most dramatically this happens in the pidginized varieties of Bantu languages where large parts of the noun class system disappear. Pidginized varieties of Swahili, for instance, have only 6 noun classes as against 15 in Standard Swahili. In Fanakalo there are traces of some of the 15 Zulu noun classes, but the only productive plural is the one that prefixes ma- to the singular word; • disappearance of agreement system, for instance by the use of invariable adjectives in pidginized varieties of Bantu languages: (6) Fanakalo bomvu mti red tree ‘red tree’ (Zulu u-muthi o-bomvu; -u, -o class prefixes) (Sebba 1997, p. 60) (7) Kituba mw-ána ya mbote/b-ána ya mbote cl1-child conn good/cl3-child conn good ‘good child/good children’ (KiKongo mw-ána ú-mbote/b-ána bá-mbote; mw-, ú-/b-, ba- class prefixes) (Mufwene 1997, p. 178) • disappearance of tense marking in the verb, sometimes replaced by independent tense markers: (8) Fanakalo mina funa lo buku manje 1s read det book now ‘I am reading the book.’ (Zulu ngi-sa-funda ibhuku; ngi-1s.sub, -sa- prog) (Kaltenbrunner 1996, p. 77) (9) Kituba béto mé(n)e di-á(k)a 1p finish eat-ant ‘We have eaten.’ (KiKongo tu-á-di-idí; tu- 1p.sub, -á- ant, -idí pf) (Mufwene 1997, p. 179) • reduction of semantic distinctions, for instance the distinction between alienable and inalienable possession in Motu, which has disappeared in Hiri Motu (Dutton 1985, pp. 10–11);

170

Kees Versteegh

• generalization of allomorphy, for instance when different forms of negation merge into a single form in the pidginized variety (Dutton 1985, p. 14); • analytic expression of possessives, if these are expressed in the lexifier as suffixes or possessive pronouns: (10) Pidgin Hawaiian mahea dala oe where money you ‘Where is your money?’ (Hawaiian i hea kau kala; i hea loc where, kau 2s.poss pron) (Roberts 1995, p. 33) (11) Eskimo Jargon awoña kammik I boot ‘my boot’ (Inuktitut kami-ga; -ga 1s.poss suffix) (Stefánsson 1909, p. 225) • analytic negation, if the negation is expressed within the verbal complex in the lexifier: (12) Fanakalo ayikona lo kati lo neg det cat cop ‘This isn’t a cat.’ (Zulu a-kuji-kati; a- cl prefix, -kuji- neg) (Kaltenbrunner 1996, p. 84) (13) Eskimo Jargon imek picuktu pitcûk water want nothing ‘He does not want water/it is water-tight.’ (Inuktitut imir-mik pisu-ngngittuq; -mik inst, -ngngit neg, -tuq 3s) (Stefánsson 1909, p. 229) • a frequent strategy concerns the use of free pronouns to indicate subject/ object rather than bound pronouns on the verb; since most Indo-European languages have free pronouns, this is not a trait often mentioned in pidgin and creole studies, but outside this restricted group it is found in almost all restructured languages: (14) Hiri Motu lau ia itaia 1s.obj 3s.sub see ‘He sees me.’ (Motu e-ita-gu; e- 3s-sub, -gu 1s-obj) (Dutton 1985, p. 9) (15) Nubi íta dúrub-u ána you hit-tr I ‘You hit me.’ (Arabic Darab-ta-ni; -ta 2s, -ni 1s.obj suffix)

Non-Indo-European Pidgins and Creoles 171 (16) Kenya Pidgin Swahili mimi ta-piga yeye I fut-hit he ‘I’ll hit him.’ (Swahili ni-ta-m-piga; ni- 3s.sub, -ta- fut, -m- 3s.obj) (Heine 1979, p. 94) The strategies or tendencies discussed here may be summarized as follows: there is a general reduction of phonological and morphological categories and semantic distinctions; there is a tendency toward analytic expression instead of synthetic; there is a tendency to get rid of redundancy. Some of the processes mentioned here, in particular the disappearance of allomorphy and the regularization of paradigms, occur in other language contact situations as well. Thus, for instance, in koineization there is always a certain degree of regularization. But it is probably fair to say that the strategies mentioned here are most frequent in those situations in which L2 learners face native speakers who accommodate them by using a foreigner talk register.

5 Contact Languages and Jargons: The Americas North American contact languages are often referred to as “jargons,” such as Chinook Jargon (Thomas 1954; Grant 1996), Mobilian Jargon, Delaware Jargon, and various Eskimo jargons (Drechsel 1981); in fact because of their stability they do not differ from other pidginized varieties. Some researchers (Silverstein 1972; Samarin 1986, 1996) maintain that these jargons were the result of contact with European settlers. Drechsel (1997, pp. 21–2, 281–2, 337–9) shows, however, that there is abundant evidence of an earlier use of Mobilian jargon as a contact language, long before the arrival of the Europeans. The newcomers were addressed in the existing contact languages and quite naturally used these in their communication with the indigenous population. Of course, this does not preclude the possibility that other jargons originated in the contact between native Americans and European traders, as Goddard (1997) maintains of Delaware Jargon, a reduced version of Unami. The North American “jargons” are characterized by a high degree of variability, especially in the lexicon, which seems to support the idea that they functioned as an interethnic means of communication whose precise form depended on the socio-political circumstances and the geographical location, each dominant group imposing or inserting its own lexical items. Yet, according to Drechsel (1997, pp. 198–9), Mobilian Jargon was stable in the sense that there was a permanent grammatical core which did not change; other “jargons” may have been more variable, such as Chinook Jargon (Silverstein 1972; but see Thomason 1983). Perhaps in different regions of America the needs for interethnic communication differed and so did the socio-political circumstances, the “ecology of the language.”

172

Kees Versteegh

In the case of the Eskimo jargons, the literature mentions several instances of trade languages both between the Inuit and neighboring groups and between Inuit and Euro-American whalers (Van der Voort 1996a). Lone travelers report that in their contacts with monolingual Inuit these developed a reduced variety on the spot in order to enable them to communicate, but in some cases the reduced register seems to have become a regular means of communication, for instance the Herschel Island Trade Jargon described by Stefánsson (1909), which was regularly used with visiting whalers and contained words from other contact languages the whalers were familiar with, even from Hawaiian (e.g., waihinni ‘woman’, Hawaiian wahine), or the West-Greenlandic Pidgin that is reported to have been used in contacts with European traders (Van der Voort 1995, 1996b). In Latin America several Amerindian languages served as lingua francas before the Spanish and Portuguese conquests. Quechua, the language of the Inca empire, is still spoken in many varieties as a lingua franca. Some of these varieties may have originated as pidginized varieties of an Inca trade language, for instance Ecuadorean Quechua (Muysken 1977). In the Amazon region Tupi and Guaraní, two related languages, were widely used among ethnic groups speaking different languages. In colonial Paraguay, where the number of Spanish immigrants was much lower than elsewhere, Guaraní became one of the two languages that everyone spoke (Lipski 1994, pp. 303–14). In the large reducciones (missions), the Jesuits standardized the language and introduced an educational system in which the language was taught as the language of the mission. After the expulsion of the Jesuits in 1768, the population remained bilingual in Guaraní and in the language of their ethnic group, without signs of extensive restructuring of Guaraní; its transmission had been relatively uninterrupted and the standard introduced by the Jesuits prevented drastic changes. In 1992 Guaraní became the second official language of Paraguay. It is not clear how jopará, a variety of Guaraní with a high degree of mixing with Spanish, should be regarded: it is sometimes called a creolized form of Guaraní (Dietrich 2002), but is perhaps better described as a kind of mixed language similar to Media Lengua. Meanwhile in neighboring Brazil the Portuguese had officially recognized a lingua franca based on another Tupi-Guaraní language, Tupinambá, as official language of Amazonia, under the name of Língua Geral. This language, also known as Nheengatu, was learned by a large number of new speakers, which led to a drastic simplification of the morphology and the syntax (Rodrigues 1996; Moore, Facundes, and Pires 1994; Jensen 1999, p. 127). In Brazil the name Língua Geral is also used for some varieties of Guaraní that were introduced as lingua francas in other regions. In Suriname several trade languages were developed in contacts between speakers of Surinamese creoles and indigenous languages, for instance the Ndyuka-Trio Pidgin (Huttar and Velantie 1997). The Ndyuka speakers are Afro-Surinamese maroons; they call this trade language Ingii ‘Indian’. The Indian Trio speakers call it Mekolo ‘Black’. This suggests that there are two

Non-Indo-European Pidgins and Creoles 173 varieties of the language, lexically differing from each other and used by both groups as the appropriate way of addressing the other group. The 30,000 Black Caribs (Garífuna) in Honduras, Belize, and Guatemala are a special case (Escure 2004). They are the descendants of African slaves who escaped to the island of St. Vincent where they mixed with the local population of Arawaks. These seem to have spoken two distinct languages: one male and one female, the latter probably derived from a Carib language. Since the escaped slaves married Arawak wives, their language was based on the female speech. In 1797 the Black Caribs were resettled by the English on the island of Roatan, whence they spread to the coastal areas (Gonzalez 1988). According to Taylor (1951) the language spoken nowadays by their descendants does not differ greatly from the Arawak language as it was spoken on St. Vincent, which would mean that any effects of restructuring on the part of the African slaves were offset by the transmission of Arawak through their wives to the next generations.

6 Trade Languages and Lingua Francas: Asia and Oceania There is surprisingly little information on pidginized varieties of the large Asian languages such as Arabic, Turkic, Persian, Japanese, and Chinese. In the Indian Ocean trade, where Arab seafarers had been active even before Islam, one might have expected the emergence of Arabic pidgins. But as Baker (1996) explains, it is not very likely that Arabic was used as a trade language with incipient pidginization in the Indian Ocean trade, because of the presence of other lingua francas, in particular Malay. Baker (1996, p. 663) leaves open the possibility that Arabic pidgins developed in some of the Arab principalities in Indonesia, or in the ports in the Arabian peninsula, although there is no evidence of this. A contemporary Arabic pidgin is spoken in the Gulf (Smart 1990) between foreign workers and Arabs. This pidgin has some of the features of pidginization mentioned above, such as the use of demonstratives instead of the definite article, invariable verbal forms, and a tendency to use independent pronouns instead of suffixed markers. All along the ancient Silk Road various trade languages must have been used, in particular Persian. In fairly recent times it was still possible to get by with Persian along the old Silk Road as far as the province of Sinkiang in western China (Wurm 1995). In Siberia and northwestern China several languages were used as lingua francas, for instance in Siberia Yakut and Chukchi (de Reuse 1996) and in northwestern China Oirat, a Mongolian language. But the most widespread languages of general communication in recent times are no doubt Chinese and Russian. Several restructured forms of Russian are currently used in Siberia, for instance in the Taymyr Peninsula (Wurm 1992, 1996). Most of these restructured forms exhibit the same characteristics, such

174

Kees Versteegh

as complete loss of declensional and inflectional endings, use of the oblique forms of the pronouns in subject position, and a synthetic system of tense marking (e.g., tebja kogda letat’ budet’ ‘when will you fly?’ with invariable budet’ ‘he/she will’ as a future marker). Between Russian and Chinese traders a trading jargon developed around the town of Kyakhta (Neumann 1966) with similar characteristics, but with a varying amount of Chinese elements. According to some reports the language was more like Russenorsk in that speakers could choose between two registers, one more Russian-lexified, the other more Chinese (Fox 1983). Apart from the lingua francas Wurm (1995) also mentions the existence of four what he calls hybridized languages in Northwestern China along the Silk Road: Ejnu in Xinjiang, Tangwang and Hezhou in Gansu, and Wutun in Tibetan Qinghai. The case of Hezhou spoken in the Linxia Autonomous Region of Gansu province in China is typical: the language has Mandarin lexicon and Uyghur grammatical structure. According to Lee-Smith (1996), this is a case of heavy borrowing from Chinese by a group of Uyghur-speakers. Wurm’s explanation of the emergence of these languages is somewhat puzzling: in his view they have emerged through a process in which the speakers of one language – Persian in the case of Ejnu, Turkic in the case of Hezhou, and Chinese in the case of Tangwang and Wutun – intermarried with women speaking another language. This led to a situation in which the mothers transmitted to their children the grammatical structure and the fathers the lexicon. But it is obvious that more details about the structure and history of these languages are needed before a satisfactory explanation of their development can be given (for more details see Wurm, Mühlhäusler, and Tryon 1996). Since there is no evidence that they have been transmitted to new speakers at all – except for Hezhou which seems to be in use as a trade language over a much larger area in Gansu and Qinghai provinces – the linguistic mixture they exhibit may be more akin to that of Mednyj Aleut and Michif Cree, which excludes them from the present survey. After Japan’s opening up by the Treaty of Kanagawa in 1854, the first contacts between foreigners and Japanese took place in a more or less conventionalized Pidgin Yokohamese (Mühlhäusler and Trew 1996, p. 376). The existence of various Japanese contact languages or colonial varieties during Japan’s occupation of Manchuria, Korea, Taiwan, New Guinea, and Micronesia is alluded to in the literature, but not much information on their structure is available (Miyawaki 2002). In Manchuria a Pidgin Chinese is mentioned. The status of the Ogasawara Mixed Language is unknown: according to Long (1999) there must have been an English-based contact language on these islands (called at the time the Bonin Islands), but after reunion with Japan a contact language based on Japanese, or possibly a mixed language composed of Polynesian and Japanese elements, developed. Korean troops in Vietnam during the Vietnam war are reported to have learned a Pidgin Vietnamese. In all these cases military occupation was the occasion of the emergence of a contact version.

Non-Indo-European Pidgins and Creoles 175 The South Asian subcontinent has been home to a large number of migrations. According to some scholars the development of the Indo-Aryan languages was a repeating history of pidginization as the Sanskrit language was learnt by peoples with different languages (Houben 1996). In this way Southworth (1971) explains the origin of Marathi, which in his view emerged as the result of imperfect L2 acquisition by Dravidian speakers and was then creolized. In modern times the vernaculars descending from Sanskrit function as regional standards, whereas Hindi together with English is recognized as a national language. Many pidginized urban varieties of Hindi are spoken as an L2, for instance in Bombay and Calcutta. Regional languages, too, exist in pidginized varieties, for instance Naga Pidgin Assamese, spoken as an L2 throughout Nagaland (Sreedhar 1974), and Vedda Creole, spoken in Sri Lanka by the indigenous population, which seems to have originated as a pidginized version of Sinhalese (Shapiro and Shiffman 1983, pp. 184–214). Strictly speaking, of course, these examples do not belong here since they represent cases of pidginization of Indo-European languages. In Southeast Asia the most important trade language was a form of Malay, sometimes called Pasar Malay. Its use is first attested as early as 700 CE and by 1500 it had spread over the entire Indonesian archipelago and beyond. Across this area different pidginized varieties of this trade language were adopted, some of which eventually became L1s (Collins 1980), such as Kupang Malay, Makassarese Malay, and Ambonese Malay (Van Minde 1997). The standard language of Indonesia, Bahasa Indonesia, is often stated to be based on the Malay trade language. Actually, during the independence struggle the nationalists chose a form of standard Malay, Riau Malay (Steinhauer 1980), as the basis for the new national language. The acceptance of the new national language was partly supported by the existing regional varieties of Malay, which continued to co-exist with Bahasa Indonesia in a kind of diglossia. For most speakers in Indonesia, however, Bahasa Indonesia is an L2, which gives rise to varying degrees of code-switching. Bahasa as spoken by the Javanese Chinese (Dreyfuss and Oka 1979) is a different case; their language probably originated when men who were L2 speakers of Pasar Malay intermarried with Javanese women. One of the linguistically most diversified areas in the world, New Guinea has a high degree of multilingualism. Since language communities are usually very small, people are accustomed to using the vernaculars of other groups (cf. the description of the situation in the Taiap speaking village of Gambun by Kulick 1992, pp. 61–91). It is not very clear what exactly this competence in other languages means. In view of the fact that people are always interested in picking up single words and commands from other language groups and using them – what is called in Tok Pisin autsait tok ‘outgroup talk’ (Kulick 1992, p. 65) – it is easy to imagine that the self-professed competence in other languages often amounts to no more than the ability to establish a communicative link with other people by using words from their language. In fact, knowing many languages is seen as a mark of prestige in New Guinea (Foley 1986, pp. 24–5).

176

Kees Versteegh

In trading it is important to use the language of the host one is trading with. Some trading languages may have been the exclusive property of individual families, as in the case of the Mekeo (Austronesian) pidgins in the Papuan Tip of New Guinea: the reduced forms of Mekeo they taught to other groups reflect their relatively more powerful position (Jones 1996). Other trading languages came to be used in a wider area (Foley 1986, pp. 30–5), such as Kâte in the Huon Peninsula and the central highlands, Yabem, Iatmul, and Kuanua (Tolai). The use of these regional trade languages was propagated by the various missionary societies, but most of them were superseded by Tok Pisin and to a lesser extent by a pidginized form of the Austronesian language Motu, formerly a law enforcement lingua franca known as Police Motu, but nowadays called Hiri Motu. It is spoken by 120,000 speakers (Dutton 1985) in the coastal areas of Papua New Guinea and has only very few L1 speakers. An important point is that the Motu differentiated the reduced form of their language; with speakers of other Austronesian languages they used (simplified) subject markers, whereas with speakers of Papuan languages, they deleted them completely. In Australia contact languages that were used over a wider area must have existed for a long time, but there is precious little information about them, since they have been wiped away by the later use of pidginized varieties of English; according to Mühlhäusler and Amery (1996), most of these were koines or lingua francas, in which some leveling took place, but little reduction or simplification. Pidgin Ngarluma, a variety without case endings and with a highly reduced morphology, was presumably used by many groups as a general contact language on the northwest coast of Western Australia. Another jargon mentioned in the literature is the Kaurna jargon that was briefly used as a contact language in the Adelaide region (Simpson 1996). In Fiji, a reduced form of Fijian had been used in outside contacts for a long time, even before the advent of the Europeans (Siegel 1987), and this must have been the case in Hawai[i, too. The picture that emerges from the extensive documentation brought together by Roberts (1995) is that of a thriving Pidgin Hawaiian that prior to the establishment of Hawaiian Pidgin English was used between Hawaiians and any other ethnic groups, including not only Chinese and Japanese laborers but also at an earlier time European whalers and foreigners who came to live on the islands. Roberts demonstrates convincingly that Bickerton was right in claiming the existence of such a pidginized form of Hawaiian (Bickerton and Wilson 1987), as against Goodman (1985). Through the whalers items from this pidginized Hawaiian found their way into other contact situations, as far north as Eskimo Jargon.

7

Urban Multilingualism and Two Creoles: Africa

In Africa there are at least two cases of non-Indo-European creoles, (Ki-)Nubi, an Arabic creole in Uganda and Kenya, and Sango, one of the two national

Non-Indo-European Pidgins and Creoles 177 languages of the Central African Republic. Arabic has long had a commercial presence in sub-Saharan Africa. In a country such as Mali, for instance, Arabic was primarily the language of trade and interethnic politics before it became associated with religion as the language of the Qur’an. But in the sub-Saharan states that emerged as the result of these contacts, local languages became the preferred lingua franca, such as Hausa in the Hausa “states,” and Kanuri in the Kanem-Bornu empire (Croft 2003, p. 23). As a result, the main language of the Islamic mission in the West African savanna region as far west as modern Ghana was not Arabic, but Hausa. In the initial contacts, Arabic-based pidgins may have been in use: the geographical literature mentions local contact languages based on Arabic, such as the Maridi Arabic pidgin that was used in the eleventh century in central Mauritania (Thomason and Elgibali 1986; but see Owens 1997, p. 132). In East Africa commercial relations along the Nile stretched as far as Sudan and Ethiopia, where undoubtedly local contact varieties of Arabic existed (Owens 1997). The origin of the modern (Ki-)Nubi, spoken by about 50,000 people in Kenya and Uganda, lies in the southern Sudan in the period of the Anglo-Egyptian conquest of the Sudan during the nineteenth century (Wellens 2005). The recruits for the victorious army came from various peoples in the southern Sudan, such as Nuer, Shilluk, Bari, and Nubi. It is not clear whether they learned a kind of barracks language (sometimes called Bimbashi Arabic) or had some previous experience with an Arabic pidgin based on the old trade language. After the Mahdist revolt the soldiers escaped together with their officers to the British colonies of Uganda and Kenya, where they settled and took local wives. In this community a native language was developed, called (Ki-)Nubi. Compared to its lexifier it is characterized by a drastic reduction of the phonemic inventory, introduction of grammatical tone to indicate such categories as plural, passive and gerund, a complete reduction of the verbal system, with a full set of tense/mood/aspect markers (Owens 1985, 1990). Meanwhile in the southern Sudan the pidginized form of Arabic remained in use as a lingua franca and in the capital of the southern Sudan, Juba, it even acquired a number of native speakers. In recent times, the influence of Khartoum Arabic has set in motion a process of decreolization (Versteegh 1993b). Another offshoot of the (west-)Sudanese contact variety is Turku, which was brought to the Chari River basin in the second half of the nineteenth century by a group of rebellious soldiers; it is now extinct (Tosco and Owens 1993). A present-day descendant of this pidgin may be Bongor Arabic (árab haná bóngor, or túrku or túrgu), which is used in south-western Chad as a lingua franca by speakers of Masa (Chadic) and Tupuri (Niger-Congo) in their contacts with Arab traders (Luffin 2008). In the Horn of Africa, Arabic is in widespread use as a lingua franca and has been affected by intensive reduction (Simeone-Senelle 2008). Sango must have been in use as a lingua franca in pre-colonial times (Heine 1973, p. 58, against Samarin 1982). It is not known to what degree its main lexifier, the Yakoma dialect of Ngbandi, was restructured during this period.

178

Kees Versteegh

Later it became the language of French expansion along the Ubangi river. Its nickname Sango si tulúgu ‘soldiers’ Sango’ suggests that the domains in which it was used in the beginning were restricted. When Sango became the second national language of the Central African Republic, together with French, its functions were expanded and its grammatical structure became more complex. In modern times the language is steadily acquiring L1 speakers, especially in the capital Bangui. The Nubi creole originated in a military environment, and the same applies to Barikanci Pidgin Hausa (Hodge 1958), which became the basis for the expansion of Hausa as an L2, for instance in Ghana where it is still associated with the military and the police (Kropp Dakubu 1997, pp. 129–38). Ewondo populaire, a pidginized variety of the Bantu language Bulu, also called Pidgin A-70, is another work-related reduced language used chiefly by traders and truck drivers along the highways and in the markets of Cameroon (Heine 1973, pp. 87–8). A different work-related context in which reduced languages are used is that of the mining system which gave rise to such languages as Fanakalo in South Africa and Town Bemba in Zambia. The former, a reduced version of Zulu (Sebba 1997; Kaltenbrunner 1996), became the chief means of communication between black workers and white supervisors in South Africa, especially in the mines; it used to be employed also in Zimbabwe (under the name chiLapalapa), the Congo, and Zambia, but was replaced by English, Shaba Swahili, and Town Bemba. Although one of the nicknames of Fanakalo, Isikula ‘language of coolies’, might lead one to suppose that its origin is connected with the arrival of Indian laborers in South Africa, the language seems to have been in use some decades before the first Indians arrived (Mesthrie 1989). Town Bemba, L2 of 2–3 million people in Zambia, is a special case. The language originated in the Copper Belt as a general means of communication between miners speaking different languages. It is derived from ichiBemba, not unlike the way Fanakalo originated from Zulu. Later on it became the general code for urban communities, most of whom speak it as an L2, although the number of L1 speakers is growing. The structure of Town Bemba as compared to its lexifier exhibits some leveling of the tense system, regularization of irregular verbs, and some reduction of tone and vowel length; besides, its speakers code-switch frequently with English. The language is generally classified as a pidgin (creole for some speakers), but Spitulnik (1999) objects to this label. In her view Town Bemba is a variety of Bemba, and she regards the label of “pidgin” as a typically Western way of denigrating the African urban way of speaking. Yet, it cannot be denied that Town Bemba’s origins look like the kind of environment in which a language is being used as a lingua franca between various languages. The generally low amount of restructuring might be explained by the fact that the languages involved in the emergence of Town Bemba were all related Bantu languages. The effect of relatedness might also play a role in explaining the difference between the development of two other varieties of a Bantu language that are widely used as a lingua franca, Lingala and Kituba. It is generally held that in the creation of Kituba as a lingua franca in the western part of the modern

Non-Indo-European Pidgins and Creoles 179 Democratic Republic of the Congo, speakers of West African languages were involved, who started to use a pidginized variety of KiKongo when they were working on the Matadi–Kinshasa railroad. Lingala originated as a lingua franca, called Mangala and is based to a large degree (30 percent of the lexicon) on Bobangi, along the river systems of northwestern Congo. With the arrival of European traders, this language became the chief means of communication; nowadays it has spread to the major Congolese cities (Knappert 1979). Since in the development of Lingala only speakers of related Bantu languages played a role until the arrival of the Europeans, its structure compared to Kituba shows a much greater retention of the original Bantu system, for instance in agreement (Rottland 1979; Mufwene 1997). This shows again the difficulty of distinguishing between pidginization and koineization. In koineization, a certain regularization takes place, but usually no simplification of the kind that is found in the restructuring of a language by speakers of unrelated languages (Siegel 1993a). The phenomenon of special urban varieties might be typical of the African situation. In the case of Fula (also Fulani, Fulfulde, Peul) the eastward conquests of the Ful have led to the emergence of lingua franca varieties which in present-day Cameroon are called Bilkiire or Kambariire, distinguished from “pure” Fula ( fulfulde laamnde). These L2 varieties are characterized by (partial) absence of implosives, loss of phonemic length and gemination, suppression of the singular/plural alternation, elimination of the noun class system, reduction of ideophones, and loss of passive and emphatic voice and most verbal extensions. In the case of the singular/plural alternation the plural is expressed by analytic means, for instance, by a marker indicating ‘many’ (Noss 1979). A new development in the Ngaoundéré region in Cameroon is the emergence of an L1 variety of Fula that differs noticeably from the “pure” Fula, borrows heavily from French, yet retains some inflectional features. In the case of the singular/plural alternation, for instance, this variety – which Noss (1979, pp. 183ff.) calls a creolized variety – introduces plural suffixes that do not exist in Bilkiire. In this respect urban Fula resembles both Town Bemba and the urban varieties of Swahili in Nairobi and Kampala described by MyersScotton (1979). Swahili is the most widespread L2 in a large part of Africa (Tanzania, Kenya, Uganda, Kongo). Since some features of vernacular Swahili seem reduced when compared to other Bantu languages, it is sometimes asserted that the language must have gone through several stages of pidginization in its history. According to Nurse (1997; Nurse and Spear 1985) heavy contact with adjacent languages during the settlement of Swahili speakers in the coastal regions of East Africa may have led to a certain degree of restructuring of the language. At a later period when Swahili trade networks were expanding inland, pidgin varieties must have sprung up everywhere along the trade routes. Some of the present-day L2 varieties of Swahili are probably descendants of these pidgin varieties (Duran 1979, p. 151). Some varieties of L2 Swahili seem to have originated as speakers of related Bantu languages adopted Swahili as their common means of communication.

180

Kees Versteegh

This seems to have been the case with Shaba (Lubumbashi) Swahili, which was introduced in 1918 in the copper mining industry in Katanga (Congo) to replace Fanakalo. The tendencies found in this variety are similar to those found in Pidgin Swahili elsewhere, but less fully implemented: the noun class agreement system and the subject/object marking in verbs still exist, though in a reduced form. In this respect Shaba Swahili resembles koineized varieties of related languages. Nowadays, Shaba Swahili has become the L1 for most children in the area (De Rooij 1995).

8

Conclusion

In her description of the language situation in Ghana, Kropp Dakubu (1997) uses the Ga expression KOOle èya ègshOè ‘Korle meets the sea’ as a Leitmotif. This expression evokes the powerful image of the water of the Accra lagoon flowing into the open water of the sea as a warning referring to non-Ga speakers listening in on conversations in Ga. Kropp Dakubu (p. 14) then concludes: This suggests that the use of second languages has provided a filter through which the inland waters can be allowed to approach the sea with minimum danger to all concerned, a safe channel for linguistic communication that does not obviously threaten the carefully constructed protection of ethnic boundaries until they can be rebuilt in different places.

We as linguists are mainly interested in the linguistic effects of this process of accommodation. For the speakers who have been referred to in the course of this chapter, finding a way to achieve this accommodation is a matter of life and death, the only safe alternative in a threatening world full of people speaking strange languages.

REFERENCES Amery, Rob (1993) An Australian koine: Dhuwaya, a variety of Yolngu Matha spoken at Yirrkala in North East Arnhemland. In: Siegel (ed.), pp. 45–64. Andersen, Roger W. (1983) Introduction: A language acquisition interpretation of pidginization and creolization. In: Roger W. Andersen (ed.) Pidginization and Creolization as Language Acquisition. Rowley, MA: Newbury House, pp. 1–56. Baker, Philip (1996) The potential for the development of Arabic-based and other contact languages along the maritime trade routes between the Middle East and China, from the start of the Christian era. In: Wurm et al. (eds.), pp. 637–72. Bakker, Peter (1997) A Language of Our Own: The Genesis of Michif, the Mixed Cree-French Language of the Canadian Métis. Oxford: Oxford University Press. Bakker, Peter and Robert A. Papen (1997) Michif: A mixed language based on Cree and French. In: Thomason (ed.), pp. 295–363.

Non-Indo-European Pidgins and Creoles 181 Belikov, Vladimir I. (1996) Use of languages in the southern part of the Russian Far East. In: Wurm et al. (eds.), pp. 1013–32. Bickerton, Derek and William H. Wilson (1987) Pidgin Hawaiian. In: Glenn G. Gilbert (ed.) Pidgin and Creole Languages: Essays in Memory of John E. Reinecke. Honolulu: University of Hawai[i Press, pp. 61–76. Bolton, Kingsley (2003) Chinese Englishes: A Sociolinguistic History. Cambridge: Cambridge University Press. Collins, James T. (1980) Ambonese Malay and Creolization Theory. Kuala Lumpur: Dewan Bahasa dan Pustaka. Croft, William (2003) Social evolution and language change. http://www.unm.edu/ ~wcroft/Papers/SocLing.pdf (accessed August 31, 2007). Dench, Alan (1998) Pidgin Ngarluma: An indigenous contact language in North Western Australia. Journal of Pidgin and Creole Languages 13, 1–61. De Rooij, Vincent (1995) Shaba Swahili. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An introduction. Amsterdam: John Benjamins, pp. 179–204. Dietrich, Wolf (2002) Guaraní criollo y guaraní étnico en Paraguay, Argentina y Brasil [Creole Guaraní and native Guaraní in Paraguay, Argentine and Brazil]. In: Mily Crevels, Simon Van de Kerke, Sérgio Meira, and Hein Van der Voort (eds.) Current Studies on South American Languages. Leiden: CNWS, University of Leiden, pp. 31–41. Dorais, Louis-Jacques (1993) From Magic Words to Word Processing: A History of the Inuit Language. Iqaluit, Canada: Nunavut Arctic College. Dorais, Louis-Jacques (1996) La parole inuit: langue, culture et société dans l’Arctique nordaméricain [Inuit speech: Language, culture and society in the North-American Arctic]. Paris: Peeters. Drechsel, Emanuel J. (1981) A preliminary sociolinguistic comparison of four indigenous pidgin languages of North America (with notes towards a sociolinguistic typology in American Indian linguistics). Anthropological Linguistics 23, 93– 112. Drechsel, Emanuel J. (1997) Mobilian Jargon: Linguistic and Sociohistorical Aspects of a Native American Pidgin. Oxford: Clarendon Press. Dreyfuss, Gail Raimi and Djoehana Oka (1979) Chinese Indonesian: A new kind of hybrid? Australian National University, Research School of Pacific Linguistics, Canberra, Papers in Pidgin and Creole Linguistics, 2, 247–74. Duran, James J. (1979) Non-standard forms of Swahili in west-central Kenya. In: Hancock (ed.), pp. 129–51. Dutton, Tom (1985) Police Motu: Iena Sivarai (Its Story). Port Moresby: The University of Papua New Guinea Press. El Aissati, Abderrahman (1996) Language Loss Among Native Speakers of Moroccan Arabic in the Netherlands. PhD dissertation, University of Nijmegen, The Netherlands. Ellis, Rod (1994) The Study of Second Language Acquisition. Oxford: Oxford University Press. Escure, Geneviève (2004) Garifuna in Belize and Honduras. In: Geneviève Escure and Armin Schwegler (eds.) Creoles, Contact and Language Change. Amsterdam: John Benjamins, pp. 35–65. Ferguson, Charles A. (1981) “Foreigner talk” as the name of a simplified register. International Journal of the Sociology of Language 28, 9–18.

182

Kees Versteegh

Ferguson, Charles A. and Charles E. DeBose (1977) Simplified registers, broken language, and pidginization. In: Valdman (ed.) Pidgin and Creole Linguistics. Bloomington and London: Indiana University Press, pp. 99–125. Foley, William A. (1986) The Papuan Languages of New Guinea. Cambridge: Cambridge University Press. Fox, James A. (1983) Simplified input and negotiation in Russenorsk. In: Roger W. Andersen (ed.) Pidginization and Creolization as Language Acquisition. Rowley, MA: Newbury House, pp. 94–108. Gingiss, Peter (1979) Commercial Dyula: A pidgin’s first cousin. In: Hancock (ed.), pp. 99–105. Goddard, Ives (1997) Pidgin Delaware. In: Thomason (ed.), pp. 43–98. Goldin-Meadow, Susan and Carolyn Mylander (1990) Beyond the input given: The child’s role in the acquisition of language. Language 66, 323–55. Gonzalez, Nancie L. (1988) Sojourners of the Caribbean: Ethnogenesis and Ethnohistory of the Garifuna. Urbana and Chicago: University of Illinois Press. Goodman, Morris (1985) Review of Roots of Language, by Derek Bickerton (1981). International Journal of American Linguistics 51, 109–37. Grant, Anthony P. (1996) Chinook jargon and its distribution in the Pacific Northwest and beyond. In: Wurm et al. (eds.), pp. 1184–208. Hancock, Ian F. (ed.) (1979) Readings in Creole Studies. Ghent: E. Story-Scientia. Heine, Bernd (1973) Pidgin-Sprachen im Bantu-Bereich [Pidgin languages in the Bantu domain]. Berlin: D. Reimer. Heine, Bernd (1979) Some linguistic characteristics of African-based pidgins. In: Hancock (ed.), pp. 89–98. Heine, Bernd (1983) The Nubi Language of Kibera: An Arabic Creole. Berlin: D. Reimer. Hodge, Carleton T. (1958) Non-native Hausa. Georgetown University, Washington, DC, Monograph Series on Languages and Linguistics 11, 57–69. Holm, John (2000) An Introduction to Pidgins and Creoles. Cambridge: Cambridge University Press. Houben, Jan E. M. (1996) Sociolinguistic attitudes reflected in the work of Bhartrhari and some later grammarians. In: Jan E. M. Houben (ed.) Ideology and Status of Sanskrit: Contributions to the History of the Sanskrit Language. Leiden: E. J. Brill, pp. 156–93. Huttar, George L. and Frank J. Velantie (1997) Ndyuka-Trio Pidgin. In: Thomason (ed.), pp. 99–124. Jahr, Ernst Håkon and Ingvild Broch (eds.) (1996) Language Contact in the Arctic: Northern Pidgins and Contact Languages. Berlin and New York: Mouton de Gruyter. Jensen, Cheryl (1999) Tupí-Guaraní. In: R. M. W. Dixon and Alexandra Aikhenvald (eds.) The Amazonian Languages. Cambridge: Cambridge University Press, pp. 125–63. Jones, Alan A. (1996) Privately owned Mekeo-based trade languages. In: Wurm et al. (eds.), pp. 219–24. Jourdan, Christine (1991) Pidgins and creoles: The blurring of categories. Annual Review of Anthropology 20, 187–209. Kaltenbrunner, Stefan (1996) Fanakalo: Dokumentation einer Pidginsprache [Fanakalo: Documentation of a pidgin language]. Vienna: Universität Wien, Institut für Afrikanistik. Keesing, Roger M. (1991) Substrates, calquing and grammaticalization in Melanesian Pidgin. In: Elizabeth Closs Traugott and Bernd Heine (eds.) Approaches to Grammaticalization, 2 vols. Amsterdam: John Benjamins, pp. 315–42.

Non-Indo-European Pidgins and Creoles 183 Knappert, Jan (1979) The origin and development of Lingala. In: Hancock (ed.), pp. 153–64. Kropp Dakubu, M. E. (1997) Korle Meets the Sea: A Sociolinguistic History of Accra. Oxford and New York: Oxford University Press. Kulick, Don (1992) Language Shift and Cultural Reproduction: Socialization, Self, and Syncretism in a Papua New Guinean Village. Cambridge: Cambridge University Press. Lee-Smith, Mei W. (1996) The Hezhou language. In: Wurm et al. (eds.), pp. 865–73. Lipski, John M. (1994) Latin American Spanish. London and New York: Longman. Long, Daniel (1999) Evidence of an English contact language in the 19th century Bonin (Ogasawara) islands. English World-Wide 20, 251–86. Luffin, Xavier (2008) Pidgin Arabic: Bongor Arabic. In: Mushira Eid, Alaa Elgibali, Manfred Woidich, Kees Versteegh, and Andrzej Zaborski (eds.) Encyclopedia of Arabic Language and Linguistics, vol. 3, pp. 634–9. Leiden: Brill. Manessy, Gabriel (1977) Processes of pidginization in African languages. In: Valdman (ed.) Pidgin and Creole Linguistics. Bloomington and London: Indiana University Press, pp. 129–54. McConvell, Patrick (2001) Mix-im-up speech and emergent mixed languages in indigenous Australia. Texas Linguistic Forum 44, 328–49. Mesthrie, Rajend (1989) The origins of Fanagalo. Journal of Pidgin and Creole Languages 4, 211–40. Mesthrie, Rajend (1993) Koineization in the Bhojpuri-Hindi diaspora with special reference to South Africa. In: Siegel (ed.), pp. 25–44. Miyawaki, Hiroyuki (2002) Colonial language policies and their effects. www.linguapax.org/congres/taller/taller1/miyawaki.html (accessed April 5, 2008). Moore, Denny, Sidney Facundes, and Nádia Pires (1994) Nheengatu (Língua Geral Amazônica), its history, and the effects of language contact. In: Margaret Langdon (ed.) Proceedings of the Meeting of the Society for the Study of the Indigenous Languages. Berkeley: University of California, pp. 93–118. Mosel, Ulrike (1980) Tolai and Tok Pisin: The Influence of the Substratum on the Development of New Guinea Pidgin. Canberra: Australian National University. Mous, Maarten (2003) The Making of a Mixed Language: The Case of Ma’a/Mbugu. Amsterdam: John Benjamins. Mufwene, Salikoko S. (1997) Kituba. In: Thomason (ed.), pp. 173–208. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mühlhäusler, Peter (1986) Pidgin and Creole Linguistics. Oxford: Blackwell. Mühlhäusler, Peter and Rob Amery (1996). Koines and indigenous lingue franche in Australia. In: Wurm et al. (eds.), pp. 17–23. Mühlhäusler, Peter and Rachel Trew (1996) Japanese language in the Pacific. In: Wurm et al. (eds.), pp. 373–99. Muysken, Pieter (1977) Syntactic Developments in the Verb Phrase of Ecuadorian Quechua. Dordrecht: Foris. Muysken, Pieter (1997) Media Lengua. In: Thomason (ed.), pp. 365–426. Myers-Scotton, Carol (1979) The context is the message: Morphological, syntactic and semantic reduction and deletion in Nairobi and Kampala varieties of Swahili. In: Hancock (ed.), pp. 111–27. Naro, Anthony J. (1978) A study on the origins of pidginization. Language 54, 314–47. Neumann, Günter (1966) Zur chinesisch–russischen Behelfssprache von Kjachta [The Chinese–Russian jargon of Kjachta]. Die Sprache 12, 337–51.

184

Kees Versteegh

Noss, Philip (1979) Fula: A language of change. In: Hancock (ed.) pp. 175–88. Nurse, Derek (1997) Prior pidginization and creolization in Swahili? In: Thomason (ed.), pp. 271–94. Nurse, Derek and Thomas Spear (1985) The Swahili: Reconstructing the History and Language of an African Society 800–1500. Philadelphia: University of Pennsylvania Press. Owens, Jonathan (1985) The origins of East African Nubi. Anthropological Linguistics 27, 229–71. Owens, Jonathan (1990) East African Nubi: Bioprogram vs. inheritance. Diachronica 7, 217–50. Owens, Jonathan (1997) Arabic-based pidgins and creoles. In: Thomason (ed.) pp. 125–72. Partmann, Gayle (1979) Some further comments on Urban Dioula. In: Hancock (ed.) pp. 107–9. Pasch, Helma (1997) Sango. In: Thomason (ed.), pp. 209–70. Reuse, Willem J. de (1996) Chukchi, English, and Eskimo: A survey of jargons in the Chukotka Peninsula area. In: Jahr and Broch (eds.), pp. 47–62. Roberts, Julian M. (1995) Pidgin Hawaiian. Journal of Pidgin and Creole Linguistics 10, 1–56. Rodrigues, Aryon D. (1996) As linguas gerais sul-americanas [The “general” languages of Latin America]. Papia, Revista de Crioulos de base Ibérica (Brasilia) 4(2), 6–18. Available at: www.unb.br/il/lablind/lingerais.htm, accessed April 5, 2008. Rottland, Franz (1979) Free variation in the concord system of written Lingala. In: Hancock (ed.), pp. 165–71. Samarin, William J. (1982) Colonization and pidginization on the Ubangi River. Journal of African Languages and Linguistics 4, 1–42. Samarin, William J. (1986) Chinook Jargon. Canadian Journal of Anthropology 5, 23–34. Samarin, William J. (1996) Arctic origin and domestic development of Chinook Jargon. In: Jahr and Broch (eds.), pp. 321–39. Samarin, William J. and James A. Walker (1994) Verb-marking in Sango predicate chains. In: Helma Pasch (ed.) Sango: The National Official Language of the Central African Republic. Cologne: R. Rüdiger Köppe Verlag, pp. 112–28. Sebba, Mark (1997) Contact Languages: Pidgins and Creoles. Houndmills and London: Macmillan Press. Shapiro, Michael C. and Harold F. Schiffman (1983) Language and Society in South Asia. Dordrecht: Foris. Siegel, Jeff (1987) Language Contact in a Plantation Environment: A Sociolinguistic History of Fiji. Cambridge: Cambridge University Press. Siegel, Jeff (1993a) Introduction: Controversies in the study of koines and koineization. In: Siegel (ed.), pp. 5–8. Siegel, Jeff (ed.) (1993b) Koines and Koineization (International Journal of the Sociology of Language, 99). Berlin and New York: Mouton de Gruyter. Silverstein, Michael (1972) Chinook Jargon: Language contact and the problem of multilevel generative systems. Language 48, 378–406, 596–625. Simeone-Senelle, Marie-Claude (2008) Lingua Franca: Horn of Africa. In: Mushira Eid, Alaa Elgibali, Manfred Woidich, Kees Versteegh, and Andrzej Zaborski (eds.) Encyclopedia of Arabic Language and Linguistics, vol. 2, pp. 268–75. Leiden: Brill. Simpson, Jane (1996) Early language contact varieties in South Australia. Australian Journal of Linguistics 16, 169–207.

Non-Indo-European Pidgins and Creoles 185 Singleton, Jenny, Jill P. Morford, and Susan Goldin-Meadow (1993) Once is not enough: Standards of well-formedness in manual communication created over three different timespans. Language 69, 683–715. Smart, J. R. (1990) Pidginization in Gulf Arabic: A first report. In: Anthropological Linguistics 32, 83–119. Southworth, Franklin C. (1971) Detecting prior creolization: An analysis of the historical origins of Marathi. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 255–76. Spitulnik, Debra (1999) The language of the city: Town Bemba as urban hybridity. Journal of Linguistic Anthropology 8, 30–59. Sreedhar, M. V. (1974) Naga Pidgin: A Sociolinguistic Study of Inter-lingual Communication Pattern in Nagaland. Mysore: Central Institute of Indian Languages. Stefánsson, Vilhjalmur (1909) The Eskimo trade jargon of Herschel Island. American Anthropologist 11, 217–32. Steinhauer, Hein (1980) On the history of Indonesian. In: A. A. Barentsen, B. M. Groen, and R. Sprenger (eds.) Studies in Slavic and General Linguistics, vol. 1. Amsterdam: Rodopi, pp. 349–75. Taylor, Douglas MacRae (1951) The Black Carib of British Honduras. New York: WennerGren Foundation. Thomas, Edward Harper (1954) Chinook: A History and Dictionary of the Northwest Coast Trade Jargon. 2nd edn. Portland, OR: Binfords and Mort. Reprinted 1970. Thomason, Sarah Grey (1983) Chinook Jargon in areal and historic context. Language 59, 820–70. Thomason, Sarah G. (ed.) (1997) Contact Languages: A Wider Perspective. Amsterdam: John Benjamins. Thomason, Sarah G. and Alaa Elgibali (1986) Before the lingua franca: Pidginized Arabic in the eleventh century AD. Lingua 68, 317–49. Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics. Berkeley and Los Angeles: University of California Press. Tosco, Mauro and Jonathan Owens (1993) Turku: A descriptive and comparative study. Sprache und Geschichte in Afrika 14, 177–267. Van der Voort, Hein (1995) Eskimo Pidgin. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 137–51. Van der Voort, Hein (1996a) History of Eskimo interethnic contact and its linguistic consequences. In: Wurm et al. (eds.), pp. 1043–94. Van der Voort, Hein (1996b) Eskimo pidgin in West Greenland. In: Jahr and Broch (eds.), pp. 157–258. Van Minde, Don (1997) Malayu Ambong: Phonology, Morphology, Syntax. Leiden: CNWS, University of Leiden. Versteegh, Kees (1984) Pidginization and Creolization: The Case of Arabic. Amsterdam: John Benjamins. Versteegh, Kees (1993a) Esperanto as a first language: Language acquisition with a restricted input. Linguistics 31, 539–55. Versteegh, Kees (1993b) Leveling in the Southern Sudan: From Arabic creole to Arabic dialects. International Journal of the Sociology of Language 99, 65–97. Wald, Benji (1981) Swahili Pre-Pidgin, Pidgin, and depidginization in coastal Kenya: A systematic discontinuity in non-first varieties of Swahili. In: Arnold Highfield and Albert Valdman (eds.) Historicity and Variation in Creole Studies. Ann Arbor: Karoma, pp. 7–26.

186

Kees Versteegh

Wellens, Ineke (2005) The Nubi language of Uganda: An Arabic Creole in Africa. Leiden: Brill. Whinnom, Keith (1971) Linguistic hybridization and the “special” case of pidgins and creoles. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 255–76. Winford, Donald (1997) Creole formation in the context of contact linguistics. Journal of Pidgin and Creole Languages 12, 131–51. Wurm, Stephen A. (1992) Some contact languages and pidgin and creole languages in the Siberian region. Language Sciences 14, 249–85. Wurm, Stephen A. (1995) The Silk Road and hybridized languages in North-Western China. Diogenes 43/3, 53–62. Wurm, Stephen A. (1996) The Taymyr Peninsula Russian-based pidgin. In: Jahr and Broch (eds.), pp. 79–90. Wurm, Stephen A., Peter Mühlhäusler, and Darrell T. Tryon (eds.) (1996) Atlas of Languages of Intercultural Communication in the Pacific, Asia, and the Americas. Berlin: Mouton de Gruyter.

FURTHER READING Arends, Jacques, Pieter Muysken, and Norval Smith (1995) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins. Contains an annotated list of pidginized/ creolized varieties all over the world classified according to lexifier and brief descriptions of a number of non-Indo-European pidgins and creoles: Eskimo Pidgin, Shaba Swahili. Mithun, Marianne (1999) The Languages of Native North America. Cambridge: Cambridge University Press. Contains short descriptions of Chinook Jargon, Delaware Jargon, Inuktitut Trade Languages, Mednyj Aleut, Michif, and Mobilian Jargon, pp. 587– 605. Sebba, Mark (1997) Contact Languages: Pidgins and Creoles. Houndmills and London: Macmillan Press. Textbook on general phenomena of linguistic contact; contains a brief description of Fanagalo. Thomason, Sarah G. (ed.) (1997) Contact Languages: A Wider Perspective. Amsterdam: John Benjamins. Contains descriptions of the following non-Indo-European pidgins and creoles: Hiri Motu, Pidgin Delaware, Ndyuka-Trio Pidgin, Arabic-based pidgins and creoles, Kituba, Sango, Pidgin Swahili, as well as descriptions of the following mixed languages: Michif, Media Lengua, Callahuaya, Mednyj Aleut, and Ma’a (Mbugu). Thomason, Sarah G. (2001) Language Contact: An Introduction. Washington, DC: Georgetown University Press. Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics. Berkeley and Los Angeles: University of California Press. Analysis of language change in the context of linguistic contacts, with case-studies on, among others, Ma’a, Michif, Mednyj Aleut, Chinook Jargon. Wurm, Stephen A., Peter Mühlhäusler, and Darrell T. Tryon (eds.) (1996) Atlas of Languages of Intercultural Communication in the Pacific, Asia, and the Americas. Berlin: Mouton de Gruyter. Contains one volume of maps of all languages of wider communication in this area, and two volumes with articles about all documented cases of such languages.

Pidgins/Creoles and Second Language Acquisition 187

Part II Perspectives on Pidgin/Creole Genesis

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Pidgins/Creoles and Second Language Acquisition 189

8

Pidgins/Creoles and Second Language Acquisition JEFF SIEGEL

1 Introduction The role of adult second language acquisition in the genesis of pidgin and creole languages has been discussed since the late nineteenth century. Earlier debate centered around whether simplification, one of the “universal processes” of second language acquisition (SLA), is responsible for the formal simplicity found in pidgins and creoles in comparison to the languages which came into contact to produce them. Later debate has also focused on whether another process of SLA, transfer, can account for the presence of various linguistic features of the substrate languages in pidgins and creoles. This chapter covers the issues concerning the role of these SLA processes in pidgin and creole (P/C) genesis – simplification in section 2, transfer in section 3. For each process, some historical background is presented, followed by a discussion of recent developments in P/C studies, especially those that have built upon research in the field of SLA. The chapter ends with a section describing additional approaches and various current controversies regarding SLA in P/C genesis (section 4).1

2 Simplification 2.1

Background

Despite the wide use of the term “simplification,” there is little agreement on its precise meaning. In P/C studies it usually refers to comparative formal simplicity – that is, simplicity in the quantitative rather than psycholinguistic sense. In other words, one variety is considered simpler or less complex than another if it has fewer components and fewer rules, but it is not necessarily easier to process. Thus, comparing one variety to another (or comparing subsets of features of one variety to another), simplification includes characteristics

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

190

Jeff Siegel

such as a smaller lexicon, a reduced phonemic inventory, fewer grammatical categories, less bound morphology, and fewer exceptions to rules. According to this definition, nearly all creolists would agree that pidgins and creoles are characterized by simplification in at least some aspects of phonology, morphosyntax, or lexicon in comparison to their superstrate (lexifier) language or their substrate languages. In an attempt to explain this simplification in P/Cs, Adolpho Coelho first proposed in the 1880s what is now known as the “imperfect learning theory” (Siegel 1987, pp. 19–20) or “imperfect second language learning theory” (Muysken & Smith 1995, p. 10). According to this theory, these simplified features are the result of incomplete or imperfect acquisition of the lexifier language, due to insufficient access to the target language, i.e., the lexifier. Similarities among various pidgins and creoles can be explained by the universal properties of simplification that occur at the early stages of the second language learning process. The theory, which was subscribed to by later linguists, such as Jesperson (1922) and Bloomfield (1933), and by early modern creolists, such as Hall (1966) and Samarin (1971), was in opposition to the view that simplification arises from speakers of the lexifier using simplified versions of their own speech (or “foreigner talk”) in talking to non-native speakers. While the field of P/C studies expanded rapidly in the late 1960s and early 1970s, the study of SLA was just starting to emerge as a separate subfield of linguistics. Thus, creolists such as Bickerton (1977) could assert that “the whole area of natural second language acquisition – a process that affects countless millions of persons – remains a vast and grotesque lacuna in our knowledge of language” (p. 54). Ferguson and DeBose (1977) also noted that there were “few studies of adult ‘natural’ acquisition” (p. 108), but they drew on the recent SLA research into the nature of learners’ grammars, i.e., transient internal representations of the structure of the target language called “approximative systems” (Nemser 1971) or “interlanguage” (Selinker 1972). Ferguson and Debose called learners’ actual productions of the target language – characterized by relative formal simplicity, a carryover of features of the first language, and great variability among learners – “broken language” (1977, pp. 108–9). They also broadened the definition of pidginization, describing it as “a process that accepts normal language as input and produces a reduced, hybridized, and unstable variety of language as output” (p. 111). It was in this atmosphere that some scholars in the nascent field of SLA looked to the field of P/C studies for explanations of the processes involved in learning a second language (L2). From 1974 to 1976, Schumann published several papers equating the processes of early SLA and pidginization. His book, The Pidginization Process: A Model for Second Language Acquisition, appeared in 1978. The basis for this model was a longitudinal study of six native speakers of Spanish learning English outside the classroom setting. One of the learners, Alberto, remained in the early stages of development with regard to the linguistic features being studied. His productions of English resembled pidgins in terms of simplification, and specifically in particular features found

Pidgins/Creoles and Second Language Acquisition 191 in English-derived pidgins – Schumann (1978b, pp. 7–5) considered data from West African Pidgin English, Neo-Melanesian (i.e., Melanesian Pidgin), an English-Japanese pidgin that appeared in Hamamatsu in the mid-1950s, American Indian Pidgin English, and an English pidgin spoken by migrant industrial workers in Australia. The relevant features include (Schumann 1978b, p. 66): 1 2 3 4 5

negatives formed by no preceding the verb; absence of inversion in questions; no auxiliaries; unmarked possessives; absence of -ed past tense marking.

Schumann’s claim was not that Alberto spoke a pidgin, but that this simplification was evidence of pidginization. Schumann surmised “that pidginization may characterize all early second language acquisition in general and that under conditions of social and psychological distance it persists” (p. 110). He also suggested, following Kay and Sankoff (1974), that the pidginized form of speech found in early SLA and pidgins may approximate the universal underlying structure of language. Schumann’s model was criticized by scholars in both P/C studies and SLA. Washabaugh and Eckman (1980) noted, among other criticisms, that creolists do not agree that the definition of pidginization necessarily includes simplification, or that a definable process of pidginization even exists! Corder (1981) took the point of view that pidginization is a linguistic process which results in the formation of a pidgin language, not a psycholinguistic process involving simplification of the superstrate, because people cannot simplify a language that they do not know. In contrast to Corder, however, Andersen (1980, 1981, 1983a) distinguished between individual pidginization, a psycholinguistic process which results in individual L2 varieties, and group (or social) pidginization, a sociolinguistic process involving communication among speakers undergoing or having undergone individual pidginization, which results in a pidgin language with its own norms. In a 1979 study, published as Andersen (1981), he made detailed comparison of the speech of Alberto and the speech of speakers of Hawai[i Pidgin English, and found features very similar to those listed above. On the basis of this study, he concluded, in agreement with Schumann and Bickerton, that “SLA and individual pidginization are really the same phenomenon viewed from different perspectives and often, although not always, occurring under different circumstances” (Andersen 1980, p. 274). There was general agreement on this point at a symposium held in 1979 on the Relationship between Pidginization, Creolization and Language Acquisition (Andersen 1983b), and soon after, Valdman (1980, p. 297) noted: “The notion that the incipient IL [interlanguage] is the product of pidginization has gained wide acceptance.” Unlike previous scholars, Andersen also addressed the issue of the specific relationship between SLA and creolization, not just pidginization. He again

192

Jeff Siegel

made a distinction between two types of creolization. Social creolization refers to the process that results in a creole language, generally thought to occur when a pidgin or pre-pidgin (i.e., a continuum of L2 varieties) becomes the mother tongue of a community of speakers. Individual creolization is defined as individual pidginization plus “the creation of form-meaning relationships which serve the creator’s (learner’s) communicative and expressive needs, but which cannot be explained as having been ‘acquired’ from the input” (Andersen 1983a, p. 9). He went on to say that expansion and complication in individual creolization occurs when the use of the original pidginized variety is extended (for example, as a language of wider communication) and when there is immediate pressure to communicate without acquired competence (p. 30). Andersen believed that both children and adults have this capacity, and by implication that certain processes of SLA may be involved (cf. 1983a, pp. 9–10). It was with regard to the expansion and complication in creolization that Andersen and Bickerton parted company. Bickerton believes that only children have the innate capacity to create linguistic systems without adequate models – thus his well-known distinction “that pidginization is second-language learning with restricted input, and that creolization is first-language learning with restricted input” (1977, p. 49). Bickerton’s view (1981, 1984a, 1988, 1999a, 1999b) is that creoles are created by children learning their first language (L1) and drawing upon their innate “bioprogram” in the absence of any fully developed linguistic model. This abrupt creolization occurs “in the space of a single generation” (1999b, p. 49), the first generation of locally born children of plantation slaves or indentured laborers. Thus, according to Bickerton, creole genesis is the result of child first language acquisition, not adult second language acquisition. Bickerton’s Language Bioprogram Hypothesis was highly controversial and certainly not universally accepted. But in part because of its impact, the main thrust of research in P/C studies in the later 1980s and early 1990s shifted away from pidgin genesis to concentrate on creoles. The focus was on the “core features” of creoles which demonstrate expansion and complication rather than on other common creole features which demonstrate earlier simplification. Thus, as Arends, Muysken, and Smith (1995, p. 320) observed: “Earlier ideas regarding possible parallels between creolization and pidginization on the one hand and natural second language on the other . . . have not been followed up by many creolists.”

2.2

Recent developments

In the 1990s, few creolists remained convinced by Bickerton’s point of view. First of all, it could not explain the clear presence of features from substrate languages in many creoles. Second, historical findings contradicted his scenario for the development of creoles, suggesting that many of them had developed gradually over several generations (see Singler 1990; Arends 1993; and various

Pidgins/Creoles and Second Language Acquisition 193 chapters in Arends 1995).2 According to the “gradualist model” which emerged from these findings, the rate of nativization was slow and speakers of the emerging creole were mainly adults. This led to a renewed interest in the role of adult SLA in creole genesis. For example, Wekker (1996, p. 146) views creolization as “a gradual process of imperfect second-language acquisition by successive cohorts of adult slaves, extending over generations.” Meanwhile, research in SLA has increased by leaps and bounds since the early 1980s, and today SLA is a much larger field than P/C studies. So while in the past the field of SLA looked toward P/C studies for data and insights, it would seem profitable now for P/C studies to look toward SLA. However, with the exception of scholars such as Mufwene (1990), Becker and Veenstra (2003a), DeGraff (1999), Clements (2001, 2003), Helms-Park (2003), Winford (2002), Field (2004), and others mentioned below, few people writing about pidgins and creoles have delved into the recent SLA literature in any detail. In a recent book about language contact, however, Winford (2003) includes a whole chapter on SLA and weaves recent SLA findings into explanations in later chapters about the genesis of pidgins and creoles.3 Several areas of SLA research have led to findings that are directly relevant to issues surrounding simplification in P/C genesis. First, detailed descriptions of the interlanguage (IL) of L2 learners have provided more evidence that the same processes operating in SLA may be responsible for features of pidgins and creoles. Second, further investigations into the role of socialpsychological factors in SLA help to explain why, if SLA was involved in their genesis, P/Cs did not continue along the developmental continuum to become more like their lexifiers, as in “normal” SLA. I will discuss each of these in turn. In addition, studies of communication strategies in L2 use have shown how adult learners can compensate for inadequate L2 knowledge by expanding their existing linguistic system, as occurs in creolization. In this regard, the distinction between L2 acquisition and L2 use has emerged as central. I will return to this in section 3.

2.2.1

Descriptions of IL and P/C features

The last decade has seen numerous detailed analyses of the IL of learners of various European languages. These have led to many more studies along the lines of Andersen (1981). Véronique (1994) describes many formal similarities between features found in the early interlanguages of first-language Moroccan Arabic learners of French as a second language and what are considered “simplified” features of Frenchlexified creoles. These include the following (p. 133): 1 2 3 4

lack of formal distinction between word classes, especially between nouns and verbs; NP and VP morphology largely lacking (compared to French); some nouns used without articles; /i/ or /li/ as a preposed predicate marker;

194 5 6

Jeff Siegel verbs developing two forms: the bare stem and stem + /e/; extended use of “presentationals” /jãna/ and /se/ ‘there is’.

He concludes that the similarities illustrate early attempts of naturalistic learners to understand and use target language input, to “crack the code” (p. 133). In an examination of the IL of West African L1 Ewe learners of French, Mather (2000) found some of these and other features similar to those of Frenchlexified creoles, such as the lack of copula, which he concludes were the result of the process of simplification in SLA. Using data from several studies on Dutch SLA, Muysken (2001) compares learners’ IL features with features of the now extinct Dutch-lexifier creole Negerhollands. He reports many similarities in formal simplicity, including rigid SVO word order, the absence of postpositions, preverbal negation, the absence of inflections on verbs, and periphrastic possessives of the type Mary her/his book. Muysken concludes (p. 169): “A number of features of Negerhollands may well be explicable as deriving from the acquisition of Dutch as a second language.” Kotsinas (1996) illustrates numerous similarities between the features of the IL of Swedish immigrants where Swedish is the L2, and those of Russenorsk, a pidgin which has closely related Norwegian as a lexifier – for example, extended use of the preposition på. In a later work (Kotsinas 2001) shows similarities between the features of L2 versions of Swedish and those typical of pidgin languages. A good general indication of IL features came out of one of the largest studies ever done of naturalistic adult SLA: the European Science Foundation (ESF) project, which took place in the 1980s. This was a longitudinal study of 40 adult immigrants with various first languages: Arabic, Italian, Finnish, Italian, Spanish, and Turkish. The target languages were Dutch, English, French, German, and Swedish (Perdue 1993). In an article about the results of the study, Klein and Perdue (1997) describe that approximately one third of the learners acquired only what they call “the Basic Variety,” characterized by numerous structural features which can be considered simplified because they are reduced, regularized, or more semantically transparent in comparison to features of the target language: 1 no inflections; 2 lexical items used in invariant form (multifunctionality); 3 invariant forms generally infinitive or nominative (but also some inflected forms); 4 lexical items noun-like and verb-like words with some adjectives and adverbs; 5 most lexical items from the L2 but some from the L1 and other languages; 6 minimal pronouns to refer to speaker, hearer, and a third person; 7 no anaphoric pronouns referring to inanimates; 8 only a few quantifiers;

Pidgins/Creoles and Second Language Acquisition 195 9 10 11 12 13

a single word for negation; only a few prepositions; no complementizers; no expletive elements (e.g., there is); use of temporal adverbs, rather than grammatical TMA markers, to indicate temporality; 14 specification of temporal relations: before, after, simultaneous, etc.; 15 “boundary markers” to express the beginning or end of some situation, such as work finish ‘after work is/was/will be over’ (p. 321).

Significantly, the majority of these features are found in most restricted pidgins including Pidgin Fijian and Pidgin Hindustani (Siegel 1987), Chinese Pidgin English (Baker 1987), Greenlandic Pidgin (Van der Voort 1996), the Hiri trading languages Eleman and Koriki (Dutton 1997), Nauru Pidgin English (Siegel 1990), Ndyuka-Trio Pidgin (Huttar & Velantie 1997), Pidgin Delaware (Goddard 1997), Pidgin Hawaiian (Roberts 1995), Pidgin French of Vietnam (Reinecke 1971), and Russenorsk (Jahr 1996). These similarities provide evidence that a large proportion of the “simplified” structural features of pidgins may be a consequence of SLA at an earlier stage of development. Some of these simplified features could also have been inherited by creoles in a later stage of development (as will be discussed in 3.2.1).

2.2.2

Social-psychological factors and “imperfect” SLA

If SLA was involved in the genesis of pidgins and creoles, we need to explain why acquisition did not progress further. In other words, why do P/C grammars remain so distinct from those of their lexifier? Baker’s view (1994, 1997, 2000) is that SLA is not relevant to P/C genesis. He believes that people of different ethnolinguistic backgrounds in contact situations are generally not interested in acquiring the language of the other groups but rather in constructing a new “medium for interethnic communication.” However, a lot of evidence exists, such as that presented above, to support the involvement of SLA (see also Mufwene 2001; Siegel 2003b), and the SLA literature itself contains a great deal of work on factors which lead to incomplete language acquisition. Early social psychological models, such as Schumann’s (1978a) “acculturation model,” considered social and psychological distance to be important factors (as mentioned in 2.1). Other models used individual factors, such as motivation and social identity as well as socio-structural factors such as relative size, status, and power of the L1 and L2 groups. (For overviews, see Ellis 1994; Siegel 2003a.) However, more recent work has criticized these models for implying that learners are free to make choices about when they interact with L2 speakers or whether they are motivated to integrate with the L2 culture, and thus blaming the learner for lack of L2 attainment. Ethnographic studies in SLA such as those by Norton Peirce (1995), McKay and Wong (1996), and Norton (2000) take into account the sociohistorical factors of power and domination which limit the choices available to learners, and also adopt the poststructuralist

196

Jeff Siegel

view that people have multiple and changing social identities, rather than the unitary static social identity of most social psychological models. Another more recent perspective in the SLA literature considers that a variety which differs from that of an idealized native speaker does not necessarily represent its speakers’ failure in attaining L2 competence. Rampton (1991, 1997), Firth and Wagner (1997), and Rampton (1991) observe that an “imperfect” variety may be used to express a particular identity of the speaker, to show solidarity with a peer group, to indicate attitudes toward society in general, even as a form of resistance. For example, stylized South Asian English is used by the adolescents in England studied by B. Rampton (1995), not because of any lack of proficiency in local varieties of English but for joking and ridiculing racist attitudes. It follows, then, that in many situations native-like proficiency is not the target of language learning. Both these recent perspectives are relevant to P/C genesis in providing explanations other than “lack of success” or “failure” in acquiring the lexifier that have been justifiably criticized (cf. Baker 1994; Sebba 1997). Rather than “imperfect SLA,” a term such as “strategic SLA” would be more appropriate.

3

Transfer

3.1

Background

The term “transfer” refers to a form of cross-linguistic influence found in SLA that involves “carrying over of mother tongue patterns into the target language” (Sharwood Smith 1996, p. 71), or more accurately, into the interlanguage. In other words, learners use linguistic features of their L1 – phonemes, grammatical rules, or meanings or functions of particular words – when speaking the L2. Such transfer can be either positive (when a feature in the L1 matches that of the L2) or negative (when corresponding features of the L1 and L2 do not match). Negative transfer, which we will concentrate on here, is sometimes called “interference” and has been given other labels as well (see below). We will focus on morphosyntactic and semantic transfer. Like the role of simplification, the degree of influence of features of the substrate languages in pidgins and creoles has been debated since the earliest days of P/C studies. One of the first scholars to ascribe this influence to language learning was Hesseling in 1933: [The African slaves] learn the surface structure of the European languages, although they make them suitable for their own manner of thinking . . . The masters hear their own words, however truncated or misshapen, while the slaves employ the foreign material in a way which is not in complete conflict with their inherited manner of expressing themselves. (Hesseling [1933] 1979, p. 69)

More recently, Alleyne (1971, p. 182) wrote along the same lines that “in attempting to speak English or French, Africans in Africa, as well as in the New

Pidgins/Creoles and Second Language Acquisition 197 World, interpreted English or French structural patterns in terms of native patterns.” However, it wasn’t until later that the notion that substrate features entered pidgins and creoles via the process of transfer in SLA was considered. On the basis of his work in Hawai[i, Bickerton (1977, p. 54) saw pidginization as a process of “relexification” where the substrate grammar is maintained but L1 lexicon is gradually replaced by superstrate words, which are “rephonologized to accord with the substrate sound system and phonotactics.” More superstrate lexicon is later acquired and for the most part “slotted into syntactic structures drawn from the substrate.” Bickerton made a clear connection between this process and early SLA, saying that “second languages are naturally acquired via piecemeal relexification, productive calquing, and the utilization of mother tongue surface structure . . . in the early stages at least” (pp. 54–5). The difference between pidginization and normal SLA lies in “the availability of target models and the amount of interaction with speakers of the target language” (p. 55). While Bickerton said that relexification may be “complete down to grammatical items” (p. 54), the examples in Bickerton (1977) and in Bickerton and Odo (1976) show mainly use of L1 phonology and word order in speaking the L2. This is what would be described as phonological and syntactic transfer in the SLA literature. Later, Bickerton (1984b, p. 152) referred explicitly to the process as learners “transferring rules of grammar from the grammar they already know to the grammar they are seeking to acquire.” In contrast to Bickerton, however, Naro (1978) claimed there was no substrate influence in Pidgin Portuguese, and therefore presumably no transfer in the pidginization process (but see Goodman 1987). Similarly, Manessy (1977) reported no substrate influence in African pidgins. Furthermore, in a work that had great influence on the field of P/C studies, Meisel (1983) argued that universal cognitive strategies of simplification rather than transfer are the most significant determinants of the features in interlanguage, and therefore transfer is not a significant factor in P/C genesis. This work reflected the then prevailing attitudes in the field of SLA regarding the role of transfer in interlanguage development. Research in the 1970s (e.g., Dulay & Burt 1973) had shown that negative transfer accounted for only a small proportion of non-target forms in interlanguage, and that there were natural sequences of morphological and syntactic development that were unconnected with the L1. As a result of these findings and others, SLA research had begun to pay very little attention to transfer and concentrate instead on the role of universal processes of language acquisition. With regard to creoles, Andersen (1983a, p. 31) believed that the substrate languages could be a possible source for structures in the expansion and complication involved in creole genesis: In the absence of clear models in the input for a linguistic form needed to consistently convey a meaning the learner wants to communicate, the learner can (probably far below the level of consciousness) use a form (or forms) available in

198

Jeff Siegel

the input to convey the meaning and/or fulfill the function of an equivalent form in his native language . . . with the distribution that form has in his native language.

But as already mentioned, Bickerton (1977, 1981, 1984a) argued that, while second language acquisition was involved in pidgin genesis, only first language acquisition was involved in creole genesis and, therefore, transfer in SLA could not be a factor in the creation of a creole. So in the 1980s and early 1990s, the combination of the earlier move away from research on transfer in the field of SLA and the impact of Bickerton’s paradigm meant that interest in the role of transfer in P/C genesis was diminished. In contrast, the field of SLA once again became interested in research on language transfer during this period, but with a new concentration on identifying conditions or constraints that determine whether or not transfer occurs (Ellis 1994, p. 315). These constraints are not absolute and may either promote or inhibit transfer. Two types of constraints were investigated. The first are situational constraints, such as the learner’s degree of proficiency and the learning context (i.e., naturalistic vs. classroom); these affect whether or not transfer occurs in general (see Siegel 1999 for details). The second are linguistic constraints, such as markedness; these affect whether or not particular linguistic features are transferred. We will explore the linguistic constraints in 3.2.1.

3.2

Recent developments

The first clear statement about the connection between the process of transfer in SLA and substrate influence in pidgins and creoles was made by Mufwene (1990, p. 2): Transfers apply putatively in the speech of multilingual speakers and/or at the stage of SLA; substrate influence is observed in a language as a relatively crystallized system. Once transfers have been replicated by different speakers, repeated by most of them, and established in the contact situation’s new linguistic system (even as variable features), they may be characterized genetically as substrate influence. The latter need not be associated synchronically with multilingual speakers and/or SLA.

By the mid 1990s, more creolists had started once again to consider the role of transfer in P/C genesis, especially grammatical and semantic transfer. For example, Wekker (1996, p. 144) describes the process of creolization as “one of imperfect second-language acquisition, predominantly by adults, involving the usual language transfer from the learners’ L1.” Winford (2000, p. 216) describes the parallels between the formation of P/Cs and the processes of SLA, which include not only universal strategies such as simplification, but also “L1 strategies,” which he calls “L1 retention” rather than transfer. (See also Migge 1998, 2000; Winford 2002, 2003.) Other scholars have also used different labels for basically the same processes. In his work on Melanesian Pidgin (MP), Keesing (1988) uses the term “calquing,” implying exact

Pidgins/Creoles and Second Language Acquisition 199 correspondence between rules in the substrate and MP (which is not really the case). Writing about Haitian Creole, Lefebvre (e.g., Lefebvre 1986, 1996, 1997) and Lumsden (e.g., 1996) use the term “relexification.” However, in recent work, they have shown its relationship to L1 transfer (Lefebvre 1998a; Lumsden 1999).4 DeGraff (2001b, p. 250) mentions the “retention and reanalysis of sourcelanguage structures in pidginization (and creolization).” Finally, Thomason (2001, p. 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 grammatical transfer.

3.2.1

Transfer mechanisms

Bickerton (1984a, p. 183) challenged substratists “to provide plausible mechanisms by which rules could have been passed from substratum to creole speakers.” He discounted the two most obvious possibilities: that “rules entered the [preceding] pidgin and were learned by creole speakers or early creole speakers were bilinguals with fluent control of the substrate languages.” In response, I proposed a “plausible mechanism” to explain how substrate features get into creoles (Siegel 1997). Briefly, in a language contact situation, one particular language may become the language of wider communication, and thus the target for learning. However, many different varieties of this language are actually used for communication, including regional dialects and foreigner talk spoken by superstrate speakers, and various interlanguage or L2 versions of the different superstrate models spoken by the substrate speakers. An existing pidgin with the same lexifier may also be used, as well as L2 versions of this pidgin. All these versions of the superstrate make up the “pool of variants” (Siegel 1997, p. 136) which are available for use by the community, and as potential models for language acquisition; the L2 versions include some features transferred from the speakers’ L1s. When the social conditions are right for the emergence of a stable pidgin or creole – for example, when speakers see themselves as a distinct community – then leveling (or koineization) occurs. In this process, some of the variants are eliminated, while others are retained and incorporated into the grammar of the new contact language. This may occur gradually as some features drop out from lack of use, or more rapidly as children acquire some features but not others. The point is that some of the retained variants may be those that originally resulted from transfer. Thus, in P/C genesis, substrate features enter via individual L2 versions – some of which are later adopted by the community or acquired by children – and not directly from the substrate languages themselves. One of the criticisms of the substrate hypothesis has been that no constraints or principles have been identified to explain why some substrate features are retained in various pidgins and creoles and others are not (e.g., Mufwene 1990, p. 6). In reaction to this criticism, and in line with the leveling model just described, I proposed that there must be both “availability constraints” that

200

Jeff Siegel

determine which substrate features actually reach the target pool of variants via transfer, and “reinforcement principles” that determine which of these transferred features are reinforced and therefore retained in the leveling process. In order to identify specific availability constraints and reinforcement principles affecting pidgins and creoles, I examined the research on various constraints on transfer proposed in the SLA literature, and tested the validity of these constraints using data from Melanesian Pidgin (Siegel 1999). I examined 11 core morphosyntactic features of the Central Eastern Oceanic (CEO) languages that are the substrate for MP. Seven of these features are found in MP, while four are not. I identified various factors which researchers in SLA have postulated as influencing which linguistic features are more likely to be transferred by individuals in SLA and related these to similar factors which researchers in P/C studies have proposed to explain which features are more likely to be “selected” or retained by the community in later stages. For the purpose of the study, six factors which have to do with the linguistic properties of both the superstrate and the substrate languages were isolated: markedness, perceptual salience, transparency, simplicity, frequency, and congruence. Each of these was examined to see whether it could account for transfer and reinforcement of some of the CEO core features but not the others. Furthermore, I found that the situational constraints which promote transfer in SLA are generally characteristic of the situations where pidgins such as MP emerge. These include a naturalistic and unfocused learning context, an immediate need to communicate, and use of the L2 mainly with social equals. With regard to the linguistic constraints, the most important factors appear to be perceptual salience and congruence. First, for a substrate feature to be transferred, it must have “somewhere to transfer to” (Andersen 1983c) – i.e., there must be a morpheme (or string of morphemes) in the superstrate that can be used or reanalyzed according to the rules of the substrate. This superstrate morpheme or string must be perceptually salient – a separate word, or words, or at least a stressed syllable – and it must have a function or meaning related to that of the corresponding substrate morpheme. Second, especially in situations where the emerging creole remains in contact with its lexifier, the substrate and superstrate morphemes should be morphosyntactically congruent, at least superficially. The absence of such a morpheme in the superstrate or the lack of structural congruence will constrain transfer, and thus the availability of the particular substrate feature. Another factor which may also play a part is transparency: transfer will be more likely if the superstrate morpheme is invariant in form and has a single function. For example, one of the 10 “minimal grammatical functions” supposedly found in all creoles (Bickerton 1988) is absent in creolized MP: a reciprocal marker. My explanation is that there is no preverbal form in the lexifier, English, that could be reanalyzed as a marker of reciprocity parallel to the preverbal reciprocal markers in the substrate languages (Siegel 1999, pp. 34–5). In other words, there is no congruent structure, and thus nowhere to transfer to.

Pidgins/Creoles and Second Language Acquisition 201 On the other hand, Australian Kriol, which also is lexified by English, does have a reciprocal marker gija (from English ‘together’), as in this example (Munro 2004, p. 94): (1) Minbala bin dokdok gija . . . 1du.excl past talk-red recip ‘We two were talking to each other . . .’ Munro (2004, p. 93) reports that all of the four key substrate languages she examined have a post-verbal marker for reciprocal (and reflexive) – for example, in Marra (based on Heath 1981, p. 205): (2) rang+bala-nyi-rlana. hit+3pl-aux.ppunct-refl/recip ‘They hit each other.’ The form together in English occurs in the same post-verbal position as the reciprocal markers of the substrate languages and has some shared function or meaning. Thus, it could be misinterpreted as corresponding to the substrate structure and transfer of function could occur. With regard to reinforcement principles affecting which of the transferred features are actually retained in MP, the most important factor appears to be frequency of the features or corresponding features in the contact environment, or what I call “substrate reinforcement” (Siegel 1998). This is closely related to observations that the greater the typological similarity among substrate languages, the greater the extent of substrate influence in a creole (e.g., Singler 1988). The predominance of speakers of certain substrate languages (numerically or socially) may also affect the frequency of particular transferred features in the contact environment (Thomason & Kaufman 1988). For example, one reason why substrate influence from Portuguese was so significant in the development of Hawai[i Creole was that a large number of its speakers were given positions of authority as plantation overseers (Siegel 2000). The combination of congruence and frequency in the contact environment can lead to the transfer and retention of features that are not commonly found in creoles because they are supposedly “superfluous” or “decorative” and not required for basic communication (cf. McWhorter 2001, 2002). For example, Tayo, a French lexified creole spoken in New Caledonia, has an evidential marker, the preverbal marker ryãke/ãke/ke (from French rien que ‘nothing but, just’), which indicates some degree of subjective reality: (3) sola ãk fe ã grã barach si larut 3pl real make a big barricade on road ‘They’ve [really] made a big barricade on the road.’ (adapted from Siegel, Sandeman, & Corne 2000, p. 92)

202

Jeff Siegel

We find this feature in Tayo because preverbal markers of asserted or insisted reality are found in the three key substrate languages – for example, in Xârâcùù: (4) è va piicè chaa péci 3sg real look.for det book ‘He is really looking for a book.’ (Moyse-Faurie 1995, p. 121) In addition, there is a salient string of morphemes in French which can occur preverbally and which could be (mis)interpreted as having a similar function (Siegel, Sandeman, & Corne 2000, p. 93).

3.2.2

Problems for transfer

There are three problems for the conceptualization of transfer just presented: (1) scarcity of evidence of early transfer in IL studies and in restricted pidgins, (2) existence of transferred structures in expanded pidgins and creoles that seemingly had nowhere to transfer to, and (3) creole features apparently from the substrate that were not found in the preceding pidgin (where textual evidence exists from both the creole and the preceding pidgin). A discussion of these problems follows. First, unlike with simplification, detailed descriptions of the IL of L2 learners have not provided evidence of the process of transfer that is supposed to be the source of some P/C features – especially those features that appear to be superstrate forms with substrate grammatical properties. For example, in the studies of the IL of learners of French, Dutch, and Swedish mentioned earlier, there is no evidence of transfer of features of the kind found in pidgins or creoles with the same lexifier. The one exception may be found in the studies comparing French ILs and French-based creoles. Although Véronique (1994) notes the absence in the ILs of the postposed determiner la which is found in the French creoles, Mather (2000) did find the postposed determiner la, modeled on the French postposed deictic là. Significantly, the feature occurred in the IL of L1 Ewe speakers, and postposed determiners are found in Ewe. (The learners of French in Véronique’s study were L1 speakers of Moroccan Arabic, which does not have postposed determiners.) Mather (2000, p. 250) cites this example: (5) L2 French, L1 Ewe: N’y a qu’à pousser auto-là. ‘All you need to do is to push the car.’ (6) Standard French: Il n’y a qu’à pousser l’auto. This is similar to the postposed definite determiner of Haitian Creole, which, as Lefebvre (1998b, p. 94) has argued, mirrors the position and functions of the definite determiner in Fongbe (a language closely related to Ewe):

Pidgins/Creoles and Second Language Acquisition 203 (7) Haitian Creole: krab la (8) Fongbe:

yo

àscn c lù crab det pl ‘the crabs (in question)’ (Lefebvre 2004, p. 91)

However, while this may be an example of L1 influence in Haitian Creole, it may not be the result of transfer so much as substrate reinforcement of one stage of the normal developmental sequence in the acquisition of French by speakers of languages with postnominal deictic markers (Zobl 1982). In the European Science Foundation (ESF) study, there was also no evidence of the transfer of L1 properties onto L2 forms in the Basic Variety. The only evidence of transfer was that of word order in the earliest stages – for example, OV rather than VO order occurred in the IL of Punjabi learners of English and Turkish learners of Dutch (reported in Kellerman 1995, p. 137). The same is true for restricted pidgins. While the transfer of word order is found in the pidginized English in Hawai[i studied by Bickerton (1977), as well as in some of the restricted pidgins listed above (e.g., Pidgin Fijian and Pidgin Hindustani), there is little if any clear evidence of superstrate forms with substrate grammatical properties. This is not surprising, considering the lack of grammatical morphology in both the Basic Variety and restricted pidgins. That word order transfer of this type occurs also presents another problem: there is no congruence between the word orders of the L1 and L2, and therefore, nowhere to transfer to. So according to the congruence constraint, such transfer should not occur. But of course, it does, and evidence of word order transfer at an earlier stage of development can be found in some pidgins and creoles, such as in the example of Sranan postpositions discussed in Bruyn (this volume). The third problem is that many of the expanded features of creoles were apparently not found in any preceding pidgin. For example, Roberts (1998, 2000) shows that three features of Hawai[i Creole – the non-punctual aspect marker stei, combinations of TMA markers (wen, stei, gon), and fo introducing clausal complements – first appear in published texts in the early 1920s and are attributed to the speech of the locally born children who were the original monolingual speakers of the creole. At first glance, this seems to preclude transfer of substrate features as a possible source and to lead to the conclusion that these features were innovations of the children. However, a detailed examination of the three features reveals striking similarities with one of the substrate languages, Portuguese (Siegel 2000). In fact, in terms of these features, Hawai[i Creole is more similar to Portuguese than to most other creoles. For example, the functions of the Hawai[i Creole non-punctual marker stei are the same as those of Portuguese estar (see Siegel 2003b, p. 197). How can it be, then, that some creoles seem to have more substrate features than their pidgin predecessors even though creole speakers, unlike pidgin speakers, could not have transferred these features because they never knew the substrate languages?

204

Jeff Siegel

One way all three of these problems can be resolved is to consider that two different kinds of transfer (other than phonological) occur mainly at different stages in SLA and in P/C development: the adoption of the word order of the L1 in learning the L2, and the use of L2 forms with L1 grammatical properties. For convenience here, I will refer to the former as “word order transfer” and to the latter as “functional transfer.” Some scholars consider the overall field of SLA to be divided into two main areas of research: second language acquisition and second language use (e.g., Ellis 1994, p. 13; Gass 1998, p. 84). L2 acquisition as opposed to L2 use is concerned with the gradual attainment of linguistic competence in the L2 – in other words, with the learning of the L2 grammar. L2 use as opposed to L2 acquisition looks at how learners make use of their existing L2 knowledge when trying to communicate in the language – and how they compensate for inadequate L2 knowledge when they are under pressure to communicate in the L2 (e.g., Tarone 1981; Poulisse 1996). The proposal here has two aspects. First, word order transfer occurs at the earliest stages of SLA and functional transfer occurs later. Second, word order transfer is involved in L2 acquisition while functional transfer is a consequence of L2 use, according to the acquisition-use distinction made here. Following is a more detailed explanation.

