Creoles, Contact, and Language Change: Linguistics and Social Implications (Creole Language Library)

Creoles, Contact, and Language Change Creole Language Library (CLL) A companion series to the “Journal of Pidgin & ...

Author: Armin Schwegler | Genevieve Escure

86 downloads 4203 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Creoles, Contact, and Language Change

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

Editors Jacques Arends

John Victor Singler

Amsterdam

New York

Editorial Advisory Board Mervyn Alleyne

Salikoko Mufwene

Kingston, Jamaica

Chicago

Norbert Boretzky

Pieter Muysken

Bochum

Nijmegen

Lawrence Carrington

Peter Mühlhäusler

Trinidad

Adelaide

Glenn Gilbert

Pieter Seuren

Carbondale, Illinois

Nijmegen

George Huttar

Norval Smith

Dallas

Amsterdam

John Holm Coimbra

Volume 27 Creoles, Contact, and Language Change: Linguistic and social implications Edited by Geneviève Escure and Armin Schwegler

Creoles, Contact, and Language Change Linguistic and social implications

Edited by

Geneviève Escure University of Minnesota

Armin Schwegler University of California, Irvine

John Benjamins Publishing Company Amsterdam/Philadelphia

8

TM

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

Library of Congress Cataloging-in-Publication Data Creoles, contact, and language change : linguistic and social implications / edited by Geneviève Escure, Armin Schwegler. p. cm. (Constructional Approaches to Language, issn 0920-9026 ; v. 27) Includes bibliographical references and index. 1. Creole dialects. 2. Languages in contact. 3. Linguistic change. I. Escure, Geneviève. II. Schwegler, Armin, 1955- III. Series, PM7831.C734 2004 417’.22-dc22 isbn 90 272 5249 1 (Eur.) / 1 58811 551 8 (US) (Hb; alk. paper)

2004055094

© 2004 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Contents Preface

vii

Part I: Historical 1. The origins of Macanese reduplication Umberto Ansaldo and Stephen Matthews



2. Court records as a source of authentic early Sranan Margot van den Berg and Jacques Arends

2

3. Garifuna in Belize and Honduras Geneviève Escure

35

4. The Nova Scotia–Sierra Leone connection Magnus Huber

67

Part II: Acquisition 5. The development of variable NP plural agreement in a restructured African variety of Portuguese Alan N. Baxter

97

6. Second language acquisition in creole genesis Fred Field

27

7. OT and the acquisition of Jamaican syllable structure Rocky R. Meade

6

Part III: Aspects of structure 8. Double-object constructions in two French-based creoles Dany Adone

89

9. Passive voice in Papiamento Eva Martha Eckkrammer

209

10. Tone assignment on lexical items of English and African origin in Krio Malcolm Awadajin Finney

22

11. TMA and the St. Lucian Creole verb phrase David B. Frank

237

vi

Contents

Part IV: Discourse/identity 12. The Limonese calypso as an identity marker Anita Herzfeld and David Moskowitz

259

13. The speech event kuutu in the Eastern Maroon community Bettina Migge

285

14. Reflexivity in French-based creoles Katrin Mutz

307

15. The role of style and identity in the development of Hawaiian Creole Sarah J. Roberts

33

Index

35

Preface This volume contains a selection of papers presented at three consecutive meetings of the Society for Pidgin and Creole Linguistics. These were held in Washington, D.C. (January 2001); Coimbra, Portugal (June 2001); and San Francisco (January 2002). Thirty-one revised papers were originally submitted for inclusion in this volume. Two or more specialists refereed each submission. We eventually accepted fifteen articles, all of which are revised and much-extended versions of the original conference papers. The fifteen articles in this volume offer a balanced sampling of creolists’ current research interests. Even though all of the contributions address issues directly relevant to pidgin/creole studies and other contact languages, their topics and approaches vary greatly. Bettina Migge and Anita Herzfeld, for instance, use pragmatic and sociolinguistic approaches, while others (Umberto Ansaldo and Stephen Matthews, Margot van den Berg & Jacques Arends, Geneviève Escure, Magnus Huber) rely primarily on sociohistorical methods; still others either combine sociolinguistic and historical approaches (see, for instance, Sarah Roberts’ study on Hawaiian Creole) or they favor predominantly theoretical orientations (see the articles by Dany Adone, Malcolm Finney, and Rocky Meade). The majority of papers address issues of morphology or syntax (Dany Adone, Umberto Ansaldo and Stephen Matthews, Alan Baxter, Margot van den Berg and Jacques Arends, Eva Eckkrammer, Geneviève Escure, David Frank, Katrin Mutz). Some of the research presented makes use of phonological analysis (Malcolm Finney, Rocky Meade, Geneviève Escure). Alan Baxter, Fred Field, and Rocky Meade study language development from the point of view of acquisition. Geneviève Escure examines the opposite, namely, attrition. A few papers examine discourse strategies and style (Bettina Migge, Katrin Mutz, Sarah Roberts), or broader issues of social and ethnic identity (Anita Herzfeld, Bettina Migge, Sarah Roberts). While this array of topics and perspectives is reflective of the diversity of the field, there is also much common ground in that all of the papers adduce solid data corpora to support their analyses. The range of languages analyzed spans the planet, as it includes Asia (Macanese), the Pacific (Hawaiian), the Indian Ocean (Morisyen, Seselwa), Africa (Krio, Tonga Portuguese), the Caribbean (Guadeloupean, Haitian, Jamaican Creole, St. Lucian Creole, Papiamento), Central America (Limon Creole, Garifuna), South America (Eastern Maroon Creole, Sranan), and North America (AAVE, Gullah, Nova-Scotian varieties). We have grouped the fifteen articles into coherent sections. Undoubtedly, a different arrangement would also have been possible, as some of them could be placed in more than one of these four categories:

viii

Preface

Historical: Acquisition: Aspects of structure: Discourse/identity:

Ansaldo/Matthews, Van den Berg/ Arends, Escure, Huber Baxter, Field, Meade Adone, Eckkrammer, Finney, Frank Herzfeld, Migge, Mutz, Roberts

Broadly defined as historical, the first section includes contributions whose primary or secondary focus is the origin or subsequent development of a contact variety. Written records as well as social, ethnic and demographic history are used to explain adstrate and substrate influences in language development. Ansaldo and Matthews describe how the convergence of Sinitic and Malay influences have conditioned reduplication in Macanese, the Portuguese-based creole of Macao. They conclude that the Sinitic adstrate was the most influential factor in morphological reduplication. Van den Berg and Arends examine 17th- and 18th-century court records as sources of early Sranan sentences. They discover that overt marking of tense (ben), mood (zal, sal, sa) and aspect (de) is attested from 1745 onward. A further finding is that copular da first emerged in presentative contexts. Escure discusses the mixed status of Garifuna and its current obsolescence. Garifuna’s various components (Arawak, Carib, French, Spanish, English, Belizean Creole) are traced to their ancestral sources via an examination of phonological and morphological processes. She also documents language attrition in terms of past tense marking — a feature that is represented differentially in Belize and Honduras. Huber presents several corpora of written records (lists of heads of households, diaries, letters) to trace the movements of ex-slaves from the US to Nova Scotia, and then to Sierra Leone. Phonological and morphological features gleaned from the written data support his claim that the transshipment of Nova Scotians to Sierra Leone may explain similarities between Gullah and Krio. The second section examines first- and second-language acquisition in an attempt to relate these findings to creole situations. Phonological, morphological, and syntactic phenomena are covered variously in the following articles: Baxter studies the growth of variable plural agreement in the noun phrase of the restructured Portuguese of the Tongas, descendants of Africans contracted on the Monte Café plantation of Sao Tomé in the 19th and 20th centuries. Basing himself on a VARBRUL analysis of plural NPs collected from three age groups, he attributes the appearance of variable plural marking in L1 Tonga Portuguese to the past role of L2 Portuguese in L1 acquisition. Field examines a sampling of English-based creoles (Hawaiian, Jamaican, Tok Pisin) that, during their formation, had various degrees of contact with the lexical source language (and its native speakers). He argues that secondlanguage acquisition processes are involved in the genesis of creole languages. To bolster his claim, he uses Pienemann’s (2000) processability hierarchy, and, for a broader perspective, he adduces data from Palenquero (Colombia). Meade discusses the acquisition of Jamaican syllable structure in several groups of Jamaican children — all learners of Jamaican Creole and English. He shows that Optimality Theory, accompanied by independently motivated constraints, can account for the

Preface

partial early inventories produced by the children, as well as for their eventual competence in the complete range of adult syllable types. The third section includes analyses of what is often considered the core of linguistic systems, namely suprasegmental phonology, syntax and semantics. The following articles illustrate a variety of approaches, some more descriptive, others more theoretical. Adone uses Wunderlich’s Lexical Decompositional Grammar to account for the distribution of dativized verbs in two French-based creoles, Morisyen and Seselwa. She offers evidence that both creoles possess two structural arguments, but that they also possess Double Object Constructions (DOC). Adone concludes that DOC constitute, in fact, one of the preferred universal structures associated with ditransitive verbs in the Morisyen and Seselwa lexicon. Eckkrammer analyzes the passive voice in Papiamento, the only Atlantic creole with a full passive. Its three passive markers (derived from Dutch, Spanish and creole) exhibit considerable dialectal variation. This observation is based on two separate written corpora of Curaçao, Bonaire and Aruba varieties. Finney adopts the position that Krio is a tonal rather than a pitch-accent language. There exists a difference in tone assignment in lexical items, depending on whether items are of English or African origin. Although the English stress pattern determines High tone in certain cases, the influence of contiguous African tonal languages cannot be discounted. In some data sets Finney finds tone to be the only contrasting element. Frank focuses on the use of TMA markers in St. Lucian Creole (SLC). He concludes that these TMA markers conform for the most part to the universal pattern proposed by Bickerton for creole languages. In some cases, however, the rules for SLC differ from those described for other Caribbean creoles. Frank suggests that to fully account for the TMA usage in SLC, discourse considerations must also be taken into account. The fourth section covers areas that broadly include socio-pragmatic issues, such as style, register, discourse context and identity. These areas have often been ignored in creolistics, but are given due attention in the contributions by Herzfeld, Migge, Mutz, and Roberts. Herzfeld discusses how calypso lyrics are representative of AfroLimonese identity among the Afro-Limonese of Costa-Rica. The author concludes that music (along with its lyrics) contributes to the maintenance of Limonese Creole, a language that is now under much pressure from Spanish. Migge investigates the social and linguistic properties of the kuutu (‘council meeting’) — a formal and highly structured event in the Eastern Maroon (Pamaka) community of Suriname and Guyane. The Eastern Maroon Creole linguistic practices of the kuutu are analyzed as face-saving strategies and social conventions that index social status. Mutz focuses on reflexivity in several French-based creoles (Haitian, Guadeloupean, Guyanese, Morisyen, Seselwa). She demonstrates that the choice of appropriate reflexive constructions is constrained by semantic and discourse factors, which can be understood as precursors of grammaticalization. Roberts focuses on the role of style and identity in the development of Hawaiian Creole. She finds evidence in data culled from 1920s Life Histories to support the claim that the linguistic elaboration attested in the development of HCE is essentially stylistic. Accord-

ix

x

Preface

ing to Roberts, modernist ideologies promoted the shift from Hawaii’s ancestral languages to Hawaiian Creole. We thank all the referees for their indirect participation in this volume, and all authors for advancing the study of creoles, contact, and language change. Geneviève Escure Armin Schwegler University of Minnesota University of California, Irvine

CHAPTER 1

The origins of Macanese reduplication Umberto Ansaldo

Universiteit van Amsterdam

Stephen Matthews

University of Hong Kong

Reduplication, a widespread feature of pidgins and creoles, is a salient feature of Macanese (the Portuguese-based creole of Macao). Like other Portuguese-based creoles of Southeast Asia, Macanese shows influence of Malay substrates. However, unlike other Portuguese-based creoles, Macanese developed in a context rich in Sinitic adstrate languages such as Cantonese and Hokkien. In this article we examine structural features of this little-known variety, and discuss Sinitic and Malay influence in the formation of Macanese reduplication patterns. We also address the issue of typological convergence in teasing out the adstrate/substrate interplay in the evolution of Macanese. Finally, we explore the origins of Macanese in the light of new data presented here.

1. Introduction* Portuguese presence in East and Southeast Asia dates back to the 16th century, and began with Portugal’s aim of gaining control of the network of naval routes and ports that supported the spice trade. From Goa, captured in 1510, the Portuguese conquered Malacca a year later and from there they traded with Malays and Javanese throughout the Indonesian archipelago. Decades later the Portuguese established a permanent base in Macao, which aided them greatly in their trade with China, Japan and the Philippines (Baxter 1996). Understanding the history of Macao and its relationship with other Portuguese colonies of East and Southeast Asia is essential in order to appreciate the development of Macanese. Macanese has been variously treated as a dialect of Portuguese or a variety of Malayo-Portuguese (see Section 3). As will become clear in this paper, our position is that the Macanese variety was once influenced by the creoles of Goa and Malacca. At the same time Macanese exhibits unique features that most likely developed in situ. To provide the proper background, in Section 2 we briefly outline the social history of Macao. In doing so we will highlight a salient aspect of Macanese life — the

2

Umberto Ansaldo and Stephen Matthews

“marriage problem” — and its consequences for the evolution of the Macanese language. Section 2 also offers a brief historical and linguistic profile of other Portuguese colonies established prior to Macao, particularly Malacca. This information is relevant for a proper understanding of the data presented in the remainder of the paper. Section 3 gives a general, structural description of Macanese in relation to its possible substrate/adstrate languages. Here we also discuss earlier accounts of this language. The bulk of our data on reduplication is discussed in Section 4, where we present our main hypothesis on typological convergence. It is self-evident that, on account of their typological similarities, Malay and Sinitic are plausible sources for several of the reduplication patterns found in Macanese. In some cases, the sources are clearly Malay. In other instances, however, they are distinctly Sinitic, as can be observed in the semantics and syntax of adverbial reduplication (Section 4.3). Cases like these are our main reason for distinguishing Macanese from other Portuguese creoles of Southeast Asia. Section 5 concludes study by evaluating our findings and their implications for reconstructing the history of Portuguese-based creoles of the region.

2. The history of Macao and the Macanese Unlike the colonies of Goa and Malacca, Macao developed without explicit authorization from either Portugal or China. Its existence was recognized after the fact by Portugal, and tolerated by the Chinese. Tradition has it that the Chinese authorities granted the Portuguese the concession in return for their role in fending off pirates in the Pearl River, though this story is not confirmed by contemporary documents. Table 1 shows key dates in the history of Portuguese settlements of Macao and other Southeast Asian colonies. Initially, contacts between Portuguese and locals were forbidden by Chinese authorities, and only a handful of Chinese traders of Hokkien origin were allowed to reside in the settlement. These were originally from Fujian. They were mostly fishermen who occupied a special role within the settlement of Macao from very early on, and formed a significant proportion of the 16th-century population. Unlike the neighboring Cantonese communities, the Fujianese traders were allowed to live inside the colony because of their essential supporting roles as providers of fish, fresh water, and shipwrights (Coates 1978: 37). These settlers were therefore in a position to influence the emerging Portuguese creole from the outset, before the relaxation of controls and migration allowed Chinese to enter on a large scale. The language of the Fujianese traders was a dialect of the Min group, related to the variety known in Singapore as Hokkien.¹ The Hokkien dialects are not mutually comprehensible with Yue dialects such as Cantonese, which were to become the dominant dialects of the Pearl River Delta area. The early presence of the Hokkien community in Macao

The origins of Macanese reduplication Table 1. Portuguese expansion in East and Southeast Asia 1510 1511 1513–1519 1542 1550 1586 16th c. 1641 1644 1652 17th c. 18th c. 19th c. 20th c.

Portuguese in Goa. Portuguese in Malacca. Portuguese traders visit the Pearl River Delta and Canton. Portuguese in Japan. Macao set up as a trading port. Macao’s status raised to that of a city by the Viceroy of Portuguese India. Mixed marriages between Malays, Hindus and the local Macanese. Chinese still have to leave town at sunset and are not allowed to buy land in the settlement. Fall of Malacca. Migration from Malacca to Macao. The Manchus invade China, push south of Peking. Manchus invade Canton. Refugees from China seek shelter in Macao. Macanese-Chinese marriages increase, but Chinese are still not allowed to live in Macao or buy property. Immigration from the Philippines and Japan (Christian converts). Chinese allowed to live in the city. Heavy Chinese immigration into Macao brings waves of Cantonese and Minnan speakers. Decreolization due to introduction of Portuguese as official language (later supplanted by English).

nevertheless must be considered a potential factor contributing to Sinitic influence in the formation of Macanese, even before Cantonese and other Yue dialects such as Zhongshan, entered the scene. All other forms of contact with Chinese were limited to a bi-monthly market supervised by the Chinese authorities. In 1563, the settlement’s population consisted of roughly 900 Portuguese and several thousand Malays, Indians, and Africans from other Portuguese colonies (Tomás 1988). The early restrictions placed on Chinese presence was gradually relaxed, and as a result two hundred years later the population had become predominantly Chinese. An outline of the demographic shift in Macao between the 17th and the 19th centuries is given in Table 2 (based on Tomás 1988). Table 2. Demographic shift in Macao between the 17th and the 19th century

Population

1622

1878

Portuguese Chinese Others

1,000 6,000 N/A

4,476 63,532 122

3

4


2.1 The marriage problem Of particular relevance to the formation of Macanese is the question of women in the Macanese community. The “marriage problem” has become a matter of some debate in the literature on Macanese culture (see Coates 1978: 44). Typically a Portuguese trader, sailor, or administrator came to Macao without his European wife. Moreover, contact with and marriage to Chinese was prohibited in the early decades of the colony by the Chinese authorities. This would suggest a limited influence of Cantonese in Macanese households. There is reason to believe that until the beginning of the 19th century, the female population of Macao came largely from other colonies in which Portuguese settlers could find wives, such as the Indian settlements and Malacca (more on this point below). Records show that several Portuguese settlers in Macao were “married in Melacca”, presumably to Malay women (Batalha 1994).² As Boxer (1965) notes, the women of the Portuguese in Macao were seldom Chinese, but rather Japanese, Malay, Indonesian and Indian. Within this culturally and racially mixed environment, Malacca was by far the most influential colony in Macao. The ties between Malacca and Macao, often remarked upon in Macanese historiography, led to cultural influences such as dressing and cooking styles of the Macanese women. While Macanese men traditionally dressed in calça moura (‘Moorish trousers’) of Indian origin, women typically wore cabaias³ of the Malay type (Amaro 1994). Malacca’s relevance in the evolution of Macao and Macanese will be considered in Sections 2.1 and 3.2 below. Bearing in mind what we said so far, it is also important to consider that official records did not take into account the presence of illegitimate offspring and local concubines (often referred to as bichas, a derogatory term literally meaning ‘worms’). However, it is likely that Chinese concubines were indeed present since contacts with Chinese women were frequent (as attested in many chronicles) and illegitimate offspring almost certainly numerous (Teixeira 1994).⁴ Chinese women would have been present as housemaids and concubines from the 17th century onwards, when heavy migrations from Southern China to Macao (and Hong Kong) took place. As mothers and nannies, these Chinese women must have played a significant role in the development of Macanese from an early stage.⁵ In summary, the native-born Macanese can be said to have originated from three main strands: Portuguese, Malacca mestiça,⁶ and — later — (Southern) Chinese (Coates 1978: 45).

2.2 Malacca and other Portuguese colonies The Portuguese presence in Malaysia (Malacca) and Indonesia (Batavia) lies at the origin of several contact languages referred to as Malayo-Portuguese creoles. These creoles evolved from a mixture of Portuguese and Malay, although not in their standard or “high” varieties, but rather derived from already simplified and possibly

The origins of Macanese reduplication

creolized varieties of each language. In particular, a simplified form of Portuguese may have been spoken early on in Malacca, probably modeled on Portuguese-based creoles developed in West Africa (Clements 1992, Hancock 1975). Such local Portuguese vernaculars came into contact with Bazaar Malay as well as with several Chinese varieties, in particular Hokkien and Baba Malay (Ansaldo and Matthews 1999). Bazaar Malay was spoken in Malacca before the Portuguese settled in the area, since the city was already a center of commerce for Chinese and Arab traders (Hancock 1975). The Portuguese took over Malacca one year after occupying Chaul, Goa and other coastal Indian towns, as the port was of strategic importance for the control of the spice trade from the Moluccas. By 1613, Malacca had a population of 7,400 Christians and 300 Portuguese. A creole variety of Portuguese known as Papia Kristang (PK, lit. ‘language of the Christians’) developed, predominantly from the intermarriage between Portuguese men and local (Malay) women. After the Dutch takeover, PK remained in use between the Dutch and the locals. PK is still spoken today by approximately 1,500 people (Holm 1989: 292). The prominent role of Malacca in the history of Macao is clearly reflected in the many features of PK that reappear in Macanese. Typical creole features include the PK pre-verbal TMA markers ja (perfective; from Port. já ‘already’, Thurgood and Thurgood 1996), bai (future; from Port. vae) and the negative future marker nadi (< Port. não ha de ‘not have to’, Holm 1989: 292). Macanese also uses ja, and nadi (Santos Ferreira 1996: 242–3). These markers are also found in Indo-Portuguese creoles from Cochin, Sri Lanka up to Korlai. Some features may have an even wider distribution. Among these is bai, also found in Melanesian pidgin. This could be a remnant of an early pidgin Portuguese that presumably played a significant role in the development of Portuguese creoles (Clements 1992).⁷ In the Indian colonies the Portuguese encouraged intermarriage between Portuguese settlers and Indian women in order to increase their numbers (Holm 1989: 285). This led to a significant number of descendentes ‘those of Portuguese descent’ in the Portuguese colonies. In Goa, for example, a colony that also appears prominently in Macanese historiography, locals of Portuguese origin numbered around 10,000 in 1540 (Marques 1976). It was within communities such as these that a variety of Indo-Portuguese arose with features that are distinct from those of Malayo-Portuguese. The influence of Indic languages was for example felt in the loss of final syllable following the accented syllable of the original Portuguese word (Dalgado 1917, in Holm 1989: 285). Examples (1)–(2) are illustrative of this change. (1) música > muz (2) água > ag In addition, Indo-Portuguese varieties such as Sri Lankan Creole Portuguese show SOV as well as SVO word-order, among other morphological and syntactic features from Tamil and Sinhala (Smith 1984).

5

6


3. Macanese Having presented the socio-historical context in which Macanese evolved, we now turn to the discussion of the linguistic sources of the Macanese language. For reasons given previously, we will be focusing on Malay and Sinitic sources, and also examine more distant but potentially relevant Indian influences.

3.1 Earlier accounts Macanese has been given little attention in pidgin and creole studies. A recent Portuguese textbook on pidgins and creoles makes only passing reference to this creole language, observing: “The Portuguese who settled there from 1557 coined the maquista language or Macao patois, thought to have been a creole, and now extinct” (Couto 1996: 238, our translation).⁸ The terms maquista (also spelt makista) and macaense refer primarily to the inhabitants of Macao with partly Portuguese ancestry, while the language is referred to as patoá or patois locally and in Macanese historiography (Batalha 1994, Pons 1999). There is little doubt that Macanese was a contact variety involving linguistic hybridization of Portuguese, Malay and Sinitic traits. This leads us to refer to Macanese as a “creole” in the sense that Macanese is a result of the mixing of different cultures and linguistic systems (Chaudenson 2001), leading to a contact language with its own innovative features, as well as significant restructuring relative to European Portuguese. We do not mean to imply that the language belongs to a synchronically definable “creole” type, as we do not consider the process or the results of restructuring of creoles to be qualitatively different from those found in other situations where language contact induced change. (Ansaldo and Matthews 2001). Macanese is now virtually extinct, at least in Macao itself. Although confirming extinction is far from straightforward (Evans 2000), it is clear that patois is no longer in everyday use. Holm (1989: 291) discusses whether Macanese can be treated as a variety of Malayo-Portuguese, alongside Papia Kristang (Malacca) and several extinct varieties once spoken in Java, Timor, Ternate and other areas formerly under Portuguese colonial influence. A similar discussion can already be found in Thompson: This dialect bears a close genetic affinity to the Portuguese dialects of Malacca and Indonesia and in its earliest form seems probably to have come to Macao “readymade” in its essentials. I take it, then, to be the creolized descendant of the Portuguese pidgin which was for two hundred years the lingua franca of the coasts of Africa and Asia from the Cape Verde Islands to Nagasaki. (1961: 238)

Ivens Ferraz (1987) and Holm (1989: 290–1) both consider the use of reduplication of nouns for the formation of plurals a distinctive feature of Malayo-Portuguese: Reduplication, a feature which in pidgins and creoles constitutes a calque of similar constructions in the substrate . . . is pervasive in the GG [Gulf of Guinea] creoles. However, neither the GG nor the UG [Upper Guinea] creoles utilize reduplication

The origins of Macanese reduplication of nouns for the formation of the plural, as is found in the Portuguese Creoles of the East . . . This construction could have spread from Malayo-Portuguese, on the Malay model, to the other Portuguese Creoles of Asia. (Ivens Ferraz 1987: 352)

Holm (1989: 291) gives these examples of pluralizing reduplication: (3) gatu-gatu ‘cats’ (4) meza-meza ‘tables’ (5) señor-señor ‘gentlemen’

(Malacca) (Malacca) (Java)

The sources for such reduplication patterns may go back as far as pidgin Portuguese (see Section 2.3, and Clements [2003]). But our structural investigation (§3.2 below) of Macanese reveals there is good reason to believe that these patterns may have other sources. We have reached this conclusion because a. Macanese shows specific affinities to other Malayo-Portuguese varieties, implying Malay influence; and b. Sinitic (Chinese) influence can be seen in Macanese not only in the lexicon, but also in the syntactic positions and semantic functions of specific reduplication patterns.

3.2 Literary sources for Macanese data Our main data source is the work of José dos Santos Ferreira, popularly known as Adé (ca. 1900–1990). Santos Ferreira (1996) is a posthumous anthology of short stories and drama scripts in “Dialecto Macaense,” originally published between 1974 and 1988. Some of the texts include Portuguese versions by the author, a tool that proves invaluable in the interpretation of creole patterns. The anthology includes a brief grammar sketch entitled “Epítome de gramática comparada” (Santos Ferreira 1996: 229–51, originally published in 1978), and a glossary. Adé was a representative of the very last generation of fluent speakers, born into a Macanese family in the early 20th century when Macanese was still widely used. As he recalls, in the first decades of this century, the Macanese dialect — by then commonly known as “patois” — was still spoken among several Macao families . . . From the ’30s onwards the dialect fell into disuse in Macao. (Santos Ferreira 1988: 10, our translation)

Adé recognized his role as one of the last speakers, and sought to contribute to preserving a dying dialect: “The Macanese dialect is, sadly, about to disappear” (Santos Ferreira 1996: 6, our translation). His work thus forms part of a conscious attempt at revitalization of the “patois”. Adé recorded several poems and other texts for Macao radio programs, some of which were included on a Compact Disc, Papiá di Macao, issued following his death in 1990. His poems also feature in songs performed by the traditional group Tuna Macaense.

7

8


While Adé was clearly a fluent speaker, it must be acknowledged that his variety may not be fully representative of earlier stages: “Macanese has been decreolizing for a century” (Baxter 1996: 327), and this process would have been under way even in Adé’s youth. This does not, however, reduce the usefulness of his texts for studying reduplication patterns. This is so because decreolization towards Standard European Portuguese has led to loss of reduplication patterns (Batalha 1994: 151). We can therefore assume that most if not all such patterns in Adé’s variety are inherited from earlier stages of the language. Adé’s metalinguistic awareness can be seen from the prefaces to his works, as well as in his grammar notes. Concerning reduplication, his comments do not go beyond reference to the salient and well-known reduplicated noun plural. Other reduplication patterns can be found in incidental entries in the glossary, but are not described in the grammar notes. We will here focus on examples from prose and drama, on the assumption that these are more representative of spoken Macanese than those found in his poetry.

3.3 Substrate and adstrate influences in Macanese Portuguese-based Asian creoles are often characterized by the fact that the tie to the local substrate was never really quite lost. This situation contrasts starkly with that of the Caribbean varieties, where African substrate languages were eventually lost (Holm 1989), except perhaps in Cuba where Yoruba and (a restructured form of) Kikongo survive to this day, albeit mostly in ritual contexts (A. Schwegler, p.c.). The Portuguese were always a minority in the Asian colonies, and Macao was no exception, as indicated above. The Sinitic influence on Macanese has, however, been downplayed (Tomás 1988), mostly because intermarriage presumably did not occur to the same degree as in other Portuguese colonies of Southeast Asia. This study will show that there is indeed evidence of Sinitic influence in the formation of the Macanese creole, as would be expected in a prolonged and intense contact situation. We will refer to this as “adstrate” rather than “substrate” influence, since the historical facts reviewed above suggest that it took effect after the initial formation of the creole. The early sociolinguistic environment in which Macanese evolves can be described as a multilingual setting comprising Hokkien, Cantonese, restructured Portuguese and other minor languages. The Portuguese variety was most likely a form of Malayo-Portuguese which, in its turn, shared some features with Indo-Portuguese (see 2.3 above). Given the historical setting outlined above, a variety of substrate or adstrate influences are possible. Based on the lexicon alone (Batalha 1977), we can establish at least two distinct substrates: A. Austronesian: Malay (especially via Malacca, but also via trade networks between Macao and the Malay archipelago), e.g.:


(6) capir ‘to clasp’ < Malay kapis (capi roupa, ‘wooden eggs’) (7) saiám ‘pain’ < Malay sayang (8) estrica ‘pressing iron’ < Papia Kristang istrika (Batalha 1994) B. Sinitic, specifically Yue and Min varieties of Chinese (Zhongshan Cantonese and Hokkien), e.g.: (9) tancá ‘boat’ < Cantonese daan1 gaa6 ‘boat people’⁹ (10) choncar ‘to bump’ < Cantonese zong6 ‘bump’ + the Portuguese verbal suffix -ar (11) faichis ‘chopsticks’ < Cantonese faai3 zi2 ‘chopsticks’

3.4 Structural features of Macanese When we look at Macanese structure there is little doubt that we are looking at a heavily hybridized contact variety involving restructuring of the original input languages (Portuguese, Malay, and Chinese dialects). Besides a typical TMA system (already noted in previous studies) that shows similarities across Malayo-Portuguese varieties (Hancock 1975, Holm 1989), we find significant morphological reductions that are typical of creolization (i.e., patterns of change typically found in contact situations, but by no means unique to them). Typical features are: a. b. c. d.

Absence of copula (12) Loss of feminine gender (13) Loss of number distinction (16) Loss of verbal morphology, with most verbs deriving from Portuguese present tense or infinitival forms (15, 16) (12) Este criada bo. (Port. Esta criada é boa.) this servant good ‘This servant is good.’

Notice the lack of agreement between noun and adjective also found between possessive pronoun and noun in (10). (13) meu mãe (compare Portuguese minha) my mother (14) Qui de gato? (qui replaces Port. onde)¹⁰ here of cat ‘Where is the cat?’ (15) Ele nã pode vem. he NEG can come ‘He isn’t coming.’

9

0


(16) Nóis nã qué vai. we NEG want go ‘We don’t want to go.’ We now turn to the various reduplication patterns found in Macanese.

4. Reduplication patterns in Macanese Because reduplication with a grammatical function is so unlike Indo-European lexifier languages, it calls for an explanation in terms of the circumstances of creole formation. Below we outline the reduplication patterns as recorded in the texts, and try to account for the most plausible substrate influences.

4.1 Nominal reduplication Adé states: “As a rule, nouns are put into the plural by doubling the word” (Santos Ferreira 1996: 231, our translation). His examples show morphemic reduplication (17) and partial reduplication (18): (17) a. home b. casa

→ hóme-hóme ‘men’ → casa-casa ‘homes’

(18) a. nhónha → nho-nhónha (or nhónha-nhónha) ‘women’ b. nhum → nhu-nhum (or nhum-nhum) ‘boys, lads’ The same rule, he observes, applies to compound nouns: (19) a. fula-papaia → fula-papaia-fula-papaia ‘papaya flowers’ b. pê-di-vaca → pê-di-vaca-pê-di-vaca ‘cows’ feet’ c. bode-vaca → bode-vaca-bode-vaca ‘clumsy children’ Such nominal reduplication in Asian creoles is generally attributed to Malay influence (Holm 1989). This conclusion is probably correct, although the possibility of influence by other Austronesian languages cannot be ruled out. Interestingly, Clements (2003) shows that some forms of nominal reduplication exist in Gulf of Guinea varieties (São Tomense, Fa d’Ambu) as well as in Indo-Portuguese varieties (Korlai, Daman). This raises the possibility that nominal reduplication was already present in the pidgin that the Portuguese brought to India, Malacca, and Macao. The partial reduplication of nouns, as in (18) above, where only the first syllable is reduplicated, can be more specifically attributed to Austronesian influences. Such partial reduplication of nouns does not appear to be attested in Indo-Portuguese, although there is partial reduplication of certain adjectives (Clements 2003). Indonesian has traces of such a reduplications, but they are no longer productive (Sneddon 1996: 21): (20) tamu ‘guest’ → tetamu ‘guest’ tua ‘old’ → tetua ‘elders’


Partial reduplication in modern Malay does not always signal “plural”; at times it merely serves to form new lexical items with specific meanings related to the root (Prentice 1990: 199): (21) laki ‘husband’ → lelaki ‘male, man’ The Macanese case does not seem to be very productive either, being restricted to high frequency items such as “men” and “women”. A second feature suggestive of Malay is the interpretation assigned to reduplicated plurals, as in the example below: (22) pintad laia-laia côr painted kind-kind colour ‘painted in various colours’ (Santos Ferreira 1996: 291) Here, the gloss ‘various colours’, consistently provided by Adé himself, strongly suggests a Malay source. In Malay, reduplication is rarely if ever obligatory for plural reference, but when it is used the plurality typically has the sense of ‘various’: (23) babi-babi (Malay) pig-pig ‘(various) pigs’ The existence of partial reduplication of nouns and the semantics of reduplicated plurals support the assumption that Malay influence has played a role in the evolution of Macanese reduplication patterns. Although other Asian creoles such as Korlai may show nominal reduplication (Clements 2003), they lack these two distinctive properties.

4.2 Adjectival reduplication Reduplication of adjectives is another productive process in Macanese. Since adjective reduplication with intensifying effect is very widespread, its presence in Macanese as in (24) does not reveal its origins. (24) Quarto iscuro-iscuro, sã qui-cuza tamêm nádi pôde olá bêm-fêto sã room dark-dark is someone also will-not able see well is nunca? not ‘In a dark room, no one will be able to see clearly, right?’ (Santos Ferreira 1996: 67) Reduplicated adjectives are also found with plural noun phrases: (25) Pa tudo nôsso amigo china-china . . . for all our friend Chinese-Chinese ‘For all our Chinese friends . . .’ (Santos Ferreira 1996: 213)



2


(26) Maior parte são gente grande-grande. greater part are people big-big ‘It’s mostly important people.’ (from a letter dated 1869; cited in Tomás 1988: 39) This pattern is evidently not obligatory, since an alternative is to reduplicate the noun, as in the following examples from Adé’s grammar sketch: (27) Home-home rico ‘Rich men’ (28) Nho-nhónha chistosa ‘Beautiful women’ (Santos Ferreira 1996: 234) A phenomenon similar to that seen in (25)–(26) occurs in Malay: “Reduplication of an adjective usually occurs when the noun it describes is plural; reduplication indicates that the characteristic indicated by the adjective applies to all the objects” (Sneddon 1996: 19). Sneddon gives the following examples: (29) Gambarmu bagus-bagus. drawing-2SG fine-fine ‘Your drawings are beautiful.’ (30) sekitar 230 pulau yang kecil-kecil around 230 island which small-small ‘about 230 small islands’ This provides further support for the hypothesis of a Malay source for pluralizing reduplication. However, such a reduplication pattern is also found in Norteiro, an Indo-Portuguese variety of Diu and Bombay (Holm 1989: 286): In Nor[teiro] an adjective may be reduplicated to convey variety [sic] of plural meaning in the noun, in the sense of several entities possessing the quality of the adjective. . . This type of reduplication likewise does not occur in Africa, nor apparently in other Asian Creoles. (Ivens Ferraz 1987: 352)

This description of the phenomenon is accurate but understates its prevalence. Clements (2003: 196) shows that adjective pluralization also exists in Daman and Korlai Portuguese, noting that precedents exist in the local adstrate languages, Marathi and Gujarati, as well as in Sãotomense.

4.3 Adverbial reduplication Another productive pattern of particular interest involves reduplication of an adverb or adjective to serve as a preverbal adverbial: (31) Ele azinha-azinha subí escada he fast-fast go-up stairs ‘He quickly went upstairs.’ (Santos Ferreira 1996: 245)


(32) Cedo-cedo sai di casa early-early go-out from home ‘(To) leave home early.’ (Santos Ferreira 1996: 313) (33) Péssa-péssa pôde cai haste-haste could happen/fall ‘Could happen/fall in a hurry.’ (Santos Ferreira 1996: 245) (34) Iou vagar-vagar já vai riva cozê mizinha. 1SG slow-slow PFV go top boil medicine ‘I slowly went up to boil some medicine.’ (Santos Ferreira 1996: 48) These adverbials are typically translated by a postverbal adverb in the Portuguese version of the same story: (35) Nhu-nhum mánso-mánso tocá “ti-ti-lim, ti-ti-lim”. man-man sweet-sweet play [onomat.] ‘The men sweetly play “ti-ti-lim, ti-ti-lim”.’ (Santos Ferreira 1996: 110) (36) Os homens tocam suavamente “ti-ti-lim, ti-ti-lim.” ‘The men play sweetly.’ [onomat.] (Portuguese version of (35)) While reduplicated adjectives do serve as adverbials in Malay, they are typically postverbal: (37) Anak itu berteriak keras-keras. child that scream loud-loud ‘The child screamed loudly.’ (Sneddon 1996: 19) Given that adverbials (reduplicated or not) are postverbal in Portuguese and Malay, neither internal development nor Malay influence predicts their preverbal position in Macanese. The preverbal adverbial patterns in (31)–(35) above provide the clearest evidence for Sinitic influence in Macanese reduplication, since preverbal reduplicated adverbs are a characteristic feature of Sinitic grammar. An adjective (or stative verb) can be reduplicated before the verb to serve as an adverbial in Cantonese, for example (Matthews and Yip 1994): (38) maan6-maan2 haang4 slow-slow walk ‘go slowly’ Here the reduplicated adverb maan-maan ‘slowly’ parallels vagar-vagar as in (34). Patterns of this kind can be found in Cantonese, and even more productively in Min dialects. In the Min dialect of Chaoyang, for example, the counterparts of (31) and (32) are well-formed:¹¹ (39) me-me khu fast-fast go ‘go quickly’

3

4


(40) tsa-tsa khu early-early go ‘go early’ As discussed in Section 2 above, a Fukienese community was present from the early years within the Macanese settlement. If influence of Min dialects is involved here, adverbial reduplication could be a substrate feature that acquires prominence because of its early presence in the evolution of the Macanese creole. Finally, among the attested reduplication patterns, there are some onomatopoeic forms: (41) Ca-ca ri ha-ha laugh ‘Laugh ha-ha’ (42) Pastro-pastro na riva di árvre pi-pi-pi ta cantá. bird-bird on top of tree pi-pi-pi PROG sing ‘Birds in the tops of the trees were singing pi-pi-pi.’ (Santos Ferreira 1996: 96) Such onomatopoeic forms are widespread. They are found for example in Cape Verde Creole (Baptista 2003), and therefore their presence does not constitute persuasive evidence for adstrate influence. However, note again that the reduplicated phrase appears in preverbal position, like the reduplicated adverbs in (31)–(35) as discussed above. Compare (41) with its Cantonese counterpart (43): (43) kaa1-kaa1 siu3 ha-ha laugh ‘laugh ha-ha’ As we have argued with respect to adverbial reduplication in general, the preverbal placement of the onomatopoeic phrase suggests the Chinese pattern.

4.4 Verbal reduplication Reduplicated verbs are less commonly found than the nominal and adverbial patterns already discussed. Examples include: (44) Venáncio . . . na cáma bulí-bulí su corpo. venancio in bed move-move his body ‘Venancio . . . moved his body about in bed.’ (Santos Ferreira 1996: 67) (45) Testa suado-mulado, coraçám batê-batê . . . forehead sweat-mottled heart beat-beat ‘His brow mottled with sweat, his heart beating away . . .’ (Santos Ferreira 1996: 68)


Both Malay and Sinitic verb reduplication patterns could be involved here. The Malay pattern is associated with a variety of aspectual and related meanings (Sneddon 1996: 20): 1. CASUAL ACTION: duduk ‘sit’: duduk-duduk ‘sit about’ 2. REPEATED ACTION: urut ‘massage’: mengurut-urut ‘stroke repeatedly’ 3. INTENSIFICATION: meminta ‘request’: meminta-minta ‘beg’ Of these three meanings, that of repeated action matches examples (44)–(45). Since these semantic functions — especially (2) and (3) — are highly iconic, it is no surprise that similar cases of verb reduplication arise elsewhere, perhaps independently. In Korlai and Daman, for example, parallel patterns are associated with the functions of intensification, sequentiality or simultaneity. Local adstrate languages, such as Marathi and Gujarati also display the sequential function (Clements 2003: 196– 7). In Macanese, too, the repeated verbs are typically used in verb phrase sequences describing simultaneous actions, as in (46): (46) Venáncio ficá cara di tolo, ôlo batê-batê, tremê bêço, nunca astrevê venancio stay face of fool eye blink-blink tremble lip not dare sai bafo. let-out breath. ‘Venancio remained looking foolish, his eyes blinking and lips trembling, not daring to breathe.’ (Santos Ferreira 1996: 67) In Cantonese, reduplication of an action verb usually co-occurs with the suffix haa5 meaning ‘to do a little’, and expresses simultaneous actions (Matthews and Yip 1994: 208), as in (47): (47) Go3 bi4bi1 sik6-sik6-haa5 laai5 fan3-zo2. CL baby eat-eat-ASP milk sleep-ASP ‘The baby fell asleep while drinking milk.’ Given such close parallels in the adstrate languages, in this case including both Malay and Cantonese, it may not be possible to differentiate between pidgin-Portuguese-derived reduplication patterns and those that have arisen locally.

4.5 Numeral reduplication Reduplicated numerals have a distributive sense, as in unga-unga ‘one by one’: (48) Unga-unga ta falá. one-one PROG speak ‘(They) are speaking one by one.’ (Santos Ferreira 1996: 72) (49) Unga-unga já virá vai casa. one-one PFV return go home ‘They returned home one by one.’ (Santos Ferreira 1996: 86)

5

6


The form dos-dos ‘two by two’ appears in the same story: (50) Êle largá pê pulá dôs-dôs degrau. s/he left to jump two-two step ‘He left jumping two steps at a time.’ Numeral reduplication has a similar distributive sense in Malay (Sneddon 1996: 21): (51) Bicaralah satu-satu. speak-PRT one-one ‘Speak one at a time.’ (52) Mereka masuk dua-dua. they enter two-two ‘They come in two at a time.’ Note, however that, as we observed in the case of reduplicated adverbs (§4.3 above), the preverbal placement of the reduplicated phrase in Macanese follows the Chinese pattern, as illustrated in (53) from Cantonese: (53) Di1 hok6saang1 loeng5 go3 loeng5 go3 jap6 lai4. PL student two CL two CL enter come ‘The students come in two by two.’ One possibility is that the Malay pattern as in (51)–(52) is hybridized with Chinese syntax as in (53) to produce the Macanese structure (48)–(50). Numeral phrases are thereby assimilated to the general Chinese pattern whereby reduplicated adverbials precede the verb (see 4.3 above).

5. Conclusion and implications Reduplication is a salient feature of Macanese that illustrates the diverse sources of what is visibly a restructured Portuguese variety. In this study, we have described the following reduplication patterns: nominal, adjectival, adverbial, verbal, numeral and onomatopoeic. Though the sources for these patterns are varied, we have shown that what is most distinctive about Macanese is the Sinitic adstrate. In the case of pluralizing nominal reduplication, the Malay adstrate provides the most plausible source on both formal and functional grounds. In the case of adverbial reduplication, Sinitic languages offer the closest parallel in structure on syntactic grounds: in particular, the preverbal position of reduplicated adverbials, numerals and onomatopoeic expressions contrasts with both Malay and European Portuguese, but is attributable to Sinitic syntax. For the remaining patterns, a number of explanations are possible, including:


1. origins in pidgin Portuguese brought from the Gulf of Guinea colonies; 2. convergence of Malay and Sinitic structures; or 3. independent development following universal patterns. Considering the origins of Macanese in time and space (§1), their close ties with Malacca, and the flourishing trade that extended over the entire Malay archipelago, we conclude that (a) convergence between Malay and Sinitic structures has taken place, and (b) pidgin Portuguese influence and/or independent development cannot be ruled out.

Notes * We thank Madalena Cruz-Ferreira for her detailed comments on the Portuguese and Macanese data, Michelle Li for Cantonese examples, and Casey Sim for her historical research. For constructive feedback on earlier versions we are grateful to two anonymous reviewers, and to participants at the SPCL 2002 meeting in San Francisco, especially Clancy Clements, Genevieve Escure and Nicolas Quint. We also thank Marlyse Baptista and Clancy Clements for sharing their forthcoming papers, and the editors of this volume for suggesting further improvements. 1. Fujian is the name of the coastal province adjacent to Guangdong to the east. Coates’ term Fukienese is based on the Cantonese reading of the same word, while Hokkien represents the pronunciation in the Min dialect of Fujian itself. 2. Regarding the Indian settlements, Boxer notes that only in exceptional cases did the Portuguese marry high-caste women: “Portuguese males preferred to consort with their slavegirls rather than enter the bonds of holy matrimony” (Boxer 1963: 76–7). 3. These are fitted, long-sleeved blouses, extending to the hip, heavily embroidered on the edges and sleeves, with floral, bird or butterfly motifs, fastened at the front with a 3-piece krosang (brooch), and worn with a sarong. 4. Macao was from early on known as “Paris of the Orient”. It was a promiscuous society where men could openly keep several mistresses in or outside the household, a practice which was not uncommon for the higher strata of Chinese society in general. 5. Women were also brought from Japan. Nagasaki, growing from a fishing village to a large town on the wealth of the Portuguese trade, had a resident Portuguese community. As conversions increased, Portuguese men married Japanese Christian women. Some of these new families had connections with Macao, where Japanese Christians engaged in the trade found it convenient to install their relatives as agents. A small Japanese community was thus an early feature of Macao, where Luso-Japanese marriages were common. Little obvious influence from Japanese can be seen in Macanese, although Batalha (1977) notes a few loan words such as auabi from Japanese awabi ‘abalone’. 6. Indo-Portuguese female of mixed blood (Boxer 1963: 77). 7. We thank two anonymous reviewers for pointing out these affinities. 8. “Os portugueses que se estabeleceram aí desde 1557 deram lugar à língua maquista ou patois de Macao que teria sido um crioulo, já extinto.”

7

8

Umberto Ansaldo and Stephen Matthews 9. Cantonese forms are given in the Jyutping romanization system developed by the Linguistic Society of Hong Kong, in which the six tones are numbered from 1 (high level) to 6 (low level). Note that the consonantal values do not match the Macanese orthographic conventions; for instance, the voiceless unaspirated consonants represented as ‘t’ and ‘c’ in Macanese appear as ‘d’ and ‘g’ in this romanization. For IPA correspondences, see Matthews and Yip (1994: 400–1). 10. Note that in colloquial Portuguese when inquiring whereabouts, “que do/da + NOUN?” can still be used, as a contraction of “que é do + NOUN?” lit. ‘what is of the + NOUN?’ (M. CruzFerreira p.c.). 11. The southern Min dialects form a dialect continuum from the Chaoyang area in eastern Guangdong to Xiamen (Amoy) and Quanzhou in Fujian province. Precisely which Min dialects were spoken by the Fujianese in Macao is unknown, except that their association with maritime activities implies a coastal origin.

References Amaro, A. M. 1994. “Sons and daughters of the soil.” Review of Culture 20: 12–67. Ansaldo, U. and S. Matthews. 1999. “The Minnan substrate and creolization in Baba Malay.” Journal of Chinese Linguistics 27.1: 38–68. Ansaldo, U. and S. Matthews. 2001. “Typical creoles and simple languages: The case of Sinitic.” Linguistic Typology 5.3: 311–26. Baptista, M. 2003. “Reduplication in Cape Verdean creole.” In Kouwenberg 2003: 177–84. Batalha, G. N. 1977. Glossário do Dialecto Macaense: Notas Linguísticas, Etnográficas, Folclóricas. Coimbra: Instituto de Estudos Românicos. Batalha, G. N. 1994. “Language of Macao: Past and present.” Review of Culture 20: 131–56. Baxter, A. N. 1996. “Portuguese and creole Portuguese in the Pacific and Western Pacific Rim.” In Atlas of Languages of Intercultural Communication in the Pacific Asia and the Americas, S. A. Wurm, P. Mühlhäusler and D. T. Tryon (eds.), Vol. II.1: 299–338. Berlin: Mouton de Gruyter. Boxer, C. R. 1963. Race Relations in the Portuguese Colonial Empire, 1415–1825. Oxford: Clarendon Press. Boxer, C. R. 1965. Portuguese Society in the Tropics: the Municipal Councils of Goa, Macao, Bahia, and Luanda, 1501–1800. Madison: University of Wisconsin Press. Chaudenson, R. 2001. Creolization of Language and Culture. London: Routledge. Clements, J. C. 1992. “On the origins of Pidgin Portuguese.” Journal of Pidgin and Creole Languages 7.1: 75–92. Clements, J. C. 2003. “An amorphous morphological account of reduplication processes in three Indo-Portuguese creoles.” In Kouwenberg 2003: 193–202. Coates, A. 1978. A Macao Narrative. Oxford: Oxford University Press. Couto, H. H. do 1996. Introdução ao Estudo das Línguas Crioulas e Pidgins. Brasília: Fundação Universidade de Brasília. Dalgado, S. R. 1917. “Dialecto Indo-Português de Negapatão.” Revista Lusitana 20: 40–53. Evans, N. 2000. “The last speaker is dead: Long live the last speaker!” In Linguistic Fieldwork, P. Newman and M. Ratliff (eds.): 250–81. Cambridge: Cambridge University Press. Hancock, I. F. 1975. “Melacca creole Portuguese. Asian, African or European?” Anthropological Linguistics 17.1: 211–36.

The origins of Macanese reduplication Holm, J. 1989. Pidgins and Creoles, Vol. 2: A Reference Survey. Cambridge: Cambridge University Press. Ivens Ferraz, L. 1987. “Portuguese creoles of West Africa and Asia.” In Pidgin and Creole Languages: Essays in Memory of John E. Reinecke, G. Gilbert (ed.): 337–60. Honolulu: Hawaii University Press. Kouwenberg S. 2003. (ed.). Twice as Meaningful: Morphological Reduplication in Pidgins, Creoles, and other contact languages. London: Battlebridge. Marques, A. H. de Oliveira. 1976. History of Portugal. New York: Columbia University Press. Matthews, S. and V. Yip. 1994. Cantonese: A Comprehensive Grammar. London: Routledge. Pons, P. 1999. Macao, un Éclat d’Éternité. Paris: Editions Gallimard. Prentice, D. J. 1990. “Malay (Indonesian and Malaysian).” In The Major Languages of East and Southeast Asia, B. Comrie (ed.), 185–207. London: Routledge. Santos Ferreira, J. dos. 1988. Macao: Jardim abençoado. Instituto Cultural de Macao. Santos Ferreira, J. dos. 1994. Poéma di Macao. Macao: Fundação Macao. Santos Ferreira, J. dos. 1996. Papiaçám di Macao. Macao: Fundação Macao. Smith, I. R. 1984. “The development of morphosyntax in Sri Lanka Portuguese.” In York Papers in Linguistics 11: 291–301. Sneddon, J. 1996. Indonesian Reference Grammar. St. Leonards, NSW: Allen & Unwin. Teixeira, M. 1994. “The origin of the Macanese.” Review of Culture 20: 157–61. Thompson, R.W. 1961.“On the Portuguese dialect of Hong Kong.” Papers presented at the Symposium on Historical, Archaeological and Linguistic Studies on Southern China, Southeast Asia and the Hong Kong Region. Hong Kong: Hong Kong University Press. Thurgood, E. and G. Thurgood. 1996. “Aspect, tense or aktionsart? The particle ja in Kristang (Melacca creole Portuguese).” Journal of Pidgin and Creole Languages 11:45–70. Tomás, I. 1988. “Makista Creole.” Review of Culture 5: 33–46.

9

CHAPTER 2

Court records as a source of authentic early Sranan Margot van den Berg


Jacques Arends


In this study we show that 17th- and 18th-century court records, in which (parts of) Blacks’ testimonies are occasionally reproduced in Sranan, contain valuable data on early Sranan. The vast majority of these records are contained in the archives of the Hof van Politie en Criminele Justitie, stored in the Nationaal Archief in The Hague. Some additional data can be found in the archives of the Sociëteit van Suriname. The earliest document mentioning a lawsuit concerning an African slave dates from 1684. The earliest Sranan words and sentences that were found in these records date from 1702 and 1707, respectively. Apart from a number of interesting metalinguistic observations, these documents contain some 500 isolated Sranan words (tokens) and some 50 Sranan sentences. We will restrict ourselves to a presentation and discussion of the latter, concentrating on aspects of the verbal system, in particular the expression of (a) tense, mood and aspect and (b) the copula system. Our findings are put in perspective by comparing them with what is known from other early sources, such as Herlein (1718), van Dyk (ca. 1765), Nepveu (1770), and Schumann (1783).

1. Introduction While the Suriname creoles are relatively well documented in their early stages of development, especially when compared to other creole languages, many of the early texts were written by European authors, for whom Suriname creoles were not native languages. In fact, the only 18th-century Suriname creole text written by native speakers are the so-called Saramaka Maroon Letters, authored by Christian Grego and Johannes Alabi. Written in a variety of Saramaccan that exhibits many traces of Sranan influence (see Arends and Perl 1995), these are the only extant 18th-century texts by native speakers. Creolists would naturally welcome augmenting this small corpus of Suriname creole texts with some truly authentic early

22

Margot van den Berg and Jacques Arends

Sranan data. Apart from the 200-word Herlein fragment from 1718 (reproduced in Arends and Perl 1995), virtually no Sranan records are available for the pre-1765 period. As was demonstrated by Roberts (1995) for Pidgin Hawaiian and Hawaiian Pidgin English, court records containing (more or less) verbatim accounts of what was said during interrogations may be particularly valuable as sources of early authentic data on creoles and pidgins. Schneider’s (2002) taxonomy of written documents as sources for investigating variation and change confirms this assumption, as he too categorizes this type of text as a close reflection of actual speech. A first indication that early data on the Suriname creoles could be found in court records was given to us by the Dutch historian Ruud Beeldsnijder, who came across a couple of Sranan sentences in the archives of the Court of Police while researching early18th-century slave culture. Although the language normally used in these court records is Dutch, Sranan is occasionally used as well, especially when the testimony is from a person of African descent. We recently examined these court records systematically to see if more early Sranan data could be uncovered; the full results of our research are given in van den Berg (2000), upon which the present paper is largely based. Court records are particularly valuable to supplement our corpus of Suriname creole texts both in terms of authenticity and period of writing. Court records are supposed to be verbatim accounts of what was said during an interrogation. Even if the ideal of 100% verbatim reporting may not always have been reached, and even though the transcribers may not have been native speakers of Sranan, these court records nonetheless are likely to reflect (close to) natural native or near native speech. A further indication of the reliability of these records can be found in some of their linguistic characteristics. For example, in these records the word kaba rather than arede is used consistently to express the notion of “already”, as in (1): (1) Mi doe langa hem caba. I do with him already ‘I’m finished with him already.’ (CR 1.05.10.02, 798, f — vo; 745)¹ In his well-known Sranan dictionary, Schumann (1783: 50) provides the following remark on the use of arede: da Bakkratongo; Ningre takki “kaba” (‘that is White Sranan; Blacks say “kaba”’). Another indication of the authenticity of these records may be found in the repeated use of a swear word such as you mama pima (‘your mama’s cunt’). Finally, on a phonological level, the occasional use of epenthetic vowels in words such as bakara ‘white person’ and masara ‘master’ (alongside bakra and masra) may also be indicative of the authenticity of the language. This type of epenthesis appears to have been typical of Blacks, especially plantation slaves (Focke 1855: xii; see also Lappe and Plag 2003, Smith 2003). Given these features and pending evidence to the contrary, we will assume that the early court records may be seen to reflect authentic early Sranan.

Court records as a source of authentic early Sranan

The great majority of the court records can be found in the archives of the Hof van Politie en Criminele Jusititie (Court of Police and Criminal Justice), while the remainder is in the archives of the Sociëteit van Suriname (Society of Suriname); both of these are stored in the Nationaal Archief (State Archives) in The Hague. The records, covering the 100-year period between 1667 and 1767, consist mainly of depositions, statements and reports of examinations concerning felonies such as murder, conspiracy, poisoning, theft, misleading patrols on the look out for runaway slaves and maroons etc. Perpetrators of light offenses such as swearing, underachievement or tardiness would not be taken to court; they would be punished on the plantation by the manager or overseer. Slave owners were not allowed to punish a slave suspected of the felonies described above. The slave would be sent to Paramaribo and imprisoned for two weeks, after which the Raad Fiskaal (prosecutor for the state) and two members of the Hof van Politie (Court of Police) would interrogate him. Subsequently, the Raad Fiskaal would file a complaint in the name of the Staten-Generaal and the board of directors of the Sociëteit van Suriname, after which the Hof van Politie would reach a verdict and sentence the slave.² The earliest document concerning an African slave dates from 1683: Attestatie off verklaringe van Frederik Kreijt ende Pieter Arlebout, ten versoecke van Willem de Bruijn, gepasseert voor den curator in de Commewijne den 23 December 1683 inhoudende, dat sekeren neger Cormantijn Claas den voorm. Kreijt heeft getragt te doorschieten. [Attestation or deposition of Frederik Kreijt and Pieter Arlebout, at the request of Willem de Bruijn, executed before the curator at the Commewijne [district] on 23 December 1683, stating that the Black Cormantijn Claas tried to shoot the aforementioned Kreijt]. No early Sranan data were encountered in these records, since the minutes of the interrogation of Claas were missing from the file.³ Apart from a number of interesting metalinguistic observations, the court records contain a small amount of purely linguistic data: some 500 isolated Sranan words (tokens) and 54 short Sranan passages (often just one or two sentences). Due to the fact that the records often contain more than one account of the same event, some of these passages occur more than once, albeit not always in exactly the same wording. Although sentence boundaries are difficult to determine in this type of material, we have made a provisional analysis, yielding a total of some 70 sentences. However, since some of these sentences occur more than once, the number of different “sentence types” is somewhat lower: around 54. As shown in Table 1, apart from ten passages dating from 1707, most passages are from records dating from the 1745–1767 period. The average sentence length is 5 words, which is comparable to that in the Herlein fragment, where it is 4,5. The total size of our text corpus (excluding isolated words) is roughly 300 words. In the remainder of this paper three topics will be discussed in some more detail. First, we will take an in-depth look at the earliest sentences found in these records. They constitute the very oldest textual Sranan material known until now, predating Herlein’s (1718) dialogues by one decade. One feature that is especially prominent

23

24

Margot van den Berg and Jacques Arends Table 1. Chronological distribution of Sranan passages in court records (1667–1767) Year

Code a

Number b

Folio number(s)

Number of Sranan passages

1707 1745 1747 1755 1757 1759 1760 1761

1763 1766

1.05.04.01 1.05.10.02 1.05.10.02 1.05.04.06 1.05.10.02 1.05.10.02 1.05.04.06 1.05.10.02 1.05.04.06 1.05.04.06 1.05.10.02 1.05.04.06 1.05.10.02 1.05.04.06

1767

1.05.10.02

234 798 929 296 942 947 309 805 312 313 806 315 808 330 331 813

257/268/271/272 – – 674 260/261/269 – 47/– – 81 307/318 – 336/347 – 116 429 –

10 10 1 1 5 6 2 1 1 4 3 5 2 1 1 1

1762

Total a b

54

‘Code’ refers to the ‘toegangsnummer’ (access number) in the Nationaal Archief. ‘Number’ refers to the ‘inventarisnummer’ (inventory number) in the Nationaal Archief.

will be investigated in particular, namely the use of the form man ‘man, be able to, have the nerve to’. In Sections 3 and 4 we will look at aspects of the TMA system (in particular the expression of counterfactuality) and the copula system in the entire 1707–1767 corpus, relating our findings to what has been found in previous studies on the basis of other textual sources. Since the sentences in the corpus are often very short, their usefulness for diachronic research on a wide array of syntactic issues is limited. At the same time, however, the corpus forms an important additional source of information on the lexicon and phonology of early Sranan.

2. Some features of the earliest (1707) sentences The main objectives of this study are, first, to introduce the court records as a new source of early creole data, and second, to illustrate what they may contribute to our knowledge of the early stages of Sranan. Since this is a preliminary investigation, only relatively few data will be examined here, and special attention will be given to features that have been the object of previous study. The first three of the 1707 sentences are part of a short dialogue between two slaves, Mingo and Waly, on the Palmeneribo plantation located on the Suriname river. At that time, Palmeneribo was the largest plantation of the colony with


148 slaves working the fields. Before it came into the hands of Jonas Witsen (1699), the plantation belonged to governor van Scharpenhuizen, who had granted the slaves many privileges that Witsen ended up taking away. When Mingo wanted to visit his wife on a nearby plantation — something that had been allowed under van Scharpenhuizen’s rule — Witsen prevented him by destroying his canoe. Angered and frustrated, Mingo left the plantation. He returned several days later, his mind set on compensation for the canoe. He consulted with his fellow slave Waly, who challenged him to go to Witsen. Mingo accepted the challenge and left for the manager’s house. Many slaves followed him, and the gathering turned into a revolt.⁴ In the records the dialogue has been transcribed in three slightly different versions ((2)–(4) below); the first is Waly’s version, the second and third are based on depositions made by Blacks other than Mingo or Waly. In lieu of glosses for each individual sentence, a literal and a free translation are given for the entire passage. (2) a. W.: mingo Joù no man. b. M.: mie man.(CR 1.05.04.01, 234, f 257 ro; 1707) (3) a. W.: Mingo joù no man. b. M.: mij man. c. W.: joù gho tham. (CR 1.05.04.01, 234, f 268 vo; 1707) (4) a. W.: joù no man. b. M.: Mi man. c. W.: joù go dan. (CR 1.05.04.01, 234, f 272 vo; 1707) Literal translation W.: Mingo, you not man M.: I man W.: You go then

Free translation ‘You don’t have the nerve.’ ‘I do.’ ‘Then, go!’

Mingo was known as a troublemaker. He often escaped from the plantation (a phenomenon known as petit marronnage), and was frequently involved in fights, not only with other Blacks, but with the plantation manager as well. This is illustrated by the following passage from a witness report: (5) . . . doch de directeur nam sijn stock sloeg en goijde die op sijn huid en den genoemde neger gebruijckte telkens dese woorden, nù wanti dat, en raapte eijndelijk de stock, die naar hem toegegoijd was, op (. . .) gingh de directeur uijt de galderije hem weder met een stock begroete; den neger gebruickte weer de vorige woorde nù wanti dat, jou no meester voor mi en gingh soo raasende en scheldende naa de negerhuysen (CR 1.05.04.01, 234, f 268 vo; 1707) . . . but the manager took his stick and hit him with it. The aforementioned slave repeatedly uttered the following words: nù wanti dat and finally picked up the stick that had been thrown at him (. . .) the manager left the

25

26


porch, again greeting him with a stick. Again the slave used the words nù wanti dat, jou no meester voor mi and yelling and swearing he went to the slave huts. Apart from the sentence jou no meester voor mi, discussed in Section 4 below, this fragment contains one other Sranan sentence: (6) nù wanti dat. now/I not want that ‘I want it now’/‘I don’t want that.’ (CR 1.05.04.01, 234, f 268 vo; 1707) This sentence is worth analyzing for some of the issues it raises. First, the subject, presumably the first person singular (normally expressed by the pronoun mi), is not overtly present. Although pro-drop is not a feature of post-1750 Sranan, absence of the pronominal subject in this early document may be indicative of the early stage of the language. If Sranan was not yet fully crystallized around 1707, the occurrence of pro-drop, a feature known to occur in a number of pidgins (e.g., Lingua Franca), would not be unusual. Another remarkable feature is the fronted position of the adverb nù ‘now’, presumably with emphatic meaning; in later stages this would have been expressed by using a cleft construction na nu . . . ‘it is now that. . .’. Perhaps clefting of adverbs, which has been attested from the 1760s (van Dyk ca. 1765) onwards, was not yet a feature of Sranan at this early stage of development. An alternative explanation for both the absent subject and the fronted adverb was suggested by John McWhorter (p.c.): nu might represent the first person singular pronoun mi ‘I’ followed by an emphatically pronounced negation no ‘not’. The remainder of this section will concentrate on the use of man in the sentences (2)–(4). Man also occurs in other Sranan sources, including in Stedman’s (1790: 103) Narrative (example (7)) and Focke’s (1855: 76) Sranan dictionary (examples (8)–(9)): (7) You man? da boy fasi. you man it.is boy manner ‘Are you a Man? You behave like a boy.’ (Stedman 1790: 103; original translation) (8) Mi no man moro. I not man more ‘I’m wasted.’ (Focke 1855: 76; orig. transl.: ‘ik kan niet meer’) (9) Ai ba’! Joe man! well brother you man ‘Hey, brother! Don’t be cheeky!’ (Focke 1855: 76; orig. transl.: ‘Nu, maat! Gij zijt bij de hand’) In the sentences above we have tacitly interpreted man (< Engl. man) as if it were a noun in a nominal predicate with “zero-copula”. Alternatively, it could be an auxiliary, meaning ‘be able to’. Either of these interpretations is possible in contempor-


ary Sranan (Winford 2000). However, the auxiliary meaning of man ‘be able to’ is absent from almost all 18th-century sources (Herlein 1718, Nepveu 1770, van Dyk ca. 1765, and Weygandt 1798), the only exception being Schumann’s (1783) Sranan dictionary (all our examples from Schumann 1783 are based on the transcription made by Adrienne Bruyn): (10) Mi no mann va hoppo datti. I not able to lift that ‘I’m not able to lift that up.’ (Schumann 1783: 185; orig. transl.: ‘ich bin nicht stark genug um das aufzuheben’) The original translation of the following example from Focke’s dictionary nicely illustrates the fact that 19th-century man could be interpreted both as a noun and an auxiliary. (11) Mi no MAN foe doe datti. I not MAN/ABLE to do that ‘I am not man enough to do that; I can’t do that.’ (Focke 1855: 76; orig. transl.: ‘Ik ben niet mans genoeg om dat te doen; ik kan dat niet doen’) Note that in (10) and as well as in (11) the complementizer va/foe appears to be obligatory, while in contemporary Sranan the complementizer fu is no longer so. (12) A diri k’falek, yu no man bai en. Mi no man. it expensive terrible you not can buy it I not can ‘It’s terribly expensive. You can’t buy it. I can’t.’ (Winford 2000: 77) These examples suggest that the 20th-century auxiliary man has its origin in an 18thcentury noun in a nominal predicate with zero-copula via a 19th-century intermediate stage. The diachronic evidence, therefore, casts some doubt on Winford’s suggestion that auxiliary man may be derived from “the modal mun found in some traditional regional dialects of English” (2000: 77). Another reason for questioning Winford’s suggestion is the difference in meaning between man ‘be able to/can’, on the one hand, and mun ‘have got to/must/will have to’, on the other. Dialect-geographic information (Orton and Barry 1971) shows that mun was used in the West Midland counties of England, an area from which few emigrants have been documented to come to the Caribbean (Norval Smith, p.c.). It is thus not likely that the English modal mun and the Sranan auxiliary man are etymologically related. The development of the auxiliary man may have been facilitated by the West African substrate languages of Sranan. One way of daring a person in Ewe, one of the major substrate languages of Sranan (Arends 1995b), is by saying ‘if you are really a man, then . . .’; the expression ‘you are (not) a man’ is used with the meaning ‘you (do not) have courage’: (13) né enyé ŋútsu if you.be man ‘If you dare.’ (J. Essegbey, p.c.)

(Ewe)

27

28


This pragmatic function of daring someone is exactly the intended meaning of Waly’s you no man in (2), (3) and (4) above. The possibility of substrate influence is even made more likely by the fact that, far from being restricted to Ewe, the idiom exemplified in (13) occurs in many Ghanaian languages. A case in point is Twi, another major substrate language of Sranan (Arends 1995b). An example from a Twi song is given in (14). (14) ɔyɛ barima aa ɔmmra. (Twi) he.be man TNS he.come ‘If he is a man (i.e., if he has courage), he should come.’ (J. Essegbey, p.c.) Finally, in the two major lexifier languages of Sranan, English and Dutch, the noun man is also associated with alleged qualities of the male gender, such as being brave. Compare for example the Dutch expression mans genoeg zijn (lit. ‘be enough of a man’), meaning ‘to be able to, to be capable of ’, used in contexts where the addressee is being challenged. In at least one English-based pidgin the use of man in a nominal predicate — but with an overt copula meaning ‘to have the nerve to’ — is encountered: (15) Yu no be man. (Ghanaian Pidgin English) you no be man ‘You don’t have the nerve.’ (J. Essegbey, p.c.) Given this evidence, it seems best to regard the 18th-century Sranan noun man in a nominal predicate with zero-copula as the result of convergence, since both the lexifier and the substrate languages contributed to its emergence. Its later grammaticalization into an auxiliary was probably an independent, language-internal development.

3. The TMA system: The expression of counterfactuality In the court records, overt marking of T(ense) (expressed by ben), M(ood) (zal, sal, sa, saa, zoe) and A(spect) (de) is found from 1745 onwards. The first combination of TMA markers is not attested before 1757. This is in agreement with Baker’s (1995) finding that combinations of TMA particles appear relatively late in the development of creoles. The earliest combination of markers involves the sequence sa ben, with the rather remarkable MT order instead of the ‘canonical’ TM order (cf. Bickerton 1981: 281; Bakker, Post and van der Voort 1995: 249). This MT order occurs in the utterance given in (16). Moreover, it is found in two slightly different statements, invariably in the sequence sa ben. This is not only the earliest but also the only case of combined TMA markers in the corpus. (16) Evie mi massra ben sendie mie go na if my master PAST send I go to


Tampatie dan mie sa ben soria dem Backara. Tempati then I FUT PAST show the whites ‘If my master had sent me to Tempati, then I would have shown the Whites.’ (CR 1.05.10.02, 942, f 260 vo; 1757) The presence of past tense ben in the main clause and the combination of past and future in the subordinate clause strongly suggest that this sentence should be interpreted as counterfactual. However, the ordering of past and future found here (sa ben rather than ben sa) is at variance with the ordering normally found both in early and contemporary Sranan: (17) Effi mi no ben takki gi ju, if I no PAST say give you ju no ben sa sabi. you no PAST FUT know ‘If I hadn’t told you, you wouldn’t have known.’ (Schumann 1783: 60; orig. transl.: ‘wenn ich dirs nicht gesagt hätte, so würdest du es nicht wissen (od. gewusst haben)’) Although the apparent idiosyncrasy of (16) above might lead us to regard sa ben as exceptional, and therefore not worthy of discussion, there are several reasons to consider it interesting. First of all, (16) cannot be simply dismissed as a “speaker’s Sranan competence issue”. While problems with speaking Sranan are sometimes reported in court records like the ones under scrutiny here, nothing of the sort is said in this particular case. Second, it should be noted that the sequence sa ben occurs in all three versions of this sentence, which renders the possibility of a transcription error rather unlikely. Third, while the sequence ben sa is clearly the preferred order in both early and modern Sranan, the reverse order is possible, although its use and meaning are different. Seuren (1981) presents several modern Sranan sentences where sa ben is used to express a past future (‘will have’), a past epistemic (‘must have’), as well as a number of other meanings. “Past future” seems to be the meaning of the two cases of sa ben that occur in the Sranan version of the (1762) Saramaka Peace Treaty, recently discovered by Hoogbergen and Polimé (2000). Compare (18) (based on the transcription published in Arends and van den Berg 2004): (18) fiscalla sa paij dem f50: Sur: moni da f42 pissie vo serem, ma effie dem kisi dem na krosibay, na wan plandasie, dem sa kisie tien piesie fo er Sch tee f42: na fasie dem SA BEN kisie dem farra weij, ofoe korosie bay foe fotto (Saramaka Peace Treaty 1762: f 180 ro). ‘the attorney will pay them f 50,– in Surinamese money, that is forty-two shilling, but if they catch them on a close-by plantation they will receive between ten and forty-two shilling [per slave?], depending on whether they WILL HAVE caught them far away or close to Paramaribo.’ While this shows that the order sa ben did occur around 1760, the fact remains that sa ben expressed past future (as it apparently does in modern Sranan) rather than

29

30


counterfactual, as it does in (16) above. We agree with an anonymous reviewer who pointed out that the speaker who uttered (16) could have been a fluent but non-native speaker of Sranan on the grounds that fluent L2 speakers of English often make mistakes with verb tenses in counterfactual clauses. Since the entire Black population of Surinam was outnumbered by newly-imported slaves every three to five years during the first fifty years of colonization (Arends 1995b), the speaker could indeed very well have been a fluent but non-native speaker of Sranan. At the same time, the fact that sa ben was not “corrected” into ben sa by the court’s clerk argues for these court records as a reliable data source.