3.3

Word order transfer and L2 acquisition

The transfer of word order has been a controversial topic in the field of SLA. Rutherford (1986) and Zobl (1983, 1986) claimed that there is usually no transfer of basic word order. But Odlin (1990) presented 11 counter-examples, such as transfer of Japanese and Korean OV word order into English. More recently, this topic has been revived in recent debates about the question whether the principles of Universal Grammar (UG), especially parameters, are available to L2 learners. Epstein, Flynn, and Martohardjono (1996) argue for what they call the “full access” hypothesis, which “asserts that UG in its entirety constrains L2 acquisition” (p. 677). This hypothesis plays down the role of transfer – in this context viewed as the use of prior linguistic knowledge in the construction of the L2 grammar. However, other hypotheses or models accept continuing access to UG but still allow for the role of transfer. Referring to White (1991), Sharwood Smith (1996, p. 75) notes that UG is relevant to SLA but that “learners assume that L1 parameter settings will work for L2 unless evidence turns up to disconfirm this assumption” (italics in original). One such model is the “Full Transfer/Full Access” hypothesis (Schwartz 1996, 1998; Schwartz & Sprouse 1996). According to this model, SLA in adults and children depends on three components – the L2 initial state, UG, and the target language (or L2) input. Significantly, the L2 initial state comprises the entirety of the L1 grammar, and therefore IL development is constrained by both the L1 grammar and UG. Thus, all the syntactic properties of the L1, including the setting for the headedness

Pidgins/Creoles and Second Language Acquisition 205 parameter which determines basic word order, are initially transferred. To support this position, Schwartz (1998) presents an impressive array of examples of word order transfer, including OV order into English IL by Turkish learners and N-Adj order into German IL by Italian and Spanish learners. This model has also received a great deal of independent support in the recent literature (e.g., Bhatt & Hancin-Bhatt 1996; Camacho 1999; Slabakova 2000). According to Schwartz (1998), the way that progress towards the L2 (or TL) takes place is that input that cannot be accommodated to the L1 grammar causes the system to restructure. For basic word order, the revision does seem to occur very rapidly – and this would explain the relative rarity of word order transfer reported in the SLA literature. This rapid adoption of L2 word order is most probably because basic word order is quite a salient structural characteristic (Odlin 1990; Comrie 1997). In other words, learners normally have metalinguistic awareness of rules for the ordering of verb and object, unless there are a large number of rules involving structural detail (as in German and Dutch). If we assume that SLA is relevant to P/C genesis, then this would explain why the basic word order of the substrate languages is not usually maintained in the resultant contact language: minimal exposure to the superstrate language would have caused rapid restructuring. But then the question is: Why would other aspects of the word order of the L1 or substrate remain? The answer may be that other L2 word order features are not so accessible to consciousness. For example, Lightbown and Spada (2000) report that because L1 English learners of French have no metalinguistic awareness of the rules they use for adverb placement, they do not notice how their English sentences differ from those of French. Thus, these learners use English patterns of adverb placement when speaking French. Lightbown and Spada conclude, therefore, that positive evidence is not always sufficient to lead to acquisition. Similarly, White (1991) argues that L2 learners, unlike L1 learners, need negative evidence to reset some parameters. In more general terms, Schwartz (1998, p. 148) asserts that “convergence on the TL grammar is not guaranteed; this is because unlike L1 acquisition, the L2 starting point is not simply open or set to ‘defaults’, and so the data needed to force L2 restructuring could be either nonexistent or obscure.” If input from the lexifier language was restricted in P/C development, as most creolists believe, then these insights from SLA theory would explain the retention of some substrate features. This point of view has been expressed in different terms by DeGraff (1996, p. 723): If target PLD [primary linguistic data] in L2A remain (moderately) below threshold T [the level needed for UG-constrained learning to occur], then the adult learner might have no other choice but to resort to the relexification-like strategies that make use of L1 settings . . .

In the case of creoles such as Sranan, which have preserved aspects of basic word order from their substrate languages in postpositions (Bruyn, this volume),

206

Jeff Siegel

we would have to assume that there was very little input from the lexifier. And this is indeed what occurred in Suriname: the lexifier, English, was the language of the colonial power only from 1651–67, after which it was replaced by Dutch. In light of the preceding discussion, the original notion of “congruence constraint” (see above) clearly needs some revision. First, if we accept Schwartz’s (1996, 1998) model of the L1 being the initial state for L2 acquisition, then there can be no constraint on transfer at the point of initial SLA. Instead, at this stage, the congruence constraint would be relevant to whether or not an initially transferred feature is retained. For example, with regard to basic word order, if a lack of congruence with the L2 is perceived, the interlanguage would be restructured to conform to the L2 data. But if an L2 feature is misinterpreted as being parallel in function and position to a feature in the L1, then restructuring will not occur. Second, it is clear that the congruence constraint cannot apply in situations where there is little or no input from the original L2 – for example in situations, such as in Suriname, where the original lexifier was removed. This analysis corresponds with that of Andersen (1990), who says: “It may be the case that, in settings where pidgin languages develop, there is not enough input from the target language available to promote transfer based on the Transfer to Somewhere Principle, and relexification is the only resource the learner has” (p. 63). The view presented here is that the Transfer to Somewhere Principle is relevant to the functional transfer that occurs later in L2 use, to which we now turn.

3.4

Functional transfer and L2 use

Many researchers in SLA think of transfer primarily as a feature of L2 performance or use rather than acquisition (e.g., Meisel 1983; Kellerman 1995). Therefore, transfer is considered to be a communication strategy, a means for overcoming communication problems. According to Sharwood Smith (1986, p. 15), cross-linguistic influence (or transfer) typically occurs in two contexts: (1) “overload” situations or “moments of stress” when the existing L2 system cannot cope with immediate communicative demands, and (2) “through a desire to express messages of greater complexity than the developing control mechanisms can cope with.” (See Jarvis & Odlin 2000 for a similar view.) These views correspond well with Andersen’s (1983a) point that the substrate languages could be a possible source for structures in the expansion and complication involved in creole genesis. If we conceive of transfer in this way, then we can understand the answer to the third problem outlined above: Why do some creoles have more substrate features than their pidgin predecessors even though creole speakers, unlike pidgin speakers, did not know the substrate languages? With regard to Hawai[i Creole, Roberts (1998, 2000) demonstrates that the first generation of locally born children of the immigrant laborers were generally bilingual in a form of English and their parents’ language, and that it was largely the second

Pidgins/Creoles and Second Language Acquisition 207 generation of locally born children, rather than the first, who became the original speakers of Hawai[i Creole. This occurred during the period from 1905 to the early 1920s, when the dominant ethnic group at this time, in terms of both prestige and numbers of the locally born second generation, were the Portuguese. Roberts (2000, p. 266) also shows that the Portuguese were the first immigrant group to shift from their ancestral language. Thus, it was adult members of the first locally born generation, bilingual in Portuguese, who adopted the existing pidgin as their primary language and passed it on to their children who acquired it as their L1. Therefore, it was likely that the grammatical innovations referred to above, such as non-punctual stei, were the result of transfer by these adults, not the inventions of their children (see also discussion in Veenstra, this volume). Another example concerns Tok Pisin (the variety of Melanesian Pidgin spoken in Papua New Guinea), an expanded pidgin which is now being acquired as a first language by large numbers of children. Sankoff (1994) shows that children who speak Tok Pisin as their first language use the preverbal particle i- with serial verbs in a pattern that is quite similar to that of the Austronesian substrate languages – unlike earlier generations of speakers of Tok Pisin, who were bilingual in the substrate languages. Sankoff believes that the substrate languages are the source of the pattern used by the children and asks (p. 312): “Why should such a development be realized by speakers who are clearly more distant from the Austronesian substrate than their grandparents or greatgrandparents were?” The explanation again seems to be that this development was an innovation of the parents, which was then passed on to and acquired by the children. As we have seen, when the use of a pidgin is extended into new areas, its grammar is also expanded. We have also seen that in order to compensate for shortcomings in their L2, speakers may fall back on their L1 as a communication strategy in L2 use. Of course, the ultimate extension of use of a pidgin L2 is when speakers shift to it as their primary language, which they then pass onto their children. So in this case, when the need arose in the parents’ generation (e.g., in using the L2, Tok Pisin, for new functions), its speakers fell back on their L1 – one of the Austronesian substrate languages – to provide missing grammatical structures. In other words, people transferred features from their L1 into their L2. Thus, Sankoff (1994, p. 314) concludes that “maximal influence is exerted from a substrate not when initial contact occurs, but just at the point of language shift.” When these innovations first occurred, they were the result of a communication strategy and not firmly established in the grammars of all adult speakers. This may be why they were not reported in published sources. Nevertheless, the innovations were added to the pool of variants which were used for communication and perhaps would have been incorporated into the pidgin over a long period of leveling. However, leveling is accelerated when nativization occurs (Siegel 1997), and thus it was the children who rapidly regularized these innovations and integrated them into the grammar of the newly emerged

208

Jeff Siegel

creole. As these features occurred more regularly in the speech of the children, they were more noticeable to observers and were subsequently recorded in published texts. This scenario would support the statement by DeGraff (1999, p. 488) following Slobin (1977) that “adults are the innovators whereas children are the regulators” (italics in the original).5 Referring to Sankoff (1994) and Newport (1999), DeGraff (1999, p. 507) concludes that “children amplify and restructure certain (substrate-based) innovations introduced through L2A [second language acquisition] by adults, and they incorporate (previously lessthan-stable) innovations into permanent and stable parts of their grammars.” In summary, we have seen that word order transfer in P/C genesis can occur in the earliest stage of development as part of L2 acquisition, but it is mostly short-lived unless there is extremely limited contact with the lexifier. We have also seen that functional transfer (the use of L2 forms with L1 grammatical properties) generally occurs in later stages of development as a strategy of expansion in L2 use. Therefore, while some evidence of word order transfer is found in restricted pidgins, evidence of functional transfer is found primarily in expanded pidgins and creoles.

4

Other approaches and controversies

While more creolists today may agree about the involvement of processes of SLA in P/C genesis, there is no consensus about exactly what these processes are and how and when they apply. For example, Koontz-Garboden and Clements (2002) view the initial stages of pidginization as the same as the initial stage of L1 or L2 acquisition. Learners first acquire lexical items which then leads to the learning (or forming) of grammar, which involves the acquisition of functional elements. They do not believe that these processes involve simplification, but they do believe that substrate transfer can affect the acquisition of functional elements (see also Clements 2003). For those creolists who do believe that simplification in SLA is an important factor, there are basically two camps. On one hand, there is the “antecedent pidgin” view. This is the conventional idea that creoles are descended from pidgins which arose in the earliest stages of language contact when, for example, slaves used incompletely acquired (and simplified) versions of the lexifier for communication among themselves (Hall 1966). Therefore, the simplification found in creoles has been inherited from their pidgin predecessors. For example, McWhorter (2000, p. 106) adopts this stand, saying that creoles “emerged as radically reduced pidgins.” He also makes clear his position that the simplification found in pidgins is the result of adult SLA (McWhorter 2001). This view includes the notion that the use of the earlier pidgin marked a break in transmission of the lexifier. On the other hand, there is the “gradual basilectalization” view. This is the idea that in the earliest stages of language contact, the lexifier was an available target and acquisition was relatively successful. Only in later stages did

Pidgins/Creoles and Second Language Acquisition 209 the lexifier become less accessible, for example during the rapid influx of new slaves. It was then that large-scale simplification began to occur as “approximations” of the lexifier language (the result of SLA) became the target. Thus, the simplification found in creoles is the result of approximations of approximations of the target language (Chaudenson 1992, 2001; Mufwene 2001; see also Crowley, this volume). Followers of this point of view maintain that, at least for French-lexified creoles, there is no evidence of a pidgin predecessor, and no break in transmission of the lexifier.6 Of course, both of these views could be true. Alleyne (2000) surmises that for English-lexified creoles, maximal restructuring occurred at the beginning of historical development through rapid pidginization, whereas for Frenchlexified creoles, it occurred at the end through cumulative divergent changes. There are also other positions which straddle both views. With regard to French creoles, Becker and Veenstra (2003a, 2003b) agree that French was the target in the early stages of language contact, but they also believe that there was a drastically simplified antecedent – the Basic Variety, which was the result of this initial L2 learning (see 2.2.1). With the influx of new slaves, the Basic Variety rather than French became the target in what they call “target shift.” Other creolists (e.g., Winford 2003) take the similar point of view that once a pidgin is established, it, rather than the lexifier itself, becomes the target for new arrivals. In all these approaches, the processes of SLA, simplification and transfer, are believed to apply at some stage. But DeGraff (2001a, 2001b) argues against the notions of a break in transmission of the lexifier – maintaining that there is no historical evidence of such a phenomenon – and a drastically simplified, pidgin-like Proto-creole. In his view, a language with such drastic simplification would be contrary to the uniform innate mental capacities human beings have for language, and thus imply that its speakers were somehow different from other humans. In addition, DeGraff (2001b, p. 265) asserts that “alongside simplification, restructuration and innovation, language contact also entails language transfer through SLA, which will inevitably carry over into the emerging contact language some of the complexity of the languages in contact.” However, we have seen earlier in this chapter that restricted pidgins exist which are drastically simplified in comparison to the languages in contact and which do not exhibit evidence of transfer. But since these 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. A similar point can be made about the existence of the Basic Variety. My view is that restricted pidgins were involved in the genesis of some creoles, and of other contact varieties as well. But in most cases the creole did not evolve directly from the pidgin. In other words, at no stage did a group of people suddenly start speaking a restricted pidgin as their first language. Rather, the pidgin was one of the many sources in the contact environment which provided what DeGraff calls the primary linguistic data (PLD) for language acquisition. This is, of course, related to the leveling view of P/C genesis that

210

Jeff Siegel

I presented earlier. Other sources of simplified features may have included various unstable L2 varieties, as well as foreigner talk. There would also have been unsimplified features in the pool, including some resulting from transfer and others directly from the various varieties of the lexifier (so that one cannot talk about a complete break in transmission). When a creole emerges, some of these features are leveled out, but others are retained, and these may include both simplified features that result from partial L2 acquisition and transferred features that result from use of a partially acquired L2. We have already seen that one factor that influences the retention of features in the leveling process is their frequency in the contact environment. Therefore, we would expect both more simplified features and more transferred features in contact varieties where, for social and/or ideological reasons, there was less complete SLA. This would explain the different degree of simplification and restructuring found in various contact languages, ranging from that of the “radical” creoles such as Saramaccan, to other creoles such as Hawai[i Creole to semi-creoles such as Brazilian Portuguese, to indigenized varieties such as Singapore Colloquial English.

5

Conclusions

This chapter has illustrated how the processes of simplification and transfer in SLA are relevant to the development of pidgins and creoles. It has also shown how SLA research can provide insights into various issues in P/C studies, such as the origins of particular creole features, the role of different kinds of transfer at different stages of P/C development, the reasons for the retention of transferred features, and constraints on substrate influence. Other work in the field of SLA could also be a source of ideas for a future research agenda in P/C studies. For example, Jarvis (2000) articulates three types of empirical evidence that can be used to test whether certain structures in IL are derived from transfer from the L1. He proposes that evidence of at least two out of the three types would be enough to verify such L1 transfer. Similar kinds of evidence could be used to test whether certain structures in P/Cs are derived from substrate languages. Thus, following Jarvis (2000), substrate influence could be verified when two of the following can be substantiated: (1) creoles and L2 varieties (or ILs) with the same lexifier or target language and the same substrate languages exhibit the same features; (2) creoles with the same lexifier language but different substrate languages exhibit different features; or (3) similarities exist between the features of the creole and the features of the substrate languages. Of course, evidence of (3) is what is usually presented in accounts of substrate influence; however, evidence of the other types is rare, and investigations in these areas would be very useful in debates about the origins of various creole features. To conclude, the field of P/C will benefit in general by being more informed about current research in the field of second language acquisition and use.

Pidgins/Creoles and Second Language Acquisition 211

NOTES 1 The chapter was written for this volume in 2002 and revised in 2004. Many of the issues discussed here are explored in greater detail in more recent publications, especially in Lefebvre, White, and Jourdan (2006a) and Siegel (2008). Further information on the development of Hawai[i Creole can be found in Roberts (2005). 2 For a detailed critique of the Language Bioprogram Hypothesis, see Siegel (2007a). 3 See also, Mather (2006) and various chapters in Lefebvre, White, and Jourdan (2006a). 4 Lefebvre, White, and Jourdan (2006b, p. 5) state, following Naro (1978, p. 337): “Relexification is a particular type of transfer.” 5 See also Hudson Kam and Newport (2005). 6 For further discussion, see Siegel (2007b).

REFERENCES Alleyne, Mervyn C. (1971) Acculturation and the cultural matrix of creolization. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge University Press, Cambridge, pp. 169–86. Alleyne, Mervyn C. (2000) Opposite processes in “creolization.” In: Neumann-Holzschuh and Schneider (eds.), pp. 125–33. Andersen, Roger W. (1980) Creolization as the acquisition of a second language as a first language. In: Albert Valdman and Arnold Highfield (eds.) Theoretical Orientations in Creole Studies. New York: Academic Press, pp. 273–95. Andersen, Roger W. (1981) Two perspectives on pidginization as second language acquisition. In: Roger W. Andersen (ed.) New Dimensions in Second Language Acquisition Research. Rowley, MA: Newbury House, pp. 165–95. Andersen, Roger W. (1983a) Introduction: A language acquisition interpretation of pidginization and creolization. In: Andersen (ed.), pp. 1–56. Andersen, Roger W. (ed.) (1983b) Pidginization and Creolization as Language Acquisition. Rowley, MA: Newbury House. Andersen, Roger W. (1983c) Transfer to somewhere. In: Susan M. Gass and Larry Selinker (eds.) Language Transfer in Language Learning. Rowley, MA: Newbury House, pp. 177–201. Andersen, Roger W. (1990) Models, processes, principles and strategies: Second language acquisition inside and outside the classroom. In: Bill Van Patten and James F. Lee (eds.) Second Language Acquisition: Foreign Language Learning. Clevedon, UK: Multilingual Matters, pp. 45–68. Arends, Jacques (1993) Towards a gradualist model of creolization. In: Francis Byrne and John Holm (eds.) Atlantic Meets Pacific: A Global View of Pidginization and Creolization. Amsterdam: John Benjamins. Arends, Jacques (ed.) (1995) The Early Stages of Creolization. Amsterdam: John Benjamins. Arends, Jacques, Pieter Muysken, and Norval Smith (1995) Conclusions. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 319–30.

212

Jeff Siegel

Baker, Philip (1987) The historical developments in Chinese Pidgin English and the nature of the relationships between the various pidgin Englishes of the Pacific region. Journal of Pidgin and Creole Languages 2, 163–207. Baker, Philip (1994) Creativity in creole genesis. In: Dany Adone and Ingo Plag (eds.) Creolization and Language Change. Tübingen: Max Niemeyer, pp. 65–84. Baker, Philip (1997) Directionality in pidginization and creolization. In: Arthur K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles. Amsterdam: John Benjamins, pp. 91–109. Baker, Philip (2000) Theories of creolization and the degree and nature of restructuring. In: Neumann-Holzschuh and Schneider (eds.), pp. 41–63. Becker, Angelika and Tonjes Veenstra (2003a) The survival of inflectional morphology in French-related creoles: The role of SLA processes. Studies in Second Language Acquisition 25, 283–306. Becker, Angelika and Tonjes Veenstra (2003b) Creole prototypes as Basic Varieties and inflectional morphology. In: Christine Dimroth and Marianne Starren (eds.) Information Structure and the Dynamics of Language Acquisition. Amsterdam: John Benjamins, pp. 235–64. Bhatt, Rakesh M. and Barbara Hancin-Bhatt (1996) Transfer in L2 grammars. Behavioral and Brain Sciences 19, 715–16. Bickerton, Derek (1977) Pidginization and creolization: Acquisition and language universals. In: Valdman (ed.), pp. 49–69. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984a) The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7, 173–221. Bickerton, Derek (1984b) The language bioprogram hypothesis and second language acquisition. In: William E. Rutherford (ed.) Language Universals and Second Language Acquisition. Amsterdam: John Benjamins, pp. 141–61. Bickerton, Derek (1988) Creole languages and the bioprogram. In: Frederick J. Newmeyer (ed.) Linguistics: The Cambridge Survey. Vol. 2, Linguistic Theory: Extensions and Implications. Cambridge: Cambridge University Press, pp. 268–84. Bickerton, Derek (1999a) Creole languages, the Language Bioprogram Hypothesis, and language acquisition. In: William C. Ritchie and Tej K. Bhatia (eds.) Handbook of Child Language Acquisition. San Diego: Academic Press, pp. 195–220. Bickerton, Derek (1999b) How to acquire language without positive evidence: What acquisitionists can learn from creoles. In: Michel DeGraff (ed.) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press, pp. 49–74. Bickerton, Derek and C. Odo (1976) Change and Variation in Hawaiian English, Vol. 1: General Phonology and Pidgin Syntax. Honolulu: Social Sciences and Linguistics Institute, University of Hawai[i. Bloomfield, Leonard (1933) Language. New York: Holt, Rinehart & Winston. Camacho, José (1999) From SOV to SVO: The grammar of interlanguage word order. Second Language Research 15, 115–32. Chaudenson, Robert (1992) Des îles, des hommes, des langues [Islands, men and language]. Paris: L’Harmattan. Chaudenson, Robert (2001) Creolization of Language and Culture (revised in collaboration with Salikoko S. Mufwene). London: Routledge. Clements, J. Clancy (2001) Word order shift and natural L2 acquisition in a Portuguese creole. In: Caroline Wiltshire and Joaquim Camps (eds.) Romance Syntax, Semantics and L2 Acquisition. Amsterdam/Philadelphia: Benjamins, pp. 73–87.

Pidgins/Creoles and Second Language Acquisition 213 Clements, J. Clancy (2003) The tense-aspect system in pidgins and naturalistically learned L2. Studies in Second Language Acquisition 25, 245–81. Comrie, Bernard (1997) On the origin of the Basic Variety. Second Language Research 13, 367–73. Corder, S. P. (1981) Formal simplicity and functional simplification in second language acquisition. In: Roger W. Andersen (ed.) New Dimensions in Second Language Research. Rowley, MA: Newbury House, pp. 146–52. DeGraff, Michel (1996) UG and acquisition in pidginization and creolization. Behavioral and Brain Sciences 19, 723–4. DeGraff, Michel (1999) Creolization, language change, and language acquisition: An epilogue. In: Michel DeGraff (ed.) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press, pp. 473–543. DeGraff, Michel (2001a) Morphology in creole genesis: Linguistics and ideology. In: Michael Kenstowicz (ed.) Ken Hale: A Life in Language. Cambridge, MA: MIT Press, pp. 53–121. DeGraff, Michel (2001b) On the origins of creoles: A Cartesian critique of neoDarwinian linguistics. Linguistic Typology 5, 213–310. Dulay, Heidi and Marina Burt (1973) Should we teach children syntax? Language Learning 23, 245–58. Dutton, Tom (1997) Hiri Motu. In: Thomason (ed.), pp. 9–41. Ellis, Rod (1994) The Study of Second Language Acquisition. Oxford: Oxford University Press. Epstein, Samuel D., Suzanne Flynn, and Gita Martohardjono (1996) Second language acquisition: Theoretical and experimental issues in contemporary research. Behavioral and Brain Sciences 19, 677–714. Ferguson, Charles A. and Charles E. DeBose (1977) Simplified registers, broken language and pidginization. In: Valdman (ed.), pp. 99–125. Field, Fred (2004) Second language acquisition in creole genesis. In: Geneviève Escure and Armin Schwegler (eds.) Creoles, Contact and Language Change. Amsterdam: John Benjamins, pp. 127–60. Firth, Alan and Johannes Wagner (1997) On discourse, communication, and (some) fundamental concepts in SLA research. The Modern Language Journal 81, 285–300. Gass, Susan (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, 83–90. Goddard, Ives (1997) Pidgin Delaware. In: Thomason (ed.), pp. 43–98. Goodman, Morris (1987) Pidgin origins reconsidered. Journal of Pidgin and Creole Languages 2, 149–60. Hall, Robert A., Jr. (1966) Pidgin and Creole Languages. Ithaca: Cornell University Press. Heath, Jeffrey (1981) Basic Materials in Mara: Grammar, Texts and Dictionary. Canberra: Pacific Linguistics. Helms-Park, Rena (2003) The implications of causative serial verbs in the interlanguage of Vietnamese ESL learners. Studies in Second Language Acquisition 25, 211–44. Hesseling, Dirk Christian ([1933] 1979) On the Origin and Formation of Creoles: A Miscellany of Articles (edited and translated by T. L. Markey and Paul T. Roberge). Ann Arbor: Karoma. Hudson Kam, Carla L. and Elissa L. Newport (2005) Regularizing unpredictable variation: The roles of adult and child learners in language formation and change. Language Learning and Development 1, 151–95. Huttar, George L. and Frank J. Velantie, (1997) Ndjuka-Trio Pidgin. In: Thomason (ed.), pp. 99–124.

214

Jeff Siegel

Jahr, Ernst Håkon (1996) On the status of Russenorsk. In: Ernst Håkon Jahr and Ingvild Broch (eds.) Languages in Contact in the Arctic: Northern Pidgins and Contact Languages. Berlin: Mouton de Gruyter, pp. 107–22. Jarvis, Scott (2000) Methodological rigor in the study of transfer: Identifying L1 influence in the interlanguage lexicon. Language Learning 50, 245–309. Jarvis, Scott and Terence Odlin (2000) Morphological type, spatial reference, and language transfer. Studies in Second Language Acquisition 22, 535–56. Jesperson, Otto (1922) Language: Its Nature, Development, and Origin. London: Allen and Unwin. Kay, Paul and Gillian Sankoff (1974) A language universal approach to pidgins and creoles. In: David DeCamp and Ian F. Hancock (eds.) Pidgins and Creoles: Current Trends and Prospects. Washington, DC: Georgetown University Press, pp. 61–72. Keesing, Roger M. (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford: Stanford University Press. Kellerman, Eric (1995) Crosslinguistic influence: Transfer to nowhere? Annual Review of Applied Linguistics 15, 125–50. Klein, Wolfgang and Clive Perdue (1997) The Basic Variety (or: Couldn’t natural languages be much simpler?). Second Language Research 13, 301–47. Koontz-Garboden, Andrew J. and J. Clancy Clements (2002) Some theoretical implications of adpositions in Spanish and Portuguese-based creoles. Paper presented at the Annual Meeting of the Society for Pidgin and Creole Linguistics, January, San Francisco. Kotsinas, Ulla-Britt (1996) Aspect marking and grammaticalization in Russenorsk compared with immigrant Swedish. In: Ernst Håkon Jahr and Ingvild Broch (eds.) Language Contact in the Arctic: Northern Pidgins and Contact Languages. Berlin: Mouton de Gruyter, pp. 123–54. Kotsinas, Ulla-Britt (2001) Pidginization, creolization and creoloid in Stockholm, Sweden. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 125–55. Lefebvre, Claire (1986) Relexification in creole genesis revisited: The case of Haitian Creole. In: Pieter Muysken and Norval Smith (eds.) Substrata versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 279–300. Lefebvre, Claire (1996) The tense, mood and aspect system of Haitian Creole and the problem of transmission of grammar in creole genesis. Journal of Pidgin and Creole Languages 11, 231–311. Lefebvre, Claire (1997) Relexification in creole genesis: The case of demonstrative terms in Haitian Creole. Journal of Pidgin and Creole Languages 12, 181–201. Lefebvre, Claire (1998a) Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Lefebvre, Claire (1998b) Multifunctionality and variation among grammars: The case of the determiner in Haitian Creole and Fongbe. Journal of Pidgin and Creole Languages 13, 93–150. Lefebvre, Claire (2004) Issues in the Study of Pidgin and Creole Languages. Amsterdam: John Benjamins. Lefebvre, Claire, Lydia White, and Christine Jourdan (eds.) (2006a) L2 Acquisition and Creole Genesis: Dialogues. Amsterdam: John Benjamins. Lefebvre, Claire, Lydia White, and Christine Jourdan (2006b) Introduction. In Claire Lefebvre, Lydia White, and Christine Jourdan (eds.) L2 Acquisition and Creole Genesis: Dialogues. Amsterdam: John Benjamins, pp. 1–14.

Pidgins/Creoles and Second Language Acquisition 215 Lightbown, Patsy M. and Nina Spada (2000) Do they know what they’re doing? L2 learners’ awareness of L1 influence. Language Awareness 9, 198–217. Lumsden, John S. (1996) On the acquisition of nominal structures in the genesis of Haitian Creole. In: Herman Wekker (ed.) Creole Languages and Language Acquisition. Berlin: Mouton de Gruyter, pp. 184–205. Lumsden, John S. (1999) The role of relexification in creole genesis. Journal of Pidgin and Creole Languages 14, 225–58. Manessy, Gabriel (1977) Processes of pidginization in African languages. In: Valdman (ed.), pp. 129–34. Mather, Patrick-André (2000) Creole genesis: Evidence from west African L2 French. In: Dicky G. Gilbers, John Nerbonne, and Jos Schaeken (eds.) Languages in Contact. Amsterdam: Rodopi, pp. 247–61. Mather, Patrick-André (2006) Second language acquisition and creolization: Same (i-) processes, different (e-) results. Journal of Pidgin and Creole Languages 21, 231–74. McKay, Sandra Lee and Wong, Sau-Ling Cynthia (1996) Multiple discourses, multiple identities: Investment and agency in second-language learning among Chinese adolescent immigrant students. Harvard Educational Review 66, 577–608. McWhorter, John H. (2000) Defining “creole” as a synchronic term. In: NeumannHolzschuh & Schneider (eds.), pp. 85–123. McWhorter, John H. (2001) The world’s simplest grammars are creole grammars. Linguistic Typology 5, 125–66. McWhorter, John H. (2002) The rest of the story: Restoring pidginization to creole genesis theory. Journal of Pidgin and Creole Languages 17, 1–48. Meisel, Jürgen (1983) Transfer as a second-language strategy. Language & Communication 3, 11–46. Migge, Bettina (1998) Substrate influence in creole formation: The origin of give-type serial verb constructions in the Surinamese Plantation Creole. Journal of Pidgin and Creole Languages 13, 215–65. Migge, Bettina (2000) The origin of the syntax and semantics of property items in the Surinamese Plantation Creole. In: John H. McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 201–34. Moyse-Faurie, Claire (1995) Le xârâcùù: Langue de thio-canala (nouvelle-calédonie): Eléments de syntaxe [The Xârâcùù language of Thio-Canala (New Caledonia): Aspects of syntax]. Paris: Peeters. Mufwene, Salikoko S. (1990) Transfer and the substrate hypothesis in creolistics. Studies in Second Language Acquisition 12, 1–23. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Munro, Jennifer M. (2004) Substrate Language Influence in Kriol: The Application of Transfer Constraints to Language Contact in Northern Australia. PhD thesis, University of New England, Armidale, NSW, Australia. Muysken, Pieter (2001) The origin of creole languages: The perspective of second language learning. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 157–73. Muysken, Pieter and Norval Smith (1995) The study of pidgin and creole languages. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 1–14. Naro, Anthony J. (1978) A study on the origins of pidginization. Language 54, 314–47.

216

Jeff Siegel

Nemser, William (1971) Approximative systems of foreign language learners. International Review of Applied Linguistics in Language Learning 9, 115–23. Neumann-Holzschuh, Ingrid and Edgar W. Schneider (eds.) (2000) Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins. Newport, Elissa L. (1999) Reduced input in the acquisition of signed languages: Contributions to the study of creolization. In: Michel DeGraff (ed.) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press, pp. 161–78. Norton, Bonny (2000) Identity and Language Learning: Gender, Ethnicity and Educational Change. Harlow, UK: Pearson Educational. Norton Peirce, Bonny (1995) Social identity, investment, and language learning. TESOL Quarterly 29, 9–31. Odlin, Terence (1990) Word-order transfer, metalinguistic awareness and constraints on foreign language learning. In: Bill Van Patten and James F. Lee (eds.) Second Language Acquisition: Foreign Language Learning. Clevedon, UK: Multilingual Matters, pp. 95–117. Perdue, Clive (ed.) (1993) Adult Language Acquisition: Cross-Linguistic Perspectives. Vol. 2: The Results. Cambridge: Cambridge University Press. Poulisse, Nanda (1996) Strategies. In: Peter Jordens and Josine Lalleman (eds.) Investigating Second Language Acquisition. Berlin: Mouton de Gruyter, pp. 135–63. Rampton, Ben (1995) Crossing: Language and Ethnicity Among Adolescents. London: Longman. Rampton, Ben (1997) A sociolinguistic perspective on L2 communication strategies. In: G. Kasper and E. Kellerman (eds.) Communication Strategies: Psycholinguistic and Sociolinguistic Perspectives. London: Longman, pp. 279–303. Rampton, M. B. H. (1991) Second language learners in a stratified multilingual setting. Applied Linguistics 12, 229–48. Reinecke, John E. (1971) Pidgin French in Vietnam. In: Dell Hymes (ed.) Pidginization and Creolization of Language. Cambridge: Cambridge University Press, pp. 47–56. Roberts, Julian M. (1995) A structural sketch of Pidgin Hawaiian. Amsterdam Creole Studies 14, 97–126. Roberts, Sarah J. (1998) The role of diffusion in the genesis of Hawaiian Creole. Language 74, 1–39. Roberts, Sarah J. (2000) Nativization and genesis of Hawaiian Creole. In: John H. McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 257–300. Roberts, Sarah J. (2005) The Emergence of Hawai[i Creole English in the Early 20th Century: The Sociohistorical Context of Creole Genesis. PhD dissertation, Stanford University. Rutherford, William E. (1986) Grammatical theory and L2 acquisition: A brief overview. Second Language Studies 2, 1–15. Samarin, William J. (1971) Salient and substantive pidginization. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 117–40. Sankoff, Gillian (1994) An historical and evolutionary approach to variation in the Tok Pisin verb phrase. In: Katherine Beals, Jeannette Denton, Robert Knippen, Lynette Melnar, Hisami Suzuki, and Erica Zeinfeld (eds.) Chicago Linguistics Society 30, vol. 2: The Parasession on Variation in Linguistic Theory. Chicago: The Chicago Linguistics Society, pp. 293–320.

Pidgins/Creoles and Second Language Acquisition 217 Schumann, John H. (1978a) The acculturation model for second language acquisition. In: Rosario Gingras (ed.) Second Language Acquisition and Foreign Language Teaching. Arlington, VA: Center for Applied Linguistics, pp. 27–50. Schumann, John H. (1978b) The Pidginization Process: A Model for Second Language Acquisition. Rowley, MA: Newbury House. Schwartz, Bonnie D. (1996) Now for some facts, with a focus on development and an explicit role for L1. Behavioral and Brain Sciences 19, 739–40. Schwartz, Bonnie D. (1998) The second language instinct. Lingua 106, 133–60. Schwartz, Bonnie D. and Rex A. Sprouse (1996) L2 cognitive states and the full transfer/full access model. Second Language Research 12, 40–72. Sebba, Mark (1997) Contact Languages: Pidgins and Creoles. New York: St Martin’s Press. Selinker, Larry (1972) Interlanguage. International Review of Applied Linguistics 10, 209–32. Sharwood Smith, Michael (1986) The competence/control model, crosslinguistic influence and the creation of new grammars. In: Michael Sharwood Smith and Eric Kellerman (eds.) Crosslinguistic Influence in Second Language Acquisition. Oxford: Pergamon Press, pp. 10–20. Sharwood Smith, Michael (1996) Crosslinguistic influence with special reference to the acquisition of grammar. In: Peter Jordens and Josine Lalleman (eds.) Investigating Second Language Acquisition. Berlin: Mouton de Gruyter, pp. 71–83. Siegel, Jeff (1987) Language Contact in a Plantation Environment: A Sociolinguistic History of Fiji. Cambridge: Cambridge University Press. Siegel, Jeff (1990) Pidgin English in Nauru. Journal of Pidgin and Creole Languages 5, 157–86. Siegel, Jeff (1997) Mixing, levelling and pidgin/creole development. In: Arthur K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles. Amsterdam: John Benjamins, pp. 111–49. Siegel, Jeff (1998) Literacy in Melanesian and Australian pidgins and creoles. English World-Wide 16, 104–33. Siegel, Jeff (1999) Transfer constraints and substrate influence in Melanesian Pidgin. Journal of Pidgin and Creole Languages 14, 1–44. Siegel, Jeff (2000) Substrate influence in Hawai[i Creole English. Language in Society 29, 197–236. Siegel, Jeff (2003a) Social context. In: Catherine J. Doughty and Michael H. Long (eds.) Handbook of Second Language Acquisition. Oxford: Blackwell, pp. 178–223. Siegel, Jeff (2003b) Substrate influence in creoles and the role of transfer in second language acquisition. Studies in Second Language Acquisition 25, 185–209. Siegel, Jeff (2007a) Recent evidence against the Language Bioprogram Hypothesis: The pivotal case of Hawai[i Creole. Studies in Language 31, 51–88. Siegel, Jeff (2007b) Transmission and transfer. In: Umberto Ansaldo, Stephen Matthews, and Lisa Lim (eds.) Deconstructing Creole. Amsterdam: John Benjamins, pp. 167–201. Siegel, Jeff (2008) The Emergence of Pidgin and Creole Languages. Oxford: Oxford University Press. Siegel, Jeff, Barbara Sandeman, and Chris Corne (2000) Predicting substrate influence: Tense-modality-aspect marking in Tayo. In: Jeff Siegel (ed.) Processes of Language Contact: Studies from Australia and the South Pacific. Montreal: Fides, pp. 75–97. Singler, John V. (1988) The homogenity of the substrate as a factor in pidgin/creole genesis. Language 64, 27–51.

218

Jeff Siegel

Singler, John V. (1990) On the use of sociohistorical criteria in the comparison of creoles. Linguistics 28, 645–59. Slabakova, Roumyana (2000) L1 transfer revisited: The L2 acquisition of telicity marking in English by Spanish and Bulgarian native speakers. Linguistics 38, 739–70. Slobin, Dan (1977) Language change in childhood and history. In: John Macnamara (ed.) Language Learning and Thought. New York: Academic Press, pp. 185–214. Tarone, Elaine (1981) Some thoughts on the notion of communication strategy. TESOL Quarterly 15, 285–95. Thomason, Sarah G. (ed.) (1997) Contact Languages: A Wider Perspective. Amsterdam: John Benjamins. Thomason, Sarah G. (2001) Language Contact: An Introduction. Washington, DC: Georgetown University Press. Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Valdman, Albert (ed.) (1977) Pidgin and Creole Linguistics. Bloomington, IN: Indiana University Press. Valdman, Albert (1980) Creolization and second language acquisition. In: Albert Valdman and Arnold Highfield (eds.) Theoretical Orientations in Creole Studies. New York: Academic Press, pp. 297–311. Van der Voort, Hein (1996) Eskimo pidgin in West Greenland. In: Ernst Håkon Jahr and Ingvild Broch (eds.) Language Contact in the Arctic: Northern Pidgins and Contact Languages. Berlin: Mouton de Gruyter, pp. 157–258. Véronique, Daniel (1994) Naturalistic adult acquisition of French as L2 and Frenchbased creole genesis compared: Insights into creolization and language change. In: Dany Adone and Ingo Plag (eds.) Creolization and Language Change. Tübingen: Max Niemeyer, pp. 117–37. Washabaugh, William and Fred Eckman (1980) Review of The Pidginization Process: A Model for Second Language Acquisition, by John Schumann. Language 56, 453–6. Wekker, Herman (1996) Creolization and the acquisition of English as a second language. In: Herman Wekker (ed.) Creole Languages and Language Acquisition. Berlin: Mouton de Gruyter, pp. 139–49. White, Lydia (1991) Universal Grammar and Second Language Acquisition. Amsterdam: John Benjamins. Winford, Donald (2000) “Intermediate” creoles and degrees of change in creole formation: The case of Bajan. In: Neumann-Holzschuh and Schneider (eds.), pp. 215–74. Winford, Donald (2002) Creoles in the context of contact linguistics. In: Glenn Gilbert (ed.) Pidgins and Creoles in the Twenty-First Century. New York: Peter Lang, pp. 287–354. Winford, Donald (2003) An Introduction to Contact Linguistics. Oxford: Blackwell. Zobl, Helmut (1982) A direction for contrastive analysis: The comparative study of developmental sequences. TESOL Quarterly 16, 169–83. Zobl, Helmut (1983) L1 acquisition, age of L2 acquisition and the learning of word order. In: Susan M. Gass and Larry Selinker (eds.) Language Transfer in Language Learning. Rowley, MA: Newbury House, pp. 205–21. Zobl, Helmut (1986) Word order typology, lexical government, and the prediction of multiple, graded effects in L2 word order. Language Learning 36, 159–83.

Creole Genesis 219

9

Creole Genesis: The Impact of the Language Bioprogram Hypothesis TONJES VEENSTRA

1 Introduction Derek Bickerton’s language bioprogram hypothesis (LBH), as developed since the mid 1970s, can be summed up thus: one can come to a better understanding of human language and its development thousands, even millions of years ago by looking at how creole languages were created. Strictly speaking, the primary goal of the hypothesis as it was conceived in Bickerton’s 1981 book Roots of Language was to provide a new and persuasive way to tackle the origin-of-language question. To achieve this, the LBH looks at possible parallels between creolization and first language acquisition. Although accounting for creole genesis was not the primary goal of the LBH, Bickerton argued forcefully for a scenario in which adult slaves or contract laborers, uprooted and transplanted to plantation societies in the Caribbean and the Pacific, were not the ones responsible for the creation of creole languages. Rather, it was their children who molded the newly emerging, highly variable interethnic vehicles for communication of these adult slaves and contract laborers into creoles. Bickerton’s claim that creolization is a sudden, catastrophic, single-generation process and has to be defined in terms of nativization – the process by which a language acquires native speakers – has been at the center of controversy in the field of creole studies ever since. It is precisely due to its controversial nature that the LBH provoked an avalanche of research and publications throughout the 1980s and 1990s with the sole aim of disproving it, often demonstrating that some of the data that had been introduced to support the LBH are incomplete, biased, erroneous, or at best inconclusive as evidence. As Arends, Muysken, & Smith (1995b, p. 322) observe, there appears to be a striking paradox: “while there is no question that Bickerton’s views are still quite influential among creolists, they are not shared by many of them in their most complete version.” Nevertheless, as this chapter demonstrates, the impact this hypothesis has had on the creolist community cannot be overestimated.

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

220

Tonjes Veenstra

What made Bickerton’s view of interest outside the field of creole studies, where he is still widely quoted, was his claim that creoles offer a special window on the human language faculty. Whereas his views often met general acceptance outside the field (cf. some of the peer commentaries to Bickerton 1984a), they were never generally accepted by fellow creolists. Also, his more recent work on the origins-of-language question, especially Bickerton (1990, 1995) and Calvin & Bickerton (2000), has attracted no attention whatsoever in the field, but features prominently in discussions on language evolution. Because the sociohistoric and linguistic situation in Hawai[i is fundamental to the LBH, we present a short overview of it in section 2. Section 3 outlines the LBH, distinguishing three distinct phases in its articulation over the years. Section 4 is devoted to the different sets of criticisms of the LBH. In the concluding section, the implications that some of these criticisms have for theories of creole genesis are spelled out.

2

The Hawaiian Situation

The works of Bickerton take the language contact situation in Hawai[i as a starting point for his model of creole genesis. From this particular case of creolization, he distils certain properties of the process of creolization, which he then generalizes to other cases of creolization, particularly those of the Caribbean. Since the Hawaiian situation is the major foundation on which the LBH rests, it is useful to start with a brief overview, primarily based on Reinecke (1969), Bickerton and Odo (1976), Bickerton (1981), Roberts (2000), and Siegel (2000). In the century following initial contact with Europeans, there were very few native speakers of English on the islands. In addition to the sandalwood trade and the whaling industry, there was a small sugar industry, with a labor force consisting mainly of Hawaiians. A revision of US tariff laws in 1876 allowed the free importation of Hawaiian sugar to the United States. As the sugar industry boomed, imported diseases caused Hawai[i’s native population to decline. To expand their operations, plantation owners began to look overseas for their labor supply. The major ethnic groups brought to Hawai[i to work the fields initially were Chinese (1872), Portuguese (1878), and Japanese (1885). Hawai[i’s annexation by the USA in 1898 resulted in restrictions on Chinese immigration (due to the Chinese Exclusion Act), after which plantation owners turned first to Koreans and Puerto Ricans (1903–5) as replacements, and later to Filipinos (1907). In all, approximately 350,000 immigrants came to work on the sugar plantations. As a result, in a few decades there came into existence a multilingual community that far outnumbered the population of Hawaiians and Haoles (whites). Turning to the linguistic situation, from the 1800s onwards a minority of indigenous Hawaiians spoke English at varying levels of proficiency, forming

Creole Genesis 221 a continuum of “interlanguage varieties” of English. The speech form situated at the pidginized end of this continuum was called hapa-haole ‘half-white’, mainly spoken in the towns, and influenced to some extent by neighboring Pacific pidgins such as Chinese Pidgin English and Beach-la-Mar. It further appears that there was more than one contact language in use among the immigrant workforce on the Hawaiian plantations: In addition to a widely spread pidgin based on Hawaiian (called olelo pa[i[ai ‘taro-language’) an English-based pidgin, Hawaiian Pidgin English (HPE), developed during the last decades of the nineteenth century (see J. Roberts 1995a, 1995b, who documents its emergence; see discussions of HPE on the basis of recordings and interviews conducted in the 1970s in Bickerton & Odo 1976; Bickerton & Givón 1976; Bickerton 1977). It is not clear to what extent this English-based pidgin was a continuation of the Hawaiian-based pidgin or whether it emerged independently alongside it, but due to the vitality of the Hawaiian-based pidgin the structural development of HPE was delayed (Bickerton & Wilson 1987; S. Roberts 1998), causing HPE to be rudimentary in structure and highly variable at the turn of the twentieth century, best characterized as a jargon in the terminology of Bakker (this volume). HPE was spoken by the immigrant part of the population. In the by now ethnically diverse locally born population in the early part of the twentieth century, another language variety emerged: Hawaiian Creole English (HCE). By the 1930s, some observers recognized the HCE of the native-born population as a variety distinct from the HPE of the foreign-born population, e.g., Kaapu (1937).

3 The Hypothesis There are a few papers by Bickerton that precede Roots of Language. Bickerton (1980 [1974]) compares HCE, Saramaccan, Guyanese Creole English, and Haitian Creole French, and discusses their similarities in, for example, TenseMood-Aspect (TMA) marking. Essentially, this paper provides a foretaste of what is to follow. Bickerton (1977) is a discussion of the relation between creolization and language acquisition, equating pidginization with second language acquisition with reduced input, and creolization with first language acquisition with reduced input. From the early 1980s onwards, one can distinguish three phases in the development of the LBH; below, we will consider these in turn.

3.1

1981: Roots of Language

In Bickerton (1981) the main focus was on semantic and, to a lesser extent, syntactic features of a genetically transferred and species-determined linguistic bioprogram, exemplified in the first place by the case of HCE and enhanced by a cross-creole comparison. The structure of Bickerton’s argument in Roots of Language is remarkably straightforward and transparent. He shows that

222

Tonjes Veenstra

a number of semantic and syntactic differences exist between HCE and its ancestor HPE: (1) a. b. c. d. e.

word order (and movement rules) (pp. 17–22); the determiner system (pp. 22–6); the TMA system (pp. 26–30); sentential complementation (pp. 30–3); relativization strategies (plus subject-copying) (pp. 33–7).

These differences, he argues, cannot be due to the contributing superstrate language (English), or to the substrate languages, which, according to Roberts (1998, p. 2), include Austronesian (Hawaiian, Micronesian, Ilocano), Asian (Chinese, Japanese, Korean), and European (Portuguese, German, Norwegian) languages. Bickerton’s choice of Japanese and the Philippine languages as the major substrate languages is based on his observation that their speakers constituted the two largest immigrant groups (but see section 4.4 for discussion). These and other features set HCE apart from its precursor HPE. We will briefly summarize Bickerton’s findings for some of the features in (1) below. Bickerton claims that there are only two possible explanations for these innovations in HCE: one is that they are produced by a general problem-solving device, the other that they are the result of an innate bioprogram of linguistic competence. In order to decide between the two alternatives, he undertakes a cross-creole comparison of the five features in (1) plus the seven features in (2), using Guyanese, Jamaican Creole, Papiamentu, Sranan, Saramaccan, Haitian, Lesser Antillean French-lexifier Creoles, Mauritian Creole, Crioulo, Sao Tomense, and Papia Kristang data: (2) a. b. c. d. e. f. g.

negation; the use of the same form for possessive and existential; copula constructions; adjectives as a subclass of verbs; yes/no questions; question words; passives as lexical diathesis.

Bickerton argues that “if all creoles could be shown to exhibit an identity far beyond the scope of chance, this would constitute strong evidence that some genetic program common to all members of the species was decisively shaping the result” (1981, p. 42). After his discussion of these features (pp. 51–72), Bickerton concludes that “out of the twelve areas, HCE is identical with all or with a large percentage of creoles [i.e. those under discussion] in eight, shows a fair degree of similarity in two, and differs sharply in two” (p. 72). This is summarized in table 9.1. HCE presents a genuine problem for those who emphasize substratal factors in creole genesis, since HCE – which completely lacks a West African substrate

Creole Genesis 223 Table 9.1 HCE and other creoles HCE displays substantial similarity with other creoles

HCE displays substantial similarity with some creoles

HCE displays little similarity with other creoles

word order (and movement rules) the determiner system TMA system sentential complementation the same form for possessive and existential adjectives form a subclass of verbs yes/no questions passives as lexical diathesis

copula constructions

negation

relativization strategies

question words

– includes in its grammar some of the features found in Caribbean creoles that have been traced to West African languages. These include the TMA system, for-complementation, verbal adjectives, and passives as lexical diathesis (e.g., Bentolila 1971; Alleyne 1980; Holm 1988; Lefebvre 1996). A substrate account needs to account not only for the similarities between creoles with unrelated substrates, but also for the absence of those features in the substrate languages of HCE (Bickerton 1987, p. 233; Roberts 1998, p. 2). Bickerton also points out a methodological problem for the substrate accounts of the 1970s and early 1980s: the Cafeteria Problem. It refers to a methodologically unsound procedure used to “demonstrate” influence from particular West African languages in the Atlantic creoles. As Arends, Kouwenberg, and Smith (1995, p. 100) aptly summarize, “in its extreme form the method consists simply of picking some element present in the creole at hand, and then going through as many dictionaries and grammars of West-African languages as necessary until some more or less plausible correspondence is found.” Because of the number of different languages in West Africa, it is simply a matter of chance that sooner or later some apparent correspondences will be encountered. Instead, Bickerton argues that these features arise from an innate bioprogram of linguistic competence triggered by the lack of adequate input in a situation of communicative need – a context which arises due to the extreme social conditions present in plantation colonies. According to him, creoles are the

224

Tonjes Veenstra

results of native language creation on the basis of a rudimentary and variable pidgin (such as HPE) by children. He thus envisions creole formation as first language acquisition and claims that creoles arose abruptly, in a single generation, as a consequence of nativization. Note, however, that this scenario is claimed to apply only to those cases of language contact in which a pidgin has not existed for more than a generation (thereby ruling out cases like Tok Pisin) and where not more than 20 percent of the population spoke the dominant (colonial) language, the remaining 80 percent speaking a variety of other languages.

3.1.1

The creole features of Bickerton (1981)

We now turn to some of the features in (1). With regard to word order, there is a substantial amount of variation in the HPE data, largely attributable to substrate influence. Bickerton reports that speakers with a Japanese background had between 30 and 60 percent SOV sentences in their HPE speech (Japanese being a strict SOV language). The HPE data of speakers with a Filipino background had between 15 and 50 percent VS sentences when the subject was a full NP (the languages of the Philippines being largely VSO). VS structures in the data from Japanese speakers and SOV structures in those of Filipino speakers did occur occasionally, but were overall extremely rare. Whereas word order patterns in HPE are highly variable due to different basic word orders within different subgroups of speakers, the basic syntax of HCE is strictly SVO for all speakers regardless of ethnic background, exceptions being due to (mostly leftward) movement of constituents for information-structural reasons. Determiners are used sporadically and unpredictably in the HPE data, with speakers of Japanese descent undergeneralizing them, and Filipino speakers overgeneralizing them in comparison to standard English. HCE does not follow either of these patterns. Instead, it makes a three-way distinction: (1) the definite article da is used for specific-reference NPs assumed to be known to the listener; (2) the indefinite article wan is used for specific-reference NPs assumed to be unknown to the listener (typically, first mention); (3) all other NPs have no article, including generic NPs, NPs within the scope of negation, and cases where, while a specific reference may exist, the exact identity is unknown to the speaker or irrelevant. Thus, the innovation consists of the (semantic) specific/nonspecific distinction, a basic ingredient of HCE grammar. HPE does not depend on preverbal TMA auxiliaries for expressing temporal/aspectual relations but uses mainly sentential adverbials such as before (past tense), by and by (future modality), more better (obligative modality), little more (imminent modality), and all time (habitual aspect), while progressive and habitual aspect are also indicated via verbal repetition. The use of these adverbials in the HPE data is unsystematic and optional, however. HCE, on the other hand, depends primarily on the systematic and obligatory use of preverbal auxiliaries such as bin (anterior), go(n) (future and irrealis modality), and stei (non-punctual). Of these, bin and go were also attested in HPE, but in HCE they are used to a much greater extent. The aspectual marker stei, wen as

Creole Genesis 225 a variant of bin, and combinations of TMA auxiliaries were not attested in the HPE data. In addition to the obligatoriness of the TMA markers, there are two innovations in the TMA system. One is semantic in nature and concerns the introduction of the punctual/non-punctual distinction in the aspectual domain marked by stei; the other is syntactic, introducing combinatorial possibilities in the TMA system. Whereas sentence embedding is virtually nonexistent in the HPE data, HCE has sentential complements introduced by fo and go. In HPE, in addition to its use as a main verb, go is primarily attested as a marker of the imperative, while fo is mainly used as a preposition. In HCE, on the other hand, both are used as complementizers, go marking realized events and fo unrealized events. Thus, the innovation in HCE has two parts: The syntactic innovation consists of taking what are basically a preposition and an imperative marker and turning them into complementizers. The semantic innovation is that the two complementizers distinguish between realized and unrealized complements. In addition, Bickerton (1981) discusses perception and causative verb constructions in order to corroborate the claim that creoles only exhibit finite complementation. He also presents serial verb constructions, arguing that they are repair strategies for Case-marking purposes. Although in principle not ruled out as cases of substrate influence in the Atlantic creoles (1981, p. 131), he notes that in HCE some serial verb constructions occur in the speech of older informants. The point he makes is that HCE does not have a serializing West African substrate, thereby weakening a substrate explanation for serialization in the Caribbean creoles.

3.2

1984: Behavioral and Brain Sciences

The LBH in the second phase, highlighted by the Behavioral and Brain Sciences target article (Bickerton 1984a), incorporates sociohistoric and demographic factors pertaining to creole genesis into the model, leaning heavily on the seminal work of Baker and Corne (1982). Bickerton admits that the surfacing of the bioprogram could be inhibited by greater influence of the superstrate language. He weighs the differences in the historical process of colonization and the greater/lesser presence of the superstrate in order to account for a larger and more heterogeneous group of creole languages that constitute “a continuum of creolization rather than the typologically homogeneous body suggested in some earlier work” (1984a, p. 178). More specifically, Bickerton proposes to quantify demographic factors by means of a Pidginization Index (PI): (3) PI = Y × P/R Y = the number of years between the start of the colony and the point at which substratum and superstratum populations achieved numerical parity; P = the total substratum-speaking population at this point; R = the yearly average of enslavements after this point.

226

Tonjes Veenstra

Singler (1990) is an in-depth study of the sociohistorical profiles of Haiti and Mauritius showing that the PI does not (and cannot) work and should be put to rest, a point which Bickerton (1992, pp. 307ff.) admits. Singler (1990) takes the strongest possible position, claiming that the factors affecting pidginization simply cannot be quantified. The LBH is explicitly linked to the Chomskian principles and parameters approach (cf. Chomsky 1986), in which Universal Grammar (UG) is defined in terms of a set of invariable principles and a set of parameters, each of which have a small number of possible settings. Different combinations of the settings of the various parameters yield all possible grammars of human languages. The parameters are set on the basis of the Primary Linguistic Data, i.e., the input the language-acquiring child is exposed to. Bickerton argues that due to its rudimentary and variable character, a pidgin does not provide adequate input in the creole situation; thus, radical (or plantation) creoles instantiate the unmarked options of parameter settings of UG. With regard to the role of substrate influence, and elaborating on the earlier noted Cafeteria Problem, he emphasizes the need for substratists to demonstrate that speakers of a claimed substrate language were in the right place at the right time, as well as to provide plausible mechanisms for transfer (Bickerton 1984a, p. 183; Siegel, this volume). He concludes, as he did before, that substrate influence is inadequate as an account for the creation of creole languages and their central properties.

3.3

1988: The Cambridge Linguistics Survey

In the third phase, although still claiming that creoles offer a special window on the human language faculty, Bickerton (1988) is no longer mainly focused on the issue of the extent to which the study of creole languages can shed light on the origins of language. Instead, he concentrates on providing a comprehensive theory of creolization. He argues that the different types of contact situations in which creole languages emerged seem to fall into three fairly broad sociolinguistic categories, each leading to different linguistic outcomes. These are defined by the two parameters shown in table 9.2. The three types of resulting creole languages do not form a homogeneous class and the main difference can be summarized as follows: “plantation” creoles will differ more radically from their superstrates than “fort” creoles, and “maritime” creoles will show more substratum influence than either of the other two types. Table 9.2

Types of contact situations

Permanent population displacement First-generation nativization

Plantation

Fort

Maritime

+ +

– +

– –

Creole Genesis 227 Furthermore, he embraces as the basic grammatical framework of the LBH the “lexical learning hypothesis” of Borer (1983), according to which there is one syntax underlying all languages, parametric variation being confined to the functional part of the lexicon. Thus, the syntax of a language largely depends on the feature specifications of the function words of that language. In this last phase, the bioprogram is completely identified with UG, and consists of those minimal grammatical functions that must be discharged in order for a new language (i.e., creole) to be an adequate native language. This shift in focus to the lexicon leads him to reinterpret creole genesis in terms of three processes: retention, loss, and reconstitution of functional (and lexical) elements. Only those grammatical elements and the functions associated with them that must, if lost, be reconstituted are part of UG. Although still skeptical about substrate influence in general, he implicitly sets the door ajar in those cases when an actual substrate language function word has been retained in a creole language. If that is the case, one can expect its associated syntax. (See Smith 1996 for a clear and convincing illustration.)

3.4

A first assessment

Some basic aspects of the LBH have remained constant during this development. First, a highly unstable and variable jargon is assumed to be the predecessor of creole languages (of which the HPE/HCE pair is the prototypical example). Second, Bickerton sees creolization as a catastrophic process, as opposed to a gradual one. Third, he sees nativization as the key factor in creolization. Finally, he downplays the role of substrate languages, despite extensive bilingualism in Hawai[i (and, therefore, possible cross-linguistic interference). Other aspects of the hypothesis have changed slightly over the course of time. I will mention two of them. First, in the first two phases of the LBH’s history, Bickerton adheres to the “pidgin-creole cycle” of Hall (1966) and uses a pidgin metaphor for the ancestor of creoles. Thus, Bickerton (1981, p. 4) considers a language to be a creole “if it arose out of a prior pidgin which had not existed for more than a generation.” In the next phase, he widens the definition to include languages “that have come into existence very early in the development of their antecedent pidgins” (Bickerton 1984a, p. 173). In the third and last phase, he completely sides with Alleyne’s (1971, 1980) position that there need not have been a pidgin stage in the development of creole languages (Bickerton 1988, p. 272). Instead, he introduces a dilution metaphor: the first cohorts of slaves still acquired a fairly adequate version of the superstrate language, but succeeding cohorts acquired ever more diluted versions of it, because of diminishing access to the target language due to a number of sociohistorical factors (this is comparable to the “approximations of approximations” model in the work of Chaudenson 1992). Second, although creolization is still seen as a sudden process, the assumption of Bickerton (1984a) that creolization is a single-generation process has

228

Tonjes Veenstra

been replaced by the assumption that creolization is a type of language formation where “new languages are produced ab ovo within the space of, at most, one or two generations” (1988, p. 268). This unmotivated addition of “or two” appears to turn the LBH into a circular hypothesis, perhaps even an unfalsifiable one. In the original formulation of creolization as a unigenerational process, it is unambigously clear that the innovations introduced by the first locally born generation must be those properties reconstituted in order to create a fullfledged language. In other words, these must be the basic properties of natural language. If, on the other hand, some of these properties can be introduced in the following generation, they cannot be part of the basic set, because if they were, they would have been introduced by the first generation. Alternatively, the implication would be that the first generation was not in possession of any full-fledged language. This surely cannot be so, given the well-documented case of Nicaraguan Sign Language (Kegl, Senghas, & Coppola 1999) where the first generation of native speakers introduced substantial innovations, turning it into a viable and full-fledged language. As it turns out, there are two issues here. The first is that the first locally born generation in Hawai[i was bilingual in the pidgin and one of the ancestral (i.e., substrate) languages, a possibility which is explicitly denied in Bickerton’s post-1981 work (e.g., Bickerton 1984b, p. 145; 1988, p. 302; but see 1977, p. 333). The second is that the second locally born generation is the first generation of native speakers. From these facts, it follows that creolization is not a unigenerational process. We will return to these issues in section 4.4 when we take a more detailed look at the sociolinguistic situation of Hawai[i as described in Roberts (2000).

4

Critiques

The LBH has been criticized from many perspectives. A number of scholars have argued that the properties of particular creoles do not fit the generalizations of the LBH. Some critiques have addressed Bickerton’s account of the relation between HPE and HCE; a counterclaim that the ancestor pidgin of HCE was not HPE, as Bickerton claims, but another structurally more complex pidgin has been put forward. A critique has been levied also against Bickerton’s views of the historical demographic context of creolization, arguing that nativization does not play a decisive role in creolization. Finally, the claim that a full-fledged language emerges in a catastrophic, unigenerational processs has been criticized. We will consider each of these issues in the following sections.

4.1

The prototypical creole features

Features that have been put under scrutiny include the determiner system (cf. Bruyn 1995; Janson 1984; Lucchesi 1993; Mufwene 1986a), the TMA system (cf. Bruyn & Veenstra 1993; Gibson 1986; Lefebvre 1996; Maurer 1985; Muysken

Creole Genesis 229 1981; Singler 1990; Winford 1993, 2000a, 2000b), verbal adjectives (cf. Sebba 1986; Seuren 1981, 1986; Winford 1997), the copula system (Arends 1989; DeGraff 1992; Arends, Muysken, & Smith 1995b), and the finiteness issue (Mufwene & Dijkhoff 1988; Kouwenberg 1990; Veenstra 1996). Over the years, it has become more and more questionable whether there exists anything like a consistently uniform creole syntax along the lines of the (1981) formulation of the LBH. Muysken (1988) argued strongly against the implicit claim that there is something like a prototypical syntax of creoles. As he notes, although creoles are remarkably similar with respect to word order, preverbal particle order – although not so much with respect to the semantic content of the TMA elements – and the paucity of inflectional morphology, they also vary, for instance with respect to the range of functions expressed by serial verb constructions and the permissibility of preposition stranding. To this, one can add the variation between creoles with respect to predicate cleft constructions (Veenstra & Den Besten 1995), and (subject) clitics and pro-drop phenomena (DeGraff 1993; Déprez 1994; Kouwenberg 1990; Muysken & Veenstra 1995; Syea 1993; Veenstra 1996). Thus, there seems to be a tension between unity and diversity of linguistic structures in creoles – a fact that creolists have to come to grips with. We have to set up scenarios specific enough to master this apparent contradiction. Sure enough, such a scenario cannot be unidimensional in nature, but should involve a multidimensional model of creole genesis (Mufwene 1986b; Corne 1994; DeGraff 1999a; Kihm 2000; Muysken 2001; Becker & Veenstra 2003). Still, there is no doubt that the very existence and overall similarity of properties such as the preverbal particle systems and forcomplementation are in need of an explanation by any theory of creole genesis.

4.2

The pidgin ancestor of HCE

In order to account for some of the similarities between HCE and Englishbased creoles in the Carribean, Holm (1986, 1989) proposes a model of diffusion which involved the transmission of a structurally complex pidgin (or creole) from the Caribbean to Hawai[i (see also Dillard 1995; Goodman 1985; Keesing 1988; McWhorter 1994; Meyerhoff, this volume). This pidgin, he claims, is the ancestor of HCE, rather than HPE. Roberts (1998) has clearly spelled out the predictions this model makes for the linguistic situation in Hawai[i: If early nineteenth-century sailors brought an Atlantic pidgin-creole to Hawai[i . . . then attestations of early-nineteenth century HPE should more closely resemble CEC’s [Caribbean English Creoles] than nearby Pacific pidgins. We would not expect to find them documenting a pidgin that developed CEC-like features much later. These features would include HCE functional categories such as TMA categories and complementizers, as well as other complex features. We would also not expect to find CEC-like features correlated with the locally born population. (p. 4)

230

Tonjes Veenstra

On the basis of an extensive database consisting of a wide range of different pidgin/creole texts covering the nineteenth and twentieth centuries, Roberts argues that none of these predictions is borne out by the data. Textual evidence shows that nineteenth-century HPE not only lacked the structure of later HCE, but also displayed stronger links with English-based pidgins from the Pacific than with the English-based creoles of the Caribbean. Furthermore, the HCE TMA system (pp. 18–26) as well as for-complementation (pp. 26–32) developed late and primarily within the population of locally born speakers. Thus, she finds little or no support for Holm’s model of diffusion. On the other hand, the patterns laid bare are grosso modo in line with the LBH.

4.3

The role of nativization

Another set of criticisms came from the perspective of specialists of the early demographic history of Caribbean creole societies. For instance, Singler (1992) points out that, due to the skewed nature of the population (i.e., the low proportion of women to men), high mortality rates, and the low cost of importing new slaves from Africa, the birth rates in these plantation societies were extremely low. Singler (1996), however, alters his (1992) position, saying that more women and children were present in early Caribbean plantation societies than he had originally thought (see also Arends, this volume). However, even with these adjustments to the age- and sex-based population numbers, Singler continues to argue against nativization as the primary mechanism in creolization. He asserts that the larger the proportion of children present in early Caribbean plantation societies, the closer the result is to the lexifier language; this, he claims is evidence against nativization (1992, 1996). Roberts (2000) has presented evidence against this claim which we will review in section 4.4 below. A further problem with Singler’s argument stems from confusion with respect to the nature of the concept of “nativization.” For Bickerton, nativization is basically psycholinguistic in nature (and, hence, at the level of I-creolization), whereas for Singler it is more sociolinguistic in nature (hence, at the level of E-creolization). (See DeGraff (1999a) for a careful discussion of the need for distinguishing the notions I-creolization and E-creolization in general.) Similar objections to the LBH’s emphasis on nativization come from Arends (1989, 1993, 1994), who claims that for the Suriname colony during the first hundred years of its existence, “no creolization – in the traditional view of nativization by children – could possibly have taken place simply because the children required to perform this process were not present” (1993, p. 375). Apart from the fact that we don’t know how many children would suffice to introduce a creole language, Arends (1995) produced new figures which caused him to significantly revise his estimate of the number of children in the early history of the Suriname colony upwards. From these new estimates, Smith (2001, p. 51) concludes that “it is clear that we do not have much basis for claiming that at some earlier period there were too few children for the LBH to operate [in Suriname]. The question that we must answer is rather, when

Creole Genesis 231 could it have operated.” The work of Jennings (1995, pp. 38–40) on the early demographic data from the Rémire plantation in Cayenne show that in 1685, only 25 years after the first slaves arrived from Africa, at least 28 percent of the plantation’s population and 21 percent of the colony’s population as a whole was locally born. Evidence from more recent contact language situations – mainly in the Pacific but not restricted to it – points to the need to distinguish between two types of scenarios, one in which nativization plays a substantial role, and another where nativization is of less (or no) importance. Thus, on one hand, as already argued above, it is clear from the studies of Roberts (1998, 1999) that the locally born population has been an important factor in the formation of HCE. On the other hand, Tok Pisin, or any of the Melanesian Pidgins (Tok Pisin being a descendant of a pan-Pacific pidgin that most probably originated during the nineteenth century; cf., e.g., Keesing 1988), does not seem to have undergone initial nativization until the 1950s (Sankoff & Laberge 1974). The issue then boils down to the question which model is more relevant for the Caribbean situation. An important difference between the Hawaiian situation and the Melanesian situation seems to be the sociohistorical context. Whereas in both situations indentured labor on plantations played a pivotal role, the crucial difference is that in Hawai[i the indentured population remained in the islands after serving their contracts, forming a new group within the population of the islands. In contrast, in Melanesia, they returned to their homelands, where they reintegrated (if they had survived the three- to fiveyear contract period, because mortality rates were high on the Queensland plantations). Romaine (1988) takes this to be the crucial difference between the two situations. Bickerton (1992, p. 310) concludes from this difference that creolization (in terms of nativization) took a long time in Melanesia, because there was no society (and presumably no individual) that needed the emerging contact language as a native language. The Caribbean situation seems to share its sociohistorical profile with the Hawaiian situation, in particular with respect to the permanency of population displacement. The conclusion that Bickerton drew from the Hawaiian case, that nativization can play a crucial role in the emergence of new languages, has been corroborated by recent research on sign languages. There are cases in which the target for language acquisition comes exclusively from non-native speakers (e.g., family home-sign systems or gestures that accompany spoken languages). In such cases, the language varieties that the younger learners develop outdo the predecessors on which they are modeled (cf. Newport 1999 on American Sign Language). The most illuminating illustration of such a scenario is provided by Kegl, Senghas, and Coppola (1999) in their documentation of the evolution of Nicaraguan Sign Language.

4.4

Abrupt creolization?

The classical LBH view is that creolization is a catastrophic unigenerational process (but see 3.4 for a caveat). This means that creoles result when the first

232

Tonjes Veenstra

locally born generation of children develops a vehicle for interethnic communication on the basis of a rudimentary pidgin where no other language would suffice for this purpose. Roberts (2000), draws on censuses, immigration, and school reports, as well as birth statistics, to provide a demographic history and a description of the sociolinguistic situation during the formative years of HCE. Her findings contest Bickerton’s “deprivation” scenario. She presents data from a corpus of 1,500 life histories written in the 1920s by schoolchildren in response to a questionnaire that had been designed to elicit data on, among other things, language use in the community and attitudes toward ancestral languages, i.e., the potential substrate languages for HCE. She shows that these data invalidate three central assumptions of the LBH: (4) a.

that the usefulness of the ancestral language declined rapidly (e.g., Bickerton 1981, p. 5); b. that the first generation of locally born children, even if exposed to their ancestral languages, did not acquire them (e.g., Bickerton 1988, p. 302); c. that the first generation of locally born children created the creole by acquiring their parents’ pidgin (e.g., Bickerton 1981, p. 5).

The evidence from these life histories shows that foreign-born parents did not refrain from using their ancestral languages with their children (contra (4a)). Members of the first locally born generation were mostly bilingual in the incipient creole and in their ancestral languages (contra (4b)), although fluency in the latter was receding in a growing number of cases – a signal of the first stages of language attrition due to a shift to the creole. The incipient creole was used regularly between siblings and was taught by the older ones to the younger ones. It was also increasingly used in parent–child communication, although the children were most often the ones initiating its use (contra (4c)). Summarizing, first-generation children received far more pidgin input from their peers than from their foreign-born parents. It was, therefore, the secondgeneration children who were in a position to learn the incipient creole from their parents, not the first generation, as in Bickerton’s scenario. The main problem with the LBH, Roberts (2000, p. 292) argues, is that it forces two processes onto one generation, namely what she calls “creole formation” (the process of expansion) and “nativization” (the process of innovation). Instead, she proposes that the creolization process in Hawai[i must have proceeded in two distinct phases involving two different generations, as originally anticipated by Reinecke (1969, p. 166). In the first (expansion) phase, before 1900, the locally born population – largely made up of the first locally born generation of Portuguese and Chinese – recruited a number of grammatical items of HPE, and integrated them into the newly emerging language. In the second phase, situated after 1910, the locally born population – now largely made up of the first locally born generation of Japanese and the second locally

Creole Genesis 233 born generation of Portuguese and Chinese – innovated new features, thus distinguishing this emerging language even more from HPE as spoken by the foreign-born population. Thus, nativization does not occur in the first generation of the locally born population as the LBH would have it, but rather in the second. Despite the growth of the locally born population, HCE did not conform to the superstrate, but instead developed basilectal features, contrary to the predictions of Singler (1992). Roberts (2000, p. 294) argues that the strong stigma attached to the use of the superstrate in in-group relations diminished the influence of the superstrate, and promoted the development of basilectal features which distinguish HCE not only from its superstrate but also from the pidgin of the foreign-born. Furthermore, due to the bilingualism of the first generation of locally born children – those responsible for the expansion phase – the possibility of substrate influence in the formation of HCE returns to the research agenda. But if substrate influence is involved, why did Bickerton (1981) not find influence on the HCE word order patterns from Japanese (head-final) and the Philippine languages (verb-initial), whose speakers constituted the two largest immigrant groups? The absence of influence from the Philippine languages can be accounted for by invoking the “Founder Principle” of Mufwene (1996). Since HCE was well established in the domain of cross-ethnic interactions by 1910 (as argued by Roberts) and substrate influence was mainly due to the first generation of locally born children, the Filipinos simply appeared too late on the scene (the first cohort arrived in 1907). However, it seems that the Founder Principle does not provide a straightforward account for the absence of Japanese influence, since the Portuguese and Japanese entered the scene only a few years apart (1878 vs. 1885, to be precise). The answer seems to lie in the composition of the first group of locally born and differences between the Portuguese and Japanese in their rates of bilingualism. As Bickerton (1984a, p. 183) noted, the distribution of bilinguals was skewed ethnically, the Japanese group having the greatest number of bilinguals. The Japanese also showed the greatest amount of substrate language maintenance, followed by the Chinese, and finally by the Portuguese. If we look at language shift, however, the order is reversed: the Portuguese shifted away from their ancestral language first, followed closely by the Chinese, and only later by the Japanese (Roberts 2000): (5) order of arrival level of ancestral language maintenance order of language shift

Chinese > Portuguese > Japanese Japanese > Chinese > Portuguese Portuguese > Chinese > Japanese

To account for the absence of Japanese influence, then, the idea is that early shifting groups exert the most influence on the emerging new language. In the first phase of creolization in Hawai[i, the locally born population was largely composed of the first locally born generation of Portuguese and

234

Tonjes Veenstra

Chinese. Therefore, these are the languages which could have had a major impact on HCE. Only in the second phase, after 1910, when HCE was already well established in the domain of interethnic interactions, did the locally born Japanese first emerge on the creolization scene. At that point, they were confronted with an already largely stabilized HCE; therefore, they could not exert their influence on the language as strongly. If this scenario turns out to be correct, it would mean that in the realm of language shift and nativization, founder effects also play a role. Initial evidence for such a scenario has been presented in Siegel (2000), where it is argued that several of the major characteristics of HCE match properties of Chinese and Portuguese (see also Siegel, this volume). Future research will have to show in more detail to what extent substratal influence shaped the form of HCE. From the discussion above, it is clear that the claim of the LBH that creolization is a unigenerational process cannot be upheld. Instead, the process of creolization of HCE is spread out over two generations. This brings the Hawaiian situation, and creole genesis in general, more in line with other – albeit less dramatic and drastic – language shift situations (cf. Veltman 1983). Furthermore, the locally born population did not create HCE in the face of insufficient input. Finally, substrate influence played a role in the creation of HCE, and such influence is subject to founder effects in language shift and nativization.

5

Conclusion

The major implication of the discussion in the last subsection is that one has to assume a multi-generational scenario for creole genesis. Such a scenario was first proposed in Corne (1994) for Tayo. Further evidence for this approach to creole genesis has been adduced by Jennings (1995) for Cayenne Creole and Becker & Veenstra (2003) for Mauritian Creole. The picture of the shift to creole is then as follows: (6) Multi-generational scenario of creole genesis (Becker & Veenstra 2003) G1 (foreign-born immigrants) L1 ancestral language(s) L2 pidgin G2 (first generation locally born)

L1 L1

ancestral language(s) pidgin/creole

G3 (second generation locally born)

L1 creole (L2 ancestral language(s))

In this scenario, the first generation (G1) consists of immigrants/slaves who pidginize the superstrate. The next generation (G2) is that of the locally born children of G1. According to Corne (1994) this is the crucial generation. As he puts it, writing about the emergence of Tayo:

Creole Genesis 235 Members of G2 retain their ancestral L1 . . . but, like the children of immigrants everywhere, they have to acquire the language of the community as a whole. Their problem is that there is no community L1 apart from pidginized L2 French, and they must therefore acquire/create L1 competence on the basis of their exposure to (some subset of) the varieties of pidginized L2 French, variable over time and depending on personal circumstances . . . It is they who participate in the creation of the nascent creole. (Corne 1994, p. 296, emphasis as in original)

Members of the third generation (G3) are only indirectly exposed to their ancestral language(s), due to the presence of G1 and G2 speakers in their community; having largely lost their ancestral language(s), they are the first monolingual speakers of the incipient creole. Thus, the process of creolization is driven by two types of agents, namely the first and second generations of locally born children. Therefore, it involves both bilingual language acquisition by G2 as well as first language acquisition by G3 (in addition to second language acquisition by the adult (foreign-born) G1 population). Some other important outcomes of this chapter’s discussion are that creole grammars do not consistently display the features of the bioprogram, as proposed in Bickerton (1981), that creolization is not the result of linguistic deprivation, and that substrate influence cannot be ruled out as a significant factor in the creation of HCE in particular and of creoles in general. However, as McWhorter (2002, p. 39) correctly points out, “none of these revisions belie Bickerton’s focal point that creoles are in some fashion more directly reflective of the heart of language than older ones [i.e., older languages].” In other words, Bickerton’s position that creoles do offer a special window on the human language faculty is maintained. What we are left with is not so much the strong version of the LBH, which has it that creoles are inventions of children growing up on the newly formed plantations, but rather a weak version of it, in which the human language faculty guides the selection or regularization of grammatical properties from the environment. For instance, in the model of DeGraff (1999a, p. 525) children are vital to the creolization process in transforming the input from adults into a stable, consistent, and fully developed linguistic system. As such, nativization still plays a role in the debates on creole genesis. I close this chapter with the observation by Kouwenberg (2000, p. 12) that, “while strong viewpoints are often at the center of controversy, and encounter harsh criticism (even more so if they are formulated in an uncompromising and exclusionist manner), they are also the much-needed catalysts of progress in ways that consensus views are not.” Although the strong view that LBH holds with respect to creole genesis cannot be upheld in its totality, nevertheless, the LBH has been instrumental in motivating research on the properties of creole grammars, and especially in leading the creolist community to many new insights regarding the emergence of creole languages. In this sense, then, the LBH has proved to be a highly successful hypothesis.

236

Tonjes Veenstra

FURTHER READING The three most important reviews of Bickerton (1981) are Hazaël-Massieux (1982), Muysken (1983), and Goodman (1985). Bickerton (1984a) is accompanied by a set of peer commentaries from different disciplines; they are compulsory reading for anyone interested in the relevance of creole studies for nativist approaches to language. Bickerton (1988) is followed by a dialogue concerning the linguistic status of creole languages between Derek Bickerton and Pieter Muysken. Few studies are devoted to the acquisition of creole languages: Adone (1994) is a cross-sectional study of the syntactic L1 development in Mauritian Creole, and Adone and Vainikka (1999) is a study of the acquisition of wh-movement in Mauritian Creole. Meade (2001) on Jamaican Creole is the first in-depth phonological study of creole L1 development. DeGraff (1999b) is the most recent collective volume linking creole studies to issues of language acquisition, development, and language change; the editor’s carefully reasoned epilogue (1999a) is highly recommended. Holm and Patrick (forthcoming) undertake an extensive cross-creole comparison. In McWhorter (1998) the notion of “prototypicality,” this time in the realm of morphology, is re-introduced in creole studies, whereas McWhorter (2001) puts the issue of complexity of creoles on the research agenda again, attributing complexity to the “bric-a-brac” of older languages, basically following Bickerton (1984a) in this regard. Although not covered in this chapter, some recent key references to the debate on the origin of language besides Bickerton (1990, 1995) and Calvin and Bickerton (2000) include Carstairs-McCarthy (1999), Deacon (1997), and Pinker and Bloom (1990). The following edited volumes survey a cross-section of modern theories, methods, and research pertaining to the origin and evolution of language: Christiansen and Kirby (2002), Hurford, Studdert-Kennedy, and Knight (1998), Knight, Studdert-Kennedy, and Hurford (2000), and Trabant and Ward (2001).

REFERENCES Adone, Dany (1994) The Acquisition of Mauritian Creole. Amsterdam: John Benjamins. Adone, Dany and Anne Vainikka (1999) Acquisition of wh-questions in Mauritian Creole. In: DeGraff (ed.), pp. 75–94. Alleyne, Mervyn (1971) Acculturation and the cultural matrix of creolization. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 169–86. Alleyne, Mervyn (1980) Comparative Afro-American. Ann Arbor: Karoma. Arends, Jacques (1989) Syntactic Developments in Sranan. PhD dissertation, University of Nijmegen, The Netherlands. Arends, Jacques (1993) Towards a gradualist model of creolization. In: Frank Byrne and John Holm (eds.) Atlantic Meets Pacific. Amsterdam: John Benjamins, pp. 371–80. Arends, Jacques (1994) The African-born slave child and creolization (a postscript to the Bickerton–Singler debate on nativization). Journal of Pidgin and Creole Languages 9, 115–19.

Creole Genesis 237 Arends, Jacques (1995) Demographic factors in the formation of Sranan. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 233–85. Arends, Jacques, Silvia Kouwenberg, and Norval Smith (1995) Theories focussing on the non-European input. In: Arends, Muysken, and Smith (eds.), pp. 99– 110. Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) (1995a) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins. Arends, Jacques, Pieter Muysken, and Norval Smith (1995b) Conclusions. In: Arends, Muysken, and Smith (eds.), pp. 319–30. Baker, Philip and Chris Corne (eds.) (1982) Isle de France Creole: Affinities and Origins. Ann Arbor: Karoma. Becker, Angelika and Tonjes Veenstra (2003) The survival of inflectional morphology in French-related creoles: The role of SLA processes. Studies in Second Language Acquisition 25, 283–306. Bentolila, Alain (1971) Les systèmes verbaux créoles; Comparaisons avec les langages africaines [The creole verbal systems: Comparisons with African languages]. PhD dissertation, l’Université de Paris V – René Descartes. Bickerton, Derek (1977) Change and Variation in Hawaiian English. Vol. 2: Creole Syntax. Honolulu: University of Hawai[i, Social Science and Linguistics Institute. Bickerton, Derek (1980) [1974] Creolization, linguistic universals, natural semantax and the brain. In: Richard Day (ed.) Issues on English Creoles: Papers from the 1975 Hawaii Conference. Heidelberg: Groos, pp. 1–18. (Originally published (1974) University of Hawaii Working Papers in Linguistics 6, 125–41.) Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984a) The language bioprogram hypothesis. Behavioral and Brain Sciences 7, 173–88. Bickerton, Derek (1984b) Creole is still king. Behavioral and Brain Sciences 7, 212–18. Bickerton, Derek (1987) Beyond roots: Knowing what’s what. Journal of Pidgin and Creole Languages 2, 229–37. Bickerton, Derek (1988) Creole languages and the bioprogram. In: Frederick Newmeyer (ed.) Linguistics: The Cambridge Survey, vol. 2. Cambridge: Cambridge University Press, pp. 267–84. Bickerton, Derek (1990) Language and Species. Chicago: The University of Chicago Press. Bickerton, Derek (1992) The sociohistorical matrix of creolization. Journal of Pidgin and Creole Languages 7, 307–18. Bickerton, Derek (1995) Language and Human Behavior. Seattle: University of Washington Press. Bickerton, Derek and Givón, Talmy (1976) Pidginization and language change: From SXV and VSX to SVX. In: Sanford B. Steever, Carol A. Walker, and Salikoko S. Mufwene (eds.) Papers from the Parasession on Diachronic Syntax. Chicago: Chicago Linguistic Society, pp. 9–39. Bickerton, Derek and Carol Odo (1976) Change and Variation In Hawaiian English. Vol. 1: General Phonology and Pidgin Syntax. Honolulu; University of Hawai[i, Social Science and Linguistics Institute. Bickerton, Derek and William Wilson (1987) Pidgin Hawaiian. In: Glenn Gilbert (ed.) Pidgin and Creole Languages: Essays in Memory of John E. Reinecke. Honolulu: University of Hawai[i Press, pp. 61–76. Borer, Hagit (1983) Parametric Syntax. Dordrecht: Foris.

238

Tonjes Veenstra

Bruyn, Adrienne (1995) Grammaticalization in Creoles: The Development of Determiners and Relative Clauses in Sranan. Amsterdam: Institute for Functional Research into Language and Language Use (IFOTT). Bruyn, Adrienne and Tonjes Veenstra (1993) The creolization of Dutch. Journal of Pidgin and Creole Languages 8, 29–80. Calvin, William and Derek Bickerton (2000) Lingua Ex Machina: Reconciling Darwin and Chomsky with the Human Brain. Cambridge, MA: MIT Press. Carstairs-McCarthy, Andrew (1999) The Origins of Complex Language: An Inquiry into the Evolutionary Beginnings of Sentences, Syllables and Truth. Oxford: Oxford University Press. Chaudenson, Robert (1992) Des îles, des hommes, des langues. Paris L’Harmattan. Revised English edition: Creolization of Language and Culture, London: Routledge. Chomsky, Noam (1986) Knowledge of Language: Its Nature, its Origin and Use. New York: Praeger. Christiansen, Morten and Simon Kirby (eds.) (2002) Language Evolution: States of the Art. Oxford: Oxford University Press. Corne, Chris (1994) Relativization and thematization in Tayo and the implications for creole genesis. Journal of Pidgin and Creole Languages 9, 283–304. Deacon, Terence (1997) The Symbolic Species: The Co-evolution of Language and the Brain. New York: W. W. Norton. DeGraff, Michel (1992) Creole Grammars and Acquisition of Syntax. PhD dissertation, University of Pennsylvania. DeGraff, Michel (1993) Is Haitian Creole a pro-drop language? In: Frank Byrne and John Holm (eds.) Atlantic Meets Pacific, Amsterdam: John Benjamins, pp. 71–90. DeGraff, Michel (1999a) Creolization, language change and language acquisition: An epilogue. In: DeGraff (ed.), pp. 473–543. DeGraff, Michel (ed.) (1999b) Language Change and Creation: Creolization, Diachrony and Development. Cambridge, MA: MIT Press. Déprez, Vivienne (1994) Haitian Creole: A pro-drop language? Journal of Pidgin and Creole Languages 9, 1–24. Dillard, Joey L. (1995) Letter to the editor (28 July 1994). Journal of Pidgin and Creole Languages 10, 219–20. Gibson, Ken (1986) The ordering of auxiliary notions in Guyanese creole. Language 62, 571–86. Goodman, Morris (1985) Review of Roots of Language by Derek Bickerton (1981) International Journal of American Linguistics 51, 109–37. Hall, Robert (1966) Pidgin and Creole Linguistics. Ithaca: Cornell University Press. Hazaël-Massieux, Guy (1982) Sur l’origine des créoles: Roots of language de Bickerton [On the origin of creoles: “Roots of Language” by Bickerton]. Etudes Créoles 5, 109–22. Holm, John (1986) Substrate diffusion. In: Pieter Muysken and Norval Smith (eds.) Substrata Versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 259–78. Holm, John (1988) Pidgins and Creoles, vol. 1. Cambridge: Cambridge University Press. Holm, John (1989) Pidgins and Creoles, vol. 2. Cambridge: Cambridge University Press. Holm, John and Peter Patrick (eds.) (forthcoming) Comparative Creole Syntax. London: Battlebridge. Hurford, James, Michael Studdert-Kennedy, and Chris Knight (eds.) (1998) Approaches to the Evolution of Language: Social and Cognitive Bases. Cambridge: Cambridge University Press.

Creole Genesis 239 Janson, Tore (1984) Articles and plural formation in creoles: change and universals. Lingua 64, 291–323. Jennings, William (1995) The first generations of a creole society: Cayenne 1660–1700. In: Philip Baker (ed.) From Contact to Creole and Beyond. London: University of Westminster Press, pp. 21–40. Kaapu, Myrtle (1937) A Study of the Influence of Japanese Syntax and Idiom upon the Spoken and Written English of a Group of Ninth Grade Pupils. MA thesis, University of Hawai[i. Keesing, Roger (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford: Stanford University Press. Kegl, Judy, Ann Senghas, and Marie Coppola (1999) Creation through contact: Sign language emergence and sign language change in Nicaragua. In: DeGraff (ed.), pp. 179–238. Kihm, Alain (2000) Are creole languages “perfect” languages? In: McWhorter (ed.), pp. 163–200. Knight, Chris, Michael Studdert-Kennedy, and James Hurford (eds.) (2000) The Evolutionary Emergence of Language: Social Function and the Origins of Linguistic Form. Cambridge: Cambridge University Press. Kouwenberg, Silvia (1990) Complementizer pa, the finiteness of its complements and some remarks on empty categories in Papiamento. Journal of Pidgin and Creole Languages 5, 39–52. Kouwenberg, Silvia (2000) Review of Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole, by Claire Lefebvre. Revue québécoise de linguistique 28, 129–36. Lefebvre, Claire (1996) The tense, mood and aspect system of Haitian Creole and the problem of transmission of grammar in creole genesis. Journal of Pidgin and Creole Languages 11, 231–311. Lucchesi, Dante (1993) The article systems of Cape Verde and Sao Tomé Creole Portuguese: General principles and specific factors. Journal of Pidgin and Creole Languages 8, 81–108. Maurer, Philippe (1985) Le système temporel du papiamento et le système temporel proto-créole de Bickerton [The temporal system of Papiamento and the prototypical temporal system of creoles as posited by Bickerton]. Amsterdam Creole Studies 8, 41–66. McWhorter, John (1994) Rejoinder to Derek Bickerton. Journal of Pidgin and Creole Languages 9, 79–93. McWhorter, John (1998) Identifying the creole prototype: Vindicating a typological class. Language 74, 788–818. McWhorter, John (ed.) (2000) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins. McWhorter, John (2001) The world’s simplest grammars are creole grammars. Linguistic Typology 5, 125–66 (followed by commentaries and author’s reply 167–412). McWhorter, John (2002) The rest of the story: Restoring pidginization to creole genesis theory. Journal of Pidgin and Creole Languages 17, 1–48. Meade, Rocky (2001) Acquisition of Jamaican Phonology. PhD dissertation, University of Amsterdam. Mufwene, Salikoko (1986a) Number delimitation in Gullah. American Speech 61, 33–60. Mufwene, Salikoko (1986b) The universalist and substrate hypothesis complement one another. In: Pieter Muysken and Norval Smith (eds.) Substrata versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 129–62.

240

Tonjes Veenstra

Mufwene, Salikoko (1996) The founder principle in creole genesis. Diachronica 13, 83–134. Mufwene, Salikoko and Martha Dijkhoff (1988) On the so-called “infinitive” in Atlantic creoles. Lingua 77, 297–330. Muysken, Pieter (1981) Creole tense/mood/aspect systems: The unmarked case? In: Pieter Muysken (ed.) Generative Studies on Creole Languages. Dordrecht: Foris, pp. 181–99. Muysken, Pieter (1983) Review article of Roots of Language, by Derek Bickerton. Language 59, 884–91. Muysken, Pieter (1988) Are creoles a special type of language? In: Frederick Newmeyer (ed.) Linguistics: The Cambridge Survey, vol. 2. Cambridge: Cambridge University Press, pp. 285–301. Muysken, Pieter (2001) The creole languages of the Caribbean. In: James Arnold (ed.) A History of Literature in the Caribbean, Volume 2: English- and Dutch-Speaking Regions. Amsterdam: John Benjamins, pp. 399–41. Muysken, Pieter and Tonjes Veenstra (1995) Universalist approaches. In: Arends, Muysken, and Smith (eds.), pp. 121–35. Newport, Elisabeth (1999) Reduced input in the acquisition of signed languages: Contributions to the study of creolization. In: DeGraff (ed.), pp. 161–78. Pinker, Steven and Paul Bloom (1990) Natural language and natural selection. Behavioral and Brain Sciences 13, 707–84. Reinecke, John E. (1969) Language and Dialect in Hawai[i: A Sociolinguistic History to 1935. Honolulu: University of Hawai[i Press. Roberts, Julian (1995a) Pidgin Hawaiian: A sociohistorical study. Journal of Pidgin and Creole Languages 10, 1–56. Roberts, Julian (1995b) A structural sketch of Pidgin Hawaiian. Amsterdam Creole Studies 12, 97–126. Roberts, Sarah (1998) The role of diffusion in the genesis of Hawaiian Creole. Language 74, 1–39. Roberts, Sarah (1999) The TMA system of Hawaiian Creole and diffusion. In: John Rickford and Suzanne Romaine (eds.) Creole Genesis, Attitudes and Discourse: Studies Celebrating Charlene J. Sato. Amsterdam: John Benjamins, pp. 45–70. Roberts, Sarah (2000) Nativization and the genesis of Hawaiian Creole. In: McWhorter (ed.), pp. 247–300. Romaine, Suzanne (1988) Pidgin and Creole Languages. London: Longman. Sankoff, Gillian and Susanne Laberge (1974) On the acquisition of native speakers by a language. In: David DeCamp and Ian Hancock (eds.) Pidgins and Creoles: Current Trends and Prospects. Washington, DC: Georgetown University Press, pp. 73–84. Sebba, Mark (1986) Adjectives and copulas in Sranan Tongo. Journal of Pidgin and Creole Languages 1, 109–21. Seuren, Pieter (1981) Tense and aspect in Sranan. Linguistics 19, 1043–76. Seuren, Pieter (1986) Adjectives as adjectives in Sranan: A reply to Sebba. Journal of Pidgin and Creole Languages 1, 123–34. Siegel, Jeff (2000) Substrate influence in Hawai[i Creole English. Language in Society 29, 197–236. Singler, John Victor (ed.) (1990) Pidgin and Creole Tense-Mood-Aspect Systems. Amsterdam: John Benjamins. Singler, John Victor (1992) Nativization and pidgin/creole genesis: A reply to Bickerton. Journal of Pidgin and Creole Languages 7, 319–33.

Creole Genesis 241 Singler, John Victor (1996) Theories of creole genesis, sociohistorical considerations, and the evaluation of evidence: The case of Haitian Creole and the Relexification Hypothesis. Journal of Pidgin and Creole Languages 11, 185–230. Smith, Norval (1996) Focus marking wE in Saramaccan: Grammaticalization or substrate. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Relating to Grammaticalization in Creole Languages. London: University of Westminster Press, pp. 113–28. Smith, Norval (2001) Voodoo Chile: Differential substrate effects in Saramaccan and Haitian. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 43–80. Syea, Anand (1993) Null subjects in Mauritian Creole and the pro-drop parameter. In: Frank Byrne and John Holm (eds.) Atlantic Meets Pacific. Amsterdam: John Benjamins, pp. 91–102. Trabant, Jürgen and Sean Ward (eds.) (2001) New Essays on the Origins of Language. Berlin: Mouton de Gruyter. Veenstra, Tonjes (1996) Serial Verbs in Saramaccan: Predication and Creole Genesis. The Hague: Holland Academic Graphics. Veenstra, Tonjes and Hans Den Besten (1995) Fronting. In: Arends, Muysken, and Smith (eds.), pp. 303–16. Veltman, Carl (1983) Language Shift in the United States. The Hague: Mouton. Winford, Donald (1993) Predication in Caribbean English Creoles. Amsterdam: John Benjamins. Winford, Donald (1997) Property items and predication in Sranan. Journal of Pidgin and Creole Languages 12, 237–301. Winford, Donald (2000a) Tense and aspect in Sranan and the creole prototype. In: McWhorter (ed.), pp. 383–442. Winford, Donald (2000b) Irrealis in Sranan: Mood and modality in a radical creole. Journal of Pidgin and Creole Languages 15, 63–126.

242

Sarah G. Thomason

10

Pidgins/Creoles and Historical Linguistics SARAH G. THOMASON

1

Introduction

The main point of intersection between historical linguistics and pidgin/creole studies is in the lively and controversial area of pidgin/creole genesis. Potential contributions to the two fields are mutual: insights from historical linguistics can shed light on processes and results of pidgin/creole genesis, and investigation of pidgin/creole genesis can also help deepen historical linguists’ understanding of linguistic diversification and linguistic change, especially contact-induced change. The two fields differ greatly in their relative stages of development. Historical linguistics emerged as a systematic field of study in the nineteenth century and was the dominant subfield of linguistics until the 1950s. The development of historical linguistic theory and methodology has been stimulated and furthered by the existence of rich ancient documentation – more than a thousand years of continuous attestation in several branches of the Indo-European language family and in a few (very few) other language families as well. Pidgin and creole studies, despite the important work of Hugo Schuchardt in the late nineteenth century and a few pioneers in the early and mid twentieth century, have been pursued vigorously by a sizable community of scholars only since the 1960s. In historical studies of pidgins and creoles the time depths available for direct investigation are shallow (no more than a few hundred years in the most favorable cases); early documentation of the languages themselves, though not entirely lacking, is sparse. Well-studied pidgins and creoles are not numerous – a few dozen at most – whereas historical linguists have massive amounts of data and extensive catalogues of changes for dozens or even hundreds of languages, so that theoretical predictions can be tested on a wide variety of situations. It is not surprising, therefore, that historical linguists are more confident about their results and more in agreement about the foundations of their field than specialists in historical pidgin/creole studies can be. On the other hand, pidgin/creole specialists often have access to nonlinguistic historical

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Pidgins/Creoles and Historical Linguistics 243 and social documentation, including demographic figures, that historical linguists rarely encounter in comparable quantity and quality, so in this respect creolists are privileged. (Pidgins attract much less scholarly attention than creoles, and both nonlinguistic and linguistic historical documentation are generally lacking, so pidgin specialists usually don’t enjoy this advantage.) The following sections will address the main issues that are relevant to both fields and that arise frequently in discussions of pidgin/creole genesis. First, section 2 considers a question that is problematic for both historical linguists and pidgin/creole specialists: does children’s first language acquisition play a crucial role in language change and in creole (but not pidgin) genesis? Second, a perennial question that has implications for both fields concerns the extent to which genesis processes can be equated with “ordinary” language change (section 3). Third, in what ways do pidgins and creoles contribute to the debate about the existence or nonexistence of mixed languages? This and a closely related question concerning the status of pidgins and creoles within the historical linguist’s framework of genetic relationship will be discussed in section 4. Fourth, is pidgin genesis essentially the same as creole genesis, or do the two kinds of contact language arise through entirely different processes (section 5)? And finally, I will argue that the criteria for establishing an instance of contact-induced language change are also applicable for establishing sources of pidgin and creole features (section 6). The brief conclusion (section 7) summarizes the mutual contributions between historical linguistics and pidgin/ creole studies. Before we consider these questions, though, some definitions are needed. It would be nice to be able to start with a foundation of agreed-upon positions, to provide a stable context in which the controversial issues could be situated. Unfortunately, this isn’t possible: I think it’s safe to say that the only thing all pidgin/creole specialists agree on is that we don’t agree on much of anything. There isn’t even an “establishment” position that can be contrasted with one or more “challenging” positions; senior as well as younger scholars tend to disagree on the most fundamental matters. This striking divergence of opinion makes it absolutely necessary to provide definitions before embarking on a discussion of issues, so that readers will at least be able to tell whether authors are debating about comparable things. Here, therefore, are my definitions of “pidgin” and “creole,” and also of “contact-induced language change,” which I will be comparing with pidgin and creole genesis. A pidgin is a mixed language that is spoken only as a second (or third, or fourth, or nth) language and that arises in a contact situation involving two or more (usually at least three) linguistic groups; the groups have no shared language, and they communicate regularly, but for limited purposes, such as trade. They do not learn each other’s language(s). The lexicon of the pidgin may, and usually does, come primarily from one of the languages in contact, called the lexifier language; the grammar, crucially, does not come primarily from that language or from any other single language (Thomason 2001, pp. 273–4). A pidgin is a fully crystallized language, with its own lexical and

244

Sarah G. Thomason

grammatical structures, so that no one, including speakers of the lexifier language, can speak the pidgin simply by producing ad-hoc simplified utterances using the pidgin’s vocabulary. Instead, each speaker of the pidgin must learn its lexical and grammatical structures in order to speak it. A new pidgin has less linguistic material than a non-pidgin, both in the lexicon (fewer words) and in the grammar (fewer stylistic resources, fewer elaborate morphosyntactic structures) (Thomason 2001, p. 274); older pidgins also have limited linguistic material, provided that they remain secondary languages used for quite limited purposes. Examples are Chinook Jargon, which thrived during the late nineteenth and early twentieth centuries in the Pacific Northwest of the US and neighboring parts of Canada, the Pidgin Delaware that was spoken in the eighteenth century in the eastern US, and Chinese Pidgin English. A creole is a “mixed language that is the native language of a speech community” (Thomason 2001, p. 262). This definition of course requires a further definition, of “mixed language”: this is “a language that did not arise primarily through descent with modification from a single earlier language” (Thomason 2001, p. 271). Examples are Jamaican Creole in the Caribbean, Hawaiian Creole English, Tok Pisin in Papua New Guinea, and Nigerian Pidgin English, which in spite of its name now has a great many native speakers. The reason I define creoles in historical terms is that the argument that creoles form a synchronically identifiable typological class, as in recent claims by McWhorter (1998), is in my opinion unconvincing. Two of McWhorter’s three criteria for creole status, (near-)absence of inflection and (near-)absence of phonemic tone, are found, as McWhorter acknowledges, in non-creole languages as well as in creoles, for instance some Mon-Khmer languages (cited by McWhorter) and Trumai (Brazil; Guirardello 1999, p. 352). His third criterion, “semantically regular derivational affixation” (1998, p. 798), is rather vague and hard to pin down: how regular does the regularity have to be? In Fa d’Ambu, for instance, the derivational morphology includes affixes with varied uses and affixes with lexically specific domains (Post 1995). Although McWhorter requires all three of these structural features to be present in a language judged to be a creole, therefore, the cross-linguistic frequency of the first two (which could easily have arisen independently in any language) and the vagueness of the third make the set a shaky foundation for creole identification. In addition, some languages that all specialists consider to be creoles do not meet all three criteria. Unlike new pidgins and some older pidgins, creoles do not have limited linguistic resources: this is the justification for making the distinction between the two types of contact language. But what about those older pidgins that have expanded beyond an initial trade or other limited social context to the point where they are used for a wide variety of functions in a community or even a whole region? Examples include some of the Pacific pidgins, e.g., Bislama (before it acquired some native speakers). If an expanded pidgin is still learned and used only as a second language, it will still be identifiable as a pidgin on this social criterion; but it will no longer be distinguishable from a creole on

Pidgins/Creoles and Historical Linguistics 245 linguistic grounds, because both types of mixed language will have unlimited linguistic resources (to the extent that any language has unlimited resources). Some specialists prefer to call these linguistically and socially expanded contact languages “creoles,” abandoning the native speaker criterion for creole status. I don’t, but this means that the category of pidgins, in my terminology, is linguistically quite heterogeneous. This terminological awkwardness doesn’t bother me, because heterogeneity is present no matter how we define these languages. It is inevitable that there will be languages that can’t be unambiguously classified as pidgins or creoles, no matter how carefully we set up our criteria (Thomason 1997). There are pidgins that border on creoles (how many native speakers are needed to constitute a speech community?), pidgins that border on foreigner-talk versions of a lexifier language (how different does a grammar have to be to count as sufficiently different from the lexifier language’s grammar?), and creoles that border on non-creoles (how much mixture is necessary before a language can no longer be considered the product of descent with modification from a single earlier language?). And also, of course, a pidgin that is in the process of expanding linguistically and socially moves from a linguistically limited state toward a linguistically unlimited state, with no sharp temporal or linguistic dividing line between the two. In fact, the first lesson to be learned by pidgin/ creole specialists from historical linguists has to do with fuzzy boundaries: in spite of the many clear cases on each side of a boundary, any situation in which change is occurring will inevitably present us with unclassifiable borderline cases (see Thomason 1997 for discussion). In historical linguistics, for instance, the question of whether we are dealing with two dialects of the same language or with two closely related languages arises again and again; the criterion of mutual intelligibility works fine in many clear cases, but not at stages where dialects are (or were) in the process of diverging into separate languages, when attitudinal as well as linguistic factors affect levels of intelligibility. The fuzziness of borderline cases should not, however, prevent us from drawing generalizations based on clear cases. Finally, here is my definition of contact-induced language change: a contactinduced change is any linguistic change that would have been less likely to occur outside a particular contact situation (Thomason 2001, pp. 61–3). This definition is quite broad. Most obviously, it includes actual interference features transferred into a language A from another language B. But it also includes a number of other phenomena that are less obvious: changes in a dying language that do not make its structure more like that of the dominant language; innovations that appear at a late stage of a chain reaction that was triggered in the first instance by structural transfer; and innovations introduced deliberately to make one language more different from another. The reason for the phrase “less likely” in the definition is that many contactinduced changes have internal structural motivations as well, and thus are instances of multiple causation; but they would have been less likely to occur without the external motivation.

246

2

Sarah G. Thomason

How Important is L1 Acquisition in Language Change and Pidgin/Creole Genesis?

Questions about the nature and degree of children’s contribution during first language (L1) acquisition are essentially the same in historical linguistics and in investigations of pidgin/creole genesis. These questions are much debated in both fields: the controversy about whether L1 acquisition plays a crucial role in creole genesis is mirrored in historical linguists’ debates about whether the primary locus of language change is L1 acquisition or adults’ linguistic changes resulting from social interactions. The debate divides fairly neatly along theoretical lines. Generative linguists believe that abrupt creoles are constructed by children during L1 acquisition, and that adult grammars can change only in superficial ways, while children can effect significant restructuring during L1 acquisition; non-generativists are more inclined to accord adults a major role in both creole genesis and language change. In order to address this issue for “ordinary” language change, we first need to distinguish between a linguistic innovation and the spread of an innovation. Any given linguistic change can in principle be studied from either perspective – how the change arose, or how it spread from its point of origin (in a single speaker’s brain, or in several or many speakers’ brains) through an entire speech community. The term “linguistic change” is ambiguous; it can be used to refer to either process. In practice, however, linguistic changes can’t be studied until they have spread well beyond the innovator’s, or innovators’, speech. There is no way to tell, when you hear an innovative form like an English plural foots or past tense sweared, whether it will become a permanent part of a single speaker’s internalized linguistic system, much less whether it will spread from that speaker throughout the speech community to replace the older form, feet or swore. This is as true of the toddler who overgeneralizes the plural or past tense formation rule as it is of the adult who usually says feet and swore but produces an innovation as a speech error. It is also true of adults’ innovations that are more likely to be directly relevant for pidgin/ creole genesis, for instance a one-time imitation of what the speaker believes to be a more prestigious dialect or a one-time importation of a sound or structure from another language. Another preliminary point that needs to be made here is that, for a period of time that may last a hundred years or more, an innovative form competes with the old form. Some variants may remain in a language for a very long time; one example is English ask vs. aks, variant forms that can both be traced back to Old English and that are not distributed neatly in different dialects (though aks is always nonstandard). A given speaker may use just the innovative form or just the old form; in modern sociolinguistic studies of ongoing change, investigators sometimes find that the oldest generation always uses an older form and the youngest generation always uses a new form. But sometimes a single speaker may use both an old and a new form more or less

Pidgins/Creoles and Historical Linguistics 247 interchangeably; common examples are the English past tense forms dreamt/ dreamed and dived/dove. And sometimes the old and new forms remain in broad use in the speech community but take on different meanings, as with English hanged (for executing people) and hung (for everything except executing people). The examples in this paragraph are all single lexical items, because those are the simplest illustrations, but competition involving added or lost rules of all kinds is also easy to exemplify. For instance, the Indic language Shina has acquired several rules from the isolate Burushaski, among them a rule of number agreement according to which an interrogative or indefinite pronoun takes plural verb agreement (Thomason forthcoming). Now, what about L1 acquisition in language change and pidgin/creole genesis? First, we aren’t talking about pidgin genesis, because one of the few points of general agreement among specialists is that pidgins are not created by children during L1 acquisition. Second, we aren’t talking primarily about the spread of an innovation through a speech community, because that is a social process that certainly includes adults as well as children, so that it is not confined to L1 acquisition per se (though children do acquire variation along with invariable rules and lexicon). The focus here is therefore on the process by which a child acquires his or her (I’ll use “her” for convenience) grammar, and potential connections between that process and both language change and creole genesis. I can’t hope to resolve the question of the role of L1 acquisition in language change here, and I won’t try. I’ll start by sketching a necessarily oversimplified picture of the generativist view of language change and of the opposing, or at least skeptical, non-generativist view of language change. Then I’ll turn to a consideration of generativist views of abrupt creole genesis. Since the 1960s, the basic generativist view of L1 acquisition has been that children make use of the input they receive to construct their grammar, primarily by choosing among options made available by Universal Grammar, the hard-wired grammatical structures inherited by all human infants. Children also learn thousands of lexical items and their associated meanings, at a stunning rate of many words per day in their early years. In the generativist view of language change (see e.g., Halle 1962; Lightfoot 1979), the input received by L1 acquirers comprises the output of older speakers’ grammars which, in addition to the grammar internalized in early childhood, includes “late” rules added after the period of L1 acquisition. According to this view, an adult does not change her initial grammar at all; instead, she may change the output of that grammar by adding a few additional rules. The child who hears the adult’s modified output then constructs her own grammar, which incorporates the adult’s changes in an orderly, less cobbled-together way. So, for instance, suppose that a language has only voiceless fricatives. Adults in the speech community may begin voicing their fricatives between sonorants, for whatever reason; but, according to the traditional generativist view, they still have an internalized grammar that produces only voiceless fricatives, and the voiced fricatives are produced by a later “adjustment” rule that changes the

248

Sarah G. Thomason

output of their early-acquired grammar. Child learners, hearing voiced fricatives between sonorants and voiceless fricatives elsewhere, then construct a grammar that contains the voicing rule to begin with. This cumbersome mechanism is perhaps not to be taken too literally, but the general idea is that adult grammars can change in only rather minor ways via supplements to the early-acquired grammar. Children do not, of course, learn the adults’ grammar per se; rather, they build their own original grammars on the bases of Universal Grammar and of adults’ and older children’s input. Language transmission thus consists of children’s turning inevitably fragmentary input into a new grammar that will match the internalized grammars of older members of the speech community closely enough for its output to be unremarkable. Non-generativist historical linguists see problems with this scenario as the primary mechanism of language change. There is solid evidence that adult grammars can change in major ways, most dramatically in deliberate changes effected by entire speech communities. The most famous cases of new mixed languages created by bilinguals, apparently for the purpose of serving as a symbol of new (sub-)ethnic groups, provide one set of examples: Michif, with its French noun phrases and Cree verb phrases, is an especially welldocumented case; another is Mednyj Aleut, with Russian finite verb inflection grafted onto an otherwise Aleut grammatical and lexical system. Children were certainly born into both communities once the new groups existed, but the mixing had to have been done by accomplished bilinguals, and neither case resembles any known instance of children’s bilingual L1 acquisition. Another set of cases can be found in language planning, either in the formation of a standard language or for less stately (and non-governmental) purposes. One especially striking example is Uisai, spoken on Bougainville Island (Papua New Guinea), whose speakers replaced all their elaborate masculine agreement markers with feminine markers, and vice versa, in order to differentiate their dialect more sharply from other dialects of the Buin language (Laycock 1982, p. 36). Similarly, though less spectacularly, speakers of a dialect of Lambayeque Quechua (Peru) systematically distorted their lexicon by metathesizing certain final VC sequences to CV sequences (David Weber, p.c. 1999, citing Dwight Shaver’s fieldwork). The reason I cite such elaborate examples is that these are cases where the role of adults in the change process cannot reasonably be doubted. Many examples can also be adduced from more ordinary cases of contact-induced language change, in contexts where the contacts are primarily confined to adults. Asia Minor Greek dialects, for instance, underwent major structural interference from Turkish over several centuries, resulting in changes at all levels of their structure (Dawkins 1916); but in some of the most-affected dialects, a Greek village’s menfolk, who worked in nearby Turkish towns, were apparently the main source of the borrowed lexicon and structure. If adult grammars can and do undergo significant contact-induced changes, it seems rash to assume that significant restructuring through internally

Pidgins/Creoles and Historical Linguistics 249 motivated change, such as the changes discussed by Lightfoot and other generativists, is effected solely by children during L1 acquisition. The relevance of this controversy within historical linguistics for the similar controversy in creole studies is that, if adults carry out significant restructuring in “ordinary” language change, it would be risky to assume that they do not participate in processes that produce creoles. The scholar whose name is most closely linked to L1 acquisition as the primary mechanism of creole genesis is Derek Bickerton (see e.g., Bickerton 1984, and see also Veenstra, this volume). According to his Language Bioprogram Hypothesis, abrupt creoles are created by children who are born into a new plantation community and whose input for their L1 acquisition is the unsystematic macaronic (pre-)pidgin spoken by their elders. The community was formed by their parents, who shared no common language and so needed a new means of communication; but the parents’ own medium of interethnic communication had no time to coalesce into a fully crystallized pidgin language before children arrived. The children, deprived of systematic input, fell back on Universal Grammar in constructing their own grammars, although their vocabulary was derived from their caregivers’ (pre-)pidgin. The (pre-)pidgin in turn took its lexicon from the language spoken by the plantation owners. Bickerton makes no claims about creoles that have developed from long-established pidgins, like Tok Pisin; his claims are confined to creoles that arose abruptly. The Language Bioprogram Hypothesis is a good fit with the generativist view of the main mechanism of ordinary language change. Here too there are skeptics, and the skeptics’ position is generally considered by creolists to be quite strong nowadays; the Language Bioprogram Hypothesis hit the peak of its popularity among creolists in the mid-1980s. Skeptics have emphasized two general kinds of objections to Bickerton’s hypothesis. First, the hypothesis predicts that all abrupt creoles will have largely identical grammatical structures, at least in the morphosyntactic features present in the proposed bioprogram (a.k.a. Universal Grammar); but various scholars have shown that there is more variation in the relevant features than the hypothesis allows for (see e.g., Bakker, Post, & Van der Voort 1995). Second, many Caribbean creoles, which were once believed to have arisen abruptly, turn out to have arisen gradually instead, after a contact period that lasted long enough to allow either for the development of a robust and fully crystallized pidgin or for truly gradual creole genesis via repeated instances of shift-induced interference in varieties of the lexifier language (see section 3 below for discussion of the latter scenario). Moreover, even in an abrupt-creole genesis setting, there are serious doubts about the validity of Bickerton’s claim that the only input to children’s L1 acquisition was an unsystematic pre-pidgin which, he believes, was the lingua franca among the adults in the new community. Without evidence of any kind, it is surely rash to deny that the caregivers’ own L1s and perhaps other languages that the caregivers knew, in addition to the pre-pidgin, comprised the children’s input for L1 acquisition (Thomason & Kaufman 1988, pp. 163ff.). If other languages formed part of

250

Sarah G. Thomason

children’s input for L1 acquisition, and especially if in a given plantation community a creole developed so gradually that it went through a robust pidgin stage, then Bickerton’s hypothesis can’t account for the proposed bioprogram features shared by the eventual creoles. Since many of the relevant creoles do share the features, some other explanation is needed to account for them; and in fact most creolists’ enthusiasm for the Language Bioprogram Hypothesis has diminished steadily as the number of creoles to which it might be applicable has decreased. Probably most creolists would agree that L1 acquisition might be an important factor in creole genesis under the right circumstances. The right circumstances would be those in which a creole arises abruptly, in a new contact situation, as a new contact language that is used more or less immediately for all ordinary communicative purposes and learned, or (in part at least) created, by children during L1 acquisition. Pitcairnese, the creole that developed after the Bounty mutineers settled on Pitcairn with their Tahitian-speaking companions, is a potential example. But many or most creolists are skeptical of the evidence that has been presented for its relevance in particular cases (and still others challenge its classification as a creole), so its significance for the development of known creoles remains in doubt. And the fact that adults must have been primarily responsible for the formation of many or most known creoles – that is, some creoles that were once thought to have arisen abruptly turn out instead to have emerged through nativization of a pidgin – indirectly supports the position of those historical linguists who see a larger role for adults in the initiation and spread of internally and externally motivated language changes. Ultimately, the arguments for significant adult contribution to language change and to creole genesis converge: examples of adults’ “ordinary” language changes, especially contact-induced changes, are similar in type to examples of adults’ creation and stabilization of new contact languages, specifically pidgins. In both fields, the demonstration that adults are responsible for some of the relevant phenomena weakens the case for children as the prime movers.

3

Do Pidgins and Creoles Arise Solely through Ordinary Processes of Language Change?

Few pidgin/creole specialists, if any, doubt that the emergence of a pidgin or creole language has much in common with linguistic change as historical linguists know it. The controversial question is whether there is anything in the process(es) of pidgin/creole genesis that can or must be sharply distinguished from ordinary language change. (At this point I’ll stop surrounding the word “ordinary” with shudder quotes. I can’t define ordinary language change in a really satisfactory way; but the term is meant to include all linguistic changes,

Pidgins/Creoles and Historical Linguistics 251 both internally and externally motivated, that are not generally thought to produce immediate extreme results.) Only contact-induced change is relevant here. Except possibly for Chaudenson (e.g., 1992) and his colleagues, no creolist would argue that pidgins or creoles arise solely through gradually accumulated internally motivated changes, given the fact that pidgins and creoles are contact languages par excellence: they are the products of contact situations in which speakers who do not all share a common medium of communication develop one. Given my definition of contact-induced change – as a change that would have been less likely to happen outside a particular contact situation – any linguistic changes that might contribute to pidgin/creole genesis would have to be contactinduced changes. Specifically, the kind of contact-induced change that is likely to be most directly comparable to processes of abrupt creole genesis (if any) and pidgin genesis is shift-induced interference, which occurs when a group of people shift from their shared native language to another language, usually that of a socioeconomically dominant group. Of course the creators of a pidgin or abrupt creole do not share a single native language – if they did, they would have no need for a contact language. What they have in common with the shifting group in a case of ordinary shift-induced interference is that neither the linguistically homogeneous shifters nor the pidgin/creole creators learn a target language perfectly: imperfect learning is the hallmark of both processes. This type of interference has traditionally been called substrate interference; but the term “substrate (language)” implies that such a language must be spoken by socioeconomically subordinate people, and this is historically inaccurate in some ordinary language shift situations and some pidgin genesis situations, so I prefer to avoid the term. The other kind of interference, borrowing in a narrow sense, is the incorporation by bilinguals or multilinguals of material from one of their languages into another of their languages (typically, but not always, their native language). This process is unlikely to play a major role in pidgin/creole genesis. The reason is that pidgins and abrupt creoles arise in contact situations that do not involve bilingualism or multilingualism that includes control of a single shared language; that’s why a new contact language, a pidgin or a creole, is needed for intergroup communication. Shift-induced interference is closely allied to the linguistic phenomena studied, primarily in individuals, by specialists in second language acquisition. Shifting speakers may carry over features from their original native language, and they may also fail to learn certain target-language features. But it must be emphasized that imperfect learning in this context has no necessary connection with trying and failing to learn the target language; it may well be that shifting speakers prefer to hang onto bits of their original language’s structure, and prefer not to adopt bits of the target language structure. In pidgin and abrupt creole genesis, there may in fact be no target language (see, e.g., Baker 1990). If speakers’ goal is to develop a common means of communication, they

252

Sarah G. Thomason

may need, and take, only lexicon from the lexifier language, making no effort to learn its structure. In some situations the lexifier language is unavailable, even if the shifting speakers do wish to learn it; in other situations, the lexifier language may be available (the shifting speakers may have enough access to native speakers of the lexifier language) but still not learned as a whole. Broad types of interference correlate robustly with the two basic kinds of ordinary contact-induced change: in borrowing, the receiving language (almost) always gets loan words before structure, while in shift-induced interference the receiving language – that is, the target language for the shifting speakers – gets more structure than vocabulary. In some cases of language shift the only interference features are structural, especially phonological and syntactic features; but in most cases, although structural interference predominates, at least some loan words also appear in the target language. The only exception to this generalization is found in some cases of superstrate shift, where the shifting group exerts socioeconomic dominance over the target-language speech community; the most famous example is the shift by Norman French speakers to English in the thirteenth century, a process that brought hundreds or even thousands of French loan words into English. (See Thomason & Kaufman 1988 for an extended discussion of the distinction between borrowing and shiftinduced interference and its relevance to pidgin/creole genesis.) In part the controversy over change vs. creation in pidgin/creole genesis has been fueled by unfortunate choices of terms. Bickerton, for instance, has said that “the issue is about distinguishing creolization from natural change” (1998, p. 191), and he is not alone among creolists in framing the distinction in these terms. Others reject this formulation, objecting to the implication that creolization is somehow unnatural. In part, however, Bickerton’s position is a substantive claim about the non-connectedness of creolization and ordinary change processes. This claim is actually rather surprising, since his Language Bioprogram Hypothesis is compatible with a view that the only difference between abrupt creole genesis and “ordinary” L1-acquisition-driven language change lies in the nature of the input: from a generativist viewpoint, language change via children’s construction of their grammar would arguably not differ in kind in the two situations. At the other extreme from Bickerton are those creolists who equate creole genesis with ordinary language change. For instance, DeGraff (2005) has a section entitled “ ‘Creole genesis’ vs. ‘language change’: a mythical dualism?”, in which he argues that certain syntactic divergences between Haitian Creole and its source languages “have recurrent analogues in language acquisition and in language change” (for instance in changes in English and some Scandinavian languages). This, he suggests, means that “key aspects of Creole genesis can be apprehended by an investigation of the same UG principles that are known to constrain paths of parameter-(re)setting in developmental patterns arising outside ‘creolization’.” From a somewhat different perspective, creolists like Mufwene (e.g., 2001) and perhaps Chaudenson (e.g., 1992, to the extent that he allows for externally motivated changes) equate creole

Pidgins/Creoles and Historical Linguistics 253 genesis with shift-induced interference, and consider the outcomes of creole genesis as straightforward derivatives of their lexifier languages. Indeed, it seems unlikely that what speakers do in ordinary contact situations is drastically different from what they do in a social context in which a pidgin or creole language emerges. This position has long been held by specialists who believe that pidgin and creole structures owe much to the languages spoken originally by their creators – that is, that pidgin/creole genesis is analogous to shift-induced interference. (But Chaudenson and Mufwene don’t belong to this group, because they would reject the notion that a formative role should be assigned to the influence of any languages other than the lexifier.) And, as noted above, generativists’ emphasis on L1 acquisition as the major force in language change and abrupt creole genesis is basically the same scenario in both areas. What may be different, in pidgin/creole genesis contexts, is the extent to which speakers create new structure more or less consciously. There are scattered examples from around the world of apparently deliberate linguistic changes (see Thomason 2001, pp. 149–52), but there is also a great deal of evidence that indicates lack of speaker awareness for probably the vast majority of ordinary linguistic changes, including contact-induced changes. There is no evidence to support a claim that contact-induced change in general or shiftinduced interference in particular is more often deliberate and conscious than internally motivated change is. But the lack of a common medium of communication makes the social contexts in which abrupt pidgins and abrupt creoles arise unusual, if not unique: speakers can hardly fail to notice their inability to communicate effectively with their interlocutors in a new contact situation when they share no common language, and this awareness may well promote a high level of conscious participation in the group’s effort to construct a new medium of communication. The process by which speakers developing a pidgin or creole select and reinforce easy-to-understand structural features in conversation has been called “negotiation” (Thomason & Kaufman 1988, ch. 6), “accommodation,” or – in a discussion of the famous Balkan linguistic area rather than pidgin or creole genesis per se – “mutual reinforcement of change” (Lindstedt 2000). I make this suggestion about possibly greater speaker awareness with some diffidence, however, because I can’t begin to imagine how one might go about testing it. The conditions available for experimental testing of hypotheses in laboratories can’t easily approach the real-life initial-contact circumstances that promote pidgin genesis, much less the disruptive and traumatic social circumstances attendant on the institution of slavery. The situation is further complicated by the fact that a pidgin or abrupt creole language, once it crystallizes as a separate language with its own grammar that must be learned by child or adult newcomers, will inevitably change, as every living language changes. The problem then is to distinguish internal or contact-induced changes within an existing contact language from the processes that create the contact language’s grammar in the first place. No simple resolution of this difficulty is possible, because both kinds of process might be

254

Sarah G. Thomason

going on at the same time during later stages of crystallization. But at least it can be said that, once a contact language clearly exists as a separate language, any changes it undergoes are just that – changes in an existing language, not part of a pidgin/creole genesis process. In fact, if Chaudenson and Mufwene are right, the processes involved in at least some instances of creole genesis are entirely subsumed under shift-induced interference, as each contingent of new slaves learns the lexifier language from earlier contingents, in an unbroken chain going back to those who learned it imperfectly from native speakers of the lexifier language. And even among creoles that do not fit the Chaudenson/Mufwene model(s) – for instance those for which there was no true target language – some have apparently developed what are considered typical creole features gradually, well after they presumably crystallized as pidgin languages (see, e.g., Arends & Bruyn 1995 and references cited there). In these cases, the late-appearing creole features must have developed through ordinary change, probably both internally motivated and contact-induced change. Whether expansion from a linguistically limited fully crystallized pidgin to a linguistically unlimited creole (or expanded pidgin) presents features that are peculiar to the expansion process is an open question, one that deserves close attention in pidgin/creole studies. For the purposes of this section it is reasonable to sum up the processes involved in pidgin and creole genesis as “ordinary processes, extraordinary results” (Thomason 1995). But, as we will see in the next section, even if most creolists would not dispute the claim about ordinary processes, some would dispute the claim that the results are extraordinary. Historical linguistics has contributed to these debates by providing a framework that distinguishes between borrowing in a narrow sense and shift-induced interference and by comparing pidgin/creole genesis to shift-induced interference.

4

Are Pidgins and Creoles Mixed Languages? Are They Genetically Related to Their Source Languages?

These two questions are intimately connected. The claim that pidgins and creoles are not genetically related to any of their source languages depends crucially on acknowledging their status as mixed languages; if they are not mixed languages, then it is probably impossible to argue that they are not genetically related to their lexifier languages. There is considerable confusion in the literature on both of these points. Some of the confusion arises from the fact that different scholars operate with different definitions of “mixed language,” and some of it, I believe, arises from the fact that most of the debaters are creolists, not historical linguists, and so have a limited understanding of the ramifications of the fundamental concept of “genetic relatedness” in the

Pidgins/Creoles and Historical Linguistics 255 field of historical linguistics. Still more confusion results (again in my quite possibly biased opinion) from a pervasive tendency to overlook the fact that both language change and language genesis are processes that take time and therefore have no clear-cut point-in-time division between “being” and “not being.” My own position on the mixed language question is evident from the definitions of “pidgins” and “creoles” as mixed languages in section 1 above. In this section I’ll argue in support of that position. The debate about language mixture dates from the late nineteenth century, when the mainstream historical linguists of the day claimed that mixed languages do not exist (e.g., Müller 1871–2, p. 86), while others, most notably Hugo Schuchardt, the first great creolist, declared that there is “no fully unmixed language” (1884, p. 5). Müller and Schuchardt may not have had the same definition of “mixed language” in mind, of course, but there is no doubt that Schuchardt’s arguments, based on evidence from creoles, constituted a serious challenge to the Neogrammarians’ bias in favor of internally motivated language change. Similarly, the question of whether creoles have genetic affiliations has recurred throughout the history of pidgin/creole studies. Weinreich, for instance, argued that lexical evidence alone – “the existence of cognates in the basic morpheme stock, with parallelism in allomorphic alternations as a powerful supplement” – served to establish genetic relationships, and that creoles must therefore be related to their lexifier languages (1958, p. 376). The definition of “mixed language” that I’ll be using in the following discussion stems from the discussion of the issue in Thomason and Kaufman 1988. Here it is: A mixed language is a language whose lexical and grammatical structures cannot all be traced back primarily to a single source (“parent”) language (see Thomason & Kaufman 1988 for arguments in support of this characterization and of the implications for genetic relationship discussed below). This definition of “mixed language” is embedded in the standard historical linguistic notion of “genetic relationship,” according to which the members of a language family (“sister languages”) are the products of descent with modification from their single parent language (see any historical linguistics textbook for elucidation of this notion, e.g., Trask 1996, pp. 176ff., or Hock 1991, ch. 18). In a process of descent with modification, an entire language is passed down from one generation to the next, structure as well as lexicon. So if a language’s lexicon – in particular its set of basic vocabulary items – and its structural subsystems don’t all correspond systematically to a putative parent language and any putative sister languages, then that language cannot be placed into a family tree: it is not genetically related to any contemporaneous or antecedent languages. The crucial evidence is a mismatch between components of a langage, either the lexicon as a whole and the structure as a whole or between major structural subsystems and the lexicon and/or the rest of the structure. Now, one can certainly argue, in a particular case, about whether there is enough systematic correspondence in all subsystems to support a claim of

256

Sarah G. Thomason

genetic relationship. But to argue that a language should be considered to be genetically related to one of its source languages – say the lexifier language – when entire structural subsystems do not correspond systematically to structures in that source language, one must abandon the standard historical linguist’s notion of genetic relationship and formulate a totally new concept (in which case it would be preferable to use a different term, so as to avoid confusion with the standard term). What’s new in historical linguistics since Weinreich recommended using lexical evidence alone as the criterion for establishing genetic relationship? First, a much larger body of evidence is now readily available about the whole range of contact phenomena in language history; and second, previous strands of thought have been pulled together, in particular Boas’s insight about the evidential weight of mismatches: “. . . the possibility of mutual influences . . . will be revealed, in part at least, by lack of correspondence between lexicographic, phonetic, and detailed morphological classifications” (1917, p. 4). These strands fit together to form a coherent picture of descent with modification and consequently of genetic relationship. Since 1988, many or most historical linguists have accepted the above definition of mixed languages, with its implications for genetic relationship (see, e.g., Jasanoff 1989; Solnit 1992, p. 195; McMahon 1994, p. 212; Fox 1995, p. 126). As in so many other areas of historical linguistics, the borderline between a mixed language, on the one hand, and on the other hand a language that is not mixed under this definition but that has undergone quite a bit of ordinary contact-induced change, is inevitably fuzzy; in borderline cases, it’s impossible to say exactly how much difference is too much. Probably the least controversial non-borderline cases of mixed languages are two-language mixtures created by bilinguals, like Michif and Mednyj Aleut (mentioned in section 2 above): no one, surely, would want to argue that the French noun phrases of Michif are descended from the same parent language as the Cree verb phrases of Michif. But the great majority of pidgins and creoles also qualify as mixed languages. If the lexicon comes almost entirely from a single lexifier language but most of the grammar doesn’t come from the same language, then the pidgin or creole is a mixed language by the definition above; it is not the product of ordinary descent with modification from a single parent language. There are, to be sure, borderline cases among pidgins and creoles. Reunionese has a considerable amount of French grammar as well as French lexicon, and Barbadian Creole English has quite a bit of English grammar as well as English lexicon (see, e.g., Bickerton 1998, p. 191, on the debate about the status of Barbadian Creole English). It’s not surprising, therefore, that reasonable people disagree about the status of these languages: creoles or not creoles? From a historical linguist’s viewpoint, the question is in principle unanswerable in a borderline case – the line between “mixed” and “unmixed but the product of much contact-induced change” can’t be drawn sharply. But the existence of borderline cases isn’t a good reason for tossing out the distinction between mixed languages and languages that are (in the standard

Pidgins/Creoles and Historical Linguistics 257 sense) descended from a single source language. The vast majority of the world’s languages have undergone ordinary genetic development; what makes pidgins and creoles, as a class of languages, so interesting to historical (and other) linguists is precisely their mixed nature, or rather the relatively unusual social (and perhaps linguistic) processes that produced the mixtures. The historical investigation of pidgins and creoles has contributed greatly to historical linguists’ thinking on these points in the last few decades, helping to clarify the limits and criteria for establishing genetic relationship. Finally, it must be emphasized that the claim that a pidgin or creole is not related to any other language (unless a creole itself has daughter languages and is therefore the parent of a language family) is not a denial that pidgins and creoles have source languages. They are not invented out of nothing, and nobody claims that they are wholly unconnected with any previously spoken language. As Jasanoff puts it (1989, p. 624), the assertion that pidgins and creoles have “no genetic ties, properly speaking, at all” is not to say that Jamaican Creole and Tok Pisin, e.g., are historically independent of English, but simply that the nature of the connection makes it inappropriate to employ the family tree metaphor. To demonstrate that languages are related in the strict sense, we must be able to show that they share features of a kind that could not possibly have survived the effects of discontinuous transmission. In practical terms, this means that systematic correspondences in grammar alone, or in vocabulary alone, are not enough; we must find both.

Pidgin and creole lexicons demonstrably come from lexifier languages, and their structures, except perhaps in an extreme version of the Language Bioprogram Hypothesis, owe much to the languages spoken by their creators or (if children really are involved in the creation of some creoles through L1 acquisition) their creators’ caregivers. Abruptly created pidgins are new languages; so are abrupt creoles. Together with bilingual mixed languages, they are the only languages in the world for which one can establish something close to a specific starting point (though the “point” must be a period of some years, because it takes time for an emerging language to crystallize).

5 Are Pidgin Genesis and Creole Genesis the Same Thing? Most pidgin/creole specialists focus on creoles rather than on pidgins, and most of the research on pidgins has focused on expanded pidgins rather than on new pidgins with limited linguistic resources. As a result, we know less about pidgins, overall, than about creoles. The question that is crucial for the present chapter is this: Are pidgin genesis and creole genesis essentially the same process, or are these two types of contact languages completely distinct historically? The answer to this question has important implications for some

258

Sarah G. Thomason

of the controversies discussed above. In particular, if the two processes are the same, then L1 acquisition can’t be the main factor in creole genesis, because it isn’t a factor at all in pidgin genesis. Unfortunately, there is no simple answer to the question, primarily because there’s more than one route to creole genesis. If I’m right in attributing a major role to adults in abrupt creole genesis and in pidgin genesis, then these two processes, at least, share many features: both are strongly influenced by the languages spoken by the creators of the new contact language, and the genesis processes are thus akin to shift-induced interference. The difference between a new pidgin and an abrupt creole would then be the presence or absence of significant linguistic limitations – present in the pidgin, not present in the creole. This difference might or might not be responsible for differences in the extent of contributions from the input languages; it seems reasonable to expect more influence in a pidgin, because its creators continue to speak their own languages regularly, but there is too little information available to generalize confidently on this point. In a scenario like Chaudenson’s and Mufwene’s, creole genesis isn’t as much like pidgin genesis, because the creole emerges as the end point in a series of imperfect learning events, whereas most pidgins appear to have emerged abruptly. Still, imperfect second language acquisition plays a major role in both processes, so again they share important features. These brief remarks are intended not so much to answer the question in the section’s title as to indicate the need for more systematic historical comparison of pidgins and creoles. We may never be able to sort out the different ways in which creoles and even pidgins arise, or to compare their genesis processes systematically, but at present we have too little information to know just what the limitations of such investigations will be.

6

Criteria for Establishing Contact-Induced Language Changes and Sources of Pidgin/ Creole Features

This section too will be very brief. I include it in order to emphasize the point that, as with the issue of the contribution of L1 acquisition, the criteria for establishing the sources of particular pidgin/creole features are essentially the same as the criteria for establishing the sources of particular linguistic changes, especially contact-induced changes. In practice, this means (in my opinion) that “the methodological constraints on studies of ordinary language change must also be applied in studying pidgin and creole genesis” (Thomason 1993, p. 293). Explicit criteria are hardly needed when, say, a borrowing-language morpheme agrees in both form and function with a source-language morpheme (e.g., English shah, from Persian), or a creole morpheme agrees in both form

Pidgins/Creoles and Historical Linguistics 259 and function with a lexifier-language morpheme – although the sources of certain grammatical morphemes have been hotly debated by creolists, and some of these morphemes may have multiple sources. But in shift-induced interference the transfer of morphemes is less common than the transfer of patterns, and this is also true of most pidgin/creole grammatical features that have been attributed to substrate languages. It is these examples that require strict criteria. In contact-induced language change, the source of a proposed contact feature can only be firmly established under the following conditions (see Thomason 2001, pp. 91–4 for detailed discussion of each criterion). First, we must look at the languages in question as wholes, not just at a single feature. If one structural feature has been transferred from a source language, it is overwhelmingly probable that other structural features will also have been transferred from that same source language; a claim of interference that rests solely on a single feature will remain unconvincing. Second, a source language must be identified – a language that has either the precise structural feature or a feature that can reasonably be argued to be derived from a feature of that language. Third, we must find shared structural features in the proposed source language and the proposed receiving language; the shared features need not be, and often won’t be, identical in every respect. Fourth, we must prove that the shared features did not exist in the proposed receiving language before it came into intense contact with the proposed source language. And fifth, we must prove that the shared features did exist in the proposed source language before it came into intense contact with the proposed receiving language, either by comparing the two languages directly or (if the proposed source language no longer exists because all its speakers shifted to the proposed receiving language) by comparing the proposed receiving language with close relatives of the proposed source language. All these criteria are designed to apply to two-language contact situations; complications arise when we’re looking at a Sprachbund, or at pidgins and creoles. Only the first of these requirements applies without modification to the task of establishing the sources of pidgin and creole features, but criteria two, three, and five are crucially relevant in modified form (see Thomason 1993, pp. 287–93 for discussion and exemplification of the criteria). All three of these criteria must be modified to include the probability of multiple source languages for many or most structural features, given the multi-ethnic composition of the relevant contact situations. Only the fourth criterion is irrelevant for pidgin/creole genesis: because pidgins and creoles are new languages, there is no possibility that the features shared by them and the proposed source language(s) existed in the pidgin or creole before the contact situation arose. The modification of three of the criteria is important, because it means that one cannot legitimately conclude, from the fact that a particular creole shares a given structural feature with a particular source language, that that language is the sole source of the creole feature. If (as is almost always the case) other

260

Sarah G. Thomason

languages were present during the period in which the creole arose, then their structures must also be taken into account. This is, I believe, another potential area of contribution by historical linguistics to investigations of pidgin/creole genesis. It is vital to know where pidgin and creole structural features come from when we want to understand the history of any pidgin or creole language; adopting these criteria should put origin hypotheses on firmer ground than they have sometimes been in the past.

7

Conclusion: A Summing-Up

Each of the preceding five sections has addressed one set of questions related to the overall issue of interconnections between historical linguistics and pidgin/creole genesis: the contribution of L1 acquisition both fields, ordinary contact-induced language change vs. pidgin/creole genesis, pidgins and creoles as mixed languages, the relationship between pidgin genesis and creole genesis, and criteria for establishing the source(s) of a proposed contactinduced change or a pidgin/creole feature. These questions are all interrelated, of course, so the separation into distinct topics is to some extent arbitrary. I’ve proposed several ways in which historical linguistics has contributed, or can contribute, to the investigation of pidgin and creole origins, and vice versa. The topically overlapping literature on children’s vs. adults’ role in language change and in pidgin/creole genesis is so closely parallel that what is valid for the one is very likely to be valid for the other; and there is solid evidence both in pidgin genesis and in contact-induced linguistic changes that adults play a major role, and in some cases the major or only role, in the genesis and change processes. Proposals that connect contact-induced language change closely with pidgin/ creole genesis are strongest, I believe, when the types of change are embedded in an analytic framework that distinguishes shift-induced interference (which is at least closely allied to what happens in pidgin/creole genesis) from borrowing in a narrow sense (which isn’t). The development of a model of contact-induced language change has in turn been partly inspired, and even in part based on, considerations that arise in studies of pidgin/creole genesis, which have helped historical linguists understand the nature and limits of the genetic relationship of languages as well as the nature of contact-induced change. Schuchardt’s challenge to the Neogrammarians’ mainstream historical linguistics did not overthrow their approach, but it ultimately pointed the way to a more realistic model of internally and externally motivated language change. And finally, historical linguists’ criteria for establishing the source(s) of a linguistic innovation, specifically a proposed contact-induced change, can contribute significantly to a more systematic study of the origins of pidgin and creole structural features.

Pidgins/Creoles and Historical Linguistics 261

REFERENCES Arends, Jacques and Adrienne Bruyn (1995) Gradualist and developmental hypotheses. In: Arends et al. (eds.), pp. 111–20. Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) (1995) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins. Baker, Philip (1990) Off target? Journal of Pidgin and Creole Languages 5, 107–19. Bakker, Peter, Marike Post, and Hein Van der Voort (1995) TMA particles and auxiliaries. In: Arends et al. (eds.), pp. 247–58. Bickerton, Derek (1984) The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7, 173–203. Bickerton, Derek (1998) Review of The Early Stages of Creolization, by Jacques Arends (ed.). Journal of Pidgin and Creole Languages 13, 188–93. Boas, Franz (1917) Introductory. International Journal of American Linguistics 1, 1–8. Chaudenson, Robert (1992) Des îles, des hommes, des langues. Paris: L’Harmattan. Dawkins, R. M. (1916) Modern Greek in Asia Minor. Cambridge: Cambridge University Press. DeGraff, Michel (2005) Morphology and word order in “creolization” and beyond. In: Guglielmo Cinque and Richard Kayne (eds.) The Oxford Handbook of Comparative Syntax. New York: Oxford University Press, pp. 293–372. Fox, Anthony (1995) Linguistic Reconstruction: An Introduction to Theory and Method. Oxford: Oxford University Press. Guirardello, Raquel (1999) Trumai. In: R. M. W. Dixon and Alexandra Y. Aikhenvald (eds.) The Amazonian Languages. Cambridge: Cambridge University Press, 351–3. Halle, Morris (1962) Phonology in generative grammar. Word 18, 54–72. Hock, Hans Henrich (1991) Principles of Historical Linguistics, 2nd edn. Berlin: Mouton de Gruyter. Jasanoff, Jay H. (1989) Review of Thomason and Kaufman, 1988. Language 65, 623–8. Laycock, Donald C. (1982) Melanesian linguistic diversity: A Melanesian choice? In: R. J. May and Hank Nelson (eds.) Melanesia: Beyond Diversity. Canberra: Australian National University Press, pp. 33–8. Lightfoot, David W. (1979) Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lindstedt, Jouko (2000) Linguistic Balkanization: Contact-induced change by mutual reinforcement. In: Dicky Gilbers, John Nerbonne, and Jos Schaeken (eds.) Languages in Contact. Amsterdam: Rodopi, 231–46. McMahon, April M. S. (1994) Understanding Language Change. Cambridge: Cambridge University Press. McWhorter, John H. (1998) Identifying the creole prototype: Vindicating a typological class. Language 74, 788–818. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Müller, Max (1871–2) Lectures on the Science of Language, vol. 1. New York: Scribner. Post, Marike (1995) Fa d’Ambu. In: Arends et al. (eds.), pp. 191–204. Schuchardt, Hugo (1884) Slawo-deutsches und Slawo-italienisches [Slavo-German and SlavoItalian]. Graz: Leuschner & Lubensky.

262

Sarah G. Thomason

Solnit, David (1992) Review of Japanese/Austro-Tai, by Paul K. Benedict (1990). Language 68, 188–96. Thomason, Sarah G. (1993) On identifying the sources of creole structures: A discussion of Singler’s and Lefebvre’s papers. In: Salikoko S. Mufwene (ed.) Africanisms in Afro-American Language Varieties. Athens, GA: University of Georgia Press, pp. 280–95. Thomason, Sarah G. (1995) Language mixture: Ordinary processes, extraordinary results. In: Carmen Silva-Corvalán (ed.) Spanish in Four Continents: Studies in Language Contact and Bilingualism. Washington, DC: Georgetown University Press, pp. 15–33. Thomason, Sarah G. (1997) A typology of contact languages. In: Arthur K. Spears and Donald Winford, Pidgins and Creoles: Structure and Status. Amsterdam: John Benjamins, pp. 71–88. Thomason, Sarah G. (2001) Language Contact: An Introduction. Edinburgh/Washington, DC: Edinburgh University Press and Georgetown University Press. Thomason, Sarah G. (forthcoming) Can rules be borrowed? In: Thomas Smith-Stark and Roberto Zavala (eds.) Festschrift for Terrence Kaufman. Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Trask, R. L. (1996) Historical Linguistics. London: Arnold. Weinreich, Uriel (1958) On the compatibility of genetic relationship and convergent development. Word 14, 374–9.

Pidgins/Creoles and Contact Languages 263

11

Pidgins/Creoles and Contact Languages: An Overview RAJEND MESTHRIE

1 Introduction: Language Contact and Contact Languages The study of language contact falls within the wider field of bilingualism (or multilingualism). The wider field stresses the fact that societies are rarely monolingual; languages exist amidst other languages. The field is concerned in the first place with how multilingualism is “managed” by speakers at a microlevel (in actual speech) and at a macrolevel by societies (educational structures, administrative bodies, etc.). The subfield of language contact is particularly concerned with the ways in which speakers of different languages influence each other and how their languages are accordingly altered. Credit for the establishment of the field of language contact usually goes to Uriel Weinreich’s monograph Languages in Contact (1953). However, studies of pidgin and creole languages predate this by a long way, especially Van Name (1869–70), Schuchardt (1882), and Hesseling (1897). Since language alteration by contact occurs over time, language contact is also an essential part of historical linguistics. The latter field is concerned with accounting for language change, especially in phonology and slightly less so in morphology and syntax; part of this enterprise is to tease out “spontaneous” changes from contact-induced changes. In reconstructing earlier stages of a language historical linguists work backward to factor out the effects of bilingualism to establish a pre-contact stage of the language. The concept of a “contact language” deals precisely with the excluded part of historical linguistics. Contact languages are those which can be shown to have come into being under conditions of contact, before which they had no existence. A basic distinction in the cross-cutting fields of language contact, creolistics, and historical linguistics is made between borrowing and interference through shift (Thomason & Kaufman 1988; Thomason, this volume). At its simplest level, borrowing occurs when speakers of a language adopt a new word from another

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

264

Rajend Mesthrie

language, e.g., the English word computer is borrowed in Xhosa, a South African language as ikompyuta. Such borrowing usually occurs under conditions of language maintenance: Xhosa continues to be spoken and is only slightly changed by the effects of occasional borrowing. While borrowing may lead to extensive structural changes in the borrowing language, this usually takes centuries under conditions of stable bilingualism. Substratum interference, on the other hand, which is associated with language shift, may take as little as a generation (Thomason & Kaufman 1988, p. 41). This applies when a community adopts a new language (a Target Language or TL) to the detriment of its own, without, however, replicating (or intending to replicate) all of the TL rules. Clearly not all language contact leads to the formation of a contact language, as the phenomenon of lexical borrowing makes clear. The prototypical contact language is the pidgin – one that is spoken, at least initially, only in situations of contact between speakers of two or (more usually) more languages, and not as an in-group (ethnic) language. Since pidgins and creoles are the main topic of the rest of this handbook, they will not be discussed at length in this chapter. It is, however, necessary to be explicit about terminology. I see the stable pidgin and the creole as key concepts in creolistics, with other varieties that have been distinguished in the literature ( jargon, expanded pidgin) as being secondary: Jargons (also known as “pre-pidgins”) are unstable, simple varieties with mixed vocabulary and minimal grammar. If contact is sustained, a jargon may turn into a pidgin, which in turn may expand into a stable pidgin: Pidgins are examples of partially targeted or non-targeted second-language learning, developing from simpler to more complex systems as communicative requirements become more demanding. Pidgin languages by definition have no native speakers, they are social rather than individual solutions, and hence are characterized by norms of acceptability. (Mühlhäusler 1986, p. 5)

A stable pidgin may eventually develop structurally into an expanded pidgin without being nativized. An expanded pidgin may in some circumstances occur side-by-side with a creolized version of it. From a synchronic point of view creoles are “normal” languages, with all the formal properties of a natural language. Their status as a contact language ceases for “ordinary” purposes – it is only for the technical historical linguist (and possibly for typologists) that their status as a contact language is at issue. Perhaps one should include the sociologist of language here, since the issue of whether a creole is a separate language or a variant of the lexifier language is of significance in terms of community attitudes, educational policies, and political developments. Stable pidgins are also referred to as “crystallized” pidgins, a term which captures the emergence of stable structure and sharply circumscribed contexts of use. If societal change affords greater incorporation of a formerly subordinate social group into the economy, allowing greater interethnic contacts, a pidgin is likely to expand, giving rise to an expanded pidgin. An example of such an expanded pidgin is Tok Pisin, as used as a second language (L2) by

Pidgins/Creoles and Contact Languages 265 many of its speakers. An example of a crystallized pidgin that has not expanded in this way is Fanakalo, as spoken – inter alia – in the South African mines. Fanakalo has a clear-cut and stable structure, and is typically used in the work domain. Since the mines operate on a migrant labor basis, with no families on the compounds and with a high degree of ethnic separation, the contexts of use are indeed sharply circumscribed. When South African Black workers need a lingua franca, they turn to other township varieties like Tsotsitaal which express male solidarity without the overtones of labor exploitation. Fanakalo shows that a pidgin can exist as a stable entity for a century and a half without being nativized. Three approaches will be delineated that make it possible to characterize and evaluate the degree of restructuring in pidgins and creoles (P/Cs) as opposed to other contact phenomena. The first approach, which draws on a further distinction made by Thomason and Kaufman (1988) between shift with normal versus abnormal transmission, is central in characterizing degrees of “contactness.” In shift with normal transmission, shifting speakers acquire the bulk of TL grammatical structures along with the TL vocabulary. Some of the grammatical features carried over from the original native language may cause significant changes in the TL grammar, but the TL as a whole reflects its genetic background in most respects. In shift without normal transmission, shifting speakers acquire so few of the TL grammatical structures that the TL cannot be said to have been passed on to the new speaker group as a set of interrelated lexical, phonological, morphosyntactic, and semantic structures. Mufwene’s (2001) reading of the typical creole-engendering plantation scene, continuing a line that goes back to, inter alia, Alleyne (1980) and Chaudenson (e.g., 1992), is different. For him restructuring of the TL into a creole was a gradual process which began with a colonial koiné (see section 2.7) spoken by Europeans in specific territories. In these early days of a colony, the percentage of slaves was comparatively low and the average size of individual farms relatively small. Thus, Mufwene sees the initial generations of slaves as adopting the koiné more or less intact. This adaptation occurred before large plantations were developed that required larger numbers of slaves. Subsequent generations of incoming slaves gradually replicated this variety less and less perfectly, resulting in a basilectal creole at one extreme. Thomason and Kaufman’s account draws a sharp distinction between creoles and indigenized colonial varieties like (East) Indian English; Mufwene’s does not. In his account the broad principles of transmission are the same. The third approach is that of gradualism, of which I take Baker as a representative, though important protagonists include Singler (1990), Arends (1993), and Carden and Stewart (1988). For Baker (1995), the formation of P/Cs is not a result of a failure to acquire a TL adequately; rather it is the result of slaves, and – to a lesser extent – their masters setting out (successfully) to form a medium of interethnic communication from whatever means possible, not necessarily by targeting the superstrate. Thus, for Baker (2000), P/C formation is radically different from second language acquisition. Not all gradualists

266

Rajend Mesthrie

share this conviction (see, e.g., Winford 2000). Baker stresses that pidgins turn into creoles gradually, taking at least two generations and very different pathways from that implied in Bickerton’s “instantaneous” (i.e., single-generational) model (see Veenstra, this volume). This chapter will use the insights arising from these three positions somewhat eclectically to characterize varieties which arise out of contact, but which usually fall outside the field of creolistics: interlanguages, second languages, foreign languages, koinés, semi-creoles, creoloids, bilingual mixed languages, and language-shift varieties.

2

Broad and Narrow Definitions of Contact Language

For Thomason and Kaufman (1988) a contact language is one that does not result from continuous intergenerational transmission and does not have the same source language for crucial subsystems (like the lexicon versus grammar or subsystems within the grammar itself). This constitutes a narrow definition of contact language, for which only three varieties qualify: “crystallized” pidgins, “abrupt” creoles, and bilingual mixed languages. Crystallized pidgins have a clear-cut structure (and hence count as new languages, even if they might not be nativized). Abrupt creoles, according to Thomason and Kaufman, are those which come into being without a prior stable pidgin stage. They exhibit contact-induced shift with imperfect learning. (It is perhaps necessary to stress that “imperfect” here implies incomplete learning because of restricted access, rather than an inability to learn.) Bilingual mixed languages are new codes resulting not from pidginization/creolization but from intimate borrowing in a situation of bilingualism (see Muysken, this volume). A broad definition of “contact language” admits any new identifiable code (however ephemeral) arising out of multilingualism. This broad definition follows from the view that if “imperfect learning” is a part of the definition of a P/C (and it is in the first two of the three views cited above), then there are degrees of “imperfection.” The scale “creole–non-creole” may well therefore be a graded one. The rest of this section provides an overview of contact varieties, apart from pidgins and creoles.

2.1

Interlanguages

The term “interlanguage” dates to the 1960s when applied linguists at Edinburgh began studying the developing competence of speakers of a second language. The term, coined by Selinker (1972), was meant to capture the insight that a second-language acquirer frequently used a system that was different from his or her L1 (native language) and from the L2 (or target language). Such learner language is variable, along fairly systematic lines, and there are specific pathways along which interlanguages develop. As subsequent research

Pidgins/Creoles and Contact Languages 267 emphasized, for the most part the trajectories of L2 acquisition are not very different from L1 acquisition by children, apart from interference/transfer effects and the possibilities of fossilization. For example, negatives in English interlanguages and in L1 acquisition begin with (1) a stage of external negation (No play); then develop to (2) simple negation within the VP (I no play; I no can play); to (3) more complex forms that negate modals (I can’t play); to (4) the eventual rise of negation with don’t and didn’t. One way in which interlanguages differ (from the TL and from each other) is in the possibility of influence of the L1, as when German learners of English use occasional post-verbal negatives (I play not) or Spanish speakers spend a longer time at stage (2) (I no play) before elaborating the interlanguage in respect of negation. In the long term the interlanguage might fossilize (when speakers reach a stage where they do not need to develop their communicative competence further, despite continuing access to the TL), hence stabilizing some non-TL structures. Interlanguages are significant in creolistics since it is difficult to differentiate a pidgin in the making from an early-fossilized interlanguage in terms of their structure (see Schumann 1978). Unstable jargons may precede and feed into pidgins as well as into “early interlanguages” (see further Siegel, this volume, for a critique of the position that pidgin = early-fossilized interlanguage).

2.2

ESL (English as Second Language, or more generally XSL, for any language X)

For the most part, interlanguage studies focused on individual learners in an “Anglo” environment (the UK or US), where the TL of the metropolis was readily available. On the whole an interlanguage was not considered a new language (it reflected an individual’s growing competence rather than a group phenomenon) and hence would hardly count as a contact language. The position is quite different outside the metropolis: if we take the spread of English as a paradigm example, we see the interlanguage concept applying to aggregates of people who would use their interlanguages with each other in certain domains. In this process new structural, lexical, and pragmatic norms stabilize. The field of “New Englishes” or “World Englishes” studies the process of indigenization and nativization of English as it spreads globally. These two processes are not clearly differentiated in the literature, and I would like to propose such a differentiation. The term “indigenization” is best used to refer to the acculturation of the TL to localized phenomena, be they cultural, topographical, or even linguistic (insofar as English becomes part of the “linguistic ecology” of a particular area by converging structurally with local languages). Nativization, on the other hand, is a psycholinguistic process, referring to the ways in which the TL, though not the chronological first language of speakers, may become like a native language in certain domains (i.e., it is processed and used like an L1). This use of “nativization” differs from its usual sense in creolistics, where it denotes the stabilization of an L2 (pidgin) into an L1 (creole) by children. There has not been much dialogue between New English

268

Rajend Mesthrie

scholars and creolists over the concept of nativization. However, there is a significant literature on the problems inherent in the concept of “native speaker” (e.g., Paikeday 1985; Singh 1998). Scholars like Singh (1998) argue that the prevailing concept of native speaker in linguistics shows a Western bias, where monolingualism is taken to be the norm. In many multilingual societies, on the other hand, a child may be said to have several native languages, with the order of acquisition not being an indicator of ability. What is such a multilingual person a native speaker of, but a polyphony of codes/languages working cumulatively (and sometimes complementarily), rather than a single firstlearned code? For Singh and others it is problematic to privilege one domain over others in determining a “native language.” ESLs may at first blush not seem to have a great deal in common with creoles, especially if we consider that they are mostly introduced via an educational system and are mutually intelligible with the superstrate. That is, ESLs are targeted toward the superstrate in a way that first generation creoles are not (Baker 1990, 1995). On the other hand, as most ESL countries are multiethnic and multilingual, L2 varieties may arise amongst people who have insufficient access to the superstrate but who need the ESL for communication in interethnic contacts fostered by colonialism. Such varieties have been claimed to have considerable overlaps with P/Cs, for example Irish English (Hickey 1995), American Indian English (Leap 1993), and Singapore English (Platt 1975). Two radically different views can therefore be found regarding the relationship between second language acquisition and creolization. These positions can best be characterized by the diametrically opposed views of Bickerton and Mufwene. Although Bickerton (1977a, p. 49) had once characterized pidginization as second-language learning with restricted input and creolization as firstlanguage learning with restricted input, in a subsequent study he was to paint them as radically divergent processes: No real connection exists between SLA [Second Language Acquisition] and creolization: they differ in almost every particular. SLA is done alone, creolization is done in groups; SLA has a target, creolization hasn’t; SLA is done mainly by adults, creolization mainly by kids . . . SLA gives you a second language, creolization gives you a first; SLA is done by people with a “normal” background, creolization with an “abnormal” background. (Bickerton 1983, p. 238).

Although Bickerton’s assertion that SLA is done alone is true, it is spurious insofar as all language acquisition is to some extent an individual psycholinguistic process. In the typical colonial ESL situation, second-language speakers have had to interact with each other in the second language, as most ESL countries were multilingual and colonization brought people together from different ethnic groups, often for purposes of labor. This is hardly what Bickerton had in mind when he referred to a “normal language background”; it appears that he was thinking of situations of language maintenance, with a first language providing a “safe home” for speakers who were in the process

Pidgins/Creoles and Contact Languages 269 of acquiring the target language. On the other hand, Bickerton’s discussion of targeting is insightful: ESL varieties were always targeted toward English in a way that creole varieties appear not to have been. However, Mufwene’s (2001) position plays down the notion of targetness. He proposes that the same formula is operative in second language acquisition, first language replication of colonial dialects, as well as what is traditionally called creolization. Such a formula (which has yet to be specified) would have to include the following: • nature of superstrate dialects brought from European country; • co-existence of European superstrate with other languages in the colonies; • demographic proportions of speakers of languages in contact during critical periods of contact; • kinds of social contact between different social and ethnic groups; • structural features of the varieties in contact; • rate of immigrations subsequent to initial crucial contacts; • social background of new immigrants; • patterns of integration of new immigrants. (Mufwene 2001) Different instantiations of the variables would account for different “new” varieties originating under contact. My own view is that ESLs are indeed different from creoles in obvious ways. These include (1) structural considerations like the radically different TMA systems of P/Cs compared to the less-restructured equivalents in ESLs; (2) intelligibility considerations, with P/Cs being far less mutually intelligible with the TL than ESLs; and (3) attitudinal considerations, with an ESL not being considered “our language” by its speakers in the way that creoles might be. Also, many situations of language shift result in outcomes that are clearly identifiable as varieties of the superstrate (a “social dialect” thereof). But there are indeed some intermediate cases which do raise interesting questions about the nature and outcome of the language contact (see section 4).

2.3

EFL (English as Foreign Language, or more generally XFL, for any language X)

EFL denotes the use of English by speakers within a particular territory solely for communication with outsiders. For all other purposes (including medium of instruction at schools and literary production), a local language is used. EFLs have been characterized as “performance varieties” (Kachru 1983, pp. 38–9), suggesting that they do not have a clear-cut status and norm, in contrast to ESLs which are “institutionalized” in educational and bureaucratic contexts. Insofar as they are not used for “internal” communication and insofar as they are the outcomes of formal teaching, XFLs are different from pidgins. However, the remoteness of the TL outside the EFL classroom may well make the English of the tour guide an interlanguage worthy of comparison

270

Rajend Mesthrie

with pidgins and jargons. Furthermore, XFLs meet their mirror image in “foreigner talk.”

2.4

Foreigner talk

Foreigner Talk (FT), the way an L1 speaker of a language X speaks to someone in a simplified version of X who has little or no competence in X, is generally characterized as a special register (or style) of X. As such, it is relevant to contact linguistics as a possible ingredient in the formation of P/Cs. That is, FT is part of the sum total of utterances that form the TL. Clearly, FT is not a contact language in itself, since only one party in the conversation uses it, and since it does not have a crystallized status. Rather it is an ad hoc simplification of language X by individual speakers of X, using basic iconic structures, repetitions, deletion of unstressed grammatical elements, gestures, etc. In some cases where unequal contact between two groups of people is sustained, FT may develop as a conventionalized code with certain grammatical regularities (Hinnenkamp 1984). Ferguson (1971, 1977) claimed FT to have some overlaps with baby talk (the way caregivers speak to babies) and pidgins. However, this is not true of Dutch FT, which is different from typical Dutch baby talk and from pidgin Dutch, notably in word order (Den Besten, Muysken, & Smith 1995, p. 97).

2.5

Semi-creoles

Not a fully established term in the literature, “semi-creole” refers to a variety which shows similarities in structure to known creoles and whose historical circumstances make it likely for something akin to creolization to have occurred. Putative cases involve the transplanted languages Reunion French (Baker & Corne 1982), Brazilian Portuguese (Holm 1988, 2000), Afrikaans (Roberge 1995), and African American English (Holm 1988; Schneider 1990). The term, which is based on Schuchardt’s (1889, p. 480) notion of Halbkreolisch, was used by Reinecke (1937, p. 559) in connection with Afrikaans. For example, in comparison to Dutch, Afrikaans exhibits the following characteristics: 1 2 3 4

5

loss of all traces of ablaut, and thus, in conjunction with (2) below, the distinction between strong and weak verbs; loss of all regular verb endings pertaining to person, number, and tense; replacement of past tense by perfective aspect, using the prefix ge-; introduction of constructions like double negation, double diminutives, and reduplication, and increase in the proportion of single diminutives; and loss of gender in the article system.

On the other hand, Afrikaans has retained some striking Germanic properties like:

Pidgins/Creoles and Contact Languages 271 1 2 3 4 5

V2 order in main clauses; main verb in final position if V2 is filled by an auxiliary (the “satzklammer”); the split order rule placing the verb in final position in subordinate clauses; a host of morphological features: ge- as perfective morpheme, nouncompounding; an array of verb-prefixation processes; and formal distinction between present and past modal auxiliaries.

Following Raidt (1983), the existence of Afrikaans as a separate language, one showing the striking linguistic changes cited above, is usually dated to about 1750–75, i.e., 100 to 125 years after the initial settlement; though “there are good reasons to be skeptical of the received chronology” (Roberge 1995, p. 72) in favor of a period shorter than a century. But there are also prominent retentions that make acrolectal Cape Dutch seem continuous with its European ancestor (Deumert 2001). A medial position seems plausible between the creolist view (e.g., Valkhoff 1966) and the “diachronic purist” view (e.g., Raidt 1983). Roberge (1995) and Den Besten (1989) articulate such a medial position in analyzing modern standard Afrikaans as a synthesis between acrolectal Cape Dutch of the nineteenth century (which was never pidginized or creolized) and a pidgin form used by indigenous Khoekhoen (and to some extent the slave population). The semi-creole nature of present-day standard Afrikaans, for these authors, lies in the accommodation of certain features of the pidgin into the superstrate, rather than a pidgin-into-creole scenario. (The latter scenario may have applied to nonstandard varieties that stabilized in the interior in the absence of superstrate speakers, though this is not at issue here.) This accommodation became accepted when a new standardizing ideology arose in the late nineteenth century that rejected “Dutch” in favour of “Afrikaans” (whose literal meaning is “of Africa”). Instead of being rejected, features from the pidgin were “ennobled” (Combrink 1978, p. 85) as manifestations of the new language. Some authors doubt the logic of the concept “semi-creole”; for example, Kaye (1990, p. 301) quips that it is tantamount to claiming that someone is partially pregnant. However, the historical juxtaposition of a superstrate itself subject to contact-induced changes with a pidgin seems to warrant the term, at least for Afrikaans. For other cases involving different types of restructuring, see Schneider (1990) and Holm (2000).

2.6

Creoloids and “language-shift varieties”

A related, and equally disputed, concept is that of the creoloid. The term dates to Platt’s (1975) discussion of Singapore English, which arose under colonialism in a highly multilingual context that included several Chinese languages, Malay, and Tamil. Although the variety owed its existence mainly to an educational setting in which the TL was made accessible, Platt noted several grammatical features of colloquial Singapore English that reminded him of creolization, most notably features of the copula system and tense marking. As checklists go, his was a bit thin; though one should point out that this was

272

Rajend Mesthrie

before Bickerton’s Roots of Language (1981, ch. 2), which posited the influential (if ultimately flawed) notion of 12 features of creole grammars. (Platt does not appear to have been influenced by the features noted by Taylor, 1971.) Nevertheless, he had put his finger on something significant which he was able to build on in subsequent work (e.g., Ho & Platt 1993, showing significant similarities between the semantics of article usage in Singapore English and Hawaiian Creole English as studied by Bickerton 1977b). Platt’s adoption of the term “basilect” for the sub-variety in question perhaps prejudged the issue of how similar early-fossilized group interlanguages were to creoles. Further research did indeed show up major differences in Singapore English between basilect and the superstrate. Ritchie (1986), for example, emphasized substrate influences from Chinese in Singapore English which included: 1

2 3

topic – comment structures occurring more frequently and in a wider variety of contexts than standard English, e.g., Certain medicine we don’t stock – in our dispensary; zero pronouns (or pro-drop), e.g., – can speak Cantonese also; lack of dummy subjects, e.g., In China, where – got people go to English school? ‘Where do you get people going to English school in China?’ – i.e., ‘There aren’t people who go to English school in China.’

Ritchie argued for a typological difference between the basilect (topic – comment) and the superstrate (subject – predicate). In view of this difference as well as features outlined in Ho and Platt (1993), the case is fairly strong that Singapore English, though always targeted toward the superstrate in an educational setting and presumably mutually intelligible with the superstrate, has more than passing resemblance to a creole (i.e., is indeed a creoloid in Platt’s terms). Platt (1975, p. 372) defined the term “creoloid” in terms of the following list of characteristics (which I have reworded slightly): 1 2 3 4 5 6

It has structural similarities to creole languages. It did not evolve from a prior pidgin. It developed from interference from several languages in contact. Its superstrate is an official language of the territory. It (the creoloid) is used as one of several “native” languages by the speech community. It is a medium of interethnic communication within the speech community.

Criterion (5) is problematic, as suggested by Platt’s scare-quotes on “native.” (Platt appears to intend the New English sense of “nativized” (i.e., used and processed like an L1), rather the creolistics sense of “L1.” I propose that a creoloid prototypically be considered an L2 with significant restructuring/ interference; in contrast to a semi-creole, which is an L1 (superstrate) with significant restructuring as discussed above. Holm (1988, p. 10) expresses skepticism about the concept of a creoloid, seeing it as a temporary

Pidgins/Creoles and Contact Languages 273 interlanguage phenomenon with limited grammatical interference. This is not the case with all ESLs. I argue in section 4 that certain creoloids that subsequently undergo language shift amidst limited access to the TL are relevant to creolistics. I use the term language-shift varieties for such ex-creoloids. Some clear-cut cases are Hiberno English, South African Indian English, and varieties of American Indian English. The position of Singapore English is perhaps more complex: whilst it is on its way to becoming a language of the home, many of the substrate languages are being maintained (Gupta 1994). What we seem to have is a shift in dominance rather than a complete language shift.

2.7

Koinés

A koiné is a variety essentially arising out of contact involving dialects of the same language. The result of such contact is typically a low prestige lingua franca which is structurally continuous with its antecedents (Siegel 1985, 1997). The Greek prototype (the Koiné) was a mixed dialect (Thomson 1972), arising out of Greek territorial expansion in the third century bc. It was initially a second dialect or second language for a majority of its speakers, co-existing in Greece with other city-state dialects, before gaining prestige on account of its association with government and administration, and later as the language of the New Testament. It eventually became the basis for the modern colloquial dialects. On most historical formulations koinés and creoles are easily distinguished from each other, the former arising largely from dialect shifts, the latter from language shifts. Koinés thus have only passing relevance to creole studies, as a means of characterizing how far-reaching the restructuring of the lexifier is under conditions of creolization. However, Mufwene (2001) has suggested two reasons why koinés might be of greater significance than this. First, he proposes that the superstrates in the colonies were themselves koinés, i.e., distinct varieties arising from accommodation between dialects of the superstrate, and hence an essential ingredient that preceded the creolization process. This appears to be untrue of most colonial varieties of Dutch (Silvia Kouwenberg, p.c.) Second, (and not many analysts agree with Mufwene on this), initial creolization was simply a continuation of the koineization process. Only later when larger numbers of slaves arrived on the plantations was there less access to the koiné and significant imperfect replication of its norms. Following arguments that go back to Chaudenson (1979) and Alleyne (1980), Mufwene emphasizes that each cohort generation of newly arrived slaves would have targeted the variety of earlier arrivals, thus getting successively further away from, rather than closer to, the original koiné. For Mufwene, colonial European koinés, creolization, and second language colonial varieties (New Englishes, etc.) are all cut from the same cloth, involving recourse to substrata and superstrata as filtered by linguistic universals (the “complementary hypothesis”) and mediated by the founder principle, which gives prominence to the role of early socially significant groups of

274

Rajend Mesthrie

learners of the koiné. Mufwene insists that all language learning and transmission is subject to this formulation. Note that in this view the particular distinction between creoloid and semi-creole would fall away (and indeed many other distinctions).