4. Aspects of the copula system An important reason for selecting the copula as a topic for this study is that the historical development of the Sranan copula had already been investigated on the basis of a large number of early sources (Arends 1989, 1995c). This makes it possible to relate our findings from the court records to those of earlier findings. These earlier findings can be summarized as follows. While early 18th-century Sranan does not seem to have used an overt copula in any type of copula construction, by the end of that century (Schumann 1783) the use of da as well as de was common albeit not categorical. The functional distribution of these items was such that da was used in presentative (e.g., ‘it’s me’) and equative sentences (e.g., ‘I am a linguist’); de appeared in equative and locative/existential sentences (e.g., ‘I am in the woods’). In van Dyk (ca. 1765), the only substantial source for mid-18th-century Sranan, da (never de) is used in equatives. Throughout its history, adjectival predicates in Sranan have been treated as if they were verbs, i.e., with no overt copula intervening between subject and adjective. This situation is summarized as in Table 2. In view of the scarcity of data for the pre-1765 period, the Sranan sentences found in the early court records, ranging from 1707 to 1767, could provide crucial additional evidence with regard to the development of the copula system in the first half of the 18th century. It is to these data that we shall now turn. Apart from the sentences with man (see (2)–(4) above), which were analyzed above as copula-less equative predicates, the 1707 passage contains one other equative sentence; see (19). Table 2. The development of the copula in 18th-century Sranan

Herlein 1718 van Dyk ca. 1765 Schumann 1783

Presentative

Equative

Locative

Adjectival

no data da da

Ø Ø/da Ø/da/de

no data de de

Ø Ø Ø


(19) jou no meester voor mi. you not master for me ‘you are not my master.’ (CR 1.05.04.01, 234, f 268 vo; 1707) The absence of an overt copula in sentences (2)–(4) and (19), all of which date from 1707, fits the picture presented in Table 2, where zero-copulas are shown to be normal in early 18th-century Sranan equatives. Compare, for example, the following nearly identical sentence from Herlein (1718): (20) Oe.som.bady Mastre vor joe? who master for you ‘Who is your master?’ (Herlein 1718: 281) The remainder of the corpus contains two other equative sentences, dating from 1761 and 1762, respectively. In both of these the copula da is used: (21) Mi da bossiman. I am bushnegro ‘I am a bushnegro.’ (CR 1.05.04.06, 313, f 318 vo; 1761) (22) Wie da ningre voor joe. we are slaves for you ‘We are your slaves.’ (CR 1.05.10.02, 806, f — vo; 1762) The use of da in these two sentences, dating from the same period as van Dyk (ca. 1765), in which da but not de is used in equatives, is in accordance with the situation presented in Table 2. This supports the idea that the split which has been observed to occur within the equative copula system, with da being used for the expression of identity and de for attribution (Arends 1986, 1989), did not commence until shortly before Schumann first recorded these two forms in his 1783 dictionary. The same copula form, da, is found twice in our corpus in presentative sentences, both dating from 1759. One of these sentences is given in (23): (23) Da him. it.is him ‘It’s him.’ (CR 1.05.10.02, 947, f — vo; 1759) Again, this is in accordance with Table 2, where it is shown that da was used in this context around 1765. The occurrence of presentative da in 1759 lends support to the claim made in previous research (Arends 1986, 1989) that the development of da as a copula (from a pronoun meaning “that”) started in presentative contexts such as these. As far as adjectival predicates are concerned, only one example occurs in our corpus; in accordance with the consistent treatment of predicative adjectives as verbs, no overt copula is used:

3

32


(24) Argus, mie dede. Argus I dead ‘Argus, I’m going to die.’ (CR 1.05.10.02, 947, f — vo; 1759) An almost identical sentence occurs in a source from a slightly later period: (25) Massera, we dede. master we dead ‘Master, we’re going to die.’ (Stedman 1790: 434) Finally, let us examine locative constructions. In all five cases in the corpus an overt copula de/dee/die is present, which is in line with the situation presented in Table 2: (26) Hoe.sambre dee? who is.there ‘Who’s there?’ (CR 1.05.10.02, 798, f — vo; 1745) (27) Voevoereman de na hosse dappe. thief is LOC house top ‘There’s a thief on the roof.’ (CR 1.05.10.02, 808, f — vo; 1763) In summary, the use of the copula as it occurs in the early court records is fully in accordance with what we know about the development of the copula system on the basis of a range of other documents.

5. Conclusions We have shown that several of the syntactic patterns found in the early court records are in accordance with what we find in other early sources. This not only means that the court records may be considered accurate reflections of early 18th-century Sranan, but it also strengthens the case for the reliability of other texts whose representativeness has at times been questioned (e.g., Bickerton 1991). In this paper we have examined only a few of the topics on which the early court records provide valuable data. We have, for example, said nothing on phonological and morphological issues. With regard to the latter, phenomena such as derivation, compounding and reduplication would be particularly interesting topics to investigate on the basis of early court records. Archival sources

(NA = Nationaal Archief) Inventaris van de archieven van de Raad van Politie (1669–1680) en de Raad van Politie en Justitie (1680–1683) en het Oud-Archief van het Hof van Politie en Criminele Justitie in Suriname (NA, code 1.05.10.02, inventory numbers 781–948). Overgekomen brieven en papieren uit het archief van de Sociëteit van Suriname, 1683–1715

Court records as a source of authentic early Sranan (NA, code 1.05.04.01, inventory numbers 212–40). Overgekomen brieven en papieren uit het archief van de Sociëteit van Suriname, 1751–1767 (NA, code 1.05.04.06, inventory numbers 286–335). Oud-Archief Suriname/Hof van Politie en Criminele Justitie (NA, code 1.05.10.02, inventory number 66).

Notes 1. CR = Court Records; f = folio. 2. Information in this paragraph is based on Beeldsnijder (1994). 3. Since Suriname was an English colony until 1667, no pre-1667 records are available in Dutch archives. 4. It is interesting to note that Mingo and Waly may be represented on Dirk Valkenburg’s 1707 painting entitled Slavendans ‘Slave dance’, which was made the year the incident occurred. Dirk Valkenburg was employed as a bookkeeper and artist on Jonas Witsen’s plantation “Palmeneribo” — the place where the incident took place. For a reproduction of the painting, see Carlin and Arends (2002).

References Arends, J. 1986. “Genesis and development of the equative copula in Sranan.” In Substrata versus Universals in Creole Genesis, P. Muysken and N. Smith (eds.), 103–27. Amsterdam: Benjamins. Arends, J. 1989. Syntactic developments in Sranan: Creolization as a gradual process. Ph.D. Thesis. Katholieke Universiteit Nijmegen. Arends, J. (ed.) 1995a. The Early Stages of Creolization. Amsterdam: Benjamins. Arends, J. 1995b. “Demographic factors in the formation of Sranan.” In Arends (ed.) 1995a: 233–85. Arends, J. 1995c. “Introduction to Part I.” In Arends and Perl, 11–71. Frankfurt: Vervuert. Arends, J. and M. Perl 1995. Early Suriname Creole Texts: A Collection of 18th-century Sranan and Saramaccan Documents. Frankfurt: Vervuert. Arends, J. and M. van den Berg 2004. “The Saramaka Peace Treaty in Sranan: An edition of the 1762 text (including a copy of the original manuscript).” Creolica: Revue du Groupe Européen de Recherches en Langues Créoles (http://www.creolica.net). Baker, Ph. 1995. “Some developmental inferences from historical studies of pidgins and creoles.” In Arends (ed.) 1995a: 1–24. Bakker, P., M. Post, and H. van der Voort, 1995. “TMA particles and auxiliaries.” In Pidgins and Creoles: An Introduction, J. Arends, P. Muysken and N. Smith (eds.), 247–58. Amsterdam: Benjamins. Beeldsnijder, R. 1994. Om werk van jullie te hebben’: Plantageslaven in Suriname, 1730–1750 [Bronnen voor de Studie van Afro-Surinaamse Samenlevingen 16] Utrecht: CLACS/ IBS Bickerton, D. 1981. Roots of language. Ann Arbor: Karoma. Bickerton, D. 1991. “On the supposed ‘gradualness’ of creole development.” Journal of Pidgin and Creole Languages 6: 25–58.

33

34

Margot van den Berg and Jacques Arends Carlin, E. and J. Arends, (eds). 2002. Atlas of the Languages of Suriname. Leiden: KITLV Press. Focke, H. C. 1855. Neger-Engelsch Woordenboek. Leiden: P. H. van den Heuvell. Herlein, J. D. 1718. Beschrijvinge van de Volksplantinge Zuriname. Leeuwarden: Meindert Injema. Hoogbergen, W. and T. Polimé. 2000. “De Saramakaanse vrede in het Sranantongo.” OSO 19: 221–40. Lappe, S. and I. Plag. 2003. “Rules versus analogy: Modeling variation in word-final epenthesis in Sranan.” In Plag, 71–90. Nepveu, J. 1770. “Annotatien op de Surinaamsche Beschrijvinge van Ao 1718.” In Arends and Perl, 77–91. Orton, H. and M. Barry. 1971. Survey of English dialects: The basic material. Vol. 2: The West Midland Counties. Leeds: Arnold. Plag, I. (ed.). 2003. Phonology and Morphology of Creole Languages. Tübingen: Niemeyer. Roberts, J. 1995. “Hawaiian court records as a source of old pidgin texts: An update on recent research.” Paper presented at the conference of the Society for Pidgin and Creole Languages, New Orleans, January 1995. Schneider, E. 2002. “Investigating variation and change in written documents.” In The Handbook of Language Variation and Change, J. K. Chambers, P. Trudgill and N. SchillingEstes (eds.), 67–96. Oxford: Blackwell. Schumann, C. L. 1783. Neger-Englisches Wörterbuch. MS. Seuren, P. 1981. “Tense and aspect in Sranan.” Linguistics 19: 1043–1076. Smith, N. 2003. “New evidence from the past: To epenthesize or not to epenthesize? That is the question.” In Plag, 91–107. Stedman, J. G. 1988 [1790]. Narrative of a five years expedition against the revolted negroes of Surinam. Ed. by R. Price and S. Price. Baltimore: The Johns Hopkins University Press. van den Berg, M. C. 2000.“Mi no sal tron tongo”: Early Sranan in court records, 1667–1767. Nijmegen: MA Thesis Katholieke Universiteit. van Dyk, P. n.d. (ca. 1765). Nieuwe en nooit bevoorens geziene onderwijzinge in het Bastert Engels, of Neeger Engels. . . Amsterdam: Wed. Jacobus van Egmont. Republished in Arends and Perl, 93–239. Frankfurt: Vervuert. Weygandt, G. C. 1798. Gemeenzame leerwijze, om het basterd of Neger-Engelsch op een gemakkelyke wyze te leeren verstaan en spreeken. Paramaribo: W.W. Beeldsnijder. Winford, D. 2000. “Irrealis in Sranan: Mood and modality in a radical creole.” Journal of Pidgin and Creole Languages 15: 63–125.

CHAPTER 3

Garifuna in Belize and Honduras * Geneviève Escure

University of Minnesota

Garifuna is a mixed language spoken in Central America, especially in Belize and Honduras. The present study offers a sociolinguistic update on the current situation in both countries — an urgent task, considering the language is rapidly losing speakers to dominant languages (Creole English, and Spanish). The linguistic scene is first examined in terms of the Garifuna group’s historical contacts and conflicts with other groups, and its ensuing diaspora from South America to Central America. The linguistic consequences of this diaspora are then discussed, with special reference to the diverse components — Arawak, Carib, French, English, Belizean Creole, and Spanish — that make up Garifuna. Garifuna is here shown to be undergoing differential changes in Belize (where it is dominated by Belizean Creole) and Honduras (where it is subjected primarily to Spanish influence). My conclusion is that morphological changes akin to creolization are currently affecting Garifuna, especially within the domain of past tense marking.

1. Introduction Rapid language loss is currently affecting the Garifuna-speaking communities of Central America (Belize, Honduras and Guatemala). Garifuna is an endangered language because it is no longer fluently spoken by the younger generations, and older generations (over 50) are regularly involved in code switching (to English, Belizean Creole, or Spanish). The present study offers a preliminary analysis of some of the ongoing linguistic changes that appear to occur in Garifuna, and how these may accelerate its decline. Garifuna has been claimed to have a primary Arawak substrate, combined with other linguistic elements, including Carib (genetically related to Arawak), Spanish, French and English (Taylor 1951, 1955, 1956a, 1956b, 1958, 1962, 1977). All Arawak (Maipuran) languages of South America are endangered, but Garifuna, a version of Arawak transported to Central America, has been described in comparison as “one of the few healthy Arawak languages” (Aikhenvald 1999: 72). Although this might have been true fifty years ago, this is unfortunately no longer the case.¹ The case of Garifuna is particularly interesting because this language is the prod-

36

Geneviève Escure

uct of multiple social contacts — due primarily to migration and colonization — that occurred in the last fourteen centuries among diverse ethnic groups. A major difficulty with conducting research on Garifuna is its areal fragmentation, as it is now distributed over several Central American countries (primarily Belize and Honduras, and also in small portions of Guatemala [see Map 1]). Moreover, the substantial emigration, mostly to the United States, of what used to be — at least during the 20th century — a tightly confined ethnic group certainly accelerates the splintering of the group, and the ensuing loss of language and identity. Table 1 offers a demographic estimate of the Garifuna population in Central America (Section 2.1). I begin with a brief overview of the sociohistorical contacts with other languages that led to the development of Garifuna, and crucially determined its lexical and syntactic idiosyncracies (Section 2). I discuss in some detail the extensive IndoEuropean (French, English, Spanish) borrowings that combined with the original Amerindian components (Arawak and Carib), and in some cases facilitated the grammaticalization of certain items (Sections 3 and 4) to give rise to the language in its current state. I then proceed to an analysis of some of the linguistic mechanisms that are currently indicative of language attrition (Section 5). My observations suggest a possible connection between language death and language birth (the latter may be occurring either as pidginization or creolization). Thus, the hypothesis that a new variety of Garifuna is now evolving will be examined in the context of certain morphological changes observed in Belize and Honduras. When I refer to creolization, I do not imply that Garifuna is a creole — nor do I mean to say that it is not — I merely suggest that phenomena akin to creolization are currently affecting the language, as it becomes increasingly subjected to the hegemony of vigorous contiguous languages.

Map1. Garifuna in Central America-

Garifuna in Belize and Honduras

2. Sociohistorical background: from Black Carib to Garinagu The Garifuna adventure began around 600 AD, as South American Arawaks migrated to the Lesser Antilles. Carib Indians — also from the Amazon area — followed the same route three or four centuries later. It is assumed that Arawaks and Caribs clashed, and eventually intermarried not long before Columbus discovered the New World (Breton 1666, Taylor 1951, Young 1795). Their descendants are called Red Carib or Yellow Carib. As Europeans brought slaves from Africa, the Caribs seem to have welcomed among them various groups of African maroons — shipwreck survivors, or escaped slaves. The offspring of this additional interethnic situation became known as Black Carib, Charaibes Noirs, Karaib Negros, Garif or Morenos, according to early texts (Breton 1665, 1666, 1667, Coelho 1955, Conzemius 1928, Gullick 1976, Taylor 1951, Young 1795). This group was not officially recognized until 1981, when Belize gained independence from England. At that time, the term Black Carib gave way to Garifuna (and Garinagu) as a sign of respect. The term Black Carib is henceforth used for any pre-1981 situation, and Garifuna/Garinagu is reserved to references posterior to the official recognition of Garifuna ethnicity in Belize.² Throughout the 18th century, the Black Caribs were involved in hostilities opposing the French and the British, who fought over possession of St. Vincent and other islands in the region. Although they sometimes sided with the British, the Black Caribs were often allies of the French, which explains the important French elements present in the Garifuna language. This ambiguous relationship led to a momentous event. In retaliation to a failed rebellion against the British (subsequent to the French withdrawal from St. Vincent), the British deported 5,000 Black Caribs to the deserted island of Roatán in 1797, the largest of the Bay Islands of Honduras. Unlike other regions along the Central American Caribbean coast — such as Belize, Nicaragua and Panama — Honduras never received large contingents of African slaves. The Spanish never settled on the Bay Islands, and British attempts were short-lived. The first permanent English settlement on Roatán was established in the 1830s after emancipation, as freed slaves and former slave owners migrated to the island from the Cayman Islands, Belize and the Mosquito Shore. Thus, some thirty years after their arrival, the Black Caribs probably interacted with English-speaking newcomers. This contact undoubtedly contributed to the multilingualism of many older Garinagu. The Black Caribs’ tribulations continued during most of the 19th and 20th centuries. First, they were caught in the struggle opposing the Spanish and the British over the possession of Central American territories. Then, they were engaged in the wars of independence of Central American nations from Spanish rule. Thereafter, they became caught up in the civil wars between liberals and conservatives in Guatemala and Honduras. The Bay Islands territory became briefly a Crown colony in 1852, but was ceded to Honduras seven years later. On the other hand, the neighboring settlement of British Honduras became a Crown colony in 1862, and

37

38

Geneviève Escure

Map 2. Garifuna in Central America (major Garifuna communities are undrscored; field sites are starred)

remained so until independence from England in 1981, at which time the country assumed the name of Belize. Black Carib survivors of a massacre perpetrated by the Honduran government moved to British Honduras in 1939, and founded the village of Hopkins. Others had also fled at various times to British Honduras, founding the cities of Stann Creek (renamed Dangriga) and Punta Gorda (near the Guatemala border), or eastward to Honduran Mosquitia and onward to what is now Nicaragua (Map 2). In the course of the 20th century, the United States increasingly interfered with local Central American politics, and became established as an economic power through the construction of the Panama Canal and the growth of the United Fruit Company. These economic opportunities triggered various population movements across Central America, and many Black Caribs achieved relative prosperity for a while through labor in lumber camps, banana plantations, or in the construction of the Panama Canal. But natural disasters (hurricanes, floods, fires), and the 1930s economic slump reversed any prior positive trends, and contributed to the great wave of northward immigration. This immigration to the United States continues today, involving all ethnic groups (Escure 1979, 1982, 1997). There are no official figures on emigration, but it is estimated that the population living outside Central America is at least as large as that residing in the region. There are constant movements in and out of the region, which makes it difficult to estimate the actual Garifuna population in Central America.


2.1 The current situation: demography and census data The Garinagu who remain in Central America live in small communities, mostly in Belize and Honduras, where the majority of the Garifuna population resides. There are a few Garifuna communities on Guatemala’s narrow Caribbean coast, such as Livingston, where the native language competes with Spanish, as it does in Honduras. Nicaragua no longer has Garifuna speakers. Population estimates vary: 98,000 for Honduras alone (Rivas 1993: 257); 30,000 to 100,000 for “Central American Arawak speakers” (Aikhenvald 1999: 72). The group’s numbers in Honduras are uncertain, as Garinagu and Afro-Caribbean people (Creoles) are often counted together as morenos (Blacks), or población negra de habla inglesa (English-speaking black population). This terminology is indeed ambiguous since the Garinagu also speak a variety of English, at least on Roatán (Escure forthcoming). The Creoles and the Black Caribs have interacted since the mid 1800s, and all the older Garinagu speak a variety of English Creole, not unlike that spoken in the Cayman Islands or in Belize. The confusion occurs even among the local population on Roatán — including the Garinagu, who often believe that they are solely of African descent, when in fact they also have Amerindian ancestry. A Honduran association for indigenous people — the Confederación de Pueblos Autóctonos de Honduras (CONPAH 1998) — estimates the Garifuna people to number 150,000, but fails to provide figures for “English-speaking blacks”, who are merely mentioned in passing. However, more recent figures obtained in Honduras in 2003 (Fiscalia especial de etnia y patrimonio cultural, Ministerio público, Honduras) give separate figures for English-speaking blacks (80,000) and Garinagu (250,000). But the figure of 250,000 seems too high for Honduras alone, especially if it excludes the Creoles, although it is probably appropriate if it includes those Garifuna expatriates who settled in the United States. Extended families are also difficult to count because of constant population movements in and out of the three Central American countries in which the Garinagu habitually reside.³ On the other hand, Belize has officially recognized the separate existence of the Garifuna group, and included it in the two latest population censora.⁴ The 2000 Belize census indicates that the Garinagu amount to 6.5% of the population, also concentrated in specific areas, as discussed below. Table 1 displays the relative proportion of Garinagu in comparison to Mestizos and Creoles in both countries. It is likely that the Central American Garifuna population is still decreasing, as emigration to the United States steadily increases. Table 1. Garinagu in Central America

Belize Honduras

Population

Garinagu

(%)

Mestizos (%) Creoles (%)

240,204 6,560,608

15,685 200,000

(6.5%) (3%)

46.4% 88.0%

27.7% 1.2%

39

40

Geneviève Escure

2.2 Honduras From the original settlement of Punta Gorda, which is located on the northeastern coast on the island of Roatán, the Garinagu moved out into the Honduran mainland soon after their arrival in 1797. They established a number of small towns and villages along the coast in the eastern provinces of Colón and Gracias de Dios (just north of Nicaragua), as well as in the northwestern provinces of Atlántida and Cortés. According to Coelho (1955), half a century ago, there were thirty-six Garifuna communities in Honduras. This investigation is limited to Punta Gorda (ca. 1,000 inhabitants), the only primarily Garifuna community on the island (see Map 2). After their deportation to this mountainous and lush area, the Black Caribs were totally isolated, at least until the Creoles’ arrival forty years later, and this isolation may have contributed to the maintenance of their language and culture. While the Garinagu inhabit the eastern side of the island, the Creole population mostly lives on the western part (in Sandy Bay, West End and Coxen Hole), although there is a Creole village (Politilly Bight) in close proximity to Punta Gorda. In the past, Black Caribs and Creoles probably interacted (as they still do) in the busy commercial ports of French Harbour and Oak Ridge, both located in the middle section of the island (see Map 3). Until the late 1980s, various missions were in charge of education, uniformly given in English. Around that time, the Honduran authorities finally addressed the long-standing neglect of the Bay Islands, as they realized the economic potential of the area. The government then began to develop a tourist industry, as well as a basic infrastructure and educational institutions. Schooling in Spanish is now obligatory, and Spanish has become the primary language of the younger generations (age 20 and below). The recent commercial development of the island has been accompanied by an increase in the Hispanic population from the Honduran mainland. Especially for the young on Roatán, today the incentives to learn American English are considerable, as the developing tourist industry (mostly upscale diving groups in a few select hotels) provides attractive jobs for locals. Young adults are increasingly socialized in outgroup cultural and linguistic systems, and as a result, they often neglect their native identity. For these various reasons, Garifuna is being replaced by the combined influences of Spanish and English. Although Spanish is not as overwhelmingly dominant on the island as it is in the country at large — the Mestizo

Map 3. Roatán island


population constitutes 88% of the overall population of Honduras — it is definitely competing with English.⁵ There are informal reports that Garifuna is more strongly rooted on the Honduran mainland. While on Roatán, I met many Garinagu from the coast (Trujillo or other Garifuna communities), all of whom are monolingual speakers of Spanish (with the exception of a few, who are generally trilingual in Garifuna, Spanish and English; these speakers are, however, all over 40 years of age).

2.3 Belize As the Garinagu spread in a northwesterly direction along the coast to Guatemala, and on to Belize, they founded several coastal communities in the two southernmost districts of Belize, Toledo District (contiguous to Guatemala), and Stann Creek District. The largest Garifuna towns are Dangriga (Stann Creek District), and Punta Gorda (Toledo District). In addition, there are four villages (Barranco, Seine Bight, Hopkins and Georgetown), all primarily inhabited by Garinagu (see Map 4). The Black Caribs were barely recognized until Belize’s independence, and eked out a meager living by working on citrus or banana plantations. Those who lived in Seine Bight had developed an economic symbiosis with the neighboring Creole village, which entailed the use of Belizean Creole by the Garinagu. Until the late 1980s, the Creoles had a flourishing fishing business (fish, conch, shrimp and lobster), and the

Map 4. Garifuna in Belize

4

42

Geneviève Escure

Garinagu worked for them, usually in subaltern positions, packing seafood for export, cutting bush or washing clothes. This dependency has now changed, as the Garinagu have gained acceptance and pride, thanks in part to the official recognition campaign launched by the new government. Many Garinagu now hold relatively prestigious jobs (police, army, teaching). But this upward mobility has also led to Garifuna language loss, as increased contacts with other groups, especially Creoles and Mestizos, have induced them to speak Belizean Creole (English-based) with growing frequency (Escure 1979, 1982, 1984). However, many Garifuna communities are still mostly destitute, and emigration to the United States seems more intense than ever. Although Mestizos now constitute the majority group overall, having surpassed the Creoles in the last fifteen years, the Garinagu mostly live in districts (given in bold in Table 2) where the Creoles predominate (Belize, and Stann Creek). In the Toledo district adjacent to Guatemala, Maya-Kekchi speakers constitute the majority, but the Garinagu primarily live in the small town of Punta Gorda (also inhabited by the Creoles), whereas the Amerindians reside in rural areas. This means that they are exposed to English varieties rather than to Spanish. The few Garinagu who live in the western (Cayo) and northern (Corozal and Orange Walk) districts are more likely to speak Spanish.⁶ This investigation was conducted in the village of Seine Bight (see Maps 1 and 4) — with a population of 500–located in the Stann Creek district, in close proximity to the Creole village of Placencia.

3. What is Garifuna? Over the last 400 years, the ancestors of the Garinagu have come into frequent language contact with peoples of varied linguistic and ethnic background. In the absence of historical records, it is difficult to assess precisely which items or structures were borrowed and when.⁷ The linguistic integration and acceptability of loanwords can, however, be evaluated on the basis of their frequency of use, native Table 2. Percentages of ethnic groups in the six districts of Belize (Escure 1997: 31, updated with Census information, 2000) Districts

Creole

Garifuna

Mestizo

Maya

Corozal Orange Walk Belize Cayo Stann Creek Toledo

7.6 7.4 67.9 23.0 25.1 5.7

1.3 1.2 5.3 1.7 36.2 10.0

74.1 71.7 18.7 58.0 23.7 11.9

5.0 9.1 1.2 8.7 8.0 62.8


language synonym displacement, morphophonemic and/or syntactic integration, and so on, as discussed below. Garifuna is generally classified as Arawak (Maipuran), but its actual status is somewhat unclear. It is variously qualified as “a Carib-based Creole” (Merrill 1995: 98), “a mixed Arawak-Carib pidgin” (Aikhenvald 1999: 75), or “un imbroglio linguistique” (Renault-Lescure 1999). One of the first French missionaries in Guadeloupe, who spent 18 years (1638–1654) close to the population of Dominica and St. Vincent, describes a single language that he calls “Carib” (Breton 1666). According to him, the Caribs attacked the Arawak settlers, exterminated the men, but spared the women.⁸ He assumed that the language of the resulting group was Carib, the victors’ language, with only limited Arawak elements used by the women and children. The Arawak component was also recognized by Adam (1877), who claimed that the language described by Breton two centuries earlier was then dying — “sur le point de passer à l’état de langue morte” (Adam 1877: ix). Following Adam, Taylor observed that the language described by Breton is Arawak, not Carib, and that the men had kept only limited elements of their ancestral Carib language (Taylor 1951, 1977). In the late 1940s Taylor focused on the variety spoken in Hopkins, British Honduras. He called that variety Island Carib, and concluded that Hopkins Garifuna had lost many of the Carib components recorded in Breton’s grammar. He assumed that the 17th-century “Carib” situation involved “a kind of diglossia”, and that “the Carib words were nurtured by men through a sort of machismo as the language of the more prestigious nation” (Taylor 1977: 98). This interpretation would suggest that, during the early stages of European colonization, Carib was the superstrate and Arawak the substrate, since the Caribs conquered the Arawaks. This phenomenon is observed in contact situations leading to the emergence of pidginization, in which the victors’ language yields the lexicon, whereas the subaltern group projects its native phonology and morphology. One might thus expect Carib lexicon and Arawak morphosyntactic patterns, or perhaps eventual dominance of Carib leading to Arawak obsolescence. However, Taylor claims Arawak dominance, at least in Hopkins “Island Carib”. This assertion has not been challenged, but there is some evidence that the Carib component is more extensive than previously assumed. It is essential to reassess the state of Garifuna, and the current nature of its components. Is it a creole, a semi-creole, an intertwined language? The objective of this preliminary investigation is to provide some answers to these questions.

3.1 Documenting a dying language Crystal defines moribund languages as having “only a handful of good speakers left, mostly very old” (2000: 21). It is a major challenge in the study of language death to determine who is a good speaker, as well as who still speaks the moribund language at all. The actual competence or fluency of the speaker is often in doubt, because it

43

44

Geneviève Escure

is extremely difficult to establish when exactly the language is changed, or decaying to the extent that it no longer qualifies as the original language. When is change loss? And when is it innovation? I encountered many Garinagu who claimed they knew Garifuna, though they had not used their native language for a long time. They sometimes produced versions deemed “incorrect” by others. I found it helpful to consult several respondents for each recorded sample, and to obtain different translations and opinions for the same spontaneous conversations. Natural discourse probably offers the most reliable method to document a language that is undergoing attrition in its community. By the same token, translations of prepared key sentences are often artificial because they are context-free, and thus likely to appeal strictly to a formal competence, and not to language use. Furthermore, surveys of self-reported competence in the endangered language (that is, questions of the type: “Do you speak Garifuna?”) are often tainted by psychological factors.⁹ Although interesting from a psychological viewpoint, responses to this simple question are almost certainly of little reliability when assessing the vitality of the language. The collection of natural discourse is a lengthy, time-consuming process, but it is also highly productive, as it makes it possible to gain crucial insights into the actual mechanisms of obsolescence. I conducted observations in Punta Gorda, the original Central American settlement of the Garinagu on the island of Roatán, Honduras, as well as in the village of Seine Bight in Belize. One of my objectives is to examine some of the (recent) contact-induced differences that may have developed between Honduran and Belizean Garifuna. I had established prior contacts in Seine Bight in the early 1980s. At that time, Seine Bight was isolated on the Placencia Peninsula, as only a beach path connected it to the closest Creole village of Placencia, located five miles south on the coast (see Map 2). My contacts with the Garifuna population in Honduras are recent (1999). Several efforts at documenting Garifuna culture and language took place during the last thirty years, mostly in Belize. The first Carib-English dictionary (Stochl 1975) is a useful but formal reference tool (containing some grammatical information) based in part on the Garifuna spoken in Seine Bight.¹⁰ Later efforts to preserve Garifuna culture and language include a Garifuna-English dictionary (R. Cayetano 1993), a grammar based on Taylor’s work (Suazo 1991), and descriptions of Garifuna history and culture (S. Cayetano 1990, Gonzales 1988, Kerns 1989).

3.2 Arawak and Carib elements Garifuna in its current state at the beginning of the 21st century has clearly preserved important elements of Arawak morphology, and also a large number of Carib items. The agglutinating patterns still present in Garifuna reflect areal traits typical of all Amazonian languages (that include both Arawak and Carib languages). Of the typological characteristics of Amazonian languages described in Adam (1893), Dixon and Aikhenvald (1999: 8–9), Fleming (1965), Hoff (1968), Matteson


(1972), Noble (1965), and Pelleprat (1656), the following features can be recognized in Belizean and Honduran Garifuna: Some Arawak features of Garifuna (preliminary overview): (i) Polysynthetic morphology. (ii) A five-vowel system [i, e, a, ø, o/u] shared by most Amazonian languages.¹¹ (iii) Grammatical gender (masculine-feminine) affecting all nouns, but now marked only by a postposed determiner (le/to) and an occasional (but rare) suffix (-e/-o) attached to some adjectives. (iv) Bound prefixed pronominal forms marking gender, person, number, and possession that are practically identical to those found in some Arawak varieties (Noble 1965: 27). Carib also has prefixed forms, though lexically different (Adam 1893, Hoff 1968: 279). However, pronominals can also be suffixed in some instances, which suggests the possibility that a split ergative system may still be represented in Garifuna, as it is in Arawak (Aikhenvald 1999: 87–9).¹² (v) TMA verbal markers occur infrequently as suffixes. Dixon and Aikhenvald (1999) refer to Amazonian “optional suffixes”. But in Garifuna, there is a shift toward preverbal markers (to be developed in Section 5 below). (vi) Adverbs and adpositions are often suffixed to nominal and verbal categories. (vii) Amazonian languages typically have a very small class of lexical numbers. Garifuna has kept only the numerals “one” and “two”, from Arawak (aban and biama, respectively), whereas “three” — ørøwa — comes from Carib (Adam 1893). Garifuna later acquired a full set of numbers through French suppletion. Breton (1667: 10) cites five “Carib” numerals, which constitutes evidence that French suppletion for “four” and “five” had not yet been implemented. The Carib inheritance is represented in lexical morphemes, but also includes regular suffixes. However, a full discussion of Carib elements cannot be given here. The following features are merely indicative of the persistent presence of Carib in Garifuna: Some regular Carib morphemes (i) A nominalizing suffix -un(i), which is common in subordinating constructions, as in eiga ‘eat’; eigini ‘food’. There is a cognate suffix -ne or -ri in some Carib languages (Derbyshire 1999: 53). (ii) A large number of evidential particles, or items marking “verification”, occur in Carib and appear to be widely represented in Garifuna. Those particles express the speaker’s degree of certainty or authority. Some of the most common suffixes include -ti (hearsay), -na (uncertainty), and -me (deductive). The features listed above are not consistently present in contemporary Garifuna. The most regular pattern appears to be that of pronominal/possessive bound prefixes and suffixes (Table 3), which almost exactly matches Arawak cognates. Current

45

46

Geneviève Escure Table 3. Pronominal and possessive prefixes and suffixes (Arawak data from Aikhenvald 1999: 8) Garifuna 1S 2S 3SMasc 3P 1P 2P 3P

Prefixes nblt whh-

Arawak Suffixes -ina -ibu -i -u -wa/iwa -wa/iwa -wa/iwa

Prefixes nu-, or ta(p)iri-, or ithu-, uwa(h)na-

Suffixes -na, or te -pi -ri, or i-thu, -u -wa -hi -na

affix data replicate almost exactly Taylor’s observations of Island Carib in Belize, which were variously discussed in Taylor’s 1950s publications, and later presented in table form (Taylor 1977: 45). On the basis of recent observations, Central American Garifuna thus appears to include synthetic Arawak morphosyntax, as well as lexical and morphological Carib elements tightly integrated into Garifuna grammar. A more precise assessment of the extent of this combination must be left to further analysis.

3.3 Indo-European elements (Spanish, French, and English) The linguistic composition of Garifuna is complicated by the input of European contact languages. This is so because (i) Indo-European items are combined with Arawak and Carib morphology, and (ii) some loans have been grammaticalized. Early contacts in the 16th century probably involved Spanish, followed by contacts with English, French, and their creoles. Breton’s dictionary contained “some six dozen loans from Spanish” (Taylor 1977: 28), but shows few French items, contrary to the extensive number represented in contemporary Garifuna, as also noted by Taylor (he refers to “several hundred”) in the Island Carib variety spoken in the village of Hopkins (Belize). A short text of 500 words exhibited “no fewer than 39 borrowed forms, some of which are free (loanwords) while others occur in construction with Arawakan affixes” (Taylor 1977: 125). Those borrowings include Carib words (aufuri ‘father’s sister’; duna ‘water’), French (busu ‘greet’ < bonjour; gumese ‘begin’ < commencer), and English (siliba ‘silver’; bugidu ‘bucket’).¹³ A more recent count of the items contained in R. Cayetano’s (1993) dictionary yielded a total of 472 non-Amerindian loanwords out of a total of 2,753 “nominal” entries, including 152 derived from French, 203 from Spanish, and 117 from English (Pury 2000). However, my Belizean and Honduran corpora suggest a much higher incidence of borrowed items. This may be because a dictionary provides only a par-


tial view of the language. Furthermore, Stochl’s (1975) and R. Cayetano’s (1993) dictionaries are both based on Belizean Garifuna. Finally, borrowed items may be so well integrated to the phonological system of Garifuna, or its antecedents, that their sources are not immediately recognizable.¹⁴

4. Integration of loanwords Lexical borrowing is the intrasentential incorporation of lexical items from a language into the lexicon of another. Loanwords can be more or less incorporated into the base language. A distinction can be made between nonce borrowings (innovative or occasional), and established items (old and better assimilated). The following discussion focuses on lexifying influences posterior to Amerindian contacts that may have contributed to the current language attrition, namely transfer from French, and Spanish, which is extensive, as well as from English, and creole. African loanwords are surprisingly scarce — with the exception of the very common mutu ‘person’ (from Bantu; e.g. Kikongo mùntu ‘person, head’ [Laman 1964: 619]), represented in (4). This absence parallels the paucity of African lexical items in Caribbean creoles, but African substratal influences at the grammatical level are of course not excluded, especially in view of the contacts the Black Caribs had with creole-speaking people. Casual Garifuna conversations read like a linguistic patchwork, as illustrated in the following sentences excerpted from my corpus. Some ancient borrowings are hardly recognizable because of their phonological integration — for example efeda (1), a reduced form of efeduha, the verbal form of fedu < FR [fεt] (fête) ‘celebration’, hence ‘to celebrate, get a treat’. More recent items are easier to trace, even unchanged, and devoid of Arawak/Carib affixes, as is the case for SP mendigando (2) or pastelito (3). The origin of certain items — especially French or Spanish — is sometimes uncertain, although it can be assumed that recent borrowings are closer to the original lexifier:¹⁵ (1) Ka-ba-sa ora l-efeda sopa le ma?¹⁶ FR-FR-FR SP A-FR SP FR [see note 15] what-FUT-PT time 3M-enjoy soup the mother? ‘When can we expect to enjoy the soup, mother?’ [Honduras M.1.3.] (2) L-un n-eigi-ni musu faluma ora le hafu n-añuru ya benabu A-A A-C-C FR SP SP FR CR A-A CR C 3M-for I-eat-PT piece coconut now PT must I-sit here door mendigando l-un l-iabi SP A-A A-A beg 3M-for 3M-come ‘In order to eat a bit of coconut, you must sit begging at someone’s doorstep.’ [Honduras M.6.10]

47

48

Geneviève Escure

(3) M-adøga-tibu-ha pastelito uguñe? A-A-A-C SP A NEG-bake-you-PT cake today? ‘Didn’t you bake little cakes today?’ [Honduras M. 1.5] (4) Dan-bwe me-ha h-eidin mutu oucha-ha Kalifuna FR-FR A-C A-A WA A-C [place name] time-PT PA-PT they-go people fish-PT Kalifuna’ ‘One time, some people went fishing in Kalifuna.’ [Belize O.11. 1] (5) Lidan-le n-aranseru-bwe l-un n-idi Merige-n irumu disi nefu ørøwa FR-FR A-FR-FR A-A A-A EN-C C FR FR FR time-PT I-ready-PT 3M-for I-go USA-to year ten nine three wein disi nefu FR FR FR twenty ten nine ‘The time that I went to America, it was in 1979.’ [Belize J.1.1]¹⁷ Old and newer loanwords are interspersed with Arawak and Carib morphology, and fully integrated into the grammar, with personal and negative prefixes, and a number of verbal suffixes and particles. The five sentences shown above include almost 50% loanwords (28) out of a total 60 morphemes (with 22 of Arawak origin, and 8 from Carib). Among the loanwords, there is a majority of French source items (18) as compared to Spanish (6) and English/Creole (3), and the lone Bantu mutu. A similar distribution occurs in longer texts, although the distribution of Spanish and English items varies according to the country.

4.1 French loanwords Contact with French was early, but in contradistinction with Spanish was discontinued in the 18th century. It is likely that the Red Caribs (before African miscegenation) and the Black Caribs interacted extensively with French missionaries and troops for as long as one hundred years from the end of the 17th century until their deportation in 1797. It is also possible that they interacted with African maroons speaking French Antillean Creole, which would explain the presence of some items incorporating the definite article, a common feature of French creole, as represented in luburuaniga < FR le brouillard (‘the fog’); or FR laj < l’ail (‘the garlic’). However items with incorporated article are relatively infrequent. French source loans can be identified in terms of the Arawak or Carib phonotactic constraints that assimilated those external items, as they are still active in today’s Garifuna. For example, the following are some of the consonantal substitutions that have applied in the course of the borrowing process: • epenthesis, apocope or consonant deletion due to a constraint on consonant clusters:


musu (murusu) < [morso] morceau = ‘piece’; disi < [dis] dix = ‘ten’. • substitution of voiced stops for voiceless stops: fedu < [fεt] fête = ‘celebration’; gumese < [komãse] commencer = ‘begin’; ‘dan [dan/ dã] < [tã] temps = ‘time, weather’.¹⁸ The above process is unusual since voiced stops are typically more marked than voiceless stops, but note that French voiceless stops are unaspirated. French voiced stops were preserved in word-initial position, as in bulesei < FR blesser = ‘hurt’; or bungiu < bon dieu = ‘God’. There is evidence that this process still applies in place names: Trujillo, a city on the Honduran coast, is Duruweyo. • substitution of the labiodental voiceless fricative for the voiceless bilabial stop: afurendeira < FR [aprãdr] apprendre = ‘learn’; furedei < FR [prete] prêter = ‘lend’. • substitution of voiced bilabial stop for voiced labiodental fricative: subudi < FR [savwar] savoir = ‘know’; bibu < FR [vivr] vivre = ‘live’. • or substitution of /w/for/v /: wiwan< FR [vivã] vivant = ‘living’, weyasu < FR [vwajaʒ] voyage = ‘travel’; weye < FR [vjø, vjej] vieux, vieille = ‘old’.¹⁹ • substitution of dental/s/for the palatal voiceless fricative [ʃ]: seru < FR [ ʃɛr] cher = ‘expensive’; sø < FR [ ʃo] chaud = ‘hot’; surusia < FR [ ʃırürʒjɛ˜]] chirurgien = ‘doctor’. A large variety of lexical items, including almost all numerals, names of days, and of most months, foods, articles of clothing, tools and other objects, feelings, as well as verbs, adverbials and conjunctions are French-derived. French or French Creole also contributed functional items such as articles, conjunctions and discourse markers, as well as phrases such as laluruei ‘king’s throne’ < FR là où (est le) roi ‘where the king is’. Common French loans in contemporary Garifuna are set out in Table 4 (p. 50).

4.2 Grammaticalization A number of established French lexical items acquired grammatical functions and became an integral part of Garifuna morphosyntax when there are no obvious cognates in Arawak or Carib. But there are ambiguous cases, as some items may have survived due to superficial similarities — phonetic, semantic, or both — between the ancestral language(s) and the loan language(s). Some possible cases of grammaticalization include the following: • dan ‘when’ (also danle ‘when in the past’, danme ‘when in the future’) is almost certainly derived from FR [tã] temps ‘time, weather’. This is supported by the fact that dan is used with the same semantic range as temps is in French, that is, meaning ‘time’ as well as ‘weather’ (see examples (4) and (5) above).²⁰

49

50

Geneviève Escure

• ka ‘what,’ a general interrogative, is probably related to FR [kwa], [kø] quoi, que ‘what?,’ as in (1). There is a morpheme ka in Proto-Arawakan which has become ga ‘have’ in Garifuna and has other (non-interrogative) functions. • sa is a polite marker that also carries other pragmatic values, such as impatience as shown in example (1) kaba sa ora lefeda sopa le?’, in which the speaker is exTable 4. French loanwords: numbers and nouns Numbers a gadørø seingø sisi sedø widø nefu disi disi sisi disi seidø disi widø dimí san san milu gadørø hati seingø hati sisi hati sedø hati widø hati

Nouns ‘four’ ‘five’ ‘six’ ‘seven’ ‘eight’ ‘nine’ ‘ten’ ‘sixteen’ ‘seventeen’ ‘eighteen’ ‘fifty’ ‘one hundred’ ‘one thousand’ ‘April’ ‘May ‘June’ ‘July’ ‘August’

< quatre < cinq < six < sept < huit < neuf < dix < dix six < dix sept < dix huit < demi cent < cent < mille < quatre < cinq < six < sept < huit

Verbs, adjectives afurendeira ‘learn’ afeidira/aferedira ‘lose’ afredeira/furedei ‘loan’ blesei ‘hurt’ redei ‘stop’ sandi ‘be sick’ subudi ‘know’ weye, weyo ‘old’ bibu ‘lively’ biniwa ‘blessed’ seru ‘dear’ buin ‘full’ aransera ‘repair’ a

leinudi wandaredi samudi dimasu weyasu seinsu simisi bisledu laj luburuaniga farusu garabali farøga lamuña muisei bungiu letu fiadørø

‘Monday’ ‘Friday’ ‘Saturday’ ‘Sunday’ ‘travel’ ‘money’ ‘shirt’ ‘gun’ ‘garlic’ ‘fog’ ‘umbrella’ ‘wind’ ‘cattle pen’ ‘shade’ ‘humor’ ‘god’ ‘porridge’ ‘limpira’

< lundi < vendredi < samedi < dimanche < voyage < cent sous < chemise < pistolet < l’ail < le brouillard < parasol < rafale < parc < l’ombre < amuser < bon dieu < lait < piastre

Conjunctions, adverbs < apprendre < perdre < preter < blesser < arreter < santé < savoir < vieil < vivant < béni < cher < plein < arranger

murusu meme koma furumien bandi sodon ka sa ba dan

‘a little’ ‘even’ ‘as’ ‘first’ ‘many’ ‘suddenly’ ‘which’ polite form fut./ imp. ‘when’ or ‘time’ or ‘weather’

< morceau < même b < comme < premier < bande < soudain < quoi < ça, c’est < va < temps

Words for months (from April on) are formed by combining the number with hati ‘month’. Taylor (1951: 51–2) claims that meme is not French, but is instead a reduplicated form of the Arawak particle me “indicating futurity or continuity”. There seem to be two types of meme. See my discussion in Section 5.1. below. b


pressing polite eagerness to eat dinner (‘When will the soup — dinner — be (finally) ready to enjoy?’). A possible source is French ça [sa], which is often used as a postposed intensifier (quand ça? ‘when is it?,’ qui ça? ‘who is it?’). Sa often occurs in interrogative sentences, but is not a required question marker. • m-/ma-, the negative prefix, which can occur in preverbal, preadjectival, or prenominal positions, has an Arawak cognate (Matteson 1972: 165, Taylor 1977: 45). The phonetic similarity with FR [pa] ‘pas’ may have reinforced its position as the primary negator, especially since there exists another negative element -ua/uati (from Carib) — that seems to be restricted (in my corpus at least) to emphatic negative contexts. Another negative element ni/nin, and various similar compounds, are clearly borrowed from FR ni . . . ni ‘neither . . . nor’. (6) M-arihi nin usari nin aguri m-arihi-ti nia ba. NEG-see NEG deer NEG aguti NEG-see-PT NEG PT ‘He didn’t see any deer, or aguti, he saw nothing at all.’ [Honduras B16.3] (7) Wagiya m-ei-ti. we NEG-meat-PT ‘We have no meat at all.’ [Honduras B16.1] • le/la [phonetic form varies] has several overlapping aspectual functions. It is a marker of completion in sentence (1) above, and in danle or lidanle ‘when’ (referring to a past event), as in (5). It also functions as a postposed particle that appears to mark structural units, and also carries emphatic value. Surprisingly, Taylor does not refer to le as having any aspectual value, but merely mentions his co-occurrence with dan, assuming a French etymology (l’heure ‘the hour’) for le. It is possible that this item developed since Taylor’s observations. I assume that it could have been derived from the colloquial post-nominal la, which in French has deictic/emphatic value. This function could then be translated into a notion of accomplishment, hence a perfective aspect. There is a sometimes homophonous, but functionally distinct form, affixal l-/le (clitic or suffix derived from Arawak) that is rarely represented phonetically as a variant [la] in its full form. This -l/le has several gender-related (masculine) functions, all relating to 3S: • pre-verbal clitic (l-aramu garabali, lit. ‘he-stop wind = the wind (M) stops/ stopped’) • pre-prepositional (l-abu, lit. ‘he-with’ = ‘with him’, l-uagu lit. ‘he-about’ = ‘about him’) • pre-nominal possessive (l-iri ‘his name’) • post-nominal deictic determiner (weye le ‘that old man’) In the last example, the masculine morpheme is redundant, as it is marked by the final -e morpheme of the root as well by deictic le. Its feminine counterpart is weyo to ‘that old woman’. This masculine form l-/le regularly alternates with feminine singular t-to (tabu ‘with her’ vs. labu ‘with him’; tuagu ‘for her’ vs. luagu ‘for him’), as well as with other persons n-/nu (1S), b-/bu (2S), w-/wa (1P), and h-/ha (2P and 3P).

5

52

Geneviève Escure

Three uses of le/l- are illustrated in (8): the first is the pre-nominal possessive, the second postnominal le is the masculine determiner (weye le), whereas the last le (non-obligatory) is a sentence-final marker, whose primary function is that of a discourse marker. (8) Ka l-iri wey-e le le? what 3M-name old-M M PT? ‘What was this old man’s name?’ [Honduras B.1.4] • bei/bwe is a very common suffix, with intriguing properties. Taylor (1977: 52) says that this item (transcribed by him as bai) includes the perfective affix -a with masculine personal suffix -i, and carries an additional injunctive or purposive function. Suazo (1991: 89) has called this same item a “present marker” or “a copula” — equivalent to Spanish. The ambiguity involved in defining the functions of this particle suggests some extensive development if Taylor’s interpretation is correct. But Aikhenvaldt (1999: 94) refers to the existence of a “quotational -bi suffix” in Arawak, which may be closer to current usage. The frequency and range of bwe in my corpus indicates that it functions like a highlighting particle. Such highlighting particles are common in creoles, and are also found in French. I hypothesize that a contributing factor that favored its development may involve the influence of FR [bjɛ˜] (casual [bε˜]) bien/eh bien ‘well’, which is phonetically and pragmatically similar to Arawak bi. This pragmatic element often modifies constituents, especially in colloquial speech (FR tu as BIEN fermé la porte? ‘are you sure you closed the door?’). In the following samples, I consistently transcribe this phonetically variable element as bwe, and refer to it as a particle (PT): (9) Aba lun-bwe l-iabi l-egeyedagu-ni t-idibu uhasi then 3M-PT 3M-come 3M-go around-PT 3M-tree clearing ‘So what he did, he walked around the clearing.’ [Honduras B.16.5] (10) M-afediru-bwe-ti buga nie-gien? NEG-lose-PT-PT PA there-from? ‘(He) didn’t get lost coming from there?’ [Honduras B.16.4] (11) bia-igi-bwe t-ani two-nose-PT 3F-POSS ‘a double-barreled (gun)’ [Honduras B.16.10] [literally, this translates as a rifle (bisledu = feminine gender) that “possesses two noses”] Ubiquitous and productive, bwe can be used in various positions — post-adverbially, as in dan-bwe ‘there was a time’ (example 4 above), and post-verbally as in n-aranseru-bwe ‘I was indeed ready’ (example 5 above). Regardless of etymological uncertainties, the multifunctionality and flexibility of bwe is reminiscent of patterns generally found in contact and creolized languages.


4.3 Other loans Taylor claims that English has also yielded items at about the same time, but evidence of older English items is relatively scarce in my recorded corpus.²¹ Some possible English loans include fulesei < please, and apakuha < pack (though a French origin cannot be excluded). There is a high frequency of discourse markers, such as well and ok, which compete with the even more common Spanish pero ‘but’ and bueno ‘well’. English items are more frequent in Belize, but primarily include modern items, and many are derived from creole, such as janikeke < Johnny cake. Considering the continuing influence of Spanish and English, the distinction between nonce and established borrowings is meaningful. But it is sometimes difficult to trace the actual source language, considering that English borrowed a large Romance vocabulary (see Table 5). The most frequent Garifuna modals appear to be directly derived from Englishbased creoles (hafu, musu, garawa/giara), and must have been part of the language for a long time, because they occur in Honduras as well as in Belize. Interaction with Creole people on Roatán may explain the occurrence of the creole modals. The case of ga is less clear-cut, as there is an Arawak reflex — the “attributive prefix ga/ka” (Taylor 1977: 41). The creole origin is reinforced by the fact that ga is primarily used as existential ‘there is/are’. Here again, the confluence between an ancestral and a new form in contact may have contributed to the current development. Early Spanish influence resumed after settlement in Honduras, thus all varieties of Garifuna include established Spanish loanwords, but Honduran Garinagu are Table 5. English and creole loanwords English (established)

English (nonce)

creole

fulesei apakuha apinida afalara tanki (niabu)

bolu busu pili wosu miki < mickey rice n beans

bredkain yanikeke musu hafu neks < next ga/gan < gja, gat

‘please’ ‘pack up’ ‘pain’ ‘follow’ ‘thank (you)’

hot n’cold clover shop food anytime container paralyze well

‘bowl’ ‘bus’ ‘pill’ ‘worse’ ‘to mock’

‘ground food’ ‘Johnnycake’²² ‘must’ ‘have to’ ‘this’ ‘have’, existential

gawara / ‘can’ giara < kyan

(discourse marker)

53

54

Geneviève Escure

much more currently involved in massive Spanish borrowing than Belizean Garinagu. This is clearly represented in the presence of non-assimilated items, whether phonetically or morphologically, as well as in extensive code-switching. As indicated above, the Belizean districts (Stann Creek and Belize) that include most Garifuna communities are less likely to be exposed to contact with Spanish speakers, whereas creole influence is predominant, both linguistically and culturally. Spanish loanwords Spanish (established) bagasu ‘cattle’ < vaca gabayu ‘horse’ < caballo gamisa ‘shirt’ < camisa eneru ‘January’ < enero pero ‘but’ < pero bueno ‘well’ < bueno

Spanish (nonce) imaginar ‘imagine’ facilitar ‘facilitate’ tal vez ‘maybe’ alergia ‘allergy’ ameba ‘worm’ consultar ‘consult’

Table 6 provides some summary indication of the relative frequencies of French, English and Spanish borrowings, comparing two sets of texts recorded in the two communities, Punta Gorda (Honduras), and Seine Bight (Belize). In both texts, loans constitute roughly half of the overall morphemes. French items, all established and integrated, are the most numerous, as they constitute about half of all borrowed items. On the other hand, English items occur in somewhat equivalent percentages in the Honduran (23%) and Belizean texts (27%). As expected, in Honduras I found Spanish items to be much more frequent (30% vs. 6% in Belize). Table 6. Lexical loans in Belizean and Honduran Garifuna Source language

Honduras

(N = 120)

Belize

(N = 301)

French English Spanish

41.6% 23.0% 30.0%

(50) (32) (38)

63.1% 27.2% 6.3%

(190) (82) (19)

4.4 The morphology of established and nonce loans Nonce items (in bold in the examples below) are not integrated to Garifuna morphology. Thus the Spanish-derived verb konsultar (SP consultar ‘to consult’) in (12) is produced in the Spanish infinitive form, rather than in the expected Garifuna form that would contain the personal prefix n-/nu ‘I’. Konsultar is preceded by the creole modal mosu ‘must’ and the irrealis-optative hamuga (whose origin is unclear). Konsultar is, however, followed by the evidential particle -ne.


(12) Mosu hamuga konsultar-ne h-uma le l-ana ubou l-un MOD IRR consult-PT they-to who 3M-POSS village 3M-to gayara-laan h-alugu n-un ederere-ha musu fulasu n-un MOD-PT they-sell I-to enjoy-PT piece place I-to ‘I would need to consult with people from the village to see if they can sell me a piece of land for me to enjoy.’ [Honduras M.14.7] On the other hand, the two English items in (13) apinidu and paralaiz appear to be well integrated, as they are combined with pronominal clitics, and with the nominalizer -ni: (13) Aba h-apinidu-ni h-ibiri paralaiz-hama larig ien me-ha sun søgø so they-pain-PT others paralyze-they after PA-PT some pass ‘So they suffered great pain, others were paralyzed afterwards, and some passed away.’ [Belize O. 11. 1–2]

5. Linguistic processes in language attrition It is well-known that dominant languages are largely responsible for the obsolescence of contiguous weaker varieties. According to studies of dying languages, Gaelic and Dyirbal reduced productivity of their affixes such as gender, case, and tense. For example, Gaelic lost future-conditional suffixes (Dorian 1981), and Dyirbal lost ergativity, case markers and its noun classification system (Schmidt 1985). Syntax is also affected by the loss of affixes, which are typically replaced by periphrastic forms. In the case of Garifuna, I hypothesize that contact with the Englishbased creole, which functions as lingua franca in Belize, and with Spanish in Honduras, may have led to morphological simplication in Garifuna. Previous descriptions of Garifuna qualify it as a synthetic language, with extremely rich affixation. Tense, modality, aspect, case, gender, person, and number markers may occur as prefixes or suffixes. Yet the contemporary language that I recorded in casual contexts displays a morphology that is much reduced when compared to Taylor’s description, the most extensive to date. Although agglutination occurs, as shown in (14), there is a clear trend in the language (a) away from synthetic morphology, and (b) toward increasing analytic morphosyntax (with preverbal morphology). Example (14) illustrates a complex sequence bou-lumuti hamuga (lumuti is a synthetic combination often attached to a verb) that marks a combination of counterfactual and irrealis aspect.

55

56

Geneviève Escure

(14) Añihen hamuga le ma-chutau l-au garinagu if IRR PT NEG-smart 3M-about Garinagu bou-l-umu-ti hamuga Salvador Honduras destroy-3M-AUX-3M IRR Salvador Honduras ‘Had the Garinagu not been smart [during the war], El Salvador would have destroyed Honduras.’ [Belize; 11/2000] Hamuga is an irrealis morpheme that can modify a verbal unit. In (14) above, the first hamuga — combined with a completive suffix — precedes a negative adjectival verb hamuga-le ma-chutau (lit.‘would-finish-not-smart = had not been smart’). Hamuga can also combine with other elements, as represented in the sequence lumuti hamuga. L+umu+ti is a compound including the counterfactual auxiliary -umu- combined with two clitics, a masculine singular subject pronominal prefix (l-) referring to ‘El Salvador’ and a masculine singular object specific pronominal suffix (-ti), referring to ‘Honduras’. Note that this group occurs after the verb bou ‘destroy’, but a more common combination — also grammatical, according to native speakers — is: Salvador lumuti hamuga bou Honduras. lit. ‘El Salvador, it could have, but didn’t do this, destroy Honduras.’ However, the speaker fronts the verb for emphasis, and moves the subject in postverbal position through some kind of right dislocation, since the subject clitic l- referring to “El Salvador” is prefixed to the auxiliary l-umu-ti. The type of complex synthetic morphology that is illustrated in (14) does not occur with high frequency in my corpus (Escure ms. further discusses this phenomenon). Based on preliminary observations, both Belizean and Honduran Garifuna varieties appear to mark aspect and tense with limited agglutinative morphology and a few regular, usually preverbal morphemes. To illustrate this trend, I will now focus on two items that are particularly noticeable in recent varieties because of their frequency as well as their syntactic behavior. I will briefly evaluate whether their occurrence is indicative of the varying influence of dominant languages on Garifuna attrition. The two morphemes in question are me and buga, both now primarily past tense/aspect markers.

5.1 Me (and variants meha and meti) Taylor (1977: 49) noted the occasional occurrence of the “syntactic tense markers future me and past buga” in the Hopkins variety of Belizean Garifuna. Cayetano (1993) and Stochl (1975: xii) also mention the presence of a future marking suffix -me besides another future marker ba that also functions as imperative auxiliary. Although -me is occasionally suffixed in a future tense context, I found this function to be restricted to co-occurrence with the temporal conjunction dan (danme ‘when’ in the future), both in Belize and Honduras. And in this case danme contrasts with danle (‘when’ in the past). However, me occurs widely as a preverbal morpheme


in a past context in my Belizean corpus, though not in Honduras. Stochl (1975: xii) also mentions that the variant meha can occur as “remote past” in Belize, and Taylor alludes briefly to uses of me that are not clearly identifiable as marking future tense. His interpretation is that me can be used to indicate a “change in the speaker’s condition” (Taylor 1977: 66). The Belizean data collected so far confirm the regular use of me as a past morpheme marker, but not as denoting a particularly remote past, except when it is repeated, as shown in (15). In this sentence, Jo refers to a situation that existed thirty years ago (he refers to a visit to Miami in the 1970s): (15) Lidan-me dan ligia bulu-to me-me-ha huhune polisman karø grin time-PA times those blue-F PA-PA-PT vehicle policeman car green toubaø stripe ‘In those days the police cars (in Miami, Florida) were blue with green stripes.’ [Belize J.1.6] Me is often combined with the particles -ha or -ti. Both occur widely as suffixes; -ha typically marks verbal intensification or repetition, as in adarara ‘to open’ vs. adaraha ‘to open repeatedly’ while -ti can be suffixed to practically any item. Breton (Adam: 1877: xx, xxi) called it a “positive particle”, and indeed it provides an aspectual notion of achievement or completeness, independent of tense. Both meha and meti, when suffixed to some other element, can occur in pre-verbal position (lines 1, 4, 6, 8), as well as in pre-nominal position (lines 7, 11), as illustrated in the short text in (16): (16) OLD GARIFUNA LIFE [Olga tells how some Garinagu peasants used to survive] Dan-bwei me-ha h-eidin mutu ouchaha kalifuna time-PT PA-PT they-go people fish Kalifuna l-un-ti h-eidin kalifuna M-to-PT they-go Kalifuna mosu ha-huñuragu-n lidan guyara must they-cross-PT in canoe ingiña me-ti h-eibuga nefu maili then PA-PT they-walk nine mile l-un-ti h-asøgøru-n lidan ichari le M-to-PT they-pass-PT in plantation PT larigien me-ti ha-bunagua gain bu-nagua baruru then PA-PT they-plant cassava you-plant plantain h-eibuga ya me-ti nefu maili harigoun they-walk there PA-PT nine mile backward luba ya me-ti h-ayabian tidan guyara also there PA-PT they-went F-in dorey

57

58

Geneviève Escure

l-un aguyu niha buri h-achøløru guñou M-to home there dark they-arrive evening h-un h-asanigu aba ha-pinidu-n they-to their-children so they-suffer-PT hibiri paralaiz-hama larigien me-ha sun søgø. others paralyzed-they after PA-PT some pass ‘A long time ago, people would go fishing in Kalifuna. In order to go to Kalifuna, they had to go across in a canoe, then they walked nine miles, until they had passed by a plantation, then they planted cassava and plantain. Afterwards, they would walk back nine miles and take a dorey [to get home]. They would get home at dark to their children. [Many] were in great pain [because of the hardships of the journey], others were paralyzed afterwards, and some passed away.’ [Belize O. 11. 1–2] In this short text, meha occurs twice, and meti four times. There is no obvious functional difference between the two items. However, one might tentatively refer to meha as a marker of habitual past, as Olga recounts the traditional rural Garinagu’s lifestyle, and to meti as a marker of sequential past actions, with a focus on change of state, as characters are presented in their successful farming activities and hardships. This interpretation is supported in other Belizean texts, which for space limitations cannot be cited here. The prevalence of me as past morpheme in Belize may well be due to the influence of creole me, a past and sometimes anterior aspectual marker (derived from English been) that occurs with great frequency in Belize as illustrated in (17), also discussed in Escure (1999, 2004): (17) Toni kaal mi tel mi dei me gat tu pipl we me wan kum in; Toni call me tell me they PA get two people that PA will come in a tel im a me gat dis trip lain-op fu tode eniwei I tell him I PA get this trip lined-up for today anyway ‘Toni called me to tell me that they had two people who might come in; I told him that I got this trip lined up for today anyway.’ [Belize 1998, ms.] Since Belizean Creole is the most widely used variety in Belize, the superficial similarity between creole me and Garifuna me is likely to have led to the current practice. It is thus not surprising to see me shifting from a future to a past reference marker.

5.2 Buga The Honduran texts collected to date do not show evidence of the use of me as past tense morpheme. The Garifuna used in Punta Gorda appears to favor instead the morpheme buga as a marker of past tense. This item is called a “recent past” by Stochl (1975), in reference to Belizean Garifuna, but it rarely occurs in my Belizean corpus. The use of buga in Honduras is illustrated in example (18):


(18) THE OLD HUNTER [Bobo tells Maria the story of an old man who went hunting for game in the bush: M. = Maria; B. = Bobo] M. Kaba-ti buga ei? what-PT PA meat B. L-erenga-ti weye mutu le he-say-PT old man this “N-eiga-ya-ti ededøgø uguñe bueno pero I-eat-here-PT meatless today well but n-agaragua n-arigoun dabiarug-un-adina I-return I-back jungle-to-I dise-dina ñei-hun.” far-I here-from M. Ma-fediru-bwe-ti buga ñei-gie? NEG-lose-PT-PT PA there-from B. Ma-fediru-bwe-ti. NEG-lose-PT-PT M. L-afara-ti buga aba hadagien megu ha? he-kill-PT PA then some monkey PT ‘What about the meat? — The old man said: “I’ll eat a meatless meal again today, well, so I’ll go back more deeply into the bush.” — Didn’t he get lost there? — He didn’t get lost. — Did he kill some monkeys?’ [Honduras #16, 15–16] Buga appears to be an original item since it appears frequently in Breton’s 1667 grammar (he refers to it as a “perfect” — un parfait). The 17th-century function has thus survived in Honduras. On the other hand, buga is rarely used as a simple past marker in Belize, probably because Belizean Garifuna has already borrowed a simple past preverbal morpheme from Belizean Creole. The two Garifuna varieties use similar strategies, but different morphemes, in their shift toward a simpler morphological system. In Belizean Garifuna, meha and meti represent a shift from “future” to “past” meaning due to the probable influence of a homophonous creole morpheme. Honduran varieties, on the other hand, preserved the older morpheme meaning “in the past”, now giving it a temporal value.

5.3 Language obsolescence and pidginization It has been shown that Garifuna has massively borrowed and fully integrated about half of its lexicon, with concomitant grammaticalization in some cases. There is also evidence of increasing analytic, preverbal morphology (at least in the case of me and buga). In conclusion, Garifuna appears to have acquired two regular past morphemes that are preferred to the more complex auxiliary-affix system of what I assume was the former Garifuna morphological structure.

59

60

Geneviève Escure

In situations of language contact, analytical morphology is generally preferred to synthetic morphology. If such morphological shifts are supported by a more complete analysis, they are indicative of language change. Some synthetic structures are replaced by others that are borrowed from the dominant language (here primarily creole, or Spanish), resulting in analytic formations. But do such changes signal decay or innovation? Probably both. As Garifuna is losing speakers, those individuals who still use it (albeit only sporadically so) create new strategies and naturally rely on live models with which they are more familiar. Ambiguity and multifunctionality are typical of language contact, and are amply documented in pidginization and creolization. It is thus not surprising to find that they occur during attrition. A language typically recedes under the pressure of a dominant language that has acquired prestige, and it does so by massively incorporating elements of the donor language. In the end, this may lead to language obsolescence and eventual language death. One should not assume, however, that there are linguistic processes that are unique to language obsolescence. Language attrition is implemented through interference and code alternation that may also occur in situations not necessarily leading to language death. Dressler points out that “interference is not a sufficient criterion for decay, since it can be reversed” (1988: 184). Several phenomena have been identified as indicators of the moribundity of a language. These phenomena include massive lexical loans from the dominant language, substitution of borrowed grammatical morphemes for synonymous indigenous elements, attrition of word-formation rules, decay of inflectional morphology, replacement of synthetic by analytic constructions, loss of syntactic subordination, and more generally, maximization of transparency (Dorian 1981, Dressler 1988: 185–7). Garifuna clearly shows evidence of many such characteristics. Massive population migrations that result in the loss of native varieties are sometimes concomitant with the development of new varieties (pidgins and creoles). Parallels have been established between language death and pidginization or creolization. McMahon (1994: 284), for instance, writes that “language death resembles pidginization and creolization” (1994: 284). Language death may lead to language birth, as language contact precipitates the transfer of linguistic units from one system to another.

6. Conclusion This article has presented a broad overview of the current state of Garifuna, and also attempted to explain a specific aspect of the process of language attrition through contact with dominant languages, namely the functioning of past marking. Earlier claims that Garifuna is primarily Arawak may hold true. But other linguistic forces (mainly language contact) have contributed to making Garifuna in Central America vastly different from its ancestral roots in South America. Garifuna speakers have integrated numerous elements — both lexical and grammat-


ical — that they accepted from other systems in the course of their peripatetic peregrinations over the last millennium. Some features of contemporary Garifuna suggest that the language is developing in a manner reminiscent of pidginization or creolization. I have here suggested that changes within the verb phrase in particular resemble the TMA marking in pidgins and creoles. My discussion has stopped short of the claim that Garifuna is a creole. Indeed, Garifuna still exhibits a morphology that is richer than that of any pidgin or creole. The Garifuna data presented in this study do exhibit some morphological complexity: grammatical gender in nouns, number suffixing — though restricted to animate nouns — number and person prefixes, as well as suffixes that are systematically linked to various grammatical categories. Contemporary Garifuna thus continues to make active use of a fairly complex synthetic grammar while simultaneously employing (more) recent analytic structures. These observed general patterns are concomitant with the impending loss of the language, now spoken fluently — in one form or another — only by those over 50 years of age.

Notes * This research was funded by an NEH Fellowship (FA-36507–01), and a University of Minnesota Graduate School Grant-in-aid. The support provided by both is gratefully acknowledged. I also want to thank Armin Schwegler, John Singler and three anonymous reviewers for their constructive comments. I am especially indebted to my Garifuna friends in Belize and Honduras, who welcomed me into their homes, and patiently endured lengthy recording sessions. I am particularly grateful for assistance provided by Joe Williams and Nattie Lopez in Seine Bight (Belize), and by Celso Zapata, Reina Martinez, and Marla Jean Martinez in Punta Gorda (Roatán, Bay Islands of Honduras). 1. Witness this statement from the proceedings of a Garifuna language workshop held a little more than a decade ago in Belize: “I am very concerned about . . . the Garifuna people as we irreversibly lose our language and replace it with Creole.” (S. Cayetano 1990: 155). 2. Garinagu is the plural form that properly refers to speakers of Garifuna (singular form). The suffix -gu is a (Carib-derived) plural morpheme restricted to human beings — as in umada ‘friend’ > umadagu ‘friends’, or uguchu ‘mother’ > uguchunigu ‘mothers’. This plural formation also applies to loanwords, as in sudara/sudaragu ‘soldier’ [from French or English], or meisturu/meisturugu ‘teacher’ [from English, French or Spanish]. 3. In Table 1, I assume an intermediate figure of 200,000. 4. Guatemala has just recognized Garifuna as one of its official languages (Anita Herzfeld, p.c.). 5. Honduran Amerindians live in separate communities in the interior. They comprise seven ethnic groups — Lenca, Miskito-Zambo, Tolupan-Xicaque, Maya-Chorti, Pech, Twahka, and Nahoa — and range in number from 400,000 (Lenca) to 816 (Nahoa). Overall, they represent 9% of the Honduran population. 6. Other ethnic groups besides the Maya-Kekchi include East-Indians (who speak Belizean Creole), Mennonite (who speak German), and a few others (Chinese, and Middle Eastern), who mostly speak the creole as lingua franca as well.