2.8

Bilingual mixed languages

Bilingualism and code-switching may appear to involve different linguistic processes from those which engendered pidgins and/or creoles, insofar as they involve language maintenance and close familiarity with the original codes in contact. However, in quite a few cases bilingualism and codeswitching have led to the stabilization of a new code that draws on the norms of the languages in intimate contact. Such mixing may no longer be ad hoc – i.e., dependent upon speaker competence, style, and choice. Rather a new code arises from which “unmixing” is not always possible (or desired). McCormick (2002) describes an urban vernacular drawing upon English and Afrikaans that arose amongst people once classified as “Colored” (as opposed to “Black”) in parts of Cape Town like the remnants of District Six. Mixing of the two codes is evident in aspects of segmental phonology and intonation (in which the norms of Afrikaans tend to prevail); in syntax where some new norms show a convergence between Afrikaans and English; and in lexis where both Afrikaans and English words are in use, the latter providing the more formal terms. What makes this a particularly intimate mix is the similarity between the lexifiers, Afrikaans and English, in which a large number of cognates can be found: e.g., eet ~ eat; son ~ sun; maak ~ make; vas ~ fast; drink ~ drink. In addition the lexicons of the two codes have become highly compatible via (1) a number of English borrowings into vernacular Afrikaans; (2) earlier Afrikaans borrowings into English; and (3) calquing (e.g., she did scold us out, based on Afrikaans uitgeskel). In respect of grammar, McCormick provides examples of similar processes of convergence. The Afrikaans word order has become more variable, admitting sentences with Subject-Verb-Object (SVO) order in main clauses that have an auxiliary, contrary to traditional Afrikaans satzklammer which requires the verb in final position in this instance. Second, SVO occurs in certain subordinate clauses, unlike standard Afrikaans which is more rigidly SOV. The complementizer dat of Afrikaans has become variable, allowing a zero alternative rather like the (cognate) English subordinator that. Areas where English converges toward Afrikaans are in the loss of verb endings in the present and in tense and aspect generally. It appears that speakers who use this code as their vernacular are, with appropriate English-medium education, able to “unmix” in formal situations requiring English; but many speakers are unable to “unmix” toward an unambiguous Afrikaans norm in informal speech. An interesting kind of language shift appears to be on the way from Afrikaans as vernacular and “H” (High status) language, to a mixed code as vernacular and English as H language.

Pidgins/Creoles and Contact Languages 275 This may well become a recurring formula in urban areas under globalization. More intriguing mixing can be found in what Bakker and Muysken (1995) and Muysken (this volume) call “intertwined languages.” The prototype is Michif – a mixed code of the fur-trading eighteenth- and nineteenth-century frontier in Canada (Bakker & Papen 1995). As spoken by Métis descendants of both French and indigenous peoples, Michif has its verbs largely from Cree and its nouns largely from French. A related type of mixing conceals rather than displays the fact of contact, showing only syntactic convergence whilst keeping the lexis of the antecedent codes distinct. This may be called the “Kupwar option,” following the classic findings of Gumperz and Wilson (1971). In Kupwar, a village in the state of Maharashtra, India, speakers may be bilingual or trilingual in the main codes Kannada, Marathi, and Urdu, which have been in contact in this particular village (but not necessarily in other parts of the state) for hundreds of years. At the lexical level the languages are kept apart with relatively little borrowing. However, in matters of syntax, intimate bilingualism has led to a restructuring of the codes toward one “deep” village syntax. Gumperz and Wilson raised the question whether such convergence was a kind of creolization; and the same could be asked more generally about urban mixed codes and intertwined languages. In Thomason and Kaufman’s terms, they do not involve pidginization (and hence no “pidgin-into-creole” cycle); they do not involve abrupt creolization (since they are the outcomes of gradual bilingualism); nor are they examples of shift-induced restructuring (since there is either no language shift, or only a partial shift to the mixed code in informal speech). However, they do satisfy Thomason and Kaufman’s requirement that contact languages (in the strict sense) show a break between transmission of the lexicon and grammar of any antecedent language. The question has to be raised whether mixed codes symbolize a new identity that draws on more than one cultural tradition. In this connection Thomason (2001a, p. 198) suggests that “whereas pidgins and creoles arise from a need for a medium of communication among groups that share no common language, the other mixed languages arise instead within a single social or ethnic group because of a desire, perhaps even a need, for an in-group language.”

3 Koinés Revisited The koiné forms of European languages that developed in plantation colonies would have drawn on a number of sources. For English colonies these would have included standard forms of English and a number of regional dialects of the UK, as well as sociolects like sailors’ and working-class dialects. Hiberno English is an interesting ingredient here, since while nominally belonging to the superstrate, it was really the language of indentured laborers who worked with the slaves. Rickford (1986) examined the role of Irish English in the formation of habitual be in African American English and Caribbean varieties of

276

Rajend Mesthrie

English, concluding that Irish English usage reinforced substrate West African language options. Ships’ English of the era of slavery – another important ingredient of the superstrate – is examined by Bailey and Ross (1988). Trudgill (1986) proposes the following characteristics of a koiné like Australian English. Unification between previously distinct groups (based on region or social factors) results in speech accommodation, which results in large degree in the neutralization of the social meaning attached to linguistic variants. More salient morphological features will be retained while less salient ones (in terms of syllable structure and stress) will be less likely to remain. Of the remaining forms, those which are more regular, and therefore more easily learnt, stand a better chance of being retained than irregular forms which will be “accommodated out.” Where several alternants exist, frequency of a particular form may also play a significant role: the more dialects a form exists in, the greater its chances of survival in the koiné. Finally, in determining who accommodates to whom, and what forms win out, demographic factors involving proportions of different dialect speakers and relative prestige of groups will also play an important role. It is not clear if all of these principles apply to the Caribbean English superstrate, in which there was less social leveling. A subordinate koiné is perhaps different from superstrate koinés in sociopolitical terms; however, the processes of koiné formation are not that different from those proposed by Trudgill, as I shall show in some detail. A particular set of koinés that is of relevance to the study of P/Cs comprises varieties of North Indian languages that amalgamated on many of the erstwhile slaveholding plantations. In the aftermath of the abolition of slavery in the early nineteenth century, plantation owners sought alternate sources of labor to compensate for the reluctance of manumitted slaves to remain on the plantations. A system of indenture arose that brought in Asians, primarily (East) Indians, on fixed contracts. Although this practice was dubbed “a new system of slavery” by Lord John Russell – a phrase taken up as his book title by the historian Hugh Tinker (1974) – it is debatable whether indenture was quite as harsh as slavery. Large numbers of villagers from north and south India were induced sometimes of their own free will (during bad seasons), and sometimes by deception, to travel to the colonies. This large-scale immigration to Mauritius (1834–1917), Guyana (1839–1916), Trinidad (1845–1917), South Africa (1860–1911), Suriname (1873–1915), and Fiji (1879–1946) was well documented by the British administration, which was obliged by the government of India to appoint a Protector of Indians in each colony, under whose office were kept ships’ lists and subsequent records pertaining to each laborer who immigrated. Records exist pertaining to the names, villages of origin, caste, age, and identifying body marks of each immigrant for the entire period of immigration – which spanned about a century (the first immigrations being to Mauritius in 1834 and the last to Fiji in 1946). From these records historians and linguists have been able to make clear inferences about the linguistic background of immigrants and evaluate subsequent innovations in their speech. Languages spoken by the indentured laborers were not always

Pidgins/Creoles and Contact Languages 277 mutually intelligible. In particular those from the north spoke an Indic vernacular (usually Bhojpuri, Awadhi, and less commonly Magahi, Bengali, Braj, etc.), while those from the south spoke a Dravidian language (usually Tamil or Telugu). The north Indian vernaculars were more or less mutually intelligible whilst Tamil and Telugu were not. Moreover, the north and south Indian languages, whilst typologically fairly similar, were mutually unintelligible. There was thus fertile ground for the possible development of an “internal” lingua franca amongst incoming Indians, yet an Indo-Dravidian pidgin did not develop. There are several reasons for this. In the case of Natal the preexisting pidgin, Fanakalo (lexified by Zulu), acted as a relatively simple lingua franca which served for vertical communication with employers as well as for horizontal communication with the indigenous Zulu people, and in some cases as a lingua franca across the Indic–Dravidian divide. Some plantation bosses with prior experience in Mauritius or India knew Hindustani. In Fiji two pidgins were used; one was a pidgin form of Fijian that predated the arrival of Indians, and the other a pidgin form of Hindi that arose as south Indians and English overseers tried to communicate with north Indians and each other. In most other colonies Indians acquired the pre-existing creole language via ordinary processes of second language acquisition. In effect, these practices meant that Indian languages survived for at least a hundred years in each overseas territory. For ease of exposition I will focus on the north Indian languages and dialects. These underwent three simultaneous processes: 1 2

3

accommodation of mutually intelligible dialects (koineization) – discussed further below; structural change contingent on social change, especially the weakening of caste distinctions – this resulted in a more egalitarian pronoun and verb system which eschewed respectful “V” forms; indigenization in a new territory – as evidenced by a large number of loan words from English and a smaller number from Afrikaans and Zulu.

As North Indians emanated from a wide area that spanned several languages and numerous sub-varieties, in each colony a compromise dialect arose, usually called Hindi or Hindustani (the terms for the umbrella standard language) in the official sources, though it was in practice a koiné based on a range of vernaculars like Bhojpuri, Awadhi, and Magahi. The antecedent vernaculars appear to have been replaced by the koiné early on, probably within one or two generations. Judging from the few remaining speakers who were alive during the time of my research in the early 1980s, incoming immigrants from India learnt the koiné, which probably became their main form of communication. (They would retain their home vernaculars and a form of Hindi as additional dialects, but not really pass them on to the next generation, which used the koiné.) Remarkably similar processes occurred in all the colonies to which North Indians went. These varieties, which may for convenience be labeled “overseas Bhojpuri-Hindi” (OBH), show considerable morphological

278

Rajend Mesthrie

reduction in the following areas: (1) gender on noun and verb; (2) stylistic options for pronouns; (3) number on nouns; (4) agreement between adjectives and nouns; and (5) verb inflections denoting “respect.” On the other hand, despite considerable variation in verb endings in the antecedent dialects, verb inflections are prominent in the koiné, with demographic factors dictating which dialect forms eventually won out over others. (The survival of verb endings in the OBH koinés, incidentally, contrasts strongly with their dramatic loss in the semi-creole Afrikaans.) These verb endings differ slightly in the various OBHs (e.g., ‘(s)he saw’ is dekh-lak in Mauritius; dekh-le in Guyana; dekh-al in Trinidad; dekh-las ~ dekh-a ~ dekh-is in South Africa; dekh-is in Suriname; dekh-is ~ dekh-a in Fiji. To a large extent such variants can be explained in terms of recruiting patterns in India. In north India recruitment was at first concentrated in the more easterly areas of Bengal Presidency and Bihar (Tinker 1974, pp. 46–53); and moved in a westerly direction into the interior. Thus Mauritius, the first colony to which indentured workers were sent, shows the most easterly character of the OBHs, whilst Fiji, the last colony to import indentured workers, shows the greatest influence of varieties spoken further into the interior of north India. Other OBH varieties fit fairly well into this “transplanted dialect continuum” framework (Mesthrie 1992b, pp. 71–7). Specific developments in OBH can be used to evaluate salient debates in creolistics. For example the “cafeteria principle,” contrary to ideas in creolistics, does hold for koinés: a blend of features from a variety of north Indian languages can be seen in each of the new colonial varieties. Gambhir’s (1988) analysis of the origins of Guyanese Bhojpuri (GB) shows this blending of linguistic features (see table 11.1). In addition, Gambhir notes the existence of several features of Bhojpuri-Hindi and Standard Hindi that are not found in GB. Of these, nine are found only in Indian Bhojpuri, five are common to all the other Bhojpuri-Hindi dialects, and one is unique to Standard Hindi.

Table 11.1 Multiple/koiné origins of Guyanese Bhojpuri (adapted from Gambhir 1988, pp. 78–81) Features of GB 21 6 2 2 1 6 4 0

common to all the Bhojpuri-Hindi dialects present in the contact situation; of the 21, 18 also occur in Standard Hindi common to all but the Western Hindi dialects and Standard Hindi common to all but Indian Bhojpuri and Standard Hindi common to all but Indian Bhojpuri found in Western Hindi only found in Indian Bhojpuri only features unique to GB features found only in GB and Standard Hindi

Pidgins/Creoles and Contact Languages 279 Table 11.1 is slightly misleading (as baldly summarized here) insofar as it conflates very specific syntactic constructions (e.g., a special negative copula form) with very broad language criteria (e.g., word order; phonological systems). The blending of features is not entirely random, however. Features common to the array of input dialects/languages are most prominent in the Guyanese koiné. Of the varieties that were the sole contributors of particular features, the majority dialect Bhojpuri (according to recruiting patterns) is the best represented, though highly marked forms from this majority dialect did not survive. Guyanese Bhojuri thus shows that the cafeteria principle is constrained by matters pertaining to demography and linguistic markedness. Finally, it is not clear if the founder principle (Mufwene 1996) is supported by OBH research. It does not seem to be the case that the first indentured immigrants entering the colonies were the ones whose dialects had an advantage over subsequent immigrants. For this to happen Bengali should have been the variety spoken in South Africa, whereas the initial majority from Purulia (speaking Bengali) in the first year were soon swamped by speakers of other languages and dialects. Within KwaZulu-Natal province in South Africa, there remain today two main dialects corresponding to recruitment patterns in India and settlement patterns within the colony, rather than the growth of one founding dialect. Likewise, by the founder principle, the variety spoken in Mauritius should not be a variety of eastern Bhojpuri, but rather the dialects of what the British called “hill coolies,” speaking “tribal” Austronesian languages (or a koiné based on them). Although the matter is deserving of further research, it appears that for OBH simple demography outweighs any founder effect.

4 Creoloids and Language-Shift Varieties Revisited In this section I focus on South African Indian English (SAIE), which is in many ways similar to Singapore English with its multiple substrates, rapid social change among immigrants, and the co-existence of basilectal (L2) varieties of English with more classroom-based forms. At one stage (when it was an L2) SAIE could have been characterized as a creoloid; today it is a “language-shift variety” (in showing the stabilization of a creoloid L2 into an L1, with little input from TL speakers). This section will emphasize some similarities between language-shift Englishes and known creoles. Indeed, Siegel (1994, p. 189) concludes that “SAIE provides a missing link in the chain of possible contact varieties,” and Mufwene (2001, p. 111) likewise suggests that SAIE “has fallen between the cracks” of existing contact typologies. Such claims can best be evaluated within the framework of Thomason and Kaufman’s (1988) characterization of language shift and abrupt creolization. Although originally advocating a clear distinction between shift with normal and abnormal transmission, Thomason’s most recent summary (2001b) plays down this distinction, and is more in line with the theory of Chaudenson

280

Rajend Mesthrie

(1979) and Mufwene (2001): “Creoles emerge through a gradual process of repeated second language acquisition, with shift-induced interference at each stage accumulating until the result is a creole language, too distant linguistically to be considered a dialect of the lexifier language that provided the original target for language learning” (Thomason 2001b, p. 462). This formulation is not accepted by all creolists, especially the links between SLA and creolization, which for analysts like Baker (1990) differ crucially in respect of “targetedness.” It is a perspective that is, however, relevant to the study of creoloids and their language-shift offshoots. Although Thomason (p.c., 2001) is not in favor of an interpretation that links creoles with what I term “language shift Englishes” (Mesthrie 1992a, pp. 3–5), it is a postulate worthy of future research that the main difference between Creole Englishes and language-shift Englishes is not so much in the original conditions of contact and restructuring as in the degree of subsequent modifications. Shift-induced interference may drift generationally into a creole in the absence of the superstrate as a viable target; or it may “relapse” generationally in the direction of the superstrate in languageshift varieties, as the TL becomes increasingly more available. Although the end products might differ (creole vs. “language-shift variety/social dialect of superstrate”), at early stages the processes of “language creation” might have been similar. Whereas this might be conjecture for a language-shift variety like Hiberno English, it is more visible in SAIE, since the policy of apartheid prevented the absorption of this shift-induced variety into the superstrate (South African English). Although the shift from the many Indian languages cited above was gradual rather than abrupt, there were speakers who were using English in a “broken form” with each other within 50 years of the first immigrations (according to M. K. Gandhi in his newspaper, Indian Opinion in 1909). The use of English was necessitated by the existence of five or more Indian languages across the Indic–Dravidian divide, none of which qualified as a language of identity for the whole community, as they were too strongly identified as ethnic languages. By the 1940s the sociolect was sufficiently established for there to be a satirical radio series, Applesammy and Naidoo, drawing on the L2 as a source of humor (Mesthrie 2002). This L2 was pressed into service first in formal domains (notably education) and then in increasingly informal domains, culminating in a domestic language shift that began in the 1960s and is reaching its conclusion at present. That is, language shift began 100 years after the initial arrival of Indians, in other words a mere 50 years after the last shipload. The shift took place with minimal input from native speakers on account of (1) initial sociocultural segregation of Indians from English speakers, and a fair bit of colonial racism; (2) rigid apartheid after 1948, which minimized social contacts between English speakers and Indians; and (3) education not being widespread in the first 75 years (Mesthrie 1992a, pp. 18–29). The result is that, although English became the main lingua franca and a home language, it did so with minimal input from native speakers. SAIE shows quite clearly the effects of an L2 stabilizing as an L1. This process throws up a great deal of variation that is clearly correlatable with a creole

Pidgins/Creoles and Contact Languages 281 continuum. The basilect shows a large measure of syntactic variation between OV-like structures from Indic and/or Dravidian and VO structures of the target language. For example, Mesthrie (1992a, pp. 71–100) documents immense variation within the relative clause, which can be categorized into four broad classes: 1 2

3

4

standard English forms inherited from the classroom; “almost standard” English dialect forms, involving nonstandard relative pronouns such as what, which, which one, and contact relatives, in which a subject NP is relativized without an overt relative pronoun, e.g., We talking about my friend Ø lives down there); substrate-induced forms, including: (a) an Indic correlative strategy, e.g., [Which-car [they supposed to give us]], someone else got it = ‘Someone else got the car which they were supposed to give us’); (b) an Indic participial strategy (e.g., [That [Neela’s knitted] jersey] is gone white = ‘That jersey which Neela knitted/knitted by Neela has gone white’); and (c) a Dravidian premodifying strategy (e.g., That’s all [[we had] trouble] = ‘That’s all the trouble we had’); and discourse-based forms.

The class in (4) seems to be neither an inheritance from the substrates nor the superstrate (either in the classroom or in informal input from TL speakers). Rather it appears to draw on universals of discourse organization, i.e., building grammar out of discourse without too much “interference” from the standard. Examples include “near-relatives,” with clause juxtaposition (e.g., I’m a man I don’ go church an’ all = ‘I’m a man who doesn’t go to church and such’) and use of resumptive pronouns (e.g., I was this girl that I always used to read in the bus = ‘I was a girl who always used to read in the bus’). Relative clauses in SAIE thus give ample evidence for speakers building a grammar out of a variety of materials, some inherited from the TL, others from substrata, and still others from “universals” of discourse. The same processes are manifest in almost every facet of its morphology and syntax. Moreover, the relationship between basilect, mesolect, and acrolect is rather similar to traditional accounts of decreolization. Broadly speaking, the basilect shows the development of a core grammar via what Valdman (1977, p. 55), building on Hymes (1971a), called in a creole context “expansion of inner form” and “complexification of outer form.” Essentially “inner form” refers to core grammatical structure like tense and aspect, maximal projections like noun and verb phrases and main clauses and the development of subordination. The “building” of relative clauses in SAIE can be seen as an instance of this process. “Outer form” refers to elements like inflections and adverbials that flesh out the core structures by marking functions of phrases and clauses and/or showing relations between them. As noted for creole languages by Valdman

282

Rajend Mesthrie

(1977), SAIE outer form frequently shows the use of TL material with non-TL meanings, functions, and distribution. Thus, time, side, part, way are used in the SAIE basilect as quasi-postpositions, with functions similar to TL prepositions like temporal at, spatial in, temporal at, and manner in, respectively (e.g., Morning-part is too busy = ‘It is very busy in the mornings’). Another example shows the replacement of the TL “outer form” both X . . . and Y by the basilectal SAIE construction X too . . . Y too (e.g., I made rice too, I made roti too (= ‘I made rice and roti’; roti = ‘flat, round unleavened bread’). In general, as I show in Mesthrie (1992a, pp. 101–27), the basilect uses a large number of features associated with OV rather than VO languages; here the postnominal coordination pattern leans toward a “Greenbergian” OV patterning. In contrast, the mesolect (which has more overt prestige) often replaces basilectal forms with target language forms on the surface, without necessarily changing the basilectal nature of the syntax. For example, one mesolectal speaker produced the following sentence in an interview: My dad was a soccerite as well, he was a musician as well. In avoiding the unprestigious basilectal form too . . . too, the speaker replaced it with the standard TL form as well, whilst unwittingly keeping the basilectal structure. Such ad hoc replacements are quite common in the mesolect: the result, as Bickerton (1975) stresses in his discussion of the Guyanese Creole mesolect, is that speakers attempting to eschew the basilect produce constructions that are no more acrolectal than their basilectal equivalents were. Schneider (1990) described a “cline of creoleness” for the creoles and semicreoles of the Caribbean. I suggest on the basis of evidence given in this section that SAIE gives considerable support for the more general notion of a “cline of contactness.” In other words, it is a missing link between creoles and second-language varieties. It falls into a class of language-shift varieties whose genesis was not abrupt, but perhaps not all that gradual either, and which involves more than one substrate, often substrates that are lexically and syntactically diverse. That they do not quite go all the way along the cline of creoleness and are, at a pinch, intelligible to outsiders may well be related to their origins in the classroom as well as the plantation and street. SAIE fits the decreolization model (e.g., De Camp 1971) reasonably well. With the gradual increase in education, mesolects developed side by side with the basilects. That is, in Mufwene’s (2001) formulation there was decreasing rather than increasing basilectalization, because of the influence of school and urbanization from the 1930s onwards. However, because of apartheid, the basilect survived in a robust form, even as speakers developed stylistic flexibility. This code symbolized a measure of solidarity while class continua arose that were responsive to the pressures of the acrolect. In one respect Mufwene’s language ecology thesis is not borne out: it is doubtful that a colonial koiné was a particularly influential part of the initial state out of which SAIE developed. The colonial form of English characteristic of settlers in Natal province bears little resemblance to the SAIE system in phonology, syntax, or discourse

Pidgins/Creoles and Contact Languages 283 preferences. Furthermore the early English inputs to SAIE were probably too diverse to exert a strong influence on the development of SAIE. These inputs included ESL missionaries from Europe with varying English competence, Scots and Irish English as well as English regional dialects, classroom English, Christian Indian teachers, and some “Colored” teachers from Natal.

5 Conclusion Pidgin and creole studies have long been the driving force in contact linguistics, providing frameworks, models, descriptions, and debates that have contributed to the description and understanding of other forms of language contact. I have suggested that the study of these other forms of contact can in turn enrich creolistics. I have done so by assessing principles of contact proposed by creolists like the cafeteria principle and the founder principle and by examining the nature of nativization in situations of shift still amenable to observation. A continuing symbiosis between P/C studies and that of “other” contact phenomena is not only in order; it is to be welcomed.

REFERENCES Alleyne, Mervyn (1980) Comparative Afro-American. Ann Arbor: Karoma. Arends, Jacques (1993) Towards a gradualist model of creolization. In: Francis Byrne and John Holm (eds.) Atlantic Meets Pacific. Amsterdam: John Benjamins, pp. 371–80. Bailey, Guy and Garry Ross (1988) The shape of the superstrate: Morphosyntactic features of Ship English. English World-Wide, 9, 193–212. Baker, Philip (1990) Off target? Journal of Pidgin and Creole Languages 5, 107–19. Baker, Philip (1995) Motivation in creole genesis. In: Philip Baker (ed.) From Contact to Creole and Beyond. London: University of Westminster Press, pp. 3–15. Baker, Philip (2000) Theories of creolization and the degree and nature of restructuring. In: Ingrid Neumann-Holzschuh and Edgar Schneider (eds.) Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins, pp. 41–63. Baker, Philip and Chris Corne (1982) Isle de France Creole: Affinities and Origins. Ann Arbor: Karoma. Bakker, Peter and Pieter Muysken (1995) Mixed languages and language intertwining. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 41–52. Bakker, Peter and Robert A. Papen (1995) Michif: A mixed language based on Cree and French. In: Sarah G. Thomason (ed.) Contact Languages: A Wider Perspective. Amsterdam: John Benjamins, pp. 295–363. Bickerton, Derek (1975) The Dynamics of a Creole System. Cambridge: Cambridge University Press. Bickerton, Derek (1977a) Pidginisation and creolisation: Language acquisition and language universals. In: Albert Valdman (ed.) Pidgin and Creole Linguistics. Bloomington: Indiana University Press, pp. 49–60.

284

Rajend Mesthrie

Bickerton, Derek (1977b) Change and Variation in Hawaiian English. Volume II: Creole Syntax. Manoa: University of Hawai[i, Social Sciences and Linguistics Institute. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1983) Comments on Valdman’s “Creolisation and second language acquisition.” In: Roger Andersen (ed.) Pidginisation and Creolisation as Language Acquisition. Rowley, MA: Newbury House, pp. 235–40. Carden, Guy and William A. Stewart (1988) Binding theory, bioprogram, and creolization: Evidence from Haitian Creole. Journal of Pidgin and Creole Languages 3, 1–67. Chaudenson, Robert (1979) Les créoles français [The French creoles]. Paris: Fernand Nathan. Chaudenson, Robert (1992) Des îles, des hommes, des languages: Essai sur le créolisation linguistique et culturelle [Islands, people and languages: Essay on linguistic and cultural creolization]. Paris: L’Harmattan. Combrink, Johan G. H. (1978) Afrikaans: Its origins and development. In: Leonard W. Lanham and K. P. Prinsloo (eds.) Language and Communication Studies in South Africa. Cape Town: Oxford University Press, pp. 69–95. De Camp, David (1971) Toward a generative analysis of the post-creole continuum. In: Dell Hymes (ed.), pp. 349–70. Den Besten, Hans (1989) From Khoekhoe foreigner talk via Hottentot Dutch to Afrikaans: The creation of a novel grammar. In: Martin Pütz and René Dirven (eds.) Wheels within Wheels: Papers of the Duisburg Symposium on Pidgin and Creole Languages. Frankfurt: Peter Lang, pp. 207–54. Den Besten, Hans, Pieter Muysken, and Norval Smith (1995) Theories focusing on the European input. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 87–98. Deumert, Andrea (2001) Language variation and standardization at the Cape (1880– 1922): A contribution to Afrikaans sociohistorical linguistics. Journal of Germanic Linguistics 13, 301–52. Ferguson, Charles A. (1971) Absence of copula and the notion of simplicity: A study of normal speech, baby talk, foreigner talk, and pidgins. In: Hymes (ed.), pp. 141–50. Ferguson, Charles A. (1977) Baby talk as a simplified register. In: Catherine Snow and Charles A. Ferguson (eds.) Talking to Children: Language Input and Acquisition. Cambridge: Cambridge University Press, pp. 209–35. Gambhir, Surendra (1988) Structural development of Guyanese Bhojpuri. In: Richard K. Barz and Jeff Siegel (eds.) Language Transplanted: The Development of Overseas Hindi. Wiesbaden: Otto Harrassowitz, pp. 70–94. Gumperz, John J. and Robert Wilson (1971) Convergence and creolisation: A case from the Indo-Aryan/Dravidian border. In: Hymes (ed.), pp. 151–67. Gupta, Anthea Fraser (1994) The Step-Tongue: Children’s English in Singapore. Clevedon, UK: Multilingual Matters. Hesseling, Dirk (1897) Het Hollandsch in Zuid-Afrika [Dutch in South Africa]. De Gids, 60, 138–62. Hickey, Raymond (1995) An assessment of language contact in the development of Irish English. In: Jacek Fisiak (ed.) Linguistic change under Contact Conditions. Berlin: Mouton de Gruyter, pp. 109–30. Hinnenkamp, Volker (1984) Eye-witnessing pidginisation? Structural and sociolinguistic aspects of German and Turkish foreigner talk. York Papers in Linguistics 11, 153–66.

Pidgins/Creoles and Contact Languages 285 Ho, Mian L. and John T. Platt (1993) The Dynamics of a Contact Continuum. Oxford: Clarendon. Holm, John (1988) Pidgins and Creoles, vol. 1. Cambridge: Cambridge University Press. Holm, John (2000) Semi-creolization: Problems in the development of theory. In: Ingrid Neumann-Holzschuh and Edgar Schneider (eds.) Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins, pp. 19–40. Hymes, Dell (1971a) Introduction to Section III. In: Hymes (ed.), pp. 151–67. Hymes, Dell (ed.) (1971b) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. Kachru, Braj B. (1983) The Other Tongue: English Across Cultures. Oxford: Pergamon. Kaye, Alan (1990) Observations on pidginistics and creolistics. Semiotica 78, 285–348. Leap, William (1993) American Indian English. Salt Lake City: University of Utah Press. McCormick, Kathleen K. (2002) Language Use in District Six, Cape Town. Oxford: Oxford University Press. Mesthrie, Rajend (1992a) English in Language Shift: The History, Structure and Sociolinguistics of South African Indian English. Cambridge: Cambridge University Press. Mesthrie, Rajend (1992b) Language in Indenture: A Sociolinguistic History of BhojpuriHindi in South Africa. London: Routledge. Mesthrie, Rajend (2002) Mock languages and symbolic power: The South African radio series Applesammy and Naidoo. World Englishes 21, 99–112. Mufwene, Salikoko S. (1996) The founder principle in creole genesis. Diachronica 13, 83–134. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mühlhäusler, Peter (1986) Pidgin and Creole Linguistics. Oxford: Blackwell. Paikeday, Thomas (1985) The Native Speaker is Dead. London: Paikeday Publishing. Platt, John T. (1975) The Singapore English speech continuum and its basilect “Singlish” as a creoloid. Anthropological Linguistics 17, 363–74. Raidt, Edit (1983) Einführung in Geschichte und Struktur des Afrikaans [Introduction to the history and structure of Afrikaans]. Darmstadt: Wissenschaftliche Buchgesellschaft. Reinecke, John (1937) Marginal Languages: A Sociological Survey of the Creole Languages and Trade Jargons. PhD dissertation, Yale University. Rickford, John (1986) Social contact and linguistic diffusion: Hiberno English and New World Black English. Language 62, 245–89. Ritchie, William C. (1986) Second-language acquisition research and the study of nonnative varieties of English: Some issues in common. World Englishes 5, 15–30. Roberge, Paul (1995) The formation of Afrikaans. In: Rajend Mesthrie (ed.) Language and Social History: Studies in South African Sociolinguistics. Cape Town: David Philip, pp. 68–88. Schneider, Edgar (1990) The cline of creoleness in English-oriented creoles and semicreoles of the Caribbean. English World-Wide 11, 79–113. Schuchardt, Hugo (1882) Kreolische Studien I: Über das Negerportugiesische von San Thome (Westafrika) [Creole studies I: On the Negro Portuguese of Sao Tomé (West Africa)]. Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften zu Wien 101, 889–917. Schuchardt, Hugo (1889) Beiträge zur Kenntnis des kreolischen Romanisch: Allgemeineres über das Indo-portugiesische (Asioportugiesische) [Contribution to the study of Romance-based Creoles: Generalizations about Indo-Portuguese (Asian Portuguese)]. Zeitschrift für Romanische Philologie 13, 476–516.

286

Rajend Mesthrie

Schumann, John (1978) The Pidginization Process: A Model for Second Language Acquisition. Rowley, MA: Newbury House. Selinker, Larry (1972) Interlanguage. International Review of Applied Linguistics 10, 209–30. Siegel, Jeff (1985) Koines and koineization. Language in Society 14, 357–78. Siegel, Jeff (1994) Review of Rajend Mesthrie, English in Language Shift: The History, Structure and Sociolinguistics of South African Indian English. Journal of African Languages and Linguistics 15, 86–90. Siegel, Jeff (1997) Mixing, leveling, and pidgin/creole development. In: Arthur K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles. Amsterdam: John Benjamins, pp. 111–49. Singh, Rajendra (ed.) (1998) The Native Speaker: Multilingual Perspectives. New Delhi: Sage. Singler, John Victor (1990) On the use of sociohistorical criteria in the comparison of creoles. Linguistics 28, 645–59. Taylor, Douglas (1971) Grammatical and lexical affinities of creoles. In: Hymes (ed.) pp. 293–6. Thomason, Sarah G. (2001a) Language Contact: An Introduction. Edinburgh: Edinburgh University Press. Thomason, Sarah G. (2001b) Contact Languages. In: Rajend Mesthrie (ed.) Concise Encyclopedia of Sociolinguistics. Oxford: Pergamon, pp. 461–4. Thomason, Sarah G. and Kaufman, Terrence (1988) Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Thomson, George (1972) The Greek Language. Cambridge: Heffer. Tinker, Hugh (1974) A New System of Slavery: The Export of Indian Labour Overseas, 1830– 1920. Oxford: Oxford University Press. Trudgill, Peter (1986) Dialects in Contact. Oxford: Blackwell. Valdman, Albert (1977) Creolisation: Elaboration in the development of Creole French dialects. In: Albert Valdman (ed.) Pidgin and Creole Linguistics. Bloomington: Indiana University Press, pp. 155–89. Valkhoff, Marius (1966) Studies in Portuguese and Creole with Special Reference to South Africa. Johannesburg: Witwatersrand University Press. Van Name, Addison (1869–70) Contributions to creole grammar. Transactions of the American Philological Association 1, 123–67. Weinreich, Uriel (1953) Languages in Contact. New York: Linguistic Circle of New York. Winford, Donald (2000) Introduction: On the structure and status of pidgins and creoles. In: Arthur K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles. Amsterdam: John Benjamins, pp. 1–31.

Creole Studies and Multilingualism 287

12

Creole Studies and Multilingualism PIETER MUYSKEN

1 Introduction Creole studies is essentially part of historical linguistics. It tries to understand how a group of new languages emerged, and how their structural and lexical features can be related to the circumstances of their genesis. In contrast, multilingualism research lies at the intersection of sociolinguistics and psycholinguistics. It tries to understand, usually in contemporary settings, how multilinguals use their languages, and what are the consequences of multilingual language use for the individuals, their speech communities, and for the languages involved. Formulated in this way, multilingualism research and creole studies are far apart, and in fact the researchers involved in the two disciplines form different scientific networks, read different journals, fight about different things, etc. Nonetheless, there are good reasons to explore the links between the two fields, since, in anybody’s reckoning, pidgin and creole genesis is a multilingual matter. Crucial, in this respect, is the adoption of the Uniformitarian Hypothesis (Labov 1972), which leads to the assumption that the sociolinguistic and psycholinguistic processes operant in the genesis of pidgins and creoles in earlier times are the same as those operant in contemporary multilingual settings. The Uniformitarian Hypothesis is important because creole genesis is all about bi- and multilingualism. If there had been no bi- or multilingualism (henceforth, “multilingualism” will be used for both), creoles would have never arisen. Thus, viewed in a slightly more general perspective, creole studies and multilingualism research are both part of language contact research – as acknowledged, for instance, in the introductions to language contact studies by Sebba (1997) and Thomason (2001). The psycholinguistic processes we will be considering here have to do with the way the representation and processing of different languages interact in our cognitive system; one model for this has been proposed by Carol MyersScotton (e.g., 2002), and there is a large and rapidly growing literature on this. The sociolinguistic processes involved relate to the dynamics of multilingual

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

288

Pieter Muysken

interaction (Auer 1998a, 1998b), the transmission of innovations in communicative networks (Milroy & Milroy 1985), the role languages play as symbolic systems in group formation (Le Page & Tabouret-Keller 1985), and the evaluations and ideological constructs surrounding language use (Gal 1989; Woolard 1998). I will begin by briefly reviewing multilingualism in the settings in which the Caribbean creoles arose, assuming that creole languages elsewhere also emerged in circumstances where many languages were spoken (section 2). Then I turn to a number of subfields in language contact studies where the creole evidence and the evidence from more contemporary contact settings may be confronted (section 3). In conclusion, the perspective that Hugo Schuchardt, one of the founders of creole studies, took on language contact will be brought into relief (section 4). It should be emphasized that sometimes the link between creole studies and other domains in multilingualism research is tentative; I will not try to aim for some grand total scheme. I limit myself here to the Caribbean creoles, leaving aside pidgins and creoles spoken elsewhere. However, the Pacific and Africa have always known multilingualism to the same extent as the Caribbean, if not more so, and it should be clear that the same general conclusions should hold there whenever the circumstances of genesis are sufficiently similar.

2 2.1

Multilingualism in Creole Genesis Evidence for multilingualism

Turning for a moment to the evidence we have for the extent and nature of multilingualism in the context of creole genesis, at least four types of data should be mentioned. Most important, of course, are the historical demographic data on the origins of the people who came to the Caribbean, nationalities of settlers, ports from which slaveships departed, etc. These data are not without problems, since, for example, the port of shipment for slaves was not necessarily their place of origin and, therefore, does not necessarily indicate the languages they spoke (see Arends, this volume). A second important source is contemporary accounts of language use. To give but one example, Van Oldendorp ([1777] 1987, pp. 200–6) lists a phrase (“Christ has loved us and has washed away our sins with His blood”), some words (God, Heaven, Sun, Moon, Human, Hand, Foot, Head), and numerals (1–8) in 27 African languages that he found speakers of in the Virgin Islands. Third, there are the words and grammar features derived from different African languages in present-day Caribbean languages. Thus, Baker (1993) identifies 126 words of Bantu origin, 88 of Kwa origin, and 37 of West Atlantic/ Manding origin in the French-lexifier creoles of the Caribbean. There is, however, one important caveat here: the presence of African lexical elements from

Creole Studies and Multilingualism 289 different language backgrounds in a creole is not necessarily evidence for extensive multilingualism, since the different African languages could have been and often were introduced at different times. Finally, there are the secret and ritual languages, remnants of earlier African languages, still in use today or until recently (Smith 1996).

2.2

Early creole communities as complex multilingual settings

It is clear that creoles emerged in multilingual settings. These were multilingual along a number of different dimensions. First of all, different native languages were spoken by the slaves. It is truly exceptional to find a slave plantation where all or a large majority of the imported slaves spoke a single language. Such a case may be Berbice in Guyana, and this is reflected in the exceptional features of the resulting creole (Smith, Robertson, & Williamson 1987; Kouwenberg 1992). More usual are cases where it appears that a number of different African languages were spoken alongside each other by the initial slave population. Moreover, many of the slaves brought as captives may have been bilingual or multilingual. Also, in several plantation communities, there may have been different contributing pidgins and creoles. Thus in the Danish Virgin Islands (now the US Virgin Islands) it is possible that a Dutch pidgin was introduced when Dutch slaveowners left St. Eustatius around 1667 and took their slaves with them to St. Thomas, the principal island of the Danish Virgin Islands (Goodman 1985). Slaves who were later brought from Curaçao and probably spoke Papiamentu also contributed to the emerging creole (Hesseling 1933). Thus we find items such as kabaj ‘horse’, parie ‘give birth’, and ka ‘perfective’ of IberoRomance origin. Finally, different European superstrate or lexifier languages were involved in many island and mainland colonies. The European powers were in competition with each other, and some colonies frequently changed hands. Furthermore, European settlers in a colony often had different nationalities. To take the case of St. Thomas again, the nationalities of the planters in 1688 are as in table 12.1. In 1688 there were 422 slaves in St. Thomas, and 317 whites. Altogether, the picture that emerges is that the plantation communities in which the creoles were formed were complex multilingual settings.

2.3

Generational effects, loyalty, and shift

Another question is: How long did multilingualism last in the creole communities? The standard perspective on language shift among immigrant groups is the three-generation model: a first generation fully competent in the native language and with various degrees of competence in the new language; a second generation competent in the new language and with various degrees of competence in the original community language; a third generation with only

290

Pieter Muysken

Table 12.1 The white population (in number of households) of St. Thomas in 1688 (from Arends & Muysken 1992, p. 51, based on Sabino 1990) Nationality

Number of households

Dutch English Danish French German Swedish Holstein Portuguese Total

66 32 20 8 3 3 1 1 134

limited passive competence in the original community language (e.g., Fishman 1965). This model is inadequate for our purposes due to a number of considerations. First of all, it conflates two dimensions which should remain separate: various degrees of competence in the second language (L2), and degrees of loss of the first language (L1). It ignores the possibility of prolonged bilingualism, hence of cross-generational competence in both languages. It also ignores large intergroup differences. The data based on Australian census data in table 12.2 illustrate this. It turns out that immigrant ethnic groups differ considerably in the extent of shift to English as a home language, hence in the degree to which the original language is maintained across the generations. This may further mean that immigrant groups also differ in the degree to which the dominant language (here, English) is acquired, but systematic information which pertains to competence is not given. These differences Table 12.2 Home language use of a number of immigrant communities in Australia (Clyne & Kipp 1997, p. 459) Country of origin of first generation in 1996 census

Home language use: percentage using English

Netherlands Germany Spain Italy Hong Kong Greece Taiwan

62% 48% 22% 15% 9% 6% 3%

Creole Studies and Multilingualism 291 suggest that other factors may play a role. For creole settings different rates of shift have been commented upon by Roberts (2000) (see discussion in Veenstra, this volume). Moreover, in creole settings, rather than simply shifting to an existing language, a new language was created, a language which could serve as a symbol for a new social identity. Thus the metaphor of “shift” is not entirely appropriate. Additionally, there was a long period during which there were no generations to speak of (cf. Arends 1995, pp. 17–18; this volume). To take an admittedly extreme case, life expectancy upon arrival in Suriname during the eighteenth century averaged between five and ten years, even though a slave who survived the first year in the colony by the middle of the eighteenth century had a life expectancy of thirty years (Arends 2001, p. 296). There was an exceptionally low birth rate and high infant mortality. Finally, African languages were maintained as secret in-group languages for many years, partly in association with religious cults. Altogether, the three-generation model seems ill-conceived for the Caribbean slave plantations. This may not come as a surprise but merits bearing in mind.

3 Evidence from Language Contact Studies 3.1

Relexification, creoles, and mixed languages

The issue of maintenance and shift becomes particularly relevant when we take mixed (“intertwined”) languages and relexification into account in possible models for creole genesis. Implicitly or explicitly these have been part of creolist thinking since the early developments of the field, e.g., in the work of Adam (1883) and subsequently Sylvain (1936), where creoles are portrayed as a particular kind of mixed language: a European lexicon grafted onto an African semantic and syntactic base. These mixed creoles were assumed to have arisen through relexification. However, definitions of relexification were less rapidly forthcoming. In the 1960s and early 1970s, when monogenesis of pidgins and creoles on the basis of a single West African Portuguese Pidgin was discussed as a serious option, relexification referred to massive lexical replacement (Whinnom 1956). It was thought that the West African Portuguese Pidgin would have had a predominantly Portuguese-derived lexicon, while its various creole successors inherited its structure but replaced its vocabulary with English-, French-, or Dutch-derived items, under the influence of the local dominant European language. A typical example of this line of thinking was Voorhoeve (1973), who attributed the higher proportion of Portuguesederived words in Saramaccan as compared to Sranan to the fact that the relexification from Portuguese to English lexicon was interrupted when the Saramaccan maroons fled their plantations. Lexical semantic considerations played a minor role. (See Smith 1987, for a different account.)

292

Pieter Muysken

The perspective on relexification changed with the discovery of Media Lengua (Muysken 1981), a form of Ecuadorian Quechua relexified with Spanish forms at some time in the course of the twentieth century. A typical example follows (adapted from Muysken 1996a, p. 366): (1) a.

uwixa-buk yirba nuwabi-shka sheep-for grass there.is.not-SD ‘There turns out to be no grass for the sheep.’ b. llama buk k’iwa illa shka c. No hay hierba para las ovejas

(Media Lengua)

(Quechua) (Spanish)

Here, the italicized forms are from Spanish, and the bold forms from Quechua. However, the lexical semantics of the Spanish roots in Media Lengua (ML) is also like Quechua. Thus the form nuwabi-shka is modeled on Quechua illa-shka ‘it turns out there is no . . .’, through the combination of Spanish no and the root habe- ‘have, existential’. The notion relexification thus involves systematic replacement of the phonetic shapes of lexical items. The model of the lexicon invoked was that of Jackendoff (1975, p. 641), where lexical items were seen as bundles of relatively independent features: (2) /phon a/ +F STEM+x stratal a +___X SELECT Y MEANING Z

phonological representation syntactic categorial feature (e.g., [+V]) morphological composition possible stratal feature (e.g., “learned”) subcategorization feature (e.g., “transitive”) selectional feature (e.g., “human agent”) meaning

Since these features are seen as independent, there are ample possibilities for dissociation. Elaborating on the original proposal for relexification in Muysken (1981), Lefebvre (1998, p. 16) claims that relexification was the central process operant in the genesis of creole languages such as Haitian. She provides the schematic representation in figure 12.1 for relexification. The primary innovation in the Lefebvre definition is that the new lexical entry can be a null form. A second claim in Lefebvre (1998) is that relexification may involve a change in the word order of the immediate environment of the relexified item. The definition given in Mous (2001) for paralexification elaborates on Lefebvre’s by suggesting that two phonological representations may become available for a single original lexical entry, one from each language in contact.

3.2

Relexification and L2 learning in creole genesis

Models for creole genesis frequently involved the idea that relexification was a special kind of L2 learning in unfavorable circumstances. Here I will explore

Creole Studies and Multilingualism 293 Original lexical entry Lexifier language [phonology]i [phonetic string]j used in [semantic feature]k specific semantic and [syntactic feature]n pragmatic contexts \ / New lexical entry [phonology]j or [Ø] [semantic feature]k [syntactic feature]n Figure 12.1 The relexification process (Lefebvre 1998, p. 16)

this possibility somewhat further. In earlier work (e.g., Muysken 1980) I systematically contrasted relexification and L2 learning as alternative contact strategies. The reason for the strict bifurcation between the two strategies was that in my fieldwork around Salcedo in the central Andes of Ecuador I had noted that the type of Spanish interlanguage spoken by incipient QuechuaSpanish bilinguals, migrant cargadores (load-bearers) in the urban center, differed considerably from the Media Lengua of the originally Quechua-speaking communities near the town. The differences are summarized in table 12.3. Presented in this way, the two contact strategies have completely different results. Relexification and L2 learning thus seem to constitute alternative routes to pidgin/creole genesis. The question now is whether the Media Lengua/ interlanguage contrast in this extreme form should be generalized to other situations as well, and particularly whether creoles can plausibly be argued to resemble Media Lengua in having resulted from relexification and not from more straightforward L2 learning, as has been assumed by many other researchers. The answer is certainly not an unqualified yes, for a number of reasons. Table 12.3 Schematic contrast between Media Lengua and Spanish interlanguage in central Ecuador

Structure Degree of stabilization Source

Function

Media Lengua (relexification)

Interlanguage (L2 learning)

complex rigid

highly simplified highly variable

Quechua morphosyntax and phonology with slight Spanish influence in-group language

Spanish morphosyntax and phonology with Quechua influence interethnic communication

294

Pieter Muysken

First of all, other contact varieties in Ecuador must have resulted from a mixture of relexification and L2 learning, notably Catalangu. Examples from this variety are given in (3a)–(4a), with (3b)–(4b) representing the local Quechua equivalents, and (3c)–(4c) the Spanish equivalents (from Muysken 1996a, pp. 413–14; italicized forms in the Catalunga examples from Spanish, bold forms from Quechua, other forms cannot be directly attributed to either Quechua or Spanish): (3) a.

a-kin-ta-pi buska-ri-ndu? acc-who-acc-emph seek-inc-ger ‘Who are you looking for?’ b. pi-ta-tak mashka-ri-ngi? who-acc-emph seek-inc-2s ‘Who are you looking for?’ c. a quién estás busca-ndo acc who cop-2s seek-ger ‘Who are you looking for?’

(4) a.

ellos-kuna Sigsihuaicu-manta es 3p.m-pl Sigsihuaicu-abl cop ‘They are from Sigsihuaicu.’ b. pay-kuna Sigsihuaicu-manta-mi 3-pl Sigsihuaicu-abl-aff ‘They are from Sigsihuaicu.’ c. ellos son de Sigsihuaicu 3p.m cop-pl of Sigsihuaicu ‘They are from Sigsihuaicu.’

(Catalangu)

(Quechua)

(Spanish)

(Catalangu)

(Quechua)

(Spanish)

In the Catalangu examples (3a)–(4a) we find Quechua suffixes such as -ta ‘accusative’, -ri ‘inchoative’, -kuna ‘plural’, and -manta ‘plural’. In addition, however, there are Spanish grammatical elements such as -ndu ‘gerund’ and es ‘copula’. (Notice, incidentally, that the form used is singular, while the subject is plural; this could be because the copula is interpreted here as a reflex of the invariant Quechua affirmative validator -mi.) It is hard to say anything about the syntax; it looks mostly Quechua, but (3a) could also be Spanish interlanguage. If L2 learning and relexification were such radically different strategies, the existence of varieties like Catalangu, which combines the results of both, would be hard to explain. Another issue is the fact that Media Lengua owes its particular structure (a radical disjunction between the roots, almost exclusively from Spanish, and the affixes and enclitics, almost exclusively from Quechua) in large part to the typological features of Quechua, with its highly agglutinative morphology, and Spanish, with its highly frequent vowel-final roots – compatible with those of Quechua, which are also overwhelmingly vowel-final. Quechua is rather exceptional in allowing free borrowing of Spanish verb stems, something we

Creole Studies and Multilingualism 295 only find either among highly isolating languages such as Bazar Malay or Chinese, or among radically agglutinative languages such as Quechua; even a textbook agglutinative language such as Turkish does not allow direct borrowing of verb roots, and most other language types have alternative verb incorporation strategies (Muysken 2000). Thus the split between the types of morpheme categories in Media Lengua may well be triggered by specific morphological properties of the pair Quechua/Spanish. A similar account is needed for the Quechua/Puquina mixed language Callahuaya (Muysken 1996b). Not surprisingly, then, other cases of relexification show different patterns, i.e., there are different types of mixed languages. One example, Hottentot Dutch, is like Media Lengua (Den Besten 1987, p. 23; italics for Dutch elements, bold for Nama elements): (5) Heeltemaal-se natuur-a-xu bedorven-he totally-ADV nature-CASE-POST rotten-PASS Hoaraga-se =ûb-a-xu gau-he Van nature helemaal bedorven. ‘totally rotten in nature’

(Hottentot Dutch) (Nama/Khoikhoin) (Dutch)

However, the mixed language Michif, also claimed to have resulted from relexification, shows a noun phrase/verb phrase division (Bakker 1997, p. 45; italics for French elements, bold for Cree elements): (6) kî-nipi-yiwa son frère aspin kâ-la-petite-fille-iwi-t PAST-die-OBV.SUB 3S.POSS.M brother since COMP-the-little-girl-be-3S ‘Her brother died when she was a young girl.’ Yet a third possible type of relexification is exhibited by languages such as Petjo (Van Rheeden 1994, p. 226), a mixture of Malay (bold) and Dutch (italics): (7) kleren njang di-wassen door di frou cloth-PL REL PASS-wash by the woman ‘the clothes that are washed by that woman’ Here both languages contribute functional elements, in different proportions for each category. In table 12.4, an overview of this is given for one Petjo corpus. All these differences imply that relexification is a heterogeneous concept and that the Media Lengua case is not necessarily illustrative of relexification in general. Moreover, none of the Caribbean creoles looks like one of these intertwined languages, which we might expect if indeed relexification was at their origin: retention of functional categories from a substrate language is very rare (cf. Arends, Kouwenberg, and Smith 1995), unless we also include the retention of null functional categories, as is assumed by Lefebvre (1998). However, the theoretical status of null functional categories remains controversial.

296

Pieter Muysken

Table 12.4 Proportions of source language elements in Petjo (based on Van Rheeden’s 1994 analysis of the writings in Petjo of Tjalie Robinson)

exclamatives, interjections relative markers forms of address demonstratives adjectives possessives nouns question words verbs personal pronouns conjunctions numerals prepositions articles

Malay

Dutch

99 95 95 53 50 40 37 23 10 5 5 5 1 –

1 5 5 47 50 60 63 77 90 95 95 95 99 100

A further problem is that at the time the original disjunction between relexification and L2 learning was proposed, L2 learning was assumed not to be characterized by transfer to any great extent. This has changed in recent years, with the work of researchers such as Ineke Van de Craats (Van de Craats, Corver, & Van Hout 2000, 2002) and Bonnie Schwartz (Schwartz & Sprouse 1996). With various caveats and nuances, these researchers claim that the grammatical skeletons erected by the projections from the functional categories of speakers’ native languages, e.g., in the DP (nominal) and CP (clausal) systems, constitute the initial hypotheses that L2 learners make about the new target language. Thus, even if the functional categories in the resulting new language are not filled with morphemes from the original source language, they may be defined by the structural features of that source language, and only receive a superficial phonetic shape from the target. In short, they are relexified. Furthermore, models of L2 vocabulary acquisition present a complex picture. When an L2 lexical form is acquired, part of its meaning is learned first, and only later on are other dimensions of meaning filled in. As Kroll and Tokowicz phrase it (2001, p. 49): “During early stages of acquisition, words in the second language, L2, may rely on their counterparts in the first language, L1, to mediate access to meaning.” Thus, relexification in its strict sense can be modeled as very initial L2 vocabulary learning without concomitant L2 syntactic learning. Treating relexification as incomplete L2 vocabulary learning has the advantage that it can be viewed as a differentiated process, in which the saliency and frequency of the vocabulary items in the L2 input can help

Creole Studies and Multilingualism 297 determine to what extent they are relexified. Malt and Sloman (2003) show that this even holds for words referring to fairly concrete concepts like ‘cup’. A relexification account, in its strict form, commits one to the transfer of semantic organization features, lexical subcategorization, and selection features of individual items. Thus, any postulated substrate feature should be tied to specific lexical elements rather than to structural properties of classes of items or properties not closely linked to lexical items. For example, the Sranan serial verb poi ‘subsequent events turn out badly’ (< English spoil) in the relexification logic needs to be derived from a West African source – which has not been identified so far. An explanation where a general “serialization” construction or parameter has been transferred from a West African source would not be sufficient in a strict relexification account. Related to this, the issue of how to model the differences between languages from a theoretical perspective also impinges directly on the relexification debate. The lexical learning hypothesis was proposed as one way to account for crosslinguistic variation. If all cross-linguistic variation is indeed fundamentally lexical (in an abstract sense) in nature, relexification can lead to structural substrate influence. If on the other hand, as in Construction Grammar, grammatical constructions have a place separate from the lexicon in accounting for differences between languages, looking for relexification to account for structural substrate effects may not be the right approach. Obfuscating the issue again is the distinction of content words versus function words. In Muysken (1988) the claim was made that “real” or “pure” relexification, without accompanying semantic change, can only involve content words. Relexification of function words would automatically involve the target language, since function words necessarily depend, for their meaning definition, on L2-internal paradigmatic and syntagmatic relationships. If L2 shapes of function words were imported, it would be hard to escape the meaning of these items in the L2; otherwise there would be no way of accessing their meaning, needed for inter-lingual identification, at all. Whatever the logic of this argument, so far it has not been tested empirically in a rigorous way (cf. Muysken & Smith, in preparation), and indeed the available evidence may point exactly in the opposite direction. Speaking more generally, the notion of relexification is linked closely to our view of the lexicon. Richer models of the lexicon immediately lead to richer models of relexification, as the combinatory possibilities of the different chunks of information that can come from different language sources increase componentially with the number of chunks. In principle, the following kinds of information can be distinguished in the lexical entry: (8) Phonetic shape Morphological structure Conjugation class Linear orientation (left/right headedness) Subcategorization of arguments

298

Pieter Muysken Semantic argument structure Event structure Category Meaning

Although it is clear that these different chunks of information are linked, different theorists differ as to the nature and strength of these links. Roughly speaking, a three-way division is often assumed: (9) a.

Phonetic shape Morphological structure Conjugation class ====================================================== b. Linear orientation (left/right headedness) Syntactic subcategorization of arguments (object NP, to NP, etc.) ====================================================== c. Semantic argument structure (Agent, Experiencer, Theme, etc.) Event structure Grammatical category Meaning

Potentially, either only the outer form (phonetic and morphological shape) of an item is affected (9a), or also its syntactic behaviour (9b). However, this is still a matter of debate. This is an area where creole studies and multilingualism research have much to contribute to our view of the lexicon. Altogether, there is good reason to assume that relexification, viewed as a special type of L2 learning, was an important process in the genesis of many creole languages. However, L2 learning involves several other strategies in addition to transfer of L1 lexical semantic patterns. These other strategies can have been equally present in creole genesis, and may, in specific circumstances, have been more prominent than relexification.

3.3

Code-switching and code-mixing

At first sight there is very little relation between creole studies and the study of code-switching and code-mixing. I use the term “code-switching” here in the strict sense that clear switching between two distinct languages is meant, and “code-mixing” as a more inclusive term referring to utterances in which elements from several languages are combined. What is involved in codemixing is generally maintenance of the grammars of at least one of the languages concerned (the “matrix language,” cf. Myers-Scotton 1993) and the clear lexical presence of at least two languages. However, it is possible to imagine that frequent code-mixing was prevalent in the early plantation contact setting leading to creole genesis. When the languages involved have relatively sparse morphology and rather similar surface constituent orders, quite complex and

Creole Studies and Multilingualism 299 intimate code-mixing can be encountered. Examples from Sranan (bold)–Dutch (italics) code-mixing and of Moluccan Malay (bold)–Dutch (italics) codemixing are given in (10) and (11), respectively: (10) wan heri gedeelte de ondro beheer fu gewapende machten COP under control of armed forces one whole part ‘An entire part is under the control of the armed forces.’ (Bolle 1994, p. 75) (11) aku nog steeds vinden akan raar kata koe bellen aku twee keer 1S still find it strange that 2S call 1S two time zonder dapat gehoor without get hearing ‘I still find it strange that you called me twice without finding anyone home.’ (Huwaë 1992, p. 49) At first sight, these examples look quite a bit like those given above of mixed languages such as Media Lengua. They differ from the mixed language examples in that the speakers who produced (10) and (11) are also capable of producing unmixed sentences in either of the languages involved and, given the right interlocutors outside of their peer network, will consistently do so. Thus (10) and (11) are instantaneous productions, whereas mixed languages are frozen language systems in their own right. What the cases in (10) and (11) illustrate is that the functional elements are retained from the matrix language, even when the lexical elements are from the embedded language: the so-called system morpheme effect (Myers-Scotton 1993). In Myers-Scotton (2002) it is hypothesized that this asymmetry could also be responsible for the substrate-based structural features of creoles. An intriguing source of information is the rigorous quantitative work on Wolof-French and Fongbe-French bilingual discourse by Poplack and Meechan (1995) on nominal structures and determiners, and by Meechan and Poplack (1995) on adjectives. In both cases the African languages provide a frame into which French elements are inserted. Following earlier work by Poplack and associates, Meechan and Poplack make a sharp distinction between codeswitching and borrowing. (12) illustrates the latter. In Fongbe nominal structures many French nouns appear, and they are not accompanied by articles, unlike in French (Poplack & Meechan 1995, p. 204; Fongbe bold, French italics): (12) et puis science xl2 m5 gbè4é 43 tonnerre hù m7 4òkpó and then science shows us never that thunder kills person one ‘And science has never shown us that thunder killed one person.’ However, they can occur with Fongbe determiners and modifiers, and do so roughly at the same rates as native Fongbe nouns. Recall that Fongbe and French are two important languages in the genesis of Haitian (Singler 1996).

300

Pieter Muysken

Here, code-mixing patterns which result from borrowing may tell us something relevant for creole genesis. Poplack and Meechan call “(nonce) borrowing” what others have termed “insertion under congruence” or “equivalence.” If nonce borrowing – as in this account of French nouns in Fongbe – lies at the basis of the genesis of Haitian, we predict a similar distribution of determiners and modifiers in Haitian as in Fongbe. This still needs to be established quantitatively. For adjectives, a different analysis is adopted, in terms of code-switching. While Fongbe has descriptive predicates (semantically equivalent to French predicative adjectives) that are used without a copula, French adjectives appear in Fongbe sentences with a semi-auxiliary Dò (Meechan & Poplack 1995, p. 186): (13) cigar, alcool, action y6t3n 4ò passagère cigar, alcohol, action POSS COP passing ‘Cigars, alcohol, their action is passing.’ Their analysis is that “it seems inescapable that the Fongbe semiauxiliary Dò is being specialized as a device for handling French-origin adjectives.” Thus French adjectives are treated very differently from their notional equivalents in Fongbe. This can be compared with Haitian, where French-origin predicate adjectives do not take a copula and behave like stative verbs, as illustrated for pwé and kôtâ (Hall 1953, p. 51, spelling as in original): (14) a.

li pwé rivé 3s ready arrive ‘He was ready to arrive.’ b. m-kôtâ wè ou 1s-glad see 2s ‘I am glad to see you.’

There are two ways of interpreting this contrast between Haitian and Fongbe code-mixing. One is that whatever happens in the type of code-mixing exemplified here has nothing to do with the process of relexification or intertwining supposedly involved in the genesis of Haitian. The other is that Haitian did not emerge via a route of relexification at all, but rather that there was transfer of a grammatical pattern that lead to (14). Even if it is unsure yet what the exact relation is between creole genesis and code-mixing – if there is any at all – it is clear that a close comparison of contemporary Fongbe/French language contact has the potential to shed interesting light on the patterns in Haitian. Similar research is imaginable and possible for code-mixing involving other West African languages that played a role in the genesis of Caribbean creoles. Amuzu (2004) provides extensive data on Ewe–English code-switching, which could also be considered from the perspective of the potential light thrown on creole genesis.

Creole Studies and Multilingualism 301

3.4 3.4.1

Borrowing Lexical borrowing

A well-established domain of contact research is the study of lexical borrowing. Just like other types of language contact, lexical borrowing tends to be strongly asymmetrical. A socially subordinate language borrows from a socially dominant language, whereas the reverse is much less frequent. Thus we find a great many borrowings from Spanish in Bolivian Quechua (Van Hout & Muysken 1994), but few borrowings from Quechua in the local Spanish, the latter typically found in specific semantic domains: local fauna and flora, local food, local textiles and clothing styles, rituals, and exclamations. Similar patterns recur in many other parts of the world. In creoles, there is evidence of considerable borrowing in the historical development of individual creoles; to name but one example, Papiamentu has borrowed extensively from Latin American Spanish, Dutch, and English since it emerged in the course of the seventeenth century. However, we can also model the genesis of the creole lexicon in terms of borrowing, in casu borrowing into an empty lexicon. The distribution of the European lexicon and the African lexicon in most creoles is reminiscent of the asymmetries we find in typical borrowing situations: African lexicon again pertains to lexical domains such as food, rituals, and exclamations, European lexicon to most other domains (Huttar 1985).

3.4.2

Stratal morphology

Another aspect of the borrowing process which ties in with the observations just presented, and which may well be of great significance for creole studies, is the existence of stratal phenomena in the morphological behavior of borrowed words (e.g., Aronoff 1976). It has often been noted that a latinate affix such as -tion in English can only be productively attached to latinate roots; hence reduction but not *breaktion. In the lexicon of various European languages certain classes of roots apparently remain tagged as [romance] or [latinate]. As far as known, this restriction does not hold in creoles, and particularly not in creoles with multiple lexical sources. Dijkhoff (1993) describes the situation in Papiamentu as follows: the Papiamentu nominalizer -shon (compare Spanish -ción) is limited to Spanish verbs, but not really productive, occurring mostly if not exclusively in words directly derived from Spanish or Portuguese. In contrast, the agentive suffix -dó (compare Spanish -dor) and the action nominal -mentu are not limited to Spanish verb roots. Thus we have with agentive -dó: (15) fèrf-dó kap-dó las-dó wèlder-dó

‘painter’ (cf. Dutch verv-en ‘to paint’) ‘cutter’ (cf. Dutch kap-pen ‘to cut’) ‘welder’ (cf. Dutch las-sen ‘to weld,’ also Papiamentu laser ‘welder’) ‘welder’ (Aruban variety of Papiamentu, cf. English welder, Papiamentu wèlder ‘to weld’)

302

Pieter Muysken

The productivity of -dó, then, is linked to the lack of stratal restrictions on its use. The absence of restrictions on the basis of lexical origin suggests that the creole lexicon is essentially open and flexible, and that we could conceptualize its formation as basically instantaneous, without the historical layers that led to the formation of lexical strata in the European languages. This is an idealization, of course. In fact, different languages contributed words at different times. Thus Ibero-Romance words in Papiamentu are much older than Dutch or certainly English words.

3.4.3

Deep borrowing and functional categories

A final remark concerning the conceptualization of the formation of the creole lexicon as massive lexical borrowing is that the relative scarcity of function words directly taken from the European lexifier languages could be viewed in terms of the relative difficulty of borrowing function words. Even though function words can be borrowed, this generally happens much less frequently and much later in the historical process of lexical influence than is the case with the borrowing of, for instance, nouns. In the following fragments of Bolivian Quechua a fairly radical case of function word borrowing is illustrated. In (16a) we have a diminutive -situ and a plural -s from Spanish, and in (16b) the emphatic negation marker ni (from Urioste 1964, p. 3; Spanish elements in italics): (16) a.

atoj-situ-s-kuna-qa fox-dim-pl-pl-top ‘the little foxes’ b. Chay-manta-qa niña ni ima mikhu-na ka-jti-n, ni that-abl-top girl neg what eat-nom be-sd-3 neg ropa ni ka-jte-n-qa, sastre-mán tukuy ima-ta clothes neg be-sd-3-top tailor-abl all what-acc sua-rqa-mu-sqa. steal-int-cis-sd ‘Then when there was nothing to eat, and neither were there any clothes, the girl went to steal everything from the tailor.’

In Bolivian Quechua a wide array of functional elements can be borrowed from Spanish (Van Hout & Muysken 1994), but even here – perhaps surprisingly – certain categories, such as pronouns, are almost never borrowed.

3.5

Creoles and Sprachbund regions or linguistic areas

There has been considerable interest in recent years in the phenomenon of Sprachbund regions or linguistic areas. Recent examples are Aikhenvald and Dixon (2001) and Thomason (2001); earlier references include Campbell (1997a,

Creole Studies and Multilingualism 303 Table 12.5 Schematic contrast between creole settings and linguistic areas

Lexicon Morphosyntax

Creole and lexifier

Linguistic area

shared not shared

not shared shared

1997b, 1998). Thomason defines a linguistic area as “a geographical region containing a group of three or more languages that share some structural features as a result of contact rather than as a result of accident or inheritance from a common ancestor” (2001, p. 99). To some extent linguistic areas are the mirror image of creole settings. In the case of creoles, one could say that the creole language does not have the morphosyntactic features of the lexifier language that the inherited lexicon would lead one to expect. In linguistic areas, on the other hand, languages do not have the lexical relatedness that the shared morphosyntactic features would lead one to expect. Schematically this can be presented as in table 12.5. On a more concrete level, however, it is clear that the creole-speaking Caribbean – and probably the same holds for parts of the pidgin/creole-speaking Pacific – is a linguistic area by any definition one may wish to employ or by any scenario giving rise to a linguistic area one may imagine. There are a number of contributing factors: (a)

There are common African substrates in creole languages with different European lexifier languages. (b) At least some Caribbean creoles (particularly the English-lexifier creoles) have a common origin in an antecedent pidgin or creole. (c) There has been extensive movement of slaves from one plantation colony to another. (d) There has been extensive subsequent contact in several cases, as evidenced by the borrowing of vocabulary between some creoles. (e) In some cases there has been language shift from one creole to another. (f) There are extensive common superstrate influences. All these factors taken together have contributed to the morphosyntactic similarities between the Caribbean creoles, quite apart from shared circumstances of genesis.

4 Conclusion: Schuchardt’s Perspective on Language Contact I want to end this fragmentary overview with a plea to return to the roots of language contact research, and particularly to the accomplishments of Hugo

304

Pieter Muysken

Schuchardt. Not only did Schuchardt work on most of the topics mentioned here, including pidgin and creole genesis, linguistic borrowing, and codeswitching, he also proposed formalisms to schematize some of his findings (Schuchardt 1890, pp. 177–80; Muysken 1999). Unfortunately, these formalisms were mostly limited to lexical structures. Central to Schuchardt’s approach is a pairing between L (Laut ‘sound’) and B (Bedeutung ‘meaning’) of items, and matches and mismatches of L and B between different languages. In addition, Schuchardt uses the notions identity “=”, similarity “~”, and difference “)(”. These elements are combined in formulas of the following kind, where the sounds and meanings of the Grundsprache ‘base language’ (l, b) and of the einwirkende Sprache ‘influencing language’ (L, B) are compared: (17) l )( L b = B -------------------L The formula in (17) marks replacive borrowing, as when vínde (< English window) has replaced fénster ‘window’ in American Yiddish. The sounds of the base language and the influencing language are different, while the meanings are identical in this case. What wins out is the sound of the influencing language, vínde in the example given. A more complicated example is: (18) l = L b ~ B -------------------[ lB (LB) ] The formula is illustrated by Schuchardt with Portuguese lá, which resembles Malay emphatic lah in some of its uses, and is used frequently in Malay Portuguese, but, as the formula indicates, mostly with its original Malay meaning, albeit variable in its pronunciation. Even though the formulas employed by Schuchardt entail a fairly simple view of the lexicon, they have the satisfying property of being able to cover a wide variety of phenomena. I hope to have shown in this essay that it is indeed profitable to look at creole languages from the wider perspective of language contact research, in the same way that Schuchardt did at the very beginning of our field, as a systematic domain of scholarly enterprise.

REFERENCES Adam, Lucien (1883) Les idiomes afro-aryen et maléo-aryen: Essai d’hybridologie [The AfroAryan and Malayo-Aryan languages: A study in hybridology]. Paris: Maisonneuve.

Creole Studies and Multilingualism 305 Aikhenvald, Alexandra Y. and R. M. W. Dixon (eds.) (2001) Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics. Oxford: Oxford University Press. Amuzu, Evershed Kwasi (2004) Ewe–English Codeswitching: A Case of Composite rather than Classic Codeswitching. PhD dissertation, Australian National University, Canberra. Arends, Jacques (1995) The socio-historical background of creoles. In: Arends, Muysken, and Smith (eds.), pp. 15–24. Arends, Jacques (2001) Social stratification and network relations in the formation of Sranan. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 291–307. Arends, Jacques and Pieter Muysken (1992) Demografische modellen in de creolistiek [Demographic models in Creole studies]. Gramma / TTT, tijdschrift voor taalkunde 1, 41–55. Arends, Jacques, Silvia Kouwenberg, and Norval Smith (1995) Theories focusing on the non-European input. In: Arends, Muysken, and Smith (eds.), pp. 99–110. Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) (1995) Pidgins and Creoles: An Introduction (Creole Language Library 15). Amsterdam: John Benjamins. Aronoff, Mark (1976) Morphology in Generative Grammar. Cambridge, MA: MIT Press. Auer, Peter (1998a) Introduction: Bilingual conversation revisited. In: Auer (ed.), pp. 1–24. Auer, Peter (ed.) (1998b) Codeswitching in Conversation: Language, Interaction, and Identity. London: Routledge. Baker, Philip (1993) Assessing the African contribution to French-based creoles. In: Salikoko S. Mufwene (ed.) Africanisms in Afro-American Language Varieties. Athens, GA: The University of Georgia Press, pp. 123–55. Bakker, Peter (1997) A Language of Our Own: The Genesis of Michif, the Mixed Cree-French Language of the Canadian Métis. Oxford: Oxford University Press. Bolle, Jette (1994) Sranan Tongo–Nederlands: Code-wisseling en ontlening [Sranan Tongo–Dutch: Code-switching and borrowing]. MA thesis, University of Amsterdam. Campbell, Lyle (1997a) American Indian Languages: The Historical Linguistics of Native America. Oxford: Oxford University Press. Campbell, Lyle (1997b) Typological and areal issues in reconstuction. In: Jacek Fisiak (ed.) Linguistic Reconstruction and Typology. Berlin: Mouton de Gruyter, pp. 45–72. Campbell, Lyle (1998) Historical Linguistics: An Introduction. Edinburgh: Edinburgh University Press. Clyne, Michael and Sandra Kipp (1997) Trends and changes in home language use and shift in Australia, 1986–1996. Journal of Multilingual and Multicultural Development 18, 451–73. Den Besten, Hans (1987) Die niederländischen Pidgins der alten Kapkolonie [The Dutch pidgins of the Old Cape Colony]. In: Norbert Boretzky, Werner Enninger, and Thomas Stolz (eds.) Beiträge zum 3. Essener Kolloquium über Sprachwandel und seine bestimmenden Faktoren. Bochum: Brockmeyer, pp. 9–40. Dijkhoff, Marta B. (1993) Papiamentu Word Formation: A Case Study of Complex Nouns and Their Relation to Phrases and Clauses. PhD dissertation, University of Amsterdam. Fishman, Joshua (1965) Who speaks what language to whom and when? Linguistics 2, 67–88. Gal, Susan (1989) Language and political economy. Annual Review of Anthropology 18, 345–67.

306

Pieter Muysken

Goodman, Morris (1985) The origin of Virgin Islands Creole Dutch. Amsterdam Creole Studies 7, 67–106. Hall, Robert A. (1953) Haitian Creole: Grammar, Texts, Vocabulary (The American Anthropologist. Memoir 74). Washington DC: American Anthropological Association. Hesseling, Dirk Christiaan (1933) Papiaments en Negerhollands [Papiamentu and Negerhollands]. Tijdschrift voor Nederlandsche Taal- en Letterkunde 52, 265–88. Huttar, George (1985) Sources of Ndjuka African vocabulary. New West Indian Guide 59, 45–71. Huwaë, Rosita (1992) Tweetaligheid in Wierden: het taalgebruik van jongeren uit een Molukse gemeenschap [Bilingualism in Wierden: The language of youths in a Moluccan community]. MA thesis, University of Amsterdam. Jackendoff, Ray (1975) Morphological and semantic regularities in the lexicon. Language 51, 639–71. Kouwenberg, Silvia (1992) From OV to VO: Linguistic negotiation in the development of Berbice Dutch Creole. Lingua 88, 263–99. Kroll, Judith F. and Natascha Tokowicz (2001) The development of a conceptual representation for words in a second language. In: Janet L. Nicol (ed.) One Mind, Two Languages: Bilingual Language Processing. Oxford: Blackwell, pp. 49–71. Labov, William (1972) Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press. Le Page, Robert and Andrée Tabouret-Keller (1985) Acts of Identity. Cambridge: Cambridge University Press. Lefebvre, Claire (1998) Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Malt, Barbara and Steven Sloman (2003) Linguistic diversity and object naming by nonnative speakers of English. Bilingualism: Language and Cognition 6, 47–67. Meechan, Marjory and Shana Poplack (1995) Orphan categories in bilingual discourse: Adjectivization strategies in Wolof-French and Fongbe-French. Language Variation and Change 7, 169–94. Milroy, James and Lesley Milroy (1985) Linguistic change, social network, and speaker innovation. Journal of Linguistics 21, 339–84. Mous, Maarten (2001) Paralexification in language intertwining. In: Norval Smith and Tonjes Veenstra (eds.) Language Theory and Creole Genesis. Amsterdam: John Benjamins, pp. 113–23. Muysken, Pieter (1980) Alternative roads to creole genesis. Paper presented at the Workshop on Creole Syntax, University of Amsterdam. Muysken, Pieter (1981) Halfway between Quechua and Spanish: The case for relexification. In: Albert Valdman and Arnold Highfield (eds.) Historicity and Variation in Creole Studies. Ann Arbor: Karoma, pp. 52–78. Muysken, Pieter (1988) Lexical restructuring and creole genesis. In: Norbert Boretzky, Werner Enninger, and Thomas Stolz (eds.) Beiträge zum 4. Essener Kolloquium über “Sprach-kontakt, Sprachwandel, Sprachwechsel, Sprachtod” vom 9.10.-10.10.1987 an der Universität Essen. Bochum: Brockmeyer, pp. 193–210. Muysken, Pieter (1996a) Media Lengua. In: Thomason (ed.), pp. 365–426. Muysken, Pieter (1996b) Callahuaya. In: Thomason (ed.), pp. 427–48. Muysken, Pieter (1999) Lexical contact phenomena: Schuchardt (and beyond?). In: Bernt Brendemoen, Elizabeth Lanza, and Else Ryen (eds.) Language Encounters Across Time and Space. Oslo: Novus, pp. 29–36. Muysken, Pieter (2000) Bilingual Speech: A Typology of Code-mixing. Cambridge: Cambridge University Press.

Creole Studies and Multilingualism 307 Muysken, Pieter and Norval Smith (in preparation) Introduction. In: Pieter Muysken and Norval Smith (eds.) The Trans-Atlantic Sprachbund: Language Contacts between Benin and Surinam. Myers-Scotton, Carol (1993) Duelling Languages: Grammatical Structure in Codeswitching, Oxford: Clarendon Press. Myers-Scotton, Carol (2002) Bilingual Encounters and Grammatical Outcomes. Oxford: Oxford University Press. Poplack, Shana and Marjorie Meechan (1995) Patterns of bilingual mixture: Nominal structure in Wolof-French and Fongbe-French bilingual discourse. In: Lesley Milroy and Pieter Muysken (eds.) One Speaker, Two Languages: Cross-Disciplinary Perspectives on Code-Mixing. Cambridge: Cambridge University Press, pp. 199–232. Roberts, Sarah (2000) Nativization and the genesis of Hawaiian Creole. In: McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 257–300. Sabino, Robin (1990) Towards a Phonology of Negerhollands: An Analysis of Phonological Variation. PhD dissertation, University of Pennsylvania. Schuchardt, Hugo (1890) Kreolische Studien IX: Über das Malaioportugiesische von Batavia und Tugu [Creole Studies IX: On the Malayo-Portuguese of Batavia and Tugu]. Sitzungsberichte der kaiserlichen Akademie der Wissenschaften zu Wien 122(9), 1–256. Schwartz, Bonnie and Richard A. Sprouse (1996) L2 cognitive states and the Full Transfer/Full Access model. Second Language Research 12, 40–72. Sebba, Mark (1997) Contact Languages. London: Macmillan. Singler, John Victor (1996) Theories of creole genesis, sociohistorical considerations, and the evaluation of evidence: The case of Haitian Creole and the Relexification Hypothesis. Journal of Pidgin and Creole Languages 11, 185–230. Smith, Norval (1987) The Genesis of the Creole Languages of Surinam. PhD dissertation, University of Amsterdam. Smith, Norval (1996) WE-focus in Saramaccan: Substrate feature or grammaticalization. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Relating to Grammaticalization in Contact Languages (Westminster Creolistics Series 2). London: University of Westminster Press, pp. 113–28. Smith, Norval, Ian E. Robertson, and Kay Williamson (1987) The Ijo element in Berbice Dutch. Language in Society 16, 49–90. Sylvain, Suzanne (1936) Le créole haitien: Morphologie et syntaxe [Haitian Creole: Morphology and Syntax]. Wetteren, Belgium: Imprimerie de Meester. Thomason, Sarah G. (ed.) (1996) Contact Languages: A Wider Perspective. Amsterdam: John Benjamins. Thomason, Sarah G. (2001) Language Contact: An Introduction. Edinburgh: Edinburgh University Press. Urioste, Jorge L. (1964) Transcripciones Quechuas [Quechua transcripts]. Cochabamba, Bolivia: Instituto de Cultura Indígena. Van de Craats, Ineke, Norbert Corver, and Roeland Van Hout (2000) Conservation of grammatical knowledge: On the acquisition of possessive noun phrases. Linguistics 38, 221–314. Van de Craats, Ineke, Norbert Corver, and Roeland Van Hout (2002) The acquisition of possessive HAVE-clauses by Turkish and Moroccan learners of Dutch. Bilingualism: Language and Cognition 5, 147–74. Van Hout, Roeland and Pieter Muysken (1994) Modelling lexical borrowability. Language Variation and Change 6, 39–62.

308

Pieter Muysken

Van Oldendorp, Christian G. A. ([1777] 1987) History of the Mission of the Evangelical Brethren on the Caribbean Islands of St. Thomas, St. John, and St. Croix. Edited by Johann Jakob Bossard, translated by Arnold R. Highfield and Vladimir Barac. Ann Arbor: Karoma. Van Rheeden, Hadewych (1994) Petjo: The mixed language of the Indos in Batavia. In: Peter Bakker and Maarten Mous (eds.) Mixed Languages: 15 Case Studies in Language Intertwining. Amsterdam: IFOTT, pp. 223–37. Voorhoeve, Jan (1973) Historical and linguistic evidence in favor of the relexification theory in the formation of creoles. Language in Society 2, 133–45. Whinnom, Keith (1956) Spanish Contact Vernaculars in the Philippine Islands. Hong Kong: Hong Kong University Press. Woolard, Kathryn A. (1998) Introduction: Language ideology as a field of enquiry. In Bambi B. Schieffelin, Kathryn A. Woolard, and Paul V. Kroskrity (eds.) Language Ideologies: Practice and Theory. Oxford: Oxford University Press, pp. 3–47.

A Demographic Perspective on Creole Formation 309

13

A Demographic Perspective on Creole Formation JACQUES ARENDS

1 Introduction Anyone uninitiated in the field may be a bit surprised to find a chapter on demography in a handbook of pidgins and creoles. However, once it is realized that the formation of a creole language is a result of the interaction between groups of speakers of different languages, it will be clear that both the numerical proportion of these groups and their internal composition are important factors in this process. Other things being equal, a language with relatively many speakers is more likely to influence the outcome of contact than a language with relatively few speakers. Since creole formation is a process of both first and second language acquisition, it is also important to know the proportion of children and adults in the creole-forming population. (Although it is customary among demographers to distinguish between infants (age 0–1), children (age 1–4), and others (age 5 and above), in this chapter the term “children” will be used to refer to infants, children, and adolescents, i.e., all persons below the age of – roughly – 18.) When precise data on age distribution are lacking – which is often the case for creole-forming societies – an indirect route must be used to obtain the necessary information. As far as the proportion of children is concerned, an important variable is the proportion of men to women, particularly those women of child-bearing age. This is so because of the obvious relationship that exists between sex ratio, on the one hand, and the number of children being born, on the other. Other factors, which we will argue below come into play as well, include the rates of child mortality (especially high among those below 4), still-births and abortions. Life expectancy, which was extremely low in many colonies, is a linguistically relevant factor too. In its turn, life expectancy is related to a number of other variables, such as morbidity (including lack of immunity to certain diseases), nutrition, and labor and living conditions. Other demographic variables that have linguistic relevance include the black-to-white ratio (or, more generally, the proportion of non-Europeans to Europeans, both locally

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

310

Jacques Arends

born and foreign-born) and the Creole-to-Bozal ratio (or the proportion of locally born versus foreign-born non-Europeans). A final demographic issue – the one that has attracted most attention – is the question of where the immigrants, especially the slaves, were coming from. This is so because in order to establish the make-up of the language contact situation that produced particular creoles, we need to know as precisely as possible which languages were involved in it. Finally, with respect to all of these issues we have to be aware of the fact that the demography of most creole-forming societies is highly anomalous. Low life expectancies, high infant mortality rates, low sex ratios, and exceptionally high immigration rates combined to produce societies whose demography was extremely unstable and remained in flux throughout much or all of the creole-forming period. Despite the diversity found in the populations involved in the formation of creoles, in most cases they fall into one of two categories: a population of Europeans on one hand, and a population of (forcibly recruited) immigrant labor on the other. This is especially true with regard to the group of creoles we will concentrate on here, namely the plantation creoles of the Atlantic area, where the second group consisted mainly of enslaved Africans. Adopting a practice that has become established among creolists, the languages spoken by the groups will be referred to as the superstrate (or lexifier) and substrate language(s), respectively. This does not mean, of course, that for every creole in the world the superstrate is always a European language, much less that substrate languages are always African (see Versteegh, this volume). Nor does it mean that these two categories exhaust the entire population involved in creole formation; other groups, such as Native Americans, may be involved as well, although it appears that these groups have played only a minor role in creolization. While later demographic developments, after a creole has been essentially formed, may have a bearing on its further development, we will concentrate here on the demographic characteristics of the formative period. We will take the first one hundred years after the onset of colonization as a rough approximation of the formative period, steering a middle ground between the most extreme instantaneous and gradualist scenarios (Bickerton 1981, and Arends 1989, respectively). In addition to such clearly demographic variables as birth and death rates, age distribution, sex ratio, and black-to-white ratio, we will also discuss some non-demographic aspects of social history, such as place of origin, place of birth (i.e., foreign-born versus locally born), and language(s) spoken. Due to limitations both in the literature and in the author’s expertise, the discussion will be focused on the English-lexicon creoles of the Caribbean. The remainder of this chapter is organized as follows: In section 2 we will discuss a number of demographic factors which are relevant for creole formation. These include the ethnolinguistic origins of the African and European inhabitants of the colonies, especially of the slaves (2.1); the black-to-white ratio, i.e., the numerical proportion of Africans to Europeans (2.2); and the Creoleto-Bozal ratio, i.e., the numerical proportion of locally born blacks (“Creoles”)

A Demographic Perspective on Creole Formation 311 to African-born blacks (“Bozals”) (2.3). Attention will also be given to these ratios’ development through time. Since exact figures are often lacking, other demographic variables will also be briefly discussed. These include the respective roles of children and adults (2.4), crop selection, in particular the demographic effects of the introduction of large-scale sugar cultivation in the New World (2.5), inter-colony migration (2.6), the role of the “founder population” (2.7), and the mulatto-to-black and free-to-enslaved ratios and their development through time within the population of color (2.8); finally, some nondemographic factors which have been adduced in the literature will be briefly discussed (2.9). In section 3 several “demographic pitfalls” are discussed along with some advice on how to avoid these. Section 4 contains a short guide to the literature on the demography of slavery; the chapter concludes with a brief Outlook.