6

62

Geneviève Escure 7. Interactions were of the following types, in chronological order: Arawak/Carib, African/ Red Carib, French and French Creole/Red Carib/Black Carib, English/Black Carib, Spanish/ Black Carib, Belizean Creole/Garifuna, and Spanish/Garifuna. 8. Garifuna is often cited as the prime example of a language that includes distinct men’s and women’s forms. Such putative differences, as described by Breton, are now extinct. Older Garinagu are aware of the existence of separate vocabularies, although they use either item indifferently, often preferring the Arawak (so-called “female”) form. If they are aware of any difference, individuals typically make contradictory statements as to what is the male and the female form. Here is a small sample of those traditional distinctions (data come from Breton 1666 and Taylor 1951; I also elicited information through such questions as “do women/men use a different forms for ‘item X’?”): Female (Arawak) nuguya hianru eyeri eyeri irahø hianru irahø

Male (Carib) au wøri wøgøri wøgøri irahø wøri irahø

Gloss ‘I’ ‘woman’ ‘man’ ‘boy, son’ ‘girl, daughter’

Female (Arawak) nui nugune

Male (Carib) yudi yagane

Gloss ‘my meat’ ‘my boat’

Related evidence occurs in the title song of the popular Garifuna song Nuguya Garifuna = ‘I (am) Garifuna’, sung by (male) Belizean singer Aziatic, who thus uses the traditional female pronoun. 9. One young Garifuna man (age 17) in Seine Bight spontaneously told me in Creole: “a shiem a no spik Garifuna” (‘I’m ashamed that I can’t speak Garifuna’). Bonner (2001) makes similar observations in the town of Dangriga, a small town of Belize, though she assumes — erroneously so — that Garifuna is still widely spoken by children in rural villages. 10. The primary informant of this dictionary was Roman Zuniga (now deceased). He taught me Garifuna in the early 1980s. Mr. Zuniga, a powerful advocate of Garifuna, was already very aware of the waning of the language. 11. Dixon and Aikhenvald (1999) and Taylor (1977) represent a high central vowel (barred ‘i’) but the Garifuna vowel appears to be closer to a mid front unrounded vowel, that I hereafter represent as [ø]. The back rounded vowel fluctuates between [u] and [o]. 12. Discussion of the putative occurrence of ergativity in Garifuna must be left to future research. However, when ergativity occurs, it seems to have a somewhat restricted range. 13. Silibu is not recognized now by my Honduran respondents, who tell me that they use Spanish plata instead. They also tell me that aufuri is a term that they no longer use. This confirms the obsolescence of Carib items (aufuri) and the replacement of older loanwords like silibi by Spanish reflexes. 14. Garifuna has a regular stress pattern: on the first syllable of a bisyllabic item, and on the second syllable of longer words. But many lexical items are stressed otherwise. Suprasegmentals are not indicated or discussed in this paper. 15. Abbreviations are as follows: Source language: A = Arawak; C = Carib; SP = Spanish; FR = French; EN = English; CR = Creole; WA = West African. Some Carib derivations may not

Garifuna in Belize and Honduras have been assessed yet. Grammatical function: M = masculine marker; F = feminine marker; NEG = negative; PT = particle; PA = past marker; FUT = future. Verb-final morphology cannot be fully discussed here, as it is subject to extensive variation. Some common verbal suffixes illustrated in the texts include -ha, -ra (intensive, or repetitive), -gøda (causative), -ba (future, imperative), -ti, -me, -bwe, -ne (intensive particles). 16. Each example is associated with four lines: (1) text; (2) source language (provenance is discussed later in the text), (3) literal translation, with some morphological information; and (4) translation. Note that no source language is given for ma in (1) since it is a universal word. 17. To refer to a year (in this case “1979”), Garifuna regularizes the use of large French numbers (here “70”) on the model of “80” — using a multiplication system: 3 (times) 20 = 60, plus 10, plus 9, thus resulting in: 10 – 9 – 3 – 20 – 10 – 9 = 10–9–3 × 20–10–9 = 1979. 18. Vowel nasalization occurs, but contrary to French, it appears to be a strictly phonetic process. 19. There is an Arawak cognate uaiha (Taylor 1977: 141). The phonetic similarity may have contributed to the selection of this item. 20. It is a natural extension to grammaticalize this item to the temporal conjunction (the expression FR au/du temps = ‘at the time’ can also be used as an equivalent of quand = ‘when’). 21. For example fagøsi < fox — included in R. Cayetano’s 1993 dictionary — does not occur in my corpus, probably because this kind of animal is not as common locally as gasigamu ‘armadillo’, gibinadu ‘gibnut’, aguti ‘wild rabbit’ gegeu ‘peccary and megu ‘monkey’ — all Arawak or Carib words. 22. A type of flat bread that remains edible for several days.

References Adam, L. 1877.“Introduction.” Nouvelle édition de la grammaire caraïbe du Père Raymond Breton [1967] publiée par L. Adam and C. A. Leclerc, ix–xxviii. Paris: J. Maisonneuve. Adam, L. 1893. Grammaire comparée des dialectes de la famille caribe. Paris: J. Maisonneuve. Aikhenvald, A. 1999. “The Arawak language family.” In Dixon and Aikhenvald, 65–106. Bonner, D. 2001.“Garifuna children’s language shame: Ethnic stereotypes, national affiliation, and transnational immigration as factors in language choice in southern Belize.” Language in Society 30: 81–96. Breton, R. 1665. Dictionnaire caraïbe-françois meslé de quantité de remarques historiques pour l’éclaircissement de la langue, composé par le R. P. Raymond Breton, religieux de l’ordre des Frères Prescheurs, et l’un des premiers missionaires apostoliques en l’isle de la Gardeloupe et autres circonvoisines de l’Amérique. Auxerre: Gilles Bouquet, imprimeur ordinaire du ROY. Breton, R. 1666. Dictionnaire françois-caraïbe. Auxerre: Gilles Bouquet, imprimeur ordinaire du ROY. Breton, R. 1667. Grammaire caraïbe. Auxerre: Gilles Bouquet, imprimeur ordinaire du ROY. Breton, R. 1999. Dictionnaire caraïbe-français. Nouvelle édition, présentée et annotée par le CELIA et le GEREC. Paris: Khartala et IRD. Cayetano, R. 1993. The People’s Garifuna Dictionary. Dangriga: The National Garifuna Council of Belize.

63

64

Geneviève Escure Cayetano, S. 1990. Garifuna History, Language and Culture. [No publisher]. Coelho, R. 1955. The Black Carib of Honduras: A Study of Acculturation. PhD thesis. Evanston: Northwestern University. Reprinted as Los negros caribes de Honduras. Tegucigalpa: Editorial Guaymaras (1981). CONPAH. 1998. Confederación de pueblos autoctónos de Honduras. Tegucigalpa: Honduras. Conzemius, E. 1928. “Ethnographic notes on the Black Carib (Garif).” American Anthropology 30: 183–205. Crystal, D. 2000. Language Death. Cambridge: Cambridge University Press. Derbyshire, D. 1999. “Carib.” In Dixon and Aikhenvald, 23–64. Dixon, R. M.W. and A. Aikhenvald (eds.). 1999. The Amazonian Languages. Cambridge: Cambridge University Press. Dorian, N. 1981. “Language Death: The Life Cycle of a Scottish Gaelic Dialect.” Philadelphia: University of Pennsylvania Press. Dressler, W. 1988. “Language death.” In Linguistics: The Cambridge Survey, F. J. Newmeyer (ed.), vol. 4: 184–92. Cambridge: Cambridge University Press. Escure, G. 1979. “Linguistic variation and ethnic interaction in Belize: Creole/ Carib.” In Language and Ethnic Relations, H. Giles and B. Saint-Jacques (eds.), 101–16. Oxford: Pergamon Press. Escure, G 1982. “Contrastive patterns of intragroup and intergroup interaction in the creole continuum of Belize.” Language in Society 11: 239–64. Escure, G. 1984. “The acquisition of creole by urban and rural Black Caribs in Belize.” In York Papers in Linguistics 11: Urban creoles, M. Sebba and L. Todd (eds.), 95–106. York: University of York. Escure, G. 1997. Creole and Dialect Continua: Standard Acquisition Processes in Belize and China (PRC). Amsterdam: John Benjamins. Escure, G. 1999. “The grammaticalization of past in creoles.” American Speech 74: 165–202. Escure, G. 2004. “Belize and other Central American varieties.” In A Handbook of Varieties of English: The Americas and the Caribbean, E. Schneider (ed.), 315–42. Berlin: Mouton. Escure, G. Ms. “From synthetic to analytic morphology in Garifuna.” Fleming, I. 1965. “Carib.” In Languages of Guatemala, M. Mayers (ed.), 303–8. The Hague: Mouton. Gonzales, N., 1988. Sojourners of the Caribbean: Ethnogenesis and Ethnocentricity of the Garifuna. Urbana: University of Illinois Press. Gullick, C. J. M. R. 1976. Exiled from St. Vincent: The Development of Black Carib Culture in Central America up to 1945. Malta: Progress Press. Hoff, B. F. 1968. The Carib Language. The Hague: Martinus Nijhoff. Kerns, V. 1989. Women and the Ancestors: Black Carib Kinship and Ritual. Urbana: University of Illinois Press. Laman, K. 1964 (1936). Dictionnaire kikongo-français (2 vols.). Ridgewood, NJ: The Gregg Press. Matteson, E. 1972. “Proto-Arawakan.” In Comparative Studies in Amerindian Languages, E. Matteson, F. Christian, F. Jackson, N. Waltz, and A. Wheeler (eds.), 160–242. The Hague: Mouton. McMahon, A. 1994. Understanding Language Change. Cambridge: Cambridge University Press. Merrill,T. (ed.). 1995. Honduras: A Country Study. Washington, D. C.: Library of Congress. Noble, K. 1965. “Proto-Arawakan and its descendants.” International Journal of American Linguistics 31: 1–128.

Garifuna in Belize and Honduras Pelleprat, P. 1656. “Introduction à la langue des Galibis.” In Relation des missions des pp. de la Compagnie de Jésus dans les isles, et dans la terre ferme de l’Amérique méridionale. Paris [no publisher]. Pury, S. de. 2000. “Le garifuna, une langue mixte” (unpublished manuscript). Renault-Lescure, O. 1999. “Le caribe insulaire, langue arawak: Un imbroglio linguistique.” In Breton 1999: XLVII–LXXXVIII. Rivas, R. 1993. Pueblas indígenas y garífuna de Honduras. Tegucigalpa: Editorial Guaymuras. Schmidt, A. 1985. Young People’s Dyirbal: An Example of Language Death from Australia. Cambridge: Cambridge University Press. Stochl, J. 1975. A dictionary of Central American Carib (revised and enlarged by Richard Hadel with the assistance of Roman Zuniga). Belize: BISRA. Suazo, S. 1991. Conversemos en garifuna. Tegucigalpa: Editorial Guaymuras. Taylor, D. 1951. The Black Carib of British Honduras. New York: Wenner-Gren Foundation. Taylor, D. 1955. “Phonemes of the Hopkins (British Honduras) dialect of Island-Carib.” International Journal of American Linguistics 21: 233–41. Taylor, D. 1956a. “Island-Carib, II: Word classes, affixes, verbs, nouns.” International Journal of American Linguistics 22: 1–44. Taylor, D. 1956b. “Island-Carib, III: Locators, particles.” International Journal of American Linguistics 22: 138–50. Taylor, D. 1958. “Island-Carib, IV: Syntactic notes, texts.” International Journal of American Linguistics 24: 36–60. Taylor, D. 1962. “Lexical borrowing in Island-Carib.” Romance Philology 16: 143–52. Taylor, D. 1977. Languages of the West Indies. Baltimore: Johns Hopkins University Press. Young, Sir W. 1795. “Account of the Black Charaibs of St. Vincent with the Charaib Treaty of 1779 and other original documents.” In Journal of a Voyage Undertaken in 1792, W. Young and B. Edwards (eds.), 259–301. London [no publisher].

65

CHAPTER 4

The Nova Scotia–Sierra Leone connection New evidence on an early variety of African American Vernacular English in the diaspora Magnus Huber

University of Regensburg

In an attempt to account for the linguistic similarities between Gullah and Krio, this article investigates the historical connection between the US and Sierra Leone. The research question is whether earlier varieties of African American Vernacular English and/or Gullah could have influenced Krio, the creole language spoken in Sierra Leone. In order to provide an answer to this question, the points of origin on the American mainland of over half of the 1,196 ex-slaves who left Nova Scotia for Sierra Leone in 1792 are identified. The finding is that about 85% of the “Nova Scotian Settlers” originally came from the South Atlantic states of Virginia, North and South Carolina, and Georgia. Virginia and South Carolina provided the bulk of the settlers — about half and a quarter, respectively. Then, an analysis of the historical and socio-statistical background of the South Atlantic states in the late 18th century provides some insight into the kind of language restructuring that was taking place in this area. The hypothesis that the Nova Scotians may have spoken an early form of AAVE or Gullah is thereby addressed as well. This background information is then used to refine the analysis of a corpus of letters written by the Nova Scotians. Although the linguistic analysis of these letters is hampered by the small number of tokens, they nevertheless provide useful insights into the nature of the Nova Scotians’ speech. It is concluded that the emigration of ex-slaves to Sierra Leone may explain the similarities between Gullah and Krio.

1. Introduction¹ Late in 1791, the agent of the Sierra Leone Company, John Clarkson, toured Nova Scotia (NS) to collect African Americans (former slaves who had fought on the British side in the American Revolution) willing to be transferred to Sierra Leone. His proposal of a free settlement in Africa was met with great enthusiasm by the Black community in Birchtown (near today’s Shelbourne), and over 550 individuals

68

Magnus Huber

volunteered for emigration. Clarkson was at great pains to point out the dangers and risks that the founding of a new colony entailed. On November 2, 1791, he entered the following episode in his diary: Took down the names of a man conditionally as before, he came originally from the coast of Africa, and spoke English indifferently; the following dialogue passed between us. “Well my friend, I suppose you are thoroughly acquainted with the nature of the proposals offered to you by His Majesty.” “No Massa me no hear, nor no mind, me work like slave cannot do worse Massa in any part of the world, therefore am determined to go with you Massa, if you please.” “You must consider that this [is] a new Settlement, and should you keep your health must expect to meet with many difficulties, if you engage in it.” “Me well know that Massa, me can work much, and care not for climate; if me die, me die, had rather die in me own Countrey than this cold place.” (Clarkson 1791–1792: 1791–11–02)

The above is a clear indication that at least one of the emigrants spoke restructured English. The present article sets out (a) to investigate the historical link between the New World and Sierra Leone as constituted by the NS Settlers, and (b) to explore whether earlier varieties of African American Vernacular English (AAVE) and/or Gullah influenced Krio, the creole spoken in Sierra Leone. While investigating the degree of affinity between 13 English-lexicon pidgins and creoles world-wide, Baker and Huber (2001) found that Gullah and Krio rank third-highest among 21 possible pairings of seven varieties in the Atlantic, second only to the closely-related Krio and West African Pidgin English (see for instance Huber 1999: 75–134), and St. Kitts Creole and Bajan (cp. Baker 1999). The varieties considered were Gullah (GUL in Figure 1 below), Jamaican Creole (JAM), St. Kitts Creole (SKI), Bajan (BJN), the Surinam varieties (SRN), Krio (KRI), and West African Pidgin Englishes (WAF). The calculations were based on 248 phonological, morphological, and lexical features. Figure 1 shows the strongest affinities between

GUL 10.9

17.8 11.3

19.6

JAM

10.9 17.4

KRI

33.5

WAF

11.9

11.3

SKI 27.1

BJN SRN

Figure 1. Degrees of affinity between 7 English-lexicon varieties in the Atlantic (adapted from Baker and Huber 2001: 185)

The Nova Scotia–Sierra Leone connection

the seven varieties, the figures and thickness of the connecting lines indicating the excess of features that pairs share in addition to those that are statistically expected. As can be seen, GUL-KRI, in addition to being the third-highest affinity, also represents the strongest cross-Atlantic link. This invites the question of how the similarities between these two creoles arose. In trying to explain the linguistic similarities between Gullah and Krio, this article provides some sociolinguistic background to a recently discovered diaspora variety of early AAVE, the language of the Nova Scotians who emigrated to Sierra Leone in 1792. Having left the southern states in the 1770s and 1780s, this group of African Americans was in fact among the earliest to emigrate from America. In contrast to other diaspora communities that still form enclaves in majority societies from which they differ both in culture and in language, the NS Settlers were assimilated into Sierra Leonean Krio society, formed by several settler groups in the 19th century (see Section 2, below). They are not, therefore, any longer discernable as a linguistically distinct group (other than that they are Krios, as opposed to indigenous Sierra Leoneans). The only direct evidence we have of the Nova Scotians’ language consists of a series of petitions and letters written between 1791 and 1800 and published in Fyfe (1991). These are among the earliest first-hand documents of the language of African Americans in the diaspora, and as such they constitute precious documentation of AAVE in the second half of the 18th century. However, the small size of the corpus (some 12,000 words), and the fact that it is formal written language somewhat limits its usefulness. Studies on surviving diaspora varieties are not hampered by these restrictions, but they are faced with the problem that their object of investigation, though generally believed to be more conservative than mainland AAVE, is the result of some 200 years of independent language history. This means that they, like the NS Settlers’ first-hand written data, present only indirect evidence in the reconstruction of early AAVE (see Montgomery 1999: 3–5 for further discussion). The Nova Scotians’ letters have already been used in the reconstruction of earlier stages of AAVE and in the debate surrounding its origins (Montgomery 1999, van Herk 1998, 1999a, 1999b, 1999c). But this article provides information on the geographical and social origin of the NS Settlers, and applies this evidence in a new linguistic analysis of the letters. However, since the Gullah-Krio connection is the primary focus of this article, I will not enter into the debate about the origins of Gullah (developed in situ, introduced from the Caribbean or ultimately even from West Africa?), nor will I discuss more than marginally (in Section 3) its diachronic relationship with AAVE (has Gullah always been a linguistic isolate, or was all AAVE once like Gullah?).² Instead of assuming that present-day AAVE is a decreolized descendant of a once more widespread Gullah, I accept the emerging consensus that, in the second half of the 18th century, variously creolized varieties existed on the Sea Islands and in the Low Country of South Carolina and Georgia, stretching about 50 to 60 km inland.

69

70

Magnus Huber

2. Historical background In the American Revolutionary War of 1775–1783, the British promised freedom to those slaves that crossed lines and fought against the Americans. In the course of the war, thousands of slaves, particularly from the southern states of Virginia and South Carolina, fled their masters in hope of a better future. When the war was lost, the British evacuated many of these African Americans along with the White Loyalists. There were several evacuation points, but the main Loyalist center was New York (cf. PRO 6/561: 407, Troxler 1974: 46, 97, Walker 1976: 8–10). In their negotiations of the modalities of the evacuation, the commanders-inchief of the British and the Americans, Carleton and Washington, agreed to record the particulars of the former slaves who were being evacuated from New York. Thus, in the course of 1783, 2,832 African Americans were recorded in the so-called Book of Negroes (1783, Ruch 1996), which provides information on individuals, including the following categories: Negro’s Names, Age, Claimants: Names + Places of Residence, and Remarks, including the residence of the former master. About 90% of the Black Loyalists leaving New York in 1783 went to Nova Scotia and New Brunswick in Canada, where they faced extreme conditions: they were unaccustomed to the harsh climate, they knew little about farming the poor soil, and many waited in vain for the support they had been promised. The abolitionist Sierra Leone Company in London heard of the problems that beset the Black Loyalists in the maritime provinces of Canada and sent their agent John Clarkson over to propose emigration to Sierra Leone. From October to December 1791, Clarkson gathered 1,196 Black Loyalists and accompanied them to Africa in January 1792. Of these, 1,131 set foot on the Sierra Leone peninsula in February and March 1792. They constituted the first major population group in the history of that settlement, as the following overview shows (for more details, see Huber 1999: 59–69, 2000: 276–7): 1786: 377 “Original Settlers” from England. By 1791, their number had dwindled to 64. 1792: 1,131 “Nova Scotians” 1800: 525 Jamaican Maroons (via Nova Scotia) 1808–1863: Liberated Africans (60,000+ from West Africa)

3. The origin of the “Nova Scotians” in Sierra Leone The extent to which different population groups contributed to the genesis of Krio is disputed. The various scenarios that have been proposed follow two main lines of argumentation. Some scholars maintain that the formation of the language essentially took place in Africa. In this scenario, present-day Krio is seen as the descendant of a hypothesized Upper Guinea Coast Creole English (the purported ancestor


of all Atlantic creoles), allegedly spoken by creole communities in Upper Guinea in the 17th and 18th centuries (Ehret 1997, Hancock 1969, 1986, 1987, 1994, 2001).³ A variant of this theory argues that Krio descends from the restructured English spoken by the Liberated Africans, freed by the British navy from slave ships captured off the West African coast and settled on the Sierra Leone peninsula between 1808 and 1863 (Nylander 1984, 1999). The prevailing view among those creolists who believe that Krio was substantially influenced by a Western Hemisphere creole is that the arrival of the Jamaican Maroons in 1800 accounts for Krio’s New World bias (this theory can be traced back to Hall 1966: 17). However, as indicated in the preceding section, the Nova Scotians arrived eight years earlier than the Maroons, and in 1800 outnumbered them by almost 2:1. The Nova Scotians were the first major contingent of settlers in the “Province of Freedom” and they are therefore a more likely historical link that could account for the similarities between Krio and restructured varieties in the New World. Note also that no link is shown in Figure 1 between Jamaican Creole and Krio.⁴ Instead, the strongest cross-Atlantic affinity is that between Krio and Gullah. As mentioned above, a substantial number of the Black Loyalists in Nova Scotia and New Brunswick had originally come from the southern Atlantic states, including South Carolina, where Gullah is still spoken today. Since a subgroup of these Black Loyalists formed the original Sierra Leone Settlers, it is conceivable that they can provide the explanation for the correspondences between Krio and Gullah. In order to determine what role the Nova Scotians played in the history of Krio, the following sections will attempt to trace the ultimate points of origin on the American mainland of those Loyalists that traveled to Sierra Leone in 1792.

3.1 Data sources Needless to say, pinpointing the origin of the NS Settlers is not easy. In the diary covering his stay in Nova Scotia in 1791–1792 (Clarkson 1791–1792, Fergusson 1971), John Clarkson mentions general musters of the emigrants (prior to embarkation) several times, but these muster lists have not been found so far. Embarkation or passenger lists of the transports sailing to Sierra Leone in January 1792 have apparently not survived either, so the identity of the emigrants to Sierra Leone had to be established indirectly, by gleaning their names from a variety of sources, discussed hereafter. • One main source is the “List of Blacks at Birch Town who gave in their names for Sierra Leone” (PRO, CO 217/63: 361–3). This list was compiled by Clarkson during his stay in Birchtown (see Section 1), and lists 155 heads of household by name. It also specifies their places of birth by state, and the number of accompanying persons. With a total of 533 emigrants plus “26 who got to Halifax by stealth” (about whom no further information is given) this represents almost half of the total number of emigrants.

7

72

Magnus Huber

• Clarkson’s diaries, the “Mission to America” (Clarkson 1791–1792, Fergusson 1971) and “Mission to Africa” (Clarkson 1927) mention 149 names of emigrant African Americans. • The series of letters and petitions written by NS emigrants to Clarkson and the Sierra Leone Company between 1791 and 1800 (mentioned in Section 1; Fyfe 1991) contains 116 names. • Two additional, minor sources were also used: in her analysis of women traders in Sierra Leone, White (1987) refers to four Nova Scotians. Two names were extracted from Fyfe (1953). After the removal of doublets, these sources provide 340 different names. To establish their origin on the American mainland, these were then searched for in the Book of Negroes.

3.2 Problems in identifying Settlers in the Book of Negroes The identification in the Book of Negroes of the names extracted from the sources may seem like a simple and straightforward task, but in a good number of cases it proved a rather time-consuming process, for a number of reasons, such as variable spelling of names and uneven historical documentation. Spellings in the sources often deviate from one another (the second variant in the following examples is that in the Book of Negroes), as indicated in Table 1 in the Appendix. For example, Ackey Dixon (from the Birchtown List) is listed as Acky Dickson, Mosis Murry (Fyfe letters) is spelt Moses Murray, and Job Alleyne (Birchtown) is Job Allen, to name but a few. Some of the variants are probably the result of mishearings or phonological spellings on the part of the scribes. Mishearings are presumably the reason for the variants Toby Castleton (Birchtown) and Toby Castington, or Luke Jordan (Birchtown) and Lake Jordan.⁵ Some spellings indicate phonological characteristics of the speaker (or of the scribe, if variants are interpreted as mishearings), e.g. Cato Birden (Birchtown) occurs as Cato Boden, which may be the result of Birden’s non-rhotic pronunciation of his name, taken down by a scribe with a rhotic accent, and Philish Halsted (Fyfe letters) appears as Phillis Halstead, showing s-palatalization. In a couple of instances the identity of individuals was obscured by wrong segmentation of the name by the scribe, as in the case of Dimp Sullivan (Birchtown), whose name is spelt Dempse Sullivan in the Book of Negroes, or Lats Laid (Birchtown), identical with Lotche Slaide — the latter case reflects a shift of the word boundary, s-palatalization and maybe /ɒ/–/ɑ/ merger. Nicknames and abbreviations posed another obstacle to identification. Thus, for example, Henry in one source occurred as Harry in another, Richard as Dick, Nathaniel as Nat, Robert as Bob, and so on. As mentioned above, New York was not the only point of Loyalist evacuation to Nova Scotia: a number of African Americans had come from Boston in 1776 and


with the Loyalist exodus from Charleston and St. Augustine in 1782 (Walker 1976: 7, 9–10, Troxler 1974: 46, 97). In addition to the majority of freed African Americans, at least 1,300 slaves were brought to Nova Scotia by White Loyalists fleeing the United States (Grant 1973: 255). There is very sparse and for our purposes insufficient documentary evidence of this latter group, which means that the points of origin on the American mainland can only be identified for those emigrants to Sierra Leone who had come via New York, and whose particulars had therefore been entered in the Book of Negroes. However, since most of the slaves were not allowed to emigrate, their omission from the records does probably not significantly affect this study. Still, as will be shown in the next section, only 54% of the Settlers could be identified in the Book of Negroes.

3.3 The emigrants’ origins and sociolinguistic background Of the 340 names found in the sources, 118 were identified in the computerized version of the Book of Negroes (Ruch 1996). These are listed in Table 1 in the Appendix by state, county, and — wherever available — town of origin on the American mainland. Added to these are another 70 for whom only the state of origin is known (under the entry “unidentified” for each state). Table 1 also indicates the percentage of the African American population for each county in the southern states of VA,⁶ the Carolinas, and GA, from censuses taken between 1770 and 1790. If the average household size of 3.4 in the Birchtown list is characteristic of the Settler community as a whole, then the 188 names in the table represent about 639 emigrants, that is, over half of the 1,196 who left for Sierra Leone in 1792 — a fairly representative sample. Figure 2 is a summary of the figures in Table 1, and shows the proportions of emigrants to Sierra Leone by US state.⁷ The segments of the southern states of VA, NC, SC, and GA are shaded. There were thus two main areas of origin: half of the Settlers came from VA, while a third was from the Carolinas and GA, with SC providing the majority among the latter group. These areas correspond to the two centers of slave concentration in the late 18th century (cf. the map for 1790 in Hilliard 1984).

VA 50.5%

SC 26.6%

Others 11.7%

NC 4.3% GA 3.2% Carib 3.7%

Southern states

Figure 2. Nova Scotian Settlers by state of origin

GA

73

74

Magnus Huber GA 4.8%

SC 20.8%

VA 40.1%

NJ 9.6% NY 9.3%

NC 2.1% CT 2.9% PA 2.8%

Others 7.6%

Southern states

Figure 3. Black Loyalists in Nova Scotia and New Brunswick

For comparison, Figure 3 shows the states of origin of all Black Loyalists evacuated to Nova Scotia and New Brunswick in 1783 (about 90% of all New York evacuees went to Canada), as recorded in the computerized version of the Book of Negroes (Ruch 1996). When the origins of all Black Loyalists (as indicated in Figure 3) are compared with those of the subgroup that moved on to Sierra Leone nine years later (Figure 2), we note that the latter shows a proportional increase in Southern state origin (in particular VA, SC, NC) at the expense of the Northern states of NJ, NY, CT, and PA (compare the shaded segments in the Figures, which indicate the Southern states). This may be due to the fact that conditions in the South with its plantation economy were harsher on the whole. The memory of the hardships suffered in captivity would still have been present among the exiles in Nova Scotia, and presumably provided a stronger motivation to emigrate to Sierra Leone for Southern than for Northern slaves. Using the Atlas of Early American History (Cappon 1976), the points of origin of the 118 Settlers identified in Table 1 were plotted on Map 1. One finding is that the majority of Settlers came from a coastal belt stretching about 50 km inland, while only few came from the interior. This is an important finding and requires explanation, given that by 1780 almost all of modern VA and SC had been settled.⁸ For SC in particular, one explanation is that between 1775 and 1781, when the NS Settlers fled to the British, the area beyond the 50 km coastal belt had a low population density, with small farms dominating. The majority of labor-intensive plantations were concentrated near the seaboard, and they would — simply by their numerical dominance — provide the bulk of runaway slaves. Another explanation again concerns motivation, in that conditions on the large coastal plantations were harder for the individual slave than in the hinterland with its (presumably more humane) homestead economy. Thus, the urge to join the British would have been greater in the Tidewater areas than in the interior. Although Gullah is today spoken in a very restricted geographical area, it is generally accepted that this creole was more widespread in the second half of the 18th century. Specifically, Gullah (or varieties showing at least some degree of creoliza-


tion) was current in the Low Country of SC and GA (e.g., Bailey 2001: 58, Kautzsch and Schneider 2000: 251, Winford 1997: 315, 317 fn. 8, Sutcliffe 2001: 150–for a discussion of possible creolization in VA see below). Gullah was present exactly on that 50 km wide coastal strip from which many of the NS Settlers from SC and GA originated. The black wedges in circles on Map 1 represent the percentage of the African American population in 1770 for the four southern states (figures from USBC 1976: 1168), from which 85% of the Settlers originated. It ranged from 35.3% in NC to 60.5% in SC. These figures may suffice as a very general indication of the Set-

Map 1. Origin of the Nova Scotians emigrating to Sierra Leone in 1792

75

76

Magnus Huber

tlers’ socio-linguistic background. However, it is well known that the Black to White ratio varied greatly within individual states, generally with higher percentages of Blacks near the coast, where the plantations were larger, and lower percentages in the hinterland with its smaller farms. For a more accurate social and linguistic picture of of the Settlers’ origins we need statistical information on those social units in which the formation of AAVE or creolization took place, i.e., the individual plantations. But as long as state-wide statistical information on these is lacking, other sources must be tapped. As to the period for which this fine-meshed information is needed: the average age of the NS Settlers identified in the Book of Negroes was 32.5 years in 1783, so we can assume that their language acquisition must have taken place around 1760. The first detailed nation-wide census was taken in 1790; before that, population figures are more difficult to come by, less reliable, and not generally available for all areas in the same quality. Table 2 shows the population data for the four states of VA, North and South Carolina, and GA that I was able to find for the 1755–1790 period. For each period, the first column (a) indicates the proportion of African Americans among the total population in the respective state as a whole.⁹ The second column (b) shows the proportion of the black population in those counties where the Nova Scotians identified in Table 1 and Map 1 actually came from. Throughout the period and in every state considered here, the percentage of the African American population was higher in the counties of Settler origin than it was in the respective state as a whole, in the majority of cases even substantially so (highlighted figures in Table 2). Unfortunately, county-level data for SC and GA is neither available for 1755 nor for the 1770s-1780s period. It may reasonably be assumed, however, that the trend here was the same. The Settlers, then, generally came from counties with a high proportion of African Americans, which again confirms the findings above that conditions for the African American population were worse (and consequently the motivation to flee their masters greater) in areas where they constituted a large proportion of the total population. Table 2. Percent of African Americans in: (a) state of origin and (b) counties of origin (Greene and Harrington 1981; USBC 1976; Dodd and Dodd 1973; Rogers 1969 [1918]) State

VA

1755

1760s–1780s

1790

State of origin

Counties of origin

State of origin

Counties of origin

State of origin

Counties of origin

58.1

59.6

53.7a

55.0a

40.9

51.4

50.3b

NC

38.3

52.2

41.2b

26.8

37.7

SC

66.6

—

60.5c

—

43.7

69.7

—

45.5c

—

35.9

56.8

GA a

50.0

1782–1783

b

1767

c

1770 — no data available


Winford (1997: 315) argues that Gullah as a full-fledged language must have emerged in coastal SC in the 1720–1775 period, which saw the establishment of a “successful plantation economy” in the southern American colonies. The time that the NS Settlers acquired their language (roughly the 1760s) falls into this period. The majority came from counties characterized by larger plantations, and about a quarter of them actually came from Gullah country. For these 25% at least, a relatively restructured, even creolized, type of English can be assumed. As for the Virginians among the Settlers, the social situation in this state (close contacts between African Americans and Whites on tobacco farms until about 1720) is usually considered to have produced closer approximations to English among the African Americans than in SC and GA. However, the speech of at least some of the Virginian Settlers may have been of a more restructured type than is usually assumed for the state as a whole: as in the more southern colonies, they came from Tidewater counties with a comparatively high proportion of the Black population (see Table 2 above), suggesting heavier restructuring of English. The possibility of pidginization and creolization in some areas of, or individuals from, VA is pointed out by Winford (1997: 320–1). A third of the slaves imported to VA in 1769–1772 came directly from Africa (2,149 in absolute numbers; USBC 1976: 1172). These would have developed L2 varieties at best, and more probably pidginized versions of English on the Tidewater country plantations that many of the NS Settlers came from. Of the individuals listed in Clarkson’s Birchtown list (see 3.1), 28.7% were born in Africa, and these can be assumed to have spoken pidginized English of the kind that Clarkson recorded in his diary (see Section 1). In sum, about a quarter of the Nova Scotians arriving in Sierra Leone in 1792 came from an area where Gullah or related varieties of restructured English were spoken. The other main region of origin was the Chesapeake area of VA with its relatively high Black to White ratio. Additional research into the possible former existence of pidginized and creolized English in Tidewater VA is needed, but given that a good number of the Virginian Settlers came from this area it is plausible that they spoke such varieties.

4. The Nova Scotians’ language I will now turn to a linguistic analysis of the Nova Scotians’ language in the light of the socio-historical and demographic background reconstructed in the previous sections. Linguistic aspects of the Settlers’ variety have already been analyzed by Jones (1991) and more recently by Montgomery (1999) and van Herk (1998, 1999a, 1999b, 1999c). But the information about the emigrants’ origins and background that we now have at our disposal allows us to put these findings into perspective, and draw conclusions about the quality as well as social and geographic variability of the Nova Scotians’ language. A word of caution is in order, since there is a certain risk of circularity: as far as

77

78

Magnus Huber

language restructuring or the genesis or existence of creoles is concerned, the events that took place in Sierra Leone and in the South Atlantic states are neither entirely clear nor undisputed in the literature. Thus it would, for example, be problematic to argue for creolization in VA just because there is a creole in Sierra Leone and many of the Nova Scotians originally came from VA. Instead, what I propose to do in the following is to look at evidence of the Nova Scotians’ language to see if this settler group, which in the previous section has been shown to originate from areas on the American mainland where the existence of restructured English is likely, can possibly account for the affinities between Gullah and Krio, as suggested in the Introduction.

4.1 The Nova Scotians’ letters As mentioned in Section 1, the only substantial first-hand evidence of the language of the Nova Scotians that has come to light so far is a collection of 41 letters and petitions, a corpus of approximately 12,000 words, written in Nova Scotia and Sierra Leone in the period 1791–1800 (published in Fyfe 1991).¹⁰ The following excerpt conveys an idea of the nature of these documents: Hond Sir It this indeed out of my power To express to you the regard most of your People Still bear toward your Hond the Natise as well as us all even those at Baullum shore; I thanks you for your Friendly advise you Give me in your letter I receive the 12 of Dec and For you satterfaction I have precur you one of the Governor Pmphle in which you will find the Movening cause of all The disturbance in the place . . . (Letter No. 33. Boston King, 1798–01–16)

This short passage illustrates the non-standard quality of the letters. However, since the writers must have had some manner of schooling or instruction to be able to write at all, the language of these letters is certainly more acrolectal than the authors’ spoken language. In spite of their phonological, morphological, and grammatical peculiarities, a number of the letters are quite formal, many being official petitions and complaints addressed to a superordinate. Only occasionally — as in the excerpt above — do they strike a personal note. The formal character of these documents also becomes clear in the use of conventionalized address and farewell formulas. Nevertheless, two passages in Clarkson’s diary make it clear that the language of at least some of the Nova Scotians was further removed from the standard than their letters indicate. On October 31, 1791, Clarkson was addressed by a prospective emigrant in Nova Scotia “in language, which is impossible for me to convey, in the simple, unaffected and forcible style in which it was delivered” (Clarkson 1791– 1792: 91). Two days later, he actually reproduced a specimen of this language in his diary: “No Massa me no hear, nor no mind, me work like slave cannot do worse Massa in any part of the world, therefore am determined to go with you Massa, if you please” (1791–1792: 96; full quote in Section 1). Note that this speaker was born


in Africa, like many others among the Nova Scotians emigrating to Sierra Leone (see 3.3). Pidginized speech such as the one quoted by Clarkson is likely to have been common among the African-born in the emigrant community.¹¹ In the following analysis, I will restrict myself to those letters whose authors’ origins are identified in Table 1. The authors of the 17 letters that I was able to trace back unambiguously to a state and, in all cases except one, to a particular county, are listed in Table 3. The letters are identified by the state abbreviation and the numbering in Fyfe (1991), together with the writers’ names and counties of origin (the figures in parentheses refer to the county identification in Table 1 and on Map 1). With some 3,000 words (NY 228, VA 525, SC 2031, GA 246), this represents roughly a quarter of the entire letter corpus. Table 3 shows that, with over 2,000 words, there is a strong SC bias in this selection — twice as much as the other three states taken together. In addition, all SC writers came from the Charleston area. Montgomery (1999: 23–4) raises the question of whether at least some of the Nova Scotians’ letters were written by scribes, since many African Americans in the late 18th century were illiterate. That illiteracy was also common among the Nova Scotians is clear from the fact that some signatures are accompanied by a cross with the addition “his mark”. Thus, unless the identity of the scribe(s) is known, such letters do not constitute documentation of the speech of any particular individual(s), but rather they are representative of the community as a whole — that is, if the scribe was a Nova Scotian. After an evaluation of the sociohistorical circumstances Montgomery suggests “that the documents derive from writers within the settler community rather than from a British or North American agent” (1999: 24). Furthermore, in the case of the SC letters in Table 3, we can be almost certain that they were actually written by the individuals signing them. We know that Boston King was literate,¹² so there was little need for him to engage scribal help. Isaac Anderson

27

228

John Cooper (1)

VA

04 06 10 17

76 35 124 290

Susana Smith (13) Daniel Cary (2) Luke Jordan (14) Miles Dixon (14)

GA

05

82

Richard Dixon (?)

08 11

128 36

Andrew Moor (1) John Cuthbert (2)

SC 18 20 21 22 32 33 34 36 41

Words

Words

NY

Letter

Letter

Writers and their county of origin

State

State

Table 3. Origin of Nova Scotians’ letter writers

314 84 156 246 294 408 146 358 25

Writers and their county of origin Isaac Anderson, Cato Perkins (6) Isaac Anderson, Cato Perkins (6) Isaac Anderson, Cato Perkins (6) Isaac Anderson (6) Boston King (6) Boston King (6) Isaac Anderson (6) Isaac Anderson, Stephen Peters (6) Isaac Anderson (6)

79

80

Magnus Huber

and Cato Perkins were sent to London by the Nova Scotian community in 1793 to lodge complaints with the board of directors of the Sierra Leone Company. Given that such a task inevitably involved correspondence at least one of them was in all probability literate. In a letter that Anderson, still in London, addressed to Clarkson on 1794–11–02 he cautions that “things must Remain silant till I Recd some answer from the peopel of Sira Lone” (SC22). The confidential tone of this speaks against the presence of a third party in the form of a scribe. Note also that there are no letters signed by Perkins alone but three signed only by Anderson. This could of course be pure accident but is possibly an indication that the mission’s correspondence lay in Anderson’s hands.

4.2 Linguistic analysis In the analysis of the letters in the subcorpus described in Table 3, I have generally disregarded formulaic greetings and farewell phrases, as these may not always be representative of the writers’ language. To make my study of the 17 letters comparable to previous investigations of the entire letter corpus, I will follow Montgomery’s (1999) agenda, and examine phonological characteristics as well as past marking, the presence of -s on verbs in third person singular and plural, as well as nominal pluralization in the area of morphology.

4.2.1 Phonology

Misspellings and hypercorrect spellings indicate phonetic and phonological characteristics of the speech of the letter writers. Rhoticity, consonant cluster reduction, /w/—/hw/ merger, replacement of the velar nasal /ŋ/ by /n/, avoidance of central vowels, and front lax vowel neutralization have already been spotted by Montgomery (1999: 8) and will shortly be illustrated here with examples from the subcorpus. Rhoticity. Montgomery argues, contra Jones (1991: 87), that “the evidence for r-lessness in the Sierra Leone documents is unambiguous, at least for some writers” (1999: 8 fn. 5). The language of the writers in the subcorpus seems to have been non-rhotic (honnah ‘honor’ VA10, yea ‘year’ SC32), but there are also two nonstandard spellings that possibly indicate the presence of post-vocalic r: senceer ‘sincere’ SC22 and plesuer ‘pleasure’ SC22.¹³ Note that both tokens are found in the same letter, written by Isaac Anderson. They may thus be idiolectal rather than characteristic of South Carolinian speech as a whole. Consonant cluster reduction. Montgomery found examples of final consonant cluster reduction, to which can be added the cases of medial reduction in the letters examined here. Examples of both types are: cappin ‘captain’ SC33 (also common in white pronunciation), the hypercorrect spelling mends ‘men’ SC41, natise ‘natives’ SC33, nourishmen ‘nourishment’ GA08, hypercorrect pends ‘pens’ SC32, hypercorrect sarvast ‘service’ SC33, sining ‘sending’ SC33. /w/–/hw/ merger. The subcorpus contains spellings that omit -h-, such as wane


‘when’ SC22, wich ‘which’ SC22, wether ‘whether’ SC41, wither ‘whether’ VA10. Hypercorrect spellings are not found. Standard English velar nasal realized as [n]. There are two tokens illustrating this process, been ‘being’ SC36 and blessen ‘blessing’ SC22. Avoidance of the central vowels /ə, ɜ, ʌ /. This can be found in consarned ‘concerned’ SC22, considaration ‘consideration’ GA08, dissalute SC36, marcy ‘mercy’ SC22, porpose ‘purpose’ SC22, sarvant ‘servant’ SC22, sarvast ‘service’ SC33, yong ‘young’ GA05. Front lax vowel neutralization before a nasal consonant. This is shown by blessen ‘blessing’ SC22, endependence ‘independence’ SC32, entention ‘intention’ NY27, senceer ‘sincere’ SC22, and sining ‘sending’ SC33. Replacement of /v/ by /b/. Not listed by Montgomery but present in my subcorpus, as in fabour ‘favor’ SC33 and possibly sabbage ‘Savage’ (surname) SC32.

4.2.2 Morphology

Montgomery (1999: 10–17) examines past marking for weak and strong verbs, verbal third person singular and plural agreement, as well as pluralization of regular and irregular nouns in the Nova Scotians’ letters. Table 4 shows the results for the same categories in the subcorpus, sorted by state and letter. The column “Past mark-

Table 4. Selected morphological features in 16 Nova Scotians’ letters [Standard English features]

NY 27 VA 04 06 10 17 SC 18 20 21 22 32 33 34 36 41 GA 05 08

Past marking

Verb agreement

Pluralization

Standardness by letter

Standardness by state

Strong Weak verb verb

3S -s 3PL -ø

Reg. Irreg. Ns Ns

Total

%

Total

%

2/2

1/1 1/1

6/6 1/1 1/1 1/3 9/14 23/25 6/6 7/8 8/11 6/18 12/18 4/7 18/20 1/3 2/2 7/7

100 100 100 33 64 92 100 88 73 33 67 57 90 33 100 100

6/6

100

0/1 9/10 1/1 3/3

2/2 1/1

0/1 2/2

1/1 0/2 1/2

4/4

4/5

1/1 1/1 2/2 0/1 0/1 0/1 0/1 1/3 1/1 2/2

3/3 0/1 0/4 3/4 1/1 2/3 1/6 4/4 2/2

1/1

1/1 1/1 7/7 6/6 2/2 3/3 5/5 5/8 5/9 1/2 10/11 4/4

0/1 1/1 2/2 0/1

1/3 1/1

12/19

63

85/116 73 9/9

100

8

82

Magnus Huber

ing” shows how many of the strong and weak verbs (excluding copular and auxiliary uses of be and have) are inflected in past contexts. “3S -s” under “Verb agreement” indicates -s-marking of verbs (including auxiliary have but disregarding be) with a 3S subject, while “3PL -ø” (including have and be) records verbal agreement with plural subjects, i.e., absence of -s. Letter GA36 does not exhibit any of the contexts investigated here and is therefore disregarded in the table. Note that, for reasons that will be explained below, the figures indicate how many tokens are marked in the Standard English manner rather than the absence of standard features. For example, the figures for “Verb agreement” in letter SC22 mean the following: 3S -s 0/1 — the one instance of a 3S lexical verb in this letter is not inflected (i.e., shows absence of Standard English -s); 3PL -ø 2/3 — of the three tokens of plural verbs two are inflected in the standard English way (i.e., they have -ø marking). Thus, although the criteria for identifying and including features in our calculations are the same, the figures in Table 4 are not directly comparable with Montgomery’s, who counts the presence of non-standard features. The figures for the categories investigated in the subcorpus are too small to allow a fruitful investigation of contextual conditioning effects, so the results have to be interpreted in some other way. The columns “Standardness by letter” and “Standardness by state” provide a measure of the conformity to the rules of Standard English of a letter or all letters written by Nova Scotians from a particular state by adding up all features attested in that particular letter or group of letters. Looking at the figures for whole states, both NY and GA are completely standard (at least in the features investigated here),¹⁴ while VA and SC show rates of 63% and 73%, respectively.¹⁵ This suggests that more restructuring of English took place in some parts of VA than is commonly assumed, a possibility raised by Winford (1997: 320–1), and that the VA letter writers came from these areas. That the Virginians among the Nova Scotians came from counties with high proportions of African Americans has already been shown in Table 2. Note also the great variability within VA and SC, ranging from 33% to 100% standardness. Again, the chance factor cannot be excluded since the number of tokens is quite small for some letters and the corpus is skewed by necessity, but it may also point to considerable intra- and inter-individual heterogeneity (cf. also Montgomery 1999: 18). Charleston is the only relevant county in SC for which I was able to find population data for 1770 (see Table 1): while in that year African Americans made up only 53.7% of the total population in Charleston county, they constituted 60.5% in SC as a whole (Table 2). This hardly comes as a surprise since the proportion of African Americans was generally higher in rural areas with their plantations than in urbanized ones. Nova Scotian letter writers from rural SC have not yet been identified, but one may surmise that their written and spoken English was even more restructured than that of the Charleston letters in the subcorpus. In his analysis of the Settlers’ letters, Jones (1991: 82, 102) concludes that they show no evidence of pidginization, but rather exhibit similarities to 18th-century lower-class white speech. Montgomery (1999: 27) similarly argues that the evidence


“is consistent with the view that the character of early African American English reflected varieties traceable to the British Isles,” but is more cautious regarding the issue of pidginization and suggests that the variability in the corpus may be due to a creole background of some of the writers and that the “everyday speech of Settlers may have had more discernible creole roots” (1999: 28). In this light, the most important findings of this section are (1) that the writers from the South (VA and SC) show a lower rate of standard features than the one who came from the North (NY), possibly pointing to a more creole-like nature of the former’s spoken language, and (2) that there was no linguistic difference between SC and VA, suggesting that the restructuring of English in some parts of the Chesapeake area of VA may have been as heavy as in coastal SC. This could be an indication that creolization of the Gullah type was more widespread than is generally assumed today. However, it should be borne in mind that these conclusions rest on a very small corpus, and that more research is needed to confirm or refute these findings.

5. Conclusions A careful analysis of historical records is indispensable in the reconstruction of early varieties of AAVE. Extrapolation from synchronic diaspora varieties data can help in the re-creation of earlier stages of AAVE but should be supplemented by original historical data wherever possible. In many cases such documents are available in the form of sociohistorical information or as records written by speakers of early AAVE, as Sections 3 and 4 have demonstrated. It is true that such data have their shortcomings, particularly for the early stages of the variety (e.g., statistical and biographical information is often rather limited, not always as reliable as one would wish, and not equally available for all areas or individuals studied; written primary sources do not directly mirror the speech of their authors, etc.), but in a historical approach one simply has to make the best use of any extant information. Collecting original historical data is often a time-consuming (and sometimes frustrating) undertaking, but if available data are analyzed with care they can add to our knowledge of the development of AAVE. This article has provided a historical and socio-statistical background to one of the earliest groups of African Americans that left the eastern United States and settled overseas, the Nova Scotians who arrived in Sierra Leone in 1792. This information was then used to refine the analysis of a corpus of letters written by these emigrants. We now have a fairly accurate idea of the point of origin of the Nova Scotians in Sierra Leone. And seventeen of the letters can now be linked to their geographic origin. While the linguistic analysis of these letters is hampered by the small number of tokens, they nevertheless provide useful insights into the nature of the Nova Scotians’ speech. Montgomery noted that in the Nova Scotians’ letters “unambiguous evidence for creole features is lacking, but the prevalence of zero morphology [. . .] is not

83

84

Magnus Huber

inconsistent with Sierra Leone having a creole-based character to some degree or for some speakers” (1999: 20). With the possible exception of a couple of phonological features, such as the avoidance of central vowels, the /w/–/hw/ merger (typical of Gullah), or the replacement of /v/ by /b/ (also found in southern white speech), the letters analyzed here do not show any features that are found exclusively in English creoles in the Atlantic. Thus, the letters do not in themselves offer firm proof that the language of the NS Settlers was Gullah-like. However, their places of origin (coastal counties with a high proportion of African Americans) as well as the internal variation found in the letters (see Section 4 above), coupled with the fact that over 28% of the Settlers were born in Africa and therefore probably spoke pidginized or African L2 varieties of English, make it quite likely that some of them were creole speakers. Especially for the 25% from traditional Gullah country this is a likely scenario, and some individuals from VA may also have spoken a rather heavily restructured variety. Regardless of whether the features in the Nova Scotians’ speech (or early African Americans’ language in general) can be traced back to British English dialects or result from the creole origins of AAVE, it is clear that the Nova Scotians introduced quite a number of non-standard features to Sierra Leone. Therefore — although the findings presented here are based on a rather small corpus and on historical information which has its limitations — this invites the conclusion that the transshipment of Nova Scotians to Sierra Leone may explain the similarities between Gullah and Krio.¹⁶

Notes 1. An earlier version of this article was presented at the Meeting of the Society for Pidgin and Creole Languages, Coimbra, Portugal, June 26–7, 2001. I am grateful to the participants for their helpful comments and suggestions. Special thanks go to two anonymous reviewers and Philip Baker, as well as to Michael Montgomery and Gerard van Herk for their interest and help, as well as willingness to share their data with me. 2. For a summary of these issues see the papers by Cassidy, Hancock, Montgomery, and Mufwene in Montgomery (1994), Kautzsch and Schneider (2000), Winford (1997), as well as the more specific discussions in Cassidy (1980), Hancock (1980), and Rickford (1998, 1999). 3. Hancock’s “componential” approach provides room for the influence of other settler groups as well, such as the Jamaican Maroons or the Nova Scotians, but the basic and earliest ingredient in the process is still assumed to be Upper Guinea Coast Creole English. 4. The calculations actually yield a slight positive score of 1.8 for JAM-KRI, but only scores higher than 6.7 (20% of the highest score of 33.5; for KRI-WAF) are shown in the Figure. 5. Some names are the result of rather tasteless wordplays by former masters, a common and psychologically cruel practice among slave owners (see Berlin 1996: 251–2). Some appellations ironically commented on the slaves’ bondship: Names of Roman emperors were in vogue, as were biblical allusions. Luke = Lake Jordan is an example of the latter.

The Nova Scotia–Sierra Leone connection 6. The two-letter abbreviations used in the following are CT = Connecticut, DE = Delaware, GA = Georgia, MA = Massachusetts, MD = Maryland, NC = North Carolina, NJ = New Jersey, NY = New York, PA = Pennsylvania, RI = Rhode Island, SC = South Carolina, VA = Virginia. 7. Note that in the Birchtown list the place of birth for 20 heads of household is given as “Carolina” only. Of these, 7 were identified in the Book of Negroes: 5 SC (71.4%), 1 GA (14.3%), 1 NC (14.3%). Assuming these proportions, 9 of the remaining 13 would have been from SC, and 2 each from GA and NC, respectively, which have been included in Figure 2. 8. Actual settlement areas (rather than areas just claimed by the colonies) are indicated on the map “Expansion of European settlement 1650–1800” at http://www.geoamericas.com. 9. The total population is here considered to be Blacks plus Whites only. The category “others”, presumably mainly referring to the Indian population and in most cases very small anyway, was disregarded for the present purposes. 10. Clarkson’s diary contains a couple of additional letters, not published in Fyfe (1991), but these are disregarded for the present purposes. 11. See Huber (1999: 72–3) for further discussion of the evidence from Sierra Leone of the Nova Scotians’ English. 12. There are biographical studies on, or autobiographies by, some of the individuals identified in Table 1, including Boston King (Blakeley 1968, Clarke 1991). King had been born 28 miles from Charleston, and moved to the capital when he was 17 or 18 years old. He had some schooling and was able to read and write, which makes it very likely that he wrote his letters himself. 13. One of the anonymous reviewers suggests the possibility that -er “is used more generally as a representation for what is schwa in English,” which strengthens the case for r-lessness. 14. That the NY letter is so close to Standard English in the categories investigated here is explainable by the low number of African Americans in the northern states, corresponding to better access to English other than plantation continuum varieties. The standardness of the GA letters remains a puzzle at this stage of research, at least for the letter by John Cuthbert, who came from the coastal region bordering on SC. 15. These figures do not indicate that the SC letters are further removed from the standard than the ones from VA: the Yate chi-square test shows that the difference between the VA and SC figures is statistically not significant (p = .526), i.e., that there is no linguistic difference between the VA and the SC letters. 16. An anonymous reviewer suggests that this “in turn can be interpreted as important evidence that the Low Country coastal strip stretching from GA to NC, together with the tide water areas of VA was an area where a wider Gullah was spoken.” One has to beware of circular arguments here (see Section 4), but the evidence presented in this article is at least suggestive of wider restructuring/creolization in the southern states.

References Bailey, G. 2001. “The relationship between African American and white vernaculars in the American South: A sociocultural history and some phonological evidence.” In Lanehart, 53–92.

85

86

Magnus Huber Baker, P. 1999. “Investigating the origin and diffusion of shared features among the Atlantic English creoles.” In St. Kitts and the Atlantic Creoles. The Texts of Samuel Augustus Mathews in Perspective, P. Baker and A. Bruyn (eds.), 315–64. London: University of Westminster Press. Baker, P. and M. Huber. 2001. “Atlantic, Pacific, and world-wide features in English-lexicon contact languages.” English World-Wide 22: 157–208. Berlin, I. 1996. “From creole to African: Atlantic creoles and the origins of African-American society in North America.” The William and Mary Quarterly 53: 251–88. Blakeley, P. R. 1968. “Boston King: A negro loyalist who sought refuge in Nova Scotia.” Dalhousie Review 48: 347–56. Book of Negroes. 1783. Microfilms file 10,427 reel M-369 National Archives of Canada, Ottawa; M-154.29 The Colonial Williamsburg Foundation, Virginia. (Original: PRO 30/55/100, in the Public Record Office, London). Cappon, L. J. (ed.). 1976. Atlas of Early American History. The Revolutionary Era. Princeton, NJ: Princeton University Press. Cassidy, F. C. 1980. “The place of Gullah.” American Speech 55: 3–16. Cassidy, F. C. 1994. “Gullah and the Caribbean connection.” In Montgomery, 16–22. Clarke, G. E. (ed.). 1991. Fire on Water. An Anthology of Black Nova Scotian Writing. Vol. 1: Early and Modern Writers 1785–1935. Lawrenceton Beach, NS: Pottersfield Press. Clarkson, J. 1791–1792. “Mission to America.” Ms, microfilm. The New-York Historical Society. Clarkson, J. 1927 (1792). Diary of Lieutenant Clarkson, R.N. Freetown: Fourah Bay College Bookshop. (=“Mission to Africa”). Dodd, D. B. and W. S. Dodd. 1973. Historical Statistics of the South 1790–1970. Alabama: The University of Alabama Press. Ehret, R. 1997. “Language development and the role of English in Krio.” English World-Wide 18: 171–89. Fergusson, C. (ed. and intr.). 1971. Clarkson’s Mission to America 1791–1792. Halifax: Public Archives of Nova Scotia. Fyfe, C. 1953. “Thomas Peters: History and legend.” Sierra Leone Studies N. S. 1: 4–13. Fyfe, C. (ed.). 1991. “Our Children Free and Happy.” Letters from Black Settlers in Africa in the 1790s. Edinburgh: Edinburgh University Press. Grant, J. N. 1973. “Black immigrants into Nova Scotia, 1776–1815.” The Journal of Negro History 58: 253–70. Greene, E. B. and V. D. Harrington. 1981 (1932). American Population before the Federal Census of 1790. Reprint 1993. Baltimore, MD: Columbia University Press. Hall, R. A. 1966. Pidgin and Creole Languages. Ithaca, NY: Cornell University Press. Hancock, I. 1969. “A provisional comparison of the English-based Atlantic creoles.” African Language Review 8: 7–72. Hancock, I. 1980. “Gullah and Barbadian — origins and relationships.” American Speech 55: 17–35. Hancock, I. 1986. “The Domestic Hypothesis, diffusion and componentiality. An account of Atlantic anglophone creole origins.” In Substrata versus Universals in Creole Genesis. Papers from the Amsterdam Creole Workshop, April 1985, P. Muysken and N. Smith (eds.), 71–102. Amsterdam: John Benjamins. Hancock, I. 1987. “A preliminary classification of the anglophone Atlantic creoles with syntactic data from thirty-three representative dialects.” In Pidgin and Creole Languages.

The Nova Scotia–Sierra Leone connection Essays in Memory of John E. Reinecke, G. G. Gilbert (ed.), 264–333. Honolulu: University of Hawaii Press. Hancock, I. 1994. “Componentiality and the creole matrix: The southwestern English contribution.” In Montgomery, 95–114. Hancock, I. 2001. “Review article: M. Huber and M. Parkvall (eds.). Spreading the Word: The Issue of Diffusion among the Atlantic Creoles.” English World-Wide 22: 303–14. Hilliard, S B. 1984. Atlas of Antebellum Southern Agriculture. Baton Rouge: Louisiana State University Press. Huber, M. 1999. Ghanaian Pidgin English in its West African Context: A Sociohistorical and Structural Analysis. Amsterdam: John Benjamins. Huber, M. 2000. “Restructuring in vitro? Evidence from early Krio.” In Neumann-Holzschuh and Schneider, 275–307. Jones, C. 1991. “Appendix: Some grammatical characteristics of the Sierra Leone letters.” In Fyfe, 79–104. Kautzsch, A. and E.W. Schneider. 2000. “Differential creolization: Some evidence from Earlier African American Vernacular English in South Carolina.” In Neumann-Holzschuh and Schneider, 247–74. Lanehart, S. L. (ed.). 2001. Sociocultural and Historical Contexts of African American English. Amsterdam: John Benjamins. Montgomery, M. 1994. “Introduction.” In Montgomery, 1–15. Montgomery, M. 1999. “Eighteenth-century Sierra Leone English: Another exported variety of African American English.” English World-Wide 20: 1–34. Montgomery, M. (ed.). 1994. The Crucible of Carolina. Essays in the Development of Gullah Language and Culture. Athens, GA: The University of Georgia Press. Mufwene, S. S. 1994. “Misinterpreting linguistic continuity charitably.” In Montgomery, 38– 59. Neumann-Holzschuh, I. and E. Schneider (eds.). 2000. Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins. Nylander, D. K. 1984. “Histoire et genèse du krio.” Journal des Africanistes 54: 125–7. Nylander, D. K. 1999. “The structure of Tense Phrase in creole languages: A case study (Krio).” In Spreading the Word. The Issue of Diffusion among the Atlantic Creoles, M. Huber and M. Parkvall (eds.), 153–61. London: University of Westminster Press. PRO = Public Record Office, London. Rickford, J. 1998.“The creole origins of African American Vernacular English: Evidence from copula absence.” In African-American English. Structure, History, and Use, S. S. Mufwene et al. (eds.), 154–200. London: Routledge. Rickford, J. 1999.“Prior creolization of African American Vernacular English? Sociohistorical and textual evidence from the seventeenth and eighteenth centuries.” In African American Vernacular English. Features, Evolution, Educational Implications, J. Rickford (ed.), 233–51. Oxford: Blackwell. Rogers, S. L. (ed.). 1969 (1918). Negro Population 1790–1915. Vol. 1. New York: Kraus Reprint Co. Ruch, J. E. 1996. Carleton’s Loyalist Index. Database of 54,568 Records. A Select Index to the Names of Loyalists and their Associates Contained in the British Headquarters Papers, New York City 1774–1783. Ottawa: United Empire Loyalists’ Association of Canada. Sir Guy Carleton Branch.

87

88

Magnus Huber Sutcliffe, D. 2001. “The voice of the ancestors: New evidence on 19th-century precursors to 20th-century African American English.” In Lanehart, 129–68. Troxler, G. C.W. 1974. The migration of Carolina and Georgia loyalists to Nova Scotia and New Brunswick. PhD thesis. Chapel Hill: University of North Carolina. United States Historical Census Data Browser. http://fisher.lib.virginia.edu/census. USBC = United States Bureau of the Census. 1976. The Statistical History of the United States. From Colonial Times to the Present. New York: Basic Books. van Herk, G. 1998. “Don’t know much about history: Letting the data set the agenda in the origins-of-AVVE debate.” Paper presented at New Ways of Analyzing Variation 27, University of Georgia, Athens, GA, October 1–4, 1998. van Herk, G. 1999a. “‘We was very much oppress’: 18th-century African American Vernacular English texts and the origins debate.” Paper presented at the meeting of the Society of Pidgin and Creole Languages, Los Angeles, CA, January 8–9, 1999. van Herk, G. 1999b. “‘Ain’t-shaped holes’ and Standard English that isn’t: Negation and literacy in early African American English letters.” Paper presented at Methods X, St. John’s, Newfoundland, August 1–5, 1999. van Herk, G. 1999c. “‘Safe arrived’: The perfect in early African American English Letters.” Paper presented at New Ways of Analyzing Variation 28, Toronto, October 14–17, 1999. Walker, J.W. St. G. 1976. The Black Loyalists. The Search for a Promised Land in Nova Scotia and Sierra Leone 1783–1870. London: Longman. White, E. F. 1987. Sierra Leone’s Settler Women Traders. Ann Arbor: University of Michigan Press. Winford, D. 1997. “On the origins of African American Vernacular English — a creolist perspective. Part I: The sociohistorical background.” Diachronica 14: 305–44.


Appendix Table 1: 188 Nova Scotian emigrants by place of origin By state, county, and town (1776 administrative borders and names, according to Cappon 1976). Exact town/location in square brackets wherever known. Numbered from north to south and from west to east within individual states; numbers refer to indications on Map 1. Except where they are used as nicknames, name abbreviations have been resolved, e.g., Jno = John etc. Absolute number of emigrants from each state and their percentage of the whole given after the name of the state. Percentage of African American population and absolute figures of B(lacks) and W(hites) together with the census year indicated after the county (figures from Greene and Harrington 1981). Abbreviations used ? B C F W

identification doubtful Birchtown list Clarkson’s “Mission to America” and “Mission to Africa” (first mention of name only) Fyfe (1991) (first mention of name only) White (1987)

VIRGINIA (95 emigrants = 50.5%) 1. FAIRFAX, 1 42.1% (1782; 5,154 W : 3,609 B) Washington, Henry (B 19) = Washington, Harry [Mt. Vernon] 2. CAROLINE, 2 79.4% (1783; 965 W : 3,712 B) Carey, Daniel (C 1791–12–08) Cary, Dannail (F 1792–06–16) = Carey, Daniel ?Simmons, Rubin (F 1794–11–19) = Simmons, Ruben [“Carlisle Co.”] 3. ESSEX, 1 53.1% (1785; 2,489 W : 2,817 B) ?Garnet, Bristol (B 6) = Garritt, Bristow 4. LANCASTER, 1 62.5% (1783; 1,541 W : 2,567 B) Steven, Anthony (F 1792–11–19) = Stevens, Anthony 5. HENRICO, 1 83.9% (1787; 1,339 W : 6,961 B) Williams, Georg (F 1792–11–28) = Williams, George 6. NEW KENT, 1 64.6% (1782; 1,617 W : 2,957 B) Castleton, Toby (B 56) = Castington, Toby 7. GLOUCESTER, 3 46.7% (1783; 3,151 W : 2,764 B) Freeman, Thomas (F 1799–08–31) = Freeman, Thomas Henry, Ralph (C 1791–12–08) = Henry, Ralph [Gloucester] Wite, Elizabeth (F 1792–11–28) = White, Elizabeth

89

90

Magnus Huber 8. NORTHAMPTON, 4 50.5% (1790; 3,181 W : 3,244 B) Smith, Jacob (F 1792–11–28) = Smith, Jacob [Horntown] Warren, Henry (C 1791–12–08) = Warren, Henry Veters, Adam (B 16) = Veters, Adam Weeks, Peter (B 73) Weak Peter (F 1797–06–1) = Weeks, Peter 9. JAMES CITY, 1 78.8% (1782; 493 W : 1,832 B) Gray, John (B 147) Gray, John (F 1792–11–28) = Gray, John [Williamsburg] 10. SURRY, 1 53.3% (1782; 2,389 W : 2,729 B) Cox, Martin (B 41) = Cox, Martine 11. ELIZABETH CITY, 5 41.6% (1782; 908 W : 648 B) Glasgow, Jonathan (B 92) Glasgow, Jonathan (F 1795–04–16) = Glasgow, Jonathan [Hampton] Kirby, Hannah (C 1791–12–08) = Kerby, Hannah [Hampton] Moore, Lewis (B 58) = Moore, Lewis [Hampton] Sullivan, Dimp (B 29) Sillavan, Dem (F 1792–11–19) = Sullivan, Dempse Weeks, George (B 20) Weeks, George (F 1792–11–28) = Weeks, George 12. ISLE OF WIGHT, 1 43.9% (1782; 3,760 W : 2,948 B) Lawrance, Dick (F 1797–06–1) = Laurence, Dick 13. NORFOLK, 21 36.6% (1782; 5,365 W : 3,096 B) Bracy, Jonas (F 1792–11–26) = Bracey, Jonas [Norfolk] Campbell, Pompey (C 1791–12–08) = Campbell, Pompey [Norfolk] ?Widow Conner (C 1792–01–9) = Connor, Dinah [Craney Island] Cook, Henry (B 86) Cook, Henry (F 1792–11–19) = Cook, Henry [Norfolk] Godfrey, Abegail (C 1791–11–13) = Godfrey, Abigail [Norfolk] Davis, Anthony (C 1792–09–1) = Davis, Anthony [Norfolk] Greene, Adam (C 1791–12–08) = Green, Adam [Portsmouth] Dixon, Ackey (B 142) = Dickson, Acky [Norfolk] ?Jordan, John (F 1794–11–19) = Jordan, Jack [Norfolk] Miller, Frank (B 66) = Miller, Frank Murray, Moses (B 135) Murry, Mosis (F 1796–05–24) = Murray, Moses Patrick, Franck (F 1792–01–12) = Patrick, Frank [Norfolk] Perth, Mary (White 1987: 21) = Perth, Mary [Norfolk] Robertson, James (B 17) Robertson, James (F 1799–08–31) = Robertson, James [Portsmouth] Saunders, Thomas (C 1791–12–03) = Saunders, Thomas [Craney Island] Smith, Susana (F 1792–05–12) = Smith, Susannah Snowball, Nathaniel (F 1795–04–16) = Snowball, Nathaniel [Norfolk] Virginia, Norfolk (B 34) = Virginia, Norfolk Waistcoat, Joseph (B 74) = Waistcoat, Joseph [Norfolk] ?Walker, Henry (B 81) Walker, Henry [Norfolk] Whitton, Samuel (B 152) = Whitten, Samuel [Norfolk] 14. NANSEMOND, 8 47.5% (1783; 2,842 W : 2,567 B) Bush, Richard (B 126) = Bush, Richard Demps, John (F 1792–11–28) = Dempse, John Dixon, Absalom (C 1791–12–12) = Dickson, Absalom Dixon, Luke (B 51) Dixon, Luke (F 1792–11–19) = Dixon, Luke

The Nova Scotia–Sierra Leone connection Dixon, Miles (F 1792–11–19) = Dixon, Myles Jordan, Luke (B 32) Jurdin, Luke (F 1792–11–18) = Jordan, Lake Turner, Nathaniel (C 1791–12–12) = Turner, Nat Wilkinson, Moses (B 53) Wilkinson, Moses (F 1794–11–19) = Wilkinson, Moses 15. PRINCESS ANNE, 1 39.9% (1783; 3,999 W : 2,656 B) ?Lawson, Saul (B 15) = Lawson, Solomon 0. (UNIDENTIFIED VA), 43 Bacclay (?), James (B 33) Ball, Richard (B 46) = Ball, Richard Bennet, Joe (B 12) Bligh, London (B 48) Brane (?), Joseph (B 70) Campbel, Ezekiel (B 36) Campbel, Samuel (B 13) Coinash, Lewis (B 121) Conner, James (?) (B 8) Cratton, Isaac (B 85) Demps, John (B 10) Francis, Peter (B 136) Godfry, John (B 112) Gordon, Frank (B 45) Grant, William (B 84) Gullet, Samuel (B 44) Gullet, Samual (F 1797–06–1) = Gullett, Samuel [“Wooster Co.” > MD?] Gusto, Shak (?) (B 95) Harrison, Robert (C 1791–12–03) = Harrison, Bobb [“James River”] Hector, James (?) (B 129) Ivy, John (B 14) James, Isaac (B 110) Johnson, Prince (B 54) Jones, Adam (B 50) = Jones, Adam [“Gold Mine”] Jones, Frank (B 96) Lawrence, Philip (B 27) Leach, Richard (B 133) McKay, John (B 25) Nicholson, Robert (B 24) Oliver, John (B 108) Poll, John (B 144) Proof, Simon (B 7) ?Prophet, Daniel (F 1792–11–19) = Proffitt, Daniel [Rappahannock Co.]/Proffott [?], Daniel [Norfolk] Richards, Richard (C 1791–12–03) = Richards, Dick [Little York]/Richards, Dick [Isle of Wight] Roger, Abby (B 62) Rutledge, Pompey (B 5) Salsbury, John (B 18) Stafford, Robert (C 1791–12–08) = Stafford, Bob

9

92

Magnus Huber Steward, Joseph (B 107) Strong, John (C 1792–09–1) = Strong, John [Cherrystone] Thompson, Charles (B 31) Williams, William (B 100) Willoughby, Edward (B 63) Wilson, Thomas (B 117) SOUTH CAROLINA (41 emigrants = 21.8%) 1. (SILVER BLUFF), 1 27.4% (1790, data from Edgefield Co.) George, David (B 1) George, David (C 1791–10–25) = George, David (Grant 1992) [Silver Bluff] 2. ST. JOHN, BERKELEY, 1 88.2% (1790; 692 W : 5,170 B) Peters, Frank (B 130) = Peters, Frank [Moncks Corner] 3. ST. JAMES, SANTEE, 2 88.4% (1790; 437 W : 3,345 B) Collins, Jacob (B 106) = Collins, Jacob [Santee River] Jones, March (B 103) = Jones, March 4. ST. JAMES, GOOSE CREEK, 1 84.2% (1790; 439 W : 2,333 B) Withers, Timothy (C 1791–12–08) = Withers, Timothy 5. ST. MICHAEL, ST. PHILIP, 2 48.7% (1790; 8,089 W : 7,684 B) Addams, Syman (F 1792–01–12) = Addams, Simon [Ashley River] Ellis, Jack (B 98) = Ellis, Jack [Ashley River] 6. CHARLESTON, 28 53.7% (1770; 5,030 W : 5,833 B) Anderson, Isaac (B 4) Anderson, Isaac (F 1792–11–19) = Anderson, Isaac [Charleston] Ash, William (B 28) = Ashe, William [Charleston] Bacchus, Thomas (B 30) Baccas, Thomas (F 1792–11–12) = Bachus, Thomas [Charleston] Banbury, Lucy (B 60) = Banbury, Lucy [Charleston] Black, Effy (B 23) = Black, Effy [Strasburg Ferry] Channel, Thomas (B 118) Croy, Abraham (B 93) = Cry, Abraham [Charleston] Elliott, William (B 71) Evans, Thomas (B 72) = Evans, Thomas [Charleston] Hazel, Abraham (B 3) Jackson, Thomas (B 89) Jackson, Thomas (F 1792–11–19) = Jackson, Thomas [Charleston] Jones, Dinah (B 122) King, Boston (F 1792–11–19) = King, Boston [Charleston] Leslie, Mungo (B 91) = Leslie, Mingo [Charleston] London, Thomas (B 87) London, Thomas (F 1792–11–28) = London, Thomas [Charleston] Lusaine, Joe (B 21) McGregor, Henry (B 88) = McGriggor, Harry [Charleston] Perkins, Cato (B 52) Perkins, Cato (F 1792–11–19) = Perkins, Cato [Charleston] Peters, Hector (F 1792–11–28) = Peters, Hector [Charleston] Peters, Stephen (B 47) Peters, Steven (F 1792–11–19) = Peters, Stephen [Charleston] Petters, Sally (Fyfe 1953: 5) = Petters, Sally [Ashepoo] Primus, John (B 68) = Primus, John [Charleston]

The Nova Scotia–Sierra Leone connection Summers, Guy (B 97) Wigfall, Jacob (F 1792–11–28) = Wigfall, Jacob [Charleston] Williams, Samuel (C 1791–12–26) = Williams, Samuel [Charleston] Williams, William (B 57) Wiring, Joe (B 22) [Strasburg Ferry] Wright, Samuel (C 1791–12–08) = Wright, Samuel [Charleston] 7. BEAUFORT, 2 76.5% (1790; 4,364 W : 14,236 B) Ficklin, Stephen (F 1792–11–28) = Fickin, Stephen [Indian Land] Ramsey, Joseph (F 1792–11–28) ?Ramsey, John (B 11) = Ramsey, Joe [Indian Land] 8. ST. JOHN, COLLETON, 1 88.9% (1790; 585 W : 4,705 B) Birden, Cato (B 37) = Boden, Cato [Johns Island] 9. ST. LUKE, 1 76.5% (1790, data from Beaufort Co.; 4,364 W : 14,236 B) ?Pandavis, Lewis (B 61) = Pandorrus, Lymus [New River] 0. (UNIDENTIFIED SC), 2 Lynch, Jacob (C 1791–12–08) = Lynch, Jacob Linch, Pheby (F 1792–11–28) = Lynch, Phebe NORTH CAROLINA (6 emigrants = 3.2%) 1. DUPLIN, 1 22.0% (1786; 4,087 W : 1,153 B) ?Crowel, Abraham (B 141) = Cromwell, Abraham 2. BLADEN, 1 33.3% (1790; 3,350 W : 1,676 B) Williams, Henry (B 148) = Williams, Harry 3. NEW HANOVER, 2 46.5% (1786; 2,698 W : 2,344 B) Moore, Daniel (C 1791–12–12) = Moore, Daniel [Wilmington] Peters, Thomas (C 1791–08–6) petus, thomas (F 1791–12–23) = Potters, Thomas [Wilmington] 0. (UNIDENTIFIED), 2 Johnson, John (B 2) Johnson, John (F 1792–11–19) ?Williams, Joseph (F 1792–11–19) = Williams, Joseph UNIDENTIFIED “CAROLINA” (13 emigrants = 6.9%) Davis, Benjamin (B 83) Francis, Adam (B 35) Hog, Thomas (B 119) Irvin, Henry (B 79) Miles, Rinte (?) (B 138) Oakum, John (B 104) Smart, John (B 40) Spear, Silas (B 128) White, Samuel (B 69) Wilkins, Andrew (B 94) Williams, John (B 26) Wood, Abraham (B 105) York, Ishmal (B 150) GEORGIA (4 emigrants = 2.1%) 1. RICHMOND, 1 36.5% (1790; 7,162 W : 4,116 B)

93

94

Magnus Huber Moor, Andrew (F 1792–08–24) = Moore, Andrew [Augusta] 48.1% (1764 Augusta only; 540 W : 501 B) 2. CHATHAM, 2 77.0% (1790; 2,456 W : 8,201 B) ?Colbert, John (B 115) Cuthbert, John (F 1792–11–19) = Cuthbert, John [Savannah] Jenkins, Charles (B 82) = Jenkins, Charles [Savannah] 0. (UNIDENTIFIED), 1 Dickson, Richard (F 1792–05–15) = Dixon, Richard NEW YORK (6 emigrants = 3.2%) 1. ULSTER, 1 Cooper, John (B 49) Cooper, John (F 1792–11–19) = Cooper, John [Livingston Manor] 2. WESTCHESTER, 1 Halsted, Philish (F 1792–11–28) = Halstead, Phillis [Eastchester] 3. NEW YORK, 1 Benson, Yaff (C 1791–12–08) = Benson, Yaft [Harlem] 4. SUFFOLK, 1 Floyd, Henry (C 1791–12–08) = Floyd, Henry [Long Island] 0. (UNIDENTIFIED), 2 Dickerson, Samuel (B 131) = Dickison, Samuel [Philipse Manor] Marshel, Janey (F 1792–11–28) = Marshal, Jane [Philipse Manor] NEW JERSEY (6 emigrants = 3.2%) 1. ESSEX, 2 Manuell, John (F 1792–11–28) = Manwell, John [Newark] Richerson, Richard (B 134) = Richardson, Richard [Elizabeth Town] 2. SOMERSET, 1 Sloane, Richard (B 114) Sloane, Richard (C 1792–01–9) = Sloane, Richard [Rocky Hill] 0. (UNIDENTIFIED), 3 Arnold, Joseph (B 113) Brown, Mary (B 124) Ogden, Primus (B 137) PENNSYLVANIA (5 emigrants = 2.7%) 1. PHILADELPHIA, 1 Smith, Abram (F 1792–11–19) = Smith, Abraham [Philadelphia] 2. CHESTER, 1 ?Wansey, Nathaniel (1799–08–31) = Wandrey, Nathaniel [New Castle] 0. (UNIDENTIFIED), 3 Miles, Jacob (B 65) Morris, Robert (B 76) Townsend, John (B 75) Townsen, John (F 1792–11–19) = Townsend, John MARYLAND (2 emigrants = 1.1%) 1. CECIL, 1 Richardson, Hannah (C 1792–08–30) = Richardson, Hannah [Head of Elk] 2. (Eastern Shore), 1 Alleyne, Job (B 132) = Allen, Job [Eastern Shore]

The Nova Scotia–Sierra Leone connection DELAWARE (1 emigrant = 0.5%) 1. (Duck Creek), 1 ?Thomas, John (B 42) = Thomas, John [Duck Creek] MASSACHUSETTS (1 emigrant = 0.5%) 1. BRISTOL, 1 Laid, Lats (B 102) = Slaide, Lotche [Swansea] RHODE ISLAND (1 emigrant = 0.5%) 0. (UNIDENTIFIED), 1 Bowler, Arther (B 149) = Bowles, Arthur CARIBBEAN/BERMUDA (7 emigrants = 3.7%) 1. ANTIGUA, 1 Wearing, John (B 59) = Warren, John 2. BARBADOS, 1 Wright, Charles (B 109) 3. BERMUDA, 2 Godfry, Thomas (B 38) Jones, James (C 1792–08–24) = Jones James 4. ST. CROIX, 1 Beverhoudt (F 1792–06–26) = Biveroudt, Henry 5. JAMAICA, 2 Linas, Cato (F 1792–11–28) = Linus, Cato [Kingston] Stephensen, John (F 1792–08–31) = Stephenson, John

95

CHAPTER 5

The development of variable NP plural agreement in a restructured African variety of Portuguese Alan N. Baxter

Universidade de Macau

This research examines the growth of variable plural (PL) agreement in the noun phrase (NP) of the restructured Portuguese of the Tongas, descendants of Africans contracted on the Monte Café plantation of São Tomé in the 19th and 20th centuries. I first provide a sketch of the sociohistorical background of the speech community, stressing that its Portuguese was acquired in a special contact situation along with simultaneous acquisition of Umbundu. Subsequently I present the results of a VARBRUL analysis of plural NP items based on data collected from three age groups, the oldest age group consisting of first- and second-generation speakers born on the plantation. Four conditioning factors are examined: (1) morphophonological saliency of the plural word; (2) the following phonological context; (3) the form class of the plural item and its position within the NP; and (4) whether informants had at least one African parent, or locally-born parents. Particular attention is paid to the development of PL marking within the NP structure. The left position adjacent to the head is found to be prominently marked, and is the point of entry of PL in the oldest age group. The analysis presented raises the possibility that an Umbundu transfer may underlie this configuration, and suggests that the growth of PL marking across the three age groups is closely linked to a growth in NP structure. The appearance of variable PL marking in L1 Tonga Portuguese is attributed to the past role of L2 Portuguese data in L1 acquisition.

1. Introduction* This paper concerns the Portuguese of the Tongas of São Tomé (West Central Africa), the descendants of Africans indentured during the Portuguese colonial system. First described by Rougé (1992), this fascinating variety of Portuguese has much to contribute to the study of contact situations, especially those involving Portuguese. Whereas Standard Portuguese has rigid agreement rules applying to items within plural NPs, Tonga Portuguese (henceforth TP)¹ has variable plural agreement. Thus, examples such as (1), in which plural marking occurs on all items

98

Alan N. Baxter

in the NP, occur in variation with examples such as (2),² in which it is incomplete: (1) Os ôtros disse the[M]-PL other-M-PL say-PRT-3S³ ‘The others said (so).’ H3DUA (2) issos coesa tudo.⁴ this[NEU]-PL thing-[F]-S all[NEU] ‘all these things.’ H3DUA This paper has two aims. Firstly, it sketches the socio-linguistic background of the Tongas, concentrating on the context for language acquisition. Secondly, it identifies the factors whereby variable plural agreement develops and spreads. This is addressed though a discussion of the profile of variable NP plural marking in the TP of the Monte Café plantation.

2. Background: indentured labor on São Tomé plantations TP has its origins in the substantial socioeconomic and demographic changes that swept São Tomé in the second half of the 19th century.⁵ The introduction of largescale cacao and coffee planting coincided with the abolition of slavery on the island, and this event provoked a severe labor shortage. In 1854, slaves indigenous to São Tomé, were liberated.⁶ Foreign slaves were eventually freed in 1876 (Neves 1929: 55). Foreseeing these events, some planters began importing indentured labor from Angola in the early 1850s, (Nascimento & Gomes Dias 1989). In due course, they drew on English and French colonies and the Portuguese colonies of Guiné, Ajudá (in modern Benin)⁷ and, eventually, Mozambique and the Cape Verde islands. Part A of Table 1 (based on Neves (1929: 63) summarizes worker numbers and origins for 1876–1879, showing a clear Angolan prominence. Neves (1929: 63–4) also reports an Angolan majority of 7,419 in 1881, and 10,411 during 1885–1892. Part B of Table 1 shows that the Angolan prominence persisted into the early 20th century. The Mozambicans, however, began arriving only in 1901 (Neves 1929: 67).

Table 1. Indentured laborers entering São Tomé A. Period of 1876–1879 Angola

Guiné

Ajudá

5,249

1

24

Cameroon/ Gold Coast Gabon 3,229 738

Sierra Leone

Liberia

TOTAL

39

516

10,341

B. Period of 1901–1928 Angola 50,444

Mozambique 43,053

Cape Verde 6,305

Other origins 19

TOTAL 99,821

Variable NP plural agreement in a restructured African variety of Portuguese

Two related facts are especially relevant to the origins of Tonga Portuguese. Firstly, prior to 1909, no Angolans were repatriated, although workers of other origins were (Cadbury 1910: 45–6).⁸ Secondly, while most workers were repatriated in the 20th century, their São Tomé-born offspring, the Tongas, became the de facto property of the plantations.⁹ After decolonization in 1974, many Tongas remained on the plantations where they were born.

2.1 Languages used on the plantations The Tongas today speak their own variety of Portuguese and display varying degrees of maintenance of their (often koineized) African languages (Rougé 1992). Knowledge of São Tomé Creole Portuguese (henceforth Sãotomense) is not widespread among the Tongas (Rougé 1992: 173). This is especially true for those who reside on outlying plantations.¹⁰ Thus, among the informants interviewed during this project, only two younger speakers had some knowledge of Sãotomense. The Tongas’ linguistic repertoire stems from their particular socio-linguistic history. The language of communication between overseers and workers on the plantations was not Sãotomense. It was Portuguese (Rougé 1992: 173), presumably with elements of African languages.¹¹ Sãotomense appears to have been absent from the plantations for at least five interconnected reasons: (i) Following abolition in the 19th century, according to Mantero (1910: 17), former slaves simply abandoned the plantations en masse. Subsequently it appears to have been a widespread policy that the local Creole population was not encouraged to work on the plantations. This is noted in Cadbury (1910: 17– 18, 26) and in the testimonies of our oldest informants. In the 20th century, until the late 1950s, most plantation owners seem to have chosen not to employ the locals, and even to have prevented their entry to the plantations.¹² (ii) Most overseers seem to have been European. This is reflected in the observations of Cadbury (1910: 30) and, for the Monte Café plantation, both in ANON (1895) and in the testimonies of elderly informants. During the 20th century, however, some overseers were Tongas. (iii) The huge numbers of indentured workers continued to speak their ancestral languages. The maintenance of African languages among their descendants, the Tongas, related to the nature of plantation administration. Workers were housed separately, and assigned to work-groups according to their origins. (iv) The hermetic nature of the plantation system did not favor outside contact. Until the mid-1940s, the indentured workers were largely confined to the plantation precincts. The worker’s day was strictly regimented, beginning with a roll call, followed by supervised work, and ending with a curfew. Many plantations had their own shops, and until the 1940s, the workforce had only one half-day off per week. Indentured workers and Tongas who were associated with the administration of the plantation would have had greater access to

99

00 Alan N. Baxter

off-plantation contexts. Yet, the average indentured worker prior to the 1940s required a permit to leave the plantation area.¹³ (v) Finally, geographical factors could disfavor off-plantation contacts. Such factors are relevant in the case of the Monte Café plantation, situated on the shoulder of a mountain in an area that was covered with forest until 1854 (Mello 1864). Even today Monte Café lies beyond the main Creole settlements. Thus, on many plantations Sãotomense may not have been a significant factor for the development of Tonga Portuguese, unless (i) it was present during the establishment of the new plantations in the 19th century; or, (ii) a plantation had employed local Creoles at any subsequent stage. For the Monte Café plantation, neither possibility can be completely discounted. However, this plantation was founded in 1854, and its owner began to import Angolans in the same period. Although no mention of the use of local labor has yet been located in surviving documents, the possibility remains open.¹⁴ Subsequently, for the 20th century, to judge from plantation paysheets, and the testimonies of old Tongas, local Creoles were not employed as field laborers until the mid-1950s, and even then, they did not reside on the plantation. If Sãotomense did not play a direct role in the development of Tonga Portuguese in the early stages of the plantation system, could the local São Tomé variety of Portuguese have exerted some influence on TP? São Tomé Portuguese is strongly influenced by Sãotomense and has features in common with TP (see Section 2.2.). Thus, Sãotomense could have indirectly influenced TP via local São Tomé Portuguese. Indeed, any communication between plantation workers and outsiders would have been more viable in Portuguese than in Sãotomense. A channel for such communication could have been contacts between outsiders and plantation workers who were more closely connected with administration. However, for many plantations in the pre-1940s period, the restrictions on outside contact are also relevant for contacts with local São Tomé Portuguese.¹⁵ After the 1940s, with greater possibilities of off-plantation movement of workers, and the presence of local workers on the plantations as of the mid-1950s, São Tomé Portuguese came into more frequent contact with Tonga Portuguese. Also in this period, Tonga children came to have general access to off-plantation education in Portuguese.¹⁶

2.2 The speech community on the Monte Café plantation On Monte Café, Angolan workers (from the Benguela region) were imported from the early 1850s until the 1870s (Nascimento & Gomes Dias 1989: 52, 68), and from the 1880s (ANON 1885)¹⁷ into the 1890s again largely from the Benguela region (Curadoria 1895–1896).¹⁸ This strong Angolan presence was to continue in the first half of the 20th century, as is evident in Table 2. For language acquisition, the consequences of this overwhelming African presence, and longstanding Angolan connection, were exceedingly complex. Table 2 indicates that, throughout this selection of monthly pay sheets for 1935–45, the over-

Variable NP plural agreement in a restructured African variety of Portuguese Table 2. Monte Café workforce 1935, 1944 paysheets

May 1935 April 1937 April 1939 July 1942 May 1945

Angola

Mozambique

Cape Verde

Ajudá

Tonga

Children

F

M

F

M

F

M

F

M

F

M

F

M

174 153 291 366 197

37 37 36 42 24

234 234 208 186 168

3 3 3 1 1

1 1 0 8 14

1 1 1 15 14

0 0 0 0 0

2 2 2 1 1

52 68 65 63 55

59 109 107 111 97

53 59 60 83 88

43 61 54 91 91

Source: Curadoria: Folhas de pagamento — roças ‘Office for welfare of indentured workers: paysheets — plantations’, São Tomé and Príncipe Historical Archive.

whelming majority of the adults were continental Africans. It may be assumed that, alongside the African languages, a continuum of varieties of L2 Portuguese were current, as were L1 varieties of TP. Informants born in the early 1900s testify to several linguistic contexts according to parental origin: , (for example from Ajudá, Cape Verde or Mozambique),¹⁹ , and . On the one hand, the first language acquired by the offspring of Angolan parents was a Bantu language, probably a variety of Umbundu. This language has been retained in a koineized form among the Monte Café Tongas. Portuguese was acquired mainly through contact with other adults and children. This occurred in three main scenarios: (i) in the field (as mothers generally took their infants to work); (ii) in the precinct of the administrative area where children conducted yard duties supervised by an indentured worker or a Tonga²⁰); and (iii) in the context of the housing complex. In each situation, children would have heard Tonga L2 and L1 Portuguese, and they would also have heard L1 European Portuguese. In the field and in the housing contexts they would have also heard and used Umbundu. In the plantation yard and housing contexts, children could also have heard and used Cape Verde Creole. The children of African parents of different origins acquired the Angolan mother’s language (which in some cases was the father’s L2), and they acquired ‘Portuguese’ based on the L2 Portuguese spoken by the parents, and also through their exposure to Portuguese in the contexts beyond the immediate influence of the parents in the context. None of the first generation Tongas interviewed acquired African languages other than Umbundu from their parents. Finally, the offspring of Tonga couples, or of Tonga plus African couples, appear to have acquired both the Umbundu koiné and Portuguese simultaneously. In this case one principal source of primary linguistic data (henceforth PLD) was Tonga L1 Portuguese. However, just as with the children of African couples, the contexts

0

02

Alan N. Baxter

beyond direct parental influence must also have affected the acquisition of both Portuguese and Umbundu. With regard to the other languages present among the workforce (from Ajudá, Mozambique and Cape Verde), it is interesting to note that the Tongas on Monte Café display only minimal evidence of acquisition of these languages. For the population that provided the data for the present study, only two informants of the middle age group claimed to have limited L2 knowledge of a Mozambican language (spoken in Quilimane), acquired as young men. Furthermore, no informant in the population sample studied here speaks Cape Verde Creole. The historical context for the acquisition of Portuguese by the oldest of the three generations in the database to be analyzed may be summarized as follows: (i) the predominant model of Portuguese among the huge worker population was workforce Portuguese L2, which, considering the constant cyclic nature of the indenturing process (with constant new arrivals of African monolinguals contracted for three to five years), must have always existed in a continuum of interlanguage varieties; (ii) children of Generations 1 and 2 learned the predominant Umbundu-based koiné prior to learning Portuguese; (iii) children learned Portuguese in the field and in the plantation compound, mainly through contact with workers’ L2 Portuguese, less so with workers’ L1 Portuguese, and marginally through contact with European overseers. The L1 Tonga Portuguese of first generation Monte Café Tongas displays substantial and consistent morphosyntactic restructuring in the NP (in referencing, number, and gender representation), in the VP (tense–aspect and person–number marking), and in the representation of case relations and negation. Those features of L1 TP display many similarities with the grammars of Creole Portuguese varieties (e.g., Cape Verde Creole Portuguese and São Tomé Creole Portuguese; see Baxter 2002). In other respects, especially in the leveling and development of agreement rules as variable rules, TP resembles the vernacular varieties of Portuguese of Brazil, Angola, Mozambique and Macao. Variable NP number agreement is also found in São Tomé Portuguese, especially among speakers from lower socioeconomic backgrounds. The following sections discuss the development of NP plural marking across three generations of Tonga Portuguese.

3. The data for the present study Fieldwork in São Tomé was conducted during February-May, 1998. Three fieldworkers were involved: the author of this paper, co-researcher Dr. Dante Lucchesi (Universidade Federal da Bahia) and Nara Barreto (Universidade de Macau). Contact with the Tonga communities was established through the kind assistance of local Tonga politician Albano de Deus. Our Tonga field assistant at Monte Café, Fran-


cisco Paulino, conducted invaluable work with the interpretation of the interviews and transcription. Research assistants Joseli Querino (Universidade Federal da Bahia) and Dr. Norma Lopes (Universidade do Estado da Bahia) helped with the codification and initial analysis of the data. At the Monte Café plantation, Labovian-type socio-linguistic interviews of 1.5 hours average were recorded with 20 Tongas in three age groups (Group 1 = 8 speakers) 61+ years,²¹ (Group 2 = 6 speakers) 41–60 years, and (Group 3 = 6 speakers) 20–40 years, equally distributed by sex. The interviews covered a range of topics relating to plantation life, including specific incidents in the history of the plantation, and the lives of the informants. In addition, a brief ethnographic questionnaire was administered. All informants live on Monte Café and most also work there.²² Only the informants in the youngest age group, and two from the middle age group, had education at the elementary level.²³ Henceforth, these three age groups are referred to as Generations 1 (60+ years), 2 (41–60 years) and 3 (20–40 years).

4. Methodology and preliminaries to analysis As a vast body of research has been conducted on variable NP plural agreement in Brazilian Portuguese, it was decided to follow the analytic framework developed in that area. This facilitates substantial comparisons and, in many respects, avoids “reinventing the wheel”. The current study builds on the works of Guy (1981), and especially on that of Scherre (1988) and of Lopes (2001), all of which were conducted within a variationist framework using the VARBRUL programs (Pintzuk 1988). Following Lopes (2001), all NPs with plural reference were codified for the dependent variable <presence of the plural marker>, and for a range of independent factor groups: (i) saliency of the morphophonological form in terms of the type of plural formation and stress, (ii) the phonological context following the place of plural insertion, (iii) the presence of plural markers prior to the item being coded, (iv) the grammatical class of the pluralizing word, (v) the functional class of the pluralizing word and its position relative both to the head of the NP and to its linear position within the NP, (vi) age of speaker, (vii) sex of speaker, (viii) African parentage, and (ix) education.²⁴ The reader should note carefully the nature of the complex factor groups (i) and (v). On the one hand, factor group (i) <saliency> permits the analysis of all the different Portuguese morphophonological plural types, and all major stress patterns. My approach permits a wide range of analytical comparisons of the differential effect on plural marking according to the form of the word — for example, the effect of metaphony, or whether a singular word ends in a vowel or a consonant.²⁵ On the other hand, factor group (v) combines three qualities. It addresses the effect of linear position in the NP while simultaneously addressing the effect of the functional class of the pluralizing word (as NP head or non-head) and, for non-nuclear items, the effect of their position relative to the NP head. This factor group is thus strongly

03

04

Alan N. Baxter

syntactic in nature and facilitates a wide range of insights into plural marking, including the interaction of plural marking with bare NPs. Discussion of the rationale behind this type of combined factor group for analyzing Brazilian Portuguese data may be found in Scherre (1988, 1998) and Lopes (2002).²⁶ For the present paper, three factor groups were not examined. Firstly, <education> was not considered because it was categorical for the youngest age group. Secondly, in initial analyses the was identified as not exerting a significant influence on the dependent variable <presence of plural marking>. Thus, it was excluded from the further analyses conducted here. The third factor group omitted is the <presence of plural markers prior to the item coded>. This factor group tested as exerting a significant effect on the plural marking of subsequent pluralizing items in the same NP. However, it is not discussed here because this paper concentrates on the effects relating to the immediate structural points at which plural marking is introduced.

5. Overall variability and the profile of development The data comprise 3,366 NP items with plural reference. Of these, 1,835 items (55%) carry the plural morpheme, which is attached with an input probability of 0.55. The broad profile of the development of the NP plural agreement rule across three age groups is demonstrated in Table 3, showing the results of the analysis for the factor group .²⁷ Given the socio-historical context of the speech community investigated, it is non-controversial to claim that the corpus collected represents the following development: Generation 1 is minimally disposed towards plural marking. Indeed, the oldest of the speakers within this group display an exceedingly low frequency of marking. It seems reasonable to assume that this reflects the nature of predominant PLD available to this generation (and prior generations) of language learners. That type of PLD would have been heavily influenced by L2 Portuguese acquired by African adults, and there was insufficient input to set NP number agreement Table 3. The effect of on PLURAL marking of NP items a Significance: p = .002b Log Likelihood = −1152.857

Generation

Items with PL marking

Total of plural items

% PL marking

Probability of PL marking

1– (61+ years) 2– (41–60 yrs) 3– (20–40 yrs)

191 524 1120

779 1118 1469

25% 47% 76%

0.11 0.42 0.8

a b

In this and subsequent tables, only cells with 10 or more data items were processed by VARBRUL. Only levels of significance below .05 are acceptable.

Variable NP plural agreement in a restructured African variety of Portuguese Table 4. The effect of the on PLURAL marking of NP items Significance: p = .029 Log Likelihood = −611.141

Origin of parents

Items with PL marking

Total of plural items

% PL marking

pr weighting of PL mark

Africa Monte Café

275 440

914 983

30% 45%

.36 .63

(henceforth Agr). Successive generations, with exposure to PLD of a different nature, and with a growth in contact with L1 varieties of Portuguese as the plantation infrastructure expanded and the colonial society changed, display acquisition of Agr as a variable rule. Additional support for the relevance of Adult L2 Portuguese in the PLD for Generation 1 is suggested by a further analysis of the data from Generations 1 and 2. Here, an additional factor group was included, classifying speakers according to whether their parents were African born (i.e., adult L2 learner-speakers of Portuguese) or born on Monte Café plantation (i.e., ostensibly L1 speakers of Portuguese, having acquired L2 before the onset of puberty). This factor group was selected by VARBRUL as exerting a significant effect on plural marking. The results are presented in Table 4. Speakers whose parents were born on the African continent disfavor plural marking, with a probability weighting (henceforth pr.) of only 0.36, whereas speakers whose parents were born on the Monte Café plantation favor plural marking, with a pr. of 0.63. As the situation depicted in Table 4 is clearly developmental, involving the acquisition of plural marking by this population, one expects to find in the data evidence of how and where the development of plural marking gradually spread in the grammar of each age group. The following sections discuss three linguistic variables which tested as exerting a significant effect on the use of the plural marker: the morphophonological saliency of plural marking, the phonological context following the plural item, and the relative positions and status of structural elements of the NP.

6. Saliency, and the following phonological context The role played by saliency in contrasting forms has long been considered relevant to the orientation of variation in agreement rules in Brazilian Portuguese, and it figures prominently in studies of variable plural marking (Guy 1981, Scherre 1988). However, saliency has also been included in the study of acquisitional processes in depidginization (Emmerich 1984), decreolization (Bickerton 1975: 142), and the

05

06

Alan N. Baxter

areas of first language acquisition (Ingram 1989: 437) and second language acquisition research (Bayley 1994: 170–72, Wolfram 1985). The latter three cases consider the potentially facilitative effect of saliency in the acquisition process and in the association of form with function. Saliency was contemplated in the present study both in relation to the phonetic distance between singular and plural forms and in relation to stress-class. The analysis considered six saliency factors relating to the phonetic nature of the plural distinction and stress patterns. Four factors consider the presence or absence of stress on the plural syllable together with an accompanying change in the root of the plural word. A fifth factor concerns unstressed monosyllables. A sixth factor caters for plurals whose corresponding singular ends in an -s.²⁸ Table 5 exemplifies these factors and presents the results of the analysis of this factor group in four separate VARBRUL runs: one for the entire Tonga data set, and one each for the three generations.²⁹ Generation 1 stands apart, with a very different pattern of variation than those of Generations 2 and 3. This is expected, as Generation 1 would have received a heavier input from L2 PLD, from adult African L1 speakers. Indeed, it is in Generation 1 that one might expect to find reflexes of morphological leveling and of transfer phenomena, which are typical of early stage L2 acquisition by adults. The forebears of Generation 1 were predominantly of southern Angolan Bantu stock. Their L1s were agglutinating languages that indicate plural through prefixal classifiers, in a concordance spreading from the noun classifiers. In such languages, the word structure is predominantly paroxitonal (penultimate syllable stress), and they are open syllable dominant. So, when such speakers acquire Portuguese as L2 through contact, two effects are expected. Firstly, they are likely to opt for unmarked settings in the construction of interlanguage, and so will the subsequent generation in L1 acquisition. Secondly, they are expected to display some transfer effects that impede their acquisition of plural morphology. Table 5 clearly shows that the more salient oppositions in Portuguese play an important role in the development of plural morphology originating from a system that initially had a very limited use of plural marking.

6.1 Saliency factors 6.1.1 Plurals attaching to a stressed final syllable

In Generation 1, the entry point for plural marking is through words with a stressed final syllable where attachment of the plural does not involve a change in the root of the word (pr. 0.93), e.g. cafés.³⁰ This is an optimal perceptual context from an acquisitional perspective. At the same time, the presence of this salient evidence in L2 PLD from adults also makes sense. The predominant L1 of the adults, Umbundu, is strongly penultimate-syllable stressed (Valente 1964: 27), so Portuguese words with final syllable stress should stand out for the learner as being markedly different. In the transition from Generation 1 (pr. 0.93) to Generation 2 (pr. 0.57), the effect of final-syllable stress in words whose plural does not require a root change is neu-

N PL

21%

10/48

0.26

50%

244/753 32%

7/14

0.40

.

59%

88%

% PL

191/194 98%

50/84

0.93

0.44

22/25

N PL

.

% PL pr. PL

0.35

0.39

0.51

0.82

0.57

0.93

pr. PL

Significance = .005 Log Likelihood = -413.246


1. Stressed PL + root change 0/5 0% - Singular in -l#: fácil ‘easy (S)’; faceis ‘easy (PL)’ - Singular in irregular -ão#: caminhão ‘truck’; caminhões ‘trucks’ 2. Stressed PL + no root change 24/111 22% café ‘coffee’; cafés ‘coffees’ irmão ‘brother’; irmãos ‘brothers’ 3. Monosyllable 103/122 84% o ‘the (S)’; os ‘the (PL)’ 4. Unstressed PL + root change 0/2 0% - Metaphonic plural [ô]vo ‘egg’; [ó]vos ‘eggs’ 63/465 14% 5. Unstressed PL + no root change - Word with penultimate syllable stress: casa ‘house’; casas ‘houses’ - Word ending in -r : jantar ‘dinner’; jantares ‘dinners’ - Word with antepenultimate syllable stressed: indígena ‘native’; indígenas ‘natives’ 1/73 1% 6. Singular in -s país ‘country’; países ‘countries’

Type of plural

Generation 2

Generation 1

Table 5. The effect of <saliency of plural formation> on plural marking in the NP

80% 0.45

93% 0.84

% PL pr. PL

77% 0.66

41/85

48% 0.28

653/932 70% 0.40

10/13

317/319 99% 0.79

74/93

25/27

N PL


Generation 3

59%

96%

51%

82%

% PL

52/207

25%

960/2150 45%

17/29

611/635

148/288

47/57

N PL

0.35

0.46

0.59

0.61

0.52

0.88

pr. PL


All three generations

08

Alan N. Baxter

tralized; and in Generation 3 (pr. 0.45), this factor is slightly unfavorable to marking. In the transition from Generations 2 through 3, as other saliency classes spread plural marking, the dominant saliency role is assumed by words which have final syllable stress and which require a root change when plural is attached.

6.1.2 Monosyllables

Monosyllabic words are slightly unfavorable to plural attachment in Generation 1 (pr. 0.44). Why? Two factors are plausibly involved here. Firstly, there is a tendency in Generation 1 to omit monosyllabic determiners or to replace them with innovative forms. Definite reference NPs are common without monosyllabic articles, as in example (3): (3) Sabe inda dia dumingo elé lanja cumida know-PRS-3S still day[M]-S Sunday[M]-S he arrange-PRS-3S food[F]-S pa muiere veio. for women[F]-PL old-M-S ‘You know, even on Sundays he used to get food for (the) old Tonga women (i.e., here on Monte Café).’ M3ALB2–569

Furthermore, in Generation 1, the function of definite article can be performed by TP, isso ‘this, that (with specific reference [masculine or feminine gender], or with indefinite global reference)’.³¹ This is a role that the definite articles o, a ‘the’ (respectively, masculine and feminine singular), and os, as ‘the’ (respectively, masculine and feminine plural) would perform in Portuguese. Example (4) demonstrates this innovation: (4) Mas teve muita confusão aqui?³² but exist-PRT-3S much-F-S confusion-[F]-S here.Q ‘But was there much confusion here?’

{Res.}

Muito! Gente no trabaia. Se você tá much-M-S.EX people/we[F] NEG work-PRS-3S if you be-PRS-3S trabaiá venha batê você. {Inf.} work-INF come-SU hit you ‘A lot! People/we didn’t work. If you were working (they) came (and) beat you.’ Hum. EX ‘Hmm.’

{Res.}

Isso foro venha batê você. this[NEU]-S creole-M-S come-SU-3S hit-INF you-S ‘The forros (i.e., the Sãotomenses) came (would come) and beat you.’ M3SIL300–3

{Inf.}


Also, in place of Portuguese <preposition+article+number> inflecting compounds, such as nos/nas ‘in/to the’ (< P. em+o/a+s [= to/in+the (masculine gender)/the (feminine gender)+plural]), one finds (variably) uninflecting ni (or na) as a locative/directional (to or from) preposition occurring with NPs without definite articles,³³ as in example (5): (5) Oto branco landja fora, ni foro, pa ficare other-M-S white-M-S arrange-PRS-3S outside in foro-M-S to stay/be-INF cu ele. with he ‘Other Europeans arranged (them) outside, among the forros, to stay with them.’ M3SIL1102 Compounds such as dos/das alternate with the use of uncontracted de introducing NPs without articles. (6) Quando contro isso mulato, mulato, when find-PRS-1S³⁴ this-[NEU]-S mulatto-M-S mulatto-M-S mulato, mulato, é di branco que tinha mulatto-M-S mulatto-M-S be-PRS-3S of white-M-S that have-PST.IMP-3S aqui. here ‘When you find these mulattoes, they are from the Europeans who were here.’ M3SIL1096 However, a second motive for the slightly unfavorable effect of monosyllables on plural marking might be sought in the adult L2 Portuguese influenced PLD for Generation 1: Umbundu has few monosyllables, and functionally, those that are not agglutinating bases tend to be prepositions or adverbs. In contrast, in Generations 2 (pr. 0.82) and 3 (pr. 0.79), the monosyllable class assumes notable importance as a significant factor for plural attachment. The explanation would appear to lie in the broad difference in plural marking strategies between Generation 1 and the other two generations. The factors just mentioned, the reliance on alternative article/determiner forms, such as, isso and uma, and the nonagglutinated alternatives to agglutinated Portuguese <preposition + article> forms, are more characteristic of Generation 1. The difference in Generation 2 might be interpreted as a reappraisal of the ability of monosyllables to be plural marked as the definite article of Portuguese is acquired.

6.1.3 Unstressed plural syllable + a change in the root

Metaphonic plurals yielded little data and no PL attachment in Generation 1, yet they display acquisitional development from Generation 2–where they are only marginally favorable to plural attachment (pr. 0.51) — to Generation 3, where they are reasonably favorable in relative weighting (pr. 0.66).

09

0

Alan N. Baxter

6.1.4 Unstressed plural syllable + no change in the root

Across the three generations, this factor is consistently unreceptive to the plural morpheme, with pr. weightings of 0.40, 0.39 and 0.40, respectively. Recall that there are four word types involved here: words with penultimate syllable stress, words ending in -r and -l, and words stressed on the antepenultimate syllable. In Portuguese, words with penultimate-syllable stress constitute the largest stressclass and the least salient opposition phonetically. In the data base considered here, this class of words accounted for over 93% of the factor . It might be argued that the saliency status of this class of word, at least for Generation 1, is further compromised by the fact that Umbundu words are stressed on the penultimate syllable and have open syllables. Where words ending in -r and -l are concerned, Generation 1 registered no cases of PL attachment. Variable PL attachment occurs only in Generations 2 and 3, with low data numbers. Some speakers, especially in Generation 1, show evidence of treating superstrate words ending in -r as ending in a vowel, so that such words are likely to be treated as having penultimate syllable stress. Finally, words with antepenultimate stress registered no data in Generation 1, and variation in Generations 2 and 3 with low data numbers.

6.1.5 Singulars in -s

Words ending in -s are quite disfavorable to plural attachment in the three generations. This negative effect is strongest in Generation 1 (pr. 0.26), somewhat weaker in Generation 2 (pr. 0.35), and again slightly stronger in Generation 3 (pr. 0.28). From an acquisitional perspective, the fact that <s> disfavors attachment of plural morphology may be interpreted either as an analogical factor, the speaker associating a final -s with plurality, or to haplology.

6.1.6 Overview of saliency

The effect of saliency on PL attachment over the three generations is reminiscent of the findings of acquisitional studies, where a growth of inflectional morphology has been observed to initially involve salient oppositions (Bayley 1984: 170–2). The least salient oppositions are acquired last. Consideration of the VARBRUL run of the three amalgamated generational data sets gives an overall view of the relative strengths of the different saliency factors for this speech community. The effect of saliency in acquisitional terms is appreciable. On the one hand, agreement is favored by the morphophonological distinctions of the more marked oppositions, the combination of a stressed pluralizing syllable and a root change associated with PL being the most potent (pr. 0.88). Next in strength come the monosyllables (pr. 0.61), reflecting the acquisition of the article in Generations 2 and 3, followed by metaphonic plurals (pr. 0.59) and words with a stressed pluralizing syllable yet no root change (pr. 0.52). On the other hand, it is disfavored by the least salient classes — words with no root change and an unstressed pluralizing syllable (pr. 0.46), and words with a singular in -s (pr. 0.35), the latter severely compromised by their final consonant.


In sum, in analyzing saliency for this population of speakers, one would expect to find some evidence of transfer phenomena, and especially evidence of acquisition guided by saliency. Indeed, this is the case. The most salient oppositions are acquired earlier, whereas the least salient are acquired later, and this process is reflected in the speech community’s profile of variation (in the analysis of the data of the three generations together).

7. Following phonological context Another dimension concerning phonological constraints on the development of plural marking, and on the pattern of variation observed in this community, relates to the phonological context following the plural item. Table 6 (p. 112) shows the relative probability weights of the effect of the on the dependent variable. Beginning with Generation 1, it can be said that several of the phonological constraints might be explained in terms of acquisitional tendencies and transfer influence in the L2 Portuguese of African adults who provided PLD to Generation 1. Here, the strongest point of entry for plural attachment is when the plural item is in the final position of the NP voicing group (pr. 0.69). The next most propitious contexts are: a following voiceless consonant (pr. 0.66), and a following pause within the NP (pr. 0.58). On the other hand, plural attachment is strongly inhibited by a following voiced consonant (pr. 0.21), although a following a vowel exerts an almost neutral effect (pr. 0.46). The cause may be the voicing rule in Portuguese which applies to /s/ prior to voiced consonants and vowels, which Umbundu speakers would have heard from European Portuguese overseers. Part of this profile may be accounted for by adult L2 input to PLD, because morphological leveling has occurred, and the L1 of these adults displays a strong open syllable phonotactic typology, thus imposing strict constraints on sequences of consonants. Thus, a pause after a plural marker would constitute a perceptually salient phonological context for that marker, especially a pause at the end of the NP voice group. The fact that a following voiced consonant or a vowel strongly inhibits marking may have some basis in Umbundu phonotactics. In Umbundu, there are no consonant sequences, but rather prenasalized stops. Furthermore, Umbundu does not have [z] in its sound inventory (Valente 1964: 383–4). This may have yielded a restriction on the sequence PLURAL + VOICED CONSONANT OR VOWEL in the Portuguese interlanguage of L1 Umbundu speaking adults. In acquiring L2 Portuguese, Umbundu speakers would experience more difficulty in inserting /s/ before a voiced consonant or before a vowel. On the other hand, the production of a sequence where the second item is unvoiced appears to have been unproblematic and was evidently strong in PLD input to Generation 1.



3. Vowel meus avós ‘my grandparents’

2. Voiced consonant as loiça ‘the crockery’

1. Voiceless consonant pra mias tias ‘for my aunts’

H1JOA

H1JOA

H1JOA

H1JOA

N PL


731/1026 71% 0.53

Generation 3

N PL.

57% 0.40

Generation 2

N PL

424/493 86% 0.55

445/779

54% 0.48

Generation 1

% PL pr. PL

209/324 65% 0.49

270/338 80% 0.45

327/603

36% 0.58

% PL pr. PL

N PL

47% 0.66

118/247 48% 0.39

202/290 70% 0.41

167/459

33% 0.54

Significance = .000 Log Likelihood = −1434.972

98/209

29% 0.21

102/194 53% 0.50

101/149 68% 0.60

160/481

% PL pr. PL

57/194

19% 0.46

58/180 32% 0.64

120/192 63% 0.54


23/119

6% 0.69

36/167 22% 0.53

% PL pr. PL

8/130

3% 0.58


4/122


Table 6. The effect of the on the use of the PLURAL morpheme in plural NP items

4. Structure-final pause dezesseis anos ‘sixteen years’

H1JOA

Following phonological context

5. Structure-internal pause pra mias tias, irmãs dele ‘for my aunts, his sisters’


7.1 The development of PL marking and its overall profile Several factors operating in Generation 1 carry over to the two subsequent generations — three factors favoring marking, and one strongly disfavoring it. The final pause context continues to be the prominent favorable context in Generations 2 (pr. 0.64) and 3 (pr. 0.60), although decreasing slightly in the latter. The slightly favorable weighting of the internal pause weakens in Generations 2 (pr. 0.53) and 3 (pr. 0.54). The context of a following voiceless consonant, the second most propitious context for marking in Generation 1, shows a similar decline to just below neutral effect in Generation 2 (pr. 0.49), and then increases in Generation 3 (pr. 0.55) to be only slightly favorable to marking. Next, the strongly unfavorable factor of Generation 1, a following voiced consonant, continues to disfavor marking, although it gradually weakens across the three age groups (with pr. weightings of 0.21, 0.39 and 0.45 respectively). In contrast, the following vowel context holds a surprise: although it decreases only slightly in weight from Generation 1 (pr. 0.46) to Generation 2 (pr. 0.50), as a neutral effect on marking, in Generation 3 (pr. 0.41) it has become an unfavorable factor.

8. The structural configuration of the NP To assess the effect of the structural configuration of the NP on plural attachment, I follow the approach adopted by Scherre (1988, 1998) and Lopes (2001), who both build on the earlier work of Guy (1981). These researchers propose a factor group that contemplates the form class, position relative to the head, and linear position. In the present study, the data were codified for eleven configurational patterns. These are exemplified with Tonga data as follows:³⁵ (7) Plural item at the left, adjacent head Os meus AVÓS nasceram aqui. the[M]-PL my-M-PL grandparent-M-PL born-PRT-3PL here ‘My grandparents were born here.’ H1JOA (8) Plural item at the left, not adjacent head Os meu PAI vivia. the[M]-PL my-M-S father-[M]-S live-IMP-3S ‘My parents were alive.’ H2ADR (9) Plural item is head in first position in NP OLHOS abertos assim. eye[M]-PL open-M-PL thus ‘Eyes open like this.’ H1FRANJU

3

4

Alan N. Baxter

(10) Plural item is head in second position in NP os HOMENS grande. the[M]-PL man[M]-PL big-S ‘the big men.’ H2JUST (11) Plural item is head in third position in NP os meus VELHOS. the.[M]-PL my-M-PL old-M-PL ‘my parents (lit. oldies).’ M2FAT (12) Plural item is head in fourth position in NP os mos primeiros IRMÃOS the[M]-PL my-M-PL first-M-PL sibling-M-PL ‘my first brothers and sisters.’ H1FRA (13) Plural item is head in fifth position in NP seus nove ou dez ANOS [ININT] 3POS-M-PL nine or ten year[M]-PL ININT ‘her (or his) nine or ten years.’ H1JOA (14) Plural item right of head, second position in NP ELES próprios 3P-M-PL self-M-PL ‘they themselves.’ H1FRA (15) Plural item at right of head, third position in NP COISAS assim desagradáveis thing.[F]-PL thus unpleasant-PL ‘unpleasant things like that.’ M1LUC (16) Plural item at right of head, fourth position in NP essas COISA assim naturais this-F-PL thing.[F]-S thus natural-PL ‘natural things like that.’ H1JOA (17) Plural item at right of head, fifth position in NP cento e vinte CONTOS³⁶ mensais hundred and twenty conto.[M]-PL monthly-PL ‘one hundred and twenty contos monthly.’ H1JOA Table 7 presents the findings of two analyses of the above configurations in Tonga Portuguese: VARBRUL run A and VARBRUL run B. In VARBRUL run A, the 11 original configurations (of examples (7) through (17) above) were assigned to seven factor groups. This reduction was motivated by small data numbers representing the 4th and 5th positions in the NP, both for the head and the post-head items. Subsequently, in VARBRUL run B, the seven configuration factors were reduced to five. The latter reduction was undertaken because low data

pr. PL 0.97 . . 0.24 0.03

N PL % PL pr. PL 310/323 96% 0.90 26/28 93% 0.75 5/5 100% . 180/734 25% 0.28 3/28 11% 0.09

Significance=.005 Log likelihood = −413.246

N PL % PL pr. PL 310/323 96% 0.89 26/28 93% 0.72 5/5 100% . 160/680 24% 0.30 20/54 37% 0.22 1/8 13%. 2/20 10% 0.07

Significance=.008 Log likelihood = −414.963

Generation 2

% PL 98% 97% 94% 61% 64% 57% 27%

pr. PL 0.83 0.70 0.71 0.27 0.25 0.22 0.09

N.PL 529/538 65/67 33/35 463/750 30/79

% PL 98% 97% 94% 62% 38%

pr. PL 0.83 0.70 0.71 0.26 0.15

Significance = .001 Log likelihood = −576.261

N.PL 529/538 65/67 33/35 402/654 61/96 16/28 14/51

Significance = .099* Log likelihood = −575.050

Generation 3

% PL 95% 94% 93% 31% 50% 46% 20%

pr. PL 0.87 0.83 0.87 0.28 0.26 0.27 0.10

N PL 1006/1054 91/97 39/42 665/2049 34/124

% PL 95% 94% 93% 32% 27%

pr. PL 0.87 0.84 0.87 0.27 0.17


N PL 1006/1054 91/97 39/42 584/1886 81/163 17/37 17/87



*On this VARBRUL run for Generation 3, the Sig. level exceeds .05 because the factor group <presence of prior PL markers in the NP> was rejected as insignificant.

Configuration factors N PL % PL 1. Left, adjacent to head 167/193 87% 0 0 2. Left, non-adjacent to head 3. Head in 1st position in NP 1/2 50% 4. Head in 2nd, 3rd, 4th or 5th position in NP 22/562 4% 5. Right of head, in 2nd, 3rd, 4th or 1/16 6% 5th position in NP


VARBRUL run B

pr. PL 0.95 . . 0.29 . . 0.01

N PL 167/193 0 1/2 22/552 0/10 0/1 1/15

% PL 87% 0 50% 4% 0 0 6%


Configuration factors 1. Left, adjacent to head 2. Left, non-adjacent to head 3. Head in 1st position in NP 4. Head in 2nd position in NP 5. Head in 3rd, 4th or 5th position in NP 6. Right of head, in 2nd position in NP 7. Right of head, in 3rd, 4th or 5th position in NP

VARBRUL run A

Generation 1

Table 7. The effect of the on the use of the PLURAL morpheme in plural NP items.

6

Alan N. Baxter

numbers in Generations 1 and 2 prevented the detailed optional analysis of items to the right of the head. As can be seen in Table 7, comparing VARBRUL runs A and B for each generation, the marginally lower (i.e., better) Significance and Log Likelihood results in run B, in contrast with run A, for Generations 1, 2 and 3, confirm that the model with five factors provides a better fit of the data for those generations. On the other hand, for the three generations combined, the better Significance and Log Likelihood results in VARBRUL run A, in contrast with run B, reveal the seven factor model as providing a better fit of the data of the community as a whole. Indeed, as will be seen in the ensuing discussion, VARBRUL run A permits the detection of the differential effect on PL-marking of certain structural facts relating to head position (in Generations 2 and 3) and post-head items, (in Generation 3) that are not visible in run B. What then do these two analyses demonstrate regarding the effect of the factor group <structural configuration> on PL marking? Overall, they demonstrate that there are different grammars for each of the three Tonga generations, and that certain tendencies introduced in Generation 1 are preserved in later generations. In Generation 1, PL marking is initially established on the left position adjacent to the head (run A: pr. 0.95; run B: pr. 0.97). The head itself, when occurring in second position (run A: pr. 0.29) or in second or subsequent positions, (run B: pr. 0.24) disfavors plural attachment. Post-head positioning (which for Generation 1 mainly involves the 3rd position in the NP) almost categorically disfavors agreement (run A: pr. 0.01; run B: pr. 0.03). Note that the NP structure is minimal in Generation 1, mainly consisting of one pre-head position,³⁷ preferably PL-marked, a head which is seldom PL-marked, and one post-head position practically unmarked for PL. The basic PL-marking rule of Generation 1 requires PL-marking to be placed in the prehead position of the NP. Subsequently, in Generation 2, the NP is structurally more developed, allowing multiple pre-head items (witness run A, factors 2 and 5). The left position adjacent to the head (run A: pr. 0.89; run B: pr. 0.90) continues as the predominant factor favoring PL marking. However, PL has now spread to the other left positions, where it is also favored, although with a lower weighting (run A: pr. 0.72; run B: pr. 0.75). The head registers unfavorable results for second, third and fourth positions, (run A: pr. 0.30, for 2nd position, pr. 0.22 for 3rd, 4th or 5th position; run B: pr. 0.28, for 2nd, 3rd, 4th or 5th position). However, the results for run A suggest an incipient distinction between 2nd position heads and heads occurring later in the NP. Post-head positions, nevertheless, are still most unlikely to be PL-marked. In sum, the PL-marking rule of Generation 2 requires PL-marking to be placed (i) on the pre-head position, and (ii) on earlier positions more than later positions. Finally, in Generation 3, the NP structure is even more complex. It is still the left position adjacent to the head (pr. 0.83, runs A and B) that is the most propitious for plural attachment. The next most favorable factor is still the non-adjacent left positions (pr. 0.70, runs A and B). Following these, the presence of the head in NP initial position (pr. 0.71, runs A and B) is the next strongest factor, attesting the effect of


the principle of preferential marking of earlier positioned items. The positioning of the head in second position (run A, pr. 0.27) or in second and subsequent positions (run A, pr. 0.26) is similarly unfavorable for PL marking, and the incipient differentiation of these two positions, noted in Generation 2 (run A) is neutralized. Indeed, the grammar of Generation 3 would be better described by merging these two positional factors, as was done in VARBRUL run B (new factor 4). However, in run A, a differentiation is evident between the degree of unfavorability of PL-marking of the adjacent post-head position (run A, pr. 0.22) and subsequent post-head positions (run A, pr. 0.09), the latter being extremely unlikely to bear PL-marking. As a consequence of this further structural growth of the NP, the rules for PL-marking in Generation 3 are also more complex than for the previous generations. Here, PL-marking is to be assigned (i) to the left in the NP, (ii) more on the pre-head structure than on the head, and (iii) more on earlier positions. What is especially clear in the above discussion is the prominent role of the left position adjacent to the head. For Generation 1, the left adjacent position seems to play a quantifying role, bracketing the head. Why should this position be the point of entry of plural morphology? And why should it remain a prominent feature? Three answers to the first question are plausible here, in connection with the type of PLD available to Generation 1: (i) the evidence for plural marking was so diffuse in PLD that what might be considered to be an unmarked solution was adopted: quantify from the immediate left of the head (which happens to be first position in the minimal NP for Generation 1); (ii) the PLD contained data from Portuguese L2 with evidence for this same structure coming from simple Portuguese L1 structures (e.g., DEMONSTRATIVE+NOUN); or (iii) the PLD contained data from Portuguese L2 with strong L1 transfer effects. At Monte Café plantation, there are no surviving L2 Tonga Portuguese speakers, so the above three factors cannot be assessed directly. Explanation (ii) might be an option as an initial trigger, but the ratio of Africans to Europeans was very high. Interviews conducted with three European Portuguese, former administrators (including a former feitor ‘foreman’)³⁸ suggest that there was some simplification on the part of the Europeans, in the sense of using more analytic forms (e.g., subject pronouns as object pronouns). Nevertheless, the opinion of these former administrators was that the Tongas were the main originators of divergent forms. Of course, the information supplied by these men does not eliminate the possibility that input at an earlier stage may have contained variation originating in varieties of European Portuguese. Nevertheless, there are good reasons to believe that the main contribution to the development of variable plural agreement has other sources. The natural developmental argument of explanation (i), involving unmarked settings triggered by a lack of clear evidence, seems plausible. In this scenario, the learner constructs minimal NPs with a quantifying item to the left. However, the

7

8

Alan N. Baxter

demographics of Monte Café until the 1940s, and some of the phonological considerations raised earlier in this paper, suggest that transfer (explanation (iii)) must be taken into account. Generation 1 is the group that received strongest PLD from the Portuguese L2 continuum, this group having a high proportion of L2 Portuguese speaking parents. L1 influence on L2 continua or interlanguage is well documented in acquisition research, especially for rudimentary stages of L2 development. Certainly the demographic facts of the Monte Café plantation appear to point very clearly to a situation that would tend to maintain a strong presence of early stage L2 Portuguese. Guy (1981: 300–1) proposed the hypothesis that, in Brazilian Vernacular Portuguese, variable PL-marking, assigned from the left of the NP, could well have its origins in transfer from Bantu-speaking slaves, via noun classifiers and their role in PL assignment in Bantu languages. This hypothesis is worth contemplating with respect to Tonga Portuguese. Umbundu, as other Bantu languages, shows plural on nouns through a prefixal nominal classifier which signals a semantic class membership, and which may also carry deictic functions. Umbundu has a prefixal concordance system that is determined by the classifier of the noun, and extends to adjectives, determiners, and some connectives. Furthermore, the plural indication in the noun classifier is the key morpheme for the assignment of plural classifiers to other NP items. An adult Umbundu speaker in rudimentary stages of Portuguese acquisition could be expected to be more sensitive to plural indication in an adjacent prenominal position. A noun final position would be a rather unlikely place for such a speaker to perceive a plural marker or to attempt to place one. Also, with certain disyllabic classifiers, the second syllable of the classifier may show the plural contrast, e.g., o-mbwa ‘(the/a) dog’ vs. olo-mbwa ‘(the-PL) dog’; ovi-ndele ‘(the/a) white (person)’ vs. otu-ndele ‘(the-PL) white (person)’. This could further draw the learner’s attention to the immediate prenominal position. There is some corroboratory evidence in Generation 1 Tonga Portuguese that certain Portuguese prenominal determiners might have been heard as classifiers. In the following example, the variable presence of u- on the word uhome (< P. o homem ‘the man’) suggests that the Portuguese definite article may have been reinterpreted phonologically as the classifier of Umbundu class 2 nouns, pertaining to persons, and prefixed with u- prior to word initial -o. (18) Agora isso uhome dexô massi³⁹ lá. now this[NEU] man[M]-S leave-PRT-3S machete[M]-S there ‘Now, the man left a/the machete there.’ F3S-1694 In her work on structural mixing, Myers-Scotton (1997, 2000a, 2000b, 2001) has demonstrated (with extensive data from bilingual aphasia, L2 acquisition, and bilingual code switching) how content morphemes from one language can be slotted into the morphosyntactic frames of another language. Furthermore, when mixing occurs, the syntactic frame acting as a matrix may contain early system morphemes — morphemes which are intrinsic to the meaning of the content morpheme slot and


which are generated together with that slot.⁴⁰ This sort of transfer is very common in the initial stages of L2 acquisition. Myers-Scotton’s perspective seems relevant here. In other words, in Generation 1, it may well be that the predominant pattern observed relates to the African L1 structure that was strongly evident in L2 PLD available to that generation. From this perspective, transfer in rudimentary adult L2 acquisition would, in part, involve a degree of morphosyntactic frame fabrication involving L1 frames and L2 content morphemes, especially at an early stage in an L2 acquisition situation where L1 is strongly present. In this case, an Umbundu speaking adult might easily associate Portuguese determiners with Umbundu classifiers. Following this approach, the prominence of the left adjacent position in Generation 1, even if it had a non-transfer source, say as an unmarked developmental solution as mentioned earlier, could be reinforced by the insertion of Portuguese content morphemes within Umbundu nominal frames in rudimentary adult L2, transmitted to Generation 1 speakers via PLD. Such nominal frames would require their plural marking to their immediate left, where a classifier would appear in Bantu languages. Subsequently, in Generations 2 and 3, this reinforced trait has been assimilated into the developing Tonga Portuguese NP structure. In the grammar of Generation 3 speakers, who are Portuguese dominant Portuguese-Umbundu bilinguals, the left adjacent position now assumes a characteristic pattern of variable plural marking in an extensively developed NP.

9. Conclusion The profile of variable plural marking in Tonga Portuguese is the result of crossgenerational development. My analysis demonstrates that this change was conditioned by factors rooted in the L2 and L1 acquisition of Portuguese in a contact situation involving a strong Portuguese L2 component already influenced by Umbundu, and simultaneous acquisition of Umbundu. The latter two languages were constantly “renewed” through the indentured labor system. In these circumstances, the role of continuity of possible variation in plural marking in varieties of European Portuguese (Naro & Scherre 2000) would appear to be of minor or null consequence. Language acquisition in the contact context was the trigger for the development of variable agreement rules. The sources of certain details of this re-configuration, such as the predominant adjacent prenominal position of plural marking, could plausibly result from the choice of an unmarked structure in the face of diffuse models, and the influence of minimal Portuguese L1 NP structures. However, the influence of African L1 transfer in this particular socio-historical context appears to have played a significant role directly and indirectly through Portuguese L2 PLD for L1 acquisition of Portuguese. The findings of this paper lend strong support to Guy’s (1981: 300–1) hypothesis that Bantu substrate is a potential source of variable PL-marking, assigned from the left of the NP, in Brazilian Vernacular Portuguese.

9

20

Alan N. Baxter

Notes * This paper was prepared as part of a research project on Tonga Portuguese supported by Australian Research Council Grant A59803382. I am grateful to two anonymous reviewers and to John Singler for their helpful criticisms and comments. Special thanks go to Gregory Guy for extensive discussion of the analysis of saliency and noun phrase configuration. 1. The Tongas, according to the traditional Colonial Portuguese use of the term, were the descendants of continental African indentured workers. 2. TP also displays variation in subject-verb agreement and gender agreement of NP items (Baxter 2002). 3. The morpheme glosses correspond to the grammatical forms of Standard Portuguese. The following abbreviations are used in this article: [] {Inf.} {Res.} 3 Agr. AHSTP

lexical gender informant researcher third person agreement Arquivo Histórico de São Tomé e Príncipe ‘the Historical Archives of São Tomé and Príncipe’ EX exclamation F feminine gender IMP preterite imperfect indicative INF infinitive ININT unintelligible word in transcription L1 first language

L1A L2 L2A M NEU PL POS pr. PRS PRT Q S SU .

first language acquisition second language second language acquisition masculine gender morpheme boundary neutral gender plural possessive probability weighting present indicative tense simple perfective indicative preterite question singular subjunctive links EX intonation to word

4. Note that TP often uses the neutral gender form tudo ‘all’ where Standard Portuguese requires the gender- and plural-inflecting forms todas ‘all (feminine gender plural)’ or todos ‘all (masculine gender plural)’. 5. The discussion in Sections 2, 2.1 and 2.3 supersedes the account in Baxter (2002). 6. São Tomé has two indigenous Portuguese-based creole languages: São Tomé Creole Portuguese (Sãotomense) and Angolar, both of which developed in the 16th century. Angolar is a maroon creole which split off from São Tomé Creole Portuguese in the 16th century (Lorenzino 1998). 7. The minuscule territory of Ajudá (Ouidah), or São João Baptista de Ajudá, was under Portuguese control until 1961. 8. Until the 20th century, the contractual conditions often amounted to a continuation of slavery. An official repatriation system was instituted in 1908. Prior to this, among the Angolans on São Tomé were people who had been sold into slavery or simply deported for petty crimes (Swan 1909). These facts were confirmed in the testimonies of elderly Tongas. For a general overview of the São Tomé labor scandal of the late 19th and early 20th centuries, see Hodges & Newitt (1988).

Variable NP plural agreement in a restructured African variety of Portuguese 9. According to the census cited by Tenreiro (1961: 191), by 1950 there were 2,696 Tongas in the province of São Tomé and Príncipe. This number seems small, and perhaps only refers to Tonga adults in the workforce. 10. Lorenzino (1998: 48) refers to Tongas who are sixty and above who are bilingual in Umbundu and Sãotomense. It would seem likely that these are Tongas from plantations close to Creole communities. 11. My oldest informants testify to the fact that some European overseers acquired a minimal knowledge of Umbundu. This type of situation probably also existed in earlier periods. 12. These comments are based on the testimonies of elderly Tongas at the Monte Café, Agostinho Neto and Praia das Conchas plantations, including three Tongas who had been overseers. 13. These circumstances were reported by the oldest Tongas interviewed at the Monte Café, Agostinho Neto and Praia das Conchas plantations, including Tongas who were former workforce overseers. 14. Two linguistic features of TP suggest an early connection with Sãotomense: ami ‘I’, and ni ‘to, at, in, from’. However, the former is also used in the Umbundu koiné and is a likely Bantu transfer in TP. See note 32 for further discussion of ni. 15. Plantations that relied on road transport to the public docks, in contrast with those that had their own narrow gauge railways and jetties, would have required frequent off-plantation presence of numbers of workers (Cadbury 1910: 28–9). Monte Café was not in the former category, as it had its own railway. 16. The language of education is the European Standard, represented in colonial times mainly by European teachers. Broadly speaking, educated São Tomé Portuguese approximates the European Standard, although, as expected, it has certain local characteristics in phonology, morphosyntax and lexicon. However, São Tomé Portuguese displays a wide spectrum of varieties that reflect socioeconomic status and acquisition type — whether Portuguese is a first or second language. 17. This document reports an Angolan predominance among the some 500 African workers (and 40 Portuguese administrators). 18. Correspondence that I located between Monte Café administrators and the Curadoria ‘Office for the Affairs of Contracted Workers’ in 1876 also attests the presence of Kru, and Mina workers (Curadoria 1876, box 82, file 2, folder 4). Thus, although it may be said that the Monte Café Tongas descend mainly from Bantu-language speakers, principally from Angola, it must be kept in mind that Kru and Kwa elements were also present in the 19th century. This point establishes an interesting comparative link between the Tongas and certain AfroBrazilian communities. 19. In the sample considered here, there is only one couple involving a Cape Verdean (father). The offspring of this couple do not speak Cape Verde Creole. 20. According to Cadbury (1910: 40), many plantations had creches for younger children, under the care of enfermeiras pretas ‘black nurses’. The creche at Monte Café, may have opened at a later date, since our middle-aged and oldest informants made no reference to it. Rather, they mentioned that they accompanied their parents to the field, or conducted yard duties under the supervision of a contract worker. The details provided by our elderly informants

2

22

Alan N. Baxter on the induction of children into the labor force through light work in the plantation administration compound are reminiscent of the observations made by Cadbury (1910: 40) regarding the situation elsewhere on São Tomé. With extremely rare exceptions, children did not have access to formal education in the earlier periods of the indenturing system (witness Cadbury 1910: 37, and testimonies of elderly informants). At the age of 14 years, children were contracted under the same conditions as their parents (Cadbury 1910: 40). 21. The oldest speakers in this group were aged 90+. A future analysis will divide the (60+) group, creating a fourth age group. 22. Men above 65 years of age and women above 60 are retired, although they may continue to do some work for the plantation. 23. In the case of the youngest generational group, all speakers had attended a minimum of five years of primary education. However, the two informants from the second generational group had attended only the year one class, sporadically. 24. One reviewer of this paper made two valid points regarding the range of the independent linguistic factors analyzed. Firstly, in view of the complex semantic structure of the Bantu classifier system, it would be of interest to evaluate whether this system is reflected in any way in Tonga Portuguese, perhaps in terms of animacy privileges with respect to plural marking. This remains a goal for future research. Secondly, as Portuguese NP items which undergo plural marking are also marked for gender, it would be interesting to evaluate the degree and nature of possible correlations between these two agreement systems. This is also a goal for future research. So far, the only work relating to this issue involves dialects of Brazilian Portuguese such as that of Helvécia (Bahia), where variable gender agreement operates somewhat differently from plural marking. Gender marking is assigned from the head outwards in the NP, whereas plural marking is assigned from outside the head, from the beginning of the NP (Lucchesi 2000: 256–8). However, pre-nuclear positions did tend to favor the presence of gender marking, much as Scherre (1988) had found in the case of the plural marker (Lucchesi 2000: 261–2). Finally, there was considerable interdependence between the rules, such that, if number agreement was categorical, gender agreement was also categorical (Lucchesi 2000: 263–4). 25. This framework did not contemplate the potential differential effect on plural marking that might be caused by words with a singular form ending in a non-nasal vowel (for example casa ‘house’) versus those ending in a nasal vowel (for example ordem ‘order’). This is a goal of future research. 26. The rationale is that an analysis which considers the form class of a pluralizing item alone will fail to account for the fact that the position of the form classes is a major determinant of plural marking. Furthermore, consideration of the position alone of a pluralizing item will not account for the fact that the form class and function of the item interplay with its position. For further discussion see Scherre (1998). 27. Note that in a VARBRUL analysis the probability weightings are to be read in interpreting the results, and not the percentages. In a study involving unevenly distributed amounts of data from each informant, the percentage is not a reliable indicator of the effects of factor goups. For a discussion of the inadequacy of percentages and the rationale for using weighting probabilities, see Sankoff (1988). 28. Only data items with stress patterns matching those of Standard Portuguese were codified. Nevertheless, few items were excluded.

Variable NP plural agreement in a restructured African variety of Portuguese 29. These results are based on VARBRUL run B, for each generation, explained in Section 8. 30. As Greg Guy has suggested (p.c.), a clarification is essential here. It is important to interpret the results in Table 5 with respect to their distribution across NP positions (see Table 7). Consider the high pr. weight for the pluralization of words with stressed final syllables, with a low % value, and the contrasting lower pr. weight of monosyllables, with a high % value. This apparent anomaly arises because most monosyllables are articles, occurring in NP initial position, where they are marked for plural at a rate of 87% to 98% across the three generations (see Table 7). It is in relation to other words that occur in first position in the NP that the monosyllables have a slightly unfavorable pr. weight. In contrast, the words with stressed final syllables are NP heads (in the traditional sense), mainly occurring in non-initial positions in the NP. In Generation 1, in Table 7, a head in non-initial position is marked at a rate of 4%, overall. So, in relation to other heads occurring in non-initial position, the 14% marking of words with stressed final syllables, for Generation 1 in Table 5, is highly favorable, and thus receives a heavy factor weighting. 31. TP isso < P. isso ‘this’ — a neutral pronoun. 32. The conversation refers to events of the 1950s, when the colonial government attempted to force the São Tomé Creoles to work on the plantations. The informant recalls that the Creoles had attempted to disrupt plantation work. 33. In the speech of my oldest informants, the preposition ni corresponds variably to ‘in’, ‘to’ and ‘from’. Motivation for the conflation of the three distinct function forms from Portuguese into one form in Tonga Portuguese is found in Umbundu. Here three prepositions express location: (i) locative and directional ku; (ii) mu, which expresses location (within), and direction (both into and out of an enclosed space), and (iii) mo which expresses location (on). The fact that two of the Umbundo prepositions have directional functions while sharing locative functions with Portuguese em may have facilitated a transfer of their overall functions onto Portuguese em. In addition, in Umbundu-influenced Portuguese, initial [d] would be [nd], and this could facilitate the formation of ni from Portuguese de [di]. Alternatively, the values of ni in TP could be accounted for by appealing to a connection with São Tomé Creole. 34. The variable use of the first person singular to represent the third person singular is a low frequency variable. 35. The plural item is underlined and the head is represented in bold type. 36. The conto is a Portuguese measure meaning “one thousand monetary units”. 37. In Generation 1 there were only three cases with an item in a non-adjacent position at the left of the head, and all were unmarked on the non-adjacent pre-nuclear items. The items adjacent to the head were determiners, demonstratives, numerals and quantifiers. 38. Interviews with João Ronco (in São Tomé), António Portugal (in Lisbon), António Carvalho Maia (former owner of the Monte Café plantation, in Lisbon), Delfim Pinho (former feitor de mato ‘field foreman’ and feitor geral ‘general duties foreman’, in Viseu), and Elias Correia (in Viseu). These men were associated with Monte Café as of the late 1940s, and there are no surviving administrators from earlier periods. In the views of the latter two interviewees, when they arrived in São Tomé the characteristics of Tonga Portuguese, some of which they were able to imitate during the interview, were already firmly in place.

23

24

Alan N. Baxter 39. Massi ‘machete’ < P. machim ‘machete’. 40. Myers-Scotton’s 4-M model proposes “that differences in when and how different morpheme types are salient in production is reflected in various surface outcomes” (Myers-Scotton 2001: 3). The model categorizes morphemes into four classes: content morphemes and three types of system morpheme. Different classes of system morphemes are activated at different levels of language production. Early system morphemes are activated at the mental lexicon level (the lemma level) as a type of satellite to content morphemes, adding particular (core) meanings, such as, for example, plural -s in English, or definiteness of reference. Late system morphemes are assigned by structural requirements (Myers-Scotton 2000a). The concept of late system versus early system morphemes may help account for some of the phenomena discussed above, where the plural agreement systems of Umbundu and Tonga Portuguese are concerned. A key issue in this case would appear to lie in the need to distinguish some plural morphemes as being more basic, hence early system, as opposed to others that would be assigned structurally.

References Baxter, A. N. 1998. “Morfossintaxe.” In Perl & Schwegler, 97–134. Baxter, A. N. 2002. “‘ Semicreolization’? — The restructured Portuguese of the Tongas of São Tomé — a consequence of L1 acquisition in a special contact situation.” Journal of Portuguese Linguistics 1: 7–39. Bayley, R. 1994. “Interlanguage variation and the quantitative paradigm: Past tense marking in Chinese-English.” In Research Methodology in Second-Language Acquisition, E. Tarone, S. Gass & A. D. Cohen (eds.), 157–81. Hillsdale: Lawrence Erlbaum Associates. Bayley, R. 1996. “Competing constraints on variation in the speech of adult Chinese learners of English.” In Second Language Acquisition and Linguistic Variation, R. Bayley & D. R. Preston (eds.), 97–120. Amsterdam: John Benjamins. Bickerton, D. 1975. Dynamics of a Creole System. Cambridge: Cambridge University Press. Cadbury, W. A. 1910. Os serviçaes de S. Thomé. Lisbon: Livraria Bertrand. DeGraff, M. 1999. “Creolization, language change, and language acquisition: An epilogue.” In DeGraff, 473–544. DeGraff, M. (ed.) 1999. Language Creation and Language Change — Creolization, Diachrony, and Development. Cambridge: The MIT Press. Emmerich, C. 1984. A língua de contato no Alto Xingu — Origem, forma e função. PhD Thesis. Universidade Federal do Rio de Janeiro. Gass, S. 1988. “Second language acquisition and linguistic theory: The role of language transfer.” In Linguistic Theory in Second Language Acquisition, S. Flynn & W. O’Neill (eds.), 384–403. Dordrecht: Kluwer Academic Publishers. Guy, G. R. 1981. Linguistic variation in Brazilian Portuguese: aspects of the phonology, syntax and language history. Ann Arbor: University Microfilms International. Hodges, T. & M. Newitt. 1988. São Tomé and Príncipe: From Plantation Colony to Microstate. Boulder: Westview Press. Ingram, D. 1989. First Language Acquisition. Cambridge: Cambridge University Press. Lightfoot, D. 1999. “Creoles and cues.” In DeGraff (ed.), 431–52. Lopes, N. 2001. Tópicos de concordância. PhD Thesis. Salvador: Universidade Federal da Bahia.