2 Demographic Factors in Creole Formation With few exceptions, the role of demographic factors in creole formation was not recognized until quite recently. As far as the African side of the equation is concerned, this was partly due to the fact that the Atlantic slave trade remained a rarely studied topic until the publication of Curtin (1969). Hence, very little data were available for creolists to proceed from. Nevertheless, as early as 1960, Robert Le Page published his study of aspects of the demographic history of Jamaica relevant for the formation of Jamaican Creole. The bulk of this work (pp. 21–84) is devoted to the ethnolinguistic origins of the slaves, but other aspects are covered as well, such as the origins of the indentured servants and other Europeans. Another early work that paid some attention to demographic issues was Rens’s (1953) study of the social history of Sranan. While Rens has little to say about the slaves’ ethnolinguistic backgrounds, he presents a number of informed estimates of the size of the white and black population of Suriname between 1651 and 1684, based on archival and other historical documents (Rens 1953, pp. 75–85). Baker’s (1982a) meticulous study of the early population history of Mauritius provided a major impetus for demographic studies. Baker used demographic evidence to refute Chaudenson’s (1974) hypothesis that Mauritian Creole is a continuation of Réunion Creole. Baker’s example was followed by scholars such as Singler (1986, 1990) and Arends (1995a), who adduced demographic evidence to argue against Bickerton’s (1981, 1984) Bioprogram Hypothesis. Since then, however, demography has been shown to be relevant for other issues as well, such as the role of the “founder population” (Mufwene 1996).

2.1

The ethnolinguistic origins of Africans and Europeans in the Caribbean

The demographic issue that has drawn most attention on the part of creolists is the question of where in Africa the slaves came from. The relevance of this

312

Jacques Arends

question for creole formation is obvious. Unless one excludes the possible contribution of the substrate languages on an a priori basis, it is imperative to establish the ethnolinguistic origins of the people speaking these languages. The first creolist to recognize this was Le Page (1960), who, based on a careful scrutiny of the available literature as well as historical sources, made an initial attempt at determining the ethnolinguistic groups that were numerically most important during different periods in the formative stages of Jamaican Creole. One of the remarkable features of Le Page’s calculations is that he not only took into account immigration of slaves but emigration as well (pp. 80–2), something which, in spite of its obvious linguistic implications, is often neglected by creolists. (Although clearly euphemistic when referring to the slave trade – since migration was forced – the terms “immigration” and “emigration” (or out-migration) will be used in this chapter.) While some of Le Page’s findings have been shown to need revision (see below), it is important to realize that he was the first creolist to make a substantial attempt to link the ethnolinguistic profile of a population to the linguistic features of the creole spoken by that population. In doing so, he sometimes (e.g., p. 76) made use of concepts which did not become widely known among creolists until much later, such as the founder effect. A study similar to Le Page’s for Jamaica but based on more extensive and reliable data was done for Suriname by Arends (1995a). Drawing on Postma’s (1990) comprehensive, quantitative analysis of the Dutch slave trade, Arends was able to determine the ethnolinguistic origins of the Africans brought to Suriname between 1675 and 1803. (Unfortunately, due to lack of data, the pre-1675 period had to be left out of the account.) The main trends during the formative period of the Suriname creoles (roughly, 1650–1750) can be summarized as follows: the numbers of speakers of Gbe and of Kikongo were initially more or less equal (1675–1700), but the number of Kikongo speakers dropped significantly during the next 50 years while the number of Gbe speakers increased. During this same period (1700–50), the number of Akan speakers rose suddenly, surpassing even the Gbe segment. These demographic data are in line with a large body of linguistic findings (e.g., Daeleman 1972; Huttar 1985; Migge 2003) which show that Gbe, and to a lesser extent Akan and Kikongo, are the African language clusters that have left the strongest imprint on the creoles of Suriname. The importance of the mutual confirmation of the linguistic evidence on the one hand and the extra-linguistic (demographic) evidence on the other, as shown for the Suriname creoles, cannot be overestimated. Although, as yet, this kind of mutual support has not been demonstrated to the same degree for other creoles, the Suriname case alone provides decisive evidence against the idea that the linguistic heterogeneity among slaves was too strong to allow a significant influence from any African language or language cluster. This notion of profound linguistic heterogeneity among slave populations in Caribbean colonies was based on the fact that West and West Central Africa are among the linguistically most diverse regions of the world, coupled with

A Demographic Perspective on Creole Formation 313 the assumption that planters adhered to a linguistic divide-and-rule strategy ensuring that slaves from different language backgrounds were mingled on plantations in order to prevent them from planning rebellions or any other kind of insubordinate activity. It is becoming more and more clear, however, that, despite the frequent mention of the divide-and-rule strategy in the contemporary literature, it constituted a desideratum on the part of the planters rather than an actual practice. In fact, in some colonies the supply of slaves was so scarce that planters did not have any choice as to the ethnolinguistic background of the slaves they bought to work their plantations. Recent research on the Atlantic slave trade confirms that the linguistic diversity of West and West Central Africa was not equally reflected among the Africans who were taken to the New World. As shown by Arends (1995a), at least some of the slave recruitment areas (e.g., the Gold Coast and Slave Coast regions) were linguistically quite homogeneous, especially when compared to other parts of West Africa. The claim that slaves brought to the Caribbean formed relatively homogeneous groups receives further support from Eltis et al.’s (1999) database (see below), which leads historians to conclude that “the distribution of peoples from West Africa in the Americas was far from random” (Eltis & Richardson 1997a, p. 8). In other words, rather than drawing slaves from all over coastal West and West Central Africa, specific regions in the New World recruited their slaves from specific regions in Africa. Since this finding is of major importance for theories of creole genesis, the passage where it is articulated most clearly is quoted here in full: On the issue of transatlantic links, the picture of African coerced migrants arriving mainly in a mix of peoples – often on the same vessel – needs revising. Like the free migrant and indentured servant trades, systematic geographic patterns existed. Scholars should now turn to exploring what these mean both for Africa and for African influences in the shaping of the New World. (Eltis 2001, p. 41)

Although this observation in itself is not new (cf. Curtin 1969), its importance lies in the fact that it is based on the Eltis et al. (1999) data set covering some 27,000 transatlantic voyages (two-thirds of the estimated total) concerning all major slave-trading nations. This database makes it possible to achieve a more reliable picture of the ethnolinguistic origins of the slaves that were brought to particular colonies in specific periods of time. Some of the conclusions that can be drawn from this data set are worth mentioning here. It should be emphasized, however, that the Eltis et al. database is limited to transatlantic traffic, i.e., it does not contain any information about intra-Caribbean shipments, on which research is still in its infancy. With this proviso in mind, Eltis (2001, p. 46, table IV) suggests that, contrary to what was hypothesized by Le Page, Gold Coast slaves formed only a small minority (around 10 percent) of the immigrants arriving in Jamaica during the first 50 years after colonization by the English. Another important finding reported by Eltis (2001, p. 46, table V) is that between 1700 and 1750 (especially 1700–25)

314

Jacques Arends

slaves from the Bight of Benin formed the single largest group among immigrants to Haiti (unfortunately the data contained in Eltis et al. for the pre-1700 period are very limited). This finding provides independent support for a basic assumption of the Relexification Hypothesis (Lefebvre 1998), namely that speakers of Gbe (a major Bight of Benin language cluster) exerted an exceptionally strong influence in the formation of Haitian Creole. This is especially important because until now no colony-specific demographic evidence had been provided in support of this claim. The figures reported by Singler (1993b, 1996) at a time when no colony-specific figures were available refer only to the general French trade for this period. Although the proportion of Bight of Benin slaves brought to Haiti in the first half of the eighteenth century was not as high as that for the overall French trade (some 45 percent versus some 55 percent), it is clear that they formed the single largest ethnolinguistic group that came to Haiti in this period. Finally, the Eltis et al. data show that the great majority of Bantu-speaking slaves from West Central Africa ended up in Brazil (Eltis & Richardson 1997a, p. 7), a plantation society where, as far as is known, no full-fledged creole ever emerged. The relative absence of Bantu speakers from the Caribbean may explain why Bantu languages have left few traces in the Caribbean creoles, especially when compared to Kwa languages such as Akan and Gbe (but see also Singler 1993b). These examples may suffice to show the importance of the Eltis et al. data for reconstructing and explaining the formation of creole languages. As far as the other (non-African) population groups are concerned, much is still unknown. What we do know, however, is that the fact that a colony belonged to a particular European nation does not mean that that nation’s language was the only, or even the major, European language spoken in the colony. Apart from the fact that the European population of many colonies was composed of people of diverse European origin, speaking different languages (French, German, Portuguese, English, Dutch, Gaelic), it is important to realize that the European languages spoken by the settlers should not necessarily be identified with the “official” metropolitan varieties that were spoken at the time. Quite to the contrary, it is much more likely that many of the Europeans who came to the colonies, especially such people as indentured servants, convicts, ex-sailors and -soldiers, and their likes, would have been speakers of nonstandard social or regional dialects. One of the first to draw attention to this important fact was Chaudenson (2001 [1992], pp. 65– 8), who found that most colonists coming to French colonies in the seventeenth and eighteenth centuries came from rural areas where regional dialects were spoken, especially to the west of the line connecting Bordeaux and Paris. At the same time, it should be realized that the situation may have been different in non-French, especially English, colonies (see Smith, this volume). One group of European origin that is especially important with regard to creole formation is that of the indentured laborers or servants (called engagés

A Demographic Perspective on Creole Formation 315 in the French possessions), who came to the colonies, especially in the early stage of colonization, to work for a fixed period of time (usually three or five years) and who, during their indenture, were more or less treated on a par with the African slaves. With regard to the servants in the English Caribbean, Le Page (1960, pp. 12ff.) found that a large proportion of them, especially in the seventeenth and early eighteenth centuries, were Irish, and that many of those who were English came from the West of England and London areas. In the course of the eighteenth century, however, the preponderance of the Irish diminished, when they came to be replaced more and more by Scots (including Ulster Scots). There is some reason to think that some among both the Irish and the Scottish servants would have been Gaelic- rather than Englishspeaking. This is so in the case of the Irish because, being poor people, they did not belong to that section of society that spoke English, and in the case of the Scots because they included many people from the Highlands, who were Gaelic-speaking at the time (cf. Le Page 1960, p. 87). This means that for them, just as for the African slaves, interethnic communication on the plantations involved the use of English as a second, not a first, language. This is especially relevant in view of the fact that indentured servants, who were the group whose social position was closest to that of the African slaves, presumably played an important role in the formation of creoles. A group whose role in creole formation, in some cases at least, was particularly disproportionate to its size is the Sephardic Jews. When the Dutch colony of Pernambuco in northeast Brazil was re-conquered by the Portuguese (1645– 54), the Sephardic Jews who had settled there when fleeing the Inquisition were dispersed over the New World, including the Caribbean. Coming from Brazil, where large-scale sugar cultivation had been practiced for several decades, the Sephardic Jews were instrumental in spreading to the Caribbean the technology needed for the cultivation and production of sugar, triggering the switch to sugar monoculture that took place in the Caribbean in the 1650s and 1660s, beginning in Barbados and quickly spreading to other colonies, such as Jamaica and Suriname, which until then had been growing less labor-intensive crops, especially tobacco. The importance of the switch to sugar lies in the fact that its introduction was accompanied by a fundamental change in the type of agricultural unit needed for its cultivation and processing, namely the large-scale, agro-industrial plantation unit working large numbers of slaves in a factory-like mode of production (Chaudenson 2001 [1992]). Because of the serious demographic effects of this shift, such as increased death and decreased birth rates, it is seen by some creolists as a crucial change in the sociohistorical matrix of creole formation. Unfortunately, much is still unknown about the linguistic profile of the Sephardic Jews in the Caribbean, who were not only active in trade and finance but in some colonies also as plantation owners. It is this latter role which explains why their primary language, Portuguese, has left its traces in the lexicons of the Suriname creoles, especially Saramaccan and Matawai, and, to a lesser extent, Ndyuka.

316

2.2

Jacques Arends

The black-to-white ratio

Another demographic issue that is considered relevant for creole formation is the numerical proportion of the black and white segments of a colony’s population. This is based on the assumption that the degree of linguistic restructuring in a language contact situation is related to the degree of access second-language learners have to first-language speakers. Even though it is impossible to establish precisely the minimum degree of access needed for second language acquisition to be successful, the idea is that this minimum was not attained in the environments where creole languages emerged. The first creolist to propose a cut-off point beyond which creole formation rather than second language acquisition will occur was Derek Bickerton (1981), who stipulated that languages only count as creoles if they arose in an environment where fewer than 20 percent of the population were speakers of the dominant language. However, this criterion remained merely “theoretical” in that it was not substantiated for any of the creoles discussed in his work (e.g., Bickerton 1981, 1984). A more empirical approach was taken by Baker (1982b, pp. 852–6; see also Baker & Corne 1986, pp. 165–8), who identified three Events (Event 1, Event 2, and Event 3) as being of particular importance for creole genesis. Although Baker later abandoned the hypothesis associated with these Events, their heuristic importance for an empirical account of creole formation is still widely recognized. Thus, it is important to consider Baker’s Events here. Event 1 refers to the point in time when numerical parity is reached between the black and white segments of the population. It is hypothesized that in the period between the beginning of colonization and Event 1, each second-language learner has sufficient access to first-language speakers to be able to learn the superstrate language. After Event 1, when increasing numbers of slaves are imported while the number of whites does not grow in proportion, it becomes increasingly difficult for second-language learners to get access to native speakers from whom to learn the superstrate language. Needless to say, the issue of access to first-language speakers is not just a quantitative matter; the quality of the interaction between whites and blacks may also have differed widely. Not enough is known, however, to specify the linguistic impact that this may have had. Baker’s Event 2 refers to the point in time when the number of locally born blacks, or “Creoles,” reaches numerical parity with the total number of whites. Event 2 may be interpreted as the point in time at which the locally born black population has increased to such a degree that a creole, rather than a restructured variety of the lexifier language, starts to emerge. This does not mean, of course, that creolization will necessarily take place, only that a condition for it has been fulfilled. Whether creolization will actually occur depends on other demographic factors, such as the rate of post-Event 2 slave imports and the demographic developments of the white population, as well as on a number of non-demographic variables (see 2.9). Finally, Event 3 refers to the point in

A Demographic Perspective on Creole Formation 317 time when the regular immigration of substrate speakers comes to an end. In most cases, the substrate languages will cease to be spoken within one or two generations of Event 3, thereby effectively removing the possibility of any further substrate influence. With regard to the black-to-white ratio, Parkvall (2000a) has attempted to establish on empirical grounds whether or not there is a relationship between the degree of access in a language contact situation and the degree of linguistic restructuring, formalizing the latter in terms of a number of linguistic features that are present in creoles but absent from the respective lexifier languages. On the basis of a survey of 45 linguistic features in 20 creoles, Parkvall concludes that “there is indeed a correlation – whether causal or not – between the degree of restructuring and the demographic ratios in the formative period” (Parkvall 2000a, p. 197). Nevertheless, he observes that demographic disproportion is neither a necessary nor a sufficient condition for creole formation to take place (p. 199).

2.3

The Creole-to-Bozal ratio

Apart from Events 1–3, several creolists, notably Singler (1986, 1992), have recognized the importance of yet another demographic event, namely the point in time when the locally born African population (i.e., people fully or partially of African descent) reaches numerical parity with the foreign-born African population. Following Roberts (1999), we will refer to this as Event X (but note that Roberts does not agree with Singler about the significance of Event X, specifically with regard to Hawai[i). According to Singler (1992, p. 326), “[t]he faster the locally born population emerges and the larger it is, the greater the influence of the lexifier language [in determining the make-up of the creole].” In this view, locally born children are seen as conduits for the transmission of the lexifier to the substrate population, thereby decreasing the structural distance between a creole and its lexifier. Since it is convenient to have a name for the onset of colonization as well, it has been suggested to refer to that point in time as Event 0 (Arends 1995b). The Events and their meanings are listed in table 13.1. It should be emphasized that it is the relative timing of the Events that is assumed to have a bearing on creole formation, not the absolute timing of the Events themselves. For example, the longer the interval between Event 0 and Event 1, the smaller the structural distance between an emerging creole and its lexifier is expected to be. Although the precise consequences of the timing of all these different Events have not been established as yet, their tentative linguistic effects are summarized in table 13.2. The Events are historical occasions whose timing can, in principle, be established empirically. However, their importance is at least as much heuristic as it is empirical. In other words, the identification of Events 0–3 and X in itself is certainly helpful in reconstructing the formative processes that led to the existence of particular creoles, but the Events’ potential for generating new

318

Jacques Arends

Table 13.1 Landmark demographic events Name

Meaning

Event 0 Event 1

Onset of colonization Substrate population reaches numerical parity with superstrate population Locally born substrate population reaches numerical parity with total superstrate population End of regular substrate immigration

Event 2 Event 3 Event X*

Locally born substrate population reaches numerical parity with foreign-born substrate population

* Event X is placed separately from Events 0–3 in this table to indicate explicitly that it does not necessarily follow Event 3 chronologically; theoretically, it may occur anywhere from before Event 1 until after Event 3, but in practice it will usually occur between Events 1 and 3.

Table 13.2 The relative timing of demographic milestones and its linguistics consequences Timing

Linguistic effects

Late Event 1 Late Event 2 Late Event 3

Less chance of a creole coming into existence Less chance of an emerging creole getting stabilized Greater chance of a creole remaining an autonomous language variety Greater chance of increased structural difference between a creole and its lexifier

Late Event X

research questions, for example with regard to the respective roles of children, adolescents, and adults in creole formation, is at least as important. In an attempt to formalize the role of demographic factors in pidginization and creolization, Bickerton (1984) introduced the notion of a “Pidginization Index,” postulating a relationship between demographic disproportion on the one hand and the degree of language restructuring on the other. The latter is defined in terms of the structural distance between a pidgin and its lexifier language. (Note that in this view pidgins are chaotic, rudimentary forms of communication which are turned into creoles by the first generation of locally born children; see Veenstra, this volume.) Bickerton’s reasoning was that the more restructured the pidgin, the more the formation of the ensuing creole would be subject to the operation of the Bioprogram. The Pidginization Index (PI) takes into account three demographic factors: the number of years

A Demographic Perspective on Creole Formation 319 between Event 0 and Event 1 (= Y); the number of substrate speakers present at Event 1 (= P); and the average annual import of substrate speakers between Event 1 and Event 3 (= R). The formula reads: PI = Y × P/R. The idea is that a slow increase of the black population during the pre-Event 1 period (i.e., Y is high) will decrease the degree of restructuring (because every substrate speaker will have enough first-language superstrate models to learn the language from), whereas a high rate of post-Event 1 immigration of slaves relative to the Event 1 population (i.e., P/R is low) will increase the degree of restructuring (because newcomers will have to rely more and more on secondlanguage speakers to serve as models). While Bickerton (1984) only used the Pidginization Index in a very loose, unquantified way, an empirical testing by Singler (1990; see also Bickerton 1992 and Singler 1992) showed it to be invalid. This does not mean, of course, that demographic disproportion is irrelevant to creole formation, only that it does not operate according to the simple formula proposed by Bickerton. Another observation by Singler (1986) related to the Creole-to-Bozal ratio is that, in certain instances at least (e.g., Jamaica), the nativization of the black population (i.e., the replacement of African-born by locally born) took place at a very low speed, to the extent that 60 years after colonization by the English only 10 percent of the black population was locally born. While the black population continued to exhibit negative natural growth (i.e., death rates exceed birth rates) until as late as the 1840s, a continuous stream of new slaves kept coming from Africa. Other factors that contributed to the low nativization rate, not only in Jamaica but in many other colonies as well (except in North America), were the low birth rates and high mortality rates, especially among infants and children. The birth rates, in turn, were influenced by other demographic variables, especially the sex ratio among imported slaves: the smaller the proportion of women of child-bearing age, the fewer children would have been born. In this connection, it is important to realize that for certain demographic variables, such as sex ratio, relatively small skewings may have relatively big consequences, e.g., in terms of the number of children being born. Suriname is a case in point: as shown by Arends (1995a, pp. 282–4), among slaves imported into Suriname between 1684 and 1803 only 40 percent were female. This provides at least part of the explanation for the low birth figures that obtained in that colony throughout the formative period. As a result of that, the 10 percent figure of locally born blacks was not reached until around 1740 (Arends 1995a). In other words, almost a hundred years after the onset of colonization, nine out of every ten blacks in Suriname were still African-born. A dramatic example is the 1680–90 decade in Suriname, when almost 10,000 new slaves were imported; at the beginning of that decade there were only some 1,500 blacks present in the entire colony (Arends 1995a, p. 264). It may be useful to pause for a moment and ponder the implications of a demographic anomaly such as this. Imagine that nine out of every ten people in your society are recent immigrants, with one out of every two babies dying before the age of five, and new immigrants pouring in continuously, to such

320

Jacques Arends

an extent that ten years from now more than three-quarters of the present population will have been replaced by new immigrants. (Note that in making this estimate we have assumed that the 1,500 “old slaves” present in 1680, most of whom were above the “critical age” of five, had a higher life expectancy than the 10,000 newcomers; in actual fact, the proportion of the “new” population may have been as high as 90 percent or more). And the same cycle is repeated again and again (although with a gradually decreasing effect). This little thought experiment may suffice to show that the consequences of such demographically abnormal circumstances, both linguistic and otherwise, can only have been enormous. It is a wonder that societies like these created any form of stability at all, in language or in any other domain of social and cultural life. Another approach that takes into account the Creole-to-Bozal ratio, but without seeing the timing of any of Baker’s Events as crucial in itself, is Chaudenson’s (2001 [1992]) “approximation model.” According to this author, creolization is the result of the approximate acquisition by increasing numbers of Bozal slaves of the approximate varieties of the lexifier language spoken by the Creoles. In other words, creolization occurs when the language variety being acquired is an approximation of an approximation of the lexifier. The crucial event in this process is the shift in the mode of production from the société d’habitation, the ‘homestead society’ with small-scale, farm-like units where a few Europeans and Africans worked side by side, to the société de plantation, the ‘plantation society’ with large-scale, agricultural units where a few Europeans had power over a hundred Africans or even more. Most colonies started off with a period during which the population contained more Europeans than Africans (the homestead society phase, Phase I). This situation changes drastically as each colony develops into a real plantation society (Phase II), when large numbers of Africans are imported to work the coffee plantations and especially the sugar plantations. While in Phase I substrate speakers have sufficient access to superstrate speakers to be able to acquire a reasonably accurate variety of it, substrate speakers arriving during Phase II will have to depend more and more on earlier arrivals speaking approximate second-language varieties of the lexifier. Where this situation continues, it is thought to lead to an ever increasing structural distance between the newly emerging language variety and the superstrate, the ultimate result being a creole. One thing that should not be forgotten, however, is that many colonies showed significant variation in the size of their plantations, with a small number of mega-estates being supplemented by a large number of medium and small establishments. This suggests that Chaudenson’s model should be interpreted with care.

2.4

Children versus adults

Although one particular theory of creole formation, Bickerton’s (1981, 1984) Language Bioprogram Hypothesis, attributes a special role to children in that

A Demographic Perspective on Creole Formation 321 process, there seems to be a growing consensus among creolists that creole formation involves both adults and children. As far as the role of children in creole formation is concerned, Bickerton maintains that their number has no significance whatsoever. In his view, even one child would be enough to get creolization going (Bickerton, 1991, p. 38). This explains why Bickerton (1991, 1992) has always denied the importance of the evidence adduced by such scholars as Singler (1986, 1990) and Arends (1989; cf. also Arends, 1995a), which shows that the number of locally born children in at least some slave communities was extremely low – but cf. Singler (1995), where it is shown that there were significant numbers of children present in Haiti and Martinique early on. Nevertheless, it seems clear that on the basis of this evidence alone the idea of creolization as a rapid, single-generation process (a basic assumption in Bickerton’s theory) should be abandoned. In this connection, it may be worthwhile to emphasize once again that the role attributed to children in the creolization process by Singler (1990, 1995) is diametrically opposed to that attributed to them by Bickerton: in Bickerton’s scenario children are responsible for making a creole less similar to its lexifier (via the operation of the bioprogram), whereas in Singler’s view they are responsible for making it more similar (because they have better access to the lexifier than adults do). According to this view, the higher the proportion of locally born children, the less restructuring will take place. This may explain why some of the most radical creoles (those showing most structural divergence from their lexifier) have emerged in societies where the proportion of locally born children was low, such as Suriname, especially when compared to other colonies such as Martinique (discussed in Arends 1995a, and Singler 1995, respectively).

2.5

Crop selection

Another factor that because of its demographic implications is relevant for creole formation and that was first brought to creolists’ attention by Singler (1993a, 1995) is the nature of the crop grown on the plantations. In a comparative study of Martinique and Haiti, Singler (1995) showed that, due to the fact that working and living conditions on sugar plantations were particularly harsh, the demographic profile in sugar colonies was considerably worse than in colonies where other crops were grown. While the demography of most plantation societies (except in North America) was anomalous in showing a negative natural growth, it was even worse in sugar colonies. As a result of that, the rate of nativization of the population was even slower in sugar colonies than in others and, presumably, creolization more radical. Whereas crop selection is certainly not the only factor to determine the degree of radicalness, it is a fact that several of the most radical creoles (Sranan, Haitian) emerged in colonies where sugar was the major crop during the period of creole formation.

322

2.6

Jacques Arends

Inter-colony migration

Despite the crucial importance of emigration (out-migration) for obtaining a reliable picture of the demographic development of any society, it is often overlooked by creolists (but cf. Le Page 1960). Apart from the question of how many slaves left a particular colony in a particular period, the issue of out-migration involves other issues as well, such as the question of where the departed slaves ended up. In dealing with this issue, not only legal but also illegal trade (smuggling, stealing) should be included (cf. Singler 1993b; Parkvall 2000b). The importance of inter-colony migration has been emphasized by Chaudenson (2001 [1992]) in connection with the distinction between primary colonization and secondary colonization (colonization by Europeans coming from an already established colony). To the extent that secondary colonization is accompanied by the importation of an already existing or emerging creole, the language of the new colony is a “second generation creole.” The question whether a particular creole is a first- or a second-generation creole clearly has important implications for the reconstruction of its formative history.

2.7

The role of the founder population

Another demography-related concept that has been popularized in creolistics more recently is that of “founder population,” a notion adapted from evolutionary biology to refer to the group of people who are the very first to settle a particular colony (Mufwene 1996, 2001). According to Mufwene, these first settlers have a disproportionately strong influence in determining the makeup of the language variety that will emerge in the colony. The idea that the founder population has an influence beyond its numbers is known in biology as the Founder Principle or Founder Effect. It is similar to the Doctrine of First Effective Settlement used in cultural geography (Mufwene 2001, p. 27). According to this principle, since all relevant colonies were settled by Europeans, the creoles that emerged were influenced to a disproportionately strong degree by the European languages spoken by these settlers, even though they were vastly outnumbered later by the Africans brought in as slaves. This line of thinking is closely connected with Chaudenson’s (2001 [1992]) view that creoles are primarily continuations of European languages rather than autonomous linguistic entities. The extent to which this view has any value for creoles other than the ones for which it was originally proposed (those lexically based on French) remains an open question.

2.8

Demographic developments within the population of color: The mulatto-to-black and free-to-enslaved ratios

The black population is often viewed as an undifferentiated class of enslaved people, even though this group was socially stratified from early on, not only

A Demographic Perspective on Creole Formation 323 in terms of the division between those born in the colony and those born elsewhere (see above), but in terms of other dimensions as well, especially color and “enslavedness” (cf. Arends 2001). With regard to the former, since mulattoes (persons of mixed black-white ancestry) were almost without exception the children of a white father and a black mother, they often occupied a position of status within the community of color, being employed in relatively prestigious functions such as domestic service. Since mulatto children were often manumitted from slavery by their white fathers, they also formed a substantial part of the group of free people of color. From early on, however, this also included free blacks, some having been manumitted by their owners but others, e.g., skilled craftsmen, having earned enough money to buy their freedom. As a result, a new group intermediate between the “ordinary slaves” and the free whites began to form, quite small at first but soon increasing in size. It is difficult to establish exactly how the emergence of this “middle class” may have influenced the creolization process, but it seems without doubt that it did play a role, for example in the process of linguistic homogenization. Occupying a social position between the black and white segments of the population, they may have spoken something in between the language varieties characteristic of these two groups (cf. the distinction between nengre tongo ‘Black Sranan’ and bakra tongo ‘White Sranan’ in Suriname).

2.9

Non-demographic factors

Obviously, demographic factors are not the only ones to determine the outcome of language contact in plantation societies. Other factors that have been proposed are: economic change, particularly the shift from homestead to plantation (Chaudenson 2001 [1992]; motivation (McWhorter 2000); the question whether or not formal education is provided (Roberts 1999, referring to Hawai[i); and a host of other factors (cf. Parkvall 2000a, p. 197). One thing, however, should be made clear: the fact that demographic disproportion is neither a necessary nor a sufficient condition for creole formation (McWhorter 2000; Parkvall 2000a) does not mean that it does not play any role at all. McWhorter’s (2000) suggestion that demographic disproportion only plays a role in the preservation of creoles, not in their formation, is based on faulty reasoning: the fact that creole formation may occur in the absence of demographic disproportion and vice versa does not mean that demographic disproportion is irrelevant to it. It only shows that other variables have to be taken into account as well. Creole formation, just like any other language contact phenomenon, is far too complex to lend itself to monocausal explanations; demographic factors are only part of the story.

3 Some Demographic Pitfalls Before we conclude this chapter, it may be useful to include a brief discussion of the potential pitfalls that present themselves to creolists who wish

324

Jacques Arends

to incorporate demographic factors into their accounts of creole formation. The discussion is based on a sampling of the errors I encountered while reviewing the literature. On the basis of these errors, I propose six golden rules, which I will deal with in turn. 1

2

3

4

5

Do it yourself. Where possible, check original sources – and read historians – yourself. All too often one finds that creolist C refers to demographic data provided by creolist B, without realizing that B in turn has taken his data from A. It is very easy to introduce errors into these chains of reference, especially in a field such as demography where numbers are of paramount importance. Be critical. Don’t assume that original (archival, documentary) sources are necessarily reliable because they are original; be aware of possible distortions due to error, political considerations, fraud, etc. Original documents were written in a specific period, at a specific place, by a specific author, who may have had specific interests. Therefore, they may have been subject to certain constraints as a result of which the representation of reality may be more or less distorted. Well-known examples are the underreporting of the numbers of slaves by their owners with the aim of evading taxes, or the deliberate misrepresentation of population figures with the aim of misleading political enemies. Be specific. Don’t confuse general slave trade figures with colony-specific figures. Be aware that the demographic profile (e.g., in terms of ethnolinguistic distribution) of slaves brought to a specific colony is not necessarily identical to that of the general trade of the supplying nation. For those cases where colony-specific figures are available (e.g., in the Eltis et al. (1999) database), general figures should always be checked against the latter. In cases where colony-specific figures are not available, general figures may, of course, have to be used, albeit with due care. Don’t confuse estimates with data. Be aware that what are presented by some authors as if they were hard data are in some cases nothing more than estimates on their part. There is nothing wrong with estimates, as long as they are well-founded and their grounds made explicit. Nevertheless, hard data, based on original documentary sources (but with the caveat noted in Rule 2), should always be preferred. Don’t confuse immigration figures with population figures. Always clearly distinguish figures regarding immigration from those regarding population. Again, this may sound trivial, but the two are sometimes confused by creolists who identify the demographic profile of the immigrant population in a particular period (i.e., those who enter a colony between two moments in time, t1 and t2) with that of the resident population (i.e., those who live in a particular colony at some point(s) in time between t1 and t2). Due to a variety of factors, the demographic profile of the resident population at some point in time tx is not necessarily identical with the profile of the cohort of immigrants who arrived in a particular period of time preceding tx.

A Demographic Perspective on Creole Formation 325 6

Don’t confuse birth figures with population figures. This too may sound trivial, but the two are sometimes confused, e.g., in cases where birth figures are available but population figures are not. Although the birth rate of a population during a particular period gives some indication of the size of that population at the end of that period, there are many other factors involved as well, such as death rate, immigration, and out-migration.

This list could probably be expanded with at least another half a dozen rules. Suffice it to say that all of the above are based on errors that were found in well-known publications, including some by highly regarded creolists. Creolists in general tend to underestimate the difficulties inherent in demographic analysis. To the amateur, demography may appear as a rather straightforward or even “easy” science, whereas it is in fact a highly complex one. Part of its complexity arises from the fact that demography, somewhat like language, is a field where “tout se tient,” where everything hangs together with everything. A relatively small change in one variable, e.g., sex ratio, may have relatively big consequences in one or more other variables, e.g., birth rate. Hopefully, this analogy between demography and language will serve to make creolists aware that they should approach demographic problems with the same care and subtlety as they do problems in linguistics.

4 The Demography of Slavery: A Short Guide to the Literature Readers who wish not only to learn more about the demography of slavery but who would also want to consult a general introduction to historical demography are referred to Livi-Bacci (2001). As to the demography of slavery, this involves two aspects: the demography of the slave trade, on the one hand, and the demography of slave societies, on the other. Both areas of research, but especially the former, have experienced enormous growth during the last three decades. Curtin’s (1969) quantitative study of the Atlantic slave trade formed the beginning of a new line of research that continues to the present day. In spite of a number of revisions that have been made in the meantime, Curtin’s overall figure of 9 million slaves who disembarked in the New World has not been modified drastically; Eltis’s (2001, p. 43) calculations come up to 11 million. The most recent study covering the entire Atlantic slave trade is Eltis (2001), which is based on Eltis et al.’s (1999) database on CD-ROM, entitled The Transatlantic Slave Trade. The database includes information on a wide variety of variables for more than 27,000 transatlantic voyages that took place in the period between 1562 and 1867 (an estimated 70 percent of the total transatlantic slave trade; the 30 percent not covered relates especially to the Brazilian trade). One should not expect a work of this magnitude to be completely free of errors (cf. Parkvall 2000b, p. 117). Nevertheless, it constitutes an important

326

Jacques Arends

database which until now has not been used widely by creolists (but cf. Parkvall 2000b, pp. 121–38). Among specialists, however, it has already begun to yield its first results, in the form of special issues of the journals Slavery & Abolition (18–1, Eltis & Richardson 1997b), and William and Mary Quarterly (third series, 58–1, Eltis & Morgan 2001). It is important to note, however, that only transatlantic shipments are included; intra-Caribbean movements are completely left out of the picture. Since the latter were extremely common, this implies severe restrictions for the demographic inferences that can be made on the basis of the CD-ROM. This is especially relevant for creole genesis, since many slaves were shipped to other destinations within the Caribbean during that period. This means that the database can only yield real linguistic profit if it is used with the necessary care. The most informative case study on a specific slave-trading nation is Postma (1990) on the Dutch trade. A recent synthesis of the Atlantic slave trade has been published by Klein (1999), containing a section entitled “Bibliographic essay,” which the reader may consult for further references. A much more elaborate bibliography has been published by Miller (1999), who also publishes regular bibliographic updates in Slavery & Abolition, together with the Journal of African History, the most important journal in this area. As to internet resources, the website of York University’s Tubman Centre (directed by slavery historian Paul Lovejoy) provides an excellent starting point. There is also the site of the Virginia Foundation for the Humanities on “The North Atlantic slave trade and slave life in the Americas” (constructed by slavery historian Jerome Handler); it contains a superb collection of images related to the topic. (For details, see the list of internet resources at the end of this chapter.) Of equal importance for creole formation as the trade from Africa are issues related to the demographic development of the plantation societies in the Americas, such as birth and death rates, black-to-white ratios, and life expectancy. Moreau de Jonnès (1987 [1842]) seems to have been the only study in this area for almost one and a half centuries. The most important publications of the past few decades are Kiple (1984) and Sheridan (1985). Volume 3 of the General History of the Caribbean, entitled The Slave Societies of the Caribbean (Knight 1997), contains excellent survey articles by Palmer (1997) and Engerman and Higman (1997). While the bulk of attention has been devoted to the African population in the New World colonies, the demography of the European population, in particular the so-called indentured servants, has also been addressed. Their role in creole formation should not be underestimated, as they formed an important part of the white population during the initial period of colonization. The most comprehensive studies on this topic are Smith (1947) and Galenson (1981). A potentially useful source for filling in the “gaps” in the documentary record, especially as regards the black population, is provided by Coale and Demeny (1983), although it should be noted that their “model life tables” are based on stable populations. As far as synthesizing studies from a specifically creolistic point of view are concerned, there are very few. In fact, the chapter on demographic issues in

A Demographic Perspective on Creole Formation 327 Parkvall (2000b, pp. 117–38) is the only study that comes anywhere near attaining that goal. Parkvall has assembled demographic data for a wide range of Atlantic creoles of all lexifiers (English, French, Dutch, Portuguese, Spanish). The reader should be aware, however, of certain limitations inherent in Parkvall’s data as well as in the way they are presented. First of all, the sources of information used by Parkvall are quite diverse, ranging from wellestablished sources such as Curtin (1969) and Eltis et al. (1999) to the author’s own estimates. In several cases, tables are based partly on one source and partly on another; or on a very small number of voyages; or partly based on a nation’s general trade and partly on colony-specific trade. These problems are aggravated by the fact that none of the tables are accompanied by captions, which enhances the risk that they may be used uncritically by other creolists. With these caveats in mind, however, the information provided by Parkvall provides a useful first survey of the demography of creolization in the Atlantic area. Additional references to the literature on the demography of slavery are given below under “Further reading.”

5 Outlook Despite some recent skepticism concerning the role of demographic factors in creole formation, in particular concerning the role of demographic disproportion (cf. McWhorter 2000), it is not appropriate to exclude demography from the set of variables that play a role in creole formation. The incorporation of demographic factors into theories of creole genesis has only just begun and much is still to be learned. Future research on the demography of specific plantation societies will be instrumental in shaping and refining our image of the societies that gave rise to the formation of creoles.

NOTE This chapter was written while the author was a fellow at the Netherlands Institute for Advanced Study in the Humanities and Social Sciences (NIAS), whose support is gratefully acknowledged. The author would like to thank Margot van den Berg and Gigi Santow as well as the editors of this volume for their detailed and pertinent comments on an earlier version of this chapter.

REFERENCES Arends, Jacques (1989) Syntactic Developments in Sranan: Creolization as a Gradual Process. PhD dissertation, University of Nijmegen.

328

Jacques Arends

Arends, Jacques (1995a) Demographic factors in the formation of Sranan. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 233– 85. Arends, Jacques (1995b) The socio-historical background of creoles. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 15–24. Arends, Jacques (2001) Social stratification and network relations in the formation of Sranan. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 291–307. Baker, Philip (1982a) On the origins of the first Mauritians and of the creole language of their descendants: A refutation of Chaudenson’s “Bourbonnais” theory. In: Philip Baker and Chris Corne (eds.) Isle de France Creole: Affinities and Origins. Ann Arbor: Karoma, pp. 131–257. Baker, Philip (1982b) The Contribution of the Non-Francophone Immigrants to the Lexicon of Mauritian Creole. PhD dissertation, University of London. Baker, Philip and Corne, Chris (1986) Universals, substrata and the Indian Ocean creoles. In: Pieter Muysken and Norval Smith (eds.) Substrata Versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 163–83. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1984) The Language Bioprogram Hypothesis. Behavioral and Brain Sciences 7, 173–88. Bickerton, Derek (1991) On the supposed “gradualness” of creole development. Journal of Pidgin and Creole Languages 6, 25–58. Bickerton, Derek (1992) The sociohistorical matrix of creolization. Journal of Pidgin and Creole Languages 7, 307–18. Chaudenson, Robert (1974) Le lexique du parler créole de la Réunion. Paris: Champion. Chaudenson, Robert (2001 [1992]) Creolization of Language and Culture (Revised in collaboration with Salikoko S. Mufwene). London: Routledge. (French original published 1992, Des îles, des hommes, des langues. Paris: L’Harmattan.) Coale, Ansley J. and Paul Demeny (1983) Regional Model Life Tables and Stable Populations. 2nd edn. New York: Academic Press. Curtin, Philip (1969) The Atlantic Slave Trade: A Census. Madison: The University of Wisconsin Press. Daeleman, Jan (1972) Kongo elements in Saramacca Tongo. Journal of African Languages 11, 1–44. Eltis, David (2001) The volume and structure of the transatlantic slave trade: A reassessment. William and Mary Quarterly 3rd Series, 58, 17–46. Eltis, David and Philip D. Morgan (eds.) (2001) William and Mary Quarterly, special issue, 3rd Series, 58, 1. Eltis, David and David Richardson (1997a) The “numbers game” and routes to slavery. In: David Eltis and David Richardson (eds.) Routes to Slavery: Direction, Ethnicity and Mortality in the Transatlantic Slave Trade. London: Frank Cass, pp. 1–15. Eltis, David and David Richardson (eds.) (1997b) Routes to Slavery: Direction, Ethnicity and Mortality in the Transatlantic Slave Trade. London: Frank Cass. (Originally published as a special issue of Slavery & Abolition 18(1).) Eltis, David, Stephen D. Behrendt, David Richardson, and Herbert S. Klein (1999) The Transatlantic Slave Trade, 1562–1867: A Database on CD-ROM. Cambridge: Cambridge University Press.

A Demographic Perspective on Creole Formation 329 Engerman, Stanley L. and Barry W. Higman (1997) The demographic structure of the Caribbean slave societies in the eighteenth and nineteenth centuries. In: Knight (ed.), pp. 45–104. Galenson, David W. (1981) White Servitude in Colonial America: An Economic Analysis. Cambridge: Cambridge University Press. Huttar, George (1985) Sources of Ndjuka African vocabulary. New West Indian Guide 59, 45–71. Kiple, Kenneth F. (1984) The Caribbean Slave: A Biological History. Cambridge: Cambridge University Press. Klein, Herbert S. (1999) The Atlantic Slave Trade. Cambridge: Cambridge University Press. Knight, Franklin W. (ed.) (1997) General History of the Caribbean. Volume 3: The Slave Societies of the Caribbean. London/Basingstoke: UNESCO Publishing/Macmillan Education. Lefebvre, Claire (1998) Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Le Page, Robert B. (1960) An historical introduction to Jamaican Creole. In: Robert B. Le Page and David De Camp, Jamaican Creole. London: Macmillan, pp. 1–124. Livi-Bacci, Massimo (2001) A Concise History of World Population. 3rd edn. Oxford: Blackwell. McWhorter, John H. (2000) The Missing Spanish Creoles: Recovering the Birth of Plantation Contact Languages. Berkeley: University of California Press. Migge, Bettina (2003) Creole Formation as Language Contact: The Case of the Suriname Creoles. Amsterdam: John Benjamins. Miller, Joseph C. (ed.) (1999) Slavery and Slaving in World History: A Bibliography. Armonk, NY: Sharpe. Moreau de Jonnès, Alexandre (1842) Recherches statistiques sur l’esclavage colonial et sur les moyens de le supprimer. Paris: Imprimerie de Bourgogne et Martinet. (Reprinted 1987, Geneva: Slatkine Reprints.) Mufwene, Salikoko S. (1996) The founder principle in creole genesis. Diachronica 13, 83–143. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Palmer, Colin (1997) The slave trade, African slavers and the demography of the Caribbean to 1750. In: Knight (ed.), pp. 9–44. Parkvall, Mikael (2000a) Reassessing the role of demographics in language restructuring. In Ingrid Neumann-Holzschuh and Edgar Schneider (eds.) Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins, pp. 185–213. Parkvall, Mikael (2000b) Out of Africa: African influences in Atlantic Creoles. London: Battlebridge. Postma, Johannes M. (1990) The Dutch in the Atlantic Slave Trade, 1600–1815. Cambridge: Cambridge University Press. Rens, Lucien L. E. (1953) The Historical and Social Background of Surinam Negro-English. Amsterdam: North-Holland. Roberts, Sarah J. (1999) Nativization and the genesis of Hawaiian Creole. In: John H. McWhorter (ed.), Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 257–300. Sheridan, Richard B. (1985) Doctors and Slaves: A Medical and Demographic History of Slavery in the British West Indies, 1680–1834. Cambridge: Cambridge University Press.

330

Jacques Arends

Singler, John Victor (1986) Short note. Journal of Pidgin and Creole Languages 1, 141–5. Singler, John Victor (1990) On the use of sociohistorical criteria in the comparison of creoles. Linguistics 28, 645–59. Singler, John Victor (1992) Nativization and pidgin/creole genesis: A reply to Bickerton. Journal of Pidgin and Creole Languages 7, 319–33. Singler, John Victor (1993a) African influence upon Afro-American language varieties: A consideration of sociohistorical factors. In: Salikoko S. Mufwene (ed.), Africanisms in Afro-American Language Varieties. Athens, GA: The University of Georgia Press, pp. 235–53. Singler, John Victor (1993b) The African Presence in Caribbean French Colonies in the Seventeenth Century: Documentary Evidence (Travaux de Recherches sur le Créole Haïtien, vols. 16–17.) Montréal: Département de linguistique, Université du Québec à Montréal. Singler, John Victor (1995) The demographics of creole genesis in the Caribbean: A comparison of Martinique and Haiti. In Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 203–32. Singler, John Victor (1996) Theories of creole genesis, sociohistorical considerations, and the evaluation of evidence: The case of Haitian Creole and the Relexification Hypothesis. Journal of Pidgin and Creole Languages 11, 185–230. Smith, Abbot E. (1947) Colonists in Bondage: White Servitude and Convict Labor in America, 1607–1776. Chapel Hill: The University of North Carolina Press.

FURTHER READING Coale, Ansley J. and Paul Demeny (1983) Regional Model Life Tables and Stable Populations. 2nd edn. New York: Academic Press. Galloway, J. H. (1989) The Sugar Cane Industry: An Historical Geography from Its Origins to 1914. Cambridge: Cambridge University Press. Hall, Gwendolyn M. (1992) Africans in Colonial Louisiana: The Development of AfroCreole Culture in the Eighteenth Century. Baton Rouge: Louisiana State University Press. Hall, Gwendolyn M. (1999) The Louisiana Slave Database and the Louisiana Free Database: 1719–1820 (CD-ROM). Baton Rouge: Louisiana State University Press. Livi-Bacci, Massimo (2001) A Concise History of World Population. 3rd edn. Oxford: Blackwell. Miller, Joseph C. (ed.) (1999) Slavery and Slaving in World History: A Bibliography. Armonk, NY: Sharpe. Shepherd, Verene A. and Hilary M. Beckles (eds.) (2000) Caribbean Slavery in the Atlantic World: A Student Reader. Kingston, Jamaica: Ian Randle. Smith, Abbot E. (1947) Colonists in Bondage: White Servitude and Convict Labor in America, 1607–1776. Chapel Hill: The University of North Carolina Press. Van Stipriaan, Alex (1993) Surinaams Contrast: Roofbouw en overleven in een Caraïbische plantage-economie, 1750–1863. [Surinamese Contrast: Depletion and Survival in a Caribbean Plantation Economy, 1750–1863]. Leiden: KITLV Press. Watts, David (1987) The West Indies: Patterns of Development, Culture and Environmental Change since 1492. Cambridge: Cambridge University Press.

A Demographic Perspective on Creole Formation 331 Wells, Robert V. (1975) The Population of the British Colonies in America Before 1776: A Survey of Census Data. Princeton: Princeton University Press.

Internet resources www.yorku.ca/nhp/areas/nhp.htm. Website of the joint York University– UNESCO “Nigerian Hinterland Project,” containing much useful information including links to other websites (Lovejoy). hitchcock.itc.virginia.edu/Slavery/. Website of the Virginia Foundation for the Humanities, containing hundreds of images related to the Atlantic slave trade and slave life in the Americas (Handler and Tuite).

332

John Victor Singler

14

The Sociohistorical Context of Creole Genesis JOHN VICTOR SINGLER

1

Introduction

Creole studies has assumed a link between sociohistorical circumstance and linguistic outcome from the outset, from the nineteenth-century work of Schuchardt and Van Name onward. Investigations of sociohistorical factors and forces continue to be central to the field, especially to scholars’ efforts to understand the nature of creole genesis. I devote the bulk of the present chapter to an investigation of the role of sociohistorical forces in creole genesis and end with a discussion of one aspect of the intersection of history and linguistic outcome, that pertaining to the diffusion of features from creole to creole. As with much else in the field, the inquiry into the nature and consequences of sociohistorical factors has focused upon languages that arose as a consequence of European global expansion, with the greatest amount of attention devoted to the new languages that resulted from European expansion in the Caribbean, lesser amounts to ones in the South Pacific, West Africa, and the Indian Ocean, and virtually no attention whatever to those that arose elsewhere under other circumstances. Further, there has frequently been in the literature a blurring between circumstances specific to European expansion and circumstances held to be universal in creole genesis. A further characteristic of the study of pidgins and creoles is that the emphasis has been overwhelmingly on creoles, with pidgins – understood here in the strict sense of varieties not spoken natively – seen as being of less importance. This is not surprising: Chomskyan linguistics privileges native speakers, and pidgins, by this definition, do not have them. The present chapter is likewise guilty of the charge of concentrating on creoles largely to the exclusion of pidgins, and – apart from a brief discussion of selected Pacific varieties – limiting its focus to Caribbean plantation creoles. Other chapters in this volume deal with the entire class of pidgins as well as “pidgincreoles” (Bakker), with pidgins and creoles whose lexifier is nonIndo-European (Versteegh), and with the spectrum of languages that arise

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

The Sociohistorical Context of Creole Genesis 333 from situations of linguistic contact without being seen as pidgins or creoles (Mesthrie). Creoles are not a tightly defined group. Efforts by McWhorter to provide a linguistic definition of creoles notwithstanding (1998), the languages of the world do not divide into creoles and non-creoles by any set of rigorously defined linguistic criteria. (Fundamental objections to the phonological aspect of McWhorter’s proposal can be seen in Smith’s chapter in this volume, and more general objections in Thomason’s chapter.) No linguistic litmus test exists for determining creole status (cf. Mufwene 1986; Singler 1990), and the question remains unanswered as to whether a sociohistorical test can be substituted. It almost seems at times as if the set of creoles is simply those languages that linguists consider to be creoles. This is not as arbitrary and circular as it sounds: the languages thus included have had comparable and interlocking histories. If creoles can be this loosely defined, then it is neither surprising nor entirely unwarranted for particular linguists to construct theories about creole genesis and the nature of creole languages on the basis of only a subset of creoles. This has often been the case, with the focus of such theories ordinarily concentrated on those creoles considered most different from their European lexifiers, so-called “radical” creoles. A final preliminary point is that the principles responsible for the creation and character of individual creole languages are principles that govern all language and all language birth. The explicit rejection of creole exceptionalism, to frame the discussion in DeGraff’s (2003) terms, is stated by Muysken (1988, p. 300): “The very notion of a ‘creole’ language from the linguistic point of view tends to disappear if one looks closely; what we have is just a language.” In the chapter that follows, I begin by reviewing the sociohistorical circumstances that led to creole formation in the Caribbean, focusing on the different historical phases of development in plantation societies which were first recognized in the study of French-lexifier creoles, the “société d’habitation” and “société de plantation” (section 2). With its link to the rapidity of occurrence of “Event 1” – the point where the white and black populations reach numerical parity (Baker 1982) – this model of the development of plantation societies has far-reaching implications for theories of creole genesis. Accordingly, I use this scenario as the basis for examining competing views of creole genesis (sections 3 and 4). I then briefly shift the focus of creole genesis to the Pacific, first to Hawai[i and then to the Solomon Islands and Papua New Guinea (section 5). After this examination of the sociohistorical settings in which creole genesis took place, I consider the ways in which modern sociolinguistic principles and research are relevant to the historical settings in which creole genesis occurred (section 6). Beyond the debate on how creoles come into existence, creolists have also concerned themselves with a second question: Why do creoles, regardless of their lexifier languages and substrata, have so many grammatical properties in common? In considering this, I begin by questioning the extent to which it

334

John Victor Singler

is true. Then, for features where it does seem to hold, I assess in some detail Baker’s (1999) attempt to invoke diffusion as an explanation for shared features, specifically in the narrower case where the varieties in question have the same lexifier language and are located in the same part of the world (section 7). In so doing, I address a range of specific concerns regarding Baker’s methodology and findings.

2

The Sociohistorical Setting: Société d’habitation, then société de plantation

In the 1960s and 1970s, it was generally assumed that creoles evolved from pidgins, an assumption that formed a central part of Hall’s (1966) “life cycle” theory of creole genesis. Indeed, a widely accepted definition of a creole – one that is still frequently heard – was that a creole was a pidgin that had acquired native speakers. Thus, it was assumed that on Caribbean plantations the very first cohorts of slaves would have been thrown together without a language in common and without sufficient access to native speakers of a given colony’s dominant language (ordinarily, the language of those governing the colony) to acquire that language. The initial result would have been a pidgin and then, when children were born into this setting, the pidgin would have undergone nativization and thereby expanded into a creole. However, the idea that every creole had passed through a prior pidgin stage was questioned by Alleyne (1971), who noted that no evidence existed for such a stage in the development of Caribbean creoles (cf. Lalla & D’Costa 1990).1 Support for the view that a pidgin stage was not a prior condition for creole genesis comes from a close examination of the sociohistorical circumstances that led to creole genesis in the French colonies of the Caribbean, most prominently in the work of Chaudenson (1992, 2001). At the outset in these colonies, Africans were less numerous than whites. Moreover, in this early stage, colonies often had a significant number of engagés, indentured servants, who worked in close contact with the enslaved Africans. In this initial period, called by Petit Jean Roget (1980) the société d’habitation, households tended to be small. Furthermore, during this period, slaves ordinarily lived under a single roof with the owner, the owner’s family, and any engagés. Gautier (1985) argues that in France at the time there was a sharp status distinction between the head of a household and everyone else in it. She sees the household structure in the early French colonies of the Caribbean as an extension of this model, with slaves simply absorbed into it. Given not only work settings in which slaves and engagés worked together, but also a household situation in which slaves lived with the other residents of the habitation, it seems highly plausible that the first cohorts of slaves in such situations would have had sufficient access to speakers of the lexifier language that they would have acquired a second-language version of it.

The Sociohistorical Context of Creole Genesis 335 Then, as detailed by Singler (1993), when a colony switched to sugar growing, its socioeconomic structure was transformed utterly. Higman (2000) examines and ultimately affirms the validity of referring to the transformation that sugar wrought in colonial Caribbean society as the “sugar revolution.” He states: “The six central elements of the sugar revolution are commonly regarded as a swift shift from diversified agriculture to sugar monoculture, from production on small farms to large plantations, from free to slave labour, from sparse to dense settlement, from white to black populations, and from low to high value per caput output” (p. 213). While other crops grown in the colonial Caribbean – tobacco, indigo, provisions – could be produced in agricultural units large or small, by large cadres of workers or by only a few, sugar requires a large initial investment and is only economically viable as a plantation crop. Furthermore, the more land and workers that planters had, the more they wanted and the more they could afford. In sugar societies, engagés ceased to be part of the labor force (Singler 2006a, p. 163 reports on the debate as to why this happened); the reliance on slave labor was now virtually absolute. The centrality of sugar to slave economies throughout the Caribbean can hardly be overestimated, for it is sugar that gave the Atlantic slave trade its enormity. Writing with reference to the Western Hemisphere as a whole, Eltis states that “from the late sixteenth century down to 1820, 90 percent of the slaves were brought across the Atlantic because of sugar” (2001, p. 35). At the level of the individual colony, the consequences of a switch to sugar growing, hence from a société d’habitation to a société de plantation, included the following: • A sharp change in the ratio of Africans to Europeans: The sugar growers’ intense demand for slave laborers resulted in a steep rise in the number of Africans without a comparable increase in the European population. • A substantial reduction in the amount of overall contact that Africans had with Europeans: Engagés ceased to be part of the labor force, and slaves were no longer part of the household, instead living in their own quarters and living lives far more separate from those of Europeans than before. Additionally, sugar growers bought out most of the small farmers, further reducing situations where black–white contact might have occurred. Another aspect of Caribbean sugar plantations that seems clearly to bear upon creole genesis is that agricultural workers there labored under desperately insalubrious conditions. In a comparison of the cotton plantations of the American South to sugar plantations in the British Caribbean, Fleischman, Tyson, and Oldroyd (2004) speak of “demographic disaster for sugar plantation slaves. Mortality was significantly higher and fertility much lower in the British West Indies as numerous observers have demonstrated statistically.”2 In few colonies of the Caribbean did the slave population reproduce itself. In all the rest, natural decrease obtained. Thus, maintenance of a slave labor force depended upon the ongoing importation of new slaves. Sheridan (1972, p. 28), in an examination of mortality and slave import figures in Jamaica, concludes

336

John Victor Singler

further that natural decrease there worsened during the period of agricultural intensification, i.e., during the shift to sugar. Debien (1974, p. 345), summarizing the views of planters in French colonies in the Caribbean, states: “Tous les habitants conviennent que les pertes qu’ils éprouvaient pendant les huit premières années de leur acquisition peuvent sans exaggeration être évaluées à 50%” (‘All the planters agreed that the losses they suffered during the first eight years after the acquisition [of a slave] could be assessed without exaggeration at 50%’; my translation). It is certainly possible that the French slaveowners were overstating the direness of the situation. However, even if they were, linguists concerned with the setting in which creole genesis occurred need to assess the likely linguistic consequences of a situation in which much of the population is perpetually newly arrived. (See also Arends, this volume.) As indicated, once sugar-growing came in, it transformed a colony, shifting it from société d’habitation to société de plantation. In the period leading up to the shift, the African population came to outnumber the European population. Creolists seem to agree that how long the European portion of the population remained in the majority had a direct bearing on how closely the colony’s vernacular resembled its metropolitan counterpart or some variety thereof. As a general rule, the shorter that time period, the more “radical” the creole, and the longer that period, the closer to the lexifier. Arends (this volume) provides a detailed discussion of population ratios and their likely linguistic import. Chaudenson rightly notes that “numerical balance between the European and slave populations does not in any way entail a magical decisive point in the evolution of these societies” (2001, p. 64). At the same time, such figures as the length of time between a colony’s founding and the occurrence of parity have proven to be significant. Moreover, the numbers have a concreteness that speculation, for example, about the attitudes of the enslaved toward the language of slaveowners cannot achieve. The significance of a habitation/plantation model (or something akin to it) for creole emergence is widely recognized. I now turn to a consideration of opposing views of the impact of these different phases on creole development. The opposition lies in the relationship posited between lexifier and creole. One view assumes an uninterrupted process of language transmission. Thus, even as the emerging variety diverges further and further from the lexifier, it remains genetically related to the lexifier as its daughter language (section 3.1). The other view sees the divergence as a consequence of an interruption in language transmission, hence a severing of genetic affiliation between lexifier and creole (section 3.2).

3 3.1

Implications for Creole Genesis The continuity hypothesis

The French who settled in colonies in the Caribbean3 came primarily from the north and west of France, above “a line running from Bordeaux to Paris”

The Sociohistorical Context of Creole Genesis 337 (Chaudenson 2001, p. 141n). Chaudenson posits their language as being “most likely nonstandard French, marked by langue d’oïl dialectal features” (p. 152).4 He sees it as probably having “undergone the kind of restructuring associated with the strategies of informal learning” (p. 152). It is important for Chaudenson that the earliest European settlers had begun the restructuring process, in which case when the earliest cohort of slaves sought to acquire the colonial variety, the task would have been facilitated by the prior restructuring done by the French colonists. The outcome of the slaves’ restructuring would – in Chaudenson’s view – have been an approximation of the colonists’ version. Thus would begin an ongoing sequence of approximations. As Mufwene (1994, p. 70) puts it, “According to Chaudenson (1992), the whole process of linguistic creolization (from the point of view of language acquisition) consisted of continuous approximations of approximations of language varieties . . .” For Chaudenson, then, the key element in creole genesis is the (nonstandard) superstrate, and the key process is its restructuring. With the transition from société d’habitation to société de plantation, locally born slaves would have become intermediate, themselves acquiring approximations of the French of the settlers and their language simultaneously being the model for imitation/approximation by those slaves newly arrived from Africa. This model – which Parkvall (2006) labels the “continuity hypothesis” – sees creolization in the circumstances of the French Caribbean as taking roughly a hundred years, by the end of which “basilectalization” had occurred in the speech of those slaves most peripheral to the colonist core. For the French colonies, Chaudenson emphasizes the length of time that the Europeans were in the majority, stating that “numerical equality between the White and Black populations was achieved after a relatively lengthy period (50–100 years)” (2001, p. 64). However, census records indicate otherwise, with numerical parity achieved in the earliest French colonies in the Caribbean within 40 years or less.5 Recall the general assumption among creolists that the shorter the time to numerical parity between the white and black populations, the less the colony’s language is expected to be like its lexifier. Mufwene (1996, 2001) follows Chaudenson in focusing on the superstrate as the source of a creole’s properties. He frames Chaudenson’s continuity hypothesis in terms of population genetics and applies the “founder principle” by which the earliest individuals had an importance far out of proportion to their small numbers (1996). Subsequently, Mufwene (2001, p. 27) notes the parallels between the founder principle and the cultural geographer Wilbur Zelinsky’s “Doctrine of First Effective Settlement” (1992 [1973]). A crucial difference between Zelinsky and Mufwene, however, is that cultural geography allows for human agency – discussed below – while Mufwene, in keeping with population genetics, explicitly rejects it. Like Chaudenson, Mufwene sees the earliest members of the enslaved populations, who had extensive contact with speakers of French, as acquiring a second language (L2) variety of the colonists’ restructured version of French, with the first slaves’ L2 variety then providing the basis for the language of succeeding cohorts. “By the founder principle, the newcomers simply found it

338

John Victor Singler

more practical to learn the vernacular spoken by the slaves who preceded them . . .” (Mufwene 2004, p. 209). Mufwene, emphasizing even more strongly than Chaudenson the diversity within the European populations of Caribbean colonies, designates the colonists’ restructured vernacular a koiné. However, he “diverges from the established position that koinés develop by leveling out differences among dialects of the same language or among genetically and typologically related languages” (2001, p. 5). Instead, he sees the varieties that emerge as more “the results of competition and selection rather than any kind of common denominators of the dialects in contact” (p. 5). As support for his assertion that the settler population was diverse and that, consequently, koineization started there, Mufwene draws attention to European indentured servants who were not native speakers of the lexifier language. According to Mufwene: “Based on Dyde (1993) and Beckles (1990), regarding St. Kitts and Barbados, respectively, one is led to speculate that the practice of recruiting labor from places other than the metropole of a colony was a common practice.” Dyde makes reference to a “large number of Irish . . .” and Beckles speaks of Scottish Protestant and Irish Catholic indentured servants. It is likely that a significant number of the Irish indentured servants (but not the Scottish Protestant ones) were monolingual speakers of Gaelic. However, the question arises as to how many speakers this would have entailed: an examination of Beckles’s full discussion shows that ordinarily the indentured servants in Barbados were English. Moreover, in contrast to Chaudenson’s arguments for dialectal diversity among early French immigrants to the Caribbean, Smith (this volume) points out that “some form and some degree of Standard English was by the seventeenth century known to most English (and Scots) speakers, whether as an active second dialect, or as a passive second dialect.” Smith concludes that “in the various colonial environments, where the native speakers of various English and Scots dialects came together, there was no need for the development of a new koiné – some form of Standard English was already waiting in the wings.” The need for koineization is, if anything, even less motivated in the colonies where Dutch creoles arose. Van Rossem (1996) notes that, during the early period of colonization, most Dutch settlers came from Zeeland and were speakers of dialects within the West Flemish/Zeelandic dialect continuum. He cites Hesseling (1905) as making “a number of remarks which point towards one regional variety of Dutch as being the superstrate of Negerhollands” and provides evidence from Vercoullie (1919) in further support of this view. I leave in abeyance the question of whether genuine and extensive koineization took place in French colonies in the Caribbean; however, the evidence argues against its occurrence in colonies where English or Dutch varieties formed the dominant superstrate. Mufwene’s model allows for the introduction of features from substrate sources but predicts that the superstrate is the more likely source, by virtue of its precedence. Crucially, the concept is termed the “founder principle,” not

The Sociohistorical Context of Creole Genesis 339 the “founder hypothesis.” The terminological distinction is relevant in that it is not clear to me that the founder principle is open to refutation. If a feature’s selection is always proof of the feature’s having had a more fortuitous combination of properties and circumstances than competing variants did, then ultimately the principle founders on its non-falsifiability. Parkvall (2006) observes that Chaudenson and Mufwene and also DeGraff (2001a, 2001b) and Alleyne (1996) hold that the French-lexicon creoles “have diachronically distanced themselves more and more from their lexifier” (p. 314). By this reckoning, having had more than three centuries to move away from French, French-lexifier creoles are or should be quite far removed from it today. However, Parkvall, drawing on data in Fattier’s (1998) linguistic atlas of Haiti, makes a compelling case that French “exercised a post-formative influence on Haitian,” i.e., that Haitian Creole more closely resembles French today than it did three centuries ago. At the very least, Parkvall’s evidence calls into question the idea that Haitian and other French-lexifier creoles have been moving away from French in the same way that French, Spanish, and other Romance languages grew further and further from Latin over time.

3.2

Disrupted transmission

The views put forward by Chaudenson and Mufwene stand in opposition to a more widely held view (though the view can hardly be said to be superior simply because it has more adherents), one that sees creoles as the outcome of a disruption of transmission, a disruption brought about because the new speakers, i.e., enslaved laborers, had scant ongoing access to native speakers of the lexifier language.6 The most fully worked out statement of this view is that put forward by Thomason and Kaufman (1988, henceforth T&K). T&K are like Chaudenson and, before him, Alleyne (1971) in rejecting the notion of a prior, fully crystallized pidgin as a necessary precondition for the emergence of a creole. Instead, creoles arise as the products of “language shift without normal transmission” (T&K, p. 147).7 Language shift with “normal” transmission involves a shift by a community of speakers of one language to another language, where members of the group have access to native speakers of the language being shifted to; while it is possible for this kind of shift to occur in as short a time as a single generation, it usually takes longer, up to several centuries. In contrast, in the type of creolizing situation that concerns T&K, the circumstances for such “normal” transmission do not obtain. Rather, multiple groups of speakers have simultaneous, immediate need for a new medium of communication without having the requisite access to speakers of the “target language.” (I return below to the concept of “target language.”) The urgency of the need for a language prompts T&K to posit the new language as emerging via “abrupt creolization,” by which they mean that linguistic expansion occurs before linguistic stabilization has been achieved. Such a sequence – expansion before stabilization – provides a setting in which interference from substrate languages readily

340

John Victor Singler

manifests itself. Thus, rather than seeing creoles as restructured continuations of the superstrate language, T&K see them as “mixed languages” (p. 3), as languages “whose lexical and grammatical structures cannot all be traced back primarily to a single source (‘parent’) language” (Thomason, this volume). T&K do not put forth their model as applying to all creoles, only to those situations where widespread communicative need requires the immediate formulation of a new medium. In societies where community members by and large have the ability to communicate with at least some other members of their community, T&K argue, the emergence of the new language is less pressing. In such cases they do not expect “abrupt creolization” but rather a more gradual process of creole development. In the “abrupt creolization” cases, T&K assert that communicative exigencies are simply too great to allow for the emergence of a “fully crystallized pidgin language,” as such a process takes time, requiring some degree of conventionalization, so that speakers learn it as a language (p. 149). Thus, T&K take issue with those scholars who “apparently . . . [assume] that you have to pidginize (i.e., simplify . . . ) the TL [Target Language] completely before you can creolize (expand) it” (p. 352n). With specific regard to the French colonies in the Caribbean, the scenario presented for these societies in 3.1 implicitly endorses T&K’s assertion that crystallized pidgins did not emerge. In the first phase the language of the slave cohort would seem to have been most appropriately characterized as an L2 variety of French. During this period, i.e., the comparatively short habitation phase (40 years or less), it seems likely that the nature and degree of Africans’ contact with Europeans, especially European workers, would have yielded sufficient exposure to the lexifier language and – arguably – some motivation to acquire it. However, the societal transformations brought on by the rapid switch to sugar would have created a setting in which exposure and motivation would have been sharply curtailed for most Africans. Generally, then, the switch to sugar had as a direct linguistic consequence the disruption of transmission of the lexifier language that T&K posit. Whether it also induced “abrupt” creolization is far less certain. After all, communicative efficacy can be achieved without the functional material that would be developed through widescale grammaticalization. No doubt some parts of the grammar fell into place readily, but others – TMA, for example – clearly did not or did not need to. A model that distinguishes between communicative efficacy and grammaticalization does not depend upon abrupt creolization, allowing instead for it to be gradual (Arends 1986, 1989, 1995a, 1995b, and elsewhere; Singler 1986, 1988, 1990, 1992, 1996, and elsewhere.); such a model, I would argue, more fully reflects what occurred in the Caribbean. (The continuity hypothesis, it should be noted, is intrinsically gradualist.) A further but separate point regarding sugar colonies in the Caribbean is that it was the case that, for decades upon decades after colonization, most of the enslaved population was African-born. Singler (1986) cites historians’ estimates that in Jamaica, for example, 60 years after English colonization

The Sociohistorical Context of Creole Genesis 341 began, only 10 percent of the enslaved population was locally born. Despite the fact that the locally born had an importance in creole genesis out of proportion to their numbers, one must also reckon with the significance of the 90 percent and the persistent “partial repidginization” (Mühlhäusler 1997, p. 10) or “successive second language acquisition” (the phrase is Don Winford’s, cited in Singler 2006b, p. 347). (Note that creolists differ as to the nature of the importance of locally born slaves. I have argued in Singler (1992, 1998, 2006a) that, ceteris paribus, the greater the proportion of locally born slaves, the closer the creole is to its lexifier language. See also Arends, this volume.) There would have been newly arrived Africans in significant numbers for the duration of the Atlantic slave trade, i.e., for two centuries and longer. Even if proceeding from the founder principle and/or the Doctrine of First Effective Settlement, one can still see this demographic distribution as producing a temporal overlap between pidginization and creolization and also possibly as constraining the direction of creolization, i.e., so as to favor that expansion which was more readily acquired by people who were early on in the secondlanguage-acquisition process. The ongoing numerical predominance of the newly arrived would have meant as well that input from substratal languages would have been present and available not only during pidginization/reduction but also during creolization/expansion.

4 Creole Genesis as Success Story Baker (1990, 1995) takes issue with both the Chaudenson and T&K scenarios for creole genesis. He argues that the two – and creole genesis scenarios more generally – define the creation of a creole as implicitly the outcome of speakers’ failure to acquire the superstrate language. Baker argues instead that, for enslaved Africans in the Caribbean, their greatest linguistic need was the ability to communicate with other Africans, specifically with those with whom one did not already have a language in common. He suggests that what developed on a plantation was a medium of interethnic communication (MIC). Over time the MICs from different plantations would fuse into a colony’s creole. The MIC/creole should be seen, Baker argues, as a successful response by slaves to their need to communicate among themselves. Baker’s hypothesis is that participants in plantation situations, motivated by the desire to solve the problem of interethnic communication rather than by a desire to learn the language of the whites, “set about creating a new language, and succeeded in that endeavour” (1995, p. 13). Baker’s arguments are fundamentally persuasive. At the same time, while it is appropriate to focus on the function of language as communication, one ought neither to erase entirely the function of language in the construction of an individual’s identity nor to gainsay the overt prestige that the superstrate language possessed (a point to which I return below). Certainly at the very beginning of a speech variety’s emergence, especially in the context of early

342

John Victor Singler

sugar plantations, communicative necessity is likely to have been paramount; indeed, it may have been the only thing that mattered. However, it seems likely that identity and the related concept of language prestige would very quickly have emerged as factors as well. In discussing creole formation as success and in rejecting the concept of the European lexifier as target language, Baker frames his views in terms of speakers’ motivation for using the new variety. Indeed, his 1995 article is entitled “Motivation in creole genesis.” This raises the question of speaker agency. To what extent did speaker choices shape the creoles that emerged? Baker and likewise Jourdan (this volume) ascribe a significant, even determinative, role to participants in making the choices that yield languages. Mufwene (2001) takes the opposite point of view, one in which conscious speaker choice plays no meaningful role in determining the character of the language that emerges. (See also Trudgill 2006, who endorses Mufwene’s point of view on the matter.)

5

The Pacific

While studies of creole genesis have largely concentrated on the Caribbean, in the 1970s Derek Bickerton placed Hawai[i Creole English (HCE) at the center of the discussion (1980 [1974]). Because HCE’s genesis was so much more recent than that of its Caribbean counterparts, it was possible, Bickerton argued, to have access not only to early creole speakers but also to speakers of the English-lexicon pidgin that had been antecedent to the creole. Sugar monoculture, as it had in the Caribbean from the seventeenth century onward, came to dominate the Hawaiian Islands in the final quarter of the nineteenth century. Beginning in 1876, growers imported large numbers of contract laborers from Asia and beyond. Bickerton draws on data from Japanese and Filipino elders who had been contract laborers in the early twentieth century and were still alive in the 1970s (Bickerton & Odo 1976). He calls attention to the profound difference between the idiosyncratic, non-systematic language of the elders,8 which he takes to represent HCE’s precursor, Hawai[i Pidgin English, and modern HCE. He argues that, the utterly chaotic nature of the putative pidgin notwithstanding, a full-blown language emerged rapidly, and asserts that the only possible explanation for this remarkable metamorphosis lies in the Language Bioprogram Hypothesis (1984). However, Roberts (2000, 2005) mines the extensive documentation of language use and life in late nineteenth-century and early twentieth-century Hawai[i to demonstrate that creolization was not the work of the first native-born generation (as Bickerton would have it) but of the second and, further, that it occurred away from the plantation, not on it, specifically in Hawaiian cities. The Japanese and Filipino elders whose data Bickerton had drawn upon had arrived too late and were in the wrong place; their speech played no role in the emergence of HCE.9

The Sociohistorical Context of Creole Genesis 343 Veenstra (this volume) discusses in detail the implications of Roberts’s scholarship for Bickerton’s claims, and Singler (2006b) draws on Roberts’s work to set out what he sees as fundamental differences between the sociohistorical setting of creole genesis in Hawai[i and the Caribbean. What is relevant for the present discussion about Roberts’s work is her compelling demonstration that HCE, whose status as a creole is unquestioned, arose in sharply divergent circumstances from those that gave rise to creole languages in the Caribbean. Along similar lines, Jourdan (1985, 1994, Jourdan & Keesing 1997) has documented the creolization/expansion of Solomon Islands Pijin over the final decades of the twentieth century. A pidgin acquired by men from the Solomons during terms of work on plantations in Queensland and Fiji in the late nineteenth and early twentieth centuries had, upon the men’s returning home, remained a secondary language, limited in its uses. However, with the development of Honiara as the capital of the Solomons in the second half of the twentieth century, individuals’ need to communicate across the full range of human interaction with people whose wantok (first language) was different from their own led to the increased use of Solomon Islands Pijin. Jourdan makes the case that the circumstances that arose in an urban setting necessitated the functional expansion of Pijin, that this functional expansion in turn generated linguistic expansion, and that the linguistic expansion in question preceded nativization of the variety rather than following from it. Sankoff and Laberge’s earlier study (1973) likewise points to urbanization rather than nativization as having promoted the expansion of Tok Pisin in Papua New Guinea. The relevance of the Pacific cases is that they show the range of possible settings in which creolization can occur. Social settings that differ dramatically from those posited as having obtained in Caribbean colonies have nonetheless yielded creole languages.

6 Sociolinguistic Considerations: The Locus of Prestige An examination of sociohistorical forces ought also to take into account sociolinguistic ones, including those that have emerged from current sociolinguistic research. The ongoing danger is that principles arrived at through the study of modern industrialized societies may have limited applicability to plantation settings, especially given the aberrant behavior required to sustain chattel slavery. Still, one element of the continuity hypothesis, particularly as articulated by Chaudenson, seems to go against fundamental aspects of the sociolinguist’s understanding of human society. This is Chaudenson’s argument that colonial society had a single locus of prestige, the colonists. That is, there was overt prestige without an attendant, oppositional covert prestige. In Chaudenson’s view, slaves varied as to how close they were to the locus of prestige (depending in part as to whether they were locally born or Africanborn), but their attraction to the source of prestige was unmistakable.

344

John Victor Singler

Chaudenson quotes d’Ans approvingly: “ . . . we must note that what structured the world of the slaves from within established the basis of prestige and justified hierarchies. This was not loyalty to ‘Africa’, but rather adoption of white behaviors” (d’Ans 1987, pp. 238–9, quoted in Chaudenson 2001, p. 128). There was, in Chaudenson’s scenario, no sense of difference between slaves and other household members in the habitation phase, hence no linguistic impetus for asserting difference, and there was no emergence of a separate locus of prestige then or subsequently in the plantation phase. Chattel slavery as an institution, most especially that practiced on sugar plantations, could only be sustained by ongoing, systematic violence of the most unspeakable sort. Whip-lashings, maiming, torture, and capital punishment were routinized practices. Yet Chaudenson would have us believe that slaves – in some unconflicted way – simply wished to be more like their torturers. If the punishments that I mentioned were infrequently applied on some or even most plantations, the fact remains that sugar plantations were rigidly hierarchized and regulated. Moreover, language is not simply about communication; it is also about identity and membership. The cost of sounding more like one’s superiors, i.e., one’s colonial masters, is that one also sounds less like one’s fellows. In modern urban society, imitation of the highest social group is strongest among those closest to it socially, i.e., those who see themselves as having a chance of acceptance into that group. However, even for those nearest the top, imitation of prestige seems to occur extensively only when social structure is not rigid. Crucially, linguistic imitation of the next-highest group does not seem to extend across society, hence does not seem to be graduated such that each group is anxious to imitate the one immediately above it. Rather, those at a further remove from the top tend not to find imitation of this sort worthwhile. Indeed, for them it comes at a cost. Among those subject to poverty and discrimination, solidarity emerges as an overriding virtue, and abandoning the language of the group in order to imitate the speech of those with greater power generates sanctions and censure from the group (Tonkin 1994: 187, summarizing Milroy 1987). I do not deny that the language of the powerful carried overt prestige; what I find implausible is the idea that it would have been the sole variety with any affective force. In the circumstances of racebased enslavement on the plantation, the notion that everyone wanted to talk as white as possible is, to say the least, improbable. In this regard, Arends’s (2001) study of the internal social stratification of the Surinamese plantation community is relevant. He points to a dichotomy between “elite” slaves and all other slaves (p. 296). The “elite” slaves, which included black overseers, skilled slaves (carpenters, coopers, sawyers, apprentices), and domestic slaves, were most likely to have frequent verbal interactions with whites. Moreover, as Arends points out specifically for black overseers, they functioned also as a buffer of sorts between masters and other slaves (p. 297) – a role probably performed by other elite slaves as well. Both the frequency of their interactions with whites and their position within the

The Sociohistorical Context of Creole Genesis 345 slave force make it possible that the elite slaves conformed to Chaudenson’s claim of identification of white forms of behavior with prestige. It is highly unlikely, however, that this was equally true of the other slaves (unskilled slaves, field slaves), whose interactions with whites would have been far more limited, and whose social position did not make white behavior particularly desirable. But even within the elite class, Arends points to the possibility of ambiguity in roles. Thus, in his discussion of the role of black overseers, he points to evidence of their leadership roles in subversive activities (rebellions, escapes) and in Afro-Surinamese religious affairs (p. 297). In sum, those within the enslaved population with the most status and the most contact with whites may well have positioned themselves ambiguously vis-à-vis whites and other blacks. The bulk of the slave population, however, would seem to have had far less incentive (let alone opportunity) for imitating white speech and might well have faced disapproval from their peers for so doing.

7 Cross-Creole Similarities 7.1

How similar are creole grammars?

A question that has long engaged creolists is the extent to which creoles, regardless of their lexifier language and substrata, have properties in common. This question has been central to creole studies in the field’s modern era, at least dating from the Mona conferences of 1959 (Le Page 1961) and, especially, 1968 (Hymes 1971). In the discussion that follows, I consider in detail diffusion as explanation for the similarities that obtain among English-lexicon creoles, particularly those in the Atlantic. However, before doing so, I feel it appropriate to ask just how similar creole grammars are to one another. In the 1988 Bickerton-Muysken debate in the Cambridge survey, Pieter Muysken titles his contribution, “Are creoles a special type of language?” As part of his argumentation that they are not, Muysken challenges the premise that creole languages are more “alike” than other languages. He acknowledges similarities in creole word order (cf. Bakker, this volume), in creole TMA systems, and in a general tendency for creoles to rely less on inflectional morphology than do their lexifier languages, but he also calls attention to the wide variation that obtains from creole to creole with regard to properties such as serial verb constructions and preposition stranding – variation that reflects the full range of possibilities that obtain in the world’s languages. Likewise, in the preface to their 1995 textbook, Arends, Muysken, and Smith state that “diversity rather than unity is taken to be the central theme” of the book: “While every creolist will acknowledge the continuous excitement of little shocks of déjà vu when exploring the grammatical patterns of other creoles, careful analysis reveals as many differences as similarities” (1995: v). As growing numbers of creolists produce in-depth grammars of individual creoles, it is the differences that seem prominent. Further, efforts to compare creoles

346

John Victor Singler

– especially if very many creoles are involved – often seem to do violence to a given creole’s system simply in the interest of facilitating comparison. Certainly there are similarities that require explanation, but they are neither so numerous nor so profound nor so unexpected as to require extraordinary theories.