Variable NP plural agreement in a restructured African variety of Portuguese Lorenzino, G. 1998. The Angolar Creole Portuguese of São Tomé: Its Grammar and Sociolinguistic History. Munich: Lincom Europa. Lucchesi, D. 2000. A variação na concordância de gênero em uma comunidade de fala afrobrasileira: novos elementos sobre a formação do português popular do Brasil. PhD Thesis. Universidade Federal do Rio de Janeiro. Maia, António da Silva. 1964. Lições de gramática de quimbundo. Cucujães: Escola Tipográfica das Missões. Major, R. C. 1996. “Markedness in second language acquisition of consonant clusters.” In Second Language Acquisition and Linguistic Variation, R. Bayley & D. R. Preston (eds.), 75– 96. Amsterdam: John Benjamins,. Mantero, F. 1910. Manual Labour in S. Thomé and Principe. Lisbon: Francis Mantero. Mello, J. P. de. 1864. Província de São Tomé e Príncipe. Produtos naturaes, e da indústria agrícola e fabril designados à Exposição Universal em Londres. São Tomé: Imprensa Nacional. Mendes, B. C. 1985. Contributo para o estudo da língua portuguesa em Angola. Lisboa: Instituto de Linguística da Faculdade de Letras de Lisboa. Myers-Scotton, C. 1997. Duelling Languages: Grammatical Structure in Codeswitching. New York: Oxford University Press. Myers-Scotton, C. & J. L. Jake. 2000a. “Four types of morpheme: Evidence from aphasia, code switching, and second-language acquisition.” Linguistics 38: 1053–1100. Myers-Scotton, C. & J. L. Jake. 2000b.“Testing the 4-M model: An introduction.” International Journal of Bilingualism 4: 1–8. Myers-Scotton, C. 2001. “Implications of abstract grammatical structure: Two targets in Creole formation.” Journal of Pidgin and Creole Languages 16 : 217–73. Naro, A. J. & M. M. P. Scherre. 2000. “Variable concord in Portuguese: The situation in Brazil and Portugal.” In Language Change and Language Contact in Pidgins and Creoles, J. McWhorter (ed.), 235–56. Amsterdam: John Benjamins. Naro, A. J., & M. Lemle. 1977. “Syntactic diffusion.” Ciência e Cultura 29: 259–68. Perl, M. & A. Schwegler (eds.). 1998. América negra: panorámica actual de los estudios lingüísticos sobre variedades hispanas, portuguesas y criollas. Frankfurt am Main: Vervuert/Madrid: Iberoamericana. Pintzuk, S. 1988. VARBRUL Programs. Philadelphia: University of Pennsylvania Department of Linguistics. Rougé, J-L. 1992. “Les langues des Tonga.” In Actas do Colóquio Internacional sobre Línguas Crioulas de Base Portuguesa, E. d’Andrade & A. Kihm (eds.), 171–5. Universidade de Lisboa. Lisboa: Colibri. Sankoff, D. 1988. “Variable rules.” In Sociolinguistics — An International Handbook of the Science of Language and Society, U. Ammon, N. Dittmar & K. Mattheier (eds.), 984–97. Berlin/New York: Walter de Gruyter. Scherre, M. M. P. 1988. Reanálise da concordância nominal em português. PhD Thesis. Rio de Janeiro: Universidade Federal do Rio de Janeiro. Scherre, M. M. P. 1998. “Sobre a influência de três variáveis relacionadas na concordância nominal em português.” In Padrões sociolingüísticos, G. M. Oliveira e Silva & M. M. P. Scherre (eds.), 85–118. Rio de Janeiro: Tempo Brasileiro. Swan, C. A. 1910. The Slavery of Today; or, The Present Position of the Open Sore of Africa. Glasgow: Pickering & Inglis. Tenreiro, F. 1961. A ilha de São Tomé. Lisbon: Junta de Investigações do Ultramar.

25

26

Alan N. Baxter Valente, J. F. 1964. Gramática umbundu — a língua do centro de Angola. Lisboa: Junta de Investigações do Ultramar. Wolfram, W. 1985. “Variability in tense marking: A case for the obvious.” Language Learning 35: 229–53. Zimmermann, K. (ed.). 1999. Lenguas criollas de base lexical española y portuguesa. Frankfurt am Main: Vervuert/Madrid: Iberoamericana.

CHAPTER 6

Second language acquisition in creole genesis The role of processability Fred Field

California State University, Northridge

Many current models of creole genesis assign a central role to adult language learners, thus suggesting that a creole language is the consequence of second/subsequent language acquisition under rather austere circumstances. It follows that an aggregate of individual interlanguage (IL) varieties may (a) develop to meet a specific set of communicative functions between groups, (b) expand (functionally and linguistically) for basic interpersonal communication within the community, and (c) then become fossilized and conventionalized as a lingua franca of sorts. This variety then evolves into a creole as it becomes the principal language of the community. This paper argues that the nature of the input for acquisition, shaped and limited by a rather unique set of social circumstances, can explain the often-noted lack of inflectional morphology from the lexifier (the principal donor of lexical material and onetime target language or TL of acquisition). In addition, I suggest that in such an emerging creole, developmental stages in acquisition are evident. That is, the availability of donor structures in a community-wide learners’ variety reflects the step-by-step sequences in which word-order patterns involved in, for instance, question formation, negation, and so on, are normally acquired in native or subsequent language acquisition. These sequences may stop at some point in the developmental process either because of a profound change in the circumstances of acquisition (e.g., the sudden or gradual absence of the TL and its native speakers), or as a result of the insurmountable social distance between cultures and communities of speakers, and its likely effect on learning. This is tantamount to saying that some more advanced stages (according to TL native-speaker norms) may not be reached by all learners. All other things being equal, the greater the quantity and quality of the input, the closer to the lexifier the resultant creole will be. I examine a sampling of English-based creoles that, during their formation, had various degrees of contact with the lexical source language (and its native speakers), and, therefore, degrees of English-like structural traits. With respect to English, stages in acquisition are assessed according to Pienemann’s (2000) processability hierarchy. For a broader perspective, I will also refer to data from Palenquero, a Spanish-lexicon Creole spoken in northern Colombia.

28

Fred Field

1. Introduction Languages are said to be in contact when two or more languages are used by the same group of speakers (Thomason 2001: 1, Weinreich 1953: 1). On the basis of contact alone, one expects to find varying degrees of bi- or multi-lingualism in any emerging creole community. This multilingualism typically involves an original native language (or languages) and an encroaching variety, the socially prominent language of colonial masters, overseers, etc. Hence, one might suspect that any variety originating in such a community will show the influence of any or perhaps all languages previously spoken by members of that community, especially if they all shared a common native language, or if their languages were similar and/or genetically related. This paper will show that the outcome of language contact in creole communities depends to a large extent on social circumstances, particularly the ways in which the community’s multiple languages are acquired and used. Thus, native and/ or non-native acquisition scenarios and the communicative functions of the emerging variety reflecting the kinds of social relationships among various groups and subgroups of speakers — e.g., between laborers and their overseers and among laborers as a discrete social grouping — will have a profound effect on the final product of acquisition. One possible view that follows from this kind of social backdrop is that adult (i.e., generally speaking, post-pubescent) language learning must have played a principal role in the emergence of creole languages. Many leading scholars now view creoles as largely reanalyzed and restructured versions of colonial languages or masters’ tongues (Arends, Muysken and Smith 1995, Siegel 1999, Thomason 2001, Thomason and Kaufman 1988). If it is indeed the case that language acquisition has played a significant role, one should expect to find a dynamic relationship between two separate sets of linguistic processes, particularly in the earlier stages of development: (a) second/subsequent language acquisition (SLA), which, according to socalled substratist positions of creole genesis figures prominently in the emergence of some if not all creoles,¹ and (b) transfer, i.e., the persistent influence of underlying native languages (NLs) during acquisition (this transfer is also called “interference” in earlier literature). Transfer figures prominently in substratist approaches as well, especially when attempts are made to determine the language(s) that may have formed the substrate of a particular creole variety. Even though transfer is customarily linked to SLA, it is important to note that they are indeed distinct phenomena. SLA involves learning or acquiring (to the extent possible) a particular target language (TL) non-natively, albeit in a predictable, step-by-step order. Pertinent social factors to consider are (a) access to the TL and its native speakers, (b) learners’ attitudes toward the target language and its native speakers (affecting motivation), and (c) the availability (and quality) of formal instruction. Transfer involves the application by the learner of native-language (NL) phonological patterns, along with other grammatical and discourse strategies, to

Second language acquisition in creole genesis

the TL. While leaving a characteristic “accent” in the performance of the learner/ speaker, previous linguistic knowledge (knowledge of the NL), nevertheless, enables the learner to use the foreign target as it facilitates the learning process. As Corder termed it, this is the positive “role of the mother tongue” in SLA (Corder 1993: 20). Moreover, the two processes — learning and transfer — may interact in relatively predictable ways. The main line of reasoning explored here is this: (a) work in SLA shows that ILs appear to be structurally similar to pidgins in many ways; and (b) ILs (in aggregate form) can reach functional efficiency for interpersonal communication among entire groups of SL learners under certain social circumstances, and, as a result, form a model for acquisition for subsequent generations of learners. Individual ILs within a community will share a blend of characteristics of the lexifier/one-time target (superstrate) and underlying native (substrate) languages. To the degree that underlying substrate languages (that may be genetically or typologically related) share grammatical characteristics, the pool of features upon which speakers can draw may be quite similar (cf. Mufwene 2001). In fact, there may be many common features and a greater degree of homogeneity from individual to individual, more so than in an international classroom, for example, and, perhaps, similar degrees of “success” in learning the superstrate. Consequently, a community-wide, mutually intelligible aggregate of learners’ varieties can, in essence, become identified with that community so as to become its principal “(target) language”. This paper is organized as follows: Section 2 briefly examines the central and still controversial question whether a pidgin lies at the source of every creole. I will not claim that pidgin origins are unlikely or impossible; I will merely propose that there is at least one other plausible source from which creole languages can spring, namely a learners’ (or IL) variety. As will become apparent, the term IL variety may have its weaknesses when applied to the speech of an entire community or particular population segments within it, but the term does manage to capture an important characteristic of this type of contact situation. For instance, in his work on Latin American Spanish, Lipski (1994) uses interlanguage quite freely and effectively in referring to particular regional and/or national varieties of Latin-American Spanish. As a technical term, interlanguage has been used more narrowly in SLA studies to signify the stages which an individual has attained in his/her attempt to attain proficiency in the target language (that stage may be situated anywhere between zero proficiency (= L20) and native-like ability (L2n); see (Selinker 1972)). Because SL learning is my primary focus, this paper contrasts various kinds of learning environments. Therefore, discussion in Section 3 moves from research on SLA in general to that of “SLA in instructed settings” (which arguably affords SL learners a better opportunity to achieve maximal levels of proficiency in the TL), and “SLA in untutored settings” (the most likely scenario in an emerging creole community). Section 4 draws parallels between the types of structures typical of learners’ varieties of English and those traditionally found in English-based creoles. My discussion will base itself largely on the processability hierarchy of Pienemann

29

30

Fred Field

(2000), which lists a number of English-language structures typically learned in sequence by SL learners. To illustrate the proposed correlation of input (as process) and outcome (as product), Section 5 examines a sampling of English-based creoles whose creators had various degrees of contact with the lexical source language and its native speakers during the formation of their creole. Languages examined include Hawaiian Creole English, which perhaps has had the most direct and continuous contact with exposure to standard English and its speakers; Jamaican Creole English, which was shaped much more profoundly by the practice of slavery; and Tok Pisin, which had relatively little contact with any kind of English after its initial exposure to its lexifier, especially as it developed into a broad-based lingua franca in Papua New Guinea. To broaden the discussion beyond an exclusively English focus, Section 6 examines pertinent aspects of the Spanish-based creole Palenquero (Colombia). In “Conclusions” (Section 7), I discuss the evidence presented, and assess whether processes of SLA have indeed played an important role in the emergence of creole languages.

2. Birds of a feather: complementary proposals There is a popular conception that a pidgin lies at the source of every creole language. This view, known as pidgin-genesis, seems to have become traditional among non-specialists despite the scarcity of creole varieties with attested pidgins in their ancestry. This “creole < pidgin” position is, however, not without its challenges. For instance, according to Bakker, there is no “historical evidence for any Caribbean . . . or Indian Ocean creole being preceded by a pidgin” (1995: 38). Many creolists also write of so-called abrupt creolization, in which new varieties are said to emerge rather quickly with no attested pidgin stage (perhaps out of a pre-pidgin jargon stage) (Arends and Bruyn 1995: 115, Holm 1988: 8, Thomason and Kaufman 1988: 48). Recent research on the various colonial settings has revealed much relevant information on the social circumstances in which creoles have arisen. For example, linking usage in various dated texts with demographic information drawn from ship manifests and other historical records (listing age, sex, place of origin, and other characteristics pertaining to inhabitants of various colonial settlements), Lefevbre (1986), Lumsden (1999), Singler (1993, 1996), and others have found that the originators of some creole varieties were in fact adults (e.g., laborers prior to the birth or arrival of children). The findings of these authors significantly affect the view that creolization invariably involves pidgin-genesis, and that “nativization” (or acquisition of native speakers) is the first-language acquisition of a pidgin by children. As such, this goes against Bickerton’s claims (e.g., 1984, 1988). In essence, Bickerton’s “Bioprogram Hypothesis” asserts that children unconsciously create their own native language by absorbing the highly variable input of pidgin-speaking parents (presumed to be their only input) and by filling in the linguistic blanks with their innate or universal linguistic knowledge.


As forms and structures become conventionalized in common usage, crystallization — the stabilization and spread of the particular variety — can occur regardless of whether the emerging language did or did not have a pidgin in its history (cf. DeGraff 1999: 4). Additional characteristics of the dominant language can accrue in the emerging variety, provided that the lexifier/superstrate continues to be present in the creole social milieu. Because of the repeated reference to pidgin genesis in the literature on creolistics, creolization is often assumed to require an earlier pidgin stage. There are, however, pidgins that have not evolved into creoles, and creoles that have no documented prior pidgin stage. Pidgin-genesis and nativization as viewed by Bickerton simply cannot apply if the vast majority of the originators of the language identified as a creole were adults and not children. Stabilization and expansion must come from somewhere else, though the role of children cannot be minimized. If there existed no prior pidgin stage, something besides a pidgin must have served as the initial model for acquisition. Most introductions to creolistics (e.g., Bakker 1995: 25, Holm 1988: 8, Romaine 1988: 154) suggest that alternatives to pidgin-genesis are possible in the formation of a creole. One of these is a pre-pidgin jargon, though it still cannot be demonstrated that every creole had either a pidgin or jargon at its source (Muysken and Smith 1995: 6).² A jargon is generally described as a linguistic variety that is even less developed than a pidgin. Presumably like a pidgin, this variety is greatly influenced by substrate grammar. Elements associated with a jargon — content items, formulaic expressions, and a relatively small number of function words — are precisely those that one expects in a variety emerging in an environment in which access to the target language (TL) and its native speakers is severely limited. It is reasonable to assume that any language variety that draws its lexicon from a foreign lexical donor is the result of learning of some kind, even if that learning is reduced to merely picking out words and phrases for basic communication, with or without the cooperation and accommodation of speakers of the lexical donor/ superstrate. But not all specialists acknowledge that the linguistic environment in a creole community must have consisted of many different versions of the colonial language, both non-native and non-standard. Scholars also disagree as to whether the creators of creole varieties were ever in the process of learning/acquiring the language of their overseers and/or colonizers. Similarly, there exists disagreement as to whether the superstrate was the TL at any time in the development of any given creole. Thomason and Kaufman take up the challenge and offer that “there are a number of creoles for which a definite pidgin stage is not attested, and whose structure can be accounted for under a hypothesis of extreme unsuccessful acquisition of a TL” (1988: 48). In contrast to second-language acquisition, native language acquisition (NLA) of any language by children is typically characterized by maximal success. Non-native acquisition generally pales in comparison. While creators of a jargon, pidgin, and/or creole may not “succeed” in masterful acquisition of a heretofore

3

32

Fred Field

foreign language, they are, nevertheless, maximally successful in constructing an entirely new variety. Thus, pidginization and creolization are better seen as taking the lexicon of one language and encasing it in a new and different grammatical system likely to have been derived from a variety of sources, including original (native, substrate) language varieties via transfer, superstrate/lexifier via learning, and entirely novel constructs (born of necessity). Therefore, with regard to creoles, what constitutes “successful acquisition” needs to be established according to a different set of criteria.

2.1 Interlanguages and pidgins contrasted It may seem natural to SLA specialists to associate individual ILs with pidgins, especially in view of the body of work that has focused on the structural similarities between ILs and pidgins (e.g., Andersen 1980, 1983, 1984, Schumann 1978a, 1978b, 1978c, Valdman 1980). As Schumann’s earlier work demonstrated, both ILs and pidgins show a noticeable lack of TL/superstrate inflectional morphology. Moreover, they share many of the same structural characteristics, for example, word order patterns typically associated with IL grammars and the observable effects of underlying, native languages via transfer. This is not a matter of coincidence since, as Cook points out, “[T]he dynamics by which pidgins and creoles come into being and the development of the L2 learners’ interlanguage might be governed by the same factors” (1993: 69). To specialists in both SLA and creolization, the definitional problem hinges primarily on two issues, i.e., (1) whether or not all learners’ varieties are pidgins, and (2) whether all pidgins are learners’ varieties.³ Although learners’ varieties and pidgins may share numerous structural characteristics, they are not the same. Not every speaker of an English-based pidgin (e.g., Tok Pisin) learns L2 English, and not every L2 learner of English speaks a stabilized pidgin. In fact, some individuals may speak both a pidgin and its lexifier (Bakker 1995: 27). Pidgins differ from ILs in terms of the kinds of processes that are involved: social in the case of pidgins, and psychological in the case of ILs. For example, pidginization is said to be the process by which native speakers of two or more separate languages create a contact variety for specific, communicative purposes (trade, inter-group communication, and the like). The social focus in pidginization differs in kind from the focus typical of ILs, particularly when the focus of the latter is ongoing psychological processes and the behavior of individuals rather than that of groups. However, it may be difficult to differentiate clearly and unambiguously the social from the psychological regarding individual learning environments and those associated with the communities in which pidgins arise. Entire communities of (individual) SL learners may share many social and psychological characteristics with creators of pidgins. These characteristics include, for instance, issues of social distance from native speakers of target/superstrate languages and the gradual psychological processes associated with acquisition/learning.


Granted, there are obvious differences. Learning processes within a community in which a relatively stabilized pidgin is in use are likely to involve the pidgin itself as a target, rather than just the principal lexifier. The lexifier (to the degree that it is present in the environment) may serve as an additional source for lexical material and/or structural characteristics, but so will substrate strategies and other innovations. As a consequence, forms and structures from pidgin speakers plus native and non-native speakers of the target (including regional and/or social varieties of the target) may compete. Taken together, the varied input can account for other points at which a pidgin may diverge from simple IL versions of the same lexifier, which may or may not have input from native/proficient speakers as its primary source. Multiple, competing targets, instead of the relatively consistent model of acquisition assumed to be typical of most situations of NLA and SLA, will undoubtedly influence crystallization and nativization. In descriptions of the linguistic processes of SLA — and in particular when these descriptions consider social factors (Schumann 1978a, 1978b, 1978c) — the term pidginization is often used freely and without terminological precision. In SLA as well as pidginization, grammatical complexity appears to be greatly reduced. “[P]idginisation thus constitutes restrictions in use accompanied by the reduction in form” (Andersen 1983: 4). The converse may be equally true: a reduction in form may be accompanied by restrictions in use. With respect to ILs, it is not difficult to picture how input (linguistic data available for acquisition or learning) can be restricted in some sense, particularly when speech is limited to registers focused on specific tasks at hand (e.g., physical activities associated with the implementation of a plantation economy). Target norms (i.e., norms typically reached by all native speakers of the TL) may not be reached, thus leaving the emerging interlanguage grammar simple(r) by comparison.

2.2 Learners’ varieties across a community When any version of a heretofore “foreign” language (pidgin, IL, or otherwise) becomes the principal, full-fledged language of a community and the principal medium of communication, it can subsequently spread and stabilize (DeGraff 1999: 4). The emphasis of our investigation shifts from the processes of language acquisition to the results of that acquisition, which is language creation and evolution. As a consequence, it is a small step to broaden Holm’s often cited definition of creole by adding the term IL: “A creole has a jargon, pidgin or IL variety in its ancestry” (1988: 9; the italics are mine). Within a community as a whole or within particular subgroups, a patchwork or aggregate of learners’ varieties (or ILs) can certainly achieve greater and greater communicative effectiveness, particularly as it develops into a lingua franca used for everyday functions. Such an outcome, in which an aggregate of ILs becomes the variety associated with a particular social group, is typical of urban societies worldwide. To communicate effectively, people do not need to be completely proficient native

33

34

Fred Field

speakers, nor do they need to achieve native-like competence. This is also the case in an emerging creole community. When children acquire the lingua franca (e.g., the IL of parents or other community members) as their primary language, and use it to communicate with each other and with adults, the likely effect will be to increase its spread and stability. When IL varieties are the sole or primary input for native acquisition by children, the developing grammars will almost certainly diverge from adult, native versions of the lexifier. All input for acquisition is potentially shaped by social circumstances. The fact that different languages are spoken within a single community provides an opportunity for acquisition, but it does not necessarily afford the best or most fruitful learning environment, particularly if the goal is mastery of the TL (attainment of native-like proficiency). Mastery, in this sense, takes time and considerable effort. Learning to use a language for limited, functional purposes takes much less time and effort. This is an especially important point if we are to give SLA a fair hearing with respect to creole genesis. All other things being equal, the greater the quantity and quality of the input (in terms of access to native or native-like speakers), the more target-like the resultant IL or creole will be. As a consequence, the structural similarities between a creole and its lexical source (the one-time target of SLA) will be proportional to (a) the quantity and quality of contact, and (b) the social relationships established among groups. Differences in learning environments can account for a disparity in the respective outcomes. To illustrate this aspect, compare the kind of linguistic input for acquisition one expects in parent-child and teacher-student relationships to the input typically available to a plantation labor force — a situation in which the input comes from slave owners, overseers, taskmasters, and other slaves. One expects a clear contrast between the success associated with NLA and limited (comparatively unsuccessful) learning associated with SLA, especially within the plantation context. Deep divisions between social groups along the lines of master and slave (plantation laborer, servant, etc.) widen social gaps, inviting the formation of separate social and linguistic identities. The expected outcome in a plantation environment would be very consistent with the often noted absence of lexifier/superstrate (the principal lexical donor and onetime TL) structures and inflectional morphology, which are typical of both (a) limited proficiency in the TL and (b) creole varieties in general. Gardner, writing of complex sets of motivational factors within a socio-educational model of language learning, asserts that near-native-like proficiency “implies a level of expertise, and research on the development of expert behavior indicates that it requires approximately 10 years of consistent and persistent practice. . .” (2001: 4). Ten years is approximately the same amount of time that it takes a child to develop adult speech natively (Ericcson, Krampe and Tesch-Römer 1993). As regards learners’ desire to identify with speakers of the socially dominant language, the so-called “integrative” motivation, Gardner states: “The variable Integrativeness reflects a genuine interest in learning the second language in order to come closer to the other language community” (2001: 5). It is easy to see how the desire to move


linguistically and socially closer to the target language community can be a factor in a classroom setting. Likewise, the consequences of the absence of this integrative motivation in plantation settings is not difficult to imagine. On plantations, learning may have been done grudgingly, at best. In describing motivation as (a) the expenditure of effort, (b) the desire to achieve the goal of native-like proficiency, and (c) enjoying the learning task, Gardner (2001: 6) identifies three factors that appear to be missing or that at best differ in significant respects in a majority of developing creole communities. Granted, there may have been notable exceptions, and the socio-cultural histories of many communities are obscured by the lack of documentation (see, for instance, Schwegler and Morton 2003 for a discussion of Palenquero speakers and their attitudes towards regional Spanish). The social milieu characteristic of many such slave communities minimized the time and opportunities necessary for practice, and greatly reduced the motivation to approximate the speech of the ethnically distinct European members of the community. In general, it seems unlikely that mastery of the colonial master’s language was the priority of every learner within the creole community. Functional proficiency rather than native-like mastery may thus have been the target. If we add that non-natives learned from other non-natives, that later waves of African laborers learned the emerging creole from the speech of SL learners of the master’s tongue, then the outcome surely had to be “something other” than the master’s tongue. Forced servitude and the linguistic “opportunities” it presented surely set up (foreign) language learning environments that differed profoundly from those found today in university or other academic institutions, where classes are purposefully designed to enhance rather than restrict acquisition.

3. Evidence for second/subsequent language acquisition Along with the sociohistorical evidence, two kinds of structural evidence should be anticipated in an emerging creole, each depending on the amount and quality of exposure speakers of the emerging variety had to the superstrate/lexifier and its native speakers. The first type of evidence is the result of acquisition, and is reflected in the availability and usage of superstrate or target forms and structures. Particular kinds of words and morphemes are typically acquired before others (e.g., nouns, verbs, and adjectives are learned before function words and/or inflections). This is true for children acquiring a language natively (NLA) as well as for individuals who acquire a language non-natively (SLA). In an individual (or aggregate) IL, certain inflections and other functional elements may not always be used according to adult native-speaker norms. Consequently, a hierarchy of forms should be evidenced in an emerging creole, with functional elements being systematically omitted (without additional exposure and perhaps instruction in TL forms). The structures available to speakers will also reflect the ways that TL structures have been learned (where in the sequence acquisition has stopped) and which syntactic constructions will not

35

36

Fred Field

be available at all. For example, question formation can be accomplished phonologically through a change in intonation (suprasegmentals), rather than syntactically via reordering rules. Consequently, inversion cannot occur if it was never learned. The second kind of evidence is transfer, the unconscious influence of native languages on SL learning which can have long-lasting, even cross-generational effects. Transfer in SLA is tantamount to substrate influence in the emergence of a pidgin variety, and is the basis for many so-called substratist approaches to creole genesis (cf. Holm 1988: 106).⁴ To the specialist in acquisition, transfer is not only real, it is a direct link to an underlying native language. Take away the presence and influence of the TL of SLA (the superstrate) and its speakers, and its influence will be diminished. This assumes only that contact with the TL provides an environment in which learning (SLA) is a possibility (cf. Bardovi-Harlig 2000: 404). The influence of the native language (or substrate), as one of the variables in acquisition, may by unrestrained if native-speaker input of the target language is absent. The longer lasting the presence of the superstrate/lexifier and its native speakers is, the more targetlike the product of acquisition will be.⁵ Superstrate and substrate appear to exist in a dynamic relationship, just like TL and NL in SLA. The final outcome will correspond to the interaction of social (historical) events and resultant linguistic interaction between or among language systems. Each of the three English-based creoles discussed here — Hawaiian Creole English (henceforward HCE), Jamaican Creole English (JCE), and Tok Pisin (TP) — has a separate history involving considerable differences in the amount and kind of exposure to English and its native speakers. For instance, both HCE and JCE speakers have had constant and uninterrupted contact with native speakers, though social conditions characteristic of their interactions differed widely. The original speakers of TP had relatively brief exposure to English and its speakers. Even a cursory glance at these three languages shows considerable difference in the ways that characteristics of underlying languages manifest themselves.

3.1 NLA and SLA contrasted Unlike native language acquisition (NLA), in which full and complete attainment of proficiency is the normal outcome, SLA is typically associated with considerable variation in degrees of “success” (however that might be measured) in the attainment of full proficiency. The expected structural consequences in a developing IL are found mostly in the sorts of forms that are not fully acquired (and are therefore missing, in a sense) and the particular structures that develop short of those exhibited in adult, NL versions of the TL. NL acquirers are exposed to the entire (target) language at once (not as sequenced lessons in a textbook or classroom). In contrast, the input for SLA is often quantitatively limited by the numbers of native speakers serving as models of acquisition (sometimes, only a teacher) and the environments in which the language is spoken (classroom or occupational setting), which restricts the forms, structures, and registers available to the learner. In a qualitative sense, the


kinds of forms and structures that naturally accompany a wide range of registers will have an impact on language development. Input that is somehow limited may affect both (a) perception and the receptive kind of knowledge required for on-line word recognition, and (b) performance and the productive knowledge required for the production of speech (cf. Nation 1990: 29–50). Such quantitative and qualitative restrictions stand to be exacerbated considerably when motivation for acquisition is low, which must have been the case in many slave settings. Another characteristic outcome of NLA is the ability to employ a variety of structures and forms for expressive and stylistic purposes. The native speaker recognizes ambiguity, and is able to paraphrase individual concepts and utterances in a variety of ways (lexically via synonyms, hyperonyms, and hyponyms; and structurally by reordering elements, as in adverb fronting, alternating utterances in the passive or active voice, and so on). This is not a guaranteed outcome for the SL learner, whose IL may have relatively few, hence, limited or restricted options, especially when compared to those of the native speaker. Ambiguity may go undetected. In fact, SL learners often prefer simplicity and seek “one form/structure for one intended meaning” and the reverse,“one meaning for a particular utterance”. As a consequence, the available alternatives in SLA may reduce to one within the speech community (groups of SL learners) in accord with Andersen’s (1989: 388) One-to-One Principle. Developmental stages in native or non-native acquisition can be observed in a number of ways, for example, (a) in the order that forms and structures are mastered (certain forms are typically acquired before others), and (b) in the order in which particular constructions are learned (e.g., gradual acquisition of negation, question formation, and so on in English). Particular forms (inflectional affixes or other functional elements) may not appear consistently in obligatory contexts (according to TL norms), and the acquisition of constructions that typically follow specific developmental sequences may cease at some point along the path, leaving remnants of those steps that have been halted, processes which led Selinker (1972) to coin the terms fossilizing and fossilization. Consequently, as regards individual ILs, SL learners may reach a certain point along a continuum from zero to full knowledge of the target language, and simply stop learning. As a structural consequence, IL varieties typically exhibit certain, identifiable features, vis-à-vis the forms and structures typically attained in NLA. While positive evidence in the absence of prior linguistic knowledge and competing alternatives from a “mother tongue” seems to be sufficient in NLA, negative feedback (error correction) is typically necessary in SLA, instructed or not. Problems are compounded in particular groups of SL learners. For instance, patterned errors may go unnoticed, particularly by students/learners with similar or identical nativelanguage backgrounds (Lightbown and Spada 1999: 141, 152). This creates the need for teacher intervention to help students continue along developmental paths, and to prevent the fossilization of errant forms. Systematic errors can create a kind of “foreigner talk” easily associated with a particular L1 (or group of similar native languages). For instance, in their interlanguage versions of English, native speakers of

37

38

Fred Field

various dialects of Spanish, a pro-drop language, may systematically omit subject pronouns. Word-form errors in L2 English may be undetectable to L1 speakers of Vietnamese, in which words do not change form (e.g., in the selection of difference, different, differently), and speakers of Chinese languages may use specific lexical devices to topicalize certain elements in an L2 English sentence (with There is/are), creating ungrammatical double-subject utterances that are virtually impossible for Chinese speakers to recognize as errors.⁶

3.2 SLA: the effects of instruction The role of instruction is particularly important to consider whenever the outcome of NLA in the normal transmission of the European language by colonists is compared to the kinds of SLA that took place among non-Europeans during the establishment of New World colonies. The reactions of Europeans (who were native speakers of the colonial languages) to the speech of African laborers who attempted to speak the European languages in Caribbean plantation societies were characteristically negative. Criticism often appeared to be racially biased. In the opening remarks to his well-known introduction to pidgins and creoles, Holm candidly sums up early reactions to pidgins and creoles: What earlier generations thought of pidgin and creole languages is all too clear from their names: broken English, bastard Portuguese, nigger French, kombuistaaltje (‘cookhouse lingo’), isikula (‘coolie language’). This contempt often stemmed in part from the feeling that pidgins and creoles were corruptions of “higher,” usually European languages, and in part from attitudes toward the speakers of such languages, who were often perceived as semi-savages whose partial acquisition of civilized habits was somehow an affront. Those speakers of creole languages who [eventually] had access to education were duly convinced that their speech was wrong, and they often tried to make it more similar to the standard. With few exceptions, even linguists thought of pidgin and creole languages as “aberrant” (Bloomfield 1933: 471) if they thought of them at all — that is, as defective and therefore inappropriate as objects of serious study. (1988: 1)

Formal (class-room) language instruction appears to have positive long-term effects on the rate of acquisition and the ultimate levels of attainment. Classroom instruction containing both positive and negative feedback seems to speed up the learning of certain structures (without influencing either the order in which forms are learned, or the sequences in which structures are acquired). Long-term or ultimate attainment may be enhanced because mastery of many of the more abstract and/or complex grammatical aspects of a target language may require instruction and didactic attention to form. Some language-specific and highly grammaticalized forms and structures may simply not be inferred from the performance of native speakers or drawn from written discourse. If a learner’s goal is to comprehend oral or written language, semantic facets of the target, not specific grammatical features, may be the focus. Moreover, once comfortable levels of performance have


been reached and further advancement becomes increasingly difficult or tedious, academic incentives (e.g., promotion to more advanced ESL classes or release from stigmatized ESL programs altogether) may not provide enough motivation. One specific facet of SLA that has engaged researchers recently is the expression of temporality (acquisition of the TL tense-aspect system). Bardovi-Harlig (2000) makes the following generalizations: (i) In the earliest stages of SLA, there is no systematic use of tense-aspect morphology. Temporal reference is established in four ways: (a) by relying on the contributions of their fellow speakers (scaffolded discourse), (b) through reference inferred from a particular context (implicit reference), (c) by contrasting events, and (d) by following chronological order in narration (p. 25). (ii) The second stage is lexical expression, the use of time adverbials and so on (p. 36). (iii) The morphological stage is reached in which inflectional morphology gradually emerges in the learner’s interlanguage (p. 93). This final stage in the learning process demonstrates the successful acquisition of TL inflectional categories and their exponents. This description of SLA — in relation to the development of tense and aspect in individual Ils — is familiar to most creolists who have concerned themselves with the so-called structural expansion of pidgins or creoles. The morphological stage appears not to be reached in creole formation. It is replaced (or made unnecessary) by the emergence of pre-verbal TMA markers characteristic of creoles, and serves as clear evidence of language creation/evolution.⁷

3.3 SLA in untutored settings An investigation by The European Science Foundation (ESF) of foreign immigrants in Western Europe focused on the (untutored) acquisition of English, German, Dutch, French, and Swedish, and paid special attention to the difficulties that these SL learners had with tense-aspect systems while trying to adapt to a new social and linguistic environment. Learners in such situations appear to be relatively successful in communicating, as they function adequately at work and in everyday conversations when using their emerging language varieties, and this despite the lack of full mastery of verbal inflections. The ESF study sought to characterize the verbal exchanges that took place between these newcomers and native speakers of the target languages. The aims of the study were to uncover some of the linguistic and communicative factors that can account for (a) varying degrees of success that adult learners display in their attempt to acquire the subsequent (foreign) language, (b) at least some of the stages of proficiency through which learners pass en route to mastery of the target, and (c) why this type of language learning often grinds to a halt at a low yet manageable level of proficiency. Dietrich and Perdue mention the frequent observation that the speech of untutored adults — and we certainly can add tutored ones,

39

40

Fred Field

as well — is relatively consistent with certain amounts of individual variation, and “simple in relation to the TL” (1995: 3). It may appear “simple” or “restricted” because only some tense and aspect markers have been mastered. But the question remains as to whether native-like standards are attainable in SLA. Fossilization of such “incomplete” TMA systems is likely when communicative efficiency (but not proficiency) has been reached (Klein 1995: 68). A few preliminary comments are necessary before attempting to relate the learning of European temporal systems in untutored situations to a creole scenario. As already mentioned, the originators of creole varieties encountered a social environment that differed considerably from that of the immigrants described in the ESF studies. Specifically, “everyday contact with speakers of the new language” was restricted from a qualitative perspective, which holds true particularly during times when laborers or slaves outnumbered the European masters.⁸ In general, one would expect the range of registers of speech to be focused on specific tasks, with a very limited range of native interlocutors, topics, settings, purposes, and so on. As a consequence, the speech of learners in a creole context would be expected to show many of the same sorts of characteristics that SL learners exhibit in other untutored settings, though for creoles the outcome is likely to be even more limited.

4. English as a second language Pienemann (1998, 2000) claims that processability is a key factor in understanding limits on second or subsequent language development. Certain morphological and syntactic milestones are reached in a predictable sequence. Consequently, by comparing what is absent in a creole (forms and structures that are unavailable) with what is present in the assumed target, one may be able to assess at which point(s) in a particular sequence a community-wide learners’ variety or IL stopped. For example, if the third-person, singular -s suffix of English present tense (marking subject-verb agreement) does not appear in a particular English-based creole, then we can say that acquisition stopped prior to the point at which it is normally acquired by speakers of the standard. Focusing on the acquisition of English as a second language, Pienemann (2000) proposes a set of 14 morphosyntactic constructions according to six levels of processability.⁹ His concept of processability is consistent with a particular view of mind/ brain architecture as the human language processor. When processability measures are applied to a variety of forms and structures, these forms and structures can be ranked according to the amount and distribution of information. In this perspective, says Pienemann, “the language processor is seen [. . .] as the computational mechanisms that operate on (but are separate from) the native speaker’s linguistic knowledge” (1998: 5). Thus, the brain’s ability to process linguistic information is dependent on the complexity of form with respect to its function or meaning, as well as, conversely, on the complexity of information expressed with respect to form.¹⁰

Second language acquisition in creole genesis Table 1. Structures according to processability (adapted from Pienemann 2000: 116)¹¹ Level

Construction

6 5

Cancel inversion (in indirect discourse) Aux/do 2nd (subj.-aux. inversion in wh-movement) 3rd person singular present tense -s (subj.-verb. agr.) Y/N inversion (subj.-aux. inversion) Particle verbs (movement of verb particle) Copula inversion (in WH-movement) Neg + V (invariable negation with no + V) Do-fronting (question formation, without agr)¹² Topicalization (S. initial non-subject NPs) Adverb(ial) fronting/movement¹³ SVO (canonical word order) Plural (on nouns) Possessive pronoun (Det) Single-word utterances

4 3

2 1

In the following, only those constructions cited by Pienemann that are relevant to the present discussion will be described. In general, Level 1 is the word (or lemma) level; 2 is phrasal (all operations occur with individual phrases); and 3 involves moving an individual element to focus position without any other consequences to the clause (or CP). Levels 4–6 involve the sentence level, what Pienemann calls interphrasal procedures. For instance, information required for obligatory subjectverb agreement involves feature/diacritic checking at the clausal (CP) level. These rankings are listed in Table 1. What follows are a few preliminary observations on how these rankings may manifest themselves in an English-based creole variety.¹⁴ First, the consistent appearance of all Level 2 constructions is not always evidenced. Nevertheless, at least some of the constructions contained in Level 3 are quite typical, in particular Neg + V (Holm 1988: 161). In fact this particular strategy is characteristic of all ILs in which English is the target. The use of Neg + V is typically the first option irrespective of native-language background (Larsen-Freeman and Long 1991: 94–6). Regarding the interplay of do-support and negation and question formation, Holm states: . . .there is nothing like do support in English-based creoles; the negator simply precedes the verb (“im no wier shuuz” ‘He doesn’t wear shoes’) and there is no inversion of the subject and the auxiliary (or verb) to form questions (“im wier shuuz?” ‘Does he wear shoes?’). Intonation alone can distinguish questions from statements. . ., but there are also question markers for emphasis. (1988: 212)

It appears that, as a group, English-based creoles are located somewhere at levels 2 and 3. They evidence acquisition of many of the most salient features of the TL grammar (those required for communicative purposes), particularly in view of the

4

42

Fred Field

fact that effective and functionally equivalent alternatives eventually developed (e.g., preverbal T/A markers). This degree of acquisition in creoles highlights two issues. First, and going beyond the issue of acquisition, creole syntax predictably differs from that of its lexifier. One can assume that sentence-level procedures are required for a complete language. As a consequence, the additional levels of grammatical complexity appear to evolve independently of TL structures. Thus, sentencelevel procedures must emerge subsequent to any pidgin or learners’ stage (i.e., ILs that may have also provided input) in what may be called “structural expansion”.¹⁵ Secondly, it seems reasonable to assume that prior linguistic knowledge (i.e., the influence of a mother tongue, NL, or substrate variety) may have provided the originators (and subsequent speakers) of these creoles with the necessary syntactic and/or semantic resources to meet functional and expressive needs. Regarding the higher levels, Particle verbs, ranked at Level 4, involve movement of the particle, as in,“Would you please turn that music down?” The particle down has moved to a position following the direct object music. Two-part verbs (so-called), or VERB + PARTICLE, are used in Jamaican Creole English, for instance, but particles invariably follow immediately after the verb. For example, one never hears “Put de ting down!” The VERB + PARTICLE always appears as “Put-down de ting” (Peter Patrick, personal communication). At the top Level 6 (the highest level of processing difficulty), cancel inversion involves two kinds of information: (a) knowing rules and processes, and (b) knowing when they do or do not apply. The inversion of subject NP and auxiliary characteristic of English question formation requires fairly high levels of proficiency (Levels 4 and 5). Cancellation goes a step further. In keeping with the well-known U-shaped learning curve applied to the native acquisition of irregular past-tense verb forms, children appear to acquire first, in sequence, the rule (process or pattern); then only subsequently in their later development do they learn exactly when or where it does or does not apply (in the latter case it is sometimes referred to as blocking; cf. Marcus et al. 1992: 8).¹⁶ Concerning level 5 and the 3S -s, both the gradual development of a simple system of temporality (discussed in 2.2, above) and processability are relevant. This particular affix is notoriously difficult for ESL learners, irrespective of native-language background. It might be tempting to predict that SL learners with tense-aspect systems thought to resemble that of English would have an easier time mastering temporality in English. However, prior knowledge of Tense and Aspect is of little help with its consistent application along TL, native-speaker norms. The 3S -s still needs to be learned via a (normally) lengthy acquisition. Concerning tense, the inflection-poor English tense system is not similar to, say, Spanish, Italian, or Portuguese with their rich verbal morphologies. English verbal morphology does not really resemble most other Germanic systems either. What is left of the original Anglo-Saxon verbal morphology is the regular past-tense marker (-ed), which does not involve subject-verb agreement, and a single present-tense marker (-s). There are several possible reasons for the difficulty learners experience with this morpheme, despite the fact that it appears frequently in the speech of users of pres-


tige or so-called standard varieties, and seems easy for native speakers to process. This notorious -s (as distinct from the plural and genitive markers on nouns) requires a great deal of knowledge of English grammar. First, 3S -s is only applied when the subject NP fits certain criteria. It can only be (a) a 3rd-person, singular pronoun (e.g., she, he, it, or one), (b) a proper noun (the name of an individual, not group of individuals), or (c) a common noun that refers to a concept or entity that is both countable (e.g., one chair, two chairs) and singular (there is only one chair). Second, and making matters even more complicated, the 3S -s also is applied when the grammatical subject is a non-count noun (e.g., mass nouns, such as rice or water). Thus, the count-mass distinction for nouns is at the heart of the problem of ESL learners with respect to subject-verb agreement. Learners must initially decide whether a noun in the subject NP slot is countable or uncountable (the inflectional category of Number is obligatorily expressed on countable nouns in English), and then, in the event that it is countable, whether it is singular or plural. Perception and the ability to respond to spoken speech does not require the application of the suffix -s. The -s is indeed redundant and, therefore, unnecessary for comprehension. The only thing required is the passive knowledge necessary for a semantic interpretation (understanding what one reads or hears). It is quite another matter to be sufficiently aware of the idiosyncratic grammatical properties of the English nominal system to produce accurate utterances consistently. Third, the learner must be familiar with the usage of the present tense of English, which differs from that of many other languages. It is idiosyncratic, too, so much so that many linguists would not even call it a true present tense. In fact, it is typically used not to make reference to time, but to express simple statements of fact, opinion, or belief that are assumed to be true in the past, present, and future. The only thing a speaker/hearer truly “knows” about the event depicted by a verb so marked is that it did not occur at a point in time in the past; hence, this tense in English is typically called the non-past. In my experience, Vietnamese L1 speakers (whose NL has no formal device to mark tense or number of any kind) and Spanish L1 speakers (whose NL possesses a tense-aspect system much richer than that of English) both have immense difficulty mastering the standard (American) English system. A number of conflicting explanations have been offered for this difficulty, one being exposure to non-standard varieties of English. The fact remains, however, that this little “nuisance” (the thirdperson, singular, present tense -s) is positioned last in the often-cited morpheme order studies (e.g., Brown 1973 regarding NLA, and Dulay and Burt 1973 regarding SLA) and near the top of the rankings with respect to processability and complexity. Surely, this is not a coincidence. Finally, the appearance of this particular affix may be relatively infrequent in the kinds of spoken varieties that many learners encounter, in particular those registers of speech that may have been the norm for adult laborers in a plantation (creole) setting. In this way, input may also shape the outcome to a great extent. In a study of rather large corpora involving both spoken and written registers of English, Biber

43

44

Fred Field

(1988) outlines a number of correlations between situations of speech — genres in Biber’s terminology — and the occurrence of registers, i.e., the co-occurrence patterns associated with particular lexical items, function words, and other grammatical characteristics. Of particular note is (a) the frequent occurrence of firstand second-person pronouns (requiring no -s on accompanying verb forms) in informal, personally involved registers, e.g., face-to-face conversations, and (b) the greatly increased frequency of third-person pronouns (the singular, which triggers the -s on the verb via agreement, and plural, which does not) in more formal and information-driven written registers, e.g., in written, academic registers. Considering, too, that plantation masters and the overseers directly responsible for the work force most likely gave commands quite often (no -s on the imperative form of the verb) and engaged in direct face-to-face discourse to get things done, redundant 3S -s may indeed have been infrequently used at best. One may wonder how critical this sibilant was for effective communication among laborers and their overseers, who may have spoken non-standard varieties or learners’ forms of English.¹⁷

5. English-lexicon creoles We turn now to Hawaiian Creole English, Jamaican Creole English, and Tok Pisin. Regarding JCE, a pidgin stage has never been reported for any English-based Caribbean creole — although some scholars (e.g., Cassidy 1971, 1980) have theorized that some of the original inhabitants of Jamaica may have been familiar with one that originated in West Africa or, perhaps, Barbados. The African-origin creole speakers today are also usually familiar with local varieties of English, and are exposed to Standard Jamaican English (based on a British educational model) in the classrooms. Both HCE and TP had attested pidgin stages, though the social circumstances in which they arose differed somewhat. For example, in TP, a wide range of native languages existed, although some may have shared a number of areal features (Mühlhäusler 1990, Reesink 1990). Originators of HCE were composed mostly of large numbers of immigrants (an imported labor force) who spoke Cantonese, Portuguese, Japanese, Korean, and Filipino languages. HCE later also spread to speakers of the Polynesian language Hawaiian, and has become an important marker of social identity to the islanders more generally.¹⁸ Africans were forcibly brought to Jamaica as a source of slave labor where they came into contact with colonial forms of British English. In such a social context, one expects the maximum social distance between slave owner and slave, with overseers having an intermediate status in the labor force. In contrast, English-speaking colonizers went to Hawai’i, bringing with them considerable social pressure to the islands’ inhabitants to adopt the colonial language. The islanders, including indigenous Hawaiians and foreign-born laborers, were not happy with their colonial status, and, as a result, they have traditionally maintained a strong sense of ethnic solidarity and separateness from colonial society. While there may be consider-


able social distance between various islanders and “mainlanders” (generally, white people, or haoles, from the continental US), the creole that emerged is not defined in terms of forcible kidnap and slavery as was the case with Africans in Jamaica, though in both cases the rise of a creole is certainly linked to a labor force and the distance that results from clearly asymmetrical social relationships. As a consequence of the linguistic diversity in areas around modern-day Papua New Guinea (PNG), TP has functioned in a unifying way, as a lingua franca, among ethnically and linguistically diverse peoples. When the labor trade developed in the southwestern Pacific, laborers took what they could glean from the language of their English-speaking employers or overseers and fashioned a functional variety to fit their communicative needs. From its origins on plantations in Queensland (Australia) and Samoa, TP was taken to the island of New Guinea where it took root and spread (Siegel 1998: 347, 1999: 9). At present, TP is spoken natively by some speakers in PNG (hence its creole designation), and has reached a relatively high level of social prominence (Litteral 1990: 376, Wurm 1980: 239). The following subsections (5.1–5.3) illustrate the superstrate structures occurring in these three English-based creoles. In each subsection, attention is drawn to Level 4 structures — a threshold of sorts.

5.1 Hawaiian Creole English In HCE, as expected, there are no past or non-past (present) inflectional markers/ affixes. HCE does, however, have preverbal elements that indicate time reference (lexical tense). Speakers of this variety may also be proficient speakers of American English (as a result of statehood and universal public education). Knowledge and usage of the creole are clearly associated with the local population and culture. The following passage of HCE comes from a translation of the New Testament book of Mark (1:21–2).¹⁹ The orthography represents a somewhat conservative effort to combine local, phonetic-like spellings with those of the standard. (1) Da guys all wen go Capernaum. the guys all PST go Capernaum ‘The disciples went to Capernaum. On da nex Res’ Day, dey wen go inside one on the next rest day they PST go inside a on the next Sabbath, they went inside a Jewish church. Jesus wen teach all da guys dea, Jewish church Jesus PST teach all the guys there synagogue. Jesus taught all the disciples there and wen shock’ em, eh? ‘Cause da stuff he and PST shock them, eh because the stuff he and shocked them, eh? Because the things

45

46

Fred Field

teach get plenny powa, man. Not like da teach get plenty power man not like the that he taught had much power. He was not like the odda teachas who suppose to know da law. other teachers who suppose to know the law other teachers who were supposed to know the law.’ From this brief example, a number of observations can be made. First, word order is far from simple. Canonical word order is evidenced in the opening sentence (Da guys all wen go Capernaum) (Level 2); adverbial phrases are fronted (e.g., On da nex Res’ Day,. . .) (Level 3). However, no structures from Level 4 and above occur in the examples cited. There is no Inversion in the formation of any questions (no Aux/do in second position), and no 3S -s. Example 2 (HCE spelled out phonetically as in the source)²⁰ illustrates these features. (2) Weə yu laɪ go den? where you like go then ‘Where would/do you like to go, then?’ (Carr 1972: 46) (3) Eh! Leave us alone! What we wen do to you? Hah eh leave us alone what we PST do to you hah ‘Eh! Leave us alone! What did we do to you? Hey, Jesus! You wen come heah fo kill us? Eh! I know Jesus! you PST come here to kill us? eh I know Jesus! Did you come here to kill us? I know who you, you know. You da kine, you God’s who you, you know. you the kind you God-POSS who you are. You are the one; you are God’s bes’ one he wen send, eh? best one he PST send, eh chosen one whom He sent.’ (4) And den everybody was so shock, dey all wen and then everybody COP so shock they all PST ‘And then everybody was so shocked that they all talk da same time: Eh! What dis? What talk the same time eh what this what talked at the same time. Eh! What is this? What kine teaching dis? So diff ’ren, yea? He so kind teaching this so different eh? he so kind of teaching is this? [It is] so different. He is so

(Mark 1:24)


powaful, dat even da bad spirits obey ‘em powerful that even the bad spirits obey him powerful that even the evil spirits obey him.’

(Mark 1:27)

The question-word weə is in sentence-initial position, consistent with Level 3 constructions, as are the question words in (3) and (4). Note, there are no instances of Inversion or Aux/do second, nor are there any other Level 4 (or higher).

5.2 Jamaican Creole English Fairly typical of the English based creoles, JCE has no third-person, singular -s, and no consistent usage of the regular past-tense suffix, -ed. A preverbal perfective marker, don, is often interpreted as indicating events in the past. In many longer samples, standard English forms may appear that are not necessarily part of the creole, or patwa. On the island, creole, vernacular, and standard varieties are distributed along a continuum that is not unlike the registers of a single language. However, the underlying systems (lects) differ much more from each other than, say, the kind of register variation found in non-creole communities. In its basilectal forms, JCE is basically unintelligible to speakers of various standards, though most JCE speakers are proficient in at least a local standard or other so-called lects, from basilectal to mesolectal to acrolectal varieties (Patrick forthcoming). Authors refer to codeswitching in those instances where speakers clearly alternate from creole to standard forms, or when switching from one lect (e.g., basilect) to another (acrolect). Initial exposure to the official language of education may occur in the classroom — but well after the creole system has been acquired natively. Taken from Le Page and DeCamp (1960: 141), example (5) below is basilectal JCE. (5) Di uol liedi sie, tan! the old lady said wait [stand] ‘The old woman said, “Wait!

JCE

a wa de go hapm? wilyam sie, wa what PROG go happen william say what What is going to happen? William says, what de go hapm. yu wil fain out. PROG go happen you will find out is going to happen. You will find out.”’ Examples (6) and (7) illustrate both word order in questions and preverbal T/A markers (from Patrick forthcoming). (6) Wa di inglish stuor did niem agen? What the English store PST call again? ‘What was the English store called again?’

47

48

Fred Field

(7) Sapuoz man ben get op an kyatch yu hin de? suppose man PST get up and catch you in there ‘Suppose somebody had gotten up and caught you in there?’ As (6) and (7) illustrate, both ben and did are used as preverbal markers of past (anterior) tense (Patrick forthcoming). Both (6) and (7) are interrogatives, but the word order remains fixed as SVO. Other Level 3 characteristics found in JCE are Neg+V, as in (8) (Patrick forthcoming): (8) A man jos a faiya shat, im no biznis huu it kyach. a man just PROGR fire shot, him no business\ who it hit ‘A man is just firing shots. He doesn’t care who they hit.’ Overall, constructions from Level 4 or higher do not occur (this includes 3S -s of Level 5). Even with the absence of higher level structures, JCE has a wide range of syntactic (tense and aspect) alternatives (e.g., preverbal markers). The absence of standard English (obligatory) word order patterns (movement) in Inversion and so on seems to have little effect on the expressive opportunities available to speakers of JCE.

5.3 Tok Pisin The following sample comes from A New Course in Tok Pisin (Dutton 1985: 27). This sample offers a general picture of TP’s structural characteristics, and how these differ from those of HCE or JCE (see Verhaar 1990 for extensive discussion of particular characteristics of the language). (9) Bihain long taim wok i pinis after PREP time work PRED finish ‘After work, Dagi i go long klab. Em i go insait Dagi PRED go PREP club him PRED go inside Dagi went to the club. He went inside. long klab. Bihain baman i askim em. PREP club after barman PRED ask him . . . the barman asked him what he wanted.’ Em i tok olsem: he PRED talk like this [thus]: ‘He said:’ BAMAN: Yes plis. Yu laik dring wanem? yes please you want drink what ‘Yes, please. What do you want to drink?’

TP


DAGI:

Bia. beer ‘Beer.’

BAMAN: Wanem kain bia, SP o Sanmik? what kind beer, SP or San Miguel ‘What sort? — SP or San Miguel?’ DAGI:

Mi laikim SP. I want SP ‘I want a SP.’

BAMAN: Yu laikim long botol o long tin? you want in bottle or in tin ‘Do you want it in a bottle or in a can?’ Once again, as in both HCE and JCE, question word order in TP remains fixed SVO: (10) Yu laik dring wanem? you want drink what ‘What do you want to drink?’ (11) Yu laikim long botol o long tin? you want in bottle or in tin ‘Would you like it in a bottle or in a can?’ Much has been written about the tense-aspect system of TP. For example, Faraclas (1990), in his comparison of Nigerian Pidgin English and Tok Pisin, describes both as having two aspectual poles, completive and incompletive; and modality splits into realis and irrealis (Faraclas 1990: 161). Examples (12)–(13) are from Faraclas (1990: 162 and 163, respectively). (12) Mi kaikai pinis I eat COMPLETIVE ‘I have eaten.’ (13) Mi go London, mi bai baim laplap. I go London, I IRREALIS buy clothing ‘If I go to London, I will buy clothing.’

5.4 Brief observations on the three English-based creoles A few preliminary observations can be made on the basis of the examples given. The rankings of Pienemann’s hierarchy appear to be relevant to the structural description of each language. While Level 3 (and below) structures are evident in all three English-based creoles, Level 4 structures, the beginning of sentence-level (inter-

49

50

Fred Field

phrasal) constructions, are noticeably absent. The evidence points to the likely role of SLA in acquiring and/or appropriating English lexical items and (some) constructions, and suggests that underlying/substrate varieties (mother tongues) have the ability to affect the eventual outcome in ways that significantly diverge from English/superstrate patterns. Also notable is the creativity of the creole originators (and subsequent speakers) to create vibrant community languages out of essentially foreign elements. Despite differences attributable to substrate influence and/or independent development, all three creoles display similar word-order patterns and other significant grammatical and lexical characteristics. As a consequence, the superstrate, English, acting in dynamic relationship with the substrate grammars, is the apparent source of input for some sort of learning/acquisition in all three cases. The lack of inversion strategies leaves bare the prevalence of basic SVO word order in declarative and interrogative sentences, demonstrating the canonical order of English, and the clear absence of any type of sentence-level operations that would alter that order in NL or native-like varieties. The absence of standard English 3S -s via subject-verb agreement, a redundant feature that, as already mentioned, is missing in some British dialects as well, points to the superfluous nature of this -s in communication. The similarities also suggest universal tendencies in acquisition, and in each case, the results are consistent with SLA scenarios in which English was once the target. That is to say, if the superstrate (some variety or amalgamation of varieties of English) was indeed the TL of some kind of learning within the social and linguistic milieu that gave rise to these varieties, then the similarities are nothing more than an expected linguistic outcome. To illustrate the likely result of arrested (slowed to a stop) language acquisition beyond a strictly English perspective, we will now turn to the Spanish-based creole Palenquero.

6. Palenquero Palenquero consists mainly of Spanish lexical items, mostly Spanish function words, and a relatively isolating-analytical verbal morphology, especially when compared with that of European Spanish.²¹ Bantu languages, notably Kikongo, played an important role in its formation, leaving behind (via transfer) numerous lexical and some phonological and grammatical remnants (Schwegler 1999, 2000, 2002, 2003).²² Whether or not Palenquero has pidgin origins is not known. Applying the tendencies of SL learners to the acquisition of Spanish, one would also expect early ILs to be composed of content items and function words — for which few learning restrictions apply — in addition perhaps to some of its inflectional morphology, Spanish-language inflectional categories, and their values or exponents. This is in fact the case with Palenquero, as seen in (14)–(16), where the Tense-Aspect system clearly differs from that of standard varieties of Spanish.²³


(14) I ta yebá planda. yo T/A llevar plátano ‘Yo voy a llevar plátano.’ ‘I’m going to carry (bring along) bananas.’

[Palenquero] [Sp. glosses] [Sp. transl] [English]

(15) ¿I and uté a hayá aggú planda? y dónde usted T/A hallar algún plátano ‘¿Y dónde encontró usted plátano?’ ‘And where did you find bananas?’ (16) I sabé nu. Depué i tan buká pekao Yo saber no. Después yo T/A buscar pescado ‘Yo no sé. Después yo voy a buscar pescado ‘I don’t know. Afterwards, I’m going to look for fish a Katajena pa yebá pa bendé. allá Cartagena para llevar para vender en Cartagena para llevar a venderlo.’ in Cartagena to bring [here] and to sell it.’ In (14)–(16) above, the Spanish lexical items in Palenquero are not marked with any kind of inflectional morphology that can be identified as Spanish (e.g., with suffixes from the major conjugation classes on verbs) (Field 1999). Without the fusional-type morphology, subject pronouns appear to be obligatory (Spanish characteristically omits subject pronouns as a so-called pro-drop language). The absence of Spanish-like verbal inflections suggests the relative lack and perhaps reanalysis of Spanish inflectional categories (person, number, tense agreement categories) and their exponents. The T/A (tense-aspect markers) evident in Palenquero suggests that Tense and Aspect categories are present (and, perhaps, semantically similar), but their forms differ markedly from their counterparts in Spanish or any other Romance language. All verb roots are invariant. For example, in (16), the standard form of the irregular Spanish verb saber ‘to know’, glossed in Palenquero as sabé, would be sé ‘I know’ (illustrated by the Spanish translation beneath). All main verbs are unmarked in (14)–(16); nevertheless, each verb form has the correct theme vowel according to the conjugation class of the Spanish verb and receives stress on the final syllable appropriate to the infinitive form: yebá (from llevar ‘to carry, bring’), hayá (hallar ‘to find’), buká (buscar ‘to look for’), and bendé (vender ‘to sell’). The fusionaltype Spanish affixes expressing number and tense on the auxiliary are absent. This absence can be construed as evidence that the processability of the entire system has been a factor, and that acquisition stopped short of mastery. Only progressive aspect remains on the morpheme/word ta (see example (14), presumably derived from the copula estar ‘to be‘). In its various irregular forms (expressing tense and agreement with the grammatical subject in person and number), estar is used with participial forms to express progressive aspect in standard varieties of Spanish. These aspectual markers appear in the usual tensed auxiliaries

5

52

Fred Field

position (INFL), but Spanish semantically and morphologically fused Person and Number distinctions do not occur in the usual suffixed position on the copula, and the typical participial endings (-ando or -iendo) are also absent in ta yebá (14). Note, also, that the identical form yebá occurs after the preposition para ‘in order to’ — which requires the infinitive form with the appropriate affix of its declension class (i.e., Spanish llevar). From the perspective of the proficient Spanish speaker, the Tense-Aspect system of Spanish has been reduced or simplified, that is, learned imperfectly or incompletely. From the Palenquero/creolist perspective, a new and distinct language has been constructed using mostly Spanish form-meaning sets (words and morphemes) and the knowledge of underlying (substrate) languages. While it is reasonable to say that the processes of SLA have taken place, it is also clear that the speakers of this variety have been able to use fairly limited resources to construct a vibrant and expressive variety of their own, not because they were incapable of learning Spanish, but because they were motivated to develop a full language of their own for communicative, expressive, and psychological purposes (as a marker of ethnic identity). Taking the hierarchy of Pienemann for English and the processing principles that underlie them, I propose the following basic generalizations. Beginning at the bottom of the hierarchy and moving up, Palenquero has clear (canonical) SVO word order, even in question formation, for instance in the following two examples (from Friedemann and Patiño Roselli 1983: 169): (17) ¿Bo é palenkero? 2s COP Palenquero ‘¿Eres palenquero?’ ‘Are you Palenquero?’ (18) ¿I ké bo tan nda fulano? y qué 2s T/A da fulano ‘¿Y qué va a dar Ud. a fulano?’ ‘And what will you give to the guy?’ In (17), the second-person pronoun bo is overtly expressed, showing a fairly rigid SVO order. In (18), the Palenquero pronoun bo precedes both the T/A marker tan and the following main verb nda (< Spanish dar ‘give’).²⁴ The Spanish glosses in (18) show that the second-person pronoun Ud. (usted) must follow the verbal complex (when expressed) according to subject-verb inversion of the standard language. According to Schwegler and Green (forthcoming), SVO word order is routinely used in declarative and interrogative constructions in Palenquero. Contrary to Spanish, subject/verb inversion is not applied in interrogatives (Schwegler and Green forthcoming). In Spanish, inversion applies to questions introduced by question words, leading to the obligatory change from SVO to SOV. In (18), the question word ké (Spanish qué) introduces the clause, evidence of fronting (as WH-fronting in the


English model), however without the resultant changes in word order required in Spanish. In English, all constructions involving changes in word order (sentence-level procedures) are located in the higher rankings according to measures of processability.²⁵ The third-person, singular, present-tense -s is similarly ranked. The absence in Palenquero of all fusional-type verbal affixes typical of standard Spanish is an indication that processability has had an effect. Broader issues of learnability also appear relevant. Concerning Palenquero, the absence of forms and structures that exhibit greater processing demands (in the view of Pienemann 1998 and 2000) is consistent with subsequent-language learning by adults, producing the results one would expect if SLA indeed played a role in its development. Again, nativization will have its effects on spread and stabilization, but not necessarily in the creation of the creole. The overt influence (phonological and lexical) of an underlying native African language (especially Kikongo) reinforces the conclusion that SLA played a principal role in the emergence of Palenquero.

7. Discussion By arguing that SLA processes may be involved in the genesis of a creole language, I am not implying that SLA is the only possible explanation for the emergence of a creole or creole-like language (i.e., in the case of other so-called restructured varieties). Nevertheless, if linguistic factors are the focus, the conclusion that SLA played an important role cannot be dismissed. In fact, it may be unreasonable to posit one and only one type of process in pidgin or creole genesis. Dorian discusses simple theoretical dichotomies within linguistics proper: Dichotomies have the effect of nudging us towards an either-or discrimination. The responsibility for this may lie with the user of the dichotomy, but it is certainly encouraged when the terms of the dichotomy are themselves antonyms, as is the case with internal and external in the phrases “internally motivated change” and “externally motivated change” (1993: 132).

One does not have to be a formal logician to capture the fallacy. Not all problems are resolved by answers that are either A or B; sometimes the answer may be both, and perhaps neither. In this particular case, creole genesis, the terms pidgin and interlanguage simply refer to different possible causes; they are not antonymous. Different sets of processes may compete, overlap or work in concert to produce even more dramatic results, for example: (1) Limits to the learnability/processability of target language constructions by untutored learners in various social settings may lead to different outcomes. Among the possibilities are (a) a jargon or pidgin, (b) a pidgin-like variety (an IL functioning in pidgin-like ways), or merely (c) IL (learners’) varieties of a dominant social group’s native language. Outcomes (a) to (c) may possibly be found in one and the same community.

53

54

Fred Field

(2) Language acquisition involving both native and non-native scenarios (with varying degrees of success according to native-speaker TL norms and NL transfer) can and do co-exist in many communities, creole and otherwise. This should be evident within displaced or enslaved populations where individuals are in the process of rapid and/or forced shift to the foreign language, and have differential access to the lexifier, target, or one-time target. A pidgin or pidgin-like IL may form an intermediate step in the emergence and crystallization of a new variety. (3) Among NL acquirers, the target may be the principal lexifier, IL grammars of their forebears or peers, pidgin or pidgin-like varieties, the creole (different lects), or various combinations. Multiple targets will influence acquisition, leaving it for children and nativization, perhaps, to establish and stabilize norms that can be associated with native-speaker status, social identity, and so on. This also suggests that creolization and nativization are dynamic processes with (a) an interplay of numerous sociolinguistic and psychological factors, and (b) a variety of possible results. Individual outcomes are likely to depend on a range of factors. Community-wide outcomes will depend on the spread, crystallization and, eventually, the stabilization of the emerging variety. Needless to say, when and how this spread, crystallization and stabilization take place depends entirely on social rather than strictly linguistic factors. Thus, stabilization may be initiated, for instance, by the community-wide acceptance of a given speech form as an ethnic marker, as seems to have occurred in Palenque (Schwegler and Morton 2003).

Notes 1. See, for example, Lumsden (1999) for one of many discussions of the role of SLA in creole formation, specifically with respect to Haitian Creole French. 2. Not every specialist uses the terms pidgin and jargon in precisely the same manner. Holm (1988: 5) indicates that jargon can refer to speech simplified by native speakers on an ad hoc basis in an attempt to communicate to those who are not native speakers (“the New Yorker in Lisbon”). The result is a collection of lexical items with more or less immediate (and, therefore, transparent) reference. A pidgin, of course involves the participation of speakers of both/all of the languages in contact. 3. The terms interlanguage and learners’ variety are used here interchangeably. The difference is primarily one of emphasis with respect to process. Learners’ variety mainly refers to the speech of SL learners at a point in time while interlanguage generally emphasizes the acquisition process and a variety in flux. 4. See Siegel (1998, 1999) for discussion of the connections between transfer and substrate. 5. For example, in Sranan, an English-based creole spoken in Suriname, massive borrowing from Dutch was partly the result of its isolation and separation from speakers of English that began in 1667, when the Netherlands assumed control of the former British colony (Holm 1989: 433ff).

Second language acquisition in creole genesis 6. See Matthews and Yip (1995: 17–30) for further illustration of instances of syntactic transfer from Chinese into English. 7. The development of preverbal T/A markers appears to be intermediate between the (2) lexical and (3) morphological stages, after the lexical stage and, perhaps, instead of the morphological stage. The emergence of preverbal T/A markers (going beyond the lexical stage) is also consistent with the idea that new, non-lexifier sentence-level procedures, discussed in the next section, are also developing. As the “new” creole language variety develops and expands to meet a wider range of social and linguistic functions, its grammar also diverges from the superstrate grammar. In this way, originators of a creole are indeed “successful” in creating a viable language. 8. Roberts (2000) discusses reduced access to speakers of the lexifier according to the “event hypothesis” of Baker (1982) and Baker and Corne (1986). Event 0 represents the founding of a particular colony, and Event 1 occurs when the number of slaves or immigrants (speakers of substrate languages) equals the number of colonial masters or members of upper-classes (speakers of the superstrate). As the population of substrate speakers increases proportionately, access to native speakers of the superstrate decreases, “causing increasing numbers of new arrivals to learn the language from other substrate speakers. This enables the emerging contact language to stabilize as a system distinct from the lexifier” (Roberts 2000: 260). 9. Pienemann (1998, 2000) puts his work squarely within the framework of Lexical-Functional Grammar (LFG). My approach here is atheoretical — in the sense that it is not consciously attached to a particular framework. As a result, some terminological compromises are made, only with the intent of presenting issues in general terms. Some use of terminology from the LFG approach is unavoidable, however, when referring to Pienemann’s work. 10. See Pienemann (1998) for a full explanation of the Processability Theory. 11. For a complete description of these structures, see Pienemann (2000: 100). 12. This is not what is traditionally called Do-support. It is merely do in clause-initial position irrespective of subject-auxiliary inversion and subject-verb agreement. 13. Movement of question words to clause initial position (without Subject-Aux inversion) is parallel to the fronting of adverbial elements (Pienemann 2000: 113). 14. One contributing factor to this may have been the absence of this affix (3d singular) in the speech of native speakers from certain parts of England with whom speakers of an emerging creole may have had contact. Trudgill states: “In the East Anglia area . . . present-tense verbs are entirely regular and have no -s endings at all . . . This was formerly true also of the Traditional Dialects of the Essex area. . .” (1990: 94). 15. For example, typical creole preverbal T/A markers can account for constructions that might require Level 4 and higher amounts of syntactic information. Examples are found in all four creole languages discussed below. 16. The overapplication of subject-auxiliary inversion, with concomitant lack of cancellation (blocking of the rule or process), also appears to be characteristic of such ethnic varieties as Chicano English (Field forthcoming), a fact that fits nicely in the discussion of ILs and a number of non-standard varieties of English, not just creoles. 17. While it is admittedly somewhat anecdotal, it is noticeable to many teachers of English that those students who have studied English formally in a classroom (e.g., in their home country) are more likely to at least know what the 3S -s is. Many who “learn it in the streets”

55

56

Fred Field through personal interaction with fluent speakers who may not speak a standard variety can become quite fluent, yet show an amazing omission of English inflectional morphology. This includes the 3S -s, regular past tense -ed, and participial forms used as adjectives or to express perfect aspect and passive voice. This is characteristic also of non-native speakers, but also native speakers of non-standard Chicano English and a variety referred to as Vietnamese English (Wolfram and Schilling Estes 1998: 114). 18. See Roberts (1999, 2000) for discussion of the origins of HCE and its early history. 19. The following quotations from the book of Mark are from an experimental draft circulated by Wycliffe Bible translators of Mahk Story (the Book of Mark). 20. The following abbreviations are used throughout the English-based creole samples: COP = copula; PROG = progressive; PST = past; ACC = accusative; PRED = predicate marker (Mühlhäusler 1990); PLUR =plural. 21. The originators were maroons who around 1700 established a community less than 100 kilometers from the former slaving port of Cartagena, Colombia. There is conclusive documentary evidence that they have been proficient creole/Spanish bilinguals for well over 200 years (Schwegler and Morton 2003). Whether or not Palenquero developed ab ovo in Palenque or is a last remnant of a once more widely spoken creole is still a matter of debate (Schwegler and Green in press, Schwegler 2002). 22. See, for example, Schwegler (1991, 1998, 1999, 2000) for comprehensive discussion of research on Palenquero and a description of the language, its speakers, and the historical circumstances under which it emerged. 23. All examples are from Schwegler (1988). The English translations are mine. 24. Note the prenasalization of the word-initial consonant in the verb nda, cited and discussed in Lipski and Schwegler (1993: 416) and Schwegler (1996: 236). 25. Pienemann (1998) also discusses and ranks similar constructions in German (116); Swedish (182), and Japanese (211).

References Andersen, R. 1980. “Creolization as the acquisition of a second language as a first language.” In Valdman and Highfield, 273–95. Andersen, R. (ed.). 1983. Pidginization and Creolization as Language Acquisition. Rowley, MA: Newbury House. Andersen, R. 1984. “The One to One Principle of interlanguage construction.” Language Learning 34: 77–95. Andersen, R. 1989. “The ‘up’ and ‘down’ staircase in secondary language development.” In Investigating Obsolescence: Studies in Language Contraction and Death, N. C. Dorian (ed.), 385–94. Cambridge: Cambridge University Press. Anderson, S. C. and B. Comrie (eds.). 1991. Tense and Aspect in Eight Languages of Cameroon. Arlington: SIL. Arends, J., and A. Bruyn. 1995. “Gradualist and developmental hypotheses.” In Arends, Muysken and Smith, 111–20. Arends, J., P. Muysken and N. Smith (eds.). 1995. Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins.

Second language acquisition in creole genesis Baker, P. 1982. The contribution of non-Francophone immigrants to the lexicon of Mauritian Creole. PhD Thesis. London: University of London. Baker, P. and C. Corne. 1986.“Universals, substrata, and the Indian Ocean creoles.” In Muysken and Smith, 163–83. Bakker, P. 1995. “Pidgins.” In Arends, Muysken and Smith, 25–39. Bardovi-Harlig, K. 2000. Tense and Aspect in Second Language Acquisition: Form, Meaning, and Use. Malden, MA: Blackwell Publishers. Biber, D. 1988. Variation Across Speech and Writing. Cambridge: Cambridge University Press. Bickerton, D. 1984. “The language bioprogram hypothesis and second language acquisition.” In Language Universals and Second Language Acquisition, W. Rutherford (ed.), 141–61. Amsterdam: John Benjamins. Bickerton, D. 1988. “Creole languages and the bioprogram.” In Linguistics: The Cambridge Survey (vol. 2), F. J. Newmeyer (ed.), 268–84. Cambridge: Cambridge University Press. Bloomfield, L. 1933. Language. New York: Henry Holt. Brown, R. 1973. A First Language: The Early Stages. Cambridge, Mass.: Harvard University Press. Carr, E. B. 1972. Da Kine Talk: From Pidgin to Standard English in Hawa’ii. Honolulu: University of Hawai’i Press. Cassidy, F. G. 1971. “Tracing the pidgin element in Jamaican Creole.” In Pidginization and Creolization of Languages, D. Hymes (ed.), 203–23. Cambridge: Cambridge University Press. Cassidy, F. G. 1980. “The place of Gullah.” American Speech 55: 3–15. Cook, V. 1993. Linguistics and Second Language Acquisition. New York, NY: St. Martin’s Press. Corder, P. 1993. “A role for the mother tongue.” In Language Transfer in Language Learning, S. M. Gass and L. Selinker (eds.), 18–31. Amsterdam: John Benjamins. DeGraff. M. 1999.“Creolization, language change, and language acquisition: A prolegomenon.” In Language Creation and Language Change: Creolization, Diachrony, and Development, M. DeGraff (ed.), 1–46. Cambridge, Mass.: MIT Press. Dietrich R. and C. Perdue. 1995. “Introduction.” In Dietrich, Klein and Noyau, 1–16. Dietrich, R., Klein W. and C. Noyau (eds.). 1995. The Acquisition of Temporality in a Second Language. Amsterdam: John Benjamins. Dorian, N. C. (ed.). 1989. Investigating Obsolescence: Studies in Language Contraction and Death. Cambridge: Cambridge University Press. Dorian, N. C. 1993. “Internally and externally motivated change in language contact settings: Doubts about dichotomy.” In Historical Linguistics: Problems and Perspectives, C. Jones (ed.), 131–55. London: Longman. Dörnyei, Z. and R. Schmidt (eds.) 2001. Motivation and Second Language Acquisition. Manoa: Second Language Teaching and Curriculum Center. University of Hawai’i. Dulay, H. and M. Burt. 1973. “Natural sequences in child second language acquisition.” Language Learning 24: 234–52. Dutton, T. (in collaboration with D. Thomas). 1985. A New Course in Tok Pisin (New Guinea Pidgin). Canberra: Department of Linguistics, Research School of Pacific Studies. The Australian National University. Ericcson, K. A., R. T. Krampe& C. Tesch-Römer. 1993. “The role of deliberate practice in the acquisition of expert performance.” Psychological Review 100: 363–406.

57

58

Fred Field Faraclas, N. 1990. “From Old Guinea to Papua New Guinea: A comparative study of Nigerian Pidgin and Tok Pisin.” In Verhaar , 91–170. Field, F. 1999. “Language mixing: The effects of a matrix system.” Southwest Journal of Linguistics 18: 47–78. Friedemann, N. and C. Patiño Roselli. 1983. Lengua y sociedad en El Palenque de San Basilio. Bogota: Instituto Caro y Cuervo. Gardner, R. C. 2001. “Integrative motivation and second language acquisition.” In Dörnyei and Schmidt, 1–19. Holm, J. 1988. Pidgins and Creoles. Volume I: Theory and Structure. Cambridge: Cambridge University Press. Holm, J. 1989. Pidgins and Creoles. Volume II: Reference Survey. Cambridge: Cambridge University Press. Holm, J. and P. Patrick (eds.). Forthcoming. Comparative Creole Syntax. London: Battlebridge Publications. Klein, W. 1995. “The acquisition of English.” In Dietrich, Klein and Noyau, 31–70. Larsen-Freeman, D. and M. Long. 1991. An Introduction to Second Language Acquisition Research. London and New York: Longman. Le Page, R. B. and D. DeCamp. 1960. Jamaican Creole: Creole Studies I. London: Macmillan. Lefebvre, C. 1986. “Relexification in creole genesis revisited: The case of Haitian Creole.” In Muysken and Smith,279–300. Lightbown, P. M. and N. Spada. 1999. How Languages Are Learned (2d ed.). Oxford: Oxford University Press. Lipski, J. 1994. Latin American Spanish. New York: Longman. Lipski, J. and A. Schwegler. 1993. “Creole Spanish and Afro-Hispanic.” In Bilingualism and Linguistic Conflict in Romance, J. N. Green and R. Posner (eds.), 407–32. Berlin: Mouton de Gruyter. Litteral, R. 1990. “Tok Pisin: The language of modernization.” In Verhaar, 375–86. Lumsden, J. 1999. “Language acquisition and creolization” In DeGraff, 1999: 129–58. Marcus, G, S. Pinker, M. Ullman, M. Hollander, T. Rosen and F. Xu. 1992. Overregularization in Language Acquisition. Chicago, Ill.: University of Chicago Press. Matthews, S. and V. Yip. 1995. “I-interlanguage and typology: The case of topic-prominence.” In The Current State of Interlanguage: Studies in Honor of William E. Rutherford, L. Eubank, L. Selinker and M. S. Smith (eds.), 17–30. Amsterdam: John Benjamins. Mufwene, S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mühlhäusler, P. 1990. “Tok Pisin: Model or special case.” In Verhaar, 171–86. Muysken, P. and N. Smith (eds.). 1986. Substrata versus Universals in Creole Genesis. Amsterdam: John Benjamins. Muysken, P. and N. Smith. 1995. “The study of pidgin and creole languages.” In Arends, Muysken and Smith, 1–14. Nation, I. S. P. 1990. Teaching and Learning Vocabulary. Boston, MA: Heinle and Heinle Publishers. Patrick, P. Forthcoming. “Jamaican Patwa (Creole English).” In Holm and Patrick . Pienemann, M. 1998. Language Processing and Second Language Development: Processability Theory. Amsterdam: John Benjamins. Pienemann, M. 2000. “Psycholinguistic mechanisms in the development of English as a second language.” In Language Use, Language Acquisition, and Language History: (Mostly)

Second language acquisition in creole genesis Empirical Studies in Honour of Rüdiger Zimmerman, I. Plag and K. P. Schneider (eds.), 99–118. Trier: Wissenshaftlicher Verlag Trier. Reesink, G. 1990. “Mother tongue and Tok Pisin.” In Verhaar, 289–306. Rickford, J. R. and S. Romaine. 1999. Creole Genesis, Attitudes and Discourse: Studies Celebrating Charlene J. Sato. Amsterdam: John Benjamins. Roberts, S. J. 1999. “The TMA system of Hawaiian Creole and diffusion.” In Rickford and Romaine, 45–70. Roberts, S. J. 2000. “Nativization and the genesis of Hawaiian Creole.” In Language Change and Language Contact in Pidgins and Creoles, J. McWhorter (ed.), 257–300. Amsterdam: John Benjamins. Romaine, S. 1988. Pidgin and Creole Languages. New York: Longman. Schumann, J. 1978a. The Pidginization Process: A Model for Second Language Acquisition. Rowley, MA: Newbury House. Schumann, J. 1978b. “The relationship of pidginization, creolization and decreolization to second language acquisition.” Language Learning 28: 367–79. Schumann, J. 1978c. “The acculturation model of second language acquisition.” In Second Language Acquisition and Foreign Language Teaching, R. C. Gingras (ed.), 27–50. Washington, D. C.: Center for Applied Linguistics. Schwegler, A. 1988. Palenquero interviews (sample transcriptions from the tape “Dos mujeres”). Mss. Schwegler, A. 1991. “Negation in Palenquero: Synchrony.” Journal of Pidgin and Creole Languages 6: 165–214. Schwegler, A. 1996. “Chi ma nkongo”: lengua y rito ancestrales en El Palenque de San Basilio (Colombia). 2 vols. Frankfurt/Madrid: Vervuert Verlag. Schwegler, A. 1998. “Palenquero.” In América negra: panorámica actual de los estudios lingüisticos sobre variedades criollas y afrohispanas, M. Perl and A. Schwegler (eds.), 220–91. Frankfurt/Madrid: Vervuert Verlag. Schwegler, A. 1999. “El vocabulario africano de Palenque (Colombia). Segunda parte: compendio de palabras (con etimologías).” In El Caribe hispánico: Perspectivas lingüísticas actuales (Homenaje a Manuel Álvarez Nazario), Luis Ortiz (ed.), 171–253. Frankfurt/ Madrid: Vervuert Verlag. [Revised version in Palenque, Cartagena y Afro-Caribe: historia y lingüística [2002], Yves Moñino and Armin Schwegler (eds.), 171–227. Tübingen: Max Niemeyer Verlag] Schwegler, A. 2000. “The African vocabulary of Palenque (Colombia). Part 1: Introduction and corpus of previously undocumented Afro-Palenquerisms.” Journal of Pidgin and Creole Languages 15: 241–312. Schwegler, A. 2002. “On the (African) origins of Palenquero subject pronouns.” Diachronica 19: 273–332. Schwegler, A. 2003. “Einige Antworten der Kreolistik an die Hispanistik.” In Portugiesisch in der Diaspora. Vorträge zum 4. Deutschen Lusitanistentag an der Universität Mainz (2001, Dieter Messner and Matthias Perl (eds.), 22–42. Germersheim: CELA (Centro de Estudios Latinoamericanos). Schwegler, A. and K. Green. Forthcoming. “Palenquero (Creole Spanish).” In Holm and Patrick. Schwegler, A. and T. Morton. 2003. “Vernacular Spanish in a microcosm: Kateyano in El Palenque de San Basilio (Colombia).” Revista Internacional de Lingüistica Iberoamericana (RILI) 1: 97–159.

59

60

Fred Field Selinker, L. 1972. “Interlanguage.” International Review of Applied Linguistics 10: 209–31. Seuren, P. and H. Wekker. 1986. “Semantic transparency as a factor in creole genesis.” In Muysken and Smith, 57–70. Siegel, J. 1998. “Substrate reinforcement and dialectal differences in Melanesian Pidgin.” Journal of Sociolinguistics 2/3: 347–73. Siegel, J. 1999. “Transfer constraints and substrate in Melanesian Pidgin English.” Journal of Pidgin and Creole Languages 14: 1–44. Singler, J. 1993. “African influence upon Afro-American language varieties: A consideration of sociohistorical facts.” In Africanisms in Afro-American Language Varieties, S. Mufwene (ed.), 235–53. Athens, Ga.: University of Georgia Press. Singler, J. 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. Thomason, S. G. 2001. Language Contact: An Introduction. Washington, D. C.: Georgetown University Press. Thomason, S. G. and T. Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Trudgill, P. 1990. The Dialects of England. Oxford: Blackwell Publishers. Valdman, A. 1980. “Creolization and second language acquisition.” In Valdman and Highfield, 297–311. Valdman, A. and A. Highfield (eds.). 1980. Theoretical Orientations in Creole Studies. New York: Academic Press. Verhaar, J. (ed.). 1990. Melanesian Pidgin and Tok Pisin: Proceedings of the First International Conference of Pidgins and Creoles in Melanesia. Amsterdam: John Benjamins. Weinreich, U. 1953 (reprinted in 1968). Languages in Contact. The Hague: Mouton. Wolfram, W. and N. Schilling-Estes. 1998. American English: Dialects and Variation. Malden, Mass: Blackwell Publishers. Wurm, S. A. 1980. “Standardisation and instrumentation in Tok Pisin.” In Valdman and Highfield, 237–44.

CHAPTER 7

OT and the acquisition of Jamaican syllable structure Rocky R. Meade

University of the West Indies, Mona

This article discusses the acquisition of Jamaican syllable structure and shows that independently-motivated Optimality Theory (OT) constraints can account for the patterning of early child structures as well as their development into adult structures. Twenty-four children representing two socio-economic groups were recorded longitudinally over two years, starting at the ages of 1;0 and 2;6. Age of acquisition varies, sometimes widely, whereas the ordering of syllable structure acquisition is significantly similar across the children’s groups. I conclude that a particular ranking of the OT constraints accounts for early structural patterns, as well as for their development into adult syllable structures. These constraints are not changed, but simply re-ranked in the course of the developmental process.

1. Introduction¹ This article examines the acquisition of syllable structure by Jamaican children and shows that Optimality Theory (OT), when accompanied by independently-motivated constraints, can account for the partial early inventories as well as for their development towards the complete range of adult syllable types. The paper presents some of the findings of an investigation into the phonological development of Jamaican children acquiring their first language. This article thus contributes to filling the current gap in studies on the acquisition of creole languages. As Section 1.1 highlights, the Jamaican speech community is highly diverse, ranging from speakers of the basic creole (the basilect) to speakers of Standard Jamaican English (the acrolect). Section 1.2 outlines relevant results from previous research on acquisition by Grunwell (1982) and Fikkert (1994). Grunwell’s findings on syllable structure development provide the primary basis for a comparison of the Jamaican data with those of other young English learners. Although Fikkert reports on the acquisition of Dutch, she discusses an interesting aspect of syllable onset development that is not covered by Grunwell, but is applicable in this Jamaican study.

62

Rocky R. Meade

In order for OT to be taken seriously as a viable theory of grammar, it must not only account for adult grammars but also for various aspects of acquisition. As is outlined in Section 2, OT captures the universal features of languages by recognizing all constraints (as opposed to rules) as universal, and is able to account for variation by a simple process of variable ranking of the constraints. It is this variability of constraint rankings that provides the key to understanding aspects of acquisition (syllable development). A brief outline of the methodology of project design, data collection, transcription and analysis is presented in Section 3. This section also describes how two fairly distinct groups of children were selected as informants in an effort to capture as wide a range of the Jamaican continuum as possible. Twenty-four children from two socio-economic groups were recorded longitudinally over two years, starting from the ages of 1;0 and 2;6. Data collected from twenty of the children are used here. The results of the data collection and data analysis are presented in Section 4. They show significant similarities in the sequence of syllable structure development, not only between the two Jamaican groups, but also in comparison to the findings of Grunwell and Fikkert, notwithstanding age differences. The theoretical analysis is covered in Section 5, in which I demonstrate that four clearly universal wellformedness constraints are applied concurrently at each stage. The combinations of these constraints are instrumental in the developmental process from the child’s first core syllable to the more complete adult inventory.

1.1 The Jamaican language situation The Jamaican speech community consists of a minimum of two main language varieties, namely Jamaican English (JE) and Jamaican Creole (JC). JE is a dialect of English that is mutually intelligible with other dialects of English, in spite of certain differences. In the creole continuum model, JE is referred to as the acrolect. JC includes the basilect — the variety most distinct from JE — and the mesolect. The mesolect encompasses a set of intermediate varieties. There are, however, mesolectal features that belong neither to the acrolect nor the basilect, for example, the anterior marker did, which is akin to the past tense. This leads to the recognition of the mesolect as a language variety distinct not only from the acrolect (JE) but also Table 1. Variation in the production of the utterance she thought this was a stick Serial

Variety

Utterance

(a) (b) (c) (d) (e)

Basilect Mesolect

[ɪm [ ʃɪ [ ʃɪ [ ʃɪ [ ʃɪ

Acrolect

(b)ɛn tɪŋk dɪd tɪŋk dɪd tɪŋk tɔ:t θɔ:t

dɪs dɪs dɪs dɪs ðɪs

a a wʌz wʌz wʌz

wan wan a a a

tɪk] tɪk]: stɪk]: stɪk]: stɪk]:

OT and the acquisition of Jamaican syllable structure

from the basilect (Devonish 1998: 10, and Patrick 1999: 293). The mesolect, which has features of both the acrolect and the basilect, thus fits into the continuum model quite well. Table 1 gives examples of some of the ways in which a single statement could be uttered in the Jamaican speech community. Basilectal and acrolectal utterances are illustrated in (a) and (e), respectively, with the variable [bεn] or [εn] tense marker of the basilect shown in (a). Intermediate utterances for the same statement are given in (b) to (d).

1.1.1 The syllable template

The template (C20 V21 C20) with a maximum of two segments for each syllable node is the model of the JC syllable adopted for this paper. The acrolect (JE) allows clusters of three consonants in both the onset and coda. However, the majority of the children acquired only two-consonant clusters during the period of data collection. The OT analysis (Section 5) accounts for this pattern, and is still able to allow for the putative occurrence of three-consonant clusters.

1.2 Previous research In order for the Jamaican acquisition data to be situated in a broader context, they are compared with those of Grunwell (1982) and Fikkert (1994). As indicated in a number of previous studies, Grunwell’s syllable structure development sequence for children acquiring English represents a general consensus. The order of syllable structure development presented in (1) is adapted from Grunwell (1982: 99). This list displays the new syllable canonical forms that the children normally produce at each stage, but excludes the emerging structures (included by Grunwell). (1) Syllable structure development (Grunwell) Age Canonical syllable forms 0;9–2;0 CV, CVCV 2;0–3;0 CVC 3;0–4;6 CVCCV(C), CVCVCV, (C)CVC(C)

Grunwell’s stages Stages I–II Stages III–IV Stages V–VII

Fikkert’s (1994) study covers the development of Dutch syllable structure. In her model, she makes a distinction between the stage of acquisition of CV syllables with compulsory (plosive) onsets and the stage of (C)V with onsetless syllables. List (2) displays stages of onset development by Dutch children (Fikkert 1994: 65). (2) Syllable structure development (Fikkert) Age² Canonical syllable form 1;2–1;6: CV (with plosive onsets) 1;6–1;9: (C)V (possibility of onsetless syllable) 1;9–on (C)V (with other types of onsets)

Fikkert’s stages Stage I Stage II Stage III

The profile of syllable structure development represented in the Jamaican children is discussed initially in terms of various syllable structure phenomena. The patterns

63

64

Rocky R. Meade

observed are then grouped into stages, and compared with the stages noted by Grunwell for English children. I also investigate evidence of more detailed stages than Grunwell’s, such as those noted by Fikkert for the onset of Dutch children.

2. Theoretical framework Optimality Theory (OT) was originally proposed to account for adult phonological systems in general, and for variation between languages in particular. However, it became apparent that this framework could provide a unified account of many phonological changes observed in child language development as well, using essentially the same tools that are employed to account for variation between languages (Tesar and Smolensky 1994, 1998). OT — the most significant recent theory of grammar (Prince and Smolensky 1993, McCarthy and Prince 1993) — is a theory of generative grammar, in that it holds that a grammar is a finite system that can produce an infinite number of all and only the correct utterances of the language. The main difference between OT and previous generative approaches is that, in OT, grammars are deemed to work solely through a mechanism of constraint interaction, where the constraints can be violated. In mainstream OT, all constraints are deemed to be universal, and it takes language-particular variation in constraint hierarchies to account for the differences between languages. In the OT variant adopted here, universal does not mean that a particular constraint is necessarily influential in every grammar, but that it exists somewhere in every grammar.³ A constraint may be highly ranked and therefore be very influential in one grammar, but be lowly ranked in another and have virtually no effect on that language. OT has been most influential in phonology, but it is increasingly impacting all modules of grammar, from phonetics to semantics (Archangeli and Langendoen 1997, Dirks 2000).⁴

2.1 Acquisition in OT OT accounts for the acquisition process in terms of the continuous reranking of constraints until the child’s phonological system matches the adult system. There are various proposals as to the detailed workings of the model. Tesar and Smolensky (1994, 1998) share the more widely accepted OT view that the constraints are innate. It is claimed that children start out with maximally unmarked or universally well-formed segmental and syllabic structures, resulting in the radical differences between the child and adult forms in these respects. OT accounts for this generalization by assuming that all the well-formedness constraints are ordered more highly than all the faithfulness constraints. This means that the child’s phonological output is phonologically well-formed (unmarked) at the cost of maintaining the phonemic distinctions and syllable structure found in adult language. As children progress in their phonological development a gradual re-ordering of well-formed-


ness and faithfulness constraints takes place, with some being demoted in the constraint hierarchy until the child’s output matches the adult output. Boersma (1998) claims that all constraints must first be learned, based on functional necessity, before they can be ranked and reranked. Bolognesi (1998) suggests that structural well-formedness constraints and faithfulness constraints are not learned per se, but result from more general principles.⁵ He, however, also holds the view that there are other constraints outside these two categories that are necessarily learned. This paper does not contribute to this debate, since the data presented here are accounted for by well-formedness and faithfulness constraints, the nature of which is subject to wide consensus in the literature. OT is still a relatively new theory and is not without shortcomings. These include the lack of a proper theory of the Generator (GEN) (Bolognesi 1998), and the continued absence of fully developed theories of segmental and syllabic structure. Also lacking are clear explorations of fundamental assumptions pertaining to the nature of constraints, to the presumed universality of all constraints, and to the precise meaning of universality and innateness. Similar objections have been raised by numerous critics of OT, including Chomsky (1995), but also by linguists working within the OT framework (Boersma 1998, Bolognesi 1998, Smith 2000, inter alia). There have also been extensive discussions on the internet-based Optimality List on some of these issues. However, consideration of possible solutions to most of these issues is beyond the scope of this paper.

3. Methodology With the primary aim of providing a descriptive analysis of the development of the phonology of Jamaican children, research questions were formulated that not only asked about this process, but also addressed the issue of variation in the Jamaican speech community. A research design that incorporates the ability to address developmental and comparative issues was therefore chosen. The project was designed to incorporate longitudinal and cross-sectional data. This was achieved by recording the same children continuously for two years, and by having twenty of the children subdivided into four groups to allow for cross-sectional comparisons. The main controlled variables in the selection of informants were the child’s age and the level of education and area of residence of the primary caregiver. In order to cover the age period 1;0 to 4;6 in the limited time available, two age groups were selected. One subgroup in each community included children who were approximately 1;0 year, and in the other subgroup, children who were approximately 2;6 years at the beginning of the data collection period. Children were selected if they were within three months of these ideal ages. In an effort to have representation from both extremes of the language continuum, children were taken from different areas and from families with different educational backgrounds. Language use in Jamaica is often predictable on the basis

65

66

Rocky R. Meade

of education and area of residence. English is the official language and language of education. Consequently, a longer educational span often results in better command of English and sometimes a concomitant reduction in use of JC. Speakers in smaller, relatively isolated rural communities are likely to be competent in a variety of JC that is closer to the basilect than are speakers in larger urban centers (Beckford-Wassink 1999). The education level of the primary caregiver was the major factor used to determine the two groups that were studied. One group included children whose caregivers have no more than primary level education, referred to as the pre-secondary (Pre-Sec) group. The other group included children whose caregivers had more than secondary level education (college or university) referred to as the post-secondary (it was anticipated that the Post-Sec group would include speakers of JE and the Pre-Sec group would include speakers of JC). This should not suggest that the presence of only monolinguals was anticipated in each group. Rather, it was expected that the Post-Sec group may include speakers more competent in JE than those of the Pre-Sec group, and that the JC of the Pre-Sec speakers would be closer to the basilect than the JC of the Post-Sec group. This language distribution assumption was also based on the fact that the Pre-Sec group was selected from a rural community and the Post-Sec group from an urban area. In an effort to reduce the effects of regional variation, two communities that met the required locale diversity requirement — urban vs. rural and relatively isolated — were selected from the same region, the parish of Saint Catherine, which borders the capital metropolis of Kingston and Saint Andrew⁶ (see Map 1). The Pre-Sec group was drawn from the relatively small, rural community of Guys Hill, located in the hills of Saint Catherine (Map 2). The town of Linstead, also in Saint Catherine, was selected as the source of Post-Sec informants. Linstead is the largest town in Saint Catherine that is relatively far removed from Saint Andrew. The two significantly larger communities in Saint Catherine, Spanish Town and Portmore, are closer to and are, in effect, suburbs of Saint Andrew. Portmore in particular has functioned as a residential community for persons who work and go to school in the capital metropolis. In addition, Linstead is only ten miles from Guys

Map 1. Map of Jamaica with parish boundaries


Map 2. Map of Saint Catherine and surrounding parishes

Hill, and serves as the main administrative and commercial center for Guys Hill residents. Therefore, to the extent that there is regional language variation, Linstead should show region-influenced varieties that are closer to Guys Hill varieties than to those of Saint Andrew. The language used by the selected children’s caregivers was essentially as anticipated, with the Pre-Sec caregivers functioning in a language range closer to the basilectal end of the continuum, and the Post-Sec group in a range closer to the acrolectal end. As far as possible, an effort was made to include equal numbers of boys and girls in each subgroup. The profiles of the Pre-Sec and Post-Sec children and primary caregivers are presented in Tables 2 and 3, respectively. Two children from each of the subgroups — a total of eight — were recorded once per month for two years from age 1;0 to 3;0, and from age 2;6 to 4;6. The remaining Table 2. Profile of Guys Hill (Pre-Sec) informants Child informant

Primary caregiver

Code Name

DOB

Age at 1st rec.

Relation to child Education Occupation

G11 G12 G13 G14 G15 G21 G22

Raymond Shakira Russell Tonya Troy Alison Bryce

01 Jul 95 30 Aug 95 18 Jun 95 17 Jun 95 15 Aug 95 05 Nov 93 01 Dec 93

1;0 1;0 1;0 1;2 1;0 2;9 2;6

Mother Mother Father Mother Mother Mother Mother

Primary Primary Primary Primary Primary Primary Primary

G23 G24 G25

Damion 07 May 94 2;5 Oshari 01 Dec 93 2;7 Winsome 24 Mar 94 2;6

Mother Mother Mother

Primary Primary Primary

Domestic Unemployed Construction worker Unemployed Vendor Unemployed Teacher’s aid (preschool) Seamstress Vendor Seamstress

Key to codes (XYZ): X – Area (G = Guys Hill); Y – Group no. (Group 1 = age 1;0–3;0, Group 2 = age 2;6–4;6); Z – Child no. (within the subgroup)

67

68

Rocky R. Meade Table 3. Profile of Linstead (Post-Sec) informants Child informant

Primary caregiver

Code Name

DoB

Age at 1st rec.

Relation to child Education

L11 L12 L13 L14 L15 L21 L22 L23 L24 L25

04 Dec 95 08 Aug 95 16 Apr 95 23 Oct 95 08 Mar 95 26 Mar 94 10 Oct 94 03 Oct 93 15 Mar 94 29 Sep 94

1;0 1;0 1;3 1;0 1;3 2;6 2;6 3;0 2;7 2;6

Mother Mother Mother Mother Mother Mother Mother Mother Mother Mother

Karl Teresa Davin Okhanna Sharon Kereen Nathan Karel Kimone Shelly

Tertiary Post-secondary Tertiary Post-secondary Post-secondary Tertiary Tertiary Tertiary Tertiary Post-secondary

Occupation Teacher Statistician Teacher Teacher Housewife Teacher Teacher Bank supervisor Teacher Accountant

Key to codes (XYZ): X – Area (L = Linstead); Y – Group no. (Group 1 = age 1;0–3;0, Group 2 = age 2;6–4;6); Z – Child no. (within the subgroup)

twelve children were recorded at quarterly intervals for the two-year period. Audio and video recordings of mostly spontaneous utterances were made for 30 to 40 minutes at each data point. Naturalistic rather than controlled data were targeted since the former are likely to provide a closer representation of typical language use and language development. When necessary, picture books were the main elicitation items; these were at times supplemented by various toys and games. Caregivers were also encouraged to use items from the home to elicit utterances.

4. Results The children’s syllable structure development profile is examined in terms of various syllable structure types, such as V, CV, CVC, and (C)CVC(C). Those general categories are grouped into stages that are then compared with the stages noted by Grunwell (1982) for British children acquiring English. This analysis shows evidence of more detailed stages than Grunwell’s, in keeping with those noted by Fikkert (1994) for the onset of Dutch children. One child is discussed in detail, then the group results are presented and compared. The basic requirement for the full acquisition of a syllable type is that there be 75% correct production of the targeted form. Attempts made to produce each of the different syllable structure types are counted as correct production, even if the segments are produced incorrectly. A group is assumed to have acquired a given syllable type if the majority of the children in the group correctly produce at least 75% of the targeted structure.


4.1 Raymond’s development of syllable-structure The data collected from one informant, Raymond, will now be discussed. The data for the remaining children will be considered only as they relate to the particular group to which they belong. Raymond (G11)⁷ is Child #1 in the one-year-old subgroup from Guys Hill (PreSec). He is an only child, but lives in a household that includes his mother, stepfather, aunt and her three children, one of whom also participated in the project. His mother is the primary caregiver. She had primary education only, and was unemployed for the first year of the recordings, except for occasional domestic jobs. Raymond’s recordings were scheduled at monthly intervals for the two years of data collection. Transcription of his data was scheduled to take place at monthly intervals for the first year (age 1;0 to 2;0) and quarterly for the second year of recordings (age 2;0 to 3;0).⁸ The great majority of Raymond’s target forms are basilectal or mesolectal creole forms, based on language use not only by the primary caregiver, but also by all members of the household. The core syllable (CV) is the first to emerge for Raymond, based on the transcribed data at age 1;0, and was fully acquired at 1;2. This is in keeping with Grunwell’s (1982: 99) observations for children acquiring English, as well as with Fikkert’s (1994: 57) findings for Dutch children. As can be seen from the examples in (3), the boy produced CV syllables in different target syllable combinations. (3) Examples of Raymond’s CV production⁹ Gloss Target form (basilect) Child’s form a. there /dε/ [dε] banana /bana:na/ [ba] b. see /sɪ/ [tɪ] mummy /mʌmɪ/ [bʌmɪ] c. that /dat/ [da] book /bʊk/ [bʊ] d. plane /plɪεn/ [pε]

Age (1;0) (1;3) (1;1) (1;3) (1;0) (1;2) (1;7)

In (3a), CV syllables match the CV target.¹⁰ In (3b), there are cases of segmental changes, for example /s/ to [t] and /m/ to [b], but the CV syllable structure is maintained. Target syllables with codas (CVC) were produced by Raymond as CV syllables (3c). Finally, in (3d) a syllable with initial cluster is a CV sequence. In Fikkert’s stage I of onset development that includes children from age 1;4 to almost 1;11, only CV syllables with plosive onsets are produced. The same applies to Raymond for the first two months of recordings up to age 1;3. Beyond that point, he acquired nasal onsets in addition to plosives. In contrast with Fikkert’s observations Raymond also produced onsetless syllables as early as the first recording, as shown in the examples in (4) below, although these are much less frequent than the CV syllables. Fikkert (1994: 57) notes that the children sometimes used what she calls a “repair strategy” to “add an onset” to onsetless syllables while in her first stage.

69

70

Rocky R. Meade

There was no evidence of such a strategy in the examples considered here. Note that Raymond apparently acquires onsetless syllables in disyllabic forms in contrast with CV syllables. (4) Raymond’s onsetless syllables Gloss Target form (basilect) a. Harry /arɪ/ b. kitty /kɪtɪ/

Child’s form¹¹ Age [a.ɪ] (1;0) [ɪtɪ] (1;3)

The next syllable structure type acquired by Raymond is that with a coda or final consonant. Evidence of coda acquisition is first seen at the age of 1;7; this acquisition is thus earlier than that noted by Grunwell, who shows closed syllables as acquired at stage III (age 2;0–2;6). Raymond’s first cluster is the syllable final /nt/ cluster at the age of 1;10, as displayed in (5) below with other types of clusters. The issue of whether a distinction needs to be made between the syllable positions of the acquired clusters is discussed in Section 4.2. (5) Raymond’s clusters Gloss Target form (basilect) a. ‘want’ /wa:nt/ b. ‘paint’ /pɪεnt/ c. ‘cry’ /kraɪ/ d. ‘brush’ /brʌʃ/ e. ‘turtle’ /tɜrkl/

Child’s form [wa:nt] [pεnt] [kraɪ] [brʌʃ ]/[bwʌʃ ] [tɜrklʊ]

Age (1;10) (2;3) (2;6) (2;6) (3;0)

The liquid /l/ in the word “turtle” has a nuclear function, that is, the liquid functions as the nucleus of the syllable, a role more typically reserved for vowels. Raymond’s attempt at producing this syllabic consonant constitutes evidence of his ability to produce a consonant cluster — /kl/ in this case — but also of his inability to produce the syllabic consonant. This interpretation is supported by the insertion of a vowel in the vowel-less syllable. Such observations suggest that consonant production in syllabic nuclei is posterior to the acquisition of regular consonantal clusters. It is now possible to see to what extent the acquisition stages represented in Raymond’s syllable structure development corresponds with those outlined by Grunwell. His first syllables are open, primarily CV syllables, but he also produces onsetless open (vowel only) syllables. This pattern emerged at or before 1;0, based on his utterances at the first recording, but it was not acquired until 1;2. He also produced disyllabic utterances at 1;2. This stage of development is similar to Fikkert’s Stages I and II, but only CV syllables with plosives are included in her Stage I. Grunwell does not directly mention onsetless syllables, so no comparison can be made for that feature in her data. However, in terms of open syllables and disyllables, Raymond’s Stage I matches Grunwell’s first level of syllable structure development (1). The next significant step in syllable structure development involves the acquisition of closed syllables, that is, syllables with a final consonant, referred to as Ray-


mond’s stage II. Grunwell also shows closed syllables as being acquired at her second level of syllable structure development. Consonant clusters are acquired next by Raymond, with syllable-final clusters preceding syllable-initial clusters. Given the small difference in time between the two points of acquisition there is no reason to assign the occurrence of clusters to distinct stages. This stage III corresponds with Grunwell’s third level of syllable structure development. Syllables with a syllabic consonant are the last to emerge in Raymond’s speech at age 3;0, at the time of the final recording. Based on the 75% rule, we cannot conclude that it had been acquired by then, but it is reasonable to assume that it would have been acquired soon thereafter. This point is classified as stage IV, and in (6), where the child’s syllable structure development profile is shown, the age of acquisition is indicated as above (>) 3;0. Grunwell does not refer to syllabic consonants in her stages of syllable structure development, hence the need to propose a modified profile for Raymond. (6) Raymond’s syllable structure development profile¹² Stage I (1;2): Open syllables (C)V Disyllabic forms CVCV Stage II (1;7): Closed syllables (C)VC Stage III Consonant clusters a. (1;10): syllable-final (C)VCC b. (1;11): syllable-initial CCV(C) Stage IV (>3;0): Syllabic consonants CC̩ The general stages of development in other children were determined using the same procedure that was employed for Raymond, and then compared with the profiles outlined by Grunwell and Fikkert, respectively, for English and Dutch children.

4.2 The basilectal group The pre-secondary group consisted of ten children whose caregivers had less than secondary education and resided in the small rural community of Guys Hill in the parish of St. Catherine. As anticipated, the range of caregiver speech in this group was closer to the basilectal end of the continuum. Syllable structure development in the Pre-Sec group is summarized in Table 4.¹³ In spite of a few large discrepancies in the ages of acquisition of each type, the general sequence of syllable structure development is relatively consistent across individuals.¹⁴ Raymond’s (G11) acquisition data are now compared to the others in the group. As mentioned above, Table 4 reflects that by the age of 1;2 he had acquired open syllables with onsets, CV; open syllables without onsets, V; and disyllabic forms, CVCV. He acquired closed syllables, (C)VC, by the age of 1;7. Consonant clusters were

7

72

Rocky R. Meade

Table 4. Syllable structure development for the Pre-Sec group Syllable types CV CVCV V (C)VC CCV(C) (C)VCC CC̩

Age G11 G12 G13 G14 G15 G21 G22 G23 G24 G25 range 1;2 1;2 1;2 1;7 1;11 1;10 *

1;3 1;3 1;6 2;3 * * *

1;0 1;0 1;0 1;9 2;0 * *

1;2 2;0 1;6 1;6 2;6 2;6 2;3

1;0 1;0 1;3 1;6 2;9 1;6 *

^ ^ ^ ^ 3;3 3;3 2;9

^ ^ ^ ^ ^ 4;6 ^

^ ^ ^ ^ 4;6 ^ 3;7

^ ^ ^ ^ 3;6 3;6 4;6

^ ^ ^ ^ ^ 3;6 4;6

1;0–1;3 1;0–1;3 1;0–1;6 1;3–2;3 1;11–3;6 1;6–3;6 2;3–4;6

Mean Median age age 1;1 1;1 1;3 1;8 2;8 2;9 3;6

1;2 1;1 1;3 1;7 2;8 2;11 3;7

Notes : * No evidence for this type at the last recording; ^ acquisition before the first recording and excluded from calculations; Age:year; month; extreme value omitted from calculations

acquired first syllable-finally, (C)VCC, by 1;10 and then syllable-initially, CCV(C), by 1;11. Up to the final recording, there was no evidence that syllabic consonants (CC̩ ) were acquired, and this absence is marked by an asterisk (*). The data from the other children in the group are similarly presented in the table. With the exception of Tonya (G 14) — an exceptionally shy child who did not produce many utterances for her first three recording sessions up to the age of 1;10–, all the children, acquired disyllabic sequences (CVCV) and CV syllables by the age of 1;3. Tonya did not show evidence of CVCV until the age of 2;0.¹⁵ All the children in the older subgroup had acquired the first four syllable structure types at the time of the first recording. Therefore, the average ages for these types are calculated with data from the younger sub-group. Except for Raymond (G11) and Russell (G13), most children acquired onsetless syllables after the acquisition of CV and CVCV sequences. This is reflected in the mean and median ages of acquisition of V, which are higher than the means and medians for the acquisition of CV and CVCV. Although the ages differ, the children all acquired closed syllables, (C)VC, after V and before consonant clusters. Again the only exception was Tonya (G14), who appeared to acquire closed syllables at the same age as onsetless syllables. However, like the others, her closed syllables were acquired before consonant clusters. The group data suggest that the acquisition points of syllable-initial and syllablefinal consonant clusters need not be distinguished. Syllable-initial clusters and syllable-final clusters are acquired in varying orders, or simultaneously. This is one of the reasons why it is useful to examine both the individual data and the averages before making a conclusion about the group’s tendency. In this case, both the mean and median ages of acquisition for the group suggest that syllable-initial clusters precede syllable-final clusters, whereas individual data show that no distinction need be made between the two types of clusters for this group. The values excluded from the calculations of the mean and median do not affect this relationship at all


73

Table 5. Syllable structure development for the Post-Sec group Syllable Types

Mean Median L11 L12 L13 L14 L15 L21 L22 L23 L24 L25 Age range age age

CV CVCV V (C)VC CCV(C) (C)VCC CC̩

1;0 1;0 1;0 1;0 1;3 1;5 1;3 1;1 3;0 1;11 2;6 2;0 2;9 *

1;3 1;0 1;3 1;0 1;3 1;2 1;8 1;9 3;0 2;2 2;8 2;6 * *

1;3 1;3 1;3 1;3 2;0 2;9 2;9

^ ^ ^ ^ ^ 2;9 2;9

^ ^ ^ ^ 4;6 4;6 4;6

^ ^ ^ ^ 3;6 4;6 4;6

^ ^ ^ ^ ^ 3;6 *

^ ^ ^ ^ 3;6 ^ 3;6

1;0–1;3 1;0–1;3 1;2–1;5 1;1–1;9 1;11–3;6 2;0–3;6 2;9–

1;1 1;1 1;3 1;5 2;9 2;8 3;7

1;0 1;0 1;3 1;3 3;0 2;8 3;6

Notes : * No evidence for this type at the last recording; ^ acquisition before the first recording and excluded from calculations; Age:year; month; extreme value omitted from calculations

since a similar value was omitted for both syllable-initial and syllable-final clusters. Four of the children — all from the younger subgroup — show no evidence of the acquisition of syllabic consonants up to the last recording. However, there are sufficient data from the remainder of the children to determine average ages of acquisition. Both the mean and median ages indicate clearly that syllabic consonants are the last syllable types to be acquired. Only two children, Tonya (G14) and Alison (G21), showed evidence of the acquisition of syllabic consonants before consonant clusters.

4.3 The mesolectal/acrolectal group The post-secondary group consisted of 10 children whose caregivers had received education beyond the secondary level (university or college) and resided in the urban community of Linstead, also in the parish of St. Catherine. As anticipated the range of caregiver speech in this group was closer to the acrolectal (JE) end of the continuum. The Post-Sec group’s syllable structure development results are summarized in Table 5.¹⁶ This group sometimes shows wide differences in the ages of acquisition, but the general sequence of acquisition is relatively consistent. Open syllables with onsets (CV) and disyllabic forms (CVCV) were acquired simultaneously by all the children. The mean acquisition age is 1;1 for both types. Onsetless syllables were acquired after CV and CVCV, except for Davin (L13) and Sharon (L15), who acquired all three types at the same data point. However, in both cases the onsetless syllables appeared during the first recording, which therefore does not reveal the sequence of acquisition of these syllable types. Except for Teresa (L12), all the children acquired closed syllables after onsetless syllables or at the same age. This is reflected in the group’s mean and median ages. All the children acquired consonant clusters after closed syllables. There is some

74

Rocky R. Meade

individual variation in the relative order of acquisition of syllable-initial and syllable-final clusters. Although on average syllable-final clusters were acquired before syllable-initial clusters, individual data indicate that six of the ten children acquired syllable-initial clusters earlier. This situation reflects the extent of individual variation rather than a group norm. Syllabic consonants are clearly the last type of syllable structure to be acquired, as evidenced by both the mean and median ages. Four of the children had not acquired syllabic consonants up to the final recording. Except for two children who acquired syllabic consonants at the same age as they acquired syllable-final consonant clusters, all the children in this group acquired syllabic consonants later.

4.4 Comparison of the groups Both groups originated from the parish of St. Catherine, thus avoiding the additional complexity of regional variation. However, acquisition varies according to the caregivers’ education level and area of residence. The net effect of these differences is that the children were exposed to different language varieties. The children in the Pre-Sec group were exposed primarily to varieties closer to the basilectal end of the language continuum, while the Post-Sec children were exposed primarily to varieties close to the acrolectal end. Whereas the previous section focused on the similarities presented by individual children, this section displays overall group results in terms of age range and average age (see Table 6). Notwithstanding the different language inputs and environments, the similarities between the two groups are striking, not only in terms of the order of acquisition of the various syllable structure types, but also in terms of the age of acquisition. Regarding the first three syllable types, the mean age of acquisition is exactly the same for both groups. In the case of open syllables with onsets and disyllabic sequences, even the age range is exactly the same. The age range is also the same in both groups Table 6. Syllable structure development averages for both groups Pre-SecGroup

Post-SecGroup

Syllable types

Age range

Median Mean age age

Age range

Mean age Median age

CV CVCV V (C)VC CCV(C) (C)VCC CC̩

1;0–1;3 1;0–1;3 1;0–1;6 1;3–2;3 1;11–3;6 1;6–3;6 2;3–4;6

1;1 1;1 1;3 1;8 2;8 2;9 3;6

1;0–1;3 1;0–1;3 1;2–1;5 1;1–1;9 1;11–3;6 2;0–3;6 2;9–

1;1 1;1 1;3 1;5 2;9 2;8 3;7

1;2 1;1 1;3 1;7 2;8 2;11 3;7

1;0 1;0 1;3 1;3 3;0 2;8 3;6


for the acquisition of syllable-initial consonant clusters. The only notable difference between the groups is the order of acquisition of consonant clusters in the different syllable positions. However, there is individual variation in the sequence of acquisition of the clusters in both groups and there is only a one-month difference between the mean cluster acquisition ages of the groups. This again suggests that a distinction need not be made between consonant cluster acquisition in the different syllable positions. Tests for statistical significance¹⁷ further show that the syllable structure development paths of the groups are statistically similar. Hence, a single profile of the stages of syllable structure development of the two groups is now plausible. In (7), the combined general pattern for all the children is shown. (7) Syllable structure development stages for all the children Stage I (≤1;3) Open syllables CV Disyllabic forms CVCV Stage II (1;0–1;6) Onsetless syllables V Stage III (1;1–2;3) Closed syllables (C)VC Stage IV (1;6–3;6) Consonant clusters (C)CVC(C) Stage V (2;3–-) Syllabic consonants CC̩ The syllable development profile in (7) can be compared with the stages outlined by Grunwell (1982) and Fikkert (1994), which were presented as (1) and (2), respectively. In some cases the age range for the acquisition of syllable structure is lower for the Jamaican children than for those studied by Grunwell. No significance is attributed to this difference for two reasons. Firstly, Grunwell’s figures are based on composite data from different studies, and do not include the full range of ages of acquisition for all the children. The age range presented for the Jamaican children includes all the ages of syllable structure acquisition for the children in the study. Secondly, it is not surprising to have individual variation between the groups as children’s age of acquisition generally varies widely even within subgroups. Focusing on the order of the stages as opposed to the ages, the profile in (7) is similar to the relevant portions of the profiles presented by Fikkert and Grunwell. Grunwell does not include onsetless syllables and syllabic consonants in her profile of stages. However, similar components are perfectly matched in terms of their order of occurrence. Although the comparison with Fikkert’s model was limited to onset-related sequences, it is clear that Fikkert’s first two stages perfectly match the first two stages represented in (7). The 20 children display consistent syllable structure development in spite of occasional individual variation in the age of acquisition. What is significant is that all the children follow a similar development path in terms of the sequence of the various stages, even with the linguistically relevant socio-economic differences. When combined into one profile of stages, as in (7) above, the syllable structure development path for these Jamaican children is similar in terms of the order of acquisition to profiles reported for English (Grunwell 1982) and Dutch children (Fikkert 1994).

75

76

Rocky R. Meade

Both Grunwell and Fikkert predict that their findings are generally applicable to all children. To the extent that they are correct, the Jamaican data is in keeping with universal expectations, certainly in terms of the order of acquisition of various syllable structure types.

5. An OT analysis The descriptive analysis of the data in the previous section provides the basis for the theoretical analysis presented in this section. The similarities in the sequence of acquisition of syllable structure by all the children allowed for the formulation of one list of stages of development (7) to represent the data from all the children. The earliest syllable patterns and the various stages of development towards the complete adult inventory will now be explained in terms of OT. First, the relevant constraints are considered.

5.1 Relevant OT constraints The OT constraints that are used to account for the development path are all derived from observed, generally preferred conditions on syllable structure composition in most languages. They either conform to structural well-formedness conditions, or to the tendency for production to match the lexicon as closely as possible. The wellformedness constraints include ONSET (syllables must have onsets), *CODA (syllables must not have codas), *COMPLEX (syllables must not have more than one segment associated directly with a node) and *C NUC (a consonant must not form the peak of a syllable). This last well-formedness constraint is proposed here in this form for precision of application, but is derived from the observation that vowels are preferred nuclei. Another constraint utilized in the theoretical analysis of the data is not based on universal structural preferences, but rather on the universal preference for the output to be similar to the input from the lexicon. There are several constraints that address the issue of input/output correspondence. These are typically grouped as the faithfulness family of constraints. The data considered here do not require the deTable 7. OT constraints used in the analysis Constraint

Definition (informal)

*ONSET *CODA *COMPLEX *C NUC *FAITH

A syllable must have an onset A syllable must not have a coda A syllable must not have a complex component A syllable peak must not be a consonant The output must correspond to the input


tailed elaboration of all the faithfulness constraints; the constraint FAITH is thus used instead in order to capture the requirement that the output correspond with the input. Table 7 includes a summary of the constraints used.

5.2 Syllable structure development stages One of the shortcomings of many OT analyses is the tendency to provide very limited examples of the operation of OT grammars. In acquisition research, Levelt, Schiller and Levelt (2000) constitutes one notable exception. This OT analysis of the acquisition of Dutch syllable types demonstrates the OT outcomes with a variety of inputs and a variety of candidates. However, it does not include all the syllable types in each example. In the case of the Jamaican acquisition data, all syllable types produced by the children are included at different stages as inputs and candidates. This comprehensive analysis has the advantage of demonstrating more clearly how the developing OT grammar allows an increasing number of candidates to form outputs from the wide range of inputs. Each stage of syllable structure development is examined separately in order to detect which output forms are allowed from the full range of inputs, and candidates that the children are likely to encounter.¹⁸

5.2.1 Stage I: Open syllables with onsets [CV]

Based on the results discussed in Section 4, at Stage I, the grammar should allow only CV syllables to be produced irrespective of the input. For this to be achieved, all the constraints that affect syllable structure must be ranked above the constraint FAITH. The relevant constraint hierarchy is presented in (8). As is shown in Tableau (8a), even with the input being of the type V, the optimal candidate for the output is CV. In this case V is the only candidate to satisfy FAITH, but FAITH is ranked too low to offset the violation of ONSET. In Tableaux (8b) to (8f), each of the other syllable types form inputs to the same constraint hierarchy, and the expected results (the CV output) are produced in each case. (8) *C NUC, *COMPLEX, *CODA, ONSET >> FAITH¹⁹ Tableaux 8a–f Various inputs resulting in open syllables with onsets –CV 8a. Input: V

V CV CVC CCVC CC̩

*C NUC

*COMPLEX *CODA

ONSET

FAITH

*!

*!

*!

*! *

* ** *** **

77

78

Rocky R. Meade

8b. Input: CV

*C NUC

V →CV CVC CCVC CC̩

*!

8c. Input: CVC

*C NUC


*!

8d. Input: CCVC *C NUC V →CV CVC CCVC CC̩

*!

8e. Input: CC

*C NUC


*!

8f. Input: CVVC *C NUC V →CV CVC CVVC CC̩

*!

*COMPLEX *CODA

*!

* ** * ONSET

FAITH

*!

** * ** **

ONSET

FAITH

*!

*** ** * **

ONSET

FAITH

*!

*** ** *

*! *

*COMPLEX *CODA

*!

*

*! *

*COMPLEX *CODA

*!

*!

*! *

*COMPLEX *CODA

*!

FAITH

*! *

*COMPLEX *CODA

*!

ONSET

** ONSET

FAITH

*!

*** ** *

*! *

5.2.2 Stage II: Open syllables with and without onsets [(C)V]

**

At stage II, the majority of the children are able to produce open syllables with and without onsets. One should expect therefore that an onsetless candidate should be optimal at least in the case of an onsetless input. The constraint that prevented the selection of the onsetless candidate in Tableau (8a), which had an onsetless input,


is ONSET. At Stage II this constraint is demoted below FAITH (9). The examples in Tableaux (9a) to (9f) demonstrate the expected results. Only simple (no cluster) open syllables, V and CV, were produced. (9) *C NUC, *COMPLEX, *CODA >> FAITH >> ONSET Tableaux 9a–f Various inputs resulting in optional onset open syllables –(C)V 9a. Input: V

*C NUC

→V CV CVC CCVC CC̩

*!

9b. Input: CV

*C NUC


*!

9c. Input: CVC

*C NUC


*!

9d. Input: CCVC *C NUC V →CV CVC CCVC CC̩

*!

9e. Input: CC

*C NUC


*!

*COMPLEX *CODA

*!

*! *

*COMPLEX *CODA

*!

*! *

*COMPLEX *CODA

*!

*! *

*COMPLEX *CODA

*!

*! *

*COMPLEX *CODA

*!

*! *

FAITH *! ** *** **

ONSET *

FAITH

ONSET

*!

*

* ** *

FAITH

ONSET

**! *

*

* ** FAITH

ONSET

***! ** *

** * *

**

*

FAITH

ONSET

**! * * **

* * *

79

80

Rocky R. Meade 9f. Input: CVVC *C NUC V →CV CVC CVVC CC̩

*!

*COMPLEX *CODA

*!

*! *

FAITH

ONSET

***! ** *

*

**

5.2.3 Stage III: Closed syllables [(C)VC]

At Stage III most of the children began producing closed syllables. The *CODA constraint, which blocked the selection of candidates with codas in the previous stages, is demoted below FAITH at this stage. The constraint hierarchy is shown in (10). It allows for the selection of a candidate with a coda, but only when there is a coda in the input form. Where open syllables form the input, an open syllable candidate is selected. This is shown in Tableaux (10a) through (10f). Note that the OT grammar, with FAITH still ranked below *COMPLEX and *C NUC, still rejects candidates with clusters or syllabic consonants. (10) *C NUC, *COMPLEX >>FAITH >> *CODA, ONSET Tableaux 10a–f Various inputs with possible closed syllable outputs –(C)VC 10a. Input: V

*C NUC


*!

10b. Input: CV

*C NUC


*!

*! **! *** **

*COMPLEX FAITH

*CODA

*!

*!

*! ** **

*COMPLEX FAITH **! *!

*!

*!

* **

ONSET *

* *

*CODA

*!

10c. Input: CVC *C NUC V CV →CVC CCVC CC̩

*COMPLEX FAITH

ONSET *

* *

*CODA

ONSET *

* *

OT and the acquisition of Jamaican syllable structure 10d. Input: CCVC *C NUC V CV →CVC CCVC CC̩

*!

10e. Input: CC

*C NUC


*!

10f. Input: CVVC *C NUC V CV →CVC CVVC CC̩

*!

*COMPLEX FAITH

*!

***! **! * **

*COMPLEX FAITH

*!

**! ** ***! ***

*COMPLEX FAITH

*!

***! **! * *

*CODA

ONSET *

* *

*CODA

ONSET *

* *

*CODA

ONSET *

* *

5.2.4 Stage IV: Consonant clusters [(C)CVC(C)]

At Stage IV the children generally begin adding clusters to their inventory. This signals the demotion of the *COMPLEX constraint. The reranked constraint hierarchy is as presented in (11). FAITH is now ranked above *COMPLEX, *CODA and ONSET, thereby allowing outputs with clusters in addition to the previously restricted outputs with codas and optional. This is demonstrated in tableaux (11a) to (11f). (11) *C NUC >>FAITH >> *COMPLEX, *CODA, ONSET Tableaux 11a–f Various inputs with possible consonant cluster outputs – (C)CVC(C) 11a. Input: V

*C NUC

FAITH


*!

*! **! ***! **

11b. Input: CV

*C NUC

FAITH


*COMPLEX *CODA

* *

* *

*COMPLEX *CODA

*!

*!

*! **! *

ONSET

ONSET *

*

* *

8

82

Rocky R. Meade 11c. Input: CVC *C NUC

FAITH

V CV →CVC CCVC CC̩

**! *! *!

* *

*

*COMPLEX *CODA

11d. Input: CCVC *C NUC

FAITH ***! **! *!

*!

**

11e. Input: CC

*C NUC

FAITH

*

* *

*COMPLEX *CODA

*

11f. Input: CVVC *C NUC

FAITH

*COMPLEX *CODA

V CV CVC →CVVC CC̩

***! **! *! *!

***

ONSET *

**! * **! **!

*!

ONSET *

*! **

V CV CVC →CCVC CC̩


*COMPLEX *CODA

ONSET *

* *

ONSET *

*

5.2.5 Stage V: Syllabic consonants [CC?]

* *

The last observed change in the syllable-type inventory of these children is the inclusion of syllables with syllabic consonants. Although the majority of the children had not fully acquired this syllable type by the end of the data collection period, most were at least showing evidence of its emergence. What is clear is that syllabic consonants were generally disallowed before this stage. This acquisitional chronology suggests that the *C NUC constraint, which disallowed syllabic consonants, is demoted below FAITH at this stage. The constraint hierarchy at this stage, as shown in (12), is what one would expect of the adult JC grammar for syllable structures considered. At this stage, all the syllable types that were examined in this research are allowed in most cases. Tableaux (12a) to (12f) exemplify this.


(12) FAITH >> *C NUC , *COMPLEX, *CODA, ONSET Tableaux 12a-f Various inputs with possible syllabic consonant outputs – CC̩ 12a. Input: V

FAITH


*! **! ***! **!

*

12b. Input: CV

FAITH

*C NUC


*! *! **! *!

*

12c. Input: CVC

FAITH

*C NUC

V CV →CVC CCVC CC̩

**! *! *! **!

12d. Input: CCVC FAITH V CV CVC →CCVC CC̩

*C NUC

* *

*COMPLEX *CODA

ONSET

*

* *

*COMPLEX *CODA

ONSET *

* *C NUC

*

* *

*COMPLEX *CODA

ONSET *

*

12e. Input: CC

FAITH

*C NUC

V CV CVC CCVC →CC̩

**! *! **! **!

*

* *

*COMPLEX *CODA

ONSET *

* *C NUC

***! **! *! ***!

*

*

**!

V CV CVC →CVVC CC̩

ONSET *

***! **! *!

12f. Input: CVVC FAITH

*COMPLEX *CODA

*

* *

*COMPLEX *CODA

ONSET *

*

*

* *

83

84

Rocky R. Meade

5.2.6 Three-consonant clusters [CCCV or VCCC]

None of the children, not even those that appear to be speaking the acrolect,²⁰ acquired three-consonant clusters during the data collection period. Notwithstanding this, an indication is given in Tableaux (13) and (14) of how OT, with basically the same constraints, would account for the acquisition of more complex clusters. This requires the refinement of the *COMPLEX constraint to specifically target (and reject) three-consonant clusters. This constraint is referred to as *CPX 3, and it is ranked higher than FAITH and the general consonant cluster constraint *COMPLEX. The constraint hierarchy that would allow the production of two-consonant clusters but not three-consonant clusters is presented in (13). The one that would allow both types of clusters is shown in (14). (13) *CPX 3 >> FAITH >> *C NUC, *COMPLEX, *CODA, ONSET Tableaux 13a–b Two-consonant cluster production 13a: Input: CCCVC

*CPX3

V CV CVC CCVC CC̩ CCCVC

*!

13b: Input: CVCCC

*CPX3

V CV CVC →CVCC CC̩ CVCCC

*!

FAITH *C NUC ****! ***! **! * ***!

*CODA

ONSET *

*

FAITH *C NUC ****! ***! **! * ***!

*COMPLEX

*

* *

**

*

*COMPLEX

*CODA

ONSET *

*

*

* *

**

*

(14) FAITH >> *CPX 3, *C NUC, *COMPLEX, *CODA, ONSET Tableaux 14a–b Three-consonant cluster production 14a: Input: CCCVC

FAITH *CPX 3 *C NUC

V CV CVC CCVC CC̩ →CCCVC

****! ***! **! *! ***!

*COMPLEX

*CODA

ONSET *

*

*

*

* *

**

*

OT and the acquisition of Jamaican syllable structure 14b: Input: CVCCC

FAITH *CPX 3 *C NUC

V CV CVC CVCC CC̩ →CVCCC

****! ***! **! *! ***!

*COMPLEX

*CODA

ONSET *

*

*

*

* *

**

*

5.3 Accounting for variation OT can also account for the variation that is observed between the children. Variation exists primarily in the age of acquisition of the various syllable structure types. In a few cases, children showed slight differences in the sequence of acquisition of the structures. The variation in age of acquisition is accounted for by the fact that the children are not restricted to any particular age for the reranking of the constraints. The age at which each constraint is reranked will be based on a number of factors that are not dependent on the theory, such as the nature of the caregiver/child interaction. What is significant is the fact that the general sequence of acquisition is similar for all the children. One of the few discrepancies observed in the acquisition sequence involves the acquisition of syllabic consonants. Two children, Tonya (G14) and Alison (G21), acquired syllabic consonants before consonant clusters, unlike the majority of the children for whom the reverse was true. This suggests that (a) there are some constraints for which the reranking sequence is optional, and (b) some constraints may move more than one place in the hierarchy. It should be noted that the sequence of acquisition is not just freely variable. There is no evidence in this set of data, for example, of any child acquiring clusters before the core CV syllable. Therefore the order of acquisition is generally consistent and often rigid, but nonetheless the theory can account for the exceptional cases where the sequence differs.

6. Concluding remarks Responding to the need for more creole acquisition data generally and for JC data in particular, the project on which this article is based recorded the speech of more than twenty children over two years to produce a database that spans most of the Jamaican language continuum. The data suggest that the acquisition presents more similarities than differences. Significantly, the chronology of acquisition amongst these Jamaican children is analogous to that observed in children learning English and Dutch. The children follow a fairly consistent pattern of starting with the core CV syllable. They then progress by acquiring optional onsets, codas, clusters and syllabic consonants.

85

86

Rocky R. Meade

OT proved to be an effective theoretical framework to not only account for the sequence of acquisition, but also for the variation in the acquisition process. This is achieved by the continuous reranking of universal constraints until the child’s phonological system matches the adult target system.

Notes 1. The project from which the data for this article are drawn was carried out primarily in the Netherlands with fieldwork in Jamaica, and was funded by the Netherlands Foundation for the Advancement of Tropical Research (WOTRO), project number 37–075. Initial results are published in Meade (2001). 2. Fikkert does not give age ranges for each stage. The ages in (2) are based on examples she presents (1994: 58ff). 3. Some linguists working within OT, such as Boersma (1998), and Bolognesi (1998), suggest that not all constraints are universal, and that universal constraints are not innate. 4. See LaCharité and Paradis (1993) for a discussion of other constraint-based approaches to generative grammar. 5. This issue is discussed by Helsloot (1995) in relation to rhythm. 6. The two parishes of Kingston and Saint Andrew form a metropolis of one city with surrounding suburbs. They will be jointly referred to as Saint Andrew. 7. Alphanumeric codes are used to identify the children. The letter represents the community of residence, G for Guys Hill and L for Linstead. The first number shows the age group, one for the group beginning at age 1;0, and two for the group beginning at age 2;6. The second number is the individual child’s number within the subgroup. 8. Two of the targeted data points for recording and transcription were missed, due to his unavailability at the time. The missed data points are at the ages of 1;5 and 1;9. 9. The English gloss is given first, followed by the transcription of the target form, the child’s form in square brackets, and the age at time of acquisition. 10. It could be argued that the vowel in the child’s version of banana is not the first vowel of the adult form, but rather the stressed second vowel (Fikkert 1994: 229). This issue is not crucial here since the focus is simply on the fact that the child produced the monosyllabic CV syllable. 11. The dot (.) in the first example in (4) is used to represent a syllable boundary. 12. Three consonant clusters (CCC) do not occur in the data. These clusters are in fact not a part of basilectal JC phonology. 13. On the left of the table are the various syllable types noted during the development of the syllable structure. The next ten columns specify the ages at which the individual children acquired the various syllable structure types, the age range over which the various types were acquired, and the mean age and the median age for each type. 14. Special symbols and shadings are used to indicate data that are not used in the calculations of group averages. Syllable structure types acquired by the older group of children (G21 to G25) at or before the first recording are marked by the symbol ʌ. Types for which there

OT and the acquisition of Jamaican syllable structure is no evidence of acquisition up to the final recording are marked by an asterisk (*). Acquisition dates that appear to be unusually late, compared with those of the other group members, are placed in shaded cells. 15. This is considered to be an exception, and is therefore excluded from the calculations of averages. Additional support for this decision comes from the fact that with the exclusion of this figure, the mean and median ages for acquisition of CVCV are exactly the same, i.e., 1;1. The age of 4;6 at which Bryce (G22) and Damion (G23) acquired syllable-final and syllableinitial consonants, respectively, were excluded from the calculations for similar reasons. 16. Three ages of acquisition were excluded from the calculations of the group averages. They include two from Nathan (L22), for both types of consonant clusters, and one from Karel (L23), for the syllable-final cluster. In each of these cases the indicated ages (4;6 in each case) occurred for types that the majority of the other children acquired by the age of 3;0. 17. See Meade (2001: 90) for details. 18. The codes used in the following tableaux are:

* ! →

violation of a constraint fatal violation output form (optimal candidate)

19. The use of commas in this and subsequent constraint hierarchies is not intended to suggest that the affected constraints are absolutely not ranked with respect to each other in the grammar, but rather to indicate that no evidence of their relative ranking is presented in this paper. 20. The acrolect (JE) has CCC clusters unlike the varieties closer to the basilect (JC), which has CC clusters as reflexes of the CCC sequences.

References Archangeli, D. and T. Langendoen. 1997. Optimality Theory: An Overview. Malden: Blackwell. Beckford-Wassink, A. 1999. A Sociophonetic Analysis of Jamaican Vowels. PhD Thesis. University of Michigan. Boersma, P. 1998. Functional Phonology: Formalising the Interactions between Articulatory and Perceptual Drives. PhD Thesis. Amsterdam: Netherlands Graduate School of Linguistics (LOT). Bolognesi, R. 1998. The Phonology of Campidanian Sardinian. PhD Thesis. Amsterdam: Holland Institute of Generative Linguistics (HIL). Dissertations in Linguistics 38. Chomsky, N. 1995. The Minimalist Program. Cambridge, MA: Massachusetts Institute of Technology Press. Devonish, H. 1998. “On the existence of autonomous language varieties in creole continuum situations.” .” In Studies in Caribbean Language II, L. Carrington (ed.), 1–12. St. Augustine, Trinidad: Society for Caribbean Linguistics. Dirks, J. 2000. OT Syntax. PhD Thesis. Amsterdam: Holland Institute of Generative Linguistics (HIL). Fikkert, P. 1994. On the Acquisition of Prosodic Structure. PhD Thesis. Amsterdam: Holland Institute of Generative Linguistics (HIL). Dissertations in Linguistics 6. Grunwell, P. 1982. Clinical Phonology. London: Croom Helm.

87

88

Rocky R. Meade Helsloot, K. 1995. Metrical Prosody: A Template-and-Constraint Approach to Phonological Phrasing in Italian. PhD Thesis. Amsterdam: Holland Institute of Generative Linguistics (HIL). Dissertations in Linguistics 16. LaCharité, P. and C. Paradis. 1993. “The emergence of constraints in generative grammar and a comparison of three current constraint-based models.” Canadian Journal of Linguistics 38 (2): 127–53. Levelt, C., Schiller, N. and Levelt, W. 2000. “The acquisition of syllable types.” Language Acquisition 8: 237–64. McCarthy, J. and A. Prince. 1993. “Prosodic morphology I: Constraint interaction and satisfaction.” Ms. University of Massachusetts and Rutgers Centre for Cognitive Science Technical Report 3. Meade, R. 2001. The Acquisition of Jamaican Phonology. PhD Thesis. Amsterdam: Holland Institute of Linguistics and Netherlands Graduate School of Linguistics. Patrick, P. 1999. Urban Jamaican Creole: Variation in the Mesolect. Amsterdam: Benjamins. Prince, A. and P. Smolensky. 1993. “Optimality Theory: Constraint interaction in generative grammar.” Ms. Rutgers University Centre for Cognitive Science Technical Report 2. Smith, N. 2000. “Dependency Theory meets OT: A proposal for a new approach to segmental structure.” In Optimality Theory: Phonology, Syntax and Acquisition, J. Dekhers, F. van der Leew and J. van de Weijer (eds.), 243–76. Oxford: Oxford University Press. Tesar, B. and P. Smolensky. 1994. “The learnability of Optimality Theory: An algorithm and some basic complexity results.” Ms. University of Colorado at Boulder. Tesar, B. and P. Smolensky. 1998. “Learnability in Optimality Theory.” Linguistic Inquiry 29: 229–68.

CHAPTER 8

Double-object constructions in two Frenchbased creoles (Morisyen and Seselwa) Dany Adone

University of Düsseldorf

In this article I discuss “dativized” verbs in two French-based creoles, Morisyen (spoken in Mauritius) and Seselwa (in the Seychelles islands). These verbs sometimes have a prepositional complement structure [NP to NP], and a ditransitive structure [NP NP]. I argue that the Double-object construction (NP NP) is the default structure associated with ditransitive verbs in these (and possibly also other) creoles. To analyze “dativized” verbs, I use Wunderlich’s Lexical Decompositional Grammar, which accounts for the linking of structural arguments to grammatical relations. Structural arguments vary from language to language. The two creoles treated here offer evidence that they possess two structural arguments, but that — like English — the creoles also have Double-object constructions. This study supports the view that the Double-object construction is the default structure associated with ditransitive verbs in the lexicon.

1. Introduction Previous studies have confirmed that in many of the world’s languages the syntactic behavior of verbs is closely tied to their meaning (Hale and Keyser 1987, Levin 1993). The close link between verb behavior and meaning on the one hand, and the classification of verbs of the same semantic type on the other hand, is manifested across languages of the world. In this article I discuss the case of a particular group of verbs that are cognitively compatible with the notion of transfer and communication. These verbs have been referred to in the literature as verbs that “dativize”, i.e., alternate between a prepositional complement structure [NP to NP] and a ditransitive one [NP NP]. I claim that the Double-object construction (hereafter DOC) is the default structure associated with ditransitive verbs in creole languages. Furthermore, findings from sign language and first language acquisition studies taken together with creole studies strongly suggest that DOC is the structure associated with ditransitive verbs in the lexicon.

90

Dany Adone

In order to support the claim as formulated above, I investigate the status of DOC in two creoles. I adopt Wunderlich’s (1997a, 1997b) Lexical Decompositional Grammar (LDG), which provides an account of how structural arguments are linked to grammatical relations.¹ Cross-linguistic studies have shown that the number of arguments that a verb can have ranges between two and three in a given language. Basque has three structural arguments. English makes use of two structural arguments except in DOCs, where it has three arguments (Mary gave him the book) in which the “dative” argument (him) is expressed as the direct object. Other languages may be restricted to one or two structural arguments if they have neither morphological case nor agreement. The two creoles investigated — Morisyen and Seselwa — here provide evidence for languages with two structural arguments. This article is structured as follows: section 2 presents a brief discussion of how grammatical relations are marked. Section 3 focuses on the distribution of DOC in creole languages. In the same section I also present an analysis of Wunderlich’s (1997a) Lexical Decomposition Grammar. Section 4 offers a brief review of DOC in first language acquisition as well as in sign language literature. This is followed by concluding remarks.

2. Grammatical relations In this paper I treat DOC as a default and unmarked structure. In terms of storage in the working memory, a default structure is said to require the least effort to be processed. Speakers can access these structures more easily when other structures are not available. Unmarked here refers strictly to cross-linguistic distribution. Unmarked structures are thus encountered more frequently across the world’s languages than marked ones. Examples are found in the asymmetries between matrix and subordinate clauses illustrated by the Germanic verb-second rule, or the ergative and nominative distinction (Silverstein 1976, Croft 1990). Natural languages have three ways to mark grammatical relations (GR): word order, morphological case, and verb agreement. Transitive clauses are those with two arguments: an agent plus a patient, goal or theme. In many European languages, intransitive clauses have one argument that equates the subject. Ditransitive clauses have three arguments (agent, patient and theme). In many languages, the goal argument (ditransitive direct object) is expressed in the same way as the patient argument (transitive direct object), and the theme argument is expressed differently. The goal and patient arguments are known as the primary objects. And the theme argument is known as the secondary object (Croft 1990: 103). The primary/secondary object distinction is expressed in agreement patterns as seen in European languages like English: I gave/sent her the book (primary/secondary object distinction), in which the goal/recipient “her” is dativized. The direct-indirect object distinction is seen as in I gave/sent the book to her. According to Croft, there is a hierarchy in marking the GR: subject < (direct) object < oblique. Croft (1990) argues that objects

Double-object constructions in two French-based creoles

precede obliques, with the position of the indirect objects (Goal arguments) varying. In many languages the direct object (Theme) is preceded by the indirect object (Goal) as in the English dative shift construction (I gave her the book). Languages behaving similarly to English use the primary/secondary object distinction, so that the order is “primary object followed by secondary object”. The GR hierarchy subject < object < oblique is said to be manifested in the linear order of the arguments. This means that the conceptual ranking of arguments as described by the GR hierarchy is mirrored in the physical sequence of arguments in the phrase. Verbs of transfer and communication are known for taking three semantic arguments: the sender, the recipient and the object/message that is being transferred. There are two automatically licensed positions, one for the subject, and the other for the direct object. The canonical thematic roles linked with these positions correspond to sender and object/message transferred. Interestingly, we see two options available for the goal/recipient argument. A. Either it is in a syntactically unmarked position, that is, external to the verb and object unit and a part of the prepositional phrase (or verb phrase in serial verb construction). Two conditions are necessary for this: the language in question must have an element such as English “to” and a mechanism of transmitting the recipient thematic role of the verb to another case assigner. Example (1) illustrates the point: (1) Claude gives the book to Karelina. B. Or, it is in a syntactically special position, that is, in a construction with the direct object (e.g., in a small clause), and its thematic role is received directly. What I call “special” here is the possibility for an argument (object or message) to behave as a predicate that gives the recipient role together with the verb. Sentence (2) is an example: (2) Claude gives Karelina the book. Bruyn, Muysken and Verrips (2000) propose that option (a) is the result of a lexicosemantic operation, while option (b) can be seen as the outcome of a syntactic operation. The first option is called “Prepositional Ditransitive Construction” or PDC, Option (b) is called “Double-object construction” or DOC.