7.2

Diffusion as an explanation for the similarities that exist: Baker’s (1999) treatment of Atlantic English Creoles

When creoles are organized by lexifier language and region, much of what is shared is perhaps more readily explained as a consequence of diffusion rather than as repeated parallel development ab initio. Thus, Chaudenson speaks of first-, second- and even third-generation creoles (see note 5 below). Hypotheses have been advanced that specify a single location from which, for example, all English- or French-lexicon creoles in the Caribbean are said to have originated. A West African source has been posited for English creoles (the Upper Guinea Coast, Hancock 1986; Coromantin on the Gold Coast, McWhorter 1999), but – as Huber (1999) convincingly demonstrates – the hypotheses in question lack historical plausibility.10 Scenarios that take more localized points of diffusion seem more promising, because they are more in line with actual dispersal patterns of people. Thus, Baker (1999) moves the focus to St. Kitts, site of the first English settlement in the Caribbean, and argues that it is “one starting point for Atlantic Creoles” (p. 346). Baker’s detailed exposition makes his work available for scrutiny. His methodology is modeled on Clark (1979); Baker himself had earlier employed it in his investigation of the relationship of Pacific Pidgin Englishes to one another (1993). It involves compiling a list of features that he “considered to be significant deviations from British English . . . – phonological, lexical, and grammatical – which I knew or suspected to be attested” prior to 1850 in one or more of nine Atlantic English Creoles (AECs): Antiguan, Bajan, Gullah, Guyanese, Jamaican, Kittitian, Sierra Leonean Krio, Sranan (which includes Sranan and Saramaccan attestations),11 and Vincentian (Baker 1999, p. 316).12 In all, Baker found evidence for 138 features in two or more of the nine AECs. Counting every feature equally – whether phonological, lexical, or grammatical – Baker then considered the number of features for which he had evidence for a given AEC, next measuring for each pair of AECs the number of features that they shared.13 To remove any bias that would obtain simply because a given AEC had more of the features to begin with, Baker used the following formula: First, calculate the number of features each pair would share if distribution of these were random, then deduct the result from the actual number of features shared by each pair. For example, 69 of the 138 features are attested in Atg

The Sociohistorical Context of Creole Genesis 347 [Antiguan] and 86 of the 138 features are attested in Bjn [Bajan]. If the distribution of features were random, they would be expected to share 69/138 × 86 = 43.0 features whereas they in fact share 54, i.e. 11.0 more than predicted. (1999, p. 337)

For Baker, the higher the number, i.e., the greater the extent to which the actual results exceed the predicted ones, the more closely related the two creoles are. Nine creoles yield 36 pairs; of these, the actual number of shared features is higher than the expected number in 33 pairs. Baker’s emphasis on diffusion as a source for shared features seems sound; nonetheless, questions arise about his methodology, his findings, and his interpretation of those findings. With reference to methodology: in calculating the number of features out of the total of 138 that are shared by a given pair of creoles, Baker includes those 23 features that were in fact shown to obtain in every one of the nine AECs. Some of these 23 are hardly Atlantic-creolespecific but are instead characteristic of “English-lexicon contact varieties” more generally, to use Baker’s phrase. He identifies the use of one as an indefinite article and the use of me as a first person singular subject pronoun as two such features. The phonological substitution by which English interdental fricatives are replaced with alveolar stops is surely another such feature. Regardless of the cause of the 23 features’ widespread distribution, if the purpose of the calculations is to establish the comparative strengths of the relationships of given pairs of creoles, inclusion of features shared by all makes no contribution (Major 2006). Table 14.1 presents Baker’s original set of figures as well as a new set arrived at by removing the 23 features in question and then recalculating the affinities. For each cell in the table, the upper number represents Baker’s original calculation, i.e., the value that he arrived at by using the algorithm presented above. (This set of numbers in table 14.1 comes directly from Baker 1999, p. 337, but I have reordered some of the rows for expository purposes.) The lower number in each cell in table 14.1, the one in bold, likewise shows the difference between the predicted and actual values, but this time the calculations have been redone with the 23 features removed. In many cases, in instances where Baker’s original table seemed to imply that a connection between two creoles was reasonably certain, the adjusted figures signal a weaker relationship or even no relationship at all.14 With the recalculation, fully one-third of the scores are negative, a result that in each case implies the absence of a substantive relationship between the members of the pair in question. Table 14.1 shows that, of the Caribbean creoles, five clearly pattern together, the creoles of St. Kitts, Barbados, Antigua, Guyana, and St. Vincent. (They are in the first five rows of the table.) Each of the five has comparatively high scores when paired with any of the others, always 4.5 or above in the recalculated set of figures. In all, among the recalculated figures, only six pair scores are higher than 7.0, and five of them involve pairs from within this quintet.

348

John Victor Singler

Table 14.1 Difference between number of features shared by each pair of creoles and the predicted number, before and after adjustment SKi+

Atg 11.5 7.8

Bjn 10.1 7.3

Guy 9.1 7.1

SVi 8.6 4.5

Gul 6.0 3.0

Jam −0.1 −1.1

Kri 0.9 −1.5

Srn −5.1 −7.5

Bjn+

SKi 10.1 7.3

Guy 10.1 7.3

Atg 11.0 5.8

SVi 10.4 4.6

Gul 6.9 2.7

Jam 2.8 1.4

Kri 4.0 0.7

Srn 1.0 −2.3

Atg+

SKi 11.5 7.8

SVi 15.0 7.4

Guy 10.5 6.8

Bjn 11.0 5.8

Gul 8.0 2.4

Jam 1.5 −0.4

Kri 1.5 −3.0

Srn −2.5 −7.0

Guy+

Bjn 10.1 7.3

SKi 9.1 7.1

Jam 7.9 6.9

Atg 10.5 6.8

Gul 9.0 6.0

SVi 9.6 5.5

Kri 6.9 4.5

Srn 1.9 −0.5

SVi+

Atg 15.0 7.4

Guy 9.6 5.5

Bjn 10.4 4.6

SKi 8.6 4.5

Jam 3.5 1.4

Gul 4.2 −2.0

Kri 2.2 −2.7

Srn 0.2 −4.7

Gul+

Kri 12.7 9.1

Guy 9.0 6.0

SKi 6.0 3.0

Bjn 6.9 2.7

Atg 8.0 2.4

Srn 4.7 1.1

Jam 1.3 −0.3

Svi 4.2 −2.0

Jam+

Guy 7.9 6.9

Srn 5.8 4.6

Kri 4.8 3.6

SVi 3.5 1.4

Bjn 2.8 1.4

Gul 1.3 −0.3

Atg 1.5 −0.4

Ski −0.1 −1.1

Kri+

Gul 12.7 9.1

Srn 9.3 6.4

Guy 6.9 4.5

Jam 4.8 3.6

Bjn 4.0 0.7

SKi 0.9 −1.5

SVi 2.2 −2.7

Atg 1.5 −3.0

Srn+

Kri 9.3 6.4

Jam 5.8 4.6

Gul 4.7 1.1

Guy 1.9 −0.5

Bjn 1.0 −2.3

SVi 0.2 −2.3

Atg −2.5 −7.0

SKi −5.1 −7.5

The top number in each cell represents the difference between the actual number of features shared by a given pair of creoles and the predicted number; these figures come from Baker (1999, p. 337). The bottom number – in bold – is the recalculated figure once the 23 features shared by all nine creoles are removed from consideration. The recalculations are provided by Kyle Major. Key: Atg = Antiguan, Bjn = Bajan, Gul = Gullah, Guy = Guyanese, Jam = Jamaican, SKi = Kittitian, Kri = Sierra Leonean Krio, Srn = Sranan, SVi = Vincentian

The Sociohistorical Context of Creole Genesis 349 By and large the five creoles do not pattern with any of the remaining four. There are two types of exceptions to this statement, one being that Guyanese displays a strong link to every language in the sample except Sranan, and the second that Kittitian, Bajan, and Antiguan (the three oldest creoles in the sample) each have scores with Gullah that suggest a meaningful though weaker connection. Even with the scores adjusted to remove the universal features, three types of problems persist. The first involves instances where scores do not correspond to what is known about settlement patterns, either because population movement is known to have occurred (and to have occurred early) yet the score for a pair is low, or vice versa, i.e., when a high score obtains despite the absence of relevant population movement. The most dramatic illustration of the latter involves Sranan and Krio. There is no evidence of early direct contact between Suriname and Sierra Leone, yet the Sranan–Krio pair is Sranan’s highest and Krio’s second highest score. (These results hold whether one uses the original calculations or the adjusted ones.) Baker recognizes that these results present a problem (p. 349). Surinamese planters settled in Jamaica, and Jamaica Maroons are one of the three groups who came from outside Africa to form Krio society, but the Sranan–Krio score (6.4) is higher than either the Sranan-Jamaican (4.6) or Jamaican-Krio (3.6) score. In a later study of “Atlantic, Pacific, and world-wide features” in 13 pidgins and creoles, Baker and Huber (2001) once again find positive scores for the Sranan–Krio pair; in that study, however, Sranan–Jamaican gets a higher score than does Sranan–Krio, but Sranan–Krio gets a better score than Jamaican–Krio in three of the four scales used in that article. These scores, both in Baker (1999) and Baker and Huber (2001), are at odds with the known history of the colonies in question. The second problem is one of chronology, and it is best illustrated by the results involving Guyanese. Strictly speaking, it is the explanation that Baker offers for the results as much as the results themselves that present the problem. As noted, except for Guyanese–Sranan, every pair that includes Guyanese gets a high score. When the recalculated scores are used, Jamaican gets its highest score when Guyanese is its partner, Gullah, Bajan, and Vincentian their second highest, and Kittitian, Antiguan, and Krio their third highest. Baker, when presenting the history of the Guyanese settlement, states, “English planters with slaves began to arrive in Essequibo and Demerara in the 1740s . . .” (1999, p. 340). Then, to account for the “relatively high betterthan-predicted figures for features shared with all the other Creoles except Srn [Sranan],” Baker suggests that these scores “may in part be due to the influence of the more than 40,000 contract immigrants from the West Indies who reached Guyana in the period 1835–1917 (Rickford 1987, p. 60)” (1999, p. 340). If this is the most apt explanation, then we are forced to reckon with a scenario in which new (core) features are introduced into Guyanese 100–150 years after the English established themselves in the Essequibo and Demerara, and 200–250 years after the first English settlements in the Caribbean. If that is

350

John Victor Singler

the appropriate scenario, then what does it mean to speak of diffusion with reference to creole genesis? If there is to be no reckoning as to when the diffusion occurred, the significance of the putative relationships between given pairs of creoles is diminished. The third problem involves a specific pair, Jamaican–Kittitian, and it becomes apparent when one compares Baker (1999) with Baker and Huber (2001). The latter work looks at seven Atlantic and six Pacific pidgins and creoles. The Atlantic varieties are the Caribbean creoles Kittitian, Jamaican, Bajan, and Sranan, plus Gullah, Krio, and West African Pidgin English. The focus of that article is limited to grammatical features and lexical items; in all, 302 features are considered, and they are classified as Atlantic, Pacific, or world-wide. Scoring by pairs is done in this article too, but the features are broken down – for the Atlantic varieties – into (a) Atlantic and world-wide, (b) world-wide features only, (c) African features within Atlantic, and (d) non-African features within Atlantic, and each pair of creoles is assigned a score for each of these categories. Baker and Huber find “significantly positive scores . . . seemingly implying close genetic links” for five of the pairs, one of which is Jamaican– Kittitian (p. 188). That assessment of the Jamaican–Kittitian pair stands in sharp contrast to Baker (1999, p. 339), where it is said that Kittitian “appears to have had no significant influence” on Jamaican. It was mentioned above that only 3 of the 36 pairs received a negative score in Baker’s calculations; Kittitian–Jamaican is one of them. Was there “no significant influence” of Kittitian on Jamaican, or do the two creoles display “close genetic links”? Neither article addresses the discrepancy between the two results. The difference in the Kittitian–Jamaican results between the two articles raises a fundamental question as to the reliability and replicability of the instrument being used. If one set of features produces one result and a different set produces a different result, then how is the “right” set of features to be determined? Baker’s work is important for its demonstration of the frequency with which features have spread from creole to creole. My comments should not be taken as gainsaying Baker’s demonstration of that vitally important point. Still, whether the problems arise from flaws in methodology, dearth of necessary data, or errors in analysis, conclusions about particular pairs of the creoles under study often prove perplexing and, in some instances at least, likely to be incorrect. There can be no doubt that diffusion was a factor in creole genesis in the Caribbean (and pidgin/creole genesis in the South Pacific), and a viable theory of genesis must take it into account. At the same time, quite apart from the concerns I have raised about the methodology and findings under discussion here and allowing for the changes likely to have arisen in the three centuries and more of individual creoles’ histories, it is still the case that some among the individual creoles vary so much one from the other that limits must be recognized to the power of diffusion as an explanatory device.

The Sociohistorical Context of Creole Genesis 351

8 Some Concluding Observations So much remains unknown or uncertain about the environments in which creole languages came to be that hypotheses about creole genesis are still properly tentative. At the same time, advances in other disciplines, most dramatically in the study of the colonial era in the South Atlantic, continue to increase what can be determined about the social circumstances that gave rise historically to creole languages. In the discussion that has been presented here – and indeed in all the literature concerning the circumstances of creole genesis – there is a dual metric. The scenario that is put forward must be consistent with what is known of the history of particular places and times, but it must also be consistent with the linguistic evidence. Indeed, no matter how compelling a scenario one presents, if it is contradicted by the linguistic evidence, then it cannot be the appropriate scenario. Thus, it seems safe to say that the scenarios that accompany and are meant to motivate the most extreme denials of a role for substrate languages are falsified by the extent of substratal influence on, for example, Caribbean creoles. Conversely, the most extreme promotions of the role of substrate languages are falsified by the limitations on the extent of substratal influence. With reference to the theories that have been discussed in the present chapter, the distance between Caribbean creoles, on the one hand, and the varieties of European languages spoken by the early settlers, on the other, is simply too great and the grammatical properties of the creoles are too distinct to motivate the claim that what obtains is simply successive approximations of approximations. It is in keeping with Baker’s discussion of motivation to point out that, were the continuity hypothesis apt and had the enslaved genuinely sought to learn the language of the masters, they would have done so.

NOTES 1

2

Chaudenson (1979, 2001, pp. 22–3) distinguishes between créoles endogènes and créoles exogènes, the former sometimes referred to as “fort creoles” and the latter as “plantation creoles.” As Chaudenson’s terminology implies, créoles endogènes developed out of contact between a local population and resident European traders, e.g., the Portuguese creole(s) that developed on part of the West African coast and on nearby islands. In contrast, créoles exogènes came into being in colonies whose labor force had been created by the massive transplanting of populations. The challenge to the idea of a prior pidgin stage comes with reference to créoles exogènes. Demonstrations of differences in mortality and fertility between the Caribbean and the American South begin with Curtin (1968, 1969) and appear in Kiple (1984) and elsewhere.

352

John Victor Singler

3 Chaudenson applies his analysis to creole genesis in the French plantation colonies of the Indian Ocean as well as the Caribbean; the present discussion, however, focuses on the Caribbean. The various theories of creole genesis presented in this chapter are assumed to extend to English and Dutch plantation colonies as well, though the history of Suriname sets it apart somewhat, perhaps – it might be argued – because it had little or no société d’habitation phase (cf. Arends 2002). It should be noted, however, that, primarily on linguistic grounds, Alleyne (2001) argues that the English-lexifier and French-lexifier creoles had very different histories. For a discussion of Spanish colonies in the Caribbean, see Lipski, this volume. 4 The language of French colonists is a matter of debate not only among linguists concerned with Caribbean and Indian Ocean French-lexicon varieties, but also among those who study the history of Québécois and other North American varieties of French. Barbaud (1984, 1994), and Chaudenson’s (2001) translation of Barbaud (1984) present three somewhat divergent characterizations of the linguistic knowledge of people in the north and west of France in the early part of the colonization era. 5 At the beginning of the 1670s, five or possibly six French colonies in the Caribbean had white populations of 500 or more, listed here with their dates of founding: St. Kitts (1626), Martinique (1635), Guadeloupe (1635), Marie Galante (1648), and St. Croix (1650), with St. Domingue (Haiti) (1659) a likely sixth. For each of the four colonies founded prior to 1650, censuses from 1671 place the non-European population at 58 percent or more, suggesting that numerical parity had been reached in each case within 40 years and probably earlier. Chaudenson himself places parity in Martinique as occurring within 29 years of the colony’s 1635 founding (1979). In 1671, St. Croix had only been a French colony for 21 years; however, the black population was already at 41 percent of the colony’s total (Archives Nationales, Section Outre-Mer (Aix-en-Provence). St. Kitts G1471; Guadeloupe G1468; the others, G1498, No. 54). Chaudenson (1979, 1981, 2001) distinguishes between “first-generation” and “second-generation” creoles: “A second-generation creole is one whose development is marked by contributions from an earlier creole introduced by immigrants from another, creole-speaking territory” (2001, p. 35). Overall this is an appropriate distinction, but it has little bearing on the point that I am making here. Even if one goes so far as to limit the “first-generation” designation to the very first colony, i.e., St. Kitts, parity was most likely achieved within 40 years of 1626 by that colony, Martinique, Guadeloupe, and even Marie Galante. 6 I see a creole as the outcome of a particular type of language contact situation, extreme in certain ways, especially in the disruption of transmission. With that said, I reject any positing of an opposition between “natural” languages and creoles: what occurred in creole genesis in the Caribbean and elsewhere is a natural outcome of the sociohistorical and linguistic circumstances that obtained. (The subtitle of Thomason 1995, i.e., “ordinary processes, extraordinary results,” makes a similar point.) 7 I note the strong exception voiced by Michel DeGraff to the “abnormal transmission” hypothesized by T&K. In DeGraff (2001c and elsewhere), he sets forth his grounds for decrying “the theoretical abnormality of abnormal transmission in Creole genesis” (2001c, p. 242n).

The Sociohistorical Context of Creole Genesis 353 8

9

10

11

12 13

14

The data that Bickerton presents from the elderly speakers is highly macaronic and – in conflict with Mühlhäusler’s (1997) definition of a pidgin – is not speech for which there were “norms of acceptability.” Consequently, others have characterized it as a “jargon” rather than a “pidgin,” e.g., Bakker, this volume. Bickerton makes the point that HCE’s substrate languages were different from those of the Caribbean creoles and yet, he claims, HCE shares core grammatical properties with Caribbean creoles, even for Caribbean creoles whose lexifier language was not English. In sum, Bickerton argues, there are parallels that can be explained neither by shared lexifier language nor shared substrate. Siegel (2000) addresses Bickerton’s claim by showing – with reference to the copula system, TMA, complementation, and more – that demographically relevant substrate languages for HCE (Cantonese and/or Portuguese and not, as Bickerton would have it, Filipino languages and/or Japanese) share crucial elements of the grammatical properties in question with the Caribbean creoles’ African substrate languages. McWhorter (2000) argues that only those European languages whose speakers had slave forts on the West Coast of Africa gave rise to creoles. Inasmuch as Spain did not have such forts, McWhorter sees this as a straightforward explanation for the general paucity of Spanish-lexifier creoles. Whatever the merit of McWhorter’s case regarding Spanish-lexicon creoles (and see Lipski, chapter 22 this volume, in this regard), no explanation of what did or did not happen in Spanish colonies in the Western Hemisphere can rescue a hypothesis for English-lexicon creoles that cannot be connected to established realities of slaving and slave forts on the West African coast. Baker (p. 316) emphasizes that he combines Sranan and Saramaccan for the purpose of the article in question, and intends nothing further regarding the relationship of the two languages. Parkvall (1995) has done comparably detailed work in support of two sources of diffusion for French-lexifier creoles, namely Senegal and St. Kitts. Assigning equal value to phonological, lexical, and grammatical features – as Baker does – would seem problematic. Certainly one would expect lexical features to move more readily from one variety to another, phonological and grammatical features less easily, but Lacy (2001) shows that differential weighting by feature type yields results virtually identical to those which Baker presents. Major (2006) critiques Baker’s procedures still further and proposes more farreaching procedural changes than I have presented here – changes that not merely yield significantly different results from Baker’s but even call into question whether or not such a procedure can be valid.

REFERENCES Alleyne, Mervyn (1971) Acculturation and the cultural matrix of creolization. In: Hymes (ed.), pp. 169–86. Alleyne, Mervyn (1996) Syntaxe historique créole [Historic creole syntax]. Paris/Schœlcher, Martinique: Karthala/Presses Universitaires Créoles.

354

John Victor Singler

Alleyne, Mervyn (2001) Opposite processes in “creolization.” In: Ingrid NeumannHolzschuh and Edgar W. Schneider (eds.) Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins, pp. 125–33. Arends, Jacques (1986) Genesis and development of the equative copula in Sranan. In: Pieter Muysken and Norval Smith (eds.) Substrata Versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 103–27. Arends, Jacques (1989) Syntactic Developments in Sranan: Creolization as a Gradual Process. PhD dissertation, University of Nijmegen. Arends, Jacques (1995a) Demographic factors in the formation of Sranan. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 233–85. Arends, Jacques (1995b) The socio-historical background of creoles. In: Jacques Arends, Pieter Muysken, & Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins, pp. 15–24. Arends, Jacques (2001) Social stratification and network relations in the formation of Sranan. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 291–307. Arends, Jacques (2002) The history of the Surinamese creoles I: A sociohistorical survey. In: Eithne B. Carlin and Jacques Arends (eds.) Atlas of the Languages of Suriname. Leiden: KITLV Press, pp. 115–30. Arends, Jacques, Pieter Muysken, and Norval Smith (1995) Preface and acknowledgements. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam, John Benjamins, pp. v–vii. Baker, Philip (1982) The Contribution of the Non-Francophone Immigrants to the Lexicon of Mauritian Creole. PhD dissertation, University of London. Baker, Philip (1990) Off target? Journal of Pidgin and Creole Languages 5, 107–19. Baker, Philip (1993) Australian influence on Melanesian Pidgin English. Te Reo 36, 3–67. Baker, Philip (1995) Motivation in creole genesis. In: Philip Baker (ed.) From Contact to Creole and Beyond. London: University of Westminster Press, pp. 3–15. Baker, Philip (1999) Investigating the origin and diffusion of shared features among the Atlantic English Creoles. In: Baker and Bruyn (eds.), pp. 315–64. Baker, Philip and Adrienne Bruyn (eds.) (1999) St. Kitts and the Atlantic Creoles: The Texts of Samuel Augustus Mathews in Perspective. London: University of Westminster Press. Baker, Philip and Magnus Huber (2001) Atlantic, Pacific, and world-wide features in English-lexicon contact languages. English World-Wide 22, 157–208. Barbaud, Philippe (1984) Le choc des patois en Nouvelle-France: Essai sur l’histoire de la francisation au Canada [The shock of the patois in Nouvelle-France: Essay on the history of the francization of Canada]. Sillery, Québec: Presses de l’Université du Québec. Barbaud, Philippe (1994) Des patois au français: La catastrophe linguistique [From patois to French: The linguistic catastrophe]. In: Raymond Mougeon and Édouard Beniak (eds.) Les origines du français québécois. Sainte-Foy, Québec: Les Presses de l’Université Laval, pp. 79–99. Beckles, Hilary (1990) A History of Barbados: From Amerindian Settlement to Nation-State. Cambridge: Cambridge University Press. Bickerton, Derek (1980 [1974]) Creolization, linguistic universals, natural semantax, and the brain. In: Richard Day (ed.) Issues in English Creoles: Papers from the 1975

The Sociohistorical Context of Creole Genesis 355 Hawaii Conference. Heidelberg: Gross, pp. 1–18. (Originally appeared 1974 in University of Hawaii Working Papers in Linguistics 6(3), 125–41.) Bickerton, Derek (1984) The language bioprogram hypothesis. Behavioral and Brain Sciences 7, 173–221. Bickerton, Derek and Carol Odo (1976) Change and Variation in Hawaiian English. Vol. 1: General Phonology and Pidgin Syntax. Honolulu: Social Sciences and Linguistics Institute, University of Hawai[i. Chaudenson, Robert (1979) Les créoles français [French creoles]. Paris: Fernand Nathan. Chaudenson, Robert (1981) Textes créoles anciens (La Réunion et Île Maurice): comparaison et essai d’analyse [Early creole texts (Réunion and Mauritius): Comparison and analytic essay]. Hamburg: Helmut Buske. Chaudenson, Robert (1992) Des îles, des hommes, des langues [Islands, people, languages]. Paris: L’Harmattan. Chaudenson, Robert (2001) Creolization of Language and Culture. (Revised in collaboration with Salikoko S. Mufwene.) Routledge, London. (Translation of Chaudenson 1992.) Clark, Ross (1979) In search of Beach-la-Mar: Towards a history of Pacific Pidgin English. Te Reo 22, 3–64. Curtin, Philip D. (1968) Epidemiology and the slave trade. Political Science Quarterly 83, 190–216. Curtin, Philip D. (1969) The Atlantic Slave Trade: A Census. Madison: University of Wisconsin Press. D’Ans, André-Marcel (1987) Haïti: paysage et société [Haiti: Land and society]. Paris: Karthala. Debien, Gabriel (1974) Les esclaves aux Antilles Françaises (XVIIe–XVIIIe siècles) [Slaves in the French Antilles (seventeenth and eighteenth centuries)]. Société d’Histoire de la Guadeloupe, Basse-Terre, and Société d’Histoire de la Martinique, Fort-deFrance. DeGraff, Michel (2001a) Morphology in Creole genesis: Linguistics and ideology. In: Michael Kenstowicz (ed.) Ken Hale: A Life in Language, Cambridge, MA: MIT Press, pp. 53–121. DeGraff, Michel (2001b) “Neo”-Darwinian creolistics: A short debugging guide. Paper presented at the Society for Pidgin and Creole Linguistics Meeting, University of Coimbra. DeGraff, Michel (2001c) On the origin of Creoles: A Cartesian critique of NeoDarwinian linguistics. Linguistic Typology 5, 213–310. DeGraff, Michel (2003) Against creole exceptionalism. Language 79, 391–410. Dyde, Brian (1993) St. Kitts: Cradle of the Caribbean. London: Macmillan. Eltis, David (2001) The volume and structure of the transatlantic slave trade: A reassessment. William and Mary Quarterly 58, 17–46. Fattier, Dominique (1998) Contribution à l’étude de la genèse d’un créole: l’atlas linguistique d’Haïti, cartes et commentaires [Contribution to the study of the genesis of a creole: Linguistic atlas of Haiti, maps and commentaries], six volumes. Villeneuve d’Ascq, France: Presses Universitaires du Septentrion. Fleischman, Richard K., Thomas N. Tyson, and David Oldroyd (2004) Somebody knows the trouble I’ve seen: A critical and comparative analysis of racial aspects of slave plantation accounting in the U.S. and the British West Indies. Paper presented at the Tenth World Congress of Accounting Historians, St. Louis, MO, and Oxford, MS.

356

John Victor Singler

Gautier, Arlette (1985) Les soeurs de solitude: La condition féminine dans l’esclavage aux Antilles du XVIIe au XIXe siècle [Sisters in solitude: The feminine condition in slavery in the Antilles from the seventeenth to the nineteenth century]. Paris: Editions Caribéennes. Hall, Robert A. (1966) Pidgin and Creole Languages. Ithaca: Cornell University Press. Hancock, Ian F. (1986) The domestic hypothesis, diffusion and componentiality: An account of Atlantic anglophone creole origins. In: Pieter Muysken and Norval Smith (eds.) Substrata versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 71–102. Hesseling, Dirk Christian (1905) Het Negerhollands der Deense Antillen: Bijdrage tot de geschiedenis der Nederlandse taal in Amerika [The Negerhollands of the Danish Antilles: A contribution to the history of the Dutch language in the Americas]. Leiden: A. W. Sijthoff. Higman, B. W. (2000) The sugar revolution. The Economic History Review 53, 213–36. Huber, Magnus (1999) Atlantic English Creoles and the Lower Guinea Coast: A case against Afrogenesis. In: Baker and Bruyn (eds.), pp. 81–110. Hymes, Dell (ed.) (1971) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press. Jourdan, Christine (1985) Supos iumi mitim iumi: Urbanization and creolization of Solomon Islands Pijin. PhD dissertation, Research School of Pacific Studies, Australian National University. Jourdan, Christine (1994) Créolisation, urbanisation et identité aux Îles Salomon [Creolization, urbanization and identity in the Solomon Islands]. Journal de la Société des Océanistes 99, 177–86. Jourdan, Christine and Roger Keesing (1997) From Fisin to Pijin: Creolization in process in the Solomon Islands. Language in Society 26, 401–20. Kiple, Kenneth F. (1984) The Caribbean Slave: A Biological History. Cambridge: Cambridge University Press. Lacy, Ken N. (2001) Investigations of unusual similarities among Atlantic English Creoles. Paper presented at the Society for Pidgin and Creole Linguistics, Washington, DC. Lalla, Barbara and Jean D’Costa (1990) Language in Exile: Three Hundred Years of Jamaican Creole. Tuscaloosa: University of Alabama Press. Le Page, Robert B. (ed.) (1961) Creole Language Studies II: Proceedings of the Conference on Creole Language Studies, Mona, 1959. London/New York: Macmillan/St. Martin’s Press. Major, Kyle (2006) Rethinking shared features among Atlantic English Creoles. Unpublished manuscript, New York University. McWhorter, John H. (1998) Identifying the creole prototype: Vindicating a typological class. Language 74, 788–818. McWhorter, John H. (1999) The Afrogenesis hypothesis of plantation Creole origin. In: Baker and Bruyn (eds.), pp. 111–52. McWhorter, John H. (2000) The Missing Spanish Creoles: Recovering the Birth of Plantation Contact Languages. Berkeley: University of California Press. Milroy, Lesley (1987) Language and Social Networks, 2nd edn. Oxford: Blackwell. Mufwene, Salikoko S. (1986) Les langues créoles peuvent-elles être définies sans allusion à leur histoire? [Can creole languages be defined without reference to their history?] Études créoles 9, 135–50.

The Sociohistorical Context of Creole Genesis 357 Mufwene, Salikoko S. (1994) On decreolization: The case of Gullah. In: Marcyliena Morgan (ed.) Language and the Social Construction of Identity in Creole Situations. Los Angeles: UCLA Center for African American Studies, pp. 63–99. Mufwene, Salikoko S. (1996) The founder principle in creole genesis. Diachronica 13, 83–143. Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mufwene, Salikoko S. (2004) Language birth and death. Annual Review of Anthropology 33, 201–22. Mühlhäusler, Peter (1997) Pidgin and Creole Linguistics, 2nd edn. London: University of Westminster Press. Muysken, Pieter (1988) Are creoles a special type of language? In: Frederick J. Newmeyer (ed.) Linguistics: The Cambridge survey, Vol. II. Linguistic Theory: Extensions and Implications. Cambridge: Cambridge University Press, pp. 285–301. Parkvall, Mikael (1995) A Dual Approach to French Creole Genesis. MA thesis, Stockholm University. Parkvall, Mikael (2006) Was Haitian ever more like French? In: Ana Deumert and Stephanie Durrleman (eds.) Structure and Variation in Language Contact. Amsterdam: John Benjamins, pp. 315–35. Petit Jean Roget, Jacques (1980) La société d’habitation à la Martinique: Un demi siècle de formation, 1635–1685 [“Habitation” society in Martinique: A half-century of formation, 1635–1685] (2 vols.; Thèse, Paris VII, 1978.) Paris: Librairie H. Champion. Rickford, John R. (1987) Dimensions of a Creole Continuum: History, Texts, Linguistic Analysis of Guyanese Creole. Stanford: Stanford University Press. Roberts, Sarah (2000) Nativization and the genesis of Hawaiian Creole. In: John H. McWhorter (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 257–300. Roberts, Sarah (2005) The Emergence of Hawai[i Creole English in the Early 20th Century: The Sociohistorical Context of Creole Genesis. PhD dissertation, Stanford University. Sankoff, Gillian and Suzanne Laberge (1973) On the acquisition of native speakers by a language. Kivung 6, 32–7. Sheridan, Richard B. (1972) Africa and the Caribbean in the Atlantic Slave Trade. The American Historical Review 77, 15–35. Siegel, Jeff (2000) Substrate influence in Hawai[i Creole English. Language in Society 29, 197–236. Singler, John Victor (1986) Short note. Journal of Pidgin and Creole Languages 1, 141–5. Singler, John Victor (1988) The homogeneity of the substrate as a factor in pidgin/ creole genesis. Language 64, 27–51. Singler, John Victor (1990) On the use of sociohistorical criteria in the comparison of creoles. Linguistics 28, 645–69. Singler, John Victor (1992) Nativization and pidgin/creole genesis: A reply to Bickerton. Journal of Pidgin and Creole Languages 7, 319–33. Singler, John Victor (1993) African influence upon Afro-American language varieties: A consideration of sociohistorical factors. In: Salikoko S. Mufwene (ed.) Africanisms in Afro-American Language Varieties. Athens, GA: University of Georgia Press, pp. 235–53.

358

John Victor Singler

Singler, John Victor (1996) Theories of creole genesis, sociohistorical considerations, and the evaluation of evidence: The case of Haitian Creole and the Relexification Hypothesis. Journal of Pidgin and Creole Languages 11, 185–230. Singler, John Victor (1998) The higher the proportion of children in the population of color, the less radical the creole: What’s going on? Paper presented at the Biennial Meeting of the Society for Caribbean Linguistics, St. Lucia. Singler, John Victor (2006a) Children and creole genesis. Journal of Pidgin and Creole Languages 21, 158–74. Singler, John Victor (2006b) Yes, but not in the Caribbean. Journal of Pidgin and Creole Languages 21, 337–58. Thomason, Sarah G. (1995) Language mixture: Ordinary processes, extraordinary results. In: Carmen Silva-Corvalán (ed.) Spanish in Four Continents: Studies in Language Contact and Bilingualism. Washington, DC: Georgetown University Press, pp. 15–33. Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Tonkin, Elizabeth (1994) Engendering language difference. In: Pauline Burton, Ketuki Kushari Dyson, and Shirley Arderner (eds.) Bilingual Women: Anthropological Approaches to Second-Language Use. Oxford: Berg, pp. 186–92. Trudgill, Peter (2006) The role of dialect contact and the sociolinguistics of language history. Paper presented at Sociolinguistics Symposium 16, University of Limerick. Van Rossem, Cefas (1996) How Dutch is Negerhollands so? In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Relating to Grammaticalization in Contact Languages. London: University of Westminster Press, pp. 187–206. Vercoullie, Jozef (1919) Negerhollands molee, Afrikaans boetie, kartjipiering, bibies, bottel, ousanna, ewwa-trewwa, foolstuis. Tijdschrift voor Nederlandsche Taal- en Letterkunde 38, 302–6. Zelinsky, Wilbur (1992 [1973]) The Cultural Geography of the United States. Revised edn. (Originally published 1973.) Englewood Cliffs, NJ: Prentice-Hall.

The Cultural in Pidgin Genesis 359

15

The Cultural in Pidgin Genesis CHRISTINE JOURDAN

1 Introduction Of course, we could ignore the social anchoring of language, and pretend that languages appear, live, and die irrespective of the cultures that sustain them and of the people who speak them. The advantage of such a decision would be clear: the “noise” of culture would disappear. But the disadvantage would be just as clear: the “noise” of culture would disappear. And with that, any hopes of enriching our understanding of how human groups become “enlanguaged” would also disappear. For specialists of pidgin and creole languages, and more particularly for scholars looking at the genesis of these languages, this would be a terrible loss: The stories of this genesis are recent examples of how cultural groups become “enlanguaged,” of how human societies create for themselves the language that becomes the medium of their new cultural life. Being “enlanguaged” means being in the world (Heidegger 1962), being part of the cultural environment that sustains the symbolic systems of human beings. The languages that have become the focus of much of the specialist literature are the ones created in the context of the plantation societies of the Atlantic, the Indian Ocean, and the Pacific. In this chapter I will limit myself to the world of plantations in the Caribbean and the Pacific. In talking about the languages that emerged there, whether those customarily designated “creoles” or those customarily designated “pidgins,” I will – rather than relying on the hybrid term “pidgin/creole” – use “pidgin” as a cover term for all the languages under consideration. That I do so reflects my own attention over the years to the genesis and subsequent trajectory of pidgin in the Pacific. My emphasis in what I have to say here is on that which is common to plantation societies in the two areas, Caribbean and Pacific, and not on what distinguishes those in one region from those in the other. Stories of pidgin genesis are invariably stories of people (pidgin makers cum pidgin speakers) whose life circumstances have set the stage, created the need, and made it possible for a new pidgin to appear. But they are also the

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

360

Christine Jourdan

stories of cultural contacts, often dramatically so (even though this does not need to be the case, as the history of some pidgins attests), and therefore of cultural change, both radical and gradual, total and partial. And, obviously, they are also the stories of the power relationships that are at the core of the social worlds that have fostered most of the pidgin and creole languages we know. These are stories of people on the move (traders, merchants, explorers), of the commodification of – and commerce in – human beings, and of plantation societies. These are also stories of repeated negotiations and accommodations of meaning, as much as of the imposition of meaning, cultural and linguistic. In this essay, I propose to examine pidgin genesis from a cultural perspective, in the hope that some of anthropology’s central concepts may shed light on the events and processes that brought pidgins into being. I will offer a series of proposals regarding social interaction, power, the ideological dimension of culture, and about human agency as “matrix” (Alleyne 1971) of culture and pidgin genesis.1 Culture, the core concept of anthropology, will be our departure point, and with it, acculturation, contact, power, and meaning. All are germane to the social worlds associated with pidgin genesis, and to some of the concepts central to the theories of pidginization such as abrupt or gradual pidginization or dialect leveling. The social background on which my analysis focuses is that of the European plantations based on slavery or indentured labor, and I do not claim that the ideas I present here apply equally to all other contexts of pidgin genesis. Nor do they apply equally to all plantation societies for that matter: plantation societies in the Pacific were very different from what they were in the Atlantic and in the same locale, plantations certainly differed from one another in terms of work conditions, cultural diversity of the workers, and general social relations. My analysis does not apply either to all periods of African slavery on European plantations but rather focuses on the very early period (cf. Singler 1993a). Yet, as Mintz reminds us, the various instantiations of slavery were sufficiently different to warrant comparisons of them: And yet while the slavery of peoples of African origin in the Americas can for many purposes be treated as a single case, any attempt to carry the analysis further requires that comparisons be made: between and among different historical periods, different slave codes, and different colonies. For even though some common values and attitudes may have underlain the slavery institution throughout the New World, local circumstances varied widely (1974, p. 67).

Keeping the above in mind, this paper takes as its main premise that the birth of a new language goes hand in hand with the birth of the new culture of which it is becoming the main medium of communication. The paper focuses on the very early period of the genesis of pidgins, not from the perspective of language formation but from that of culture formation. One of its central questions is this: At what time in the history of the social context being considered

The Cultural in Pidgin Genesis 361 here can we say that a pidgin is born? Elements of the answer lie in the identification of the necessary cultural conditions for a pidgin to evolve. This essay starts with a discussion of the role of culture in pidgin genesis (Section 2), with particular attention to the centrality of work and work-related activities in promoting and shaping the development of pidgin (2.1); work is presented here as the common cultural denominator that provides pidgin makers with the necessary cultural matrix for the development of a pidgin. Section 3 presents a discussion of the role of social interactions in the development of linguistic and cultural meaning. An argument is made that pidgins are the result of a negotiation of linguistic meaning that takes place along a negotiation of cultural meaning. In Section 4, I visit the role of power relationships in the production of culture and focus on the place of human agency in the birth of a new language. Pidgin genesis is looked at as a form of resistance and empowerment. The paper concludes that the development of a pidgin is concomitant with that of the culture that sustains it. It is my hope that what I provide here will prove useful to researchers on pidgin genesis.

2 Culture Anthropologists have defined culture in different ways and disagree on how best to talk about it. To simplify matters we will say here that culture is a knowledge system that makes it possible to understand the world in which we live. No single individual is a true exemplar of the cultural group in which she/he lives. Rather, each of us is a repository of a small part of the knowledge system of that group. Cultural behavior is often taken as the visible dimension of culture, as the proof that culture exists. But some anthropologists will claim that culture does not equal behavior, but consists rather in a system of rules for understanding behavior. We will say that, whether we conceive of culture as a system of knowledge that is essentially cognitive as Goodenough (1971) and Keesing (1981) insist it is, or whether it is known only through its public manifestations as Geertz (1973) puts it, culture is produced by social relations that are in turn produced by it and by “structuring structures,” i.e., social institutions and established social practices (Bourdieu 1977, p. 72). We will add that culture is also the result of a constant negotiation and creation of symbolic meaning. In culture, nothing is given; everything has the potentiality of being challenged, revisited, redefined, created, and modified by a Bakhtinian dialogue (Bakhtin 1984) between individuals and groups that in turn transforms them. Central to this view is the role of human agency in the creation of culture (Foucault 1980): agency is what makes it possible for people to engage with culture, and to make it change. In this conceptual framework, culture is not understood as being an immutable entity that would remain impervious to contact with others and unaffected by it. Nor is it immune to the actions of its members. Culture is never pristine. It is always changing, although at times more than at others. Given individual agency, enculturation

362

Christine Jourdan

– the process by which adults transmit their culture to their children – is never totally successful. Culture is something that people have to work at. Each generation makes culture its own with the material at hand: that of their parents, that of their friends, that which they imagine and create. If the concept of culture is going to be heuristic for our understanding of pidgin genesis, one must therefore ask: What were the cultures of these social worlds? What kinds of social worlds were they? Historical, demographic, and socioeconomic research by Mintz (1974), Price (1983), Postma (1990), Singler (1993a, 1993b, 1995), Arends (1995), and Moitt (2001) for the Atlantic and the Caribbean; by Thompson (1975) and Galenson (1986) for English America; and by Corris (1973), Saunders (1974), Moore (1985), and Siegel (1987) for the Pacific shows that they shared many characteristics: • the labor force was in many cases forcibly removed from its home countries; • the ratio of men to women in the initial period was rather unequal on the Pacific plantations (see Corris 1973), more equal in other places such as Haiti and Martinique, as shown by Singler (1995); Suriname, by Postma (1990) and Arends (1995); and Guadeloupe, by Moitt (2001); • contact with the home country and the home language was maintained by a succession of cohorts of workers, so that parts of the culture of home were kept alive; • most of the social activities were initially essentially work-related activities (Mintz 1982); and • not all members of the home society were represented: for one thing, elders were absent initially as only young people were taken away, and families, if any, were often dispersed. In addition, and as is typical of other self-contained social worlds such as prisons, institutions, hospitals, and boarding schools, social life was restricted and took place within the structure, usually around activities that were defined from above (Foucault 1975). Rigid timetables, physical constraints and in some cases violence, psychological trauma, lack of privacy – all this negated the individual needs of the workers while seeking to transform them into pliable, if not well-adapted, members of the work community (cf. Thompson 1975; Morissey 1989; and Moitt 2001). In addition, as Thompson (1975, p. 38) remarks, some measure of isolation from the outside world was essential initially, and its effects were pervasive. To isolate is normally to continue previously formed habits and customs. But when a new settlement begins as a collection of individuals drawn from widely diverse backgrounds, as is generally true of plantation settlement, it becomes necessary to destroy in order to re-create. To break the wills and habits of others in the formation of a new institution, to redirect and maintain a new direction, requires a situation of isolation.

The Cultural in Pidgin Genesis 363 We can take the view that plantations functioned essentially as bounded social worlds, with a world view that was predominantly local. That was true for many of them. But we can also, as Arends (2001) does for Suriname for the period 1700–75, take the view that there existed contact and network relations between plantations. And that was certainly true for some plantation settings. Still, the degree to which workers on a plantation interacted with workers of neighboring plantations is still a matter of speculation, particularly for the early period of the plantation system, about which much historical work still needs to be done. But there were clearly situations in which workers had opportunities to establish links with workers from other places, through workrelated activities for instance. Arends lists a few: work, trade, leisure. The first two of these are particularly relevant to the arguments I am presenting here. We can surmise, though, that the nature and the intensity of contact probably varied through time. And of course, there was marronage, itself an extreme type of channel of exchange and diffusion of information among workers of different plantations. We can also surmise that the long work hours typical of slavery and indentured labor (Moore 1985; Morissey 1989; Moitt 2001) did not leave much room or much opportunity for the development of other types of group-based cultural relationships to be established among the laborers on the plantation initially. In short, plantations societies seem to have been truncated cultural units, linked and focused as they were to the economic exploitation of the land, in which workers found themselves locked into power relationships with overseers and owners on the one hand, and with fellow workers on the other. There was also limited time, and opportunity, for individual and collective symbolic expression, even though one expects (as will be shown below) that laborers retained their original cultural knowledge and kept whatever practices from home that they could keep in the new setting. At any rate, these workers were in no position to bring about changes in the structure of the system. Whatever form of locally produced culture emerged did so within the confines of that system. It is important to make a distinction between hegemonic structural conditions that serve as the locus of social life, and modes of life within these hegemonic structures that leave intersticial spaces, or “loopholes,” where people can engage with one another and can build on their exchanges. Importantly, and in addition to the trauma of slavery and/or indentured labor and isolation from the home place, the social world that the newly arrived workers once knew might seem to have had no immediate direct relevance for the new world they were thrown into. Indeed, if we look only on the surface of things, we might think that their knowledge systems, including the organizing categories of their languages, could not help them make sense of the new social world they experienced. Their cultural models (such as, for instance, religious beliefs, clan ideologies, social organization, and gender relations) would now have been at odds with the world they encountered. Yet if we examine the situation more clearly, we see that it is through the prism of their cultural knowledge that workers understood the new life they were

364

Christine Jourdan

thrown into, and were able to assess, understand, interpret, and engage with it. Culture is one of the frames through which we experience the world. For instance, in their article on the expectations that African slaves had of slave life in the Americas, Lovejoy and Trotman (2002) show that plantation workers in the Caribbean knew what slavery was back in Africa, and had expectations about this practice. However, they argue, the type of slavery workers encountered in the Americas, based as it was on racial categories, was very different from the African slavery they were aware of. Newly arrived workers reacted culturally: thus, whatever interpretations and meanings they were going to give to the new social world they encountered were done from the perspective of their own cultural experience, comparatively speaking, and from tapping their own cultural and individual skills. Individual and collective cultural interpretations probably took place quickly. On the one hand plantation societies were social worlds where various knowledge systems, those of the workers2 and those of the Europeans, met and confronted each other. On the other hand, it was also a space where they engaged each other: each group assessed and understood the other from its own cultural vantage point. Clearly, the new cultures that were progressively developing on the plantations were not starting in a vacuum. Even in the cases of dramatic displacement of people typical of the slave trade of the Atlantic or the indentured labour of the Pacific, the cultures that the workers brought with them – their vernacular cultures – were not totally lost, just as the languages they brought with them – their vernacular languages – were not totally lost. They remained alive through the continual arrival of new slaves or laborers and, most importantly, through peoples’ individual and sometimes collective memories. This made it possible for the more symbolic dimensions of culture to keep existing. Following on the work of Herskovits, Alleyne (1971) claims that the more symbolic dimensions of African social life such as religion, music, magic, and forms of amusement endured on plantations, while the more structural, such as technology and forms of political organization, did not. Singler (1993a, p. 210) talks about “continued Africanness.” Historical work done in Queensland by Corris (1973), Moore (1985), and Saunders (1974), shows clearly that in the initial period of their arrival on the Queensland plantations, starting around 1860, workers sought to keep alive some cultural dimensions of their home islands. Thus, Moore (1985), describing Solomon Islands indentured laborers in the town of Mackay in Queensland, writes about “Melanesian Mackay” and describes the co-existence there of Melanesian religious rituals and Christian beliefs. Overall, the environment and the social context dictated what workers could keep, of course, and what they could not. As far as possible, they planted small vegetable gardens the way they did at home, they propitiated ancestors and practiced divinations the way they did at home, and cooked food the way they did at home, roasted on the fire, or wrapped in banana leaves (Fatnowna 1989). Some men even kept men’s houses, where traditionally men secluded themselves away from women to hold rituals or simply to talk. But they could

The Cultural in Pidgin Genesis 365 not uphold the rules of marriage (the preferred form was clan exogamy) for the good reason that not all members of their societies were there and that membership in clans and lineages was difficult to reckon when clans and lineages themselves had ceased to be meaningful social units. Writing about African slaves in the New World, Mintz and Price (1992 [1976]) make it clear that, even if ideas about institutions could be maintained, the institutions themselves could not. Their analysis also holds for the Pacific. All these efforts made to maintain a link with the culture of the past, and to find a niche for it in the cultural world in which the slaves found themselves, are at the basis of many syncretic aspects of Caribbean and South American cultures today. Whatever new culture was likely to evolve among the slaves was bound to be a construct of the old cultural world that the workers knew, truncated and transformed as it were by displacement and reanalyzed in the light of the new cultural reality they discovered in situ. Again, in view of the cultural heterogeneity that often characterized many plantations, it may be appropriate to think of new cultural worlds, in the plural, rather than as a totalizing cultural world. But if these new cultural worlds drew part of their meanings from, and were shaped by, the ideology of slavery, they were also limited by the social practices of the plantation economy. Whether plantations developed their own particular culture to the extent that Thompson (1975) claims is a point of debate best left to historians. The new cultures that were emerging out of the cultural encounter had to have been limited in scope, at least initially, given the nature of the social space that served as their matrix. Still, whatever they were like, these cultures were the product of a group of workers who had “ceased to be a crowd of individuals and had become a community” (Mintz & Price 1992 [1976]). Not that it was a simple process: in many cases, multiculturalism (and attendant multilingualism among the workers) along with the structural organization of plantation life hindered for some time the transformation of such a group of individuals (and their individual responses to the new conditions of life) into a cultural community. One should be careful not to argue for a purely creativist approach to culture genesis, given that the resulting cultural formations owed more to reactive processes than to purely creative ones. Keeping the above in mind, we can identify four characteristics shared by plantation cultures in the early plantation period: (1) the cultures that were emerging were essentially centered around work or work-related activities – the control established over the laborers’ timetable and whereabouts ensured that hardly any time, or hardly any space, was left for socializing; (2) the cultures that were emerging were constrained by the lack of vernacular cultural depth in situ,3 by the structure of the plantation system, and by lack of access to European cultural depth, so that true acculturation did not take place; (3) the cultures that were emerging developed initially in relative isolation, in self-contained social and economic worlds, even though some measure of contact between different plantations and their plantation workers clearly existed, around sugar processing mills for instance; and (4) the cultures could

366

Christine Jourdan

emerge as such only when a “crowd of individual workers became a community of workers,” yet “community” should not be equated with “cultural homogeneity.” With these new cultural worlds appeared also new cultural subjectivities and identities that were grounded in specific practices workers had not experienced before. Writing about the role of agency and cultural logics in the transformation of subjectivity, Holland et al. propose: This is our objective here: to respect humans as social and cultural creatures and therefore bounded, yet to recognize the processes by which human collectives and individuals often move themselves – led by hope, desperation, or even playfulness, but certainly by no rational plan – from one set of socially and culturally formed subjectivities to another. (1998, p. 7)

2.1

Cultural worlds centered on work

What is the relationship between these new cultural formations and the emergence of pidgins? A look at work and work-related activities is crucial to the answer to this question. In addition to being the economic “raison d’être” of plantations (or mines or trading posts), work and work-related activities also provided the only social context where regular, repeated, and sometimes prolonged communication could take place between workers and plantation overseers and owners (be they “petits blancs” in Haiti and Martinique, or “mulatto” intermediaries), on the one hand, and among laborers themselves. This is true also of other types of social contexts that fostered the development of pidgins elsewhere or the restructuring of languages such as Sango (Samarin 1982) and Katanga (Shaba) Swahili (Fabian 1986). Part of the argument I am making here involves paying more attention to cultures as barriers to, and facilitators of, communication. If a local culture is likely to develop, it will be out of a cultural space shared by workers on a regular basis. As we have seen above, workers shared some measure of expectations of their new circumstances, their experience of cultural and spatial dislocation. But they shared also their experience of work conditions in a plantation setting: the same pressures from a rigid timetable and the long hours of work;4 the same hardship on the body; and the same relationship, or lack thereof, with Europeans. Work was the common space, the common denominator that opened the door for exchange, cultural and linguistic; it provided the necessary conditions for the creation of a pidgin. The culture of work became the locus for the exploration of meaning, through trial and error. Be it in the fields or in the sugar mills, it was also the cognitive center of the plantation community around which much of the meaningful daily social life initially revolved. By its very nature (physical activities, gestures, movements, planning), work is an ideal locus for the birth of vocabulary. Regular and sustained contact between peoples, including workers and overseers, made transfer of technical and practical vocabulary possible and easy, whatever the occupations of the laborers in the plantation economy.

The Cultural in Pidgin Genesis 367 And, by its very nature, work fostered the development of communicative collaboration: without collaboration, how could one interpret an order, learn to execute what was expected, explain to others how to perform tasks, coordinate actions so that the work could be done in a speedier (or slower?) fashion? Work provided the initial social space where individual identity could be reshaped and group consciousness could develop: all workers were defined in relation to it, and the hierarchical nature of the plantation social world came to the fore in those moments and activities. With the above in mind, one can propose that it is around work-related activities that pidgin varieties initially developed in a given plantation, and that from that cultural sphere they then spread to the other cultural spheres of plantation social life, however limited they may have been. This process does not mean that there existed a long lag between the time of pidgin creation in work-related settings and its expansion/application to other settings. Nor does it mean that no collective or personal activities outside of work eventually developed that did not also favor the development of pidgin (such activities as cooking, sharing meals, the practice of daily life, establishing friendship links, seeking support, and the like). The incipient pidgin progressively became central to the social life of work, where most intergroup social interactions took place, while the vernaculars remained all the while rooted in the cultural depth associated with the places of origin. Thus, while pidgin developed in response to, or in assocation with, local sociolinguistic conditions and demands (multilingualism, necessity, pragmatism, and group consciousness, among others), the vernaculars were central to the cultural memory of the workers, to those vernacular cultural practices that they were able to maintain, and to the individual and personal contact that workers may have had with members of their ethnic groups. They also remained vital to any cultural consciousness anchored in the past and certainly facilitated the psychological and emotional transition into the present for those workers fortunate enough to find on the plantation people from their own ethnic group. As frontier communities plantations have always been places in which populations of differing ethnicity and worldview confronted each other . . . As contact communities – settings of ethnic and “racial” encounter – plantations have provided rather rigid, sometimes oppressive, contexts within which newcomers had to work out a different style of life. (Mintz 1974, p. 53).

By virtue of the co-existence of these cultural worlds, the incipient pidgins and the workers’ vernaculars also co-existed. To sum up, in accordance with historical evidence, at the time of pidgin genesis, plantations were the loci of several co-existing cultural worlds: (1) truncated European cultures and social worlds sustained by the superstrate language and its social dialects; (2) truncated vernacular cultures of the workers, sustained by their respective vernaculars, and lived and experienced by individuals who became progressively removed from them; and (3) a culture of

368

Christine Jourdan

plantation work sustained by an incipient pidgin, which was progressively developing and stabilizing, adopting structural features and interpreting symbolic dimensions from the European culture(s), transferring features from the vernacular cultures, and building on both. But what about pidgins? To conclude, and this is only an educated guess, but arguably a very reasonable one, I propose that work offered the unifying leveling cultural context that made it possible for pidgin to develop (or to stabilize in cases where workers had picked up some smattering of pidgin on board slave ships or other types of ships, as Keesing (1988) reports for the Pacific for instance), despite the co-existence of partial “vernacular” cultures. Just as the vernacular cultures shaped the cultural worlds that were developing within the plantation’s social structure,5 the vernacular languages shaped the incipient pidgin varieties that were developing on each plantation.

3

Meaning

Meaning, linguistic or cultural, is not given. It is the product of social relationships and social interactions, the results of cultural debates and negotiations, and in the end the result of the development or imposition of what functions as a consensus. Consensus is a big word here and one should be wary of assuming that consensus equals universal acceptance of the choices that have been made. Following on Habermas (1975), for consensus to exist there is a need for two parties to establish communication: those who are establishing the issue around which consensus is to be built, and those who engage in the process of consensus building, and whose passive or active acceptance of this process is essential to its success. It is this consensus that in turn serves as frame of reference for further social interactions and which, paradoxically, ensures that parties can disagree on the readings of events. In this light, cultural and linguistic meaning are therefore best understood as negotiation (McWhorter 1992 talks about compromise) and as collaboration. We shall return to this point later. On the other hand, how meaning emerges is often linked to the interpretations people make of cultural phenomena, interpretations that are mediated by their own experiences. The interpretation is, as Taylor reminds us, “la tentative de rendre clair l’objet étudié, de lui donner un sens” (1997, p. 137).6 In situations of cultural contact, a great deal of the initial contact period is spent trying to understand the behavior of the “Other” and trying to render it meaningful. Interpretation of the “Other” is not limited to situations of cultural contact with an exotic other or to radical otherness: we would be right to consider the ongoing socialization that characterizes human experience as a continual effort of interpretation of yet new contexts and of new “Others.” As was explained above, not all aspects of vernacular cultural domains, and not all aspects of European cultural domains, were present on plantations (or in the other social contexts that fostered the development of pidgins, even

The Cultural in Pidgin Genesis 369 though we are not talking about these explicitly here). As a result, many semantic domains having to do with cultural knowledge and practice did not translate either. Given that the workers came to the cultural encounter with the European-ruled plantation world from the perspectives of their vernacular cultures, they came to realize that many of their own semantic categories were inapplicable. This, in my view explains in part why, even with widespread calquing (Keesing 1988) and relexification (Lefebvre 1999), pidgins are not totally similar to their substrate languages. Thus, Solomon Pijin is not Kwaio with English words (Keesing 1988), and Haitian Creole is not Fongbe with French words (Lefebvre 1999). I do not mean to suggest that these authors would want us to believe that the new languages are exact copies of the old ones. They, too, are aware that in the transition between the home world of the workers and the world of the plantations, too many cultural changes had taken place for whole semantic domains and lexical categories to be systematically transposed from vernaculars to the incipient pidgins. In addition, new cultural phenomena developed locally that warranted the creation in the pidgin of new lexical items that were not present in the vernaculars. It can be emphasized that pidgins are not simply the results of second language acquisition processes combined with the effects of Universal Grammar, or the result of direct calquing and relexification. Such a scenario would deny the possibility of the workers’ own imprint, their own creative stamp, on the new language they were in the process of “developing.” Pidgins are the results of a process of cultural translation inherent to all instances of contact situations (see Jourdan 2001). And in these types of translations, cultural reanalysis and interpretation are just as important to the language that is being created as is linguistic reanalysis. This is so in particular when the cultures in contact do not share the same type of cultural categories (a situation typical of the colonial encounter). Cultural interpretations and cultural translations also imply recreation: the resulting forms are not copies, but appropriations of meaning and a re-casting of meaning in different terms. They are driven and made possible by individual agency and collective group consciousness. These propositions are general in that they do not preclude variation. Indeed, a cultural interpretation model of pidgin genesis makes it clear that varieties of the same pidgin can co-exist in the same place, at the time of genesis, and after. As we very well know, the presence of variation and the presence of dialects do not prevent efficient communication from taking place. Each speaker accommodates to the variety he or she is exposed to. In the following, and using an example taken from the kinship terminology of Solomon Islands Pijin, I will illustrate the effect that cultural reanalysis may have on Pijin lexicon and cultural categories. A careful comparative analysis of a corpus of kinship terms from a group of vernacular languages from the Solomon Islands with kinship terms from Solomon Island Pijin (cf. Jourdan 2000) reveals that the semantic categories of Solomon Islands Pijin kinship terminology are not identical to those of the vernaculars. Rather, they are closer to what is known in anthropology as the

370

Christine Jourdan

Hawaiian system. The resemblance is a surprise in some ways as none of the vernacular kinship systems found in the corpus displays typical Hawaiiantype kinship terminology. In such a system, all members in the kinship group of the same generation are referred to or addressed by the same label (all women of the generation of the mother will be addressed as “mother,” all boys of one’s generation will be addressed as “brother,” etc.); there is no distinction between members of the nuclear family and collaterals;7 and there exist reciprocal terms of address for grandparents and grand children, and for aunts/uncles and nieces/nephews. These features also exist in Solomon Island vernaculars in general, but other features such as a distinction between elder and younger sibling, cross and parallel cousins,8 etc., are also present and render these kinship systems more complex. The resemblance of Pijin kinship terminology to Hawaiian kinship terminology raises an interesting question with regard to the genesis of Melanesian Pidgin, and of its daughter language Solomon Islands Pijin. If relexification, or calquing, is to explain the semantic transposition of vernacular terms into Pijin, one wonders why Pijin terminology has not followed the kin category labels found in the vernaculars. The probable reason for this lack of fit is linked to the cultural dimension of pidgin usage at the time of the development of Melanesian Pidgin. One needs to remember that the cultural world where this pidgin stabilized and became dominant (the plantations of Queensland) did not justify the precise labeling of kin that the vernaculars have. For one thing, the culture that sustained this new language was radically different from vernacular culture (Moore 1985), and not all members of the kin groups were present. Second, vernacular languages existed in Queensland in parallel to the developing pidgin. Paradoxically, it is the co-existence of vernaculars and pidgin in the pidgin-speaking cultural world that makes the lack of fit possible. Be they from Vanuatu or the Solomon Islands, the vernaculars were used among wantoks9 whenever important cultural matters were discussed and precise cultural terms were needed (Fatnowna 1989). Kinship is an important cultural matter in Melanesian societies, structured as they are around clans and lineages: it organizes the social life of people, gives access to land, and regulates exchange obligations. On the plantations of Queensland, the social make-up was such that there was no need for the pidgin to label cultural categories of kin that were not relevant to the new social life. Labels such as cross cousin or parallel cousin, elder brother or elder sister, maternal uncle, all of them typical of Melanesian kinship terminology, were generally irrelevant when most of the people who could have been referred to or addressed by such terms were not present. When a need to refer to one of them did arise, then the vernacular term could be used in private conversation. A generic label from the pidgin could also be used to refer to these absent kin, and noun qualifiers could be added to general pidgin kin terms, so that gender and other features could be expressed. This linguistic strategy is still used in contemporary Solomon Islands Pjin, and we can expect that it held in earlier times as well.

The Cultural in Pidgin Genesis 371 Looking at Pijin kinship data, one observes that a selection and retention has been made of only those labels that had cultural resonance in the new social context, and were shared by all indentured workers concerned. The resulting system is stripped down from the labels that have become culturally irrelevant, but is nevertheless in continuity with the vernacular kinship system in that it has kept its most salient features, such as contrast between generations. The system is also totally different from the kinship system of English, its lexifier, which typically distinguishes members of the nuclear family from other kin. English labels have been borrowed directly when they fit the vernacular kinship categories (dadi, mami, brata, sista), but have been also reanalyzed so that they can be mapped on to the vernacular system: grani is reciprocal for grandparent and grandchild (and it applies irrespective of gender), a feature that does not exist in English kinship terminology. Mami refers to one’s mother and to all women in the generation of one’s mother to whom one has kinship ties, be they related by blood or marriage. The system applies in the same way to dadi and likewise to sista and brata. Cultural reanalysis has led to the development of a terminology that is neither English nor vernacular, but is in continuity with the cultural logic of the latter, and yet makes allowances for the social conditions of enunciation and use. Some streamlining of the most culture-heavy terms took place and pre-emptive leveling in the speech community (see the section 4 on Power below) led to a selection of the more regular and universal semantic categories of vernacular kinship terminology.10 Another point is worth mentioning. Linguistically, the categories are simple, but the underlying cultural cognitive categories may remain as complex as before, thus making it necessary for people to be more precise when circumstances require it. This observation leads to another, one that is directly linked to the coexistence of Pijin and vernaculars in Queensland: the longer the co-existence of these languages, the less there was a need for the transfer of cultural categories from the vernacular world to that of the pidgin speaking world, particularly if the culture that was in the process of being developed bore little resemblance to that sustained by the vernaculars. Reciprocally, the more the cultural importance of the vernaculars diminished in the light of an expanding plantation culture based on pijin, the more the pijin lexicon expanded into other semantic fields required by a developing new culture. Those tokens that were not relevant to the new cultural world did not make it into the new language. Obviously, semantics is a good place to start looking at the congruence of old meaning systems with new ones. As we are dealing with the transferability of semantic features across cultures, we need to remind ourselves of the tight bonds that link languages to the cultures that sustain them. Cultural contacts produce cultural reanalyses and attendant linguistic reanalyses: people actively seek to interpret the cultural situation with which they are faced in light of what they know, and in light of what they need, starting with the language they know.

372

4

Christine Jourdan

Power

Scholars of language ideology have shown forcefully that language is not a socially neutral tool (among others see Bourdieu 1975, and more recently Woolard & Schieffelin 1994, and Schieffelin, Woolard, & Kroskrity 1998). The production of speech is itself power-producing, in that it is reality-producing: speech takes place within social relationships that are constructed or reinforced through it. Through language, speakers engage culture, shape it, and are shaped by it. In some sense, culture is the condition of enunciation, the context within which another layer of meaning is added to speech. Building on the work of Foucault (1980), anthropology has recognized that culture, as an ideological system resting on the meaning of symbols, is the locus par excellence of power relationships. So, likewise, is language, inasmuch as we accept that the production of language is a social reality and not simply a construction of the mind. In the case that concerns us here, the birth of a new language cannot be separated from the social forces that produce it. Such forces involve power relationships typical of the colonial worlds: between the workers and their masters, and among workers themselves vying for whatever distinction could bring recognition and possible advantages. In addition, power relationships on plantations were couched in terms of race relations, with the result that the discourse on race became central to the discourse on language within colonial relationships. It should not come as a surprise that the languages created in these circumstances would be gathered into, and confined by, a label such as pidgin, the meaning of which would be associated by observers (and sometimes speakers) with inadequacy. In its attention to the conditions of multilingualism and to the nature/amount of access to the superstrate obtained by vernacular speakers on plantation, some of the pidgin and creole literature has shown that it takes the power argument seriously. It also takes seriously the power to be found in numbers: the number of people involved at the time of genesis, the types of languages that were spoken, careful demographic analyses of gender ratio, age pyramids, and size of cohorts of slaves have proved essential to our understanding of pidgin genesis; in several cases, this research has considerably altered the views that were prevalent in the discipline until then (see Arends 1993; Baker & Corne 1982; Singler 1993a, 1993b). Certainly the presence of structural conditions left laborers little room for cultural and linguistic maneuvering, particularly in the early phases of the labor trade to the Americas. Still, once they were past the initial shock caused by their new life conditions, the laborers can be expected not to have remained passive for long. Human agency is such that “non-places” are eventually transformed into cultural places, however limited in number, scope, and type the possible forms of cultural expression. This transformation occurs within the

The Cultural in Pidgin Genesis 373 confines of the system in which people find themselves, using what it offers, with individuals adding their own imprint to change things, exploit cultural loopholes, or take whatever cultural space can be taken, sometimes in the form of covert resistance, even in the form of language. A power-centered approach to the genesis of pidgin is not satisfied with an explanation of their birth that would focus solely on language as a tool of communication. Granted, cultural and linguistic communication gaps had to be bridged. And the emergent pidgin certainly played this role. But we should ask ourselves whether the emergence of pidgins might be construed as a form of resistance to hegemony. In some ways, certainly. For not only did pidgins allow for communication to take place among workers and cause the breaking down of their linguistic isolation, but the developing pidgins also allowed for the crystallization of a nascent cultural identity, different from, and yet in relative continuity with, that of the European cultures found locally and that of the vernacular cultures of the workers. Resistance is to be understood also as an engagement with the conditions of domination, and their transformations, and not only as refusal of this engagement. Pidgins as an avenue to empowerment? Why not? Whether the workers wanted it or not, whether they were conscious of it or not (and it is most likely that they were not conscious of it, particularly if by consciousness we mean deliberateness), the development of pidgins proved to be one of the most important symbolic and pragmatic actions that they could undertake. It was the first step toward empowerment, the first step toward a subversion of hegemony, the true product of cultural agency. I am not suggesting here that pidgins developed only out of a deliberate desire on the part of workers to challenge the hegemonic conditions that controlled some aspects of their life. Yet, to speak is to create; it is to represent; it means establishing a link between one’s self and the world; it is to live. Speaking about the centrality of language in the history of humanity, Foucault (1977, p. 53) states “speaking so as not to die is a task undoubtedly as old as the word.” I could not agree more. To speak a word is to appropriate its meaning, and its power. By virtue of having a new language at their disposal that was in continuity with the vernaculars and made use of significant elements of the superstrate, pidgin makers set themselves on a course of linguistic independence that changed their relationships to the world and shaped their own identity. I am reminded here of Baker’s argument (1995) on motivation in creole genesis: it falls squarely in the analysis I am presenting, even though I am not prepared to go as far as he does on the degree of deliberate social consciousness that he attributes to pidgin makers in the process of pidgin genesis. Resistance to hegemony is not limited to the use of force; it often has to do with seizing whatever space is devoid of controls and claiming it. In this case, the void was linguistic. All the while, pidgin makers were developing a new cultural identity that could now be expressed in words. Bourdieu’s structuring structures are useful again to make us understand how practices and representations (here linguistic)

374

Christine Jourdan

can be objectively regulated and regular without in any way being the product of obedience to rules, objectively adapted to their goal without presupposing a conscious aiming at trends or an express mastery of the operations necessary to attain them and, being all this, collectively orchestrated without being the product of the orchestrating action of a conductor. (Bourdieu 1997, p. 72)

Except for some notable exceptions (Arends 2001 and Siegel 1997 among the most recent ones), symbolic power seems to have been ignored in our studies, and so has the power derived from knowledge. Yet one can easily imagine the prestige obtained by the first slaves who seemed to be able to decode some of the meanings of the world they were thrown into and could interpret it for others. Arends’s example of the place of the black overseer, the basya, in Surinamese plantation is a case in point: “Combined with his knowledge of Afro-Surinamese culture, this gave the basya a degree of power among the slaves which sometimes may have surpassed that of the white overseer” (2001, p. 297). The prestige derived from this knowledge was invaluable: whether it was derived from regular and sustained access to this world, whether it was inferred from observation or from direct and regular contact with superstrate culture, it certainly gave some individuals a degree of prestige that conferred upon them some measure of power and influence. (Consider, for example, the difference in access and contact between house slaves on a Caribbean plantation and field slaves.) We can surmise that the smaller the slave’s speech community, the greater would have been the importance of such individuals in the shaping of pidgin, but as Arends reminds us, this importance was probably mitigated by the networks to which such individuals had access. However, in the scenarios that have been built about pidgin genesis, it may be the case that too much attention has been paid to conflicting power relationships and not enough to the types of collaborative behavior that may have also existed alongside the conflictual ones. Collaborative behavior was probably just as important; in many cases it would have been the key to physical survival and psychological health. Paradoxically, one place where collaboration is certainly needed is in the development of a new language: there it must have taken different forms, and probably worked at various levels. A basic question concerns the diverse loci where collaboration is needed in any situation of communication – collaboration in the interpretation of meanings, in letting people talk, in proposing words for ideas, in understanding the intention of the interlocutors, in allowing phonological and syntactic variations, in conceding that one’s choice of words become secondary or even eliminated, and so forth. Communication is a story of concessions and collaboration as well as of turn-taking, stealing the floor, and shouting matches. And certainly communication is what made it possible for social and linguistic accommodation to take place. Along these lines, we can consider foreigner talk as a form of linguistic collaboration, as much as an expression of power on the part of speakers. We can also look at the social principles guiding language mixing or koineization (see Siegel 1997) as another form of cooperation: if efficient

The Cultural in Pidgin Genesis 375 communication is what guides the selection and use of linguistic features, then, mutatis mutandi, those factors that promote efficiency are more likely to appear in the new language. In his brilliant article on the principles that guide mixing and leveling, Siegel (1997) shows that the relevant linguistic factors are unmarkedness, transparency, regularity, frequency, economy. One of them, salience, can be construed in linguistic terms but also in symbolic ones. Of the many phases and principles involved in pidgin genesis, dialect leveling is certainly one of those that are the result of a combination of collaborative and non-collaborative power-related behaviors among pidgin makers. If we conceive of dialect leveling not only as a step toward efficient communication but also as a direct product of cultural jelling, i.e., the progressive development of cultural patterns and consciousness, it ensues that some measure of collaboration must have played a role in it, alongside the more conflictual and less consensual leveling. The latter is usually the product of social forces at play, such as the progressive adoption of one linguistic form over another by the linguistic community because of the sheer number of speakers using this form or because of hegemonic forces that shape the direction of linguistic change. And what is dialect leveling if not a form of linguistic change, a linguistic harmonization with the social pressures and cultural values at play? The literature dealing with leveling seems to indicate that it follows upon calquing, relexification, and reanalysis, in other words after individual varieties of pidgins have already appeared (for instance Keesing 1988, Lefebvre 2001). Siegel has emphasized the causal relationship between community formation and leveling: “When the social context leads to unification of the community and vernacularization of the newly developing variety, then leveling occurs” (Siegel 1997, pp. 143–4). And more recently, “When the community begins to shift from their L1 to the contact variety, leveling occurs” (Siegel 2003, p. 187). I agree with him that this is what must have happened, just as leveling keeps on taking place in all speech communities today. But I think that we must also entertain other possibilities. First, we do not need to think that leveling always takes place after the fact, as a form of correction. We would gain much understanding of leveling if we made room for some form of pre-emptive leveling that would be a response to cultural constraints, social pressures, and demands of communication at the time when individual pidgin makers/speakers start experimenting with communication in the medium they are in the process of developing. (For arguments on the “making” or the “creating” of pidgins, see Jourdan (1991) and Baker (1994).) All the principles that guide mixing also guide leveling (Siegel 1997). But an important factor, efficiency, talks to the speakers’ agency. What do I need to do in order to be understood? What do I need to do in order to understand? For this to happen, pidgin makers have to position themselves in the same communication frame and in the same temporality of intention as their interlocutors. This is what Fabian (1983) calls coevalness. Another important factor for leveling is the psychological context of enunciation: speakers’ individual needs for individual recognition, for belonging,

376

Christine Jourdan

for social positioning. These are needs that are made even stronger by situations of liminality. Fulfilling them may start with a search for conformity or difference, cultural and linguistic, and may lead individuals to quickly monitor their own speech in relation to that of their interlocutors. In that perspective, leveling not only rides on the coat-tails of community formation, as it certainly does, but also contributes to the formation of the speech community. As Trudgill (1986) and Siegel (1997) make clear, leveling does not imply a resulting overarching homogenization. While some form of convergent dialect emerges out of everyday communication within the speech community in association with vernacularization, this does not result in the erasure of co-existing varieties. Variation is still present in the speaker’s mental lexicon and grammar, and it is likely to resurface in different communicative events, and in cultural contexts where the speaker’s speech may be given more space or accorded more legitimacy.

5

Conclusion

In this essay I have tried to show the relevance of the anthropological concepts of culture, meaning, and power to our understanding of pidgin genesis. I did not try to solve problems, but rather to raise questions. Focusing on plantation settings, and starting with culture, I argued that the development of pidgins was concomitant with the development of local cultures. In doing so, I also argued that individual agency drives cultural reanalysis and cultural creation. Faced with a social situation on the structure of which they seemingly had no control and using the tools they had, the workers – slaves or indentured laborers – engaged the social world that controlled their life and put their imprint, linguistic and cultural, on it. The point of departure of these new linguistic cultures is the culture of work that developed on the various plantations. Just as the new cultures are the results of negotiations of meaning that start with the individual, so are the new pidgins. Using the concept of meaning as a reference point, I proposed that, when cultural worlds are in contact, as they were on the plantation, individuals and groups need to accommodate to what we now call in anthropology “otherness.” The making of the “Other” involves the interpretation of the “Other.” This cultural interpretation takes place in the light of one’s own cultural and personal experience, and within the ideological confines that govern cultural life and social relationships. This often results in different cultural outlooks that individuals have on their new cultural world. Focusing on Solomon Islands Pijin kinship terminology, I showed that when contact takes place between cultural groups who do not share the same type of cultural categories and dominant ideologies, not all original cultural domains are re-interpreted: only those that are relevant to the local conditions are identified as salient and are “negotiated.” The transferability of substrate and superstrate semantic features to the pidgins that are emerging in these social conditions is directly linked to the cultural interpretations made by pidgin makers.

The Cultural in Pidgin Genesis 377 Finally, a discussion on the dialogical nature of power allows for different analyses of the social relationships likely to foster pidgin genesis. In addition to the conflictual relationships archetypal of colonial worlds, more consensual relationships are also necessary for pidgins to appear. In situations of cultural alienation or cultural liminality, the creation of a new medium of communication can be seen as a form of resistance to hegemonic social conditions, as much as an expression of identity. In this light the birth of a pidgin language is as much the result of the pragmatic need to break cultural and linguistic isolation as it is a form of empowerment on the part of its makers. Just as cultures develop over time, so do the pidgin languages that become their linguistic medium. Languages cannot exist without the cultures that sustain them, and they cannot develop before the cultures that sustain them develop; the two go hand in hand, in a form of constant feedback, through which social groups become encultured and enlanguaged.

NOTES I am grateful to Rachel Selbach, Jeff Siegel, and Kevin Tuite for comments on an earlier draft of this paper, and to the editors for very helpful suggestions along the way. Shortcomings, of course, are mine alone. 1 It is interesting to note that one can find correspondences between some contemporary linguistic theories of pidgin genesis and anthropological explanations of culture change. For instance, the linguistic theories that emphasize relexification, transfer, and calquing of substrate features in the new pidgin find an echo in a model of acculturation that was prevalent in 1950s American anthropology to explain the African features of Caribbean societies, and is best represented by the works of Melville Herskovits (1941). This model sought to identify the residual African cultural features of Caribbean societies and fell under criticism because it failed to account for symbolic salience. The theories that emphasize the role of the superstrate in pidgin genesis find an echo in theories seeking to explain the role of hegemonic forces in cultural reproduction and economic development in contact situations linked to colonization. This last model leaves very little room for individual agency and is best represented by the works of Wallerstein (1974). Finally it is worth noting that some anthropologists are looking toward the theories of pidginization and creolization to understand social cultural change in societies in rapid social flux. The main proponent of this approach is Ulf Hannerz (1987). 2 We need to recall that the knowledge systems of the slaves, either in the Atlantic or in the Pacific, were heterogeneous. The slaves came from different cultural backgrounds, and, even though generalities could be established, these backgrounds were different, thereby creating de facto multiculturalism in single plantations. See also Mintz and Price (1992 [1976], pp. 14–15). 3 Cultural depth refers to the body of cultural rooting over at least a few generations of practices and ideologies. It is through cultural depth that languages acquire social legitimacy. See also Jourdan (1996). 4 Moore (1985, p. 123) notes that Melanesians in Mackay (Queensland) worked “under supervision in open fields for twelve hours a day, six and a half days a

378

5

6 7

8

9

10

Christine Jourdan week.” Moitt (2001) notes that in the French Caribbean the Code Noir of 1685 placed no limits on a slave’s workday, but prohibited work on Sundays and other Christian holidays. In his article on family structure and plantation systems in the western hemisphere, Smith (1959) also makes a distinction between social relationships linked to the production process (work, here), and other relationships: “I therefore assume that the involvement of the plantation population in systems of social relationships other than those involved in the process of production must be considered capable of variation from one plantation to another in a way that the relations involved in the process of production are not” (p. 149). ‘the attempt to clarify what one studies, to give it meaning’. Collaterals are relatives on both sides of the parents’ family and include aunts and uncles, and their descendants. In a Hawaiian system, all female relatives belonging to the same generation as one’s mother would be addressed by the term “mother,” and all male relatives belonging to the same generation as one’s father would be addressed by the term “father.” Grandparents and grandchildren would address each other by the term “granny.” “Cross cousins” are the cousins born to the brother of one’s mother and the sister of one’s father; “parallel cousins” are born to the sister of one’s mother and to the brother of one’s father. These distinctions are crucial in societies that allow crosscousin marriage but prohibit parallel-cousin marriage. Wantok is a Pijin word from English one talk. It is literally someone with whom one shares the same language. An expansion of the semantic field includes also “members of one’s village,” “members of one’s clan”, “friend,” and – more rarely – any person to whom one is linked in a particular way. This explanation is reinforced by observations made in some parts of the Solomon Islands (P. Maranda, p.c.) and Papua New Guinea (E. Schwimmer, p.c.) where individuals perform some unballasting of the vernacular cultural categories in two particular instances: between people who entertain good relationships; and when distance away from the cultural center of reference (usually the home village) removes the constraints on individuals and the need for them to respect the prescribed kinship labels. In addition, distance creates a form of desacralization of the categories, and thus makes it possible for them to be dropped.

REFERENCES Alleyne, Mervyn (1971) Acculturation and the cultural matrix of creolization. In: Dell Hymes (ed.) Pidginization and Creolization of Languages. Cambridge: Cambridge University Press, pp. 169–86. Arends, Jacques (1993) Towards a gradualist model of creolization. In: Francis Byrne and John Holm (eds.) Atlantic Meets Pacific: A Global View of Pidginization and Creolization. Amsterdam: John Benjamins, pp. 371–80. Arends, Jacques (1995) Demographic factors in the formation of Sranan. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 233–85. Arends, Jacques (2001) Social stratification and network relations in the formation of Sranan. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 293–307.

The Cultural in Pidgin Genesis 379 Baker, Philip (1994) Creativity in creole genesis. In: Dany Adone and Ingo Plag (eds.) Creolization and Language Change. Tübingen: Max Niemeyer, pp. 65–84. Baker, Philip (1995) Motivation in creole genesis. In: Philip Baker (ed.) From Contact to Creole and Beyond. London: University of Westminster Press, pp. 3–15. Baker, Philip and Chris Corne (1982) Isle de France Creole: Affinities and Origins. Ann Arbor: Karoma. Bakhtin, Mikhail (1984) Esthétique de la création verbale [Aesthetics in verbal arts]. Translated from the Russian by Alfreda Aucouturier. Paris: Gallimard. Bourdieu, Pierre (1975) Le fétichisme de la langue [Linguistic fetishism]. Actes de la recherche en sciences sociales 4, 2–32. Bourdieu, Pierre (1977) Outline of a Theory of Practice. Cambridge: Cambridge University Press. Corris, Peter (1973) Passage, Port and Plantation: A History of Solomon Islands Migration, 1870–1940. Melbourne: Melbourne University Press. Fabian, Johannes (1983) Time and the Other: How Anthropology Makes Its Object. New York: Columbia University Press. Fabian, Johannes (1986) Language and Colonial Power. Berkeley: University of California Press. Fatnowna, Noël (1989) Fragments of a Lost Heritage. North Ryde, Australia: Angus and Robertson Ltd. Foucault, Michel (1975) Surveiller et punir: Naissance de la prison [Discipline and punishment: The birth of the prison]. Paris: Gallimard. Foucault, Michel (1977) Language, Counter-memory, Practice: Selected Essays and Interviews by Michel Foucault. Edited by Donald Bouchard. Ithaca: Cornell University Press. Foucault, Michel (1980) Power and Knowledge. New York: Pantheon Books. Galenson, David (1986) Traders, Planters and Slaves: Market Behaviour in Early English America. Cambridge: Cambridge University Press. Geertz, Clifford (1973) The Interpretation of Culture. New York: Basic Books. Goodenough, Ward (1971) Culture, Language and Society. Reading, MA: Addison-Wesley. Habermas, Jürgen (1975) The Legitimation Crisis. Boston, MA: Beacon Press. Hannerz, Ulf (1987) The world in creolization. Africa 57, 546–59. Heidegger, Martin (1962) Being in the World. New York: Harper & Row. Herskovits, Melville (1941) The Myth of the Negro Past. New York: Harper & Brothers. Holland, Dorothy, William Lachicotte, Jr., Debra Skinner, and Carole Cain (1998) Identity and Agency in Cultural Worlds. Cambridge, MA: Harvard University Press. Jourdan, Christine (1991) Pidgins and creoles: The blurring of categories. Annual Review of Anthropology 20, 187–209. Jourdan, Christine (1996) Legitimacy of Solomon Island Pijin. Anthropological Notebooks 2 (special issue on Multiple Identities, ed. Borut Telban), 43–54. Jourdan, Christine (2000) “My nephew is my aunt”: Features and transformations of kinship terminology in Solomon Islands Pijin. In: Jeff Siegel (ed.) Processes of Language Contact. Montreal: Fides, pp. 99–121. Jourdan, Christine (2001) Contact. In: Alessandro Duranti (ed.) Key Terms in Language and Culture. Oxford: Blackwell, pp. 40–4. Keesing, Roger (1981) Cultural Anthropology: A Contemporary Perspective. New York: Holt, Rinehart, Winston. Keesing, Roger (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford: Stanford University Press. Lefebvre, Claire (1999) Creole Genesis and the Acquisition of Grammar. Cambridge: Cambridge University Press.

380

Christine Jourdan

Lefebvre, Claire (2001) The interplay of relexification and levelling in creole genesis and development. Linguistics 39, 371–408. Lovejoy, Paul E. and David V. Trotman (2002) Enslaved Africans and their expectations of slave life in the Americas. In: Verene A. Shepherd and Glen L. Richards (eds.) Questioning Creole: Creolisation Discourses in Caribbean Culture. Kingston, Jamaica: Ian Randle Publishers, pp. 67–88. McWhorter, John (1992) Substratal influence in Saramaccan serial verb constructions. Journal of Pidgin and Creole Languages 7, 1–53. Mintz, Sydney (1974) Caribbean Transformations. Chicago: Aldine Publishing Company. Mintz, Sydney (1982) Caribbean market places and Caribbean history. Radical Historical Review 27, 110–20. Mintz, Sydney and Richard Price (1992 [1976]) The Birth of African-American Culture: An Anthropological Perspective. Boston, MA: Beacon Press. Moitt, Bernard (2001) Women and Slavery in the French Antilles, 1635–1848. Bloomington: Indiana University Press. Moore, Clive (1985) Kanaka Maratta: A History of Melanesian Mackay. Port Moresby: Institute of Papua New Guinea Studies and the University of Papua New Guinea Press. Morissey, Marietta (1989) Slave Women in the New World: Gender Stratification in the Caribbean. Lawrence: University Press of Kansas. Postma, Johannes (1990) The Dutch in the Atlantic Slave Trade, 1600–1815. Cambridge: Cambridge University Press. Price, Richard (1983) First-Time: The Historical Vision of an Afro-American People. Baltimore: The Johns Hopkins University Press. Samarin, William (1982) Colonization and pidginization on the Ubangi river. Journal of African Languages and Linguistics 4, 1–42. Saunders, Kay (1974) Uncertain Bondage. PhD dissertation. University of Queensland, Brisbane. Schieffelin, Bambi S., Kathryn A. Woolard, and Paul V. Kroskrity (1998) Language Ideologies: Practice and Theory. Oxford: Oxford University Press. Siegel, Jeff (1987) Language Contact in a Plantation Environment: A Sociolinguistic History of Fiji. Cambridge: Cambridge University Press. Siegel, Jeff (1997) Mixing, leveling, and pidgin/creole development. In: Arthur K. Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles. Amsterdam: John Benjamins, pp. 111–50. Siegel, Jeff (2003) Substrate influence in creoles and the role of transfer in second language acquisition. Studies in Second Language Acquisition 25 (special issue on Reconsidering the Role of SLA in Pidginization and Creolization), 185–210. Singler, John Victor (1993a) The cultural matrix of creolization: Evidence from Goupy des Marets. In: The African Presence in Caribbean French Colonies in the Seventeenth Century: Documentary Evidence (Travaux de recherche sur le créole haïtien, nos. 16– 17). Montréal: UQAM, Groupe de recherche sur le créole haïtien, pp. 187–224. Singler, John Victor (1993b) The setting for creole genesis in France’s Caribbean colonies: Evidence from seventeenth century Marie-Galante. In: The African Presence in Caribbean French Colonies in the Seventeenth Century: Documentary Evidence (Travaux de recherche sur le créole haïtien, nos. 16–17). Montréal : UQAM, Groupe de recherche sur le créole haïtien, pp. 225–36. Singler, John Victor (1995) The demographics of creole genesis in the Caribbean: A comparison of Martinique and Haiti. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 203–32.