3. Double-object construction and prepositional ditransitive construction in creole languages Before I discuss the nature of DOCs, I would like to address the distribution of DOCs and PDCs in creole languages. Many creoles whose European lexifier languages have double-objects also display DOC. This is true of Negerhollands with Dutch as lexifier language, as well as of Tok Pisin with English as lexifier language. Examples 3 and 4 (taken from Bruyn, Muysken and Verrips 2000: 330) illustrate the point:

9

92

Dany Adone

(3) Negerhollands Ham a gi di mam si gout. 3SG PAST give DET man 3POS gold ‘He gave the man his gold.’ (4) Tok Pisin Mi soim yu banara bilong mi. 1SG show 2SG bow PREP 1SG ‘I show you my bow.’ However DOC also occurs in creole languages whose European lexifiers have no DOC, such as Portuguese and French. Examples come from Saramaccan with Portuguese (5), or Seselwa and Morisyen, both of which have French as lexifier language (see (6)- (7), respectively). (5) Saramaccan Mi ke pindja I wan soni. 1SG want tell 2SG one thing ‘I want to tell you something (in secret).’ (6) Seselwa Komela mo pe don u larzan. now 1SG PROG give 2SG/PL money ‘I am giving you money now.’ (7) Morisyen Asterla mo pe don twa larzan. now 1SG PROG give 2SG money ‘I am giving you money now.’ Table 1 makes clear that, as argued by Bruyn, Muysken and Verrips (2000), DOC is present in almost all these creole languages. When we check the distribution of DOC in the creole languages given in Table 1, we see some complementary relation with other constructions such as PDC and Serial Ditransitive Constructions (also known as SDC). In the survey conducted by Bruyn, Muysken and Verrips (2000), DOC is available in almost all creoles, while PDC and the SDC are more limited. Following behind DOC, PDC is the second most widespread construction, while SDC occurs only occasionally. However, in the case of Morisyen, Ngukurr Kriol, and Seselwa all three constructions occur.

4. DOC and PDC in Morisyen and Seselwa The data presented in this paper have been gathered by the author during her fieldwork in Mauritius and in the Seychelles in 2000. I collected spontaneous speech samples, and also obtained data from native speakers through elicitation. Also

Double-object constructions in two French-based creoles Table 1. Distribution of DOC, PDC, and SDC in creole languages⁸

Sranan Saramaccan Jamaican Creole Tok Pisin Negerhollands Berbice Dutch Afrikaans Haitian Creole Guyanais St. Lucian Louisiana Creole Seselwa Papiamentu Palenquero Guinea-Bissau Kryol Fa d’Ambu Principense Mayo-Portuguese Sri Lanka Portuguese Morisyen Ngukurr Kriol

DOC

PDC

SDC

+ + + + + + + + + + + + + + + + + − + + +

+ − + + + − + − − − + + − − − − − + + + +

+ + − − − + − + − − − − − − − + + − − + +

included here will be Seselwa data from Bollée (1977) and Corne (1977). Examples (8)–(10) are illustrative of how ditransitive done ‘give’ is used in Morisyen: (8) Mari pe don so papa tabak. Mary PROG give 3POS father tobacco ‘Mary gave tobacco to her father.’ (9) Mari pe don tabak pu so papa. Mary PROG give tobacco for 3POS father ‘Mary gave tobacco for her father.’ (10) Mari pran tabak don so papa. Mary take tobacco give 3POS father ‘Mary gave tobacco to her father.’ The same constructions are also witnessed in Seselwa. Example (8) is a DOC. Example (9) is a PDC with the preposition pu ‘for’. Example (10) is a Serial Ditransitive Construction (SDC) with pran/don ‘take/give’ (see Bickerton 1989; and Seuren 1990 for Seselwa). The PDC is available for benefactives, and is dependent upon the presence of a preposition similar to English “to”.

93

94

Dany Adone

Bollée (1977) and Corne (1977) provide detailed analyses of Seselwa DOC. They both observe that DOC occurs with verbs of transaction and communication (see also Levin (1993) for a typology of verbs in English). Note that the criteria and the Seselwa examples (11) to (13) are also valid for Morisyen. (11) Seselwa Si mo vje dalon torti pa ti donn mwa en buse manze `. . . if my old friend turtle NEG TNS give me a bit food ‘If my old friend Turtle did not give me a bit to eat . . .’ (12) Rakont u madam zistwar sa zako. tell 2POS wife story that monkey ‘Tell your wife the story of that monkey.’ When one of the two objects is a pronoun and the other a full NP, the order seems to be fixed as Pronoun-NP (li-so papa): (13) Seselwa a. Mari don li so papa. Mary give 3SG 3POS father ‘Mary gave it to her father.’ b. *Mari don so papa li. Mary give 3POS father 3SG ‘Mary gave it to her father.’ It is interesting to note that, when presented with the NP-Pronoun pattern (so papali), native speakers of both Morisyen and Seselwa reject sentences like (13b). If both objects are pronouns, the word order is in most instances INDIRECT-DIRECT + OBJECT (mwa-li): (14) Seselwa Don mwa li. give me it ‘Give it to me.’ This order is seen not only in imperatives (14) but also in declaratives (15): (15) Zan fin don mwa/twa li. John ASP give me/you it ‘John gave it to me /you.’ Native speakers unanimously accepted this order. When asked to judge (16), where the order of the pronouns is reversed, informants consistently rejected it as ungrammatical: (16) *Zan fin don li mwa/twa. John ASP give it me/you ‘John gave it to me/you.’


(17) *Mo vwasen ti rakont li mwa/twa my neighbor TNS tell it me/you ‘My neighbor told it to me/you.’ When both objects are lexical NPs as in semiz Klod ‘Claude’s shirt’, the order is free. (18a) is, however, potentially ambiguous, as semiz Klod could also be interpreted as “the shirt of Claude”. (18b) with Klod semiz is clearer, and seems to be the preferred order (INDIRECT-DIRECT + OBJECT ORDER): (18) Seselwa a. Mo ’n don semiz Klod. 1SG ASP give shirt Claude ‘I gave the shirt to Claude.’ b. Mo ’n don Klod semiz. 1SG ASP give Claude shirt ‘I gave this shirt to Claude.’ Compare the following sentences (18c–e): c. Mo ’n don Klod sa semiz. 1SG ASP give Claude this shirt ‘I gave this shirt to Claude.’ d. Mo ’n don Klod ban semiz. 1SG ASP give Claude PL shirt ‘I gave the shirts to Claude.’ e. Mo ’n don Klod en semiz. 1SG ASP give Claude a shirt ‘I gave a shirt to Claude.’ In (18c–e) the preferred order is also INDIRECT + DIRECT OBJECT. This suggests that with demonstrative, plural and indefinite inanimate NPs the order is restricted (indirect NP before direct NP). Other verbs (verbs of transaction) such as those classified under “give”, “future having”,² “bring”, “take”, “carry”, “throw”, “transfer” are also used with a DOC format. A closer look shows that most verbs of communication have DOC, and are thus similar to examples (19) and (20) where demande ‘ask’ is the verb used: (19) Seselwa Deman kas u misie. ask money 2POS husband ‘Ask for money from your husband.’ (20) Deman u misie kas. ask 2POS husband money ‘Ask your husband for money.’ At the same time we note two exceptions in Seselwa with the verbs of communication, demande avek ‘ask’ and dir ek ‘say with’. Bollée (1977) pointed out that these two

95

96

Dany Adone

verbs also allow PDC as illustrated by examples (21) and (22) (my examples): (21) Seselwa Mo ti deman avek Zan Klod . . . 1SG TNS ask with Jean Claude ‘I asked Jean Claude.’ (22) Mo ti dir avek Gabriel . . . 1SG TNS say with Gabriel ‘I said to Gabriel . . .’ Note that this is an important difference between Seselwa and Morisyen. These two verbs (demande avek and dir avek) have a different format in Morisyen, where it is more common to use demande in a DOC format, which I take to be the default format. The construction demande avek/ek in example (21) is used in rural Morisyen, albeit to a lesser extent. The existence of demande avek/ek/ar in Morisyen suggests that this construction may have been present in early Morisyen. A look at some old texts from Baissac’s stories compiled by Chaudenson (1981) seems to confirm this. One possible reason for the rarity of such sentences today could be French influence, and regularization in the lexicon. (23) Li dimande av coce. he ask with coachman ‘He asked the coachman.’ (Chaudenson 1981: 136) Dir avek in Morisyen today seems to be used when it is stressed that something has been said to a particular person: (24) Mo ti deman ek u talerla si u 1SG TNS ask with 2PL before if 2PL pu aste sanala. MOD buy this one ‘I asked you before if you wanted to buy this one [fish].’ (25) Mo ti deman Zan Klod . . . 1SG TNS ask Jean Claude ‘I asked Jean Claude . . .’ (26) Mo ti dir ek Gabriel, pa ek Zan. 1SG TNS say with Gabriel, NEG with John ‘I said (it) to Gabriel, not to John.’ Table 2 offers an overview of some high-frequency verbs with a DOC format in both Morisyen and Seselwa. It is interesting to note that, as illustrated in (27)–(29), adult native speakers of Morisyen and Seselwa dativize productively novel verbs with double-objects such as “to fax”,“to e-mail” and “to sms”. The same is also observed in other languages such as English, French and Dutch.

Double-object constructions in two French-based creoles Table 2. Some ditransitive verb classes and members in Morisyen and Seselwa Ditransitive verb classes

Verb members in Morisyen and Seselwa

“Give” verbs

Done ‘give’, lue ‘rent’, prete ‘lend’, ranburse ‘pay back’, vande ‘sell’, peye, ‘pay’, repeye ‘repay’

Future “having” verbs⁹

Garanti ‘guarantee’, ofer ‘offer’, dwa ‘owe’, promet ‘promise’, vote ‘elect’

“Bring“ and “take” verbs

Pran ‘take’, amene ‘bring, take’

“Send” verbs

Avoye ‘send’, poste ‘mail, post’

Verbs of instantaneous causation of ballistic motion

Avoye ‘throw’, tape ‘kick’

Transfer of message verbs

Montre ‘show, teach’, dir ‘say’, ekrir ‘write’, demande ‘ask’, lir ‘read’, site ‘cite’

Instrument of communication verbs

Kable ‘send a cable’, faxe ‘fax’, signal ‘signal’, telefone ‘phone’, emel ‘e-mail’

(27) Mo pu faxe u apre dezener. I MOD fax you after lunch ‘I will send you a fax after lunch.’ (28) Pyer ti email li en dokimen ek viris. Pyer TNS email him/her a document with virus ‘Peter e-mailed her a document with a virus (in it).’ (29) Ban zanfan sms zot kamarad sak kut. PL child sms their friend each time ‘Children sms their friends all the time (= too much).’ Both Seselwa and Morisyen have the two options for DOC and PDC. Things are somewhat complicated by the fact that there is only one preposition to express both pu benefactive and pu dative (30). Pu in Morisyen and Seselwa is not exactly the equivalent of English “to” because if it were so, then the following sentence type would be possible: (30) a. *Mari fin don liv pu Zan pu mwa. Mary ASP give book to Zan for me ‘Mary has given the book to John for me.’ Sentence (30a) is ungrammatical because pu is used with benefactive but not with dative arguments. The correct corresponding sentence is: (30) b. Mari fin don Zan liv pu mwa. Mary ASP give Zan book for me ‘Mary has given John the book for me.’

97

98

Dany Adone

In (30b) the indirect object (Zan) must come before the direct object (liv), which in turn precedes the benefactive phrase (pu mwa). In sentence (30c), the direct object can precede the indirect object: (30) c.

Mari fin don liv Zan pu mwa. Mary ASP give book Zan for me ‘Mary has given the book to John for me.’

But the sentence is then ambiguous, as it could also mean “Jean’s book”. Furthermore, sentences such as (29) with four arguments illustrate how these arguments are encoded. Both Morisyen and Seselwa may use two means to express the arguments: word order as well as a preposition. Examples (31)–(33) illustrate the various possibilities in both Morisyen and Seselwa. In (31) and (32) the DOC is used to express the ditransitive alternation seen in English. Examples (31) and (32) illustrate that the double-object frame has two forms here: NP1 V NP3 NP2, which represents the classical DOC, and NP1 V NP2 NP3 (with zero-marked preposition) in both Morisyen and Seselwa. (31) Morisyen/Seselwa Mari ti avoy Zan Klod en koli/parsel. Mari TNS send Jean Claude a parcel ‘Mary sent Jean Claude a parcel.’ (NP1 V NP3 NP2) (32) Mari ti avoy en koli/parsel Zan Klod. Mari TNS send a parcel Jean Claude ‘Mary sent a parcel to Jean Claude.’ (NP1 V NP2 NP3) (33) Mari ti avoy en koli/parsel pu Zan Klod. Mari TNS send a parcel for Jean Claude ‘Mary sent a parcel for Jean Claude.’ (NP1 V NP2 FOR NP3) In English there is an alternation between the prepositional frame NP1 V NP2 to NP3 and the double-object frame NP1 V NP3 NP2. The NP (the object of the preposition to in the prepositional frame) becomes the first object in the DOC. The DOC format in both Morisyen and Seselwa is complementary to the structure in (32). There are several ways to interpret example (32). One of them is to posit a zero-preposition marking ditransitive, thus making the NP1 V NP2 NP3 pattern a PDC. A second possibility is to view it as the result of a rightward scrambling rule (cf. Bruyn, Muysken and Verrips 2000). A third possibility is to interpret example (31) as a DOC. I here assume (31) to be derived from the French model NP1 V NP2 PREP NP3, but without the preposition à ‘to’. Thus, it is a DOC, but not the classical DOC. This development to DOC is probably related to the fact that during creolization most prepositions (including à, used to express dative arguments in French) were lost. Among the rare retained prepositions we count pu < Fr. pour ‘for’.


Levin (1993) observes two restrictions or constraints on the ditransitive alternation in English. One is called the Latinate restriction in which the morphological or phonological shape of the verb undergoes the alternation. It appears that verbs of Latin origin found in the NP1 V NP2 to NP3 frame do not have DOC.³ The second restriction involves the nature of the goal phrase. There seems to be an animacy restriction, as seen in the contrast between the English examples (34) and (35): (34) Bill sent a package to Tom/London. (35) Bill sent Tom/*London a package. The notion of animacy can be extended. As a result, example (35) is only possible if London refers to an organization or corporate body such as the “London office”. The same holds in Morisyen and Seselwa (36–7). (36) Morisyen/Seselwa Bill ti avoy en koli Zan/Lond. Bill TNS send a parcel Jean/London ‘Bill sent a parcel to Jean/London.’ (37) Bill ti avoy Zan*Lond en koli. Bill TNS send Jean/London a parcel ‘Bill sent Jean/London a parcel.’ (38) Bill ti avoy biro Lond en koli. Bill TNS send office London a parcel ‘Bill sent the London office a parcel.’ Example (37) above is normally interpreted as “Bill sent a parcel to London” (destination), and it is assumed that the benefactive argument has been mentioned before, or will be mentioned during the conversation. Example (38) shows that the addition of biro ‘office’ renders the sentence grammatical. Let us now turn to the group of verbs that take PDC. A closer look at the verbs in both Morisyen and Seselwa shows a considerable overlap between ditransitive verbs with DOC and the PDC pattern. These constructions of the type [pu X] are used for the benefactive alternation, as already hinted at above. According to Levin (1993), the benefactive alternation differs from the ditransitive alternation in that it involves the benefactive preposition for rather than the goal preposition to. In English, the benefactive alternation is found with verbs that can be broadly defined as “verbs of obtaining” or “verbs of creation”, such as bake, cook, make, draw, sing, buy, etc. (cf. Levin 1993: 48–9). Some of these verbs in Seselwa have an alternation between the [pu X] structure (see examples (39), (41), (43)) and DOC examples (40), (42), (44)). (39) Seselwa Zan ti aste en bul pu Mari. Zan TNS buy a ball for Mary ‘Jean bought a ball for Mary.’

99

200 Dany Adone

(40) Zan ti aste Mari en bul. Zan TNS buy Mary a ball ‘Jean bought Mary a ball.’ (41) Mo mama pu kwi en gato pu Rachel. 1SG mom MOD cook a cake for Rachel ‘My mom will bake a cake for Rachel.’ (42) Mo mama pu kwi Rachel en gato. 1SG mom MOD cook Rachel a cake ‘My mom will bake Rachel a cake.’ (43) Gabi pe desin en ziraf pu marmai. Gabi PROG draw a giraffe for kids ‘Gabi is drawing a giraffe for the kids.’ (44) ?Gabi pe desin marmai en ziraf. Gabi PROG draw kids a giraffe ‘Gabi is drawing the kids a giraffe.’ To recapitulate, we can say that Morisyen and Seselwa have two types of DOCs to express the ditransitive alternation seen in English. Further, a PDC accompanied by avek/ek ‘with’ also exists, but it seems to be a marked construction, limited to only a few verbs. A limited set of verbs has the benefactive alternation with DOC, and a fixed [pu X] construction. Most “verbs of obtaining” and “verbs of creation” are allowed with a PDC format with pu. At this juncture it may be beneficial to apply the NP accessibility hierarchy (Keenan and Comrie 1977) to the creole data. Similar to English, Morisyen and Seselwa allow NPs to be relativized, no matter what their grammatical relation is to the verb. Examples (45a-f) illustrate this point: (45) Seselwa/Morisyen a. Sa zom ki . . . fin kit lakaz. DET man REL ASP leave home ‘The man that . . . left home.’ b. Liv ki mo fin lir . . . book REL I ASP read ‘The book that I read . . .’ c. Fem ki mo fin avoy en rapor . . . woman REL I ASP send a repor ‘The woman I sent a report to . . .’ d. Parti kot mo fin ale . . . party where I ASP go ‘The party where I went to . . .’


Seselwa (but not in Morisyen): e. Dimun lekel mo fin al parti avek . . . person that I ASP go party with ‘The person I went to the party with . . .’ Morisyen (but not in Seselwa) f. Dimun avek lekel mo fin al parti . . . person with whom I ASP go party ‘The person I went to the party with . . .’ Except for (45e) and (45f), all examples are valid in both Morisyen and Seselwa. The Seselwa example (45e) seems to have preposition stranding like English. Morisyen (45f) follows the French pattern.⁴ Thus, on the NP accessibility hierarchy, we have the following for English, Morisyen and Seselwa: NP Accessibility Hierarchy: Subject < direct object < indirect object < oblique Two additional factors play a role in the expression of grammatical relations: animacy and definiteness. The animacy hierarchy is based on several grammatical dimensions, including person hierarchy in which the 1st person and 2nd person outrank the 3rd person. Another dimension is the NP-type of hierarchy, in which pronouns outrank common nouns, with proper names occupying an intermediate position. Finally, there is the animacy hierarchy in which humans outrank nonhuman animates, which in turn outrank inanimates. The data clearly show that in Morisyen and Seselwa the person hierarchy does not play a role. That is, 1st person and 2nd person do not outrank the 3rd person. However, the ranking order based on animacy⁵ does so, as a comparison between (46a/b) and (47a/b) shows. (46) Morisyen a. Mari pe don so papa tabak. Mary PROG give 3POS father tobacco ‘Mary gave tobacco to her father.’ b. Mari pe don tabak so papa. Mary PROG give tobacco 3POS father ‘Mary gave her father tobacco.’ (47) a.

Mari pe don so lisyen lezo. Mary PROG give 3POS dog bone ‘Mary gave her dog (a) bone.’ b. Mari pe don lezo so lisyen. Mary PROG give bone 3POS dog ‘Mary gave (a) bone to her dog.’

Example (47) provides evidence that Morisyen places animate objects after inanimate objects in the DOC. The same occurs in the Seselwa data. In Morisyen the

20

202 Dany Adone

pattern in (48a) is somewhat more common than that in (48b). Both sentences were considered to be well-formed by all of my native informants. (48) Seselwa a. Rakont u madam zistwar sa zako. tell 2POS wife story that monkey ‘Tell your wife the story of that monkey.’ Morisyen b. Rakont zistwar sa zako u madam. tell story that monkey 2POS wife ‘Tell the story of that monkey to your wife.’ Now compare (49a) and (49b): (49) a.

Mari in avoy so garson so wazen Mary ASP send 3POS boy 3POS neighbor ‘Mary sent her son her neighbor.’ b. Mari in avoy so wazen so garson Mary ASP send 3POS neighbor 3POS boy ‘Mary sent her neighbor her son.’

In this case, Morisyen native speakers tend to prefer (49b) over (49a). However, I was told that (49a), although possible, would be avoided because of its ambiguity. As such, if the pattern in (49a) is adopted, a preposition like pu would normallly be inserted (49c) so as to disambiguate the sentence. (49) c.

Mari in avoy so garson pu so wazen. Mary ASP send 3POS boy for 3POS neighbor ‘Mary sent her son for her neighbor.’

In (49c), pu could be regarded as a case marker on the Goal argument. This means that both creoles use either word order, or a preposition that is developing as a case marker to mark the Goal argument. In (50), we find the same pattern with animate and inanimate objects: (50) a.

Mari in don so wazen so loto. Mary ASP give 3POS neighbor 3POS car ‘Mary gave her neighbor (to) her car.’ b. Mari in don so loto so wazen. Mary ASP give 3POS car 3POS neighbor ‘Mary gave her car (to) her neighbor.’

Again, both sentences are considered grammatical. Native speakers did not show a preference for one construction over the other. Pragmatic knowledge most likely helps speakers interpret the sentence as ‘Mary gave the car to her neighbor’ rather than ‘her neighbor (to) her car’. In addition, we may imagine the following situation: Mary has a cat and a mouse at home. One day she decides to feed the mouse to the cat:

Double-object constructions in two French-based creoles 203

(51) a.

Mari in don so sat so suri. Mary ASP give 3POS cat 3POS mouse ‘Mary gave (to) her cat her mouse.’ b. Mari in don so suri so sat. Mari ASP give 3POS mouse 3POS cat ‘Mary gave her mouse to her cat.’

Speakers prefer (51a) over (51b). But here, it is only pragmatic knowledge that would allow one to deduce who is given as food to whom. In other words, the mouse is being given to the cat as food, and not the other way around. When one of the two objects is a pronoun and the other a full NP, the order is fixed as Pronoun-NP (li-so papa): (52) Morisyen/Seselwa Mari don li so papa. Mary give 3SG 3POS father ‘Mary gave it to her father.’ In (52) above, the NP-type hierarchy, in which a pronoun outranks a common noun, is observed. If both objects are pronouns, the order is always INDIRECT + DIRECT OBJECT (mwa-li) in both Morisyen and Seselwa, as seen in (53): (53) Seselwa Don mwa li! give me it ‘Give it to me!’ The above pattern reveals that two creoles prefer the Goal argument to precede the Theme (direct object). The two languages thus have a preference for the PRIMARY OBJECT + SECONDARY OBJECT strategy. Following Wunderlich (1997a, 1997b), I adopt the view that the Semantic Form (SF) and Conceptual Form (CF) are on two different levels in the lexicon. The SF is taken to be the computational level of grammar, and the second the level of reasoning that draws on mental operations. Since SF mediates between the conceptual structure inherent in the meaning of a word and its syntactic structure, there are some specific predictions concerning the syntactic realization of arguments. In contrast to other accounts of argument linking, which claim that the basic word order is fully determined by default linking, Wunderlich proposes a structural linking mechanism for structural arguments linked to grammatical relations (subject, direct object or indirect object). Both abstract argument positions of a lexical item and the morpho-syntactic properties of the linkers are expressed by a single feature system. This system consists of two binary features [+hr] for ‘there is a/no higher role’ and [+lr] for ‘there is a/no lower role’. These features encode the Theta Structure positions representing structural arguments as seen for one-to-three-place verbs. This linking mechanism is illustrated with the set of annotated-structures (Table 3).

204 Dany Adone Table 3. The structural linking mechanism for structural arguments

Intransitive verbs

Transitive verbs

Ditransitive verbs

_x −hr −lr NOM

_y +hr −lr ACC

_z +hr −lr ACC

_x −hr +lr NOM

_y +hr +hr DAT

_x −hr +lr NOM

An accusative argument always matches with the lowest argument of the verb. An accusative is possible only if there is a nominative. The same is valid with a structural dative. The latter is possible only if an accusative is present, as illustrated in the two German sentences in (54a = grammatical) and (54b = ungrammatical): (54) a.

Peter gab einen Kuchen. Peter gave a cake-ACC ‘Peter gave a cake.’ b. *Peter gab dem Bettler. Peter gave the beggar-DAT ‘Peter gave the beggar.’

The markedness order is then nom < acc < dat in the acc-nom patterns, and abs < erg < dat in the erg-abs patterns. The advantage of this structural linking mechanism lies in its power to lead to an automatic association between argument hierarchy and syntactic hierarchy. Depending on the type of language, the linkers available can be morphological case, agreement markers on the verb, or specific positions with respect to the verb. English uses positional linking, while the pronouns in Morisyen and Seselwa show morphological case. According to Wunderlich’s model, a language without morphological case and agreement should have only one or two structural positions. Morisyen and Seselwa confirm this, as they both display nominative and accusative (or non-nominative) case. A look at PDCs with the preposition pu ‘for’ to mark benefactivity in both Morisyen and Seselwa shows that the Recipient or Goal is marked with the preposition pu. In this type of construction the Recipient or Goal is marked by a case marker, and is equivalent to dative case in languages with morphology. The DOC2 (NP3 NP2) in Morisyen and Seselwa also shows that ditransitive verbs have the argument hierarchy Actor > Goal/Recipient > Patient. The subject or Agent is in the nominative. The argument nearest to the verb is the Goal/Recipient, and the furthest argument from the verb is the Patient. This could be explained by the fact that the patient argument is [+lr] marked. This means that it has some control, as compared to Recipient/Goal, which is [+hr]. Since the two creoles have developed neither morphological case nor agreement markers on the verb, they use positional linking.

Double-object constructions in two French-based creoles 205

Another interesting feature is the causativization of ditransitive verbs like “give” in Morisyen. The following examples illustrate how Morisyen and perhaps also Seselwa exploit the maximal structural potential positions by using positional linking, and semantic linking as an additional procedure: (55) a.

To ti don mwa kas. 2NOM TNS give 1ACC money ‘you gave me money.’ b. Misie X ti fer twa don mwa kas. Mr. X TNS make 2ACC give 1ACC money ‘Mr. X made you give me money.’

In (54a and 54b), we see the pronouns twa and mwa in the non-nominative form. If another argument is added, then it happens via a PP with a benefactive preposition: (55) c.

Misie Pol ti fer twa don mwa kas pu zot. Mr. Pol TNS make 2ACC give 1ACC money BENF they ‘Mr. Pol made you give me money for them.’

Examples (55b) and (55c) are both causativized examples of the verb done ‘to give’. In Morisyen,⁶ to (subject in 55a) becomes object twa, thus showing that it receives the same case as the Recipient/Goal mwa. If another argument is added in (55c), then it can only be expressed through a PP with a benefactive preposition, which counts as a semantic linker.

5. Concluding remarks The preceding sections have shown that creole languages have DOC associated with ditransitive verbs. Analyzing early data on Sranan and Negerhollands, Bruyn, Muysken and Verrips (2000) reach the conclusion that the DOC already appeared in the early stages of creolization. There are several arguments supporting the claim that DOC is the default structure associated with ditransitive verbs in creoles: (1) DOC’s early appearance in creole formation (Bruyn, Muysken and Verrips 2000 for Saramaccan), (2) its use in many creoles (Table 1), and (3) its early appearance in child language. DOC has received a lot of attention ever since Baker’s (1979) influential paper on the logical problem of language acquisition. Baker and others have shown how some argument-structure alternations (such as the ditransitive alternation) in English are limited in their productivity. Initially Baker assumed that children never make errors like that given in example (55), but this assumption was mistaken. (56) *John donated the library a book. Baker showed that children do not get negative evidence that would make them

206 Dany Adone

realize that (56) is ungrammatical. Other studies also showed that children do make errors with DOCs, as they use these for verbs that do not allow it. Mazurkewich and White (1984) gave examples of children’s utterances such as I’ll brush him his hair (cf. also Gropen, Pinker, Hollander, Goldberg and Wilson 1989). Pinker (1989) showed that children do retreat from overgeneralized argument-structure alternations (including the ditransitive alternation) in English. A recent study of the acquisition of DOC and PDC in Chinese interestingly confirms the surprising ease with which Chinese children acquire DOC with verbs of obtaining (e.g., “to steal”, “to buy”) (Ting Ting and Gordon 1998). In summary, these studies confirm that DOC is early in first language acquisition (cf. Gropen, Pinker, Hollander, Goldberg and Wilson 1989, Pinker 1984).⁷ Eisenbeiß (1994) confirmed that German-speaking children acquire the canonical acc-nom pattern first, and then overgeneralize it to two-place dative verbs. This means that they use accusative instead of dative with verbs such as helfen ‘to help’ and nominative instead of dative with verbs such as gefallen ‘to like’, both of which are lexical exceptions in German. Eisenbeiß’s study is especially interesting because it shows that children have a specific SF template that they use for all verbs. The over-generalization of the acc-nom in German is confirmed by creole acquisition data. Adone (2001) reports numerous DOC (NP2-NP3) patterns among groups of children between 3;5 and 4;5 years of age. Further evidence comes from Schulman, Mylander and Goldin-Meadow (2001). They argue that deaf children’s gestures provide evidence for ergative structures as these children mark patient and intransitive actor more than transitive actors. Taken together, these findings support the claim that the Patient argument is a structural argument linked to accusative, and is fully exploited by children. Evidence from creoles, sign language as well as first language acquisition provide ample evidence that DOC is one of the preferred universal structures associated with ditransitive verbs. The two creoles investigated in this paper seem to have only two structural arguments. That is, Seselwa and Morisyen — both of which lack morphological case or agreement — use mainly positional linking to mark their arguments.

Notes 1. Structural arguments refer to subject, direct object, or indirect object. See Wunderlich (1997a) for more detail. 2. In the sense of “commitments that a person will have something at some later point” (cf. Levin 1993: 46). 3. However there exist exceptions noted to this particular generalization (see Levin 1993: 48). 4. An anonymous reviewer pointed to the fact that preposition stranding is also possible in popular (spoken) French.

Double-object constructions in two French-based creoles 207 5. Human versus nonhuman, as previously seen in section 2 (examples 18). 6. The same sentence works also for Seselwa, which does not, however, have a separate 2nd person singular form (Seselwa u is 2nd person singular and plural). 7. See Snyder and Stromswold (1997), who confirm the view that children acquire DOC before PDC. 8. This table is an adapted version of Bruyn, Muysken and Verrips (2000). Both Morisyen and Kriol have been added to the table by the author. 9. In the sense of “commitments that a person will have something at some later point” (cf. Levin 1993: 46).

References Adone, D. 2001. A Cognitive Theory of Creole Genesis. Unpublished Habilitationsschrift. University of Düsseldorf. Baker, C. L. 1979. “Syntactic theory and the projection problem.” Linguistic Inquiry 10: 533– 81. Bickerton, D. 1989. “Seselwa serialization and its significance.” Journal of Pidgin and Creole Languages 4: 155–83. Bollée, A. 1977. Le créole français des Seychelles. Tübingen: Niemeyer. Bruyn, A., P. Muysken and M. Verrips2000. “Double-object constructions in the creole languages: Development and acquisition.” In Language Creation and Language Change. Creolization, Diachrony, and Development, M. DeGraff (ed.), 329–73. Cambridge, MA: MIT Press. Chaudenson, R. 1981. Textes créoles anciens (La Réunion et l’île Maurice). Hamburg: Kreolische Bibliothek. Helmut Buske. Corne, C. 1977. Seychelles Creole Grammar. Tübingen. Gunter Narr Verlag. Croft, W. 1990. Typology and Universals. Cambridge: Cambridge University Press. Eisenbeiß, S. 1994. “Auxiliaries and the acquisition of passive in German.” In Proceedings of the 25th Child Language Research Forum, E. Clark (ed.), 235–42. Stanford: CSLI Publications. Gropen, J., S. Pinker, M. Hollander, R. Goldberg and R. Wilson 1989. “The learnability and acquisition of the dative alternation in English.” Language 65: 203–57. Hale, K. L. and S. J. Keyser. 1987. “A view from the middle.” Lexicon Project Working Papers 10. Cambridge, MA: Center for Cognitive Science, MIT. Keenan, E. and B. Comrie. 1977. “Noun phrase accessibility and universal grammar.” Linguistic Inquiry 8: 63–99. Levin, B. 1993. English Verb Classes and Alternations. Chicago: University of Chicago Press. Mazurkewich, I. L. and White. 1984. “The acquisition of the dative alternation: Unlearning overgeneralizations.” Cognition 16: 261–83. Pinker, S. 1984. Language Learnability and Language Development. Cambridge, MA: Harvard University Press. Pinker, S. 1989. Learnability and Cognition: The Acquisition of Argument Structure. Cambridge, MA: MIT Press. Schulman, B. W., C. Mylander and S. Goldin-Meadow. 2001. “Ergative structure at sentence and discourse levels in a self-generated communication system.” In Proceedings of the 25th Annual Boston University Conference on Language Development (BUCLD; 25.2),

208 Dany Adone A. H.-J. Do, L. Domínguez and A. Johansen (eds.), 815–24. Somerville, MA: Cascadilla Press. Seuren, P. A. M. 1990. “Still no serials in Seselwa: A reply to “Seselwa serialization and its significance” by Derek Bickerton.” Journal of Pidgin and Creole Languages 5: 271–92. Silverstein, M. 1976. “Hierarchy of features and ergativity.” In Grammatical Categories in Australian Languages, R.M.W. Dixon (ed.), 112–71. Canberra: AIAS. Snyder, W. and K. Stromswold. 1997. “The structure and acquisition of English dative constructions.” Linguistic Inquiry 28: 218–317. Ting Ting, R. C. and P. Gordon. 1998. “The acquisition of Chinese dative construction.” In Proceedings of the 22nd Annual Boston University Conference on Language Development, A. Greenhill, M. Hughes, H. Littlefield and H. Walsh (eds.), Vol. 1: 109–20. Sommerville, MA: Cascadilla Press. Wunderlich, D. 1997a. “Cause and the structure of verbs.” Linguistic Inquiry 28: 27–68. Wunderlich, D. 1997b. “Argument extension by lexical adjunction.” Journal of Semantics 14: 95–142.

CHAPTER 9

Passive voice in Papiamento: A corpus-based study on dialectal variability Eva Martha Eckkrammer University of Salzburg

This study offers an analysis of the passive voice in Papiamento, the only Atlantic creole with a full passive. More specifically, I explore two sub-corpora of the Core Corpus of Papiamento (CCP) in order to shed light on the use of three passive markers in two dialectal varieties of the language. At the outset, I address methodological issues pertaining to corpus-based grammatical descriptions of creoles in general as well as to the compilation of the CCP. Thereafter I examine the distribution, use and status of the different passive markers applied in written Papiamento, viz. Dutch-derived wòrdu/wordo/worde, Spanish-based ser and creole-derived keda. The passive voice with the respective markers is here shown to be a frequently applied linguistic device in written genres. On the diatopic level, the three passive markers exhibit considerable dialectal variation, which goes against the general assumption that the Curaçaon variety is more Dutch influenced than the Aruban variety. In the conclusion, I sum up the most salient results and emphasize the necessity to integrate the analysis of corpus data in the current language planning process.

1. Introduction It has often been said that Papiamento and other creole languages do not have any explicit way of expressing passive voice, and that they therefore lack this grammatical feature. Authoritative publications (for example Romaine 1988) draw attention to this point, and advance it as an argument for classifying creoles as a distinct group of languages. More recently, scholars (e.g., LaCharité and Wellington 1999 for Jamaican creole) have addressed the discursive (rather than morphosyntactic) mechanisms on which creoles draw to express passive voice. Papiamento has long been known as “the only Atlantic Creole known to us with a full passive” (Markey and Fodale 1983). Exactly how this creole expresses the passive voice has, however, never been examined closely, in spite of several excellent studies on its grammatical features (cf. Maurer 1988 for the verbal system; or Muller 1989 for syntax).

20

Eva Martha Eckkrammer

This paper offers an analysis of the passive voice in Papiamento. More specifically, I examine the occurrences of three passive markers that are used in two different varieties of the creole, viz. Aruban Papiamento and Bonairian/Curaçaoan Papiamento, in an attempt to shed light on diatopic preferences of these markers. The passive markers examined in this study are (1) the Dutch-derived wòrdu/wordo/ worde-passive, (2) the Spanish-derived ser-passive¹ and (3) the keda-passive. The following examples from the Core Corpus of Papiamento (henceforth CCP) illustrate the three types:² (1) E hòmber a wòrdu detené ayera. the man PFM PSM arrest yesterday ‘The man was arrested yesterday.’ (2) kantidat grandi di karni a ser tirá na landfill. quantity big of meat PFM PSM throw in landfill ‘a large quantity of meat was thrown into the landfill.’ (3) esaki lo keda tratá awe mainta. this FM PSM treat today morning ‘this (topic) will be addressed this morning.’ The study also discusses how the Papiamento passive was described in early as well as recent sources (e.g., Lenz 1927, Martinus 1997), and examines the present-day status of the three passive markers in written genres such as articles in the press, administrative, legal and religious documents, and literary prose. Data for this paper have been taken from the digitized CCP, which is currently being compiled at the University of Salzburg. The paper is divided into four parts. The first part addresses methodological issues pertaining to corpus-based grammatical description of creoles in general (Section 2). The second part examines the distribution, use and status of the different passive markers applied in written Papiamento (Section 3). This analytical part of the paper is divided into three subsections. The first (§3.1) provides the framework for the discussion of the data by referring to previous publications on the topic. The second part (§3.2) describes the corpora used for the study, and the third (§3.3) then states the most salient results. In Section 4, I discuss the results and correlate them with the opinions of local linguists. Section 5 sums up the outcome of the study and relates the findings to current language planning.

2. Corpus-based grammatical description How do we determine how a language, still lacking a detailed grammatical description, truly works? Historically, Papiamento has been illustrated on the basis of limited sets of spoken and/or written data, i.e., with one informant and a couple of reference books in the early and adept description by Lenz (1927),³ or several recordings

Passive voice in Papiamento

in the pioneering study on phonological variation by Andersen (1974). Sometimes the source of the primary data is omitted, which at times makes it difficult to evaluate the usefulness of earlier works (cf. early grammars and text books like Evertsz 1898), or even contemporary linguistic descriptions (cf. Dijkhoff 1991 and 1993, Munteanu 1996). However, the growing presence of Papiamento in schools, along with an increasing usage of the creole in official settings, calls for an extensive reference grammar. Such a grammar has to be acceptable to all speakers and learners of the creole. Acceptance represents a crucial issue because Papiamento is not yet (fully) standardized, and displays substantial dialectal variation. Additionally, the creole has separate spelling systems, both of which were made official after the political disputes between Aruba and Curaçao in the 1970s. Attempts to standardize Papiamento began in the early 1990s, but these originally promising attempts are currently encountering numerous obstacles (Eckkrammer 1999). It may well be that an authoritative reference grammar of Papiamento can only succeed if (a) it is descriptive, (b) based on a large reference corpus, and (c) obtained from written and spoken Papiamento as found in all relevant territories, viz. Aruba, Bonaire, Curaçao and the Netherlands. The CCP contributes to these goals. The CCP is a digitized corpus that covers Papiamento texts between 1825 and today. Designed as a dynamic compilation of written and spoken Papiamento, it includes a wide array of genres and registers from the first translated Bible chapters to modern genres, such as online chats or contributions to “virtual” discussions. The corpus is intended to mirror the actual usage — the parole — of Papiamento. So far, the data base consists of approximately 600,000 words of predominantly written Papiamento. The genres thus far included are for the most part prayers, bible chapters, poetry, medical information texts, administrative texts, political manifestos, articles from the daily press, contributions to virtual guest books, song texts and, to a limited extent, academic texts. The corpus texts are coded according to (a) “applied mode” (written, spoken, etc.),“medium mode” (face-to-face-communication, radio, television, newspaper etc.), genre (private everyday conversations, radio news, interviews, news articles, poetry, religious prose, legal documents etc.), (b) register (formal, informal etc.), and (c) the year of production/publication. The corpus thus makes it possible to compile individual sub-corpora, which can then be searched for specific grammatical features, registers, language change, and so forth.

3. Analysis of the passive voice 3.1 Preliminary remarks By passive voice I understand a verbal category which, from a semantic point of view, indicates that the subject is the patient or recipient of the denoted action. This definition implies that the choice of a passive over an active clause is not functionally

2

22


restricted to the expression of an action without reference to the person or entity performing the action (the subjects often remain unspecified but logical). In addition, the use of a passive structure permits the focus or promotion of specific objects/subjects. Most European languages possess passive voice, and several morphological means to express a passive commonly coexist (this is the case, for instance, in the Scandinavian languages). Generally speaking, the active voice is deemed to be shorter, more direct, and more dynamic than the passive voice. If, however, (a) the agent of the action is unimportant, unknown or common knowledge (and therefore already an active “topic”), or (b) the register or genre used traditionally give preference to the passive over the active voice, then passive voice is frequently used. In their concise grammatical description of Papiamento, Kouwenberg and Murray state that “Papiamentu is unique among Caribbean Creoles in having a passive construction similar to that found in the European lexifiers” (1994: 37). As stated in Section 1 above, the creole essentially possesses three different passive constructions. Bickerton (1981: 71–2) mentions this fact, and offers references to other creoles with passive markers (e.g., Mauritian creole and Seychellois gay, or the get passive in English creoles). He states, however, that these passive constructions are marginal or “relatively recent superstrate borrowings” (1981: 172). In early written sources of Papiamento (19th and early 20th century), the passive voice is frequently used in written genres. Passive constructions, predominantly the Dutch-derived wòrdu/worde/wordo passives, are remarkably common in the early translations of the Bible. Lenz (1927: 130) deems these passives to be characteristic of Dutch-influenced Papiamento. He explains Papiamento’s preference for the Dutch model by pointing to the ambiguity of the lexical item ta,⁴ and to the absence, in Spanish, of a passive equivalent of Dutch worden, German werden or French devenir (Lenz 1927: 130). The most salient part of his findings on the passive voice in Papiamento (Lenz 1927 often referring to Evertsz 1898 and Hoyer 1918) is that Lenz mentions bira (derived from Ptg. virar) as a passive marker or auxiliary. But there is no firm evidence for this passive use of bira.⁵ It is noteworthy that in the section on the passive, Lenz (1927: 129–33) omits mentioning keda as a passive marker. Moreover, he does qualify ser as a castellanismo that, as he says (p. 132), rarely appears in verbal functions since it is predominantly applied as a noun (see also footnote 1). Goilo (1972), in the Spanish version of his textbook for Papiamento learners, distinguishes between progressive and conclusive passive voice (e carta ta ser skirbí ‘the letter is being written’ versus e carta ta skirbí ‘the letter is written’), but does not mention the Dutch type with wòrdu. The Dutch version (Goilo 1968: 114) of the same textbook is remarkable in that it explains both the Dutch- and the Spanishderived types of passive. It is, therefore, very likely that Goilo consciously omitted the Dutch-derived type in the Spanish version, thus rendering Papiamento more accessible to Spanish speakers. Maurer’s meticulous study (1988) of the Papiamento verbal system mentions three types of passive constructions: (1) wòrdu (for Curaçao and Bonaire) with the variant wordo (for Aruba), (2) ser and (3) keda (for all territories). Maurer consid-


ers keda to be a recently introduced passive construction — one that is functionally equivalent to the two other types.⁶ The latter point of view contradicts Munteanu (1996: 344), who assumes that the keda marker is linked to specific tense, mode and aspect sequences. Nevertheless, in a preliminary corpus-based study, I noted a high frequency of the keda passive in subjunctive mode constructions (cf. Eckkrammer 2002). Dijkhoff (1993: 19) states that passives are a “rather typical construction”. However, she equally considers passives with one of the three different markers (ser, wòrdu/wordo/worde and keda) to be highly marked passive constructions, which usually have unspecified logical subjects. Her statement shows that she also believes that Papiamento generally prefers active over passive voice. This is certainly correct for spontaneous informal speech. For written genres, however, further study is needed.

3.2 Corpora description A preliminary study based on a small corpus (Eckkrammer 2002) has already revealed useful insights, which are now reassessed on a larger set of data. The present analysis is based on two special corpora extracted from the CCP, each limited to a variety of written genres, but not to a single genre, as this approach would have required a meticulous hierarchical functional delimitation of the respective genres (cf. Dressler and Eckkrammer 2001).⁷ The data were divided up into two separate corpora: the first consists of approximately 140,000 words, and is made up of written text from Curaçao and Bonaire (henceforth CB-variety); the second contains written language from Aruba (henceforth A-variety) and consists of approximately 125,000 words. The genres of these written texts are to a large extent articles from the daily press (La Prensa, Bon Dia Aruba, Diario di Aruba, and so forth). Each corpus is complemented by a broad variety of other written genres, including poems, prayers, advertisements, administrative texts, political party platforms, medical information leaflets, etc. The insular provenience of a given text can usually be identified on the basis of its orthographic tradition.⁸

3.3 Analysis of dialectal variation The analysis of the corpus gives ample evidence that both varieties of the creole frequently make use of passive constructions, with markers derived from Dutch (wòrdu/worde/wordo), Spanish (ser) and creole (keda). In the 265,000 tokens analyzed, there are 1,290 occurrences of the explicit passive constructions with one of these three markers.⁹ The Dutch-derived marker wòrdu applied in the CB-variety is frequently written without the compulsory accent (*wordu) or with incorrect accentuation (*wordú).¹⁰ Aruban wordo/worde requires accent, but is occasionally spelt incorrectly as wordó.

23

24


An examination of the passive without attention to dialectal variation reveals that the Dutch-derived wòrdu/wordo/worde type exhibits a relative frequency very close to that of the lexically Spanish-derived ser type (495 items for the former, versus 494 items for the latter, which amounts to, respectively, 38.4% and 38.3% of all recorded passive constructions). Remarkably, the creole-derived passive marker keda, though less frequent than the aforementioned passive constructions, is far from negligible in terms of its frequency of usage (301 items in all texts, or 23.3%). As regards dialectal variation, an initial finding is that Aruban Papiamento seems to use passive constructions more frequently than the CB variety. If we mathematically equalize the slight difference in corpus size, we note that out of 100 recorded cases 55.7 items belong to the A-variety, and the rest to the CB variety. Basic diatopic differences can be observed with regard to the preference of the passive markers in the different regional varieties. The Dutch-derived type with wòrdu/wordo/worde + PAST PARTICIPLE is most popular in Aruba, where it ranks highest among passive constructions (46.2%). This is surprising since Aruban Papiamento tends to prefer Spanish borrowings, which means that in this case the ser marker would be expected. The Dutch-derived pattern is less frequent in the CB-corpus data (29.6%). The CB variety operates with the ser passive in 31.1% of the passive occurrences, which means that the ser type is slightly more frequent than the wòrdu type in the CB variety. In Aruban texts, the Spanish-derived ser marker has a higher rate of occurrence, and is comparable in terms of frequency to the Dutch-derived construction (44.7%). The CB variety operates with the ser passive in 31.1% of the passive occurrences, which means that the ser type is slightly more frequent than the wòrdu type in the CB variety. Yet, the most striking result is the use of the creole-derived keda construction: in the CB variety the keda marker is used with the majority of passive constructions (39.3%), and generally does not convey a specific aspect. Conversely, the A variety uses it only rarely (9.1% of all occurrences), and when it does so, a durative aspect seems to accompany it (4–6). (4) E auto cu a keda confisca. the car which PFM PSAM confiscate ‘The car which was (and remained) confiscated.’ (5) El a keda deteni como complice. he PFM PSAM arrest as accomplice ‘He was (and remained) arrested for complicity.’ (6) caminda patient nan ta keda interna. where patient PM PRM PSAM hospitalize ‘where patients are (and remain) hospitalized.’ An explicitly expressed agent is rare in both varieties, and there are no striking differences in the frequency of the Dutch-derived agent indicator dor di ‘by’, whose use is often criticized by prescriptive grammarians (cf. Section 4). Similarly, the prepos-


ition pa ‘by’ (deriving from Ptg. or Span. para) is applied rarely to indicate the agent (example 7). (7) E hóben a ser yudá pa algun konosí. the young man PFM PSM help by some friend ‘The young man was helped by some friend.’ In Aruban orthography, generally no accent marks are used to indicate the changed accentuation when a verb is turned into the Past Participle (PastPart; see examples 4–6). Following the norms of the CB-variety, this shift of stress to the last syllable has to be indicated by an accent mark (see yudá in example 7).¹¹ Nevertheless, the accent mark is frequently lacking in the CB-variety texts. Only in the case of keda-passives is the number of PastPart items without the respective accent mark extremely low.¹² As we have seen, in written Papiamento the passive voice plays an important role in clauses that (a) either have an unimportant, unknown or common-knowledge subject, or (b) omit the agent on purpose, and foreground the patient or recipient of the conveyed action as the sentential subject.¹³

4. Discussion The analysis reveals that Aruban Papiamento prefers the “traditional” wòrdu and ser passives that draw on European models. In my corpus, the wòrdu type is used even more frequently than the ser-type, which contradicts the often-stated preference of Spanish borrowings in the A variety of the creole. The keda passive generally plays a less important role in Aruban texts, and closer observation reveals that its use generally depends on the durative aspect. More than half of all my data samples convey an action that (a) has (or had) been completed, (b) remains completed, and/or (c) shows a result (cf. keda in the role of passive and aspect marker as shown in examples 4–6). Nevertheless, the use of keda is far from being obligatory when durative aspect must be expressed, since there are numerous examples of wòrdu and ser in similar durative constructions. In the A variety, the choice between wordo/worde and ser seems free or, perhaps, tied to style. As for contextual preferences, it is, for instance, noteworthy that usa ‘to use, apply’ is more frequently used with the passive marker wordo/worde (28 items) than ser (5 items).¹⁴ Contrary to the distribution of passives in the A variety, the CB variety has a tripartite system of passives. In this variety the three markers have a roughly equivalent frequency. The Dutch derived wòrdu type is less frequent than the Spanish- and creole-influenced constructions. This strongly suggests that the Curaçaon variety is less Dutch-influenced than is generally believed (cf. for instance the extracts from Aruban news articles cited in Maduro 1991: 69f). The dominance of the keda passive in the CB variety could, however, also be interpreted as a consequence of a strongly negative attitude towards the use of the other two passive markers, and in particular the wòrdu passive.

25

26


If one compares the results of the foregoing analysis with opinions of local experts on Papiamento, one notes a wide range of viewpoints, as well as a continuum in terms of degree of prescriptiveness. Maduro, a Papiamento specialist, who has long been known as the (purist) authority with regard to the use of the creole, favors the strict avoidance of passives, and thus stands at one extreme of the continuum. He suggests the use of periphrasis to indicate that the subject is the patient of the denoted action, and explicitly criticizes the use of prepositional dor di and dor di ku, frequently applied to indicate the subject of the denoted action in passive constructions (he refers to them as linguistic monstruonan ‘monsters’).¹⁵ To exemplify the correct usage of Papiamento in this context, he rewrites clauses taken from different publications (dominantly press articles; cf. Maduro 1992: 65 and 1991: 32).¹⁶ His arguments against the passive markers wòrdu/wordo/worde and ser are based on his view of the passive voice as being “un-creole”. Maduro subjects the Dutch-like model to fervent criticism when he states that worde hañá — C. a worde hañá morto na Amsterdam — E redaktor akí ta skirbi barbaridat akí den su korant; ma na su kas e ta bisa: nan a haña C. morto na Amsterdam. ‘worde hañá — C. was found dead in Amsterdam — The correspondent writes this barbarity here in his newspaper, but at home he says: they found C. dead in Amsterdam.’ (Maduro 1991: 162, my translation).

At the other end of the continuum, local linguists like Muller (1982) are far less prescriptive, as they do not object to the use of passive constructions with wòrdu/ wordo/worde and ser. Muller simply suggests a moderate use. Taken together, these recommendations against the use of wòrdu passives may have resulted in a shift to the keda-passive. It should be kept in mind, however, that, as stated earlier, the application of the passive is conditioned by both genre and register, and that the written genre of the newspaper article favors the occurrence of passive constructions. This strong presence of passive constructions in certain genres might be due, on the one hand, to the large amount of translations from Dutch and English into Papiamento and, on the other hand, to the multilingualism of many writers, who are often skilled writers in Dutch, Papiamento, English, and Spanish. This multilingualism may induce writers to draw on morphosyntactic devices from Spanish, Dutch and English, thus introducing into Papiamento passive constructions that would otherwise have a far lower frequency of use.

5. Conclusions This paper has shed light on the dialectal variation of the passive voice in different varieties of Papiamento. Three common types of passive constructions have here been shown to be deeply rooted in the creole, particularly in its written registers. Based on corpus data, this study has revealed clear-cut dialectal preferences for cer-


tain passive markers. These findings have at times stood in stark contrast to earlier assertions on the subject. Back in 1983, in the first published collection of papers of a conference on Papiamento, Andersen concluded that detailed linguistic research is needed on variational aspects of Papiamento “so that educational and language planning decisions about the forms and uses of language in the public domain are an accurate reflection of the complex society they will serve” (1983: 81).¹⁷ At present, and in view of the results presented in this paper, this statement can only be confirmed and reiterated. Especially in light of the current socio-linguistic situation — in which discussions on language policy figure prominently — additional research is needed that centers on spoken as well as written forms of the creole (in all its registers and dialectal variations). As the CCP continues to be expanded substantially, it may guide us further in establishing norms that in the end, it is hoped, will be satisfactory to all of the Papiamento-speaking territories.

Notes 1. It is interesting to observe that ser is also found in other functions (cf. Martinus 1997), e.g., as a lexicalized noun in expressions such as ser humano ‘be human’ or ser feminino ‘be female’. Similar uses can be found in the Cape Verdean creole, the creole of Guinea Bissau, and Palenquero. 2. The glosses applied in examples (1)–(7) are the following: PFM = perfect tense marker, PM = plural marker, PRM = present tense marker, PSM = passive marker, PSAM = passive and aspect marker, FM = future tense marker. 3. His observations are primarily based on a corpus of written and spoken texts produced by his informant, the Curaçaon cook Natividad Sillie. 4. Ta has a variety of grammatical functions, e.g., it is a copula and a present-tense marker. Lenz (1927) was unaware that these two ta are differentiated by tone, which makes them separate lexical items. 5. His examples, quoted hereafter, cannot be considered instances of passive voice: Ora nan a mira e strea, nan a bira terribel contentu (‘When they saw the star, they were extremely joyful’, Mat. 2, 10), E lo bira akusá (‘He will be accused’, La Cruz), Satanas a bira saká di su reino (‘Satan has been taken from his kingdom’, La Cruz) (Lenz 1927: 132, my translations). A corpus check shows clearly that bira does not have this grammatical role. The most frequent verbal collocation is bira konosí/conosí ‘to become known’, as in e autor a bira conoci dor di su canticanan cu letranan cinico ‘the author became known by his songs full of cynical lyrics’. 6. Contrary to wòrdu/wordo and ser, which do not perform other verbal functions than that of an auxiliary, keda also continues to be used as an independent verb (“to stay”) or as a copula. 7. Besides, a restriction to a single genre or register would have greatly limited the value of the study. 8. This implies that, if a newspaper is published for all three Papiamento-speaking islands, the orthography applied by the local news editors diverges. Books and leaflets are usually published in the local variety.

27

28

Eva Martha Eckkrammer 9. On average, every 205th word is a passive marker. 10. The accent mark in wòrdu indicates “open vowel”. 11. The tone remains the same, viz. low-high, i.e. komersiantenan ta bende hopi kos ‘merchants sell a lot of things’ vs. hopi kos ta ser/wòrdu/keda bendé (dor di/pa komersiantenan) ‘many things are sold (by merchants)’. 12. PPs are frequently not only applied verbally, but turned into adjectives. Martinus relates this use to the “profusive use of the passive with unspecified subject” (1997: 87). 13. In some clauses the choice of register seems to dictate whether preference is given to a passive construction. This is particularly the case in administrative and legal documents, as well as in medical texts. 14. As regards the verb haci ‘to make’, the distribution is more balanced (17 occurences with wordo/worde versus 14 with ser). 15. Maduro (1992, 65) similarly criticizes the use of debí na ‘because of ’, pa motibo di ‘by reason of, because of ’ or pa medio di ‘by, through the means of ’ in other mostly Dutch-influenced constructions. 16. For instance, E obra a ser realisá dor di Alemán is rewritten as E obra ku Alemán a realisá (o: produsí) ‘The work/piece was produced by Alemán’ _ ‘The work/piece that Alemán has produced’ or E atakantenan a ser mirá dor di un guardia becomes Un guardia a mira e atakantenan ‘The aggressors were observed/seen by a guard’ _ ‘A guard observed/saw the aggressors’. Maduro concludes his criticism by stating that “esta un alivio bo ta sinti ora bo kambia e nònsèns ku táki riba pa loke ta papiamentu outéntiko” (1992: 65) (‘what a relief one feels when one changes the nonsense cited above to authentic Papiamento’). 17. Andersen (1983) drew on results from his fieldwork for his doctoral thesis (Andersen 1974).

References Andersen, R.W. 1974. Nativization and Hispanization in the Papiamentu of Curaçao N. A.: PhD Thesis. University of Texas at Austin. Andersen, R.W. 1983. “One norm or several? Linguistic variation in Papiamentu and its role in language planning.” In Papiamentu. Problems and Possibilities, 63–84. Zutphen: De Walburg Pers. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Dijkhoff, M. B. 1991. Gramátika modèrno di Papiamentu. Curaçao: ILA/KOMAPA. Dijkhoff, M. B. 1993. Papiamentu Word Formation. A Case Study of Complex Nouns and their Relation to Phrases and Clauses. Amsterdam: Proefschrift. Dressler, W. U. and Eckkrammer, E. M. 2001. “Functional analysis in contrastive textology.” Logos and Language 2: 25–43. Eckkrammer, E. M. 1999. “The standardisation of Papiamentu: New trends, problems and perspectives.” In Les langues minoritaires en contexte. Minderheitensprachen im Kontext. Vol I. Les langues minoritaires entre diversité et standardisation. Minderheitensprachen zwischen Vielfalt und Standardisierung. A.-A. Dazzi Gross and L. Mondada (eds.), 59–74. Neuchâtel: Institut de Linguistique de l’Université de Neuchâtel. Eckkrammer, E. M. 2002.“Vielfalt der Modelle: Aktuelle Entwicklungen in der Morphosyntax des Papiamentu im Spannungsfeld zwischen Spanisch, Niederländisch und Kreolisch.”

Passive voice in Papiamento In Romanische Sprachen in Amerika. Festschrift für Hans Dieter Paufler zum 65.Geburtstag. J. Klare and K. Störl-Stroyny (eds.), 365–77. Frankfurt a.M.: Lang. Evertsz, J. N. 1898. Compendio de la gramática del Papiamento, ó sea método para aprender á hablarlo y á escribirlo en corto tiempo. Curaçao: A. Bethencourt e Hijos. Goilo, E. 1968. Papiamentu Leerboek. Aruba: De Wit. Goilo, E. 1972. Hablemos papiamentu. Aruba : De Wit. Hoyer, W. M. 1918. Woordenlijst en samenspraak; Holandsch-Papiamentsch-Spaansch. Curaçao: A. Bethencourt e Hijos. Kouwenberg, S. and E. Murray, 1994. Papiamentu. München — Newcastle: Lincom Europa (= Languages of the World Materials 83). LaCharité, D. and J. Wellington, 1999. “Passive in Jamaican Creole: Phonetically empty but syntactically active.” Journal of Pidgin and Creole Languages 14: 259–84. Lenz, R. 1927. El papiamento la lengua criolla de Curazao (la gramática más sencilla). Santiago de Chile: Balcells and Co. Maduro, A. J. 1991. Papiamentu. Indagando i ilustrando. Curaçao: Drukkerij Scherpenheuvel. Maduro, A. J. 1992. Papiamentu di un palu pa otro. Estudio i uso korekto di su lenga materno ta promové sivismo i patriotismo i ta enaltesé un pueblo. Curaçao: S. Joubert/Drukkerij Scherpenheuvel. Markey, T. L. and P. Fodale, 1983. “Lexical diathesis, focal shifts and Passivization: The creole voice.” English Word-Wide 4: 69–84. Martinus, F. 1997. A Kiss of a Slave. Papiamentu’s West-African Connections. Curaçao: De Curaçaosche Courant. Maurer, Ph. 1988. Les modifications temporelles et modales du verbe dans le papiamento de Curaçao (Antilles Néerlandaises). Hamburg: Buske. Muller, E. 1982. Papia Kòrsou. Curaçao: IPEP. Muller, E. 1989. Inleiding tot de syntaxis van het Papiamentu. Curaçao: self-edited. Munteanu, D. 1996. El papiamento, lengua criolla hispánica. Madrid: Gredos. Romaine, S. 1988. Pidgin and Creole Languages. London: Longman.

29

CHAPTER 10

Tone assignment on lexical items of English and African origin in Krio Malcolm Awadajin Finney

California State University, Long Beach

The nature of tone marking on words of English origin in English-based creoles is a highly debated issue. In creoles defined by some as pitch-accent languages, high tone in words that are derived from stress (accent) languages generally coincides with primary stress. I adopt herein the position that Krio is a tonal language, rather than a pitch-accent language. Tone is generally specified in the lexicon, particularly for lexical items of African origin. Tone assignment on disyllabic lexical items of English origin is unpredictable, but it is predictable in a limited set of polysyllabic lexical items: In the latter, high tone corresponds with the primary or secondary stress that is closest to the end of the word. Finally, I propose a tonal rule of high tone deletion and spreading of low tone on the initial components of compounds of English origin.

1. Introduction Factors contributing to the expansion of creole grammars have been the subject of much debate, with researchers adopting one of two polar views. The Universalist view argues for a prototypical creole grammar that accounts for grammar and lexicon primarily through the application of universal rules. Such grammars fail to incorporate any reference to tonal patterns (Bickerton 1975, 1977, 1988; McWhorter 1998; Seuren and Wekker 1986). The Substratist view, on the other hand, maintains that West African substrate languages (especially those belonging to the Kwa language subgroup) have predominantly influenced the suprasegmental, grammatical and lexical properties of creoles (particularly certain Atlantic varieties). There is thus no consensus on the extent to which properties of creoles (including Krio) have been influenced by properties of substratal languages, and in particular by suprasegmental features such as tone. Suprasegmental features of lexical items constitute a major area of controversy in Krio, an Atlantic creole and the lingua franca of Sierra Leone. Strevens asserts that the pronunciation of Krio speakers in Freetown, Sierra Leone, is closest to Standard English, possessing “a system of stress and intonation of the same nature as that of

222

Malcolm Awadajin Finney

Received Pronunciation” (1966: 116). He claims that in spite of a large number of Yoruba borrowings, the tonal system of Yoruba has disappeared from Krio, which now exhibits a stress and intonation pattern not unlike that of Received Pronunciation — Standard British English. A contrary position maintains that Krio exhibits the characteristics of a tonal language, and that the most significant phonological feature distinguishing Krio from English is tone, which is inherent and generally unpredictable for lexical items, including English borrowings (Berry 1961, 1970a, 1970b; Fyle and Jones 1980; Johnson 1974; Jones 1971). I support the view that Krio is a tonal language, and that tone assignment is unpredictable on lexical items of African origin. Tone is also unpredictable in disyllabic words of English origin. English borrowings with three or more syllables are, however, subject to certain predictable tonal rules, with high (H) tone generally corresponding with the primary or secondary stress that occurs in the ultimate or penultimate syllable. On the other hand, Low (L) tone does not necessarily correlate with unstressed syllables in English, as is proposed for pitch-accent languages (Berry 1959, Bickerton 1975, Devonish 1989, Hall 1966, Holder 1991, Schneider 1966, Strevens 1966). In order to explain Krio suprasegmental patterns, I further propose tonal rules of H deletion and L spreading on initial segments of compounds of English origin.

2. Stress, tone, and pitch-accent languages 2.1 Characteristics of stress, tone, and pitch-accent languages Languages are classified under one of three suprasegmental categories: stress, tone, and pitch-accent. Stress languages assign primary stress to only one syllable in a word. The vowel of this syllable is pronounced with more intensity (longer, louder, stronger pitch) compared to vowels contained in other syllables. Vowels in unstressed syllables are usually reduced. English, the lexifier language of Krio and of other creoles, is categorized as a stress language (Goldsmith 1990, Poldauf 1984). A tonal language has been defined as having “lexically significant, contrastive, but relative pitch on each syllable” (Pike 1948: 3), which means that pitch (tone) may be used to distinguish between meanings of words, and every syllable carries at least one significant pitch unit. However, this characterization does not apply to all tonal languages. In some cases, tones are not freely assigned to syllables. Also, there may be restrictions on the combination of tones (i.e., the number of tonal patterns), as is the case for Mande languages. Such languages are generally categorized as Restricted Tone (Tone Harmony) languages. In most lexical tone languages, including a number of West African languages, pitch is unpredictable and specified in the lexicon (Leben 1973, Woo 1972). It is used contrastively to minimally distinguish one lexical item from another, and/or distinguish one grammatical category from another. Languages vary in the number and kinds of tones they allow.

Tone assignment on lexical items of English and African origin in Krio

Pitch-accent languages are intermediate between stress and tone languages. The designation of some creoles as pitch-accented is largely due to the fact that, although creoles tend to be tonal, a substantial portion of the lexicon is borrowed from a source language that is generally a stress language (Alleyne 1980, Bickerton 1975, Holder 1991). The assumption is that creoles substitute H tone for primary stress for lexical items derived from the source language, while unstressed syllables are automatically assigned L tone: “The earliest forms of all Afro-American dialects [i.e. creoles] [. . .] had a characteristic high pitch, which generally coincided with the main stress of cognate words in English” (Alleyne 1980: 73). In pitch-accent languages, tone is not inherent in lexical entries, but it is assigned post-lexically (Hyman 1978, Pulleyblank 1986), as is also the case in stress languages.

2.2 Pidgins and creoles as pitch-accented languages Suprasegmentals have been rarely studied in pidgins and creoles, and there is a striking absence of detailed analyses of the interaction of stress and pitch in the pronunciation of lexical items derived from different sources. Alleyne proposes that although tone became distinctive in creoles with the introduction of vocabulary from the substratal language, tone became “less and less prominent as a distinctive feature” as English continued to exert its influence on Krio, Saramacan, Sranan, as well as on Jamaican Creole, and Guyanese Creole (Alleyne 1980: 71, 73). Devonish (1989) similarly claims that, in Djuka and Saramaccan, H tones in words borrowed or derived from stress languages generally coincide with the location of primary stress. Berry (1959) and Strevens (1966) make a similar assertion for Krio. Hall likewise suggests that stress position in Krio “is automatic, as all words are stressed on the first syllable except for recent loan words from English” (Hall 1966: 34–5). He attributes this to the fact that primary stress in English (and other Germanic languages) is generally assigned to initial syllables. He further claims that Haitian Creole assigns stress at the end of the word, as is the case in its source language, French. Schneider (1966) echoes a similar view in his analysis of Cameroonian Pidgin English. Bickerton (1975) and Holder (1991) support the proposal of an interaction of stress and pitch in Guyanese Creole. Holder maintains that Guyanese Creole is a pitch-accent language with surface pitch generally corresponding to stress in Standard English. He proposes that suprasegmentals are generally non-distinctive in Guyanese Creole, and that accent placement is largely predictable from syllabic and grammatical information. On the other hand, Carter (1987: 227) identifies a number of mismatches between stress pattern and pitch assignment on some English loan words in Guyanese Creole. She points out that schwas as well as syllabic liquids and nasals, receive primary stress in Guyanese Creole, although they are invariably unstressed in English. Holder acknowledges these as exceptions, but maintains that pitch remains largely predictable and that there is a correlation between stress in English and pitch in Guyanese Creole for the vast majority of the Guyanese lexicon.

223

224 Malcolm Awadajin Finney

The designation of creoles as either pitch-accent or tonal languages is still debated, and the lack of consensus is evident in discussions of Krio suprasegmentals. The primary question addressed is whether and to which extent tone is predictable. The following issues are examined: (i) Lexical specification of tone in words of English and African origins. (ii) The predictability of tone across different lexical and grammatical categories. (iii) The influence of the source language (English) on predictability.

2.3 Is Krio a pitch-accent or a tonal language? Early proposals sparked the debate over the status of Krio as a pitch-accented or a tonal language (Berry 1959, Strevens 1966). According to Berry, the placement of H tone in Krio matches the placement of primary stress in English. He claims that there is a correlation between Krio H tone and English primary stress, and between Krio L tone and British English unstressed syllables. Strevens in his discussion of the pronunciation of English in West Africa also asserts that the English pronunciation of native Krio speakers in Freetown, Sierra Leone, is similar to Standard English pronunciation because of the close relation between the phonological patterns of the two languages. Strevens maintains that Krio has derived from RP its stress and intonation system, as well as many phonological properties, and that therefore, native Krio speakers do not acquire English as a foreign language. However, the pattern attributed to pitch-accent languages — H tone coincides with main stress, and L tone with the unstressed syllables of cognate words in the source language — is not borne out in lexical items borrowed into Krio from English. Stress placement is not always predictable from syllabic information, and a number of mismatches do exist between the stress pattern of English and tone assignment on some English loan words in Krio. There are a number of English borrowings in Krio in which syllables with H and L tones do not necessarily correspond to, respectively, stressed and unstressed syllables in English. As a result, a number of linguists — mostly native Krio-speaking linguists — have maintained that Krio is a tonal language, and that tone assignment is generally unpredictable and inherent in lexical entries (Berry 1961, 1970a, 1970b; Fyle and Jones 1980; Jones 1971; Johnson 1974). According to Pike’s (1948) criteria outlined in Section 2.1., Krio can be categorized as a tonal language. Tone is specified in the lexicon and is usually unpredictable. There is no correlation between syllable structure, or vowel quality and tone marking. In addition, tone can be used contrastively in minimally distinguishing between lexical items of both African and English origins.

2.4 The influence of West African languages on Krio Krio has a rich history of contact with other West African languages, dating as far back as the abolition of the trans-Atlantic slave trade, and the resettlement of some


of the freed slaves, including recaptives (those captured in the High Seas) in Freetown, Sierra Leone. These slaves were primarily from West Africa, and brought along a variety of West African languages, Yoruba being the most influential. The language and traditions of the Yoruba settlers have had a strong influence on the language, social life and customs of Krio speakers in Freetown. A large number of lexical borrowings in Krio from West African languages, particularly Yoruba, are well documented (Bradshaw 1966, Fyle and Jones 1980, Jones 1971). Yoruba is second only to English as the largest contributor to the Krio lexicon (Bradshaw 1966, Fyle 1994). Lexical items of African origin are usually borrowed along with their tonal specifications, with occasional minor phonemic variations. H and L tone marking generally remains unchanged in Krio. Bradshaw (1966) contains an impressive list of Krio lexical items borrowed from Yoruba, which supports the correspondence of H and L tones between the Yoruba words and their borrowed reflexes in Krio. However, Yoruba Mid tones are sometimes assigned H or L tones in the Krio equivalents, although there is no apparent pattern. The influence of West African languages on Krio is evident not only in lexical borrowings but also in borrowed grammatical properties, such as compounding. Krio continues to borrow lexical items from African languages, particularly from Mende and Themne, the most widely spoken languages in Sierra Leone after Krio.

2.5 The interaction of stress and tone in Krio In spite of the influence of African languages on the tonal specifications of lexical items in Krio, the assignment of H tone nevertheless does coincide with placement of stress (primary or secondary) in some polysyllabic lexical items (containing three or more syllables) borrowed from English. In these lexical items, H tone corresponds with the primary or secondary stress closest to the end of the word, although L tone is not necessarily assigned to unstressed syllables. Thus my position is that Krio is a tone language, with the following specifications: (a) Tone is used contrastively in words of both African and English origins. (b) Tone assignment is unpredictable on words of African origin, but is predictable in a restricted set of polysyllabic items of English origin. (c) H tone does not necessarily coincide with primary stress, and neither is an unstressed syllable always assigned L tone, as proposed for pitch-accent languages.

2.6 Data sources The primary source for the Krio data used in this paper is a Krio-English dictionary compiled by Fyle and Jones (1980). According to the authors, who are native Krio speakers, the meanings and pronunciations (including tonal specifications) assigned to lexical entries represent accurately the standard usage of Krio-speaking

225


residents of Freetown. Some lexical entries in the dictionary contain multiple tonal specifications reflecting pronunciation variations among native speakers. Bradshaw (1966) is used as a supplementary source for Krio words of Yoruba origin. Examples used in this paper are limited to those whose tonal specifications are supported by my own intuitions as a native Krio speaker.