The Cultural in Pidgin Genesis 381 Smith, Richard (1959) Family structure and plantation systems in the New World. Plantation Systems of the New World (Social Science Monograph, no. 7). Washington, DC: Pan American Union, pp. 148–59. Taylor, Charles (1997). L’interprétation et les sciences de l’homme [Interpretation and human sciences]. In: La liberté des modernes. Paris: Presses Universitaires de France. Thompson, Edgar. T (1975) Plantation Societies, Race Relations, and the South: The Regimentation of Populations. Durham, NC: Duke University Press. Trudgill, Peter (1986) Dialects in Contact. Oxford: Blackwell. Wallerstein, Immanuel (1974) The Modern World System: Capitalist Agriculture and the Origins of the European World Economy in the Sixteenth Century. New York: Academic Press. Woolard, Kathryn A. and Bambi S. Schieffelin (1994) Language ideology. Annual Review of Anthropology 23, 55–82.

Grammaticalization in Pidgins and Creoles 383

Part III Pidgins/Creoles and Linguistic Explanation

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Grammaticalization in Pidgins and Creoles 385

16

Grammaticalization in Pidgins and Creoles ADRIENNE BRUYN

1 Introduction Processes of grammaticalization may be considered central to the development of pidgins and creoles (P/Cs). By its very nature, the expansion of pidgin varieties concerns the elaboration of the repertoire of grammatical forms and devices. Similarly, to the extent that creoles are languages newly emerged from a disruptive stage – be it a pidgin phase, or accumulated effects of second language acquisition – their initial histories involved more or less radical reduction of the morphosyntactic apparatus vis-à-vis the lexifier language, which subsequently may have been compensated by processes of grammaticalization. During the past decades, grammaticalization has received a growing attention in the fields of historical linguistics and typology. This area of research is relevant to the study of P/Cs in that it provides a framework for the interpretation of developments in such languages and may thus contribute to our understanding of their origins. At the same time, the investigation of the ways in which grammaticalization proceeds in P/Cs may provide insights into the nature of grammaticalization by broadening its scope to issues pertaining to young languages and language contact. This chapter will review some developments in P/Cs as they have been studied in a grammaticalization perspective, compare them with similar changes in languages with a longer history, and attempt to distinguish between language-internal and contact-induced grammaticalization. We will see that processes of grammaticalization always involve reanalysis, but that there are instances of reanalysis without grammaticalization, involving the reanalysis of a lexifier form which already has a grammatical function. In addition, several cases that appear to be instances of grammaticalization turn out to have taken shortcuts in their development which are modeled on patterns in the substrate. In the extreme case, that of “apparent grammaticalization,” no

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

386

Adrienne Bruyn

internal process of grammaticalization appears to have taken place but rather a copying of substrate functions (polysemy copying). After an introduction to the grammaticalization framework (section 2), an example of grammaticalization in Melanesian Pidgin provides the starting point for a discussion of the way P/C expansion may proceed (section 3). Section 4 presents some case studies of grammaticalization in creoles which serve to illustrate the various kinds of developments that may occur and the issues involved in their interpretation as grammaticalization. The possible contribution of substrate languages and the interaction between contact-induced and language-internal grammaticalization processes are the topics of section 5. I will argue that, while the grammaticalization framework enables us to arrive at a more precise assessment of the processes involved in P/C development, it is especially useful insofar as it enables us to identify P/C properties that are somehow unexpected and thus require explanation. In some such cases, it may become particularly clear that substrate patterns determined the route of development in a P/C; in other cases the account may involve the nature of the relationship between a creole item and its lexifier source form.

2

Grammaticalization

While grammaticalization as a concept can be thought of in various ways, the starting point here is the tradition where it is defined as “the change whereby lexical items and constructions come in particular linguistic contexts to serve grammatical functions, and, once grammaticalized, continue to develop new grammatical functions” (Hopper & Traugott 2003, p. xv). It characteristically involves correlating changes on the semantic-pragmatic, morphosyntactic, and, sometimes, phonological levels, leading an item to shift rightward on a cline of grammaticality: (1) content item > grammatical word > clitic > inflectional affix Among the concerns of the grammaticalization approach are the possible sources of grammatical forms and the trajectories they go through, and the nature of the semantic and pragmatic changes involved and their underlying cognitive principles. By way of illustration, let us briefly look at the marking of future in the Romance languages (Hopper & Traugott 2003, pp. 8–9, 52–5). In contexts involving gerundives, the Latin lexical verb habere ‘to have, possess’ acquired a sense of obligation or at least future orientation, presumably transferred from the gerundive, itself once expressing obligation. Possessing something may imply certain obligations, or future intentions (see for details Hopper & Traugott 2003 and references cited there; Bybee, Perkins, & Pagliuca 1994, pp. 258ff.). Furthermore, infinitives began to replace gerundives, in particular with speech act verbs like dicere ‘to say’:

Grammaticalization in Pidgins and Creoles 387 (2) Quid habes dicendum / dicere? what have-2s say-ger say-inf ‘What do you have to say?’ In Late Latin, the inflected form of habere could follow its infinitival complement, and a structure such as [[cantare] habeo] was reanalyzed as a single clause [cantare habeo]. (3) cantare habeo sing have-1s ‘I have to/will sing.’ Through fusion across morpheme boundaries and phonological reduction, habeo developed from a verb in periphrastic constructions as in (3) into the French affix -ai occurring in inflectional futures such as je chanter-ai ‘I will sing’. The whole development – which affected other forms, and other Romance languages as well, compare, e.g., first person plural cantare habemus becoming French (nous) chanter-ons and Italian cantar-emo ‘we will sing’ – exhibits various aspects that are considered typical for grammaticalization. The semantic shift from ‘to possess’ to the grammatical meaning of future was based on pragmatic inferences induced in contexts involving gerundives, as in (2). Through frequent usage the inferred sense of future orientation conventionalized and became part of the meaning of the item. On the structural level as well, the change proceeded in a specific configuration. The position of the inflected form of the main verb habere immediately after its infinitival complement is a prerequisite for the reanalysis as a clausal unit. This, in turn, provided the possibility for habeo to cliticize and to develop subsequently into a suffix, as in French. The eventual fusion and phonological reduction are typical for the later stages of this kind of development. Processes of grammaticalization always involve reanalysis, whereby a hearer assigns structural or semantic properties to a form or string that are different from those as construed by the speaker (Hopper & Traugott 2003, pp. 50ff.). In the case of future marking in the Romance languages, the rebracketing of [[cantare] habeo] to a unit without dependency relations is an instance of reanalysis, as is the subsequent reinterpretation of the verb habeo as an affix expressing future. Note, however, that reanalysis may well occur without grammaticalization. For example, blackbird is a lexicalized item resulting from the reanalysis of a phrase consisting of a modifying adjective black and a noun bird. In section 4, we will see how a principled distinction between grammaticalization and reanalysis as proposed by Detges (2000) may help to differentiate between types of developments in P/Cs. While habere grammaticalized into a future marker and lost its autonomy, the original lexical verb ‘to have, possess’ continued to exist. Such a split, or divergence, is not uncommon, even though the connection between a grammaticalized form and its lexical source is not always immediately

388

Adrienne Bruyn

transparent anymore to the language user. When we say ahead of the crowd we do not think of “heads,” nor do we necessarily feel there to be a connection between while as a conjunction and the noun still occurring in phrases such as a little while. The same applies to the French negator pas – as in je ne sais pas ‘I don’t know’ – and the noun pas ‘step’. Cross-linguistically, some meanings are more liable than others to be used to express other, more abstract meanings, or particular communicative intents. This explains why the same lexical items recur as sources for grammatical markers in certain domains. For instance, in languages from different parts of the world – including Sranan, see section 5 – words denoting body parts have come to indicate location, the verb meaning ‘to give’ to express a beneficiary role, and ‘to go’ to refer to the future. Whether an item actually grammaticalizes or not is arbitrary, but if it does, there are likely paths along which it may proceed. The general tendency is for concrete meanings to become more abstract. On a universal scale of metaphoric abstraction such as that in (4), there may be a shift from the domain of space to the more abstract domain of time but not the other way round, and extension from, for example, person to time is unlikely to proceed directly (Heine, Claudi, & Hünnemeyer 1991, p. 48). (4) person > object > activity / process > space > time > quality Such restrictions on the paths of change result from the nature of pragmatic inferences such as metaphor and metonymy. On the morphosyntactic level, items generally move down the cline of grammaticality given in (1), undergoing decategorization and loss of autonomy. Thus, a content word may develop into a function word, which may in turn become a bound morpheme, whereas the reverse development occurs only exceptionally. Grammaticalization is thus typically unidirectional. The various points along the route of development form a cline of grammaticalization, reflecting, for example, that a noun does not change into a locational postposition at once. Rather, it gradually loses the properties typical for a noun, for example the ability to be marked for number and case, to be modified by a demonstrative, et cetera (Heine et al. 1991, pp. 133–4). Typically, the morphosyntactic changes correlate with phonological reduction and with semantic shifts, often involving weakening or bleaching of aspects of the original meaning, especially in the later stages. Such clines of grammaticalization, also referred to as chains, or scales, may also capture the synchronic distribution of an item that has certain lexical properties in one context but is more functional in another. Processes of grammaticalization may extend over a considerable period. The development in the Romance languages of inflectional affixes indicating future was, for instance, a matter of centuries rather than decades. Such processes can be considered gradual in that they consist of various small changes following each other – e.g., when a noun loses its nominal properties one by one. Furthermore, the innovation gradually increases in frequency while

Grammaticalization in Pidgins and Creoles 389 it spreads to other linguistic contexts, and through the speech community (Hopper & Traugott 2003, pp. 46–9, 232). Grammaticalization appears to be brought about by a complex interplay between the need for expressiveness and creativity on the one hand, and for regularization and routinization on the other. For example, speakers of French initially used pas to emphasize a negation, but by being used often in that manner, it became the regular negator (see Gamillscheg 1957, in Hopper 1991). Under this view, grammaticalization is motivated by speaker–hearer interactions and communicative strategies. As Hopper and Traugott (2003, p. 74) point out, however, it is problematic to regard communicative need as a factor that induces grammaticalization. To assume that grammaticalization occurs in order to fill functional gaps in the grammar leads to the normally unwarranted implication that a language is in some respect inadequate before the grammaticalization took place. In the case of P/Cs, however, things could be considered different in this respect. Pidgins with a limited function are generally structurally limited as well, and emerging creoles may lack grammatical material that was not transmitted from the lexifier. It may therefore be the case that grammaticalization in the early stages of P/C development sometimes involves the creation of new categories or constructions in order to fill functional gaps. To the extent that languages vary widely in what is coded by grammatical means, it is difficult to establish on independent grounds what would count as a functional inadequacy. Given that there are languages without, for instance, an indefinite article, we would be hard pressed to claim that the grammaticalization of such an article in a creole language like Sranan (see section 4) took place in direct response to a particular functional gap. The same would apply to, for example, the morphosyntactic marking of relative clauses. Still, the conglomeration of newly emerging categories and constructions in P/Cs is unusual in comparison to what is found in languages with a longer history. The evolution of grammatical constructions and relations is also among the concerns of work in grammaticalization. Although this often goes hand in hand with the grammaticalization of individual items, this is not necessarily the case. An example would be the development from topic to syntactic subject through the conventionalization of discourse strategies and word order. This may, but need not, involve the grammaticalization of, for example, a pronoun into an agreement marker. Givón (1979), who uses the term “syntacticization” to refer to the development of loose, pragmatic discourse structures into tight, grammaticalized syntactic structures, notes a shift from pragmatic to syntactic mode on various levels (p. 208): (5) diachronic ontogenetic [macaronic] pidgins/creoles register level

loose parataxis → tight syntax early pragmatic mode → later syntactic mode nongrammar → grammar unplanned/informal speech → planned/ formal speech

390

Adrienne Bruyn

While diachronic developments may help to understand synchronic phenomena, synchronic facts may, conversely, provide indications of processes of grammaticalization in the past. Along the lines of Givón’s (1971, p. 413) statement “Today’s morphology is yesterday’s syntax,” synchronic data are often interpreted as reflecting diachronic developments. However, without actual data to support a particular development we can never be sure it has taken place in the alleged way. As will be shown below, this is particularly relevant in the case of P/Cs. Much more could be said about grammaticalization and the various perspectives from which it is approached. There are currently debates on fundamental issues such as the claim of unidirectionality, how grammaticalization relates to reanalysis, whether it can be distinguished from other kinds of linguistic change, and what the status is of a theory of grammaticalization (Campbell 2001; Fischer, Norde, & Perridon 2004). Nevertheless, there is sufficient agreement on what may be considered typical for processes of grammaticalization to compare developments in P/Cs with those in languages with a less turbulent history.

3

Pidgin Expansion: Grammaticalizing ‘by and by’

In rudimentary pidgins, word order may vary, arguments may remain unexpressed, semantic relations are indicated by circumlocution or not overtly expressed at all, and, on the whole, grammatical elements are scarce. Consider the following examples: (6) a.

Papuan Pidgin English (mi) sik, mi sindaun 1s sick 1s stay ‘When I was sick I stayed at home.’ (Mühlhäusler 1997, p. 151) b. Hiri Trade Language (Kerama) Aie, na nava pene na navai. Rae imo. gee 1s fish some 1s eat insides good ‘If I could only eat some fish, I’d feel a lot better.’ (Dutton 1983, p. 90)

Neither tense nor aspect is expressed here, and while the clauses are juxtaposed in iconic order reflecting the sequence of events, the precise nature of the relationships – temporal in (6a), conditional in (b) – is not indicated by subordinating conjunctions or otherwise. Linguistic communication of this type, where interpretation depends heavily on intonation, linguistic context, extra-linguistic situation, and shared knowledge, belongs to Givón’s (1979) pragmatic, or pre-syntactic, mode. Such a mode of communication may suffice in situations where the pidgin serves only a limited function, covering a relatively narrow range of topics anchored in the actual context. When a

Grammaticalization in Pidgins and Creoles 391 variety comes to be used in a wider array of functions and domains, however, it may expand through grammaticalization of individual forms as well as through syntacticization: word order becomes fixed, the syntactic component becomes more complex, and grammatical elements evolve, including conjunctions, relative clause markers, number marking, and tense, mood and aspect (TMA) markers. How, and why, do such processes of expansion in pidgins and young creoles proceed, and how does this relate to the pathways of grammaticalization usually followed in older languages? And to what extent are they affected by, or even dependent on, nativization? With these questions in mind, let us consider the development of the adverb baimbai ‘afterwards, later’ (< English by and by) in Tok Pisin, the Papua New Guinea variety of Melanesian Pidgin (Sankoff & Laberge 1973; Romaine 1995). It appears that the shift from adverb to phonologically reduced preverbal future/irrealis marker may be captured nicely in terms of grammaticalization. Disregarding various complications, some of which will be addressed below, the grammaticalization scenario may be sketched as follows. As used in (7a), baimbai is an optional clause-initial adverb. In (7b), the form is reduced to bai and appears between subject and verb, a position normally not occupied by adverbs in Tok Pisin. (7) a.

Baimbai mi go. later 1s go ‘Later I’ll go.’ b. Mi bai go. 1s fut go ‘I’ll go.’

While baimbai has continued to exist as a sentential adverb, the usage of the reduced form bai as a tense marker has become determined by grammatical rule. The original adverbial meaning ‘afterwards, later’ has been generalized and developed into the grammatical meanings of future and irrealis, and bai has become a closed-class item belonging to the emerging paradigm of TMA particles. The possible co-occurrence of bai with lexically expressed future meaning – for instance by an adverb such as klostu ‘soon’ – also indicates a certain degree of grammaticalization: semantically, bai is redundant in such cases, and its presence is rather a matter of grammar. As argued by Romaine (1995), the grammaticalization of baimbai/bai follows in many respects the universal paths of development for futures predicted by Bybee, Pagliuca, and Perkins (1991) at the same time as it intersects with many of the key routes leading to modality. However, she also concludes that the synchronic variation in the position of baimbai/bai is not a straightforward reflection of diachronic stages, that reduction and position are separate issues, and, furthermore, that the grammaticalization is not dependent on nativization or urbanization. The latter point is explored by Sankoff and Laberge (1973). They compared adults who were fluent second-language speakers of Tok Pisin and children

392

Adrienne Bruyn

who had acquired it as a native language with respect to their use of baimbai/ bai and the extent to which it functioned as a grammatical future marker. They concluded that, while the change from adverb to future marker was well under way prior to the existence of considerable numbers of native speakers, the children played an important role in carrying it through. However, whereas the children turned out to be ahead of the adults with respect to stress reduction, no differences were observed between the groups with respect to redundancy and the position of bai vis-à-vis the verb – as would have been expected if the children for whom Tok Pisin was the first language were leading an ongoing change induced by nativization (cf. Sankoff 1991). Taking both late-twentieth-century and historical data into account, Romaine (1995) similarly emphasizes that the grammaticalization of bai as a preverbal tense-aspect particle proceeded independent of nativization or urbanization (cf. also Jourdan 1985, on baebae/bae in the Solomon Islands variety of Melanesian Pidgin). Romaine’s data show that present-day second-language speakers in rural areas use bai relatively more often in preverbal position (as in 7b above) than do children. Another notable finding is that the option of using the long form baimbai preverbally has been available for at least a century, indicating not only that nativization was no prerequisite for this construction, but also that syntactic position and phonological reduction do not correlate. Such deviations from the chronology of the grammaticalization scenario as sketched above might be taken to suggest that the mother tongues of the secondlanguage speakers have induced the development of baimbai/bai – a possibility to which we will briefly return in section 5. The case of baimbai/bai illustrates that innovative developments are not confined to creole varieties but take place as well in extended pidgins, such as the non-native varieties of Melanesian Pidgin. Grammaticalization in these pidgins has affected various areas of the grammar. Some more examples from Tok Pisin include (see also Mühlhäusler 1997, pp. 166–75): the development of the third person singular pronoun im to a transitive enclitic or a causative suffix (Sankoff 1977; Mühlhäusler 1980); the reanalysis of paratactically conjoined clauses to embedded structures along with the emergence of complementizers (Woolford 1979; see also Meyerhoff 2002, on Bislama); the development of hia ‘here’ toward an emphatic marker and further as a marker of definiteness, while being reduced to ya, and the extension of the same form to a bracketing device at the beginning and end of relative clauses. Regarding the latter development, Sankoff and Brown (1976) show how the bracketing evolves from discourse strategies more generally used for information structuring. In all these cases, items have assumed a morphosyntactic function they did not have originally and moved rightward along the cline of grammaticality in (1). Significantly, these new functions can all be thought of as new categories, previously unexpressed. This is less often the case in non-P/Cs. While these developments appear to be similar to processes of grammaticalization in non-P/Cs, the issue of whether they should be regarded as gradual is a matter of perspective. On the one hand, they have proceeded

Grammaticalization in Pidgins and Creoles 393 rather faster than is the case with documented cases of grammaticalization in older languages, which may be explained by their being prompted by particular communicative needs. On the other hand, they can be considered gradual in the sense that they span a century or so, and that, significantly, various stages of development are discernible. For instance, for his investigation of number marking in Tok Pisin, Mühlhäusler (1981, 1997, p. 168) made use of historical documents going back to the end of the nineteenth century as well as modern texts from speakers of different varieties. He observes that, after a jargon stage in which there is no formal means for marking plurality, the marker ol (< English all) begins to be used optionally, with a preference for nominals referring to humans and those appearing in subject position, then becomes increasingly redundant, first with pronouns and animate nominals, then with quantifiers, and variably so with inanimates, to finally be used rather generally. (8) human > animate > inanimate This kind of stepwise progression along a scale of pragmatic saliency or topicality is characteristic for grammaticalization in general (cf. Givón 1984, pp. 364ff.; Corbett 2000, pp. 55ff., 266–7). As mentioned above, however, Romaine’s (1995) findings regarding baimbai/bai in Tok Pisin indicate that the actual diachronic developments deviated somewhat from the route which might have been expected on the basis of the synchronic variation. This indicates that some caution is called for in regarding patterns that suggest grammaticalization as constituting evidence for actual developments. Below we will see that this caveat applies more generally.

4 Grammaticalization in Creoles The previous section referred to quite a number of studies dealing with developments in varieties of Melanesian Pidgin that, albeit perhaps telescoped in time, proceeded along lines rather similar to processes of grammaticalization in languages with a longer history. Until the 1990s, few studies appeared that approached the creole languages of the Atlantic area and the Indian Ocean in a similar perspective. A possible explanation for this difference may lie in the relatively recent expansion of Melanesian Pidgin, providing the opportunity to investigate both historical stages and ongoing changes. As more studies on grammaticalization in creoles have become available recently, it becomes possible to address the issue of how processes of grammaticalization may proceed in creoles and which role such processes play in their development. Below, some such case studies are presented in order to illustrate the kind of developments that may occur and the issues involved in their interpretation as grammaticalization. One of the issues concerns the time span involved in processes of grammaticalization in creoles, another how to distinguish

394

Adrienne Bruyn

grammaticalization from other processes of change. The interplay between grammaticalization and substrate influence is discussed in section 5. In creole studies, the issue of gradualness has been a matter of debate: do creoles develop through nativization of a pidgin within one or two generations, as claimed by Bickerton (1988), or is the process less dependent on children and proceeds rather more gradually, as defended on various grounds by Arends (1993) and Singler (1992)? It may be noted here that aspects of Bickerton’s Bioprogram Hypothesis, in particular the “reconstitution” of morphemes which were lost during the dilution of the lexifier language, could be regarded as a special kind of grammaticalization; thus, reconstitution could in principle draw on the universal tendencies and the underlying cognitive strategies of grammaticalization. As a simple example, the fact that several creoles share an irrealis marker based on the verb ‘to go’ is in accord with its semantic potential and has many parallels cross-linguistically. For the Bioprogram Hypothesis, however, the fact that irrealis is marked is more relevant than which form specializes in doing so. The semantic distinctions and the ways in which these are morphosyntactically implemented in the creole language are considered to be prompted by the unfolding bioprogram and to be pre-determined by the cognitive and structural principles of the human language faculty. On this point, of course, the Bioprogram Hypothesis differs crucially from the general tenets of the grammaticalization approach, where it is assumed that the developments are driven by discourse strategies as well as cognitive processes of a more general kind. Also, as pointed out in section 2, processes of grammaticalization are, in contrast to the effects of the bioprogram, taken to be gradual and range over a longer period. It is thus important to learn more about the time span within which new morphosyntactic elements actually evolve in creole languages in order to gain insight into how creoles develop, and how grammaticalization in creoles relates to comparable changes in languages with a longer history. There is increasing awareness that there is much diversity among creole languages and in the ways in which they emerge. In the case of the articles in Sranan, to which we turn now, different kinds of developments within one and the same language are at issue.

4.1

The grammaticalization of articles in Sranan: Gradual versus abrupt

Sranan developed on the plantations of Suriname, which became an English colony in 1651. While its lexicon is primarily derived from varieties of English, it also contains quite a large number of Dutch-derived words, due to Dutch becoming the colonial language in 1668. Sranan emerged in a context where a break in transmission made English unavailable as a target (Smith 2001); it is thus a radical creole. Among the various English function words that were lost

Grammaticalization in Pidgins and Creoles 395 are the articles the and a(n). Other items came to function as articles in Sranan, but the way this happened is not quite the same for the definite and the indefinite forms (Bruyn 1995, 2007). The development of the definite articles in Sranan offers a textbook case of grammaticalization. The forms na (or a) (singular) and den (plural), both occurring prenominally, as in modern Sranan a man/den man ‘the man’/‘the men’, developed over time from determiners with demonstrative value toward more generally used definite articles marking identifiability. In eighteenth-century Sranan, the singular form is not (n)a but da, derived from the English demonstrative that in its adjectival function, as in that man; the plural form den, earlier also dem, is a reflex of English them. In accordance with their provenance, both da and den clearly have a deictic potential and must sometimes be regarded as demonstratives rather than definite articles in the eighteenth-century text sources. This has its counterpart in the fact that in this stage of the language, it is not unusual for noun phrases with definite reference to appear as bare nouns. Over time, however, the use of da/(n)a and den gained considerably in frequency and the demonstrative value of the forms diminished, implying that they became even more generally used. At present, the forms have lost their demonstrative potential and are used with most nominals the identity of which is assumed to be known to or recoverable by the hearer. In the case of the singular form, there has been a phonological change as well, from da to na and reduced a. The functional change of da/(n)a and den proceeded hand in hand with an increasing use of disi ‘this, these’ and dati ‘that, those’ as nounmodifying demonstratives. All this corresponds nicely to an often attested process of grammaticalization whereby demonstratives evolve into articles – indeed, demonstratives are overwhelmingly the most common source of definite articles (Lyons 1999, p. 331). It may be noted however, that Sranan (n)a and den are not fully grammaticalized as articles, to the extent that even in the present-day language, definite NPs sometimes occur as bare nouns, i.e., without a determiner in contexts where one would be required in a language like English. This includes NPs with referents of which the individual identity is identifiable but irrelevant, but also to cases where the referent is an important participant in the discourse which can be identified in a particular context. Whether or not the grammaticalization of na and den will proceed further, at the cost of bare nouns, is difficult to predict, but certainly the possibility cannot be excluded. For its indefinite singular article, Sranan again drew on a cross-linguistically common source to mark indefinite singular nouns, namely wan ‘one’ (< English one). Givón (1981) argues that ‘one’ is uniquely fit to perform the complex task of introducing a new argument into the discourse. As a quantifier, ‘one’ implies existence in the universe of discourse, i.e., referentiality, but not identifiability, or definiteness. Being the lowest numeral, it can refer to ‘one of many’, ‘one of the group’, or ‘one of the type’, making it possible for the hearer to recognize the referent as belonging to a certain type by its connotational or generic properties. While the functional extension of ‘one’ toward indefinite

396

Adrienne Bruyn

article is a universal one, languages may differ in the extent to which this development has proceeded. Givón (1981) postulates a developmental continuum the starting point of which is a situation in which no systematic distinction is made between referential indefinite and nonreferential NPs. This may apply to a rudimentary pidgin where article usage is unpredictable, as in Hawaiian Pidgin English (Bickerton 1981). In the next stage of the development, manifested by the typical creole language according to Bickerton as well as Givón, ‘one’ marks specific or pragmatically referential NPs only, where pragmatic referentiality has to be understood as “meant by the speaker to have important reference in the universe of discourse.” Referents of which the exact identity is unknown or irrelevant to the point at issue would appear as bare nouns, as do semantically nonreferential arguments, such as generic NPs and NPs in the scope of negation. A careful look at the actual use of wan in Sranan shows, however, that it does not exactly correspond to the system supposed to be typical for creoles: both in early and in present-day Sranan wan occurs quite often with NPs that are neither semantically referential nor pragmatically important, as illustrated by the following examples. In (9a), the NP is a generic predicate in the scope of negation, hence semantically nonreferential; (9b) is an instance of wan occurring with an NP with low pragmatic referentiality. (9) a.

a no kann komm wan Discipel va mi 3s neg can become indef.s disciple of 1s ‘He cannot become a disciple of mine.’ (Schumann 1781, p. 182) b. Dan den nengre e bron wan korfayapatu heri neti. then def.pl negro cont burn indef.s charcoal-pot whole night ‘Then the slaves burnt a charcoal stove the whole night.’ (De Drie 1985, p. 50)

The main function of wan is to mark indefiniteness, or singularity, or both, rather than referentiality. With regard to the expression of singularity, the role of wan is all the more important given the fact that, in Sranan, nouns are not marked for number. At the same time, wan is not obligatory with all indefinite singular nominals, and whether or not it occurs is certainly influenced by referentiality or specificity; this cannot, however, be considered the primary distinctive factor. Thus, the original numeral wan – which itself remains functional as a numeral and an intensifying element – has developed into a non-obligatory article marking nominals as singular and unknown to the hearer. In its function of article, the original quantificational feature has weakened, but singularity persists as a core feature of the article’s value. Referentiality, on the other hand, while playing some role in whether wan is used or not, has bleached considerably, to the extent that the article can appear with nonreferential arguments. Somewhat unexpectedly, a quantitative study of its distribution in Sranan texts from various periods leads to the conclusion that wan did not become a

Grammaticalization in Pidgins and Creoles 397 more grammaticalized indefinite article over the course of the eighteenth, nineteenth, and twentieth centuries. Had this been the case, the use of wan would have been extended, and the significance of referentiality in determining its occurrence would have diminished over time. However, there has been no increase in the overall use of wan, nor a rise in the occurrence of wan with semantically nonreferential NPs or with NPs that are pragmatically unimportant. Thus, wan developed into the article as it currently exists within a short time span during the early stages of Sranan, without bleaching further after that. In respect to the absence of further diachronic development, the case of wan contrasts not only with that of the definite articles in Sranan, but also with, for example, the development of English a(n). This form began as a numeral in Old English, then became a pragmatically strong presentative, and eventually an unrestricted indefinite article. As shown by Hopper and Martin (1987), this change proceeded slowly but steadily during the tenth through to the twentieth century. Notwithstanding the fact that in the case of Sranan wan the grammaticalization concerns only part of the trajectory, it must be considered a relatively fast development. Because wan grammaticalized only to a certain extent during the early stages but did not develop further after that, the development may be characterized as abrupt or instantaneous.

4.2

Tense markers in French creoles: Grammaticalization or reanalysis

With the Sranan articles, the starting points for the grammaticalizations were forms that were evidently part of the language. The definite articles evolved from the early forms attested as da and den/dem, and in the case of wan, a split occurred between the original numeral and the grammaticalized indefinite article. Now, consider a form such as ben or bin, derived from English been, which functions as a past or anterior tense marker in many English creoles. For Sranan in any case, there are no traces of ben other than as a preverbal tense marker. This implies that, rather than a development based on the usage of the form within Sranan itself, there has been a shortcut from the English past participle, which has already some grammatical value, to the function of tense marker in the creole language. This development can be seen as an instance of restructuring, a term used to refer rather broadly to the structural reorganization of linguistic material affecting and more or less radically altering the lexifier input (Neumann-Holzschuh & Schneider 2000; Mufwene 2001, pp. 12ff., 27ff.). The question is, however, whether it is insightful to regard this type of development as similar to processes of grammaticalization in non-P/Cs, as Mufwene does (1996, 2001, pp. 28, 54). In order to discover how P/Cs have come into being, it appears useful to distinguish cases such as Sranan ben < English been from those such as the Sranan articles (n)a, den, or wan, where grammaticalization affected forms that were part of the emerging creole. With grammaticalization, the source item normally continues to be used in its original function alongside the grammaticalized variant at least for

398

Adrienne Bruyn

some time. This, however, is not at issue in the case of Sranan ben, which is not attested in any other function than that of tense marker. Detges (2000), taking a cognitive-pragmatic approach, argues with respect to similar items in French creoles that restructurings of the ben kind are better considered as reanalysis not involving grammaticalization. Detges draws a fundamental distinction between the two processes. In contrast to grammaticalization, reanalysis is a general mechanism of change, without a strong directionality. It plays a part in processes of grammaticalization in that grammaticalizations brought about by speakers’ usages are ratified by reanalyses by hearers (Detges & Waltereit 2002). Detges (2000) applies this distinction to the development of various tense and aspect markers in French creoles. His interpretation of three of these cases is summarized below, beginning with a clear instance of grammaticalization ( fini), then turning to instances of non-grammaticalizing reanalysis. Detges discusses fini (< French fini(r) ‘to finish’) in five French creoles: Haitian, Guyanais, Mauritian, Seselwa, and Réunionnais (2000, p. 139ff.). It occurs as a lexical verb ‘to finish, to bring something to its end’, as well as in various grammaticalized functions. By its use in rhetorical discourse strategies, the meaning of the lexical verb has given rise to metonymical, associative extensions. For example, if a speaker utters “I finished the job,” an obvious inference is that the speaker is finished with it because the job is done. Such inferences may lead the meaning of fini to extend to the resultant state of some preceding event, or finishedness, and from there to complete(ly). In several of the creoles the latter inference became conventionalized as an independent meaning, with fini functioning as an adverb of degree that can modify verbs and adjectives with which the original verb fini ‘finish, be finished with’ normally cannot co-occur. In the Haitian Creole example provided by Detges (2000, p. 140), fini modifies the stative verb kônê ‘to know’: (10) M’ pa t’ a- vlé l’ finkônê sa 1s neg ant fut want 3s completely know that ‘I wouldn’t want him to know this completely/the whole story.’ (Sylvain 1979 [1936], p. 92) The meaning complete(ly) may change further into perfect(ly), while at the same time finishedness may induce fini to develop into a past marker indicating that something has been done, or into an adverb ‘already’. It is not the case that the various possible changes – which are attested for other languages, including other P/Cs – have occurred in each of the French creoles, but in every one of them there has been some development. The result, within each variety, is divergence, or layering: synchronically, fini exhibits polysemy, reflecting different points on the scale of grammaticalization. All in all, the case of fini can be classified as “ordinary” grammaticalization. The preverbal tense marker té or ti, by contrast, is considered to result from non-grammaticalizing reanalysis by Detges (2000). While the form has been

Grammaticalization in Pidgins and Creoles 399 suggested to derive from été, the past participle of être ‘to be’, Detges assumes that it goes back to the past tense form était – in particular as used in il était à écrire ‘he was writing’, the periphrastic expression of the imparfait, a tense typically used for background information in narratives. Detges shows how té/ ti may have survived in the creoles as an anterior marker through the reanalysis of était (à). While était (à) is by itself not pragmatically salient, it is commonly accompanied by temporal adverbs in certain narrative patterns, and the temporal frame of background past events as established by such adverbs was reinterpreted as a function of était (à). The result is the form té/ti, which bears a certain similarity to its French etymon in expressing anterior tense but behaves differently syntactically. Markers resulting from this kind of reanalysis typically do not display layering: they have only one function and they are, in Detges’s words, conceptually isolated. Similar to ben in Sranan, té/ti never had a function other than that of a preverbal anterior marker in the French creoles, nor was it used for expressive purposes similar to the rhetoric uses that fueled the grammaticalization in the case of fini. Cases of non-grammaticalizing reanalysis like té/ti thus contrast with those involving grammaticalization, where a form may have less and more grammaticalized uses alongside each other, especially in the earlier stages of the change. By assessing the creole facts from the perspective of the grammaticalization approach in this manner, it becomes possible to tease apart different kinds of processes playing a role in creole development. In some cases, there has indeed been grammaticalization within the creole language: an item belonging to the language assumes new meanings and grammatical functions. This applies to fini in the French creoles as well as to the Sranan articles. However, in cases such as té/ti in French creoles and English creole ben/bin, the grammatical forms rather result from non-grammaticalizing reanalysis: a lexifier form which already has some grammatical function is adopted by the creole, including some of its grammatical meaning, but with another categorial label assigned to it – in these cases that of preverbal tense marker. Even if the creole item is innovative in its morphosyntactic behavior, grammaticalization in its developmental, usage-based sense is not at issue. The last case of French tense and aspect markers to be considered here, that of the future marker pou in Mauritian French Creole, serves to illustrate that an item can be affected by both non-grammaticalizing reanalysis and by grammaticalization. In seventeenth-century French, the periphrastic form être pour (lit. ‘to be for’) had, starting from the concept destination, developed the contingent meaning of near future, as a first step on the path of discourse-based grammaticalization. Both the old and the new meaning are retained in pou as it occurs in various French creoles, namely as a preposition ‘for’, and as a marker of near future. Thus, Detges argues that the future marker pou did not arise through grammaticalization of the preposition in the various French creoles, but was adopted from French in a similar way as té/ti, i.e., by reanalysis. This must primarily have involved the preposition which is more lexical and

400

Adrienne Bruyn

thus more salient conceptually, but once the meaning of the preposition was recognized as ‘for’, the more grammatical meaning near future, being conceptually related to destination, could have been recognized as well. In any case, the co-existence of the two meanings in the creoles can be attributed to grammaticalization having proceeded in French rather than in the creoles themselves. In some of the French creoles, there has been no further development. In Mauritian French Creole, however, the future marker pou has risen in frequency and developed from indicating near future to definite future, i.e., away from being anchored in the present situation, and thus further down on the grammaticalization path (cf. Bybee et al. 1994, pp. 253ff.). Thus, the grammaticalization cline in Mauritian French Creole results partially from non-grammaticalizing reanalysis, and partially from language-internal grammaticalization. It is beyond the scope of the present chapter to discuss further details of pou and other French creole forms, or alternative analyses thereof. However, the interpretation of pou as having grammaticalized only partially in Mauritian French Creole is significant, as it indicates that even if a creole form displays several uses and meanings that can be plotted on different points of a grammaticalization scale, that does not necessarily mean that the entire development has actually taken place within the language itself. In the case of pou, the initial change appears to have been in the lexifier language. As will be discussed in the next section, we also need to take into account the possibility that grammaticalization patterns in the substrate provided a model for developments in the creole.

5

Substrate Influence and Grammaticalization

While grammaticalization is often viewed as a language-internal process, the role of language contact receives increasing attention. Heine and Kuteva (2003) review a variety of cases of grammaticalization involving contact, and propose to distinguish between the following processes: (11) a.

(i) ordinary contact-induced grammaticalization (ii) replica grammaticalization b. polysemy copying

In “ordinary contact-induced grammaticalization,” speakers of language A “notice” that language B, the model or source language, has a grammatical category, and, using material available in language A, gradually develop an equivalent category by drawing on universal strategies of grammaticalization, independent from the particular way the category came into being in the model B. One of Heine and Kuteva’s examples concerns Tayo, the French-based creole of New Caledonia. Here, the numeral de ‘two’ (< French

Grammaticalization in Pidgins and Creoles 401 deux) has been recruited to form dual personal pronouns, e.g., nu-de (lit. ‘we-two’) ‘we (dual)’, apparently in an attempt to reproduce the semantic category of dual which is obligatorily expressed in the Melanesian substrate (Heine & Kuteva 2003, after Corne 1995). In Melanesian Pidgin, a dual was created in a similar way, but as part of a more elaborate system that also included trial number and distinguished inclusive from exclusive (Keesing 1988, p. 141). To the extent that the respective substrate forms appear themselves to be related to numerals too (Heine & Kuteva 2003, p. 562, n. 6), it seems however that some of these cases could also be seen as instances of Heine and Kuteva’s second type of contact-induced grammaticalization: “replica grammaticalization.” In such cases, the speakers of the replica language repeat a grammaticalization process they assume to have taken place in the model language. An example is found in Basque, which, following the model of Spanish and French, developed a ‘have’-perfect using ukan ‘to have’ (after Haase 1992). For several other cases, including for example the use of interrogative pronouns to mark relative clauses in Basque, Heine and Kuteva offer an alternative analysis, namely that of “polysemy copying.” In such cases, there has not been a replication of the grammaticalization process, but rather of the polysemy pattern, by copying the initial and the final stages of the process. As argued by Heine and Kuteva, instances of polysemy copying – similar to what I refer to as “apparent grammaticalization” (see below) – do not actually involve a grammaticalization in the sense of a stepwise diachronic process. Let us now turn to Sranan again, and see how the case of locative items may be interpreted in terms of Heine and Kuteva’s classification.

5.1

“Apparent” and “replica” grammaticalization

The case of locative items in Sranan may serve to illustrate how a phenomenon may look like grammaticalization without actually resulting from a stepwise, language-internal process of grammaticalization. Many of the English locative prepositions (to, on, behind, etc.) were not retained in Sranan, and from the earliest Sranan sources onward until the middle of the twentieth century, simple locative prepositions with semantic content do not occur. Locative complements are either marked merely by the all-purpose preposition na (or a), which does not add anything to the meaning already expressed by the verb, or they take the form of a complex prepositional phrase (PP) containing a specifying locative item alongside na. The locative item – tappe (now spelled tapu) in example (12) – may either precede or follow the NP; it is, however, only the postnominal type that concerns us here. (12) go lydom na pappaja tappe go lie(down) loc (rush)mat top ‘Lie down on the mat.’ (Van Dyk, c.1765, p. 73)

402

Adrienne Bruyn

Some of the specifying locative items are derived from English nouns and can be fully lexical in Sranan too. These include tapu ‘top’, baka ‘back(part)’, and fesi ‘face’. When appearing in locative PPs, they are less lexical, expressing the relational meanings ‘on’, ‘behind’, and ‘before’, respectively. This layering of more lexical and more functional uses suggests that there has been grammaticalization of the kind described by Heine et al. (1991) for Ewe and other languages: body part (e.g., ‘face’) or landmark nouns (e.g., ‘top, sky’) becoming adpositions, thereby shifting from the domain of concrete, object concepts to the domain of space, and from there to the temporal domain. However, as argued in Bruyn (1995, 1996, 2003), there actually has not been a diachronic, language-internal development along these lines in the case of Sranan. The main argument concerns the fact that there has been no change in the distribution of the postnominal items between the eighteenth and the twentieth century. They did not develop further into true postpositions, as shown by the fact that they may not occur without na preceding the NP. Semantically, the eighteenth-century uses are no less abstract than the twentieth-century ones. Most importantly, it is not only items related to body part or landmark nouns that appear in eighteenth-century complex PPs, but also ini ‘in’, ondro ‘under’, and abra ‘opposite, across’ (< English in, under, over). (13) sinsi a komm na hosso inni since 3s come loc house in ‘since she entered the house’ (Schumann 1781, p. 96) In contrast to items such as tapu, the forms ini, abra, and ondro do not have nominal source forms, nor do they function as full nouns by themselves in Sranan. Thus, a grammaticalization starting with ini, ondro, and abra as lexical items can be ruled out. The fact that these items nevertheless only appear in complex PPs in the eighteenth-century text sources is further evidence that such PPs were well established quite early. At the same time, the question arises why these forms are used in complex PPs at all, and not as simple prepositions, like their English source forms. The answer lies in the way location is expressed in an important substrate for Sranan, viz. Gbe languages. Ewe, Fon, and other languages belonging to the Gbe cluster employ postnominal locative items. In view of the fact the Gbe languages constitute a substrate for Sranan, it appears no coincidence that the Sranan items remind us of the grammaticalization chain proposed by Heine et al. (1991) based on Ewe and other languages. Examples from Ewe include megbé, which has nominal properties, occurring both as a full noun ‘back (part)’ and as a postposition ‘behind’, as in (14), and té ‘under’, which according to Heine et al. (1991, p. 134ff.) is less lexical in terms of distribution and morphology. (14) é no déha le xO megbé 3s drink palm wine (be)at house behind ‘He drank palm wine behind the house.’ (Heine et al. 1991, p. 141)

Grammaticalization in Pidgins and Creoles 403 It appears, then, that the ambivalent categorial status of the locative elements in the Gbe languages – itself the result of grammaticalization – provided the model for Gbe speakers for the use of English-derived forms in similar ways in Sranan. The conceptually based universality of the metaphorical extension from body part and landmark nouns to adpositions would evidently have favored the transfer, but no gradual transition need be assumed to have taken place within Sranan itself. The assumption that the substrate provided the model can account as well for the fact that ini ‘in’, ondro ‘under’, and abra ‘opposite, across’ came to function in complex PPs in Sranan rather than as simple prepositions, something which would otherwise be difficult to explain. In previous work I have used the term “apparent grammaticalization” to denote this kind of development: grammaticalization is at issue in the sense that lexical items have assumed a relational function, but without the usual implication of a language-internal process of grammaticalization. According to Heine and Kuteva, the fact that items such as Sranan baka ‘back, behind’ as well as their substrate models are ambiguous between noun and relational item would reflect the process of replica grammaticalization. However, as there are no indications that the locative items actually developed in the typically gradual, stepwise fashion, I would rather argue that their notion of polysemy copying is appropriate here. The fact that the inventory of the Sranan items and their morphosyntactic status does not correspond exactly to their counterparts in the Gbe languages does not appear to be an argument against polysemy copying. Given the differences in detail between the various Gbe languages, which may have contributed to some variation among emerging varieties of the creole, together with the fact that, obviously, not all involved were speakers of Gbe languages, a certain degree of leveling and restructuring of the early creole varieties is only to be expected. It may also be the case that the polysemy copying did not involve individual items, but rather a systematic pattern of polysemy holding between postnominal locatives and their lexical sources in Gbe. At any rate, the cases of replica grammaticalization presented by Heine and Kuteva generally involve longterm contact and large-scale bilingualism, characteristics not applicable to the situation in which Sranan developed. It could be argued that polysemy copying, of which Heine and Kuteva suggest that it is less usual, is favored in the specific circumstances in which a radical creole evolves with a relatively homogeneous substrate. Substrate influence and internal development need not exclude each other, and creole features originally instigated by grammaticalization patterns in the substrate may develop further language-internally. A case in point is the complementizer function of taki in Sranan. Modern Sranan taki (< talk) can be a lexical verb ‘to say, to talk’, and a complementizer, ‘that’. It has been suggested, notably by Plag (1993, 1995, but cf. Plag 2002), that the latter function derives from a mainly language-internal grammaticalization process, which started with speech act verbs and then

404

Adrienne Bruyn

extended to other uses. An alternative interpretation of the historical data is, however, that the Gbe languages played a significant role in the first crucial step of the grammaticalization, viz. that taki was selected to function as a complementizer by means of polysemy copying, and that this was subsequently followed by further grammaticalization, mainly as a consequence of later stylistic expansion of the language (Bruyn 1995, 1996, 2003). Naturally, a precise assessment of the role of the substrate is all the more difficult if, as in this case, it concerns a grammaticalization process which is widely attested. As emphasized by Heine and Kuteva (2003), contact-induced grammaticalization and internal grammaticalization may proceed jointly, and they should not be regarded as mutually exclusive. This should not deter us however from trying to find out which kind of grammaticalization has been responsible in individual cases. In particular in P/C studies, ignoring the possibility of a contribution of the substrate may yield a biased picture of the development of a language. In this connection it should be mentioned that substrate patterns may also have played a part in some of the changes discussed in the preceding sections. With respect to the anterior marker té, which Detges (2000) relates to a French model, Lefebvre (1998) argues that its function in Haitian is, rather, modeled on the Gbe substrate. Furthermore, it may be worthwhile to compare the various grammaticalizations of fini as exhibited by the French creoles with the respective substrates. With regard to several of the developments in the varieties of Melanesian Pidgin mentioned in section 3, various authors have pointed out similarities with the substrate (e.g., Sankoff & Brown 1976; Meyerhoff 2002; see also Sankoff 1994). Thus, Keesing (1988, pp. 182ff.; 1991) argues that a common Oceanic pattern has led to shortcuts in the grammaticalization process of baimbai/bai in Solomons Pidgin and other Melanesian varieties. Indeed, a contribution by the substrate may provide an explanation for Romaine’s (1995) finding for Tok Pisin that the synchronic variation does not exactly reflect the diachronic changes, and that rural second-language speakers are ahead in the development in the preverbal position of baimbai/bai. Yet another indication that the substrate may well have played a role is that, although it does occur elsewhere, an adverb ‘later’ is not an often attested source for future or irrealis markers (Bybee et al. 1994; Heine & Kuteva 2002).

5.2

A case of reversal of direction

The last case to be discussed may serve to illustrate how polysemy copying may also account for a development that has proceeded in a direction contrary to the one that is typical for grammaticalization. Some Sranan verbs expressing location (including motion) have their etymological source in English or Dutch forms that are not verbs but rather prepositions, verbal particles, or adverbs. Some of these are given in (16):

Grammaticalization in Pidgins and Creoles 405 (16) abra ‘to go over, to cross’ < English over (preposition, particle) doro ‘to pass’, ‘to perform’, < Dutch door ‘through’ (preposition, ‘to arrive’ particle) opo ‘to rise’, ‘to raise, to lift up’ < English up (Dutch op) (preposition, particle) romboto ‘to surround’ < English roundabout (adverb, adjective) lontu ‘to surround’ < Dutch rond ‘(a)round’ (adverb, preposition) Given that, normally, we would expect verbs to develop into prepositions, particles, or adverbs, but not the other way around, this evidently calls for an explanation. Even though English adverbs such as up or down can be used as verbs – for example, to up the ante, to down a beer – such occurrences are too marginal to be a likely source for the Sranan cases, some of which do not correspond to items functioning as verbs in English in any case. However, in the Gbe languages, some verbs have grammaticalized into so-called verbids: forms which have a verbal origin but have lost some of their verbal properties when appearing after a main verb (Heine et al. 1991; Lord 1993). It appears that here as well, a substrate grammaticalization pattern provided a channel for the transfer into Sranan (Bruyn 2003). In contrast with the cases reviewed above however, in this case there has been a reanalysis in the opposite direction, i.e., contra the process of grammaticalization as it is assumed to have taken place in the substrate.

6 Conclusion Somewhat paradoxically, the insights of the grammaticalization approach may contribute to our understanding of P/C development precisely in those cases where a development in a P/C in some manner deviates from what would have been expected on the basis of the grammaticalization framework. In various of the cases reviewed above, such unexpected properties or developments become understandable when substrate influence is taken into account. In the case of the reanalysis of particles into verbs as just discussed, where the direction of the entire development is unexpected, it is unlikely that this would be (wrongly) seen as internal grammaticalization. Other cases however, such as that of the postnominal locative items in Sranan, and the dual function of taki as a verb and a complementizer, appear to invite an interpretation as language-internal grammaticalization, whereas actually the substrate has played a more or less decisive role. It is thus important to bear in mind that a synchronic pattern that appears to reflect grammaticalization, even when supported by cross-linguistic evidence, is not sufficient evidence that grammaticalization has actually taken place as a language-internal process. In order to determine whether “ordinary” grammaticalization is at issue in a particular case, or that the development is in some way induced by the substrate, or that ordinary

406

Adrienne Bruyn

and substrate-induced grammaticalization have reinforced each other, we need to establish the starting point of the development, trace the change through time, and determine which part of the change has proceeded in which language. In the context of P/C studies, it would be unrevealing if no distinction were made between internal and contact-induced grammaticalization, even if that sometimes proves to be difficult. Similarly, it is important to establish the extent to which grammaticalizations in a P/C are continuations of developments already begun in the lexifier language, as part of the broader issue of how to assess the relationship between a creole and its lexifier language. Distinguishing reanalysis from grammaticalization in the way suggested by Detges (2000) brings some clarity in this area. As for the grammaticalization processes that have proceeded internally in the P/Cs, there is no reason to consider them unique to P/C development. Still, there are certain aspects in which they differ from similar processes in languages with a longer history. They appear to proceed faster than is normally the case, which can be understood in light of the fact that P/C development involves the expansion of a more or less reduced language system. This same situation also accounts for the fact that many of the changes discussed involve the creation of new categories, something which is uncommon in non-P/Cs. Furthermore, contact, in this case substrate influence, plays a role in relatively many instances of grammaticalization in P/Cs. Processes may start out on the basis of a substrate model and subsequently develop further within the P/C, with or without further shortcuts. In the extreme case, that of polysemy copying or apparent grammaticalization, where the entire development can be seen as a large shortcut, it may be argued that grammaticalization as a process is no longer at issue. In this chapter much attention has been paid to cases that differ in one way or another from processes of grammaticalization as they normally take place in older languages. This should certainly not distract from the fact that ordinary grammaticalization occurs in P/Cs too. However, the mere existence of a grammaticalization pattern is not sufficient evidence that ordinary grammaticalization has indeed proceeded within the language. Therefore, it is important to trace the actual developments in individual cases, in order to establish the sources of the P/C features and the nature of the processes and mechanisms involved. In that way, we may discover more about the ways in which the speakers of P/Cs shaped their language.

NOTE Support from the Netherlands Organization for Scientific Research (NWO) enabled me to write this chapter. I am grateful to the editors for their useful comments on earlier versions.

Grammaticalization in Pidgins and Creoles 407

REFERENCES Arends, Jacques (1993) Towards a gradualist model of creolization. In: Francis Byrne and John Holm (eds.) Atlantic Meets Pacific: A Global View of Pidginization and Creolization. Amsterdam: John Benjamins, pp. 371–80. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1988) Creole languages and the bioprogram. In: Frederick J. Newmeyer (ed.) Linguistics: The Cambridge Survey. Vol. 4, Linguistic Theory: Extensions and Implications. Cambridge: Cambridge University Press, pp. 268–84. Bruyn, Adrienne (1995) Grammaticalization in Creoles: The Development of Determiners and Relative Clauses in Sranan. Amsterdam: IFOTT. Bruyn, Adrienne (1996) On identifying instances of grammaticalization in creole languages. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Related to Grammaticalization in Contact Languages. London: University of Westminster Press, pp. 29– 46. Bruyn, Adrienne (2003) Grammaticalisation, réanalyse et influence substratique: quelques cas du Sranan [Grammaticalization, reanalysis, and substrate influence: some cases from Sranan]. In: Sibylle Kriegel (ed.) Grammaticalisation et réanalyse: Approches de la variation créole et française. CNRS Éditions, Paris, pp. 25–47. Bruyn, Adrienne (2007) Bare nouns and articles in Sranan. In: Marlyse Baptista and Jacqueline Guéron (eds.) Noun Phrases in Creole Languages: A Multi-Faceted Approach. Amsterdam: John Benjamins, pp. 339–81. Bybee, Joan, William Pagliuca, and Revere Perkins (1991) Back to the future. In: Traugott and Heine (eds.), pp. 17–58. Bybee, Joan, Revere Perkins, and William Pagliuca (1994) The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press. Campbell, Lyle (ed.) (2001) Grammaticalization: A Critical Assessment. Special issue of Language Sciences 23 (2/3). Corbett, Greville G. (2000) Number. Cambridge: Cambridge University Press. Corne, Chris (1995) A contact-induced and vernacularized language: How Melanesian is Tayo? In: Philip Baker (ed.) From Contact to Creole and Beyond. London: University of Westminster Press, pp. 121–48. Detges, Ulrich (2000) Two types of restructuring in French creoles: A cognitive approach to the genesis of tense markers. In: Neumann-Holzschuh and Schneider (eds.), pp. 135–62. Detges, Ulrich and Richard Waltereit (2002) Grammaticalization vs. reanalysis: A semantic-pragmatic account of functional change in grammar. Zeitschrift für Sprachwissenschaft 21, 151–95. De Drie, Aleks (1985) Sye! Arki tori! [Silence! Listen to my story!] Compiled by Trudi Guda. Paramaribo: Afdeling Cultuur Studies van het Ministerie van Onderwijs, Wetenschappen en Cultuur. Dutton, Thomas E. (1983) Birds of a feather: A pair of rare pidgins from the Gulf of Papua. In: Ellen Woolford and William Washabaugh (eds.) The Social Context of Creolization. Ann Arbor: Karoma, pp. 77–105. Fischer, Olga, Muriel Norde, and Harry Perridon (eds.) (2004) Up and Down the Cline: The Nature of Grammaticalization. Amsterdam: John Benjamins.

408

Adrienne Bruyn

Gamillscheg, Ernst (1957) Historische französiche Syntax [Historical French syntax]. Tübingen: Max Niemeyer. Givón, Talmy (1971) Historical syntax and synchronic morphology: An archeologist’s field trip. Papers from the Seventh Regional Meeting, CLS 7, 394–415. Givón, Talmy (1979) On Understanding Grammar. New York: Academic Press. Givón, Talmy (1981) On the development of the numeral “one” as an indefinite marker. In: Hagit Borer and Youssef Aoun (eds.) MIT Working Papers in Linguistics, vol. 3: Theoretical Issues in the Grammar of Semitic Languages. Cambridge, MA: Massachusetts Institute of Technology, pp. 233–55. Givón, Talmy (1984) Syntax: A Functional-Typological Introduction (vol. 1). Amsterdam: John Benjamins. (Revised edn. published 2001.) Haase, Martin (1992) Sprachkontakt und Sprachwandel im Baskenland: die Einflüsse des Gaskognischen und Französischen auf das Baskische [Language contact and language change in the Basque Country: Gascon and French influence on Basque]. Hamburg: Buske. Heine, Bernd and Tania Kuteva (2002) World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. Heine, Bernd and Tania Kuteva (2003) On contact-induced grammaticalization. Studies in Language 27, 529–72. Heine, Bernd, Ulrike Claudi, and Friederike Hünnemeyer (1991) Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press. Hopper, Paul. J. (1991) On some principles of grammaticization. In: Traugott and Heine (eds.), pp. 17–35. Hopper, Paul. J. and Janice Martin (1987) Structuralism and diachrony: The development of the indefinite article in English. In: Anna Giacalone Ramat, Onofrio Carruba, and Giuliano Bernini (eds.) Papers from the Seventh International Conference on Historical Linguistics. Amsterdam: John Benjamins, pp. 295–304. Hopper, Paul. J. and Elizabeth Closs Traugott (2003) Grammaticalization, 2nd edn. Cambridge: Cambridge University Press. Jourdan, Christine (1985) Creolization, nativization or substrate influences: What is happening to Bae in Solomon Islands Pijin? Pacific Linguistics, A72: Papers in Pidgin and Creole Linguistics 4, pp. 67–96. Keesing, Roger M. (1988) Melanesian Pidgin and the Oceanic Substrate. Stanford: Stanford University Press. Keesing, Roger M. (1991) Substrates, calquing and grammaticalization in Melanesian Pidgin. In: Traugott and Heine (eds.), pp. 315–42. Lefebvre, Claire (1998) Creole Genesis and the Acquisition of Grammar. Cambridge: Cambridge University Press. Lord, Carol (1993) Historical Change in Serial Verb Constructions. Amsterdam: John Benjamins. Lyons, Christopher (1999) Definiteness. Cambridge: Cambridge University Press. Meyerhoff, Miriam (2002) All the same? The emergence of complementizers in Bislama. In: Tom Güldemann and Manfred von Roncador (eds.) Reported Discourse: A Meeting Ground for Different Linguistic Domains. Amsterdam: John Benjamins, pp. 341–59. Mufwene, Salikoko S. (1996) Creolization and grammaticization: What creolistics could contribute to grammaticization. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Related to Grammaticalization in Contact Languages. London: University of Westminster Press, pp. 5–28.

Grammaticalization in Pidgins and Creoles 409 Mufwene, Salikoko S. (2001) The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mühlhäusler, Peter (1980) Structural expansion and the process of creolization. In: Albert Valdman and Arnold Highfield (eds.) Theoretical Orientations in Creole Studies. New York: Academic Press, pp. 19–55. Mühlhäusler, Peter (1981) The development of the category of number in Tok Pisin. In: Pieter Muysken (ed.) Generative Studies on Creole Languages. Dordrecht: Foris, pp. 35–84. Mühlhäusler, Peter (1997) Pidgin and Creole Linguistics, expanded revised edn. London: University of Westminster Press. Neumann-Holzschuh, Ingrid and Edgar W. Schneider (eds.) (2000) Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins. Neumann-Holzschuh, Ingrid and Edgar W. Schneider (2000) Introduction: “Degrees of restructuring” in creole languages? In: Neumann-Holzschuh and Schneider (eds.), pp. 1–18. Plag, Ingo (1993) Sentential Complementation in Sranan. Tübingen: Max Niemeyer. Plag, Ingo (1995) The emergence of taki as a complementizer in Sranan: On substrate influence, universals, and gradual creolization. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 113–48. Plag, Ingo (2002) On the role of grammaticalization in creolization. In: Glenn G. Gilbert (ed.) Pidgin and Creole Linguistics in the Twenty-First Century. New York: Lang, pp. 229–46. Romaine, Suzanne (1995) The grammaticalization of irrealis in Tok Pisin. In: Joan Bybee and Suzanne Fleischman (eds.) Modality in Grammar and Discourse. Amsterdam: John Benjamins, pp. 389–427. Sankoff, Gillian (1977) Variability and explanation in language and culture: Cliticization in New Guinea Tok Pisin. In: Muriel Saville-Troike (ed.) Linguistics and Anthropology. Washington, DC: Georgetown University Press, pp. 59–74. (Reprinted in Sankoff 1980, pp. 257–70.) Sankoff, Gillian (1980) The Social Life of Language. Philadelphia: University of Pennsylvania Press. Sankoff, Gillian (1994) An historical and evolutionary approach to variation in the Tok Pisin verb phrase. In: Katharine Beals (ed.) Papers from the Thirtieth Regional Meeting of the Chicago Linguistic Society. Vol. 2: The Parasession on Variation in Linguistic Theory, CLS 30, 293–321. Sankoff, Gillian (1991) Using the future to explain the past. In: Francis Byrne and Thom Huebner (eds.) Development and Structures of Creole Languages. Amsterdam: John Benjamins, pp. 61–74. Sankoff, Gillian and Penelope Brown (1976) The origins of syntax in discourse: A case study of Tok Pisin relatives. Language 52, 631–6. (Reprinted in Sankoff 1980, pp. 211–55.) Sankoff, Gillian and Suzannne Laberge (1973) On the acquisition of native speakers by a language. Kivung 6, 32–47. (Reprinted in Sankoff 1980, pp. 195–209.) Schumann, Christian L. (Transl.) (1781) Die Geschichte unsers Herrn und Heilandes Jesu Christi, aus den vier Evangelisten zusammengezogen, durch S. Lieberkühn; in NegerEnglische Sprache übersezt, zum Gebrauch bey der Neger-Gemeine [The story of our Lord and Saviour Jesus Christ, based on the four Evangelists, by S. Lieberkühn; translated into Negro-English, to be used by the black congegration]. Manuscript, Paramaribo (Moravian Archives Utrecht 12/617).

410

Adrienne Bruyn

Singler, John V. (1992) Nativization and pidgin/creole genesis: A reply to Bickerton. Journal of Pidgin and Creole Languages 7, 319–33. Smith, Norval (2001) Voodoo Chile: Differential substrate effects in Saramaccan and Haitian. In: Norval Smith and Tonjes Veenstra (eds.) Creolization and Contact. Amsterdam: John Benjamins, pp. 43–80. Sylvain, Suzanne (1979 [1936]) Le créole haïtien: Morphologie et syntaxe [Haitian Creole: Morphology and syntax]. Geneva: Slatkine. Traugott, Elizabeth Closs and Bernd Heine (eds.) (1991) Approaches to Grammaticalization. Vol. 2: Focus on Types of Grammatical Markers. Amsterdam: John Benjamins. Van Dyk, Pieter [ca 1765]. Nieuwe en nooit bevoorens geziene onderwyzinge in het Bastert Engels, of Neeger Engels, zoo als het zelve in de Hollandsze Colonien gebruikt word [New and unprecedented instruction in Bastard English, or Negro English, as it is used in the Dutch colonies]. Amsterdam: Erven de Weduwe Jacobus van Egmont. Woolford, Ellen (1979) The developing complementizer system of Tok Pisin: Syntactic change in progress. In: Kenneth C. Hill (ed.) The Genesis of Language. Ann Arbor: Karoma, pp. 108–24.

FURTHER READING Baker, Philip and Anand, Syea (eds.) (1996) Changing Meanings, Changing Functions: Papers Related to Grammaticalization in Contact Languages. London: University of Westminster Press. Bruyn, Adrienne (1995) Relative clauses in early Sranan. In: Jacques Arends (ed.) The Early Stages of Creolization. Amsterdam: John Benjamins, pp. 149–202. Jourdan, Christine and Keesing, Roger (1997) From Fisin to Pijin: Creolization in process in the Solomon Islands. Language in Society 26, 401–20. Kriegel, Sibylle (ed.) (2000) L’évolution grammaticale des créoles: Problèmes et perspectives [Grammatical evolution in creoles: Problems and perspectives]. Special issue of Études Créoles 23(2). Kriegel, Sibylle (ed.) (2003) Grammaticalisation et réanalyse: Approches de la variation créole et française [Grammaticalisation and reanalysis: Approaches to variation in creoles and French]. Paris: CNRS Éditions. Meyerhoff, Miriam (2001) Another look at the typology of serial verb constructions: The grammaticalization of temporal relations in Bislama (Vanuatu). Oceanic Linguistics 40, 247–68. Michaelis, Susanne (2000) The fate of subject pronouns: Evidence from creole and non-creole languages. In: Neumann-Holzschuh and Schneider (eds.), pp. 163–83. Romaine, Suzanne (1999) The grammaticalization of the proximative in Tok Pisin. Language 77, 322–46.

Creoles, Markedness, and Default Settings 411

17

Creoles, Markedness, and Default Settings: An Appraisal ALAIN KIHM

1 Introduction: Which Markedness? The notion that creole languages are “unmarked” languages is a widespread one. What sense can be made of such a claim is not easy to decide, however, because “(un)marked” actually has more than one meaning or use (Moravcsik & Wirth 1986; Battistella 1996). First, there is the meaning the term was given in phonology when it was first introduced by Trubetzkoy (1939) and then extended by Jakobson, whose compact definition we may quote: One of the essential properties of phonological correlations consists in the fact that both members of a correlation pair are not equivalent: one member includes the relevant mark, while the other does not; the former is thus defined as marked, the latter as unmarked. (Jakobson 1932/1966, p. 22, my translation; also see Anderson 1985, pp. 106ff.)

Creole languages are certainly not globally unmarked in the sense that they include only unmarked phonemes, an obvious falsehood. What is true, on the other hand, is that whenever the lexifier language shows highly marked phonemes, such as the French front rounded vowels /y/ and /ø~œ/, these phonemes are not usually present in the related creole (see Smith, this volume). Marked syllable types such as CVCC syllables are also regularly unmarked in the direction of CV or CVC (compare French quatre vs. Haitian kat ‘four’, English first vs. fos and fosi in a variety of English-based creoles, etc.). There is thus a sense in which creole languages are locally less marked in the Jakobsonian use of the term than are their lexifiers. I will have little to say about phonological markedness, however, because I find its import as far as creolization is concerned hard to assess. Indeed, the sample on which we base our observations is biased by the fact that the lexifiers are mainly Indo-European languages with rather marked phonologies in their standard and/or present varieties.

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

412

Alain Kihm

(I qualify thus because other varieties may be less marked on some counts – e.g., [kat] for quatre was standard in seventeenth-century French and is now common in colloquial French when the following word begins with a consonant, and even when it does not – so the choice of a standard of comparison is itself open to question.) A few Indo-European-lexified creoles present highly marked phonemes, e.g., /D/ and /θ/ in Portuguese-based Angolar of São Tomé (see Rougé 2004). These can be ascribed to transference from a Bantu substrate. Similarly, the implosive, prenasalized, and coarticulated stops of the Portuguese and English-lexified creoles Saramaccan, Ndyuka, Angolar, Principense, can be ascribed to transference from a West or Central African substrate (see Smith, this volume). When the lexifiers belong to other stocks with less marked phonologies, as in Bantu- or Oceanic-based creoles (e.g., Kituba, Hiri Motu), nothing much of interest seems to happen in this regard, except for the frequent, but by no means unexceptional, loss of lexical tones when the lexifier has them. In Arabic-based creoles, the highly marked emphatic consonants did not survive, but neither did they in Maltese which cannot be called a creolized language by any reasonable definition of what a creole is. Another sense of markedness is statistical (Greenberg 1966). Properties observed in all or at least in most known languages are said to be unmarked; properties which only occur in a minority subset of languages are regarded as marked. Here, markedness criteria are no longer limited to phonology (and morphology). For instance, having nouns and verbs certainly is an unmarked property since no known language seems to be entirely devoid of this contrast. Having a clear-cut open class of adjectives, on the other hand, may be considered a marked trait, since in many languages words denoting qualities are morphologically and syntactically nondistinct from nouns (e.g., Arabic) or verbs (e.g., Mandarin Chinese, Wolof). The Indo-European languages with their attributive adjectives which show special inflections (when they inflect, as in German or Russian) and which enter a particular construction with the noun they modify distinct from both noun-plus-noun and noun-plus-relative clause constructions, seem definitely to belong to a minority. I chose this example on purpose because, as it turns out, Indo-European-based creoles can hardly be considered unmarked according to the criterion it sets, insofar as they seem to share a well-defined adjective category with their lexifiers (e.g., Damoiseau 1996; DeGraff forthcoming, on Haitian). Although this would obviously require a more extensive inquiry than I can attempt here, Greenbergian statistical markedness thus does not seem to contribute much that would be specific to creoles or to creolization per se. Finally, there is a third sense of markedness, clearly emerging from the following quotation from Chomsky (1981, p. 126): It is reasonable to suppose that UG [Universal Grammar], understood to be one aspect of the genetically-determined human biological endowment, determines a set of core grammars, and that what is actually represented in the mind of an individual even under the idealization to a homogeneous speech community

Creoles, Markedness, and Default Settings 413 would be a core grammar with a periphery of marked elements and constructions. A core grammar, once again, is determined by fixing the parameters of UG in one or another of the permitted ways.

Chomsky exemplifies his point by noting that John read his books is unmarked with respect to UG because the possessive pronoun his is free in its governing category, i.e., the NP [his books], as it may be co-indexed with John or with some other person not mentioned in the sentence. In contrast, They read each other’s books is marked because the anaphor each other is not locally bound within the NP [each other’s books]. Markedness so understood would appear to be a theory-internal concept, since it has to do with the way theoretical principles – here the binding module of Government and Binding (GB) theory – are supposed to apply. Yet, there is a connection with the previous, statistical sense, insofar as UG-marked constructions can be shown to be rare constructions as well, as pointed out by Chomsky himself (1981, pp. 140–1). If the proposition that creoles, or rather, creole grammars are unmarked is true, then it will have to be in this latter sense rather than in the sense of substantial or formal markedness in correlation pairs as defined by Jakobson, since, as mentioned, creole languages do show marked phonemes or syllable types, even if they generally eschew “highly” marked phonemes or syllables, and they include marked categories such as adjectives. What unmarkedness in the third sense implies is that creoles should contain no “peripheral,” noncore constructions, with the possible upshot that all the constructions they do contain are common across the world’s languages. I say “possible” because the correlation between being unmarked-core and being widespread is not a necessary one. This sets Chomskyan markedness apart from the Greenbergian, statistical version of the concept. Indeed, one may conceive of a state of affairs such that core grammar properties are actually represented in only a small group of languages, including the creoles or perhaps comprising them exclusively, whereas all other languages would resort to peripheral devices for these properties (on markedness and frequency, see Battistella 1996, pp. 49– 53). The likeliest cause for such a state of affairs is language change. Of course, the empirical task of showing that the minority grammars do express core grammar might prove formidable, but this is a different issue than the point of principle I am now making. These reflections invite us to consider again the Bickertonian view that creole grammars directly express the genetic program (“bioprogram”) for language. At first glance this view simply removes the issue, since it makes creole languages unmarked by definition. But note that this distinctive character of creole grammars follows from the very special way in which they are supposed to have emerged, viz. in a situation where primary data “provide no basis for the setting of parameters” (Bickerton 1999, p. 55). The hypothetical state of affairs just described then becomes the one to be expected. In all situations but the creole situation children are exposed to primary data that more often than not do not actually reflect the default parameter settings of

414

Alain Kihm

the unadulterated Bioprogram. A somewhat paradoxical consequence would be then that creole languages turn out to be (statistically) marked by the very fact of being unmarked amidst languages all of which were marked (pun intended) by the scars of history. Nothing of importance follows from it, however, because – and this is the gist of Bickerton’s (1999) article – the whole markedness issue simply is irrelevant for creole languages if the Bioprogram is assumed.

1.1

The logic of markedness

All this points to a deeper issue. The logic of markedness makes it a binary, or at most ternary relation: [+] contrasts with [–], and both may further contrast with being unspecified for the property in question. True, this claim can be challenged. Battistella (1996) shows that in the transition from the Jakobsonian to the Chomskyan paradigm markedness lost its strictly binary character and became multivalued. Yet, it remained a generally acknowledged truth that there must be at most one unmarked (or least marked) option, insofar as having several unmarked members in the contrast would definitely void the notion. But note that it would be as severely voided if the number of marked (or less unmarked) options was allowed to have no upper boundary and to vary in an unconstrained fashion. In other words, for there to be a meaningful contrast, the marked options must be limited in number and, more importantly, they must bear a definite inherent relation to the unmarked option. This contrast of u = 1 vs. {m}, {m}, a closed set whose members can be predicted given u, and vice versa, we may call extended binarism. A different type of relation is set up by the concept of default which Bickerton uses. If one instantiation of a given property, say anaphoricity, is chosen by default – i.e., obligatorily, unless there is a definite reason not to do so – then there is no a priori limit on the number and types of the nondefault instantiations of the property. To put it another way, [+voice] contrasts uniquely with [–voice]; unmarked Nominative Case contrasts nonuniquely with all marked cases, but they “count” as one unit as far as markedness is concerned. Should we agree, on the other hand, that each other’s books is a marked construction, this does not tell us what would be the unmarked construction to express the same idea. The possibilities are numerous: using a phrase that is semantically, but not syntactically a reciprocal anaphor (cf. French Ils ont lu les livres les uns des autres – not bad, but not great!), expliciting the problem away (cf. French Chacun a lu le(s) livre(s) de l’autre/des autres, or, more ambiguously, GuineaBissau Portuguese Creole Kada un lei libru di utru ‘Each one read the book of the other’), affixing a reciprocal morpheme to the verb, etc. More important than the sheer number of alternatives is the fact that they do not form a natural class by any standard. Naturally there may be cases where the distinction between default and markedness blurs, because there happens to be only one nondefault option. But the logic of the distinction should remain clear.

Creoles, Markedness, and Default Settings 415

1.2

“Markedness” versus “default” in the syntax

This of course raises the issue of syntactic markedness. Does the notion make sense? An important question in general, but a vital one from our standpoint since, as we saw, creoles’ global unmarkedness – as distinct from being more or less marked in some areas – can only be assumed at the syntactic level, not in phonology or morphology. It is therefore a crucial step to determine whether the relevant concept for syntax is “markedness” (as in phonology and morphology), or the much less constrained notion of “default.”1 Two rather different views of syntactic variation and acquisition follow. Under one view, UG includes a set of ideally binary parameters whose values must be set to [+] or [–], i.e., to marked or unmarked, by the children in accordance with the primary data they perceive. Since the values are not preset in the child’s mind, and being marked is not a reason for a value not to be chosen, it follows that syntactic parameters cannot be set in the absence of processable primary data. Under the other view, in contrast, one value or instantiation of the parameter is given by default, that is to say, the child is born with it. This is the one she/he will select unless some cue in the primary data (perhaps in the sense of Lightfoot 1999) forces her/him to modify the default setting in a certain way. The crucial upshot of the latter view is that it allows for language acquisition in the absence of primary data or, more accurately, in the presence of radically degenerate, unprocessable primary data, possibly reduced to no more than a syntactically unstructured vocabulary – the bare minimum, it seems, since children do not create words ex nihilo. (Why this is so is a fascinating question that ought to be pursued.) There is thus an inherent connection between the choice of default over markedness and the possible abruptness of creolization or language emergence in general, i.e., the assumption that children can tap their language faculty directly when the primary data they are exposed to are degenerate beyond what is usually the case. In return, the fact that full-fledged language does emerge in such adverse conditions – and the best evidence so far is for sign language (Kegl, Senghas, & Coppola 1999; Newport 1999) – is an indication that viewing markedness in the creoles in terms of unmarked languages is not the correct approach, that is unless “unmarked” actually means “default.”

2 Diagnostics for Syntactic Markedness Supposing the possibility that creole languages represent a case of syntactic unmarkedness in the sense of Chomsky’s quotation above, how would we know that they do?2 Viewed in current generative (minimalist) terms, syntax is the combination of two operations: Merge, i.e., combining two syntactic objects to yield another syntactic object; and Move, i.e., overtly displacing a syntactic object from one position to another. (Note that covert movement at

416

Alain Kihm

Logical Form does not count as real displacement.) Since Merge is the necessary minimum for there to be a syntax, it cannot be marked (i.e., it “comes for free”). Move must therefore be the marked operation, which makes sense from an economy of derivation viewpoint – as far as possible do nothing but the bare minimum – and also if displacement is considered an “imperfection” of language design (Chomsky 1998).3 But which kind of Move are we to consider? Even if Move is generally marked with respect to doing nothing but Merge, instances of movement that are present in both the creoles and their lexifiers clearly tell us nothing as far as the former’s unmarkedness relative to the latter is concerned. This eliminates (overt) wh-movement from consideration, since nearly all creoles show obligatory fronting of wh-words, even when it is optional in the lexifier as is the case in French. NP movement won’t qualify either, as some creoles even include instances of NP movement not present in their lexifiers: compare French (FR) with Haitian Creole (HC): (1) FR le chat qui est assis sur ma table det cat comp is seated on my table (2) HC chat ki chita nan tab mwen an cat comp sit loc table 1s det both ‘The cat that is sitting on my table.’ Under current accounts, the derivation of the clause-final position of the HC determiner must involve movement of the NP [chat ki chita nan tab mwen] over the determiner LA = [an].4 We are thus left with verb movement as a possible diagnostic tool. It has indeed been argued that in creole languages verbs never move from their base position as head of VP to adjoin to a higher Tense and/or Agreement head – known as “V-to-I” – even when they do so in the lexifiers. Note that, leaving aside non-Indo-European creoles, on which data are insufficient for our purposes, this criterion actually limits the relevant sample to Romance-based creoles, since verbs are not supposed to move in English either (Pollock 1989; Baptista 2002 on Portuguese-based Cape Verdean; DeGraff 1997 on Haitian and other French-based creoles).5 That verbs do not leave their base position in creole languages seems to be a robust fact. In order for the comparison to be significant, then, the question is: Do verbs really move in Romance lexifiers such as French and Portuguese? To try and answer this question, I will examine a proposal by Roberts (1999), who purports to demonstrate creole unmarkedness precisely with respect to verb movement (along with a few other syntactic phenomena, which I will briefly review later on). Relying on Williams’s (1994) reassessment of Emonds’s (1978) and Pollock’s (1989, 1997) analyses of the differences between English and French in this domain, I will show that Roberts’s arguments do not carry through because (1) they do not apply beyond French-based creoles – not

Creoles, Markedness, and Default Settings 417 even in Portuguese-based creoles; (2) the usual tests of adverb and negation placement are not reliable and yield contradictory results in any event; (3) his French and Haitian data are flawed. I will therefore conclude that the verb’s position cannot be taken as a clue to creole syntactic unmarkedness, and that, since it was the only clue, creoles do not differ from their lexifiers on that count. Yet, it is still a reasonable assumption that creole languages instantiate the default option for syntactic settings in a way that does not involve movement, but rather a measure of isomorphy between syntactic structure and morphophonological realization along lines set up in Muysken (1981b). This is what I will proceed to argue in section 4.

3 Verb Movement: A Test for Syntactic Markedness? The assumption that French finite main verbs and auxiliaries raise and adjoin to AGRS (the subject agreement head of the split inflectional projection IP) – and from there possibly move higher – whereas English finite main verbs stay in their basic position and only auxiliaries move, is founded on well-known contrasts involving predicate adverbs and negation (Pollock 1989, 1997). Thus, (3) and (4) illustrate the French order – main verb < negation/adverb – and the unacceptability of this order in English (ENG): (3) FR Jean (n’)embrasse pas souvent Marie. ENG *John kisses not often Mary. (4) FR *Jean (ne) pas souvent embrasse Marie. ENG John (does not) often kiss(es) Mary. (5)–(6) illustrate the same order – finite auxiliary < negation/adverb < nonfinite main verb – in French and English. (5) FR Jean (n’)a pas souvent mangé de chocolat. ENG John has not often eaten chocolate. (6) FR *Jean (ne) pas souvent a mangé de chocolat. ENG *John not often has eaten chocolate. The same parallelism is further observed in (7)–(8), where both French and English display negation < infinitive: (7) FR (Ne) pas manger de chocolat est mauvais pour la santé. ENG Not to eat chocolate is bad for your health.

418

Alain Kihm

(8) FR *manger (ne) pas . . . ENG *to eat not . . . Roberts’s (1999) account of this evidence relies on Feature Theory (Chomsky 1993). Infl contains strong or weak Verb features the function of which it is “to check the morphological properties of the verb selected from the lexicon” (Chomsky 1993, p. 28). Strong features, as in French, are phonologically “visible.” This means that they must be interpreted at Phonological Form (PF). Because they are not directly associated with phonological matrices they must overtly attract the main verb in order to be spelled out with it.6 In contrast, weak features, as in English, have no phonological visibility. This means that they need not – and Economy sees to it that they do not – overtly attract the main verb to be eliminated. Strength being correlated with the distinctive power of V-features, there is a relation, albeit not a straightforward one, between it and relative “richness” of inflection. That is why the auxiliaries have and be still move overtly to Infl in Modern English, because they are more richly inflected than main verbs and modals, i.e., they involve strong features. The difference does not exist in French, because all verbs are richly inflected. In his (1997) follow-up on his (1989) seminal article, Pollock gives a more precise definition of the impressionistic notions “rich inflection” and “strength,” whereby an inflectional category is said to be strong (“morphologically identified”) if “it alternates unambiguously [within a paradigm] with a least one distinct morpheme of the same inflectional category” (Pollock 1997, p. 269). Given this, it seems a natural, nay, a necessary step to assume that weak V-features are unmarked and strong V-features marked. Now consider the following Haitian examples from Roberts (1999, pp. 304–5) and DeGraff (1997, pp. 68, 75): (9) HC

Bouki deja pase rad yo. Bouki already iron cloth the.pl/3p ‘Bouki has already ironed the(ir) clothes.’

(10) HC Boukinèt pa renmen Bouki. Boukinèt neg love Bouki ‘Boukinèt does not love Bouki.’ A typical creole, Haitian shows even more inflectional poverty than English – in fact it shows no inflection at all. And it clearly parallels English, while diverging in an equal measure from its lexifier language, French:7 (11) FR Bouki repassait déjà le linge/ *déjà repassait (12) FR Boukinèt (n’)aime pas Bouki/ *pas aime

Creoles, Markedness, and Default Settings 419 DeGraff (1997, pp. 68–70) demonstrates the same parallel or contrast with other time or quantifying adverbs such as toujou ‘always’, preske ‘almost’, trè ‘a lot’, etc. (see also Lefebvre 1998, pp. 351–5). Roberts’s and DeGraff’s conclusion is that Haitian, like Modern English and unlike French, lacks V-to-I movement, from which Roberts draws the additional conclusion that Haitian, like Modern English, instantiates the unmarked parameter setting for V-feature strength. He further claims that the causes for the difference between Early Modern French and Haitian are the same as those for the change from Middle English to Modern English: loss of inflections, and the presence in Infl of a class of verbal elements – auxiliary have and be and do-support in English, Tense-Mood-Aspect markers similar to auxiliaries in Haitian – leading the child to infer that main verbs need not move, hence do not move.

3.1

Verb placement in Portuguese and Portuguese-lexifier creoles

There are several problems with this account. One is that it cannot be empirically generalized beyond French-based creoles – assuming all of them to behave like Haitian in this respect, for which there is no clear-cut evidence to date. For instance, Gadelii’s (1997) careful study of Lesser Antillean French Creole suggests that items like ja ‘already’ (equivalent to Haitian deja) actually are “aspectuals” rather than typical adverbs, and that identifying the latter class in these languages is not such an easy job (pp. 195–202). English-based creoles of course are irrelevant, as already mentioned. The obvious source of supportive data are the Portuguese-based creoles. Unfortunately, Pollock’s tests for V movement do not readily apply to Portuguese (PT). First, Portuguese has a preverbal negation that gives no clues as to verb movement: (13) PT O João não passou a roupa a ferro. ‘John did not iron the clothes.’ Second, adverb placement in Portuguese is a complex matter. Directly relevant for us is the fact that adverbs unambiguously modifying the predicate may precede the main verb (Costa 1996), again offering no primary evidence that V has to move to Infl: (14) PT O João sempre passa a roupa a ferro. ‘John always irons the clothes.’ Naturally, this is not to deny that main verbs may be considered to move in Portuguese. Thus, (15) illustrates inversion: (15) PT Passou o João a roupa a ferro? ‘Did John iron the clothes?’

420

Alain Kihm

The point is there is no hard-and-fast test for specific V-to-I movement, also known as short V movement, as opposed to long V movement to C (Complementizer) as in inversion.8 Portuguese-based creoles reflect this state of affairs, in particular Cape Verdean Creole (CVC). Negation in Cape Verdean precedes main verbs and all TMA markers as shown in (16) (Baptista 2002, p. 107): (16) CVC E ka ta pode djuda-m. 3s neg asp can help-1s ‘She cannot help me.’ Recall that negation also precedes main verbs in Haitian (as in (10)). These data suggest a weak V-feature in Infl, but this says nothing as to the hypothesis that creolization basically consists in substituting unmarked features for marked ones, since Portuguese looks just as unmarked – if that is what there is to it. Compare the Portuguese placement of negation in (17): (17) PT Ela não pode ajudar-me. ‘She cannot help me.’ There is one exception in Cape Verdean Creole, and that is the copula e ‘be’ which precedes negation (Baptista 2002, p. 105): (18) CVC João e ka padri. João cop neg priest ‘João is not a priest.’ The reverse order *ka e [neg cop] is ungrammatical. Whatever the explanation may be for these contrastive orderings, Roberts would have to agree that Cape Verdean exhibits markedness precisely where his theory expects none.9 Adverb placement is equally inconclusive. Baptista (2002, pp. 127–31) distinguishes six classes of adverbs with respect to their distribution. Class I adverbs are the most versatile. Thus, class I senpri ‘always’ may occur in all the bracketed positions shown in (19) (Baptista 2002, p. 128): (19) CVC (Senpri) João (senpri) bebe (senpri) vinhu (senpri). (always) John (always) drink (always) wine (always) ‘João always drank wine.’ Baptista points out that clause-initial senpri – and presumably preverbal senpri as well – quantifies over the event (‘It was João’s habit to drink wine’), whereas post-verbal senpri quantifies over the predicate “drink wine” (‘It was wine that João had always been drinking’), and that clause-final senpri is neutral between these two interpretations.

Creoles, Markedness, and Default Settings 421 These data are significant on several counts. First, they demonstrate that the relative position of the verb and the adverb cannot be considered the mechanical consequence of moving the former over the latter in order to meet checking requirements – in that case, we would not expect the relative positions to have anything to do with the adverb’s scope. (The same is probably true for English and French – see Costa 1996). Second, even keeping to the V-feature account, Cape Verdean would appear now marked ( João bebe senpri vinhu), now unmarked ( João senpri bebe vinhu), exactly like its lexifier (cf. Portuguese O João bebe sempre vinho, O João sempre bebe vinho).

3.2

Adverb placement

The discussion above shows that Portuguese-based creoles offer no independent support for Roberts’s markedness account of the formation of Haitian from French. Nor do, it seems, other cases noted in the literature: Berbice Dutch (see Kouwenberg 1994 and Lightfoot’s 1999, pp. 168–70 account of why Berbice Dutch is an SVO language with two OV lexifiers, Dutch and Ijo) and Arabicbased creoles such as Nubi (Owens 1985). Another difficulty lies in the fact that DeGraff’s (1997) Haitian data, on which Roberts bases his account, are no less idealized and partial than Pollock’s (1989) French data. Concerning the latter, I do not see how one could dismiss the fact that, to my and other native speakers’ ears, sentences like (20) are acceptable, and show no sign (e.g., intonational) of souvent ‘often’ not being in its “normal” syntactic position, left-adjacent to the verb: (20) FR

Marie souvent embrassait son père avant de partir pour l’école. ‘Mary often kissed her father before leaving for school.’

Various factors seem to play a role, including the adverb’s scope properties and idiosyncrasies of particular adverbs, as well as the tense of the main verb, as the finite/non-finite contrast is certainly too crude. Adverb placement thus appears just as inconclusive a diagnostic for the movement of finite verbs in French as in Cape Verdean, casting serious doubt on the whole issue of verb movement. I will return to the question whether the relative placement of verb and negation is more useful in this regard in section 3.3. Returning now to Haitian, it is true that adverbs such as deja ‘already’ or toujou ‘always’ precede main bare verbs, as in (9) above. But DeGraff (1997) does not say anything about their position vis-à-vis TMA markers. According to Joseph (1999) the possibilities are the following:10 (21) HC

Minis la (deja) ap (deja) li diskou a (deja). minister det (already) asp (already) read speech det (already) ‘The minister is already reading the speech.’

422

Alain Kihm

With resultative adverbs such as gravman ‘gravely’, possibilities are even more numerous: (22) HC

Ansasen an (gravman) te (gravman) blese (gravman) murderer det (gravely) tns (gravely) wound (gravely) doktè a (gravman). doctor det (gravely) ‘The murderer gravely wounded the doctor.’

Adverbs of time like raman ‘rarely’ may also occur in all these positions, and clause-initially as well. Adverbs like òdinèman ‘usually’, in contrast, are more limited, as they can only occur post-verbally and at the end of the clause: (23) HC Bouki pase (òdinèman) rad yo (òdinèman). Bouki pass (usually) cloth the.pl/3p ‘Bouki usually irons the(ir) clothes.’ No sweeping generalization based on adverb placement may be countenanced, therefore. The relative position of verbs, TMA markers and adverbs crucially depends on the semantics of the adverbs, not on some blind syntactic mechanism.11 A significant observation, moreover, is that all the positions shown in (22) are similarly accessible to the equivalent French adverbs, except the leftmost one: (24) FR L’assassin (*gravement) a (gravement) blessé (gravement) le docteur (gravement). ‘The murderer gravely wounded the doctor.’ Assuming that Haitian TMA markers (Tense marker te in (22)) and French auxiliaries (here a ‘has’) occupy the same syntactic position, all we may conclude with some certainty is that French auxiliaries, unlike Haitian TMA markers, must move over adverbs. This difference, however, need not have anything to do with the strength of features in Infl, inasmuch as auxiliaries and TMA markers – the latter even more than the former – may be assumed to be basegenerated in (split) Infl. The more proper generalization about surface order in French is, then, that predicate adverbs follow the first finite verb, be it an auxiliary (25a) or a main verb (25b). However, even that may be a hasty generalization because, as noted, the reverse order is not ungrammatical (25c) (see also example (20)), pace Pollock (1989) – although it may require a more particular discourse context to be felicitous than does its “moved” equivalent. Note further that souvent following the non-finite verb (25d) is also grammatical, as is its clause-final (25e) and clause-initial positioning (25f), with subtle informational nuances between the various constructions. Even souvent

Creoles, Markedness, and Default Settings 423 preceding the finite auxiliary (25g) without intonational breaks framing souvent does not strike me as odd. (25) FR

3.3

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

Jean a souvent embrassé Marie. Jean embrasse souvent Marie. Jean souvent embrasse Marie. Jean a embrassé souvent Marie. Jean a embrassé Marie souvent. Souvent Jean a embrassé Marie. Jean souvent a embrassé Marie.

Negation and verb movement

Recall that a comparison of Cape Verdean and Portuguese with regard to the placement of negation relative to the verb provides no evidence of a contrast in regard to verb movement. On the other hand, the contrast between Haitian (9)–(10) and its French translation (11)–(12) strongly suggests that French finite verbs move over the negator pas, whereas their Haitian equivalents remain in their base position with respect to pa. However, the difference is only significant insofar as French pas is analyzed as a negative head, like Haitian pa, or, as in Pollock (1989), as the specifier of the negation projection NegP headed by ne, thus initially to the left of the verb. There is an alternative account, though. According to Williams (1994), pas in [(ne) . . . V pas] constructions (where the dots show the position of object clitics) is neither a negative head (Neg°), nor the specifier of NegP. Instead, it is an adverb with negative meaning similar to German nicht. The negative head is optional ne, which Williams analyzes as a scope marker, and which occurs in the precise position occupied by pa in Haitian.12 As an adverb, pas has the subcategorization property that it cannot precede Tense.13 Its distributional properties follow, in particular the strong ungrammaticality of *Jean (ne) pas vient. There, Tense is fused with the main verb vient ‘comes’, so pas cannot precede the verb, as it would then precede Tense. That such fusion results from V-to-I movement of course remains as a theoretical possibility (see note 8), but the fact is that the placement of pas does not support it – even to the point of making it a dubious solution, since there is a plausible account of its position that does not depend on movement. Moreover, verb inflection can be taken care of by other means than overt Vto-I movement; after all, finite verbs inflect in English although they are not supposed to move – see Embick and Noyer (1999) for a morphological treatment of the issue. Louisiana Creole (LOU) data, which have been the object of recent detailed scrutiny (Neumann 1985; Rottet 1992; DeGraff 1997; Becker & Veenstra 2003; Kihm 2003), demonstrate the same point. Louisiana Creole has long and short verb forms; it is assumed that long forms, such as mõzhe ‘eat’, are uninflected for Tense, whereas short forms, such as mõzh ‘eat(s)’, include Tense. Thus, the

424

Alain Kihm

contrast shown in (26)–(27) appears fully parallel to the colloquial French contrast of J’ai pas mangé vs. Je mange pas: Tense precedes pa/pas. (26) LOU mo pa mõzhe. 1s neg eat ‘I didn’t eat.’ (27) LOU mo mõzh pa. 1s eat neg ‘I don’t eat.’ Other data show pa to follow Tense and to precede Aspect, as in (28); compare dialectal French J’étais pas après manger: (28) LOU Mo te pa ape mõzhe. 1s tns neg asp eat ‘I was not eating.’ This leads to the full generalization that Louisiana Creole pa, like French pas, is an adverb with negative meaning that follows Tense, rather than a negative head which, given its scope, ought to precede it. Compare this with Haitan M pa te ap manje [1s neg tense asp eat], where pa should indeed be analyzed as a negative head unlike both Louisiana Creole pa and French pas (see Lefebvre 1998, pp. 208–11). Similar to (28), the structure of Mo pa mõzhe in (26) must then be [mo tns pa asp mõzhe] where Tense is given no morphological exponent when Aspect is valued as perfective; hence, it is phonologically null.14 Williams (1994) notes further that French pas is also independently a head in addition to being an adverb; cf. his example Jean est arrivé pas heureux ‘Jean came not happy’, where pas’s headhood and local scope over the adjective are shown by the fact that ne would be ungrammatical (*Jean est arrivé ne pas heureux). This may have been an important factor in the process whereby French pas came to be reanalyzed as a head in Haitian, where pa occupies the position of the evanescent negative head ne of French.

3.4

A non-movement account of adverb placement

To sum up, there is no real evidence that negation and adverb placement depend on verb movement. The location of nonnegative adverbs appears to be governed by principles that may have little to do with syntax proper. As for negation, its contrastive placement in French and Haitian can at least as plausibly be explained by assuming a change in category from adverb to negative head. Note in passing that this account has the additional attraction of being readily extendable to English-based creoles where no has the same status as a negative head as Haitan pa, in contrast with English not, which is likely to be

Creoles, Markedness, and Default Settings 425 an adverb in the same measure as French pas is one (cf. Jamaican Im no de ya [3s neg cop here] with its English translation ‘She/he is not here’).15 Since negation and adverb placement actually constitute the main argument in favor of verb movement – morphological considerations not being compelling – and since the presence or absence of verb movement is in turn the main reason we would have to consider creole languages syntactically unmarked, Roberts’s hypothesis that they are syntactically unmarked thus appears to rest on very unstable ground. Before we start looking for firmer ground, let me sketch an alternative explanation for the position of Haitian deja and like adverbs – since no global explanation covering all adverbs is possible or desirable. Suppose that two topological fields are open to these adverbs (see Kathol 2000 for this notion): a final external field that seems to be always available for all types of adverbs, and an internal field which I will call the “verb cluster,” following Kathol (2000). For Haitian, we may define the verb cluster as everything that occurs between the negative head pa and the main verb. Leaving the unselective final field aside, we conclude that adverbs like deja must be inserted in the verb cluster. Deja may thus precede TMA markers, as in (21), provided it is not higher than negation (29), which constitutes the leftward boundary of the verb cluster; and it may follow TMA markers, as in (30), provided it does not cross the rightward boundary, the verb: (29) HC

Minis la (*deja) {VC pa (deja) ap li} diskou a minister det (already) neg (already) asp read speech det ‘The minister is not already reading the speech.’

(30) HC

Minis la {VC pa ap (deja) li} (*deja) diskou a minister det neg asp (already) read (already) speech det ‘The minister is not already reading the speech.’

Rather than invoking verb movement to explain the position of deja before bare main verbs, as in (9), repeated here for convenience as (31), we will therefore simply consider that deja is inside the verb cluster there as well, but that TMA is not overtly marked, meaning IP (or TP) is projected in the syntax but it is given no morphological realization. (31) HC

Bouki deja pase rad yo Bouki already iron cloth the.pl/3p ‘Bouki has already ironed the(ir) clothes.’

Why do deja and adverbs in its class show such partiality to the verb cluster? The reason may be semantic: deja ‘already’, toujou ‘always’, janm ‘never’, as well as process quantifiers like preske ‘almost’, trè ‘a lot’, two ‘too much’, etc., clearly belong to the tense-aspect domain. On the other hand, òdinèman ‘usually’ is also clearly aspectual, but it is excluded from the verb cluster, whereas

426

Alain Kihm

gravman ‘gravely’, a manner adverb not obviously related to aspect or mood, may occur in it. Other factors must therefore come into play. This should not surprise us, however, with such a complex, multidimensional matter as adverb placement. It is certainly no accident that man-adverbs in general probably are relatively recent borrowings from French, while deja, toujou, etc. must be as old as Haitian Creole itself. What should seem uncontroversial, in any event, is that the difference between Haitian and French in this matter cannot be one of markedness. No binary contrast of an elementary feature will account for such a complex distribution.

4

Markedness or Default?

In a pioneering article about morphosyntactic markedness Muysken (1981b, p. 451) proposes the following: “. . . interpretation which leads to a lack of parallelism in complexity between the syntax and Logical Form is marked, interpretation which preserves the parallelism is unmarked.” And he adds: “At this point we can do no more than regard it as the starting point for further research.” This seems to be a valuable starting point, indeed. Given a substantive, still largely programmatic, theory of Logical Form (LF), parallelism with morphosyntax is something that can in principle be assessed unambiguously. That is to say, there ought to be one and only one way for it to obtain. In contrast, the ways in which morphosyntax and LF are liable to mismatch cannot be bounded a priori. We expect them to be limited, and empirical research shows that they are indeed limited, but it also shows that even within such narrow boundaries there is room for rich variation. This suggests the same point as was made before, namely that “markedness” is not the proper notion to be used. Rather, parallelism in complexity between syntactic and LF representations is the “default” setting, which children apply unless they get evidence that it does not apply. Let me illustrate this point by taking up again the issue of verb placement with respect to negation and adverbs.

4.1

The acquisition of nondefault settings

Let us assume an overall architecture of grammar basically as in the theory of Distributed Morphology (Halle & Marantz 1993). The semantic component of grammar – as distinct from, although related to, general semantics or cognition – is the lexicon.16 The lexicon is the universal repertoire of languagerelevant concepts or elements, of which there are two sorts: substantial and functional. Substantial elements have denotations and they may be called “roots.” Roots are not syntactically categorized as verbs or nouns (predicators or arguments), they become so by combining with the functional elements v (verbhood) and n (nounhood). Functional elements (v, n, negation, tense, etc.)

Creoles, Markedness, and Default Settings 427 are semantico-logical representations operating on roots and/or other functional elements. The picture is sketchy and much too crude, but it will suffice for our purposes. Lexical elements combine in the lexicon or in syntax. Syntax (nondistinct from LF) is organized by the relation “A asymmetrically c-commands B” (A ^ B) (see Kayne 1994). Syntactic objects built up from (possibly complex) lexical elements are interpreted at the morphosyntactic interface (spellout) where the relation A ^ B translates into the relation “A linearly precedes B” (A < B) and the elements are associated with phonological matrices. Spellout takes place as soon as a syntactic object is assembled which the morphological component interprets as a free form (Beard 1995; Chomsky 1999). Let us now suppose, reasonably enough it seems, that the default setting for grammar is one where complexity is minimized. This means that there is parallelism à la Muysken (1981b), i.e., isomorphism between the syntactic/LF organization of lexical elements and their linearized phonological layout.17 Departures from isomorphism lead to more complex mapping types. Negation is surely a functional lexical element. As such, the default syntactic/LF object including negation is such that negation c-commands what it operates upon and linearly precedes it after spellout. Considering that “I did not see you at the party” does not negate the nuclear predicate [see you], but the tensed predicate [past [see you at the party]], it follows that, by default, negation ought to c-command and precede Tense. Consequently, Haitian (32), which shows Neg < Tense does indeed represent the default structure which we may assume children are born with: (32) HC

M pa te wè ou nan pati-a. 1s neg tns see 2p loc party-det ‘I did not see you at the party.’

In other words, the default setting for the functional lexical element {Negation} is to treat it as a head c-commanding the Tense head. In Haitian both heads can be immediately delivered to morphology because they are associated with free forms – again probably the default case as it warrants isomorphy. If the primary data the child is faced with do not match the default setting, she has to adjust to whatever these data turn out to be as long as they fall within the range of what is possible for human languages. Were we in a logic of markedness, the adjustment would be wired-in as well. That is to say, the child’s innate core grammar would also include alternative settings to fall back on as soon as the primary data tell her she cannot choose the unmarked option. Minimal exposure to unexpected data, perhaps one occurrence, should then be enough to trigger the switch, since the mechanism is supposed to be there and ready to ignite. This is what Pinker (1989) calls “default markedness.” But given this kind of mechanism, how do we explain Haitian? Obviously, children were exposed to some sorts of French data – where would they have got pa from otherwise? These French data must have included cases of

428

Alain Kihm

“marked” negation placement, as in Je comprends pas ‘I don’t understand.’18 Yet, they staunchly kept to the default setting, despite all indications that they should not. What this points to is that rather heavy exposure to nondefault data is actually required before children will give up the innate setting – what Pinker (1989) calls “strength markedness.” This in turn suggests that the alternative nondefault solution they finally settle on is not previously available in their minds as such. An account of negation placement in terms of default might then run as follows. Default categorization for the lexical functional element {Negation} is as a negative head, a particle (i.e., a free form). French departs from defaultness because it categorizes {Negation} as an adverb, linearized relative to Tense by a special rule VT/VAux < Neg (Williams 1994). This is the grammar for colloquial French to which children are likely to be exposed first. Later, when they already have switched from the default setting, they learn there is another negative element, call it {Negation Scope}, spelled out ne, that this is a head and that they may insert it (but are not obliged to, stylistic considerations apart) in the position where they would have put negation had they been allowed to.19 Clearly, the initiators of Haitian – or of any French-based creole – never had access to ne, because the variety of French they were exposed to did not include it or vanishingly so (see note 19). Nor were they sufficiently exposed to French to give up the default setting, which they applied to pa having recognized it as the phonological form of {Negation}. They may have been helped in this by cases where French pas may indeed be categorized as a negative head (e.g., pas heureux ‘not happy’), and perhaps also by sporadic occurrences of broken French forms (see note 18). In contrast, as we saw, the grammar of negation in Louisiana Creole is basically the same as in Colloquial French, indicating longer exposure to the lexifier language or, more radically, that creolization proceeded along a different route than it did in Haitian and other French-based creoles.

4.2

Further instances of “default” creole features

Other features of the creole languages can be explained by the logic of default rather than by markedness. An obvious candidate is the system of preverbal TMA particles. These constitute morphologically free realizations of functional lexical elements having tense-aspect values as their meanings and which are syntactically inserted in a c-commanding position with respect to the VP – the default option for such functional elements. The lexifiers, in contrast, apply the nondefault option of morphologically merging (some of) these lexical elements with the verb. The default character of preverbal TMA particles is strongly supported by the fact that they are present in Arabic-based creoles, where it does not seem they can be accounted for by substratal influence as has been

Creoles, Markedness, and Default Settings 429 argued, not unreasonably but never entirely convincingly, for many IndoEuropean creoles (see, e.g., Lefebvre 1998 for Haitian; for Arabic based-creoles, see Holm 1989, pp. 568–74 and references therein). I will return to the issue of interpretation when no TMA marker appears in section 4.3. SVO order, which nearly all creoles share, may also be considered the default setting, and that is the gist of Kayne’s (1994) approach to phrase structure (see also Moro 2000). Alternative orders are nondefault rather than marked because deriving them involves much more than switching the value of a discrete feature from minus to plus. Another feature considered diagnostic of “unmarkedness” is absence of referential null subjects: creoles are not “pro-drop” languages except in cases of nonargumental or expletive subjects: (33) HC

Te fè frèt. tns make cold ‘It was cold.’ (Roberts 1999, p. 312)

Concerning the status of pro-drop, divergent views have been expressed. Hyams (1986) considers pro-drop the default case because she analyzes pronouns as sets of phonologically null features in syntax (D-structure in her terms), so not having them in Phonological Form maximizes isomorphism. According to this view, creoles, alongside French and English, would actually be marked languages. Roberts (1999) links the possibility of referential null subjects to the marked process of V raising to Agr necessary to identify the features of the null subject pro. As he himself remarks, however, not all languages with putative V raising exploit this possibility – for instance, French doesn’t – which shows that other factors are active, thus opening the door to other accounts that might dispense with V movement altogether. Let me just add that the distribution of nonreferential null subjects in creole languages is more complex than Roberts’s examples suggest. Cape Verdean seems to allow them more or less in the same contexts as does Haitian: (34) CVC

Sta faze kalor oji. prog make heat today ‘It’s hot today.’ (Baptista 2002, p. 254)

Closely related Kriyol of Guinea-Bissau (KR), in contrast, does not permit null subjects in such sentences; it requires a generic noun, not an expletive pronoun, as a subject, as in (35) (Kihm 1994, pp. 49–50). Only with the raising verb parsi ‘seem’ is a null subject acceptable, as in (36) (Kihm 1994, pp. 237–9; this is equally acceptable in Cape Verdean – Baptista 2002, p. 254). In this case, however, the null subject is arguably not pro, but a variable bound by the logical subject, namely the clause Jon sta na ospital:

430

Alain Kihm

(35) KR

Kaw kinti awos. place hot today ‘It’s hot today.’

(36) KR

Parsi kuma Jon sta na ospital. seem comp Jon cop loc hospital ‘It seems that Jon is in the hospital.’

Papiamentu, although it is not a fully pro-drop language like Spanish or Portuguese, accepts null subjects in still more contexts than does Cape Verdean (Kouwenberg 1990). Finally, it is worth noting that subjects of weather and raising verbs are precisely the only subjects that can be colloquially left unexpressed in the otherwise strongly anti-pro-drop language that French is: (37) FR

Fait froid aujourd’hui. Faut qu’elle vienne. Paraît qu’elle est venue.

Again, the differential occurrence of referential and nonreferential null subjects appears to be regulated by factors which, unclear as they are, cannot be reduced to the binary logic of markedness. A possible account in terms of default would be to say that subjects must be expressed – the default option – unless their content can be retrieved elsewhere or need not be retrieved, in which case non-expression would be the default. This would explain the default character of pro-drop in richly inflected languages as well as in entirely uninflected languages such as Chinese or Japanese where retrieval is discursive (Huang 1984). It would also explain the possible nullity of semantically empty subjects in languages that do not tolerate null subjects in other contexts, and finally, last but not least, the nondefaultness of obligatorily expressed subjects in languages with enough inflectional contrasts to make them unnecessary (e.g., German).

4.3

The default interpretation of bare verbs

Finally, we must address an issue that figured prominently in debates about creoles and markedness, namely whether they represent the unmarked case as far as the expression of tense, mood, and aspect is concerned. As we saw, the morphological realization of TMA as free form verbal particles may constitute the default setting in this domain. But that leaves open the case of finite bare verbs, i.e., uninflected verbs not in the scope of an overt verbal particle which nonetheless receive a tense-aspect interpretation. Those verbs are unmarked in the material sense that no bound or free marker is present to which the function of bearing the interpretation can be assigned. Hence the question of whether the tense-aspect value of finite bare verbs is also an unmarked one and, by extension, whether the overall TMA system of creole languages, including

Creoles, Markedness, and Default Settings 431 overtly marked values, is unmarked with respect to some conceivable universal set of binary TMA values. This issue is addressed in such early works as Bickerton (1981) and Muysken (1981a). Muysken shows that Bickerton’s attempt to align all creole systems on an unmarked set of three binary values – [α anterior] for tense, [β irrealis] for mood, and [γ nonpunctual] for aspect – does not do justice to the actual complexity of the facts, and his conclusion has been repeatedly confirmed in subsequent works. What seems well established, on the other hand, is that finite bare verbs are always given the same values. Using Muysken’s (1981a) labels, these are indicative-simple past-perfective or indicative-present-imperfective, depending on the verb’s aspectual class in the sense of Dowty (1979): bare verbs referring to states cannot be past, whereas bare verbs referring to activities or telic eventualities may be (cf. the Haitian examples (9) and (10); see also Holm et al. 2000). Languages differ as to the possible interpretation of activity or telic eventuality bare verbs. In Kriyol, for instance, they only have simple past perfective meaning, as in (38), whereas present imperfective requires an overt marker, as in (39) (Kihm 1994). (38) KR

Jon kebra karu. Jon break car ‘Jon broke the car.’

(39) KR

Jon ta kebra karu. Jon asp break car ‘Jon breaks the car [every time he drives it].’

In Haitian, simple past perfective and present imperfective are both possible, but the latter requires special contexts such as the presence of a generic object (Lefebvre 1998, p. 134; DeGraff forthcoming): (40) HC

Bouki vann chat-la. Bouki sell cat-det ‘Bouki sold the cat.’

(41) HC

Bouki vann chat. Bouki sell cat ‘Bouki sells cats.’

Although marking conventions can be used, what seems to be really at issue is the question how bare verbs should “normally” be comprehended given their inherent meanings. This is the question of their default interpretation, in other words. In a Distributed Morphology framework, a bare verb consists in the combination of a root and the functional element v carrying verbhood, and nothing more. Since roots are categorially unspecified, aspectual classes must be a property of v, meaning that the lexicon includes a set {v} whose members

432

Alain Kihm

are endowed with distinctive aspectual properties (Arad 1999). Not only does v set the aspectual class of the resulting verb, but it defines the roles born by the arguments of this verb. It is well known that the external argument of a state verb, for instance, does not bear the same role as the external argument of an activity verb: the latter may be an Agent or a Causer, while the former is an Experiencer or a Theme depending on the semantics of the root, i.e., whether we are dealing for instance with a psychic state such as “love,” or a property such as “be tall,” or a natural event such as “fall,” etc. The concept of causality – not really distinct from agenthood – is an excellent candidate for being part of the genetically determined cognitive endowment of the human species and probably other animal species as well (e.g., Menzel & Johnson 1976; Worden 1998). Its basic tenet is that our minds are genetically geared to causes recognized or imagined from their observed effects (see Pinker 1997, ch. 5). Therefore, it seems rather natural that in the absence of an explicit indication to the contrary – in the default case – verbs referring to activities should be interpreted as simple past, i.e., as describing a state of affairs where the effect of the activity is accomplished and visible (cf. “John broke the car”). An example of an explicit indication to the contrary is having a generic object (as in Bouki vann chat) which does not impose an end point on the process (see Lefebvre 1998, p. 134). States, in contrast, must be inferred from an examination of the experiencer or undergoer, hence the present as the default interpretation (cf. “John loves Mary”). If the foregoing speculations, crude and sketchy as they are, make some sense, we should not be surprised that such default interpretations are not accessible in the languages that served as lexifiers for the creoles. Indeed, in these inflected languages, Indo-European and Semitic, verbs are never bare in this sense, i.e., they never consist of merely the root and v. This means that tense-aspect values are always overtly marked on them, and cannot be deduced by default from the semantics of the verbalized root.

5

Conclusion

Let me first repeat what constitutes the thrust of this chapter: the hypothesis that creole grammars are syntactically unmarked must be rejected by the very logic of markedness theory, which implies a priori sets of one unmarked and one or more marked values for every feature, with an inherent relation between the unmarked and the marked. Such sets do not seem to exist for core syntactic properties. What there is instead is a default setting for these properties, which I assume to be innate, and an indefinite array of nondefault settings which children can only acquire by being exposed to processable primary data. Of course, this conclusion does not preclude the (actually evidenced) possibility that creole languages may show unmarkedness or reduced markedness in the Jakobsonian sense of the term relative to their lexifiers, in areas where the notion is applicable, that is phonology and perhaps morphology. That the

Creoles, Markedness, and Default Settings 433 very same process, i.e., creolization, that veers grammars toward the default option also favors unmarked values where they exist is indeed to be expected. The obvious question to ask, then, is that of the extent to which creole languages manifest the default grammar resulting from the application of all default settings. This entails two questions: (1) Do all the languages we call creoles equally embody the default grammar, if at all? (2) To what degree of detail is the default grammar specified, and/or does it potentially provide for all the morphosyntactic properties of natural languages? The first question harks back to the old and as yet unsettled debate whether creole languages constitute a natural class, calling for one theory of their origin, or whether different sociohistorical situations made for distinct geneses (see, e.g., Bickerton’s distinction between “fort” and “plantation” creoles). In Kihm (2000) I argued for a mixed answer, along the following lines: Creole languages as a whole are characterized by the fact that the children who were their initiators had to tap their innate grammatical capacities more than is usually necessary, which accounts for the parallel divergences from the lexifiers such as common recourse to preverbal particles. (Adults are probably unable to do this, so if there exist creole languages initiated by adults, we expect them to vary more widely according to the adults’ first languages.) But this “more than usual” may itself be to various degrees, and so many particular factors have to be entered into the calculus that the outcome may indeed look like “order from chaos” (Lang 2000; for the degree of particularity of the factors involved, see Singler 2000). In other words, I only partially endorse Bickerton’s hypothesis that all true (i.e., plantation) creoles instantiate the – by definition default – language bioprogram, with a few departures due to subsequent history. Creole languages show more of default grammar than do languages acquired in more ordinary circumstances. Yet, to be all default grammar would require a near total absence of processable primary data, perhaps even including lexical items, which is nowhere observed.20 The second question is tougher, and I will do no more here than briefly illustrate what it points to. An intriguing and rarely discussed peculiarity of creole languages is that they are more diverse in their nominal systems than they are in their verbal systems. To take one example, French-based creoles have an NP-final definite determiner, which is a departure from isomorphism as it implies movement of the NP over the determiner. Kriyol, on the other hand, has no definite determiner at all, again a departure from isomorphism if we assume the innate lexicon to include a functional element {D} having definiteness – however defined – as its content. This difference seems to be rooted in the fact that the creole initiators were unable to parse the definite articles of the respective lexifiers, French and Portuguese, for reasons of lack of phonological salience. French offers an easily parsable alternative item that could be used to spell out {D}, namely the postposed deictic adverb là of cette femme-là ‘that woman’ or (colloquial) la femme là ‘the woman in question’ (see Lefebvre 1998, pp. 79–84, who also argues for substrate influence). No such item is readily available in Portuguese, so {D} remained unrealized in Kriyol.

434

Alain Kihm

Individual creoles appear thus to be heavily dependent on idiosyncratic features of the lexifiers for the structure of their noun phrases, to an extent that verb phrases are not. This suggests that there may be no default setting in this domain; if there were one, the initiators of what was to become Kriyol would have found a way to express definiteness – e.g., using a demonstrative as a definite determiner as in Cape Verdean, where this is a recent and partial development, however (Baptista 2002, pp. 27–30; to appear). In contrast, they did develop alternative expressions in the TMA domain, for which the lexifier also offers phonologically elusive morphemes. That is to say, it seems that the default grammar does not specify the inner structure of arguments, whereas it is rather more explicit as far as the inner structure of the predicate is concerned – which is a different hypothesis than assuming UG to be divided into a core and a periphery. If there is some reality behind this speculation that would give us an interesting insight into the language faculty, it is noteworthy that only a reasoning in terms of “default” is able to open it for us.

NOTES 1 A third possibility that figured prominently in early generative grammar is that markedness is (part of) an evaluation metric over whole grammars (see Battistella 1996, ch. 4). It does not seem to be directly relevant to our problem. 2 Mufwene (1989, 1990) presents still another view according to which creoles can only be said to be syntactically unmarked with respect to their particular lexifiers. Insofar as he means that markedness is not a relevant notion at the level of UG, I tend to side with him. However, the features he uses to illustrate his point, e.g., isolating morphosyntax as compared with agglutination, seem to me better to fall under the logic of default, as we shall see. Also see the discussion in Lefebvre (1998, pp. 72–5). 3 But note that Chomsky comes to the conclusion that there is nothing “imperfect” in displacement after all, since it is the way syntax has of making itself legible to the phonological and conceptual components. 4 LA stands for the abstract form of the definite determiner realized as /la/, /a/, /an/, or /nan/, depending on the phonological definition of the final segment of the preceding word. 5 In view of the variable word order of Dutch, with both clause-final and clausemedial verb placement patterns, Dutch-lexifier creoles such as Negerhollands and Berbice Dutch may constitute another case that is worth investigating from this perspective (e.g., Kouwenberg 1992, 1996). 6 More accurately, in order to be eliminated (“checked out”) so that what is delivered to phonology in the position of Infl, i.e., the inflected verb, is a legitimate PF object. 7 Actually, starring Bouki déjà repassait le linge is an oversimplification, as we shall see with regard to adverb placement in French more generally in section 3.2. 8 Naturally, short V-movement can still be hypothesized if that is how verbs are supposed to pick up or to check off the inflection markers they bear. However, this

Creoles, Markedness, and Default Settings 435

9

10 11

12 13

14 15

16 17

18

then becomes a strictly theory-internal decision, without empirical evidence to back it up. Baptista assumes that e involves a third person feature which forces it to raise to the subject agreement projection AgrS, much like English auxiliaries. Yet, if checking person features was enough to induce overt verb movement, we would expect person inflected verbs always to raise above negation, contrary to fact (cf. English third person inflection). A more plausible explanation is that e is historically, and perhaps synchronically as well, related to the third person subject pronoun. Similar facts obtain in English-based creoles including those of Suriname, Guyana, and Jamaica (e.g., Winford 1993). I am grateful to John Lumsden for bringing Joseph’s thesis to my attention. Pollock (1989, 1997) himself insists on the need to carefully select the adverbs before drawing any conclusion. In that sense, I am not arguing against him, especially because his main goal in his articles is to demonstrate the Split Inflection Hypothesis (SIH), which is quite peripheral to my own concerns. (Creole languages tell nothing for or against the SIH in any event.) That I do not share his grammaticality judgments only reveals that adverb placement is indeed a complex matter, about which native French speakers, as Pollock and I both are, can disagree. Nor do I share his commitment to Minimalism. It is a type 1 negative marker in the terminology of Rowlett (2001), to whose more detailed study I direct the reader. Williams (1994) makes it a general property of French adverbs that they cannot precede Tense. However, as we saw, this generalization cannot be maintained. In fact, many of the French examples which Williams stars, e.g., Jean récemment parlait à Pierre ‘Jean recently spoke to Pierre’ (p. 191), I and other native speakers find acceptable. It must therefore be a special property of pas and a few other adverbs such as toujours ‘always’ and jamais ‘never’ that they are unable to precede Tense. I am grateful to Tonjes Veenstra for discussing Louisiana Creole data with me. Pollock (1997, p. 271), following Roberts (1993), mentions that English not was a negative adverb during a “short-lived” period in the language’s history, when sentences like John not came were fine. Actually, this rather suggests that English briefly resorted to the default option, and it may be the source of the negation marker in the English-based creoles. The reason Pollock suggests for why not V became ungrammatical is not convincing, as it involves the quite controversial affix-hopping. As is well known, the French negator pas comes from a postverbal adverbial expression involving nouns of measure such as pas ‘step’, point ‘dot’, goutte ‘drop’, or mie ‘crumb’. Only the first one made it into contemporary French. This view can be attributed to Borer (1983) as noted by Bickerton (1999, p. 56). Isomorphism between lexical concepts and surface forms does not in any way imply one-to-one mapping from semantic structures which, as Seuren and Wekker (1986) correctly remark, is not a reasonable condition to impose upon natural language. On isomorphism also see Hyams (1986). There is no evidence that the French addressed the African slaves in so-called petit nègre where pas precedes the verb when the latter appears in the infinitive (cf. Moi pas comprendre). In fact, this foreigner talk variety probably did not exist at the time, as it seems to have been created much later, in the nineteenth century, in French West African colonies. And even if occasional instances of [pas Vinf] were heard, this must have been among many instances of [V pas] given the inherent variability of such situations, so we would still have to explain why the initiators

436

Alain Kihm

of Haitian consistently chose one option over the other. The substrate may also have played a part here, since the Fongbè negation mà occupies the same slot as Haitian pa (see Lefebvre 1998, p. 210). 19 I am not aware that there are specific enquiries into this matter, but it seems safe to wager that many French speakers never actively acquire this item. Nor are French children likely to encounter before school the few relic expressions in which ne may function as sole negation (cf. Je ne saurais dire ‘I could not tell’). 20 More accurately, it is observed, it seems, but in connection with sign language (see above), not spoken language. Now, if that observation is correct, it implies that deaf-mute children not exposed to meaningful signs are able to invent them. Spoken words, in contrast, are never made up out of nothing, as if there were a “block” against using phonology creatively (glossolalia is not language, and twin languages seem to be always parasitic on a surrounding language – see Bickerton 1990).

REFERENCES Anderson, Stephen R. (1985) Phonology in the Twentieth Century. Chicago: The University of Chicago Press. Arad, Maya (1999) On ‘little v’. In: Karlos Arregi, Benjamin Bruening, Cornelia Krause, and Vivian Lin (eds.) Papers on Morphology and Syntax, Cycle One. MIT Working Papers in Linguistics 33, 1–25. Baptista, Marlyse (2002) The Syntax of Cape Verdean Creole: The Sotavento Varieties (Linguistik Aktuell/Linguistics Today, vol. 54). Amsterdam: John Benjamins. Baptista, Marlyse (to appear) Cape Verdean Creole. In: Marlyse Baptista and Jacqueline Guéron (eds.) Bare Nouns and the Structure of DP in Creole Languages. Amsterdam: John Benjamins. Battistella, Edwin L. (1996) The Logic of Markedness. Oxford: Oxford University Press. Beard, Robert (1995) Lexeme-Morpheme Base Morphology: A General Theory of Inflection and Word Formation. Albany: State University of New York Press. Becker, Angelika and Tonjes Veenstra (2003) Creole prototypes as basic varieties and inflectional morphology. In: Christine Dimroth and Marianne Starren (eds.) Information Structure, Linguistic Structure and the Dynamics of Learner Language. Amsterdam: John Benjamins, pp. 235–64. Bickerton, Derek (1981) Roots of Language. Ann Arbor: Karoma. Bickerton, Derek (1990) Language and Species. Chicago: The University of Chicago Press. Bickerton, Derek (1999) How to acquire language without positive evidence: What acquisitionists can learn from Creoles. In: DeGraff (ed.), pp. 49–74. Borer, Hagit (1983) Parametric Syntax. Dordrecht: Foris. Chomsky, Noam (1981) Markedness and core grammar. In: Adriana Belletti, Luciana Brandi, and Luigi Rizzi (eds.) Theory of Markedness in Generative Grammar: Proceedings of the 1979 GLOW Conference. Pisa: Scuola Normale Superiore di Pisa, pp. 123–46. Chomsky, Noam (1993) A minimalist program for linguistic theory. In: Kenneth Hale and Samuel J. Keyser (eds.) The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press, pp. 1–52. Chomsky, Noam (1998) Minimalist inquiries: The framework. MIT Occasional Papers in Linguistics 15.

Creoles, Markedness, and Default Settings 437 Chomsky, Noam (1999) Derivation by phase. MIT Occasional Papers in Linguistics 18. (Reprinted in Michael Kenstowicz (ed.) (2001) Ken Hale: A Life in Language. Cambridge, MA: MIT Press, pp. 1–52.) Costa, João (1996) Adverb positioning and V movement in English: Some new evidence. Studia Linguistica 50, 22–34. Damoiseau, Robert (1996) Les adjectivaux en créole haïtien [Adjectival items in Haitian Creole]. In: Daniel Véronique (ed.) Matériaux pour l’étude des classes grammaticales dans les langues créoles. Aix-en-Provence: Publications de l’Université de Provence, pp. 151–61. DeGraff, Michel (1997) Verb syntax in, and beyond, creolization. In: Liliane Haegeman (ed.) The New Comparative Syntax. London: Longman, pp. 64–94. DeGraff, Michel (ed.) (1999) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press, pp. 49–74. DeGraff, Michel (forthcoming) Haitian Creole. In: John Holm and Peter Patrick (eds.) Comparative Creole Syntax. London: Battlebridge. Dowty, David R. (1979) Word Meaning and Montague Grammar. Dordrecht: Reidel. Embick, David and Rolf Noyer (1999) Locality in post-syntactic operations. In: Vivian Lin, Cornelia Krause, Benjamin Bruening, and Karlos Arregi (eds.) Papers on Morphology and Syntax, Cycle Two. MIT Working Papers in Linguistics 34, 265–317. Emonds, Joseph E. (1978) The verbal complex V’-V in French. Linguistic Inquiry 9, 151–75. Gadelii, Karl E. (1997) Lesser Antillean French Creole and Universal Grammar (Gothenburg Monographs in Linguistics 15). Gothenburg: University of Gothenburg Department of Linguistics. Greenberg, Joseph H. (1966) Language Universals, with Special Reference to Feature Hierarchies. The Hague: Mouton. Halle, Morris and Alec Marantz (1993) Distributed morphology and the pieces of inflection. In: Kenneth Hale and Samuel J. Keyser (eds.) The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press, pp. 111–76. Holm, John (1989) Pidgins and Creoles. Vol. II: Reference Survey. Cambridge: Cambridge University Press. Holm, John et al. (2000) The creole verb: A comparative study of stativity and time reference. In: John McWorther (ed.) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins, pp. 133–61. Huang, C.-T. James (1984) On the distribution and reference of empty pronouns. Linguistic Inquiry 15, 531–74. Hyams, Nina (1986) Language Acquisition and the Theory of Parameters. Dordrecht: Reidel. Jakobson, Roman (1932/1966) Zur Struktur des russischen Verbums [On the structure of Russian verbs]. In: Eric P. Hamp Martin Joos, Fred W. Householder, and Robert Austerlitz (eds.) Readings in Linguistics II. Chicago: The University of Chicago Press, pp. 22–30. Joseph, Sauveur Joseph (1999) Etude sémantico-syntaxique des adjectifs et des adverbes en créole haïtien. PhD dissertation, Université du Québec à Montréal. Kathol, Andreas (2000) Linear Syntax. Oxford: Oxford University Press. Kayne, Richard S. (1994) The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Kegl, Judy, Ann Senghas, and Marie Coppola (1999) Creation through contact: Sign language emergence and sign language change in Nicaragua. In: DeGraff (ed.), pp. 179–237.

438

Alain Kihm

Kihm, Alain (1994) Kriyol Syntax: The Portuguese-Based Creole Language of Guinea-Bissau. Amsterdam: John Benjamins. Kihm, Alain (2000) Are creole languages “perfect” languages? In: McWhorter (ed.), pp. 163–99. Kihm, Alain (2003) Inflectional morphology in creole languages. In: Ingo Plag (ed.) Phonology and Morphology of Creole Languages (Linguistische Arbeiten, vol. 478). Tübingen: Niemeyer, pp. 333–63. Kouwenberg, Silvia (1990) Complementizer pa, the finiteness of its complements, and some remarks on empty categories in Papiamento. Journal of Pidgin and Creole Languages 5, 39–51. Kouwenberg, Silvia (1992) From OV to VO: Linguistic negotiation in the development of Berbice Dutch Creole. Lingua 88, 263–99. Kouwenberg, Silvia (1994) A Grammar of Berbice Dutch Creole (Mouton Grammar Library 12). Berlin: Mouton de Gruyter. Kouwenberg, Silvia (1996) Grammaticalization and word order in the history of Berbice Dutch Creole. In: Philip Baker and Anand Syea (eds.) Changing Meanings, Changing Functions: Papers Relating to Grammaticalization in Contact Languages (Westminster Creolistics Series 2). London: University of Westminster Press, pp. 207–18. Lang, George (2000) Chaos and creoles: Towards a new paradigm? In: McWhorter (ed.), pp. 443–57. Lefebvre, Claire (1998) Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Lightfoot, David (1999) The Development of Language: Acquisition, Change, and Evolution. Oxford: Blackwell. McWhorter, John (ed.) (2000) Language Change and Language Contact in Pidgins and Creoles. Amsterdam: John Benjamins. Menzel, Emil W. and Marcia K. Johnson (1976) Communication and cognitive organization in humans and other animals. In: Stevan Harnad, Horst Dieter Steklis, and Jane B. Lancaster (eds.) Origins and Evolution of Language and Speech. New York: The New York Academy of Sciences, pp. 131–42. Moravcsik, Edith A. and Jessica R. Wirth (1986) Markedness: An overview. In: Fred R. Eckman, Edith A. Moravcsik, and Jessica R. Wirth (eds.) Markedness. New York: Plenum Press, pp. 1–11. Moro, Andrea (2000) Dynamic Antisymmetry. Cambridge, MA: MIT Press. Mufwene, Salikoko S. (1989) Some explanations which strike me as incomplete. Column, Journal of Pidgin and Creole Languages 4, 117–28. Mufwene, Salikoko S. (1990) Creoles and universal grammar, Linguistics 28, 783–807. Muysken, Pieter (1981a) Creole tense/mood/aspect systems: The unmarked case? In: Pieter Muysken (ed.) Generative Studies on Creole Languages. Dordrecht: Foris, pp. 181–99. Muysken, Pieter (1981b) Quechua causatives and logical form: A case study in markedness. In: Adriana Belletti, Luciana Brandi, and Luigi Rizzi (eds.) Theory of Markedness in Generative Grammar: Proceedings of the 1979 GLOW Conference. Pisa: Scuola Normale Superiore di Pisa, pp. 443–73. Neumann, Ingrid (1985) Le créole de Beaux Bridge, Louisiane: Morphosyntaxe, textes, vocabulaire. Hamburg: Helmut Buske. Newport, Elissa L. (1999) Reduced input in the acquisition of sign languages: Contributions to the study of creolization. In: DeGraff (ed.), pp. 161–78.

Creoles, Markedness, and Default Settings 439 Owens, Jonathan (1985) The origins of West African Nubi. Anthropological Linguistics 27, 229–71. Pinker, Steven (1989) Learnability and linguistic theory. In: Robert Matthews and William Demopoulos (eds.) Learnability and Linguistic Theory. Dordrecht: Kluwer, pp. 107–27. Pinker, Steven (1997) How the Mind Works. London: Penguin Books. Pollock, Jean-Yves (1989) Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry 20, 365–424. Pollock, Jean-Yves (1997) Notes on clause structure. In: Liliane Haegeman (ed.) Elements of Grammar. Dordrecht: Kluwer, pp. 237–79. Roberts, Ian (1993) Verbs and Diachronic Syntax. Dordrecht: Reidel. Roberts, Ian (1999) Verb movement and markedness. In: DeGraff (ed.), pp. 287–327. Rottet, Kevin (1992) Functional categories and verb raising in Louisiana Creole. Probus 4, 261–89. Rougé, Jean-Louis (2004) Dictionnaire étymologique des créoles portugais d’Afrique. Paris: Karthala. Rowlett, Paul (2001) French ne in non-verbal contexts. In: Yves D’Hulst, Johan Rooryck, and Jan Schroten (eds.) Romance Languages and Linguistic Theory 1999. Amsterdam: John Benjamins, pp. 335–53. Seuren, Pieter A. M. and Herman Wekker (1986) Semantic transparency as a factor in creole genesis. In: Pieter Muysken and Norval Smith (eds.) Substrata versus Universals in Creole Genesis. Amsterdam: John Benjamins, pp. 57–70. Singler, John V. (2000) Optimality theory, the minimal-word constraint, and the historical sequencing of substrate influence in pidgin/creole genesis. In: McWhorter (ed.), pp. 335–51. Trubetzkoy, Nikolay S. (1939) Grundzüge der Phonologie [Principles of phonology] (Travaux du cercle linguistique de Prague 7). Prague: Cercle Linguistique de Prague. Williams, Edwin (1994) A reinterpretation of evidence for verb movement in French. In: David Lightfoot and Norbert Hornstein (eds.) Verb Movement. Cambridge: Cambridge University Press, pp. 189–205. Winford, Donald (1993) Predication in Caribbean English Creoles. Amsterdam: John Benjamins. Worden, Robert (1998) The evolution of language from social intelligence. In: James R. Hurford, Michael Studdert-Kennedy, and Chris Knight (eds.) Approaches to the Evolution of Language. Cambridge: Cambridge University Press, pp. 148–66.

440

George Huttar

18

Semantic Evidence in Pidgin and Creole Genesis GEORGE HUTTAR

1

Introduction

The evidence from semantic structure with which this chapter deals – semantic evidence, for short – must be clearly distinguished from lexical evidence. Lexical evidence is about which lexemes in a particular pidgin or creole, or in pidgins in general or creoles in general (PCs), come from which languages, and from which kinds of languages – substrate, superstrate, adstrate. It is thus concerned with identifying the sources of the phonological shapes of lexemes. For example, the source of Ndyuka mofu ‘mouth’ is English mouth. Semantic structure, by contrast, looks at the range of meanings of lexemes (regardless of their source), and the relationship among lexemes within a given semantic domain in terms of shared and unique features. For example, as a body-part term, mofu participates in paradigmatic relations with other Ndyuka lexemes denoting body parts, like nosu ‘nose’ and ain ‘eye’. Under semantic evidence can also be included some aspects of compounding and other processes of lexeme combination; the focus here is on which lexemes combine with each other and with what semantic result, not on the morphosyntactic processes involved. On the borderline between semantic and morphosyntactic data we have the morphosyntactic environments in which particular lexemes occur. For example, mofu participates in noun-noun compounds as in these examples: (1) mofu ‘mouth’+ buba ‘skin’ > mofu buba, buba mofu ‘lip’ bobi ‘breast’ + mofu > bobi mofu ‘nipple’ pasi ‘path’ + mofu > pasi mofu ‘end of path’ lansi ‘spear’ + mofu > lansi mofu ‘spear point’ liba ‘river’ + mofu > liba mofu ‘mouth of river’ bataa ‘bottle’ + mofu > bataa mofu ‘bottle opening’ doo ‘door’ + mofu > doo mofu ‘doorway’ yali ‘year’ + mofu > yali mofu ‘end of year’

The Handbook of Pidgin and Creole Studies Edited by Silvia Kouwenberg and John Victor Singler © 2008 Blackwell Publishing Ltd. ISBN: 978-0-631-22902-5

Semantic Evidence in Pidgin and Creole Genesis 441 Mofu also combines with adjective gaan ‘big’ to form the adverbial phrase gaan mofu ‘impudently’, as in (2). While I consider such combinations of lexemes to fall under the rubric of “semantic structure,” I exclude some other distributional characteristics of lexemes such as their subcategorization features. But the border between semantic and morphosyntactic data is not sharp. (2) Den lobi piki sama gaan mofu. 3p love answer person big mouth ‘They like to answer people impudently.’ It is clear that in many creoles (we know little about pidgins in this area), the source of lexemes is not always the same as the source of their semantic characteristics. Thus in (1), all the lexemes except buba ‘skin’ (which has an African source; see Huttar 1985, p. 56) are from English; yet only some of the combinations shown in (1) have corresponding combinations in English: English may speak of the mouth of a river and the mouth of a bottle, but not the mouth of a year. As with other features of PCs not clearly attributable to a superstrate source, explanations of the source of such combinations cover the usual range of substrate, incomplete second language acquisition, innate universals and the like, as laid out elsewhere in this volume. For example, attributing these combinations to incomplete second language acquisition would mean extending to semantic structures what Thomason says about grammatical structures: The people in the new contact situation learn to communicate with each other by deploying the new vocabulary with grammatical structures they hope will be understood by their interlocutors. The idea is that people’s “right” guesses about what the others will understand become part of the emerging contact language. (Thomason 2001, p. 180)

To the extent that such guesses are based on substrate speakers’ native languages, this explanation would be compatible with appeal to substrate influence. Lefebvre (e.g., 1998) might attribute these combinations to relexification of parallel combinations in Fongbe or another likely substrate language. (Note that Thomason (2001, p. 180) considers Lefebvre’s relexification approach to have to do with borrowing, rather than with incomplete second language acquisition.) An innatist position would presumably claim that these combinations are the result of the human inborn Language Acquisition Device, although I don’t know of anyone who has made such a claim for all the combinations in (1). Yet the nature and use of evidence from semantic structure in explanations of pidgin/creole genesis are not well understood compared to evidence from lexicon, phonology, and morphosyntax – both a cause and an effect of the relative scarcity of studies of semantic structures of PCs, and of languages in general. Parkvall’s (2000, p. 113) statement, “[s]ubstrate influences in semantics remain a severely underexploited area of study,” is an accurate assessment

442

George Huttar

reflected in that volume itself (cf. Arends, Kouwenberg, and Smith 1995, pp. 108–9. for a similar assessment). The general principles, problems, and major hypotheses of pidgin/creole genesis described elsewhere in this volume of course apply, but just what is appropriate use of semantic evidence still remains vague, for reasons worth exploring at least briefly. After making such an exploration and discussing the nature and use of semantic evidence (section 2), I deal with methodological issues (section 3), summarize research to date (section 4), and apply its results to questions of pidgin/creole genesis (section 5). Precisely because there has been little research on creole semantic structures so far, the programmatic sections 2 and 3 will appear more substantial than the report on what we know so far in section 4. This chapter is mostly about creoles rather than pidgins, because pidgins usually don’t have as extensive a lexicon as creoles do, hence don’t exhibit as much semantic structure; and the semantic structures of pidgins are generally even less documented than those of some creoles. Moreover, research such as that discussed in sections 4–5 suggests that the second language status of pidgins precludes semantic elaboration even in the case of extended pidgins. Another bias of this chapter is that most of the examples in it come from Atlantic creoles, since they are the creoles I am most familiar with. While data from other creoles are also included, the applicability to other creoles of the principles proposed here should be treated as a question for empirical investigation. Certainly Mufwene (2001), for example, is right to stress that each pidgin or creole must be looked at in terms of its unique ecology; but to the extent that principles described here turn out to be valid for all creoles regardless of their sociohistorical background, to that extent these principles express a general, universal framework within which the specifics of various languages’ ecologies operate. Before we proceed any further, a terminological note is in order: Frequently the term “lexifier” is used as a synonym for “superstrate.” I use the term “superstrate” and avoid “lexifier,” because the latter term is often used in a way that perpetuates the idea that the vast majority of the lexemes of a particular pidgin or creole come from the superstrate language (e.g., Mufwene 1993a, p. 204, n. 2; Thomason 2001, p. 159). We may in fact know that to be the case for many pidgins, and even for many creoles. But there are also some indisputable creoles for which that is not the case – see Price (1975) on Saramaccan, Bilby (2000) on Aluku, and Holm’s (2000, p. 115) reference to Bollée’s experience with Haitian Creole “that the proportion of non-French-derived vocabulary grew with the completeness of the lexicon.” Rather, the respective proportions of superstrate-, substrate-, and adstrate-derived lexemes in a particular language is best treated as a question for empirical research.

2

The Nature and Use of Semantic Evidence

Most basically, semantic structures have been little studied by creolists probably because they have been little studied by linguists generally, compared to

Semantic Evidence in Pidgin and Creole Genesis 443 phonology and morphosyntax. It is often anthropologists or linguists whose research interests include exploring the relations between language and culture, rather than linguists more generally, who have contributed much of what we currently know about semantic structures in the sense I am talking about here. Mühlhäusler’s (2000, p. 339) comment is pertinent here: “Creolists on the whole are not familiar with the language of cognitive anthropology.” True, lexical semantics is a flourishing field within linguistics, and has not been entirely ignored within creole studies. But the formal treatment of language structures that is assumed in much recent linguistics to be the only paradigm for studying language, with semantics thought of by some as largely or exclusively a matter of formal semantics with a logical calculus of predicates, arguments, quantifiers, and the like (e.g., Seuren & Wekker 1986, p. 63), has limited some of this work to semantic topics for which formal mechanisms have already been developed. Finally – and this topic will move us to the nature and use of semantic evidence – another reason that semantic structures of PCs have been relatively little studied is precisely because their position along some important continua is not well understood compared to other parts of a language. Consider the continuum of “borrowability” – the ease, as measured by frequency, with which certain kinds of features are borrowed from one language to another. It is generally agreed that (with other factors such as the typological similarity of the languages in question and the sociolinguistic ecology of the situation kept constant) individual content lexemes are relatively easy to borrow from one language to another, inflectional affixes are relatively difficult to borrow, and other kinds of language features occupy various points between these two extremes (Thomason & Kaufman 1988; Thomason 2001; for a thorough discussion of scales of borrowability, see Field 2002, on which much of the discussion here depends). Quantitatively, this means that for two languages A and B in contact, A will have borrowed more content lexemes than inflectional affixes from B. Temporally and implicationally, it means that A will have borrowed content lexemes from B before borrowing inflectional affixes, and therefore its borrowing of inflectional affixes implies its having previously borrowed content lexemes. Sociolinguistically, it means that borrowing of content lexemes requires less intense contact between A and B than does borrowing of inflectional affixes: (3) content lexemes > . . . > inflectional affixes For many kinds of language features, their relative positions on the scale in (3) are well established. Thus (3) can be filled in to include additional kinds of features, examples of which are given in (4): (4) content lexemes > function words (e.g., conjunctions) > clitics > inflectional affixes Such scales, whether quantitative, temporal, or sociolinguistic, are agreed to be more than arbitrary empirical generalizations with no reason for being

444

George Huttar

ordered the way they are rather than some other way. Rather, they reflect a scale of structural integration: “less tightly structured features are easier to borrow than features that fit into tightly integrated closed structures” (Thomason 2001, p. 69; cf. Chaudenson 2001, p. 312). But is “borrowability” the dimension most relevant to pidgin and creole genesis? Thomason argues, with strong empirical evidence, that “shift-induced interference,” involving the introduction of interference features into a second language, shows a scale very different from those above: “unlike borrowing, it does not start with the lexicon. Instead, it starts with phonology and syntax” (Thomason 2001, p. 75). The crucial question for pidginization and creolization, of course, is whether these processes are more akin to borrowing or to shiftinduced interference. Thomason claims that “pidgin/creole genesis is akin (though not identical) to shift-induced interference” (2001, p. 158; cf. Thomason 1993, p. 285; cf. Mufwene 1997, p. 63). In the case of New World plantation creoles, Thomason’s position predicts that if substrate speakers (slaves) were in fact trying to learn the local superstrate language, their resulting variety of the superstrate would include structural interference from their native languages more than lexical interference. And even if the slaves were not trying to learn the superstrate, but were only striving to acquire some means or other to communicate with each other and with their masters (e.g., Baker 1996), the resulting language should still reflect the substrate languages more in structure than in lexical stock. And that is of course what we find: although the lexical predominance of superstrate sources in creoles is often exaggerated because of a focus on everyday, nonspecialist vocabulary, nevertheless the superstrate origin is much clearer for much of their lexicon than for their phonological, morphosyntactic, and semantic structures. Or, to put it the other way around, it is in their structures, not in their lexical stock, that creoles are most clearly different from their superstrates. Similar conclusions hold for creoles like Solomons Pijin and Vanuatu Bislama, where the structural parallels to the substrate languages are even clearer (Keesing 1991), or for the Portuguese-based creole of Guinea-Bissau (Kihm 2000). The relative borrowability scales or hierarchies shown in (3) and (4), or the very different ones that could be drawn up for shift-induced interference, omit many language features, including phonological, pragmatic, and discourse features, and, pertinent here, semantic structures: where do they fall on such scales – nearer the pole of greatest structural integration, or nearer the other end, compared with, say, order of clause constituents (VO vs. OV) or relativization strategies? Taking first an a priori approach to answering that question, we can start with the assumption that the more arbitrary a component of a language is, then the less structurally integrated it is with other parts of the language (cf. Mühlhäusler 1986, p. 128). In the case of the specific associations of phonological shapes with meanings in a language’s lexicon, we are dealing with the most arbitrary language features of all. Our assumption implies that these associations are the least structurally integrated parts of a language. Given the

Semantic Evidence in Pidgin and Creole Genesis 445 application of shift-induced interference to pidginization and creolization, that assumption fits with the fact that substrate influence is more evident in structures, and superstrate influence more evident in the phonological shape of lexemes. But associations of phonological shapes with meanings are not all there is to a language’s lexicon. A lexicon is also characterized by just what meanings are lexicalized in a given language in the first place, and by paradigmatic and syntagmatic relations among lexemes and among lexicalized meanings – clearly structural matters, and arguably less arbitrary than the phonological shapes associated with particular meanings. Evidence for their being less arbitrary is that many meanings are lexicalized in all, or nearly all, languages (a fundamental assumption behind the various “Swadesh lists” and the atheoretical notion of “basic vocabulary”) and that the same general structural principles of hyponymy (generic–specific hierarchy), part–whole relationships, synonymy, various kinds of oppositeness, etc. are characteristic of the lexicons of all languages. It is in fact likely that the lexicalization of certain concepts, and the kinds of relationships among lexemes, are constrained by universals of human perception and cognition (Miller & Johnson-Laird 1976; Langacker 2002) – but let us take that idea here simply as a possible explanation for the universally attested lexicalizations and relationships, not as evidence for their lack of (total) arbitrariness. Mühlhäusler’s inclusion of “lexical semantics” (which he uses to mean primarily semantic structures) in his list of areas of a language where “substratum influence will be most pronounced” (1986, p. 129), supports the nonarbitrariness of semantic structures. On the other hand, Thomason and Kaufman (1988, pp. 74, 121) conjecture that “lexical semantics” comes near the less structured end of things, although it is not clear that they are thinking so much of semantic structures as of the meanings associated with individual words. If such lexicalizations and interlexeme relationships were completely universal, all of them found in every language, then they would of course be found in pidgins and creoles merely as a result of “universal grammar.” They would, thus, cease to be of interest for sorting out the roles of universal grammar, superstrate, substrate, etc. in pidgin/creole genesis. But unless one holds that they have been overridden in the superstrate languages, they are obviously not all universal – or we wouldn’t be noticing in the first place that some semantic structures of PCs are not the same as those of their respective superstrates, and asking where these semantic structures come from. Our considerations so far lead us to expect that many come ultimately from substrate sources, whether from specific languages, from genetic or areal groups of languages, or from historical processes of language and culture development that start from such sources. Nevertheless, probably most linguists assume that the whole set of semantic structures of a language is nearer the loosely structured end of a continuum of language features. First, whatever integration there may be within the semantic structures of a language, we are used to thinking of such things as

446

George Huttar

unrelated to other aspects of language structure. For example, no one has demonstrated a correlation between head-first vs. head-last constituent order with taxonomy of plant names, number of basic color terms, or whether a language lexicalizes the difference between ‘hand’ and ‘arm’. As Mufwene (1993a, p. 193) puts it, “I know of no language and culture study which has shown correlation of culture with, for example, constituent order, internal morpheme structure in a word, or the particular morphosyntactic strategy used for expressing genericness.” (An exception to this general expectation of lack of relation between (largely if not completely) culturally determined semantic categorization and a feature of a language is constituted by systems of noun classifiers – where morphosyntax and semantic structure are interrelated to a great degree – perhaps made most famous by Lakoff’s (1987) Women, Fire and Dangerous Things). Furthermore, although we may ask why we should expect a correlation between, say, the structure of the color domain and that of the kin domain, connections between seemingly disparate semantic domains are easy to document – such as that between political structure and classification of birds in the Wayampi culture and language of Brazil (Jensen 1988): A village chief, tovijã, is thought of not so much in terms of power but as being a prototype of a good villager (surpassing others in wisdom and industriousness); similarly, the tovijã of a set of bird species (e.g., parrots) is the prototype of that set. Other villagers are spoken of as in the domain of the village tovijã, and other parrots as in the domain of the parrot tovijã. Some villagers are more like their tovijã than are others, and may be spoken of as tovijã miti ‘chief small’, while some parrots are more like their tovijã than are others, and may also be spoken of as tovijã miti. Other villagers, and other parrots, are simply emigwai ‘subordinates’. The same structure is found in other semantic domains, such as numbers, among which ‘100’ is considered a tovijã with certain other numbers its emigwai ( Jensen, p.c., 2002). Another instructive example is Hill and Hill’s (1997) documentation in Hopi and the Tepiman languages of a morphologically realized correlation, based on an extended notion of protrusion/intrusion, between certain body parts, features of the physical landscape, human artifacts, animals, and kin categories – the last of these varying from one language to another in a way that leads the authors to posit an “anti-Whorfian” effect of culture determining linguistic structure, in particular semantic structure. More generally, recent work in cognitive linguistics has demonstrated structural parameters that apply across a wide range of semantic domains within a language (see Jackendoff 1996). We can also take an empirical, a posteriori, approach to answering the question of where semantic structures fall on the scale of structural integration. To the extent that methodologically appropriate study demonstrates creole semantic structures that are also found in languages known on other grounds to be reasonable candidates for being substrate languages for a particular creole, while the same semantic structures are not found in that creole’s superstrate(s) and are not typologically common, to that extent we can conclude that these

Semantic Evidence in Pidgin and Creole Genesis 447 structures have substrate sources, and, by implication, demonstrate a corresponding degree of structural integration. We turn, then, to methodological issues in such research, and then to a summary of research so far, results of which provide a provisional empirical answer to the question of the extent to which creole semantic structures derive from substrate sources.

3 Methodology We start with some issues raised by the historical nature of our research. For semantic structure research as for research into any other feature of PCs, determining possible superstrate and substrate sources ideally means comparing a particular PC, its possible superstrates, and its possible substrates as they were at the time the pidgin or creole came into being. But in most cases we have to infer those earlier states from what the languages are like now, constraining our inferences by what is known about language change generally. Since a language, besides expressing and passing on much of a culture, also is a part of a culture, it should also be worthwhile to take into account what is known about culture change generally. This consideration is not a usual part of diachronic linguistics, nor has it played much of a part in PC studies. But in the case of semantic structures, which often reflect cultural values and assumptions more closely than do other features of a language, it seems especially appropriate (pace Huttar 1975, pp. 692–3). In the case of at least some creoles, we can draw on what has been proposed about the historical development of cultural institutions other than language by societies confronted with the necessity of creating a new culture, including a new language, such as those societies who created the Maroon creole languages of Suriname. Clearly, in considering possible changes in a creole well after the initial period during which it arose and became stabilized we are no longer dealing directly with the matter of creole “genesis” that is the focus here. Yet changes subsequent to the initial period must be somehow factored out if we want a clear picture of what was going on during the genesis of the creole, in semantic structures or any other part of the language. Much of the historical discussion here primarily concerns the circumCaribbean creoles. Just how the cultural institutions of the societies creating these languages were established is a matter of controversy, centering around the relative extent of retention of African cultural features as opposed to innovation of institutions built perhaps on very general African models. One point of view is represented by Bilby and N’Diaye (1992, p. 57): [The struggle for survival] required the creation of whole new societies, cultures and languages . . . In devising methods of subsistence, military strategies, systems of authority and shared languages, maroons typically selected from the full range of cultural resources available to them: African, Native American and European. The breadth of this spectrum of resources improved their chances for survival.

448

George Huttar

The claims here have been made more often and with extensive documentation by Richard Price and his co-authors (e.g., Mintz & Price 1976; Price & Price 1980; Price 1992), in works which stress the cultural and linguistic heterogeneity of the Maroon societies: One of the most notable features of the initial groups of escaped slaves that coalesced in the Suriname forests was their cultural heterogeneity. Each such group was composed of Africans from a large number of ethnic (“tribal”) and linguistic backgrounds, many of whom had been slaves – for varying periods of time – on a particular plantation in coastal Suriname. (Price & Price 1980, p. 194)

What is important for semantic structure research in these claims, besides the reminder about linguistic heterogeneity, is the idea that this heterogeneity meant that Suriname Maroon societies developed their cultural institutions by drawing on a large number of cultures which, for all the areal similarities they may have had, also differed enough that the cultural institutions familiar to some subset of the ex-slaves involved in forming a new society could not be transplanted directly, but only in negotiation with the cultural memories and experiences of other ex-slaves from numerous other cultural backgrounds. There is no a priori reason to exclude language completely from the cultural institutions that were developed in such a way. And especially for those aspects of language that most directly reflect cultural structures, such as systems of kin terminology, but also semantic domains where the relation to culture is not so apparent to the casual observer (as in the examples mentioned above from Jensen 1988 and Hill & Hill 1997), it is reasonable to suppose that often we would search in vain for a specific African substrate language (or culture) to which to attribute a particular semantic structure. If the position of Price cum suis is correct, then we are faced here with more than the familiar problem of areal features of West African languages that sometimes force us to attribute some creole feature to any of a group of substrate languages, rather than finding a more specific source. Cultural institutions that developed out of a negotiation among several cultures came to reflect no specific culture, but merely a vague, yet intuitively unmistakable African character. To take one example from Price (1992, p. 64): In the development of the kinship system of the Ndjuka Maroons of Suriname, writes André Kobben, “undoubtedly their West-African heritage played a part . . . [and] the influence of the matrilineal Akan tribes is unmistakable, but so is that of patrilineal tribes . . . [and there are] significant differences between the Akan and Ndjuka matrilineal systems”.

The example of the Ndyuka kinship system is of course reflected linguistically in part in the structure of the semantic domain of Ndyuka kin terms. If the cultural system does not go back to a single source, we cannot expect the lexical system to do so. For example, Ndyuka has a single term pai (< Portuguese pae ‘father’) with the meanings ‘father-in-law’ and ‘son-in-law’, and

Semantic Evidence in Pidgin and Creole Genesis 449 another term mai (< Portuguese mãe ‘mother’) with the meanings ‘motherin-law’ and ‘daughter-in-law’. Yet, so far, I have found no language of the relevant part of coastal West and Central Africa with such a pattern – not even Fongbe, the language variety currently most often cited with reference to the Suriname creoles’ substrates (see Migge & Smith 2007). Given the crosscultural and cross-linguistic processes that went into Ndyuka society and culture, it would perhaps be surprising if we ever did. Besides the forming of cultural institutions, including language, through negotiation among several specific African cultures, what we observe today is the result of two to three centuries of continual development since the initial period of creating a stable society with cultural and linguistic norms: We can assert with some confidence, then, that within the earliest decades of the African presence in Suriname, the core of a new language and a new religion had been developed; subsequent centuries of massive new importations from Africa apparently had little more effect than to lead to secondary elaborations . . . However, we do not mean to imply that in some special situations, late-arriving Africans were not able to exert considerable influence on local Afro-American institutions. (Mintz & Price 1976, p. 26)

There are two points worth noting here. The first is the potential for change over the centuries to a degree and in a direction often unknown, a special case of the general problem in PC origin studies mentioned at the beginning of this section. This is of course not dependent on the tie between culture change and language change, being merely the outworking of normal diachronic change in any human language, whatever its origin. The second – assuming that we can extrapolate from general cultural processes to linguistic ones, at least as far as language features closely tied to culture are concerned – is the potential for late arrivals in a given Maroon society to have an effect not only on some part of the culture, but in particular on the language, including some semantic structures. The first of these does not entail the second, for it is logically possible that later changes are constrained by initial conditions having to do with who the first generation (or two) of speakers were – cf. Mufwene’s (2001) “founder principle.” If latecomers could in fact have some effect on the semantic structures of a creole, then methodologically that means that slave shipping records and accounts of escapes of slaves from plantations from periods well after those when the respective Suriname creoles took their basic form may still need to be included in our research on substrate sources of creole semantic structures. But a different view from that of Price and others stresses the continuity of specific African cultures – those borne by members of dominant groups of slaves in particular New World settings – and their associated languages, rather than innovation. Thus, Alleyne (1993, p. 178) states that “Dahomey Fon characteristics predominate in Haiti, Yoruba in Cuba, Bantu in several parts of Latin America, and Twi-Asante in Jamaica and Barbados.” Alleyne’s position also questions the high degree of cultural and linguistic diversity among

450

George Huttar

founders of particular circum-Caribbean creole-speaking societies emphasized by Price cum suis. Note that the degree of homogeneity of a particular group that created a creole is an entirely separate issue from that of the creole’s subsequent development over the centuries, and is one that in principle and in fact varies a great deal from one creole to another, as recent demographic studies have shown (see Arends, this volume, and works cited there). It should also be noted that the position of Price cum suis, while claiming that most cultural complexes are not derived from a specific group, does not preclude some being retained from such a specific source. Yet the two positions may not be far apart, for both end up arguing for a combination of continuity and creative innovation. But if the position represented by Price is in fact overstated (and some of Alleyne’s points cannot be gainsaid), the implications for research into the genesis of circum-Caribbean creole semantic structures are two: for some such structures we can be more optimistic about identifying specific substrate languages, or at least closely similar groups of languages; and we can focus our attention on the earliest arrivals into what became a particular creole-speaking community, giving less weight to the putative contribution of those coming on the scene later. With regard to specificity of origin of some cultural features, an example of retention of a specific cultural complex related to language is provided by the fact that a well-known Ghanaian ethnomusicologist readily claimed to understand the message of the talking drums played by the court drummer of the paramount chief of the Ndyuka on a visit to the latter’s seat at Drietabbetje (Joseph Grimes, p.c., 1967). With regard to the length of time over which new arrivals in a creole-speaking community could permanently affect cultural and linguistic structures, especially semantic structures, creolists need to keep an open mind in the debate, ready to acknowledge and investigate the possibility of some of the features of today’s creoles being due not only to the input of “founders,” but also to that of speakers of other languages who joined a particular society some decades after we believe the creole to have become fairly stable. At any rate, as far as the languages themselves go, in most cases we have to proceed with the current states of the languages in question. For even for creoles such as Sranan for which we have fairly early documentation, that documentation almost never focuses on questions of semantic structures, leaving us to infer bits and pieces about such structures from example sentences and texts as best we can. The same can be said by and large for earlier descriptions of potential substrate languages of Africa; for European superstrate languages we are in a much better position. What, then, are the methodological principles we need to apply in working with the languages at whatever stage we can access them? First and most familiarly to creolists, as with any other area of PC origin research, we must include a range of languages that allows us to distinguish the effects of universals, substrates, superstrates, effects of first or second language acquisition, sociohistorical setting, etc. Specifically, beyond comparing a

Semantic Evidence in Pidgin and Creole Genesis 451 particular creole with a particular substrate language and a particular superstrate language, we must look at semantic structures in a representative sample of creoles and typologically diverse non-creoles from around the world – the latter showing us which semantic structures attested in one or more creoles could in fact be due to universal factors, as suggested by their occurrence in typologically and geographically diverse languages other than creoles. Huttar (1975, 1991), for example, includes PCs from the circum-Caribbean area, West Africa and elsewhere, non-PCs of language families represented in the environment of the Suriname creoles, non-PCs from West Africa, and nonPCs from around the world. A serious defect in the language sample in both these studies is the lack of any Kwa language (i.e., a member of Greenberg’s 1963 wider Kwa grouping, or of the more recently accepted narrower grouping as in Williamson 1989), let alone any variety of the Gbe group. The results of a comparison of these six kinds of languages is discussed below (section 4). With reasonable care and effort we can ensure that a methodologically adequate sample of these various kinds of languages is part of our database, preventing us, for example, from attributing to some specific substrate language or language family a semantic structure found in a much wider range of languages of Africa or, for that matter, elsewhere in the world. For example, it is clear that the use of a single morpheme for both ‘hand’ and ‘arm’ and another morpheme for both ‘foot’ and ‘leg’ is characteristic not only of many New World creoles but also of many languages of West Africa. But we cannot readily infer that the semantic structure in the creoles therefore came from some specific or general African substrate, once we recognize that the same structure is widely found throughout the world, especially in languages spoken in tropical and subtropical areas (Witkowski & Brown 1985). Similarly, the use of ‘door’ plus ‘mouth’ for ‘doorway’ in both New World creoles and many West African languages diminishes in significance when we recognize that the same phenomenon is found in many languages of Middle America (Campbell, Kaufman, & Smith-Stark 1986, p. 553). In both cases, the frequent occurrence of these phenomena across widely separated unrelated languages raises the possibility that universals played a role in their occurrence in the creoles – most likely in addition to, rather than instead of, substrate influence. In the case of a single lexeme, derived from Eastern Ijo, being used for both ‘face’ and ‘eye’ in Berbice Dutch and the lexeme for ‘eye’ having become the lexeme for ‘face’ in Eastern Ijo (Kouwenberg forthcoming), the situation is slightly different, for two reasons: the strong Eastern Ijo contribution to Berbice Dutch is indisputable on other linguistic grounds (Smith, Robertson, & Williamson 1987; Kouwenberg 1992); and the use of one lexeme for ‘face’ and ‘eye’ is not as widely distributed among the languages of the world as are the other examples just mentioned. It is attested, however, for Greek and for Sango, Tarascan, and more than 30 Mayan languages (Andersen 1978, p. 354); so even in this case the possibility of universals of semantic structure affecting the selection of particular substrate features cannot be ruled out. (I assume, with Kouwenberg, that the Eastern Ijo source lexeme was in

452

George Huttar

transition and was used for both ‘face’ and ‘eye’ at the time that some of its speakers were forcibly removed to the Guianas.) Perhaps more difficult than using an appropriate sample of languages is obtaining an appropriate set of semantic structures to investigate, for it is clearly desirable to go beyond one or two simple examples like those in the preceding paragraph to a broad range of semantic structures found in one or more creoles. Our sample should eventually include domains with clear relationships to culture, such as kinship, cooking, music, and biological domains (plant and animal names); domains where the relation between semantic structure and culture seems more indirect, such as expressions referring to emotions and personality; and domains where the relation between language and culture is more opaque, if it exists at all, such as the range of meanings of Ndyuka mofu ‘mouth’, which in addition to the sample given in (1) is used also to refer to food and to spoken messages. Obtaining an appropriate sample of languages is largely a practical matter, since we probably already know the theoretical basis for the kinds of languages the sample should include. Obtaining data on an appropriate set of semantic structures is also a practical matter, once we choose on good theoretical grounds what kinds of domains should be included, such as those suggested above. More difficult, however, is how to treat degrees of difference among various meanings of a given creole or non-creole lexeme when we are trying to measure the similarity between a creole and some other language or groups of languages – Kwa languages, West African creoles, Eastern Oceanic Austronesian substrate languages, or some other relevant set. For example, suppose a given language uses one lexeme, as is true of Ndyuka bee, to refer to the abdomen, to the palm of the hand, to the womb, and to pregnancy. How many different senses of the lexeme in question are represented here? Does the use of the same lexeme in some other language in, say, both the ‘womb’ context and the ‘pregnancy’ context count as much as does the use of the same lexeme in yet another language in both the ‘womb’ context and the ‘palm of hand’ context? How do we measure the degree of similarity between that first language and Ndyuka, and how do we compare that similarity with the similarity between the other language and Ndyuka? Until we have a solid semantic framework for answering such questions, the best we can do is to assume and hope that the errors introduced by treating all uses of a particular lexeme as equidistant in semantic space will cancel each other out, given a broad enough sample, rather than having a systematic effect in a particular direction. This last point exemplifies another aspect of sound methodology in research into the origins of semantic structures (or into other features of PCs), namely the need for statistical rigor to assess the probability of our results being due to chance, to help us avoid “findings” that owe more to our initial assumptions than to the nature of the data (nonparametric measures are used in Huttar 1975, 1991). A more sophisticated approach using multiple analysis of variance would go beyond testing for statistical significance by teasing apart the effects on semantic structures of the several possible factors – substrate,

Semantic Evidence in Pidgin and Creole Genesis 453 superstrate, culture and extra-linguistic history, second language acquisition universals, language universals, and the like. Finally, all of the above considerations add up to an obvious general point for any research into the relative effects of a number of potential factors, namely the necessity for careful research design.

4 Research So Far Published work on creole semantic structures ranges from isolated and anecdotal comments to more extensive comparison of one creole with its substrates to a few studies trying to approximate some of the methodological requirements outlined above. Even the first of these has some value in illustrating and suggesting the possibility, if not likelihood, of substrate influence on semantic structures. Thus when Holm (2000, p. 116) refers to Twi, Ibo, and Yoruba each having one morpheme covering ‘tree’, ‘wood’, and ‘stick’, and to Bahamian Creole English as having the lexeme stick with the same range of meanings, an explanation for the resemblance is called for. Such examples are usually chosen to show a semantic pattern different from that of the superstrate, in this case English. The suggestion that this association of what from an English point of view are three meanings with one form is due to substrate influence is of course by itself no more than a suggestion. To test this suggestion or hypothesis, we need to find out how widespread such a union of meanings for one lexeme is among other creoles, especially among creoles with no historical connection with Bahamian Creole English or with any other creole with African connections, and among other languages of the world (and, for that matter, whether it is attested for seventeenth- or eighteenthcentury varieties of English). Of course we also need to be sure that one or more of the African languages mentioned, or at least languages close to them geographically or genetically, were likely spoken by some of the people involved in the rise of the PC in question. Turning to more systematic investigation of semantic structures, we note first that most of the work has focused on structures that appear different from those of the respective superstrates. This is to be expected, since structures parallel to those in the superstrates do not seem to require explanation. But note that we have here, in fact, two opposing biases: on the one hand, the idea that superstrate-like structures do not require explanation reflects an assumption that creole structures in general derive from superstrate sources; on the other hand, research into parallels between creole and substrate sources has tended to arise in fact, though not by theoretical necessity, from the assumption that it is the substrates that are the source of the creole structures. We look first at the work of Mühlhäusler on semantic structures of Tok Pisin. In Mühlhäusler (1979), we have an extensive longitudinal study of the development of the Tok Pisin lexicon, including data on some semantic structures. Mühlhäusler shows how the addition of new lexemes to Tok Pisin in its

454

George Huttar

expansion stage (which in Mühlhäusler’s scheme follows its stabilization stage and precedes its creolization stage) results in a slight increase in structural complexity of various domains by introducing another level along the generic–specific dimension. For example, corresponding to a general term of “sensing,” harim, in the stabilized stage, we have the three specific verbs smelim ‘to smell something’, pilim ‘to feel something’, and harim ‘to hear something’ in the expansion stage. Corresponding to the noun rip ‘reef’ in the earlier stage, we have in the later stage karanas ‘coral, coral rubble’, siparam ‘reef in open sea’, and rip ‘reef’. In this instance the older term, unlike harim, has retained its generic use. Note that this example from Tok Pisin involves only the semantic “restructuring” of an increase in the number of levels in a taxonomic hierarchy, a process that is presumably universal and therefore has no necessary connection with any substrate or superstrate language. The specific distinctions made between, for example, ‘coral rubble’ and ‘reef in open sea’, however, are obviously not universal, but reflect a culturally salient distinction in the physical environment. (According to Peter Mühlhäusler, p.c., 2002, today’s young Tok Pisin speakers do not use karanas or siparam, but only generic rip, illustrating a change of the semantic structure of a domain in the opposite direction, toward a decrease in the number of taxonomic levels.) Mühlhäusler (2000) is pioneer work, documenting carefully the development of Tok Pisin plant and animal names from the early to the late twentieth century – with comments also on the period 1860–1900 – in terms of the various levels of their respective taxonomic hierarchies and the principles governing such hierarchies discussed by ethnosemanticists. He concludes that Tok Pisin semantic structures in the two domains illustrated here reflect neither those of English nor those of the local, potential substrate, languages. Rather, “the development of life form classification is largely an internal process in Tok Pisin” (2000, p. 355). By “internal” he apparently means without reference to other languages in the environment, for he also states “that the development of the structured lexicon is governed not by inherent forces but largely by external pragmatic ones and that postulated universal hierarchies for the development of lexical structures do not apply in the case of Tok Pisin” and that this confirms his findings for other contact languages such as Mauritan Creole, Pitkern/Norfolk, and St. Helena English (2000, p. 356). Mühlhäusler’s results for Tok Pisin are of interest in demonstrating changes in semantic structures, chiefly by some increase in number of taxonomic levels, through the four stages he posits for Tok Pisin: “jargon stage,” “stabilization,” “expansion phase and increasing creolization,” and “post-pidgin/creole phase” (2000, p. 354). He argues that the rich taxonomic hierarchies of the substrate languages are nowhere reflected in Tok Pisin. This result contrasts sharply with the findings of Huttar (1975, 1991) for substrate influence in semantic structures in Ndyuka. The significance of this contrast is taken up in section 5. Besides his own research, Mühlhäusler has also brought to our attention some work on another Pacific creole, Australian Kriol. The following shows the structure of the semantic domain of kinship in Australian Kriol, a structure

Semantic Evidence in Pidgin and Creole Genesis 455 paralleled in local Aboriginal languages (see also Jourdan, this volume, for a comparative analysis of a corpus of kinship terms from a group of vernacular languages from the Solomon Islands with kinship terms from Solomon Island Pijin): (5) dedi mami anggul andi greni

< < < <