3. Contrastive use of tone in Krio Contrastive use of tone in Krio is evident in lexical items of both African and English origins as the following examples illustrate. Most of the words of African origin are borrowed from Yoruba and other languages widely spoken in Sierra Leone. In the following sets, High tone (H) is marked as [´] and Low tone (L) as [`]: (1) Contrastive pairs of words of African origin àlè ‘skin-irritating herb’ a. àlé ‘go away’ b. bábá ‘a type of drum’ bábà ‘a barber’ bàbá ‘a young boy’ c. gbáŋgbá ‘a wide open public gbàŋgbá ‘a medicinal herb’ place’ d. káŋgá ‘a species of fish’ káŋgà ‘a personal name’ kàŋgà ‘magic’ e. pátá ‘baby panties’ pàtá ‘a slap’ (N); ‘to hit with palm of hand’ (V) f. sósó ‘only’; ‘nothing but’ sósò ‘so-so; not too good, not too bad’ sòsó ‘Susu’ (language of Sierra Leone) ‘a native speaker of Susu’ Tone is sometimes used contrastively for disambiguation purposes, to distinguish between the meanings of lexical items of English origin with identical segmental properties. In most cases, the lexical items have a polysemous relationship, and tone is used to distinguish the meaning of one from the other(s), as shown below: (2) Contrastive pairs of words of English origin a. bèbí ‘a baby’; ‘a doll’ bébì ‘girlfriend’; ‘an attractive young woman’ b. brdà ‘brother’ brdá ‘an elder brother or older male relative’ c. fádá ‘God’ fádà ‘father’ fàdá ‘a Catholic priest’ d. kntrì ‘a country’ kntrí ‘someone who hails from the countryside’


e. f. g.

sístà sìstá skwáyà skwàyá tε´ŋkì tε`ŋkí

‘sister’ ‘an elder sister or older female relative’ ‘a square’ (shape) ‘a dull and inexperienced person’ ‘thank you’ ‘gratitude’

The examples in (1) and (2) illustrate the contrastive use of tone in words of African and English origins. In addition, H or L tone assignment is neither influenced by vowel quality, nor by syllable weight, nor by syllable placement in the sentence.

4. Tone assignment on words of African origin in Krio Tone assignment on monosyllabic lexical items of African origin is predictable in Krio, as it invariably falls (ˆ) in sentence-final position and in citation form, but is realized as H in non-sentence-final position (Fyle and Jones 1980). Polysyllabic lexical items in Krio generally carry level tones — H (´) and L (`) — and are lexically specified. Hence, a number of combinations of H and L tones are theoretically possible, as illustrated below: (3) Two syllables káb ‘welcome’ kàŋgà ‘magic’ kókó ‘a bump on the body’ pàdí ‘friend’ (4) Three syllables árárá ‘nothing at all’ émìná ‘plant with an edible yam-like tuber’ dʒε`kùtε` ‘elephantiasis’ màràbú ‘a Muslim’ máslásì ‘mosque’ òkúrú ‘rabies’ rákpàlà ‘engage in rough and tumble’ wàhálà ‘trouble’ (5) Four syllables àbíám ‘woman with motherly feelings’ àláfíà ‘peace of mind’ àlákòrí ‘Good for nothing person’ gbògbògìà ‘thick grass’ kítíkàtà ‘rushing furiously’ kàkàtúà ‘the ring leader’ óbìátà ‘a type of sauce’

227

228


mláŋké sáŋkámàrún súmáŋgámá táwákáltù wàràwèré

‘handcart’ ‘a poor imitation of jewelry’ ‘incest’ ‘to be determined’ ‘suddenly; in a flash’

As the above examples illustrate, tone assignment on words of African origin cannot be predicted by vowel quality, syllable weight, or position of syllable in the sentence. Pitch has to be lexically specified and multiple combinations of H and L are possible. All four possible combinations of H and L tones — HH, HL, LL, LH — are evident on disyllabic lexical items in (3). All eight possible combinations of H and L tones on three syllable words — HHL, LHL, LLH, LLL, HHH, HLH, HLL, LHH — are evident in (4). Likewise, the examples in (5) are illustrations of twelve out of sixteen (theoretically) possible combinations of H and L tones: LHHL, LLHL, HHHL, LLLH HLHL, LHLH, LHHH, HHLL, LLLL, HHHH, HHLH, LHLL, LLHH.

5. Tone marking on lexical items of English origin in Krio For Krio lexical items of English origin, tone assignment is to a large extent predictable in monosyllabic items but unpredictable for disyllabic lexical items. Tone assignment is partially predictable for words of three or more syllables, with H tone coinciding with the primary or secondary stress closest to the end of the word.

5.1 Tone assignment on monosyllabic lexical items of English origin Tone assignment on monosyllabic lexical items of English origin in Krio, like those of African origin, invariably falls ( ˆ ) in sentence-final position and in citation form, and is realized as H in non-sentence-final position (Fyle and Jones 1980). Monosyllabic grammatical items (prepositions, conjunctions, auxiliaries) of English origin are exceptions to this pattern. They have an inherent low tone regardless of context (Fyle and Jones 1980). The following grammatical items are all assigned L tone regardless of the context in which they are used, with the exception of /dɔn/ (‘Perfective’), which has a high tone: (6) Items with L tone bin ‘past tense marker’ go ‘future’ bɔt ‘but’ kin ‘conditional’ de ‘progressive’ lεk ‘comparative (like/as)’ εn ‘and’ na ‘locative (on, in, at)’

Tone assignment on lexical items of English and African origin in Krio 229

5.2 Tone assignment on polysyllabic lexical items The strongest evidence of Krio as a tonal language primarily centers on the assignment of tone on polysyllabic lexical items, which, unlike pitch-accented creoles, is mainly unpredictable. Pitch-accented creoles are said to have no more than one underlying H tone, whose location on polysyllabic lexical items is predictable, usually coinciding with the location of primary stress on the cognate word in the source stress language. In effect, primary stress is reinterpreted as H tone, and unstressed syllables are automatically assigned L tone. Stress in English, the source language of Krio, is to a great extent predictable. Primary stress is largely determined by syllable weight (the number of morae that the syllables contain), and is generally assigned to a heavy syllable (CVV, CVC, CVCC, etc.). Tone assignment in Krio is not determined by syllable weight, and words derived from English may be assigned more than one H tone, which may or may not coincide with primary stress in English. Unstressed syllables are not automatically assigned L tone. Thus, different processes govern tone assignment in Krio and stress assignment in English

5.3 Unpredictability of tone assignment on disyllabic lexical items Tone marking on disyllabic grammatical items of English origin is not predictable and must be specified for each item. For example: (7) lε`kε` ‘comparative (like/as)’ ndà ‘under’ pàntáp ‘above/on top of ’ There are a number of disyllabic lexical items of English origin, including recent borrowings, in which English primary stress coincides with H tone in Krio: “Certain English-derived words considered ‘superior’ or as yet unassimilated keep their English pitch patterns, which then become Krio tonal patterns” (Fyle and Jones 1980: xxii). However, for the vast majority of disyllabic lexical items, the tonal specifications do not predictably correlate with the English stress pattern. A HL or LH sequence cannot be predicted by the position or weight of the syllable, or by the semantic category (noun, verb, etc.) of the lexical item. In the following examples, primary stress coincides with H tone on an initial syllable in (8), and with H tone on the final syllable in (9). In (10), however, H tone conflicts with primary stress location, and it is not possible to predict when H tone may coincide or conflict with primary stress: (8) Correspondence of primary stress and high tone English Krio ármy ámì búcket bókìt cíty sítì


búsiness Chrístmas márket ómelette péople

bíznε`s krísmε`s mákìt mlε`t pípùl

(9) English advíce allów befóre command demánd expéct respéct supplý

Krio àdvaís àlaú bìfó kmànd dìmánd `εkspε´kt rε`spε´kt splái

(10) English mónkey wáter dárling párty dáddy bíscuit bódy básket brácelet grávy cándy chócolate blánket máttress cápsize báptize

Krio mŋkí wàtá dàlíŋ pàtí dàdí bìskít bdí bàskít brèslε´t grèví kàndí tʃklét blàŋkít màtrε´s kàpsaí bàptaíz

It is important to note that the weights (including vowel types) of some of the final syllables with L tones in (8) and the final syllables with H tones in (10) are the same. The mismatch between English stress and Krio tone is further evident in the basilectal variety in disyllabic personal names derived from English. H tone in the mesolect and acrolect varieties generally coincides with English stress. In the following examples, an initial stressed syllable in English corresponds to a L tone in the basilectal Krio equivalents while an unstressed syllable is assigned H tone. Thus a LH (rather than a HL) tonal pattern emerges:


(11) Disyllabic personal names derived from English English Krio (first stress on initial syllable) (H tone on final syllable) Thomas tms Alfred àlfrε´d Agnes àgnε´s Sally sàlí David dèvíd Mary mèrí Joseph dʒòsε´f Frances frànsε´s Rachel rètʃε´l Wallace wlés Peter pìtá Gertrude gàtrúd This phenomenon of tone shift — a disyllabic lexical item receiving syllable-initial stress in English but syllable-final stress in creoles — is observed by Devonish (2002), not only in West African pidgins and creoles (including Krio), but also in a number of creole varieties in the Americas (including Guyanese Creole), as a result of contact between speakers of the two varieties. The lexical items involved are almost identical in form and meaning in both varieties. The only difference is that unlike Krio (and other West African pidgins and creoles) in which the initial stress is realized as L tone, in Guyanese Creole and other Caribbean creoles, it is observed as a falling tone in equivalent lexical items.

5.4 Partial predictability of tone on polysyllabic lexical items Tone assignment on English-derived polysyllabic lexical items with three or more syllables is partially predictable in Krio. In English, one syllable of polysyllabic lexical items is assigned primary stress while additional syllables may receive secondary stress or remain unstressed. When such items are borrowed into Krio, the general pattern is for the first indication of stress (primary or secondary) from the end of the word to be automatically converted into H tone. That is, H tone corresponds with the primary or secondary stress that is closest to the end of the word. However, L tone is not necessarily assigned on unstressed syllables. If the final syllable of an English lexical item is assigned primary or secondary stress, the Krio equivalent is assigned H tone, and all the preceding syllables, regardless of their stress status in the English equivalent, are assigned L tone by default. For example: (12) English télephòne cálculàte

Krio tε`lìfón kàlkyùlét

23

232


críticìze ísolàte éxercìse éstimàte úndermìne cómpensàte légalìze únderstànd clárify congrátulàte domésticàte

krìtìsáiz aìsòlét `εksàsáiz `εstìmét ndàmaín kmpε`nsét lìgàláiz ndàstán klàrìfái kŋgràtyùlét dòmε`stìkét

If one or more non-final syllables are assigned stress (primary and secondary) in English, the stress closest to the end of the word is converted to H tone, and all preceding syllables (if any) are automatically assigned L tone. The final syllable, generally unstressed in English, is also assigned L tone. The H tone is then copied perseveratively (left to right) on to following non-final syllables (if any). The process of perseverative tone copying, whereby a syllable receives its tone from an adjacent syllable to its left, is a common process among West African tonal languages (including Yoruba, Nupe, Guari, and Ngizim), and is said to be a very productive tendency in tonal languages (Hyman and Schuh 1974). In the examples in (13) below, only one syllable is stressed. All the others are unstressed. In (13a), H tone corresponds with stress on the middle syllable while the final syllable, unstressed in English, is assigned L tone in Krio. The preceding initial syllable is assigned L tone by default. In (13b), H tone corresponds with stress on the initial syllable, and is copied on to the only non-final syllable (the middle syllable). The final syllable, unstressed in English, is assigned L tone in Krio. In (13c), stress on the antepenultimate syllable is converted to H tone, which is copied on to the penultimate syllable. The final syllable is assigned L tone, as is the initial syllable. In (13d), the stress on the initial syllable is converted into H tone, which spreads on to the antepenultimate and penultimate syllables. The final syllable is assigned L tone: (13)

English eléction cassáva banána chamélion b. árticle médicine términal géneral c. América condítional biólogy a.

Krio ìlε´kʃn kàsádà bànánà kmíε`l átíkùl mε´rε´sìn támínàl dʒεńε´ràl àmε´ríkà kndíʃnàl bàyldʒì


d. dífficulty dífíkltì díctionary díkʃnε´rì nécessary nε´sísε´rì The examples in (14) contain both primary and secondary stress. The stress on the penultimate syllable, which is closest to the end of the word, coincides with H tone in Krio. The final unstressed syllable is converted into L tone. All other syllables preceding the H tone assigned on the penultimate syllable receive L tone by default: (14)

English àpplicátion èducátion èleméntary b. examination congràtulátion dissàtisfáction èxtraórdinary respònsibílity c. hélicòpter télevìsion a.

Krio àplìkéʃn `εdyùkéʃn `εlìmεńtrì `εgzàmìnéʃn kŋgràtyùléʃn dìsàtìsfákʃn `εkstràdínárì rε`spnsìbílítì `εlìkptà tε`lìvíʒn

5.5 Exceptions There are some exceptions to the pattern proposed for words with three or more syllables. Only a few of these exceptions are listed in the Krio-English dictionary (Fyle and Jones 1980). These counterexamples include tri-syllabic words with primary stress on the initial syllable in English, but in Krio they have H tone on the middle syllable: (15) English cálendar ófficer hóspital báptism Ábigail párable

Krio kàléndà físà spítùl bàptízìm àbígε`l pàrébùl

6. Tone assignment on Krio compounds of English origin In English, the compound stress rule generally assigns primary stress on the initial member of a compound while the second member is assigned secondary stress. Fyle and Jones (1980) observe that for English-derived compounds consisting of monosyllabic constituents, the initial constituent generally carries a L tone while

233

234


the final constituent is assigned a H tone. This is similar to the pattern of tone assignment in non-compounds, in which the stressed (primary or secondary) vowel closest to the end of the word is assigned H tone, and all other preceding vowels are automatically assigned L tone. For example: (16) English bláckbòard fóotbàll báthtùb clássròom líving-ròom géarbòx práyer-bòok státion-màster

Krio blàkbód fùtbl bàt-tb klàsrúm lìvìn-rúm gìàbks prèà-búk stèʃn-mástà

There are other compounds in Krio whose individual constituents are derived from English. The pattern described above applies to them as well, but only when the initial member of the compound has a HL sequence, as shown in (17). That is, it is restricted to compounds whose initial members are disyllabic with a HL sequence. But this pattern also applies to some monosyllabics, in which the falling tone in Krio is arguably a HL sequence. However, the initial member of a compound with LH sequence is not affected by this process, as shown in (18). Apparently, a LH sequence on an initial constituent of a compound blocks this process: (17) àfrìkà-mán áfríkà (Africa) + mân (man) àmε`rìkà-mán àméríkà (America) + mân (man) fìŋgà-pkìt fíŋgà (finger) + pkìt (pocket) gànà + mán gánà (Ghana) + mân (man) mε`rε`sìn-mán mε´rε´sìn (medicine) + mân (man) rtìn-bε`lε´ rtìn (rotten) + bε`lε´ (stomach) = (18) màmí-wàtá màmí (mother) + wàtá (water) bèbí- êd bèbí (baby) + êd (head) krìyó-by krìyó (Krio) + bɔˆy (boy) sàlón-mán sàlón (Sierra Leone) + mân (man)

‘an African’ ‘an American’ ‘pickpocket’ ‘a Ghanaian’ ‘a witchdoctor’ ‘gluttonous’ ‘mermaid’ ‘a sore toe’ ‘a male Krio speaker’ ‘a Sierra Leonean’


7. Discussion and concluding remarks Contrary to claims that Krio exhibits characteristics of pitch-accent languages (Alleyne 1980, Berry 1959, Hall 1966, Strevens 1966), the evidence presented here supports the view that Krio is a tonal language. Tone-assignment is largely unpredictable in polysyllabic lexical items of both African and English origin. The contrastive and lexical nature of tone assignment is also supported by the fact that tone is used to distinguish a small set of English and African loanwords. Furthermore, tone assignment is not conditioned by the position or weight of the syllable. In addition, the proposed correlation of primary stress with H tone on the one hand, and of unstressed syllables with L tone on the other hand — a correspondence that is characteristic of pitch-accent languages — , is not realized in polysyllabic Krio lexical items. H tone on disyllabic lexical items is generally unpredictable. It is partially predictable on lexical items of three or more syllables with H tone coinciding with the primary or secondary stress closest to the end of the word. Nevertheless, L tone is not automatically assigned on vowels in unstressed syllables. All vowels preceding the H tone that corresponds with English primary or secondary stress are assigned L tone regardless of their stress status in English. This H tone is also copied on to following non-final unstressed syllables, resulting in more than one H tone assigned on some polysyllabic lexical items. These processes apply to compounds as well. Furthermore, tone is used contrastively on English-derived words for disambiguation purposes, as the examples in (2) illustrate. It is likely that the stress pattern of English has determined to a certain extent the correspondence between H tone and primary or secondary stress in some words of English origin. And indeed, this influence continues to be exerted on new English loanwords. However, Krio, the lingua franca of Sierra Leone, is extensively exposed to the influence of contiguous tonal languages, which has led to the incorporation of tonal rules into Krio lexical items.

References Alleyne, M. 1980. Comparative Afro-American. Ann Arbor: Karoma Publishers. Berry, J. 1959. “The origins of Krio vocabulary.” Sierra Leone Studies 12: 298–307. Berry, J. 1961. “English loanwords and adaptations in Sierra Leone Krio.” In Creole Language Studies 2, R. Le Page (ed.), 1–16. London: Macmillan. Berry, J. 1970a. “A note on Krio tones.” African Language Studies 11: 59–60. Berry, J. 1970b. “A note on the prosodic structure of Krio.” International Journal of American Linguistics 36: 266–7. Bickerton, D. 1975. Dynamics of a Creole System. Cambridge: Cambridge University Press. Bickerton, D. 1977. “Pidginization and creolization: Language acquisition and language universals.” In Pidgin and Creole Linguistics, A. Valdman (ed.), 49–69. Bloomington: Indiana University Press. Bickerton, D. 1988. “Creole language and the bioprogram.” In Linguistics: The Cambridge Survey. Volume 2: Linguistic Theory: Extensions and Implications, F. Newmeyer (ed.), 268– 84. Cambridge: Cambridge University Press.

235

236

Malcolm Awadajin Finney Bradshaw, A. T. von. 1966. “A list of Yoruba words in Krio.” Sierra Leone Language Review 5: 61–71. Carter, H. 1987. “Suprasegmentals in Guyanese: Some African comparisons.” In Pidgin and Creole Languages, G. Gilbert (ed.), 213–63. Honolulu: University of Hawaii Press. Devonish, H. 1989. Talking in Tones: A Study of Tone in Afro-European Creole Languages. Kingston, Jamaica: Karia Press, Caribbean Academic Publications. Devonish, H. 2002. Talking Rhythm, Stressing Tone: The Role of Prominence in Anglo-West African Creole Languages. Kingston, Jamaica: Arawak Publications. Fyle, C. 1994. “Official and unofficial attitudes and policy towards Krio as the main lingua franca in Sierra Leone.” In African Languages, Development and the State, R. Fardon and G. Furniss (eds.), 44–54. London: Routledge. Fyle, C. and E. Jones. 1980. A Krio-English Dictionary. Oxford: Oxford University Press. Goldsmith, J. 1976. Autosegmental Phonology. Bloomington: Indiana University. Goldsmith, J. 1990. Autosegmental and Metrical Phonology. Oxford: Blackwell. Hall, R. 1966. Pidgin and Creole Languages. Ithaca: Cornell University Press. Holder, M. 1991. “Towards an integrated model of pitch and stress in Guyanese.” Paper presented at the Canadian Linguistic Association Annual Conference, Carleton University, May, 1991. Hyman, L. 1978. “Historical tonology.” In Tone: A Linguistic Survey, V. Fromkin (ed.), 257–69. New York: Academic Press. Hyman, L. and R. Schuh. 1974. “Universals of tone rules: Evidence from West Africa.” Linguistic Inquiry 5: 81–115. Johnson, A. 1974. A Linguistic Survey of Tones in Sierra Leone Krio. MA Thesis, Leeds University. Jones, E. 1971. “Krio: An English-based language of Sierra Leone.” In The English Language in West Africa, J. Spencer (ed.), 66–94. Harlow: Longman. Leben, W. 1973. Suprasegmental Phonology. PhD Thesis, M. I. T., Cambridge, MA. McWhorter, J. 1998. “Identifying the creole prototype: Vindicating a typological class.” Language 74: 788–818. Pike, K. 1948. Tone Languages. Ann Arbor: University of Michigan Press. Poldauf, I. 1984. English-Word Stress: A Theory of Word-Stress Patterns in English. Oxford: Pergamon Press. Pulleyblank, D. 1986. Tone in Lexical Phonology. Dordrecht: D. Reidel Publishing Company. Schneider, G. 1966. West African Pidgin-English: A Descriptive Linguistic Analysis with Texts and Glossary from the Cameroon Area.” PhD Thesis, Hartford Seminary Foundation, Athens, OH. Seuren, P. and H. Wekker 1986. “Semantic transparency as a factor in creole genesis.” In Substrata versus Universals in Creole Genesis, P. Muysken and N. Smith (eds.), 57–70. Amsterdam: John Benjamins. Strevens, P. 1966. Papers In Language Teaching. London: University Press. Woo, N. 1972. Prosody and Phonology. Bloomington: Indiana University Linguistics Club.

CHAPTER 11

TMA and the St. Lucian Creole verb phrase David B. Frank SIL International

The use of preverbal TMA (tense–mood–aspect) markers in St. Lucian French Creole to denote time reference is predictable in terms of the distinction between stative and nonstative verbs and in terms of contextual factors. The syntax and semantics of the basic set of TMA markers for St. Lucian Creole are presented and compared with those of other Caribbean creole languages. Context-sensitive constraints are then described and identified in a corpus of narrative data.

1. Introduction TMA (tense–mood–aspect) has been a dominant topic in creole studies ever since Bickerton’s (1974, 1975, 1981) claim that patterns of TMA marking figured among creole universals. Despite a number of critical studies that purport to present counterexamples to some of Bickerton’s claims, it is widely recognized that there is indeed a general tendency among Caribbean creole languages for the TMA systems to function along the lines of a stative/nonstative distinction. As a complement to current research on TMA patterns in creole languages, this article examines the correlation between the TMA system and the stative/nonstative distinction among verbs in St. Lucian Creole (henceforth SLC).¹ The main thesis is that the usage patterns of TMA preverbal markers can be recognized as rulegoverned once the relevant contexts are sorted out. The primary factor governing the TMA patterns is whether the verb in question is stative or nonstative. Other factors are whether the verb phrase is in a dependent or independent clause, whether or not it is used in reported speech, and other contextual considerations. While the analysis presented here is complementary to the body of literature on TMA patterns among creole languages in general, the TMA patterns in St. Lucian Creole differ in some respects from patterns that have been described for other creole languages.

2. Background Carrington (1968, 1984) was the first to describe the TMA system of SLC in detail.² He observed that “the presence or absence of predicative particles is one of the

238

David B. Frank

devices by which St. Lucian Creole expresses the grammatical categories of tense, aspect and mood” (Carrington 1984: 116). The pattern Carrington found for SLC is consistent with a pattern that became increasingly evident in other creole varieties. Carrington also noted that a stative/nonstative distinction is factored into the TMA usage patterns, though he uses a different terminology. He considers a “zero particle” to denote completive aspect. But this forces him to address what from an Englishspeaker’s perspective might appear to be notable exceptions: /mwẽ malad/ is viewed as the completive result of a process of becoming ill in the same way as /mari mwẽ la_glitè/³ is viewed as the completive result of my husband’s going to England. They are hence aspectually exact equivalents of /i prã zafe i/ in which the subject has completed the action of taking. Adjectival and locative predicates as well as predicates involving a small number of verbs such as /sav/ ‘to know’, /vle/ ‘to want’, /simie/ ‘to prefer’, /ni/ ‘to have’ will therefore tend to be glossed in English by present tenses despite their completive aspect. (Carrington 1984: 117)

Bickerton describes a strikingly similar pattern for TMA in Guyanese, an English-lexifier creole: The functions of the stem form in the Guyanese system depend on the stative — non-stative distinction. In its commonest function, with non-statives, it signifies ‘unmarked past’ — that is, a (usually) single action that happened at a moment in the past that may or may not be specified but should not predate any action simultaneously under discussion [. . .]. With stative verbs, however, the stem form signifies non-past. (Bickerton 1975: 28–29)

Bickerton later generalizes his observations of TMA, comparing Guyanese Creole (GC) to Hawaiian Creole English (HCE) and a variety of other creole languages: In the typical system — which HCE shares with GC, Sranan (SR), Saramaccan (SA), Haitian Creole (HC), and a number of other creoles — ranges of meaning of the particles are identical: the tense particle expresses [+Anterior] (very roughly, pastbefore-past for action verbs and past for stative verbs); the modality particle expresses [+Irrealis] (which includes futures and conditionals); while the aspect particle expresses [+Nonpunctual] (progressive-durative plus habitual-iterative). The stem form in isolation expresses the unmarked term in these three oppositions, i.e., present statives and past nonstatives. (Bickerton 1981: 58)

In his description of the verb systems of certain African language families,Welmers (1973) makes a distinction between “stative” and “active” verbs that, though it differs in some details, is essentially the same system that Carrington (1968, 1984) describes for SLC and Bickerton (1975) for creole languages in general. Welmers presents what he calls a “factative” construction whereby “some languages use a single construction to refer to past time for active verbs and present time for stative verbs” (1973: 348).⁴ In the case of Yoruba, for example, this is accomplished by the use of the bare verb (Welmers 1973: 346).

TMA and the St. Lucian Creole verb phrase

Bickerton argues that “the basic stative — non-stative distinction [. . .] plays no part (at least no overt part) in the English system and therefore [an English creole] cannot in any sense represent a simplification of that system” (1975: 46). Welmers, however, demonstrates that stativity does play a role in the English predicate system. Thus the present of an active verb such as walk is “I am walking” (rather than “I walk”), whereas the corresponding form for a stative verb such as know is “I know” (rather than “I am knowing”). He concludes that “a number of languages distinguish two types of verbs, in general ‘stative’ and ‘active’ [. . .]. For statives, a reference to present time may use the same construction that refers to past time if an active verb is used” (1973: 346). Quirk, Greenbaum, Leech & Svartvik also consider a difference between what they call “stative” and “dynamic” to be a fundamental distinction in English grammar (1972: 94). They explain that this distinction is most obvious in the verb system in the use of the progressive (1972: 39), but also in the use of imperatives, pseudoclefts, DO pro-forms, and causatives (1972: 94). Spears (1990, 1993) takes exception to Bickerton’s analysis of the Haitian TMA system. While for Bickerton, the unmarked verb form indicates past reference for nonstative verbs and present reference for stative verbs, Spears denies that absence of marking in Haitian explicitly indicates time reference. Spears places greater emphasis on the role of context and claims that “in the case of HC, it is clear that the present-stative/past non-stative hypotheses on the meaning of zero forms, following the semantics of what has been referred to as the classic creole TMA system, do not work” (1993: 262). He argues for a tendency rather than an absolute rule, and maintains that a quantitative analysis would be needed to explain the distribution of the unmarked verb (1993: 263). In his survey of the TMA systems of Caribbean English creoles, Winford agrees with Bickerton’s view that “the Stative/Non-stative distinction is crucial to the interpretation of temporal and aspectual meaning” (Winford 2001: 5). However, he adds that all creoles allow combinations that depart from Bickerton’s prototype (2001: 14). He thus explains apparent exceptions to the general rule: TMA categories in every language typically have a range of meanings and uses, i.e., interpretations in discourse. Every TMA category has a dominant meaning and often has other secondary meanings. In general, the dominant meaning of a category is represented in its primary or prototypical uses, while secondary meanings are interpretations that arise from contextual uses of the category. (Winford 2001: 3–4)

Since there is some debate on how well various Caribbean creoles fit Bickerton’s general rule for TMA, it is good to examine the facts on SLC. The findings presented here are that the TMA patterns of usage for SLC are more than statistical tendencies, and that the rules governing these patterns emerge more clearly when different contexts are taken into consideration.

239

240 David B. Frank

3. Overview of St. Lucian creole clause patterns In terms of TMA and the stative/nonstative distinction, there are three basic patterns for independent clauses in SLC: the equative pattern, the descriptive pattern, and the transitive/intransitive pattern.

3.1. The tense–mood–aspect markers By way of introduction to the topic, the main tense, mood and aspect markers will now be examined. According to Chung & Timberlake, [t]ense, aspect, and mood are all categories that further specify or characterize the basic predication, which can be referred to as the event. Tense locates the event in time. Aspect characterizes the internal temporal structure of the event. Mood describes the actuality of the event in terms such as possibility, necessity, or desirability. (1985: 202)

Time reference, as distinguished from tense, can be expressed by a combination of TMA markers (see Section 3.2 below). The three main TMA markers for SLC are as follows: té past tense/anterior tense/anti-perfect/relative past kay future/prospective mood/irrealis ka noncompletive/nonpunctual aspect In addition, there is the morpheme sa that could be characterized as deontic mood (cf. Chung & Timberlake 1985: 246–247), though Carrington puts it in the category of Predicative Auxiliaries. It means something like “can” or “be able to” as in Mwen pa sa wè’w ‘I cannot see you’. Additionally there is the morpheme pé, which partially overlaps in meaning with sa but can also mean ‘may’ or ‘might’.⁵ Carrington (1984: 119) labels té as “past tense”, but this label needs further specification. It is “past” in relation to some particular frame of reference. For stative verbs and constructions, it means “past” with respect to the present time, the time of telling. For nonstative verbs it is “past” with respect to a previously established past frame of reference, i.e., “past-before-past” or “pluperfect”. Bickerton prefers the term “anterior” (1981: 58), and Winford “relative past” (2001: 5). Spears (1993: 262) argues in favor of the term “anti-perfect” for té in Haitian Creole. The exact term applied to this tense marker matters less than a recognition of its use in different constructions.⁶ The particle kay may at first glance appear to function as a future tense marker, and indeed it does commonly indicate future time reference. However, it is closer to being a modal than a tense marker, since kay refers to an event that is not (yet, at least) an actuality (Chung & Timberlake 1985: 241). Carrington gives kay the label “prospective mood” (1984: 118), which includes not only futurity but also intention and potential. Bickerton prefers the term “irrealis” (1981: 58), which would include future and conditionals.


Rather than indicating past, present or future tense, the particle ka marks aspect, which has to do with “the relationship of a predicate to the time interval over which it occurs” (Chung & Timberlake 1985: 213). Carrington calls ka “non-completive” (1984: 118), whereas Bickerton labels it “nonpunctual” (1981: 59). But while ka does not denote tense per se, it does indicate time reference in common usage. The sentence mwen vini, as an independent clause without any TMA markers, means ‘I came’ or ‘I have come’ and has past time reference. Mwen ka vini, on the other hand, means ‘I am coming’, with present time reference.⁷ According to Comrie,“since the present tense is essentially used to describe, rather than to narrate, it is essentially imperfective, either continuous or habitual, and not perfective” (1976: 66).

3.2. The equative clause The equative clause pattern involves the copula verb sé joining a subject noun phrase with a complement noun phrase that refers back to the subject:⁸ (1) Misyé Pòl sé vwézinaj mwen. Mr. Paul COPULA neighbor my ‘Mr. Paul is my neighbor.’ Several observations can be made about the equative clause type. A simple copula predicate consists of sé ‘am/is/are’ only. The negative marker pa and the TMA markers (té “past” or “anterior”, kay “prospective” or “irrealis”, ka “nonpunctual”, sa “abilitative”) can co-occur with the copula-type predicate, subject to certain rules. One such rule is that té + sé → sété (or just té without sé). If sé occurs outside the normal order of subject-predicate-object, as in a question, it takes the form yé, as in Ki moun i yé? ‘Who is he?’ If TMA markers other than té occur in an equative clause, the copula verb sé is omitted, as in pa + té + sé → pa té; sa + sé → sa. The combinations permitted in an equative clause are limited in natural speech: a maximum number of two tense, mood, aspect and copula morphemes can co-occur in a given clause. Finally, equative clauses follow the stative pattern for TMA. That is, except for some particular contexts noted below, the copula without té denotes present reference. Here are some further examples of the equative clause type, with the predicates underlined: (2) Ou sé jan mwen. Ou té jan mwen. Ou kay jan mwen. Ou pa té kay jan mwen. Ou sa jan mwen. Ou pa sa jan mwen. Ou pa té sa jan mwen. Non mwen sé Tjals. Tjals sété an Endyen.

‘You are my friend.’ ‘You were my friend.’ ‘You will be my friend.’ ‘You wouldn’t be my friend.’ ‘You can be my friend.’ ‘You cannot be my friend.’ ‘You couldn’t be my friend.’ ‘My name is Charles.’ ‘Charles was an Indian.’

24

242 David B. Frank

Ki sa sa yé? Sa sé an kannòt.

‘What is that?’ ‘That is a dugout canoe.’

3.3. The descriptive clause Another clause type similar to the equative clause (with copula) is the descriptive clause, which lacks a copula, even in present reference. A descriptive clause consists of subject NP immediately followed by a complement modifying it. (3) Mwen las. I tired ‘I am tired.’ (4) Mari an jaden-an. Mary in garden-DEF ‘Mary is in the garden.’ Although this clause type is characterized by the absence of a verb, the predicate can include the negative marker and the TMA markers. What distinguishes a descriptive from an equative clause, besides the absence of a copula verb, is that the complement referring back to the subject is an adjective, an adjective phrase (AP), a prepositional phrase (PP), or a locative (Loc). In other words, if speakers of SLC want to equate one thing with another, they use NP + COPULA + NP, but to describe something they use NP + AP/PP/Loc without the copula. The clause I ka malad toulé Lendi ‘He is sick every Monday’ includes the nonpunctual aspect marker ka, but no verb.⁹ If the word order in a descriptive clause is other than the normal subject-predicate-object, then the verb is realized as yé. Again, there is a limit to the number of TMA markers that can occur in the verb phrase. It has been proposed that adjectives function as verbs in creole languages.¹⁰ Bickerton states categorically that “adjectives are surface verbs in creoles [. . .] and therefore require no copula [. . .]. I know of no creole where an alternative analysis of adjectives would be required” (Bickerton 1981: 67, 69). Seuren (1986) argues effectively against such a proposal in the case of Sranan.¹¹ An adjective-as-verb analysis in the case of SLC would fail to account for the presence of a prepositional phrase or a locative as substitutes for the adjective in complement position. As with the equative clauses, descriptive clauses follow the stative pattern for TMA. The following are some further examples of descriptive clauses: (5) Ti manmay-la fen. Ti manmay-la té fen. Ti manmay-la kay fen. Ti manmay-la pa té sa fen. Ti manmay-la ka fen (toulé swè).

‘The child [is] hungry.’ ‘The child was hungry.’ ‘The child will be hungry.’ ‘The child could not be hungry.’ ‘The child is habitually hungry (every night).’


Ti manmay-la té ka fen. Ti manmay-la byen anbétan. Ti manmay-la sa byen anbétan. Ti manmay-la andidan kay-la. Ti manmay-la Vyé Fò. Kouman ou yé? Mwen byen. Sa bon.

‘The child used to be hungry.’ ‘The child [is] quite troublesome.’ ‘The child can be quite troublesome.’ ‘The child [is] in the house.’ ‘The child [is in] Vieux Fort.’ ‘How are you?’ ‘I [am] well.’ ‘That [is] good.’

Since it lacks a verb, a descriptive clause is similar to — and sometimes indistinguishable from — a noun phrase. Context is one distinguishing factor between descriptive clauses and noun phrases, but there are other differences in terms of internal composition as well. A TMA marker can be inserted between the subject and complement in the case of a descriptive clause but not in the middle of a noun phrase. Out of context, a noun followed by a modifier can be either a noun phrase or a descriptive clause, but a proper noun or a pronoun followed by a modifier can only be a descriptive clause. The placement of the definite article provides another clue to sentence structure in that the article is placed at the end of a noun phrase, as in (6)–(7): (6) bway lèd-la. boy ugly-DEF ‘the ugly boy.’

(noun phrase)

(7) Bway-la lèd. (descriptive clause) boy-DEF ugly ‘The boy [is] ugly.’

3.4. The transitive/intransitive clause Besides the equative and descriptive clause types, there is a regular clause type that may include a transitive or intransitive verb. Whereas both the equative and the descriptive clauses follow the stative pattern with respect to time reference (with the unmarked form for present reference), transitive and intransitive clauses can include either a stative or nonstative verb.¹² The basic SLC clause consists of a subject NP followed by a predicate and, optionally, an object NP, as illustrated in (8)–(9): (8) Chat-la kouwi. cat-DEF run ‘The cat ran.’ (9) Machin-an tjwé tifi-a. truck-DEF kill girl-DEF ‘The truck killed the girl.’

243

244 David B. Frank

Comrie notes that, generally speaking, we find that verbs tend to divide into two disjoint (nonoverlapping) classes, those that can appear in the progressive forms, and those that cannot. Moreover, this distinction corresponds to that between stative and nonstative verbs. (1976: 35)

He adds that “different languages are free to choose, essentially as an arbitrary choice, whether such verbs are classified as stative or not” (1976: 35). In other words, the distinction between stative and nonstative verbs is widespread among languages of the world, but the particulars are language specific. In the case of SLC, the set of stative verbs is relatively small, and they all express what Welmers describes as “private actions” — that is, “actions of which only the actor is the proper judge” (1973: 345). Some SLC verbs that might be considered private actions also pattern as nonstative, however. Table 1 presents sample verbs, arranged according to their transitivity and stativity. Table 1. Stative and nonstative intransitive and transitive verbs Stative

Nonstative

Intransitive

dakò mélé

‘to agree’ ‘to care’

alé asiz diwé dòmi doubout kouwi palé pawèt

‘to go’ ‘to sit’ ‘to last’ ‘to sleep’ ‘to stand’ ‘to run’ ‘to speak’ ‘to appear’

Transitive

anvi bizwen enmen konnèt kopwann kontan kwè ni pè sav simyé vlé

‘to want’ ‘to need’ ‘to love’ ‘to know’ ‘to understand’ ‘to like’ ‘to think’ ‘to have’ ‘to fear’ ‘to know’ ‘to prefer’ ‘to want’

achté anmasé apèsivwè apwann apwésyé bat bay bo chaché chanjé chayé chwazi délivwé détwi kwè plè wè wéyalizé

‘to buy’ ‘to gather’ ‘to perceive’ ‘to learn’ ‘to appreciate’ ‘to hit’ ‘to give’ ‘to kiss’ ‘to seek’ ‘to change’ ‘to carry’ ‘to choose’ ‘to deliver’ ‘to destroy’ ‘to believe’ ‘to please’ ‘to see’ ‘to realize’


Some of the verbs in Table 1 can have either transitive or intransitive uses. For example, pè can be glossed as ‘to be afraid’ or ‘to fear (something)’. Interestingly, the verb kwè is stative in the sense of ‘to think’, but it is nonstative in the sense of ‘to believe’, as in Mwen kwè ou ni tò ‘I think you are wrong’ (unmarked, present reference) vs. Mwen kwè’y ‘I believed him’ (unmarked, past reference). Bickerton similarly observes that the Guyanese Creole word mek ‘make’ is stative in the GC equivalent of “two and two make four” (i.e.,‘equal four’ or ‘are four’) but is nonstative in the GC equivalent of “they made him stop”. In each of these GC sentences mek is unmarked for tense, but in the former it has present reference while in the latter it has past reference. This evidence suggests to Bickerton that the categories of stative and nonstative “apply to propositions irrespective of their lexical content,” which in turn suggests “that semantics is generative rather than interpretive” (1975: 30). Bickerton fails to acknowledge that words can simply have multiple meanings. As is the case with SLC kwè, one meaning is stative while the other is nonstative. The following list illustrates different stative and nonstative verbs and some of the contexts in which they can occur: (10) Stative Mwen dakò. Mwen pa dakò. Mwen pa mélé. Mwen enmen diwi. Mwen té ka enmen diwi. Mari ni an jip nèf. Mari té ni an jip nèf. Mari té ka ni an jip nèf. Mari kay ni an jip nèf. Mari sa ni an jip nèf. Mari té sa ni an jip nèf. Mari pa kay sa ni an jip nèf. (11)

Nonstative Nonm-lan dòmi. Nonm-lan té dòmi. Nonm-lan kay dòmi. Nonm-lan ka dòmi. Nonm-lan té ka dòmi. Nonm-lan té kay dòmi. Nonm-lan kay ka dòmi. Nonm-lan pa té sa dòmi. Nonm-lan kay sa dòmi. Yo ba li an chouponm. Yo té ba li an chouponm. Yo kay ba li an chouponm.

‘I agree.’ ‘I do not agree.’ ‘I don’t care.’ ‘I like rice.’ ‘I used to like rice.’ ‘Mary has a new skirt.’ ‘Mary had a new skirt.’ ‘Mary used to have a new skirt.’ ‘Mary will have a new skirt.’ ‘Mary can have a new skirt.’ ‘Mary could have a new skirt.’ ‘Mary will not be able to have a new skirt.’ ‘The man slept.’ ‘The man had slept.’ ‘The man will sleep.’ ‘The man is sleeping.’ ‘The man was sleeping.’ ‘The man would sleep.’ ‘The man will be sleeping.’ ‘The man could not sleep.’ ‘The man will be able to sleep.’ ‘They gave her a cabbage.’ ‘They had given her a cabbage.’ ‘They will give her a cabbage.’

245

246 David B. Frank

Yo ka ba li an chouponm. Yo té ka ba li an chouponm. Yo té kay ba li an chouponm. Yo pa kay sa ba li an chouponm.

‘They are giving her a cabbage.’ ‘They were giving her a cabbage.’ ‘They would give her a cabbage.’ ‘They will not be able to give her a cabbage.’

Note that the construction té ka + V is ambiguous: Nonm-lan té ka dòmi can mean either ‘The man was sleeping’ or ‘The man used to sleep’. It may be possible to use té + kay + ka in combination, but such a combination would be rare. Bickerton argues repeatedly that “statives are distinguished from nonstatives by the fact that the nonpunctual marker never attaches to the former” (1981: 160). This statement does not hold true in the case of SLC. Consider example (12): (12) Jan Sent Lisi pa ka enmen jan Babad. ‘St. Lucians don’t usually/generally like Barbadians.’ Here the nonpunctual marker ka precedes the stative verb enmen. Carrington gives several examples as well (1984: 118), such as I ka bèl délè ‘It [the sea] is lovely sometimes.’ More common is the use of ka with stative verbs in combination with té as in Nou té ka ni pli dlo ‘We used to have more water.’

4. The meanings of the parts and of the whole The different TMA markers have been given different names and characterizations by creolists. In Section 3.1 the principal TMA markers for SLC were presented and described. It is difficult or even impossible to assign a single, consistent label to each of these morphemes. This is so because no single label can succinctly capture their wide range of meanings and uses. When combined with stative or nonstative verbs and with each other, these particles exhibit their more specific meanings. But more important than arriving at the meanings of the particles themselves is characterizing the meanings of the verb phrases that contain them. The following list displays some basic verb phrase combinations. Table 2 summarizes the meanings of the different TMA combinations. PRESENT REFERENCE Stative: unmarked (13) Mwen las. (14) Mwen ni yonn.

‘I am tired.’ ‘I have one.’

Nonstative: VERB + ka (15) Mwen ka palé. (16) Mwen kay fè’y.

‘I am speaking.’ ‘I am doing it.’

TMA and the St. Lucian Creole verb phrase 247

SIMPLE PAST REFERENCE Stative: VERB + té (17) Mwen té las. ‘I was tired.’ (18) Mwen té ni yonn. ‘I had one.’ Nonstative: unmarked verb (19) Mwen palé. (20) Mwen fè’y.

‘I spoke.’ ‘I did it.’

PLUPERFECT Stative: (not distinguished from simple past reference) Nonstative: VERB + té (21) Mwen té palé. (22) Mwen té fè’y.

‘I had spoken.’ ‘I had done it.’

HABITUAL/OCCASIONAL/CONTINUOUS: VERB + ka (23) Mwen ka ni yonn. ‘I usually/sometimes have one.’ (24) Mwen ka fè’y. ‘I do it.’ IMPERFECT/PAST PROGRESSIVE: VERB + té + ka (25) Mwen té ka ni yonn. ‘I used to have one.’ (26) Mwen té ka fè’y. ‘I was doing it’ or ‘I used to do it.’ FUTURE/INTENTION/POTENTIAL: VERB + kay (27) Mwen kay ni yonn. ‘I will/shall have one.’ (28) Mwen kay fè’y. ‘I will/shall do it.’ CONDITIONAL: VERB + té + kay (29) Mwen té kay ni yonn. ‘I would have one.’ (30) Mwen té kay fè’y. ‘I would do it.’ FUTURE PROGRESSIVE Stative: (not applicable) Nonstative: VERB + kay + ka (31) Mwen kay ka palé. ‘I will be speaking.’ (32) Mwen kay ka fè’y. ‘I will be doing it.’ CONDITIONAL PROGRESSIVE Stative: (not applicable) Nonstative: VERB + té + kay + ka (33) Mwen té kay ka palé. (34) Mwen té kay ka fè’y.

‘I would be speaking.’ ‘I would be doing it.’

248 David B. Frank Table 2. The meanings of possible TMA combinations té

kay

ka PRESENT for statives PAST for nonstatives PAST for statives PLUPERFECT for nonstatives

+ +

FUTURE/INTENTION/POTENTIAL +

+

+

CONDITIONAL +

IMPERFECT/PAST PROGRESSIVE

+

+

FUTURE PROGRESSIVE

+

+

CONDITIONAL PROGRESSIVE

+ +

HABITUAL for statives PRES PROG/HAB for nonstatives

In assigning labels to particular morphemes, the term “anterior” is a convenient reminder that preverbal té denotes “past” according to a certain frame of reference — either “past” or “past past”. As already noted, in some contexts té + VERB denotes “past” but in other contexts té + VERB denotes “pluperfect”.¹³ The summary of TMA usage given above indicates that simple past reference for stative verbs is realized as VERB + té, but since descriptive clauses are verbless, the predicate consists of té by itself. In other instances of descriptive clauses as well, VERB + TMA should be interpreted as simply TMA. According to Comrie, “[t]he perfect indicates that the past situation has current relevance (i.e., relevance at the present moment), while the simple past does not carry this element of meaning” (1985: 25). In SLC as in many other languages, perfect is not a distinct category. Because of limitations on the complexity of the predicate, some of the putative combinations of TMA particles do not occur in natural speech. For example, as noted in the preceding list, Mwen kay ka palé ‘I will be speaking’ can be used for future progressive, but in natural discourse the simpler future form Mwen kay palé ‘I will speak’ is preferred. Sequences of all three TMA particles (té + kay + ka) are rare. Spears (1990: 125) establishes a distinction between Haitian definite and indefinite futures, which are marked, respectively, by ap (which seems to correspond to SLC ka), and va (corresponding to SLC kay). However, this distinction does not apply to SLC. Although preverbal ka in SLC can refer to a future action in discourse, a general rule distinguishing indefinite from definite future does not obtain in SLC. Other French creoles such as Guadeloupean seem to have two future markers (ké and kay, as shown in Poullet, Telchid & Montbrand 1984), but SLC does not.

TMA and the St. Lucian Creole verb phrase 249

Spears argues that the unmarked form is sometimes used for present reference with nonstative verbs in Haitian. This goes against Bickerton’s claim that unmarked forms are used for past reference with nonstative verbs in creole languages. Spears provides as evidence examples such as Li palé franse ‘He speaks French’ (1993: 261). This usage would be more accurately described as “habitual/occasional/continuous” than as “present tense”. In SLC ka is used in this context: I ka palé fwansé. The stative/nonstative pattern is consistent in SLC, with some discourse-related exceptions as noted in Section 5. Note also that the patterns that have been described here hold true for independent clauses, but other specific TMA rules apply in various types of dependent clauses. For adverbial clauses beginning with pou ‘in order to, in order that’, lè ‘when’, padan ‘while’, etc., the meanings of the TMA particles in these contexts and the cooccurrence rules remain to be determined, but they are not necessarily the same as in independent clauses.¹⁴

5. Discourse considerations The meanings listed in Section 4 pertain to what Winford (2001: 3) calls the dominant, primary, or prototypical uses of the TMA particles. But Winford adds that there are sometimes other “interpretations that arise from contextual uses” (Winford 2001: 4). Or as Spears puts it, in the use of the TMA markers there is “non-isomorphism in some cases between form and meaning” (1993: 261). One discourse pattern that has to be taken into consideration when analyzing a text is that reported speech, even in the case of an indirect quotation, does not follow the same time orientation as the rest of the text. Anything that follows the VERB di ‘said’ as its object will follow its own temporal deixis until the reported speech is completed. (For examples, see Text 1, sentences 2, 5, 10, 16, 30.) In English one might say My mother said she didn’t know what happened, but in SLC one would say, Manman mwen di i pa sav ki sa ki fèt, which can be translated literally as ‘My mother said she doesn’t know what happened’ (Text 1: sentence 30). Pollard (1989) offers an elegant argument that the preverbal marker en in Jamaican Creole marks background information as opposed to foreground information in discourse. The analog of JC en in SLC would seem to be the té tense marker. While it is true that when applied to nonstative verbs, té would have the meaning of “pluperfect” and would thus refer to background information, Pollard’s argument cannot be extended to explain the pattern reflected in SLC. There are, however, noteworthy correspondences between TMA patterns and discourse context in SLC. A narrative text that has an overall time orientation of “past” can begin with an “opening as preview” or “opening as stage” component that is given as if it were present time reference, even though it refers to past events (Frank 1990). Text 2 provides several examples. There, sentences 1–8 constitute the opening of the narrative before the main part of the plot begins. This part of the narrative is told as

250

David B. Frank

present reference, even though the person whose actions are being described was long dead at the time of telling. Note that the present reference pattern applies to both stative and nonstative verbs. The stative verbs sé in sentences 1 and 8, ni in 3 and 8, konnèt in 4, and kontan in 6 are given in present reference, as are the nonstative verbs ka kwiyé in sentences 1 and 5, ka alé in 2 and 3, ka antwé, ka bwè and ka manjé in sentence 3, and ka fè in 4.¹⁵ In this same text, there is a definite shift to the normal, expected past time reference beginning in sentence 9 with the words yon jou ‘one day’.¹⁶

6. Conclusions In SLC, the preverbal tense, mood and aspect markers work in a predictable way in conjunction with a stative/nonstative distinction to denote time reference. The SLC pattern described here conforms for the most part to the universal pattern proposed by Bickerton for creole languages. In some cases, however, the rules for SLC differ from those that have been described in analyses of other Caribbean creoles. In order to fully account for TMA usage in SLC, the relevant context must be taken into consideration, such as whether the verb phrase in question is in an independent or dependent clause, in reported speech, or in a marked portion of a narrative.

Notes 1. The findings reported here are based on fieldwork conducted in St. Lucia from 1984 through 2000 under the auspices of SIL International. I wish to thank my wife, Lynn, SIL colleagues Paul and Cynthia Crosbie, and colleagues Peter Samuel and Mano Leon, who are native creole speakers, for their contributions to our understanding of St. Lucian Creole. I wish to thank two anonymous reviewers, and Geneviève Escure and Armin Schwegler for their helpful suggestions on the analysis and presentation of this article, and also Betty Eastman for editorial help. And I especially wish to thank May Joseph and the late Ma Benjamen for their friendship and contribution to this study in the form of textual language data. 2. Carrington’s fieldwork was carried out in St. Lucia from 1964 to 1967. His grammatical description of St. Lucian Creole was completed in 1968 and then published as a monograph in 1984. 3. I left “la_glitè” as specified in the original text. The first vowel is probably nasalized. 4. Welmers uses the term factative to describe a pattern he found in many Niger-Congo languages, explaining that “the construction expresses the most obvious fact about the verb in question, which in the case of active verbs is that the action was observed or took place, but for stative verbs is that the situation obtains at present” (1973: 346–347). He refers students of African languages to an earlier publication (Welmers 1946), where he first uses this term. 5. For more on the distinction between sa and pé, see Carrington (1984: 120–122). There are other SLC functors as well, including ja ‘already’, dwé ‘must’, and fin ‘just’, that should probably be analyzed as aspect or mood markers. But all of these, including sa and pé, are outside the focus of this article and therefore are not examined here.

TMA and the St. Lucian Creole verb phrase 6. Sometimes té indicates mood, referring to a hypothetical world, such as in a dependent clause after si ‘if ’ as in Si mwen té sav sa . . . ‘If I had known that. . .'. This underscores the notion that it is not the meaning of the specific morphemes that we need to clarify in cases like this, but rather the meaning of the constructions in which they occur. 7. There are exceptions to this general rule according to the particular discourse context, described in Section 5 below. 8. I constructed all of the sample sentences presented in this paper. These sentences are based on my intuition as a second-language speaker of SLC with nearly seventeen years of immersion in creole. The examples were then checked by native speakers. TMA patterns are best illustrated in natural speech data, such as the two texts presented in the appendix to this article. 9. Carrington labels ka “non-completive”. Again there is a problem with attempting to create a label to cover each of the meanings and uses assumed by a TMA marker. While getting sick can be “punctual”, being sick would never be punctual (in the technical sense of aspect terminology), so it does not make sense to say that ka makes “being sick” nonpunctual. Carrington’s label “non-completive” does not work much better, except that he does explain that ka “suggests habitual, semi-habitual, occasional, continuous or incomplete states, attributes, actions or locations” (1984: 118). 10. See, for example, Bickerton (1975: 30). 11. Seuren gives several arguments, including some that are specific to Sranan and some that apply to SLC as well. Among these is the fact that a form of the copula appears at the end of the sentence in the case of WH-questions. Seuren is not reacting to Bickerton specifically, but to earlier claims made by Voorhoeve (1957, 1962). 12. This broad category of clauses includes a variety of subtypes depending on the occurrence of such factors as direct objects, indirect objects, complements, etc., but such considerations are outside the scope of the present analysis of TMA markers, time reference, and the stative/nonstative distinction. 13. “The meaning of the pluperfect,” according to Comrie, “is that there is a reference point in the past, and that the situation in question is located prior to that reference point, i.e. the pluperfect can be thought of as ‘past in the past’ ”(1985: 65). 14. Maher notes that in St. Barth Creole, as in Dominican Creole, “té normally expresses anteriority but after pou it does not carry that meaning” (1993: 413). The same holds true in SLC purposive clauses. 15. Several verbs in the opening are in dependent clauses in sentences 1–8 and thus do not follow the patterns for independent clauses: alé in 3 and 8, kwiyé in 6, and jwenn in 8. 16. For further examples of this sort of shift in time reference from present to past as a discourse strategy after the opening of a narrative, see Frank (1990). 17. The text by May Joseph was originally published in a book of vernacular stories entitled Jou Lavi Nou [“Days of our Lives”] (Frank 1989).

25

252

David B. Frank

References Bickerton, D. 1974.“Creolization, linguistic universals, natural semantax and the brain.” University of Hawaii Working Papers in Linguistics 6.3: 125–141. Bickerton, D. 1975. Dynamics of a Creole System. New York: Cambridge University Press. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Byrne, F. & J. Holm (eds.). 1993. Atlantic Meets Pacific: A Global View of Pidginization and Creolization. Amsterdam: John Benjamins. Carrington, L. D. 1968. St. Lucian Creole: A Descriptive Analysis of its Phonology and MorphoSyntax. PhD Thesis, University of the West Indies. Carrington, L. D. 1984. St. Lucian Creole: A Descriptive Analysis of its Phonology and MorphoSyntax. Hamburg: Buske. Chung, S. & A. Timberlake 1985. “Tense, aspect, and mood.” In Language Typology and Syntactic Description. Volume 3: Grammatical Categories and the Lexicon, T. Shopen (ed.), 202– 258. Cambridge: Cambridge University Press. Comrie, B. 1976. Aspect. Cambridge: Cambridge University Press. Comrie, B. 1985. Tense. Cambridge: Cambridge University Press. [Frank, D. B., compiler]. 1989. Jou Lavi Nou. Castries, St. Lucia: SIL. Frank, D. B. 1990. “The structural organization of St. Lucian French Creole narrative texts.” Work Papers of the Summer Institute of Linguistics in St. Lucia 2. Maher, J. 1993. “Antillean Creole on St. Barthélémy.” In Byrne & Holm, 409–417. Pollard, V. 1989. “The particle en in Jamaican Creole: A discourse-related account.” English World-Wide 10: 55–68. Poullet, H., S. Telchid & D. Montbrand. 1984. Dictionnaire des expressions du créole guadeloupéen. Fort-de-France: Hatier Antilles. Quirk, R., S. Greenbaum, G. Leech & J. Svartvik. 1972. A Grammar of Contemporary English. London: Longman. Seuren, P. 1986.“The adjectival copula in Sranan.” Paper presented at the Sixth Biennial Conference of the Society for Caribbean Linguistics, Trinidad. Spears, A. K. 1990. “Tense, mood and aspect in the Haitian Creole preverbal marker system.” In Pidgin and Creole Tense–mood–aspect Systems, J. Singler (ed.), 119–142. Amsterdam: John Benjamins. Spears, A. K. 1993.“Stem and so-called anterior verb forms in Haitian Creole.” In Byrne & Holm, 261–275. Voorhoeve, J. 1957. “The verbal system in Sranan.” Lingua 6: 374–396. Voorhoeve, J. 1962. Sranan Syntax. Amsterdam: North-Holland. Welmers, W. 1946. A Descriptive Grammar of Fanti [Language Dissertation 39]. Language 22.3 supplement. Welmers, W. 1973. African Language Structures. Berkeley: University of California Press. Winford, D. 2001. On the Typology of Creole TMA Systems [Society for Caribbean Linguistics Occasional Paper 29]. St. Augustine: SCL.


Text 1. An Bagay Ki Fèt Lè Mwen Té Fèt A narrative told by Ma Benjamen (1914–1999) (1) Manman mwen di mwen kon sa, lè mwen fèt piti ti manmay i té mayé épi papa mwen. (2) I di mwen, pwèmyé moun i té enmen sé papa mwen. (3) Papa mwen fè zanfan épi’y avan yo té mayé. (4) Fanmi papa mwen pa té vlé i mayé épi manman mwen pas yo di, akwèdi pou yo menm yo té gwan jan èk manman mwen té twò malèwèz. (5) Mé papa mwen di i pa ka mayé épi pyès kalité moun pasé Lousi paski sé li i konnèt. (6) Lè i mayé épi manman mwen, manman mwen di i fè tout ich li èk papa mwen.

My mother told me thus, when I was made [i.e., born] a little child she was married with my father. She told me, the first person she loved is my father. My father made children with her before they were married. My father’s family did not want her to marry with my father because they said, as though for them they were big people and my mother was too poor. But my father said he is not marrying with any kind of woman other than Lucy because it is she he knows. When he married with my mother, my mother said she made all her children with my father. (7) Lè i vini akouché mwen — dènyé ich papa When she came to give birth to me — the mwen — tout ich papa mwen fèt blan, mé last child of my father — all of of my mwen fèt nwè ék bèl chivé pla épi fidji father’s children were made white, but I was made black with nice straight hair mwen té benyen favowi. and my face was full of sideburns. (8) Papa mwen pa té la an monman-an, i té My father was not there at the time, he an bitasyon. was on the estate. (9) Manman mwen di lè i akouché i tann an My mother said when she gave birth moun di, “Hmmm! Man Alèksann fè an she heard someone say, “Hmmm! Mrs. kou.” Alexander made a coolie.” (10) I di lè i tann sa, tjè’y pa soté pis i di i sav i She said when she heard that, her heart pa fè pyès kou. did not jump because she said she knows she did not make any coolie. (11) Moun ki pa té janmen vini wè’y vini wè’y People who had never come to see her lè i akouché. came to see her when she gave birth. (12) Yo vini ka gadé mwen. They came looking at me. (13) Yo ka gadé mwen épi yo ka di, “Hmmm.” They are looking at me and they are saying, “Hmmm.” (14) Manman mwen di mwen papa mwen My mother told me my father told her di’y moun mouté an bitasyon èk yo di’y, people came up to the estate and they told “Misyé Alèksann, madanm ou fè an ti him, “Mr. Alexander, your wife made a manmay, an ti fi . . . mé awa!” child, a girl . . . but oh no!” (15) Papa mwen di, “Zòt sa di an wenn si zòt My father said, “You can tell a queen vlé. Mwen konnèt zafè mwen.” [idiom] if you like. I know my business.”

253

254

David B. Frank (16) Pis tout ich li ka fèt blan, èk mwen menm, manman mwen di mwen sété an Endyen pi èk i té ni Endyen ki ka twavay bay papa mwen. (17) Lè papa mwen vini, i di papa mwen pa janmen fè sa lè i akouché ich li. (18) I di papa mwen vini, i débatjé asou chouval li. (19) I mété abousak li la, i antwé an chanm-lan. (20) Pwèmyé bagay i fè i dékouvè mwen asou kouch-la. (21) Lè i dékouvè mwen i anbwasé manman mwen. (22) I déviwé kouvè mwen, épi manman mwen di mwen papa mwen anbwasé’y èvèk i bo’y, i bo’y, i bo’y épi i di manman mwen “Ou fè byen!” (23) Mé manman mwen pa té sav sa vlé di sa. (24) Èvèk i sòti andidan-an, èk i viwé antwé ankò. (25) Fèt mwen fèt wi, manman mwen di mwen apwochan dimi jouné i té ni dépi mwen té fèt. (26) I pwen mwen, i voplé mwen èk i di manman mwen, “Mwen ka vini.” (27) Manman mwen di, “Koté ou ka alé épi ti manmay mwen-an Alèksann?” (28) I di, “Mwen di’w mwen ka vini!” (29) Èk i alé épi mwen. (30) Manman mwen di i pa sav ki sa ki fèt dèyè do’y. (31) Lè i viwé papa mwen viwé mennen mwen ba li. (32) Èk i bo mwen ankò. (33) Èk i di, “Lousi, an tout ich mwen, mwen simyé sala pasé pyès.” (34) Manman mwen di, “Ès si sé kon i nwè-a?” (35) I di, “Sé kon i nwè-a mwen enmen’y kon sa. Èvèk i ni bèl chivé.” (36) Nou tout ni bèl chivé mé mwen té Endyen pi. (37) Pis yo jik di papa mwen sé Tjals ki papa mwen.

Because all of his children are made white, and I myself, my mother said I was a pure Indian and there were Indians who are working for my father. When my father came, she said my father never did that when she gave birth to her children. She said my father came, he got down off of his horse. He put his haversack there, he entered the room. The first thing he did, he uncovered me on the bed. When he uncovered me he embraced my mother. He covered me back, and my mother said my father embraced her and he kissed her, he kissed her, he kissed her and he told my mother, “You did well!” But my mother did not know what that wanted to say [i.e., meant]. And he went out, and he returned entered again. Made I made yes [i.e., I had just been born], my mother said approaching a half day it was since I was made. He took me, he wrapped me and he told my mother, “I am coming [right back].” My mother said, “Where you are going with my little child Alexander?” He said, “I told you I am coming!” And he went with me. My mother said she doesn’t know what happened behind her back. When he returned my father returned brought [i.e., brought back] me to her. And he kissed me again. And he said, “Lucy, among all my children, I prefer that one over any.” My mother said, “If it is [i.e., Is it] because she is black?” He said, “It is because she is black that I love her like that. And she has nice hair.” We all have nice hair but I was pure Indian. Because they even told my father it is Charles who is my father.

TMA and the St. Lucian Creole verb phrase (38) Tjals sété an Endyen ki té ka twavay bay Charles was an Indian who was working papa mwen. for my father. (39) Apwé sa papa mwen di manman mwen i After that my father told my mother he mennen mwen bay papa’y Vélinò. brought me to his father Velinor. (40) Sa sé gwanpapa mwen. That is my grandfather. (41) I di lè i mennen mwen bay papa’y, i di He said when he brought me to his father, papa’y, “Mi ich mwen, mwen mennen he told his father, “Look at my child I ba’w.” brought to you.” (42) I di, “Alèksann ou pa janmen mennen He said, “Alexander, you did not ever pyès ich ou ban mwen en ou mennen sa’a bring any of your children to me and ban mwen — sé an kòz?” you brought this one to me — is there a reason?” He said when my grandfather uncovered (43) I di lè gwanpapa mwen dékouvè mwen anba ti lenn-an, i di dé zyé papa’y vini me under the little cloth, he said my plen dlo èvèk i bo mwen èk i di, “Bondyé, father’s two eyes [be]came full of water and he kissed me and he said, “God, I was mwen té ka mandé’w an ich kon sa èvèk asking you for a child like that and you ou ban mwen’y.” gave it to me.” (44) I di, “Sa sé manman mwen tjwa, tjwa, tjwa! He said, “That is the spit and image of my Ou kay bay non’y Mari en! Èvèk alé èk mother! You will give her the name of kité’y ban mwen. Sa’a sé sa mwen.” Mary! And go and leave her to me. This one is mine.” (45) Mwen yo té di ki pa ich yo pyès-la, pa [It is] I they had said who is not their ich papa mwen, mé gwanpapa mwen vlé child at all, not the child of my father, but mwen. my grandfather wants me. (46) Papa’y di mwen ka sanm manman’y. His father said I resemble his mother. (47) I di tout fanmi’y sòti koulè’y mé He said all his family came out his color manman’y té menm koulè èk mwen. but his mother was the same color with me. So my mother said, since that day my (48) Kon sa manman mwen di, dépi jou sala papa mwen enmen mwen, i ka apwésyé father loves me, he appreciates me mwen magwé moun di mwen pa ich li, although people said I am not his child, mwen sé ich yonn yo ka kwiyé Tjals Kouli. I am the child of one they call Charles Coolie. (49) Lè mwen koumansé vini gwan mwen When I began to [be]come big I began to koumansé chanjé koulè abwézan. change color now. (50) N’a vini wouj a mizi. I am [be]coming red gradually. (51) Chak lè papa mwen ka vini i ka di, “Ich Every time my father comes he says, “My mwen, wété nwè on.” child, stay black, okay?” (52) M’a té nwè, nwè, nwè mé mwen té an ti I was not black, black, black but I was a koulè lapo nwè. little black color skin. (53) Èk lè papa mwen sòti Sent Lisi i alé And when my father left St. Lucia he Kayenn, i èkwi, i di bay manman mwen, went to Cayenne, he wrote, he said to my yon sèl wigwètman i ni, i oblijé alé mother, only one regret he has, he must Kayenn, i kay kité mwen mé i pa kay wè go to Cayenne, he will leave me but he mwen ankò. will not see me again. (54) Mé i enmen mwen. But he loves me.

255

256

David B. Frank (55) I di sé mwen ki kay ni pou fèmé zyé manman mwen lè i mò. (56) Épi kon ki di kon ki fèt. (57) Lè manman mwen mò, lanmen’y té anlè tèt mwen, i di mwen i anvi dòmi. (58) Mwen tiwé lanmen’y alé tèt mwen èvèk i mò. (59) Ès si ou wè mannyè moun ka di sa yo vlé? (60) Mwen pa té ich papa mwen pyès davwa mwen pa sòti menm kon sé lézòt-la. (61) Mé manman mwen di, papa mwen té enmen mwen pasé pyès. (62) Pli mwen ka vini gwan sé pli mwen ka sanm papa mwen.

He said it is I who will have to close my mother’s eyes when she died. And as was said is as it happened. When my mother died, her hand was on my head, she told me she wants to sleep. I removed her hand from my head and she died. If you saw [i.e., Do you see] how people say what they want? I was not my father’s child because I did not come out the same as the others. But my mother said, my father loved me more than the others. The more I am [be]coming big is the more I resemble my father.

Text 2. Mimin, ‘Soup Labé’ A narrative told by May Joseph ¹⁷ There is an old woman of Choiseul they are calling Mimin. She is always going into the presbytery, she will see the priest. But before she went see the priest, she is entering into the kitchen of the priest and she is drinking the soup of the priest and she is eating all the priest has there. So nou konnèt sa i ka fè-a. So we know what she is doing. Nou ka kwiyé’y “Soup Labé”. We are calling her, “Priest’s Soup”. Lè nou kwiyé’y “Soup Labé”, i pa kontan When we called her “Priest’s Soup”, she non sala. doesn’t like that name. Sa i ni, sa i fwapé nou. What she has [is] what she hit us [with]. Mé pou i jwenn nou byen sé lè nou alé But for her to get us good is when we go anba mango’y, piskè i sé yon moun ki ni under her mango [tree], because she is a an chay pyé mango. person who has a lot of mango trees. Yon jou, mwen désann anba pyé mango-a. One day, I went down under the mango tree. Mwen té ka anmasé mango. I was gathering mangos. An anmasé mango, mwen santi an wòch On gathering mangos, I felt a rock fall tonbé dèyè mwen, “Boup!” behind me, “Boop!” Lè mwen tounen, mwen gadé, mwen wè When I turned, I looked, I saw Mimin Mimin ja douvan mwen èk an koutla. already in front of me with a cutlass. Tout sa mwen té sa fè sé kouwi désann an All I could do is run down in the rocks of wòch wavin-lan. the ravine. Mwen kouwi, Mimin kouwi, mwen kouwi, I ran, Mimin ran, I ran, Mimin ran. Mimin kouwi.

(1) I sé on vyé madanm jan Chwazèy yo ka kwiyé Mimin. (2) I toujou ka alé an pwèzbitè, i kay wè pwètla. (3) Mé avan i alé wè pwèt-la, i ka antwé an tjwizin pwèt-la èk i ka bwè soup pwèt-la épi i ka manjé tout sa pwèt-la ni la. (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

TMA and the St. Lucian Creole verb phrase (15) Mwen bay Mimin yon détou, èk Mimin pa wè koté mwen ay séwé. (16) Épi sa fini pasé, mwen vini mouté isi. (17) (18)

(19) (20) (21)

I gave Mimin a detour [i.e., ‘the slip’], and Mimin did not see where I went to hide. And that finished happening, I came up here. An jou mwen désann. One day I went down. Lè mwen wè Mimin, i di mwen, “Bon, jòdi- When I saw Mimin, she told me, “Good, a ou sé an gwan fi, mwen ja gwan moun. today you are a big girl, I am already Pou anbété ou té ka anbété mwen, ban an old person. For trouble you used to mwen kéchòy!” trouble me, give me something!” Mwen di’y, “Ében, sa vwé.” I told her, “Well, that’s true.” Mwen fouyé lanmen mwen an pòch I dug my hand into my pocket and I gave mwen épi mwen bay Mimin on ti Mimin a little change I had there. lanmonné mwen té ni la. Mwen ba’y on ti lajan pou i otjipé kò’y. I gave her a little money for her to take care of herself.

257

CHAPTER 12

The Limonese calypso as an identity marker Anita Herzfeld

The University of Kansas in collaboration with

David Moskowitz

The University of South Dakota

This paper deals with the relation between language and identity among the AfroLimonese of Costa Rica, a people immersed in a Hispanic milieu. Specifically, calypso lyrics are here considered as one of the major factors that aid in implementing the maintenance of Limonese Creole. We explore the hypothesis that Limonese Creole will endure in spite of the Spanish influence, partly because of the Afro-Caribbean musical tradition, which provides a platform for the renovation of the identity of that group. First Limonese Creole is presented within a socio-historical framework, then the paper focuses on music as an identity factor, and on calypso as a musical genre. The final section analyzes several Limonese calypso lyrics in an attempt to substantiate the above-mentioned claim.

1. Introduction¹ In spite of long-standing claims of physical and cultural Spanish homogeneity, anthropologists and other social scientists have recently made great strides in raising Costa Rican (tico) consciousness about the population’s mestizo origin and African ancestry (Bozzoli 1993).² However, the diversity of languages spoken in Costa Rica is hardly recognized beyond academic circles.³ In fact, the shift from indigenous and creole languages to Spanish — Spanish dominance over other mother tongues — is considered an ineluctable fact, and is viewed as a marker of tico (‘Costa Rican’) identity. As a counterpart to the impending linguistic homogeneity, linguists’ efforts to document and maintain linguistic diversity have given rise to an important movement of social action meant to rescue ethnic pluralism.⁴ Multilingualism and ethnic diversity are often perceived by the authorities as impediments to the development of a national identity, because once the link between the dominant official language

260 Anita Herzfeld and David Moskowitz

and the nation is strongly forged, it galvanizes that relationship, condemning linguistic diversity. Thus instead of feeling appreciated, the speakers of languages that are not the official languages of Latin American countries are threatened by homogenizing educational campaigns. For example, as part of an effort to acculturate Jamaican immigrants and to nationalize the Limón region, in 1953 the Ministry of Education of Costa Rica prohibited the so-called “little English schools”. These were mostly church-sponsored schools where Limonese parents sent their children, whose mother tongue was Limonese Creole, to learn Standard English. Furthermore, the Ministry insisted that the children attend public schools where classes were taught in Spanish. Limonese Creole (LC), an English-based language spoken by the Afro-Caribbean minority that settled on the eastern coast of Costa Rica, in the Province of Limón, underwent severe criticism, even by its own members. This paper deals with the possible preservation of the relation between language and identity among the Afro-Limonese, a people immersed in a Hispanic milieu. Specifically, calypso lyrics are here considered as one of the major factors that aid in implementing the maintenance of the language. We explore, in a speculative fashion, the hypothesis that Limonese Creole will endure, in spite of the Spanish influence, partly because of the Afro-Caribbean musical tradition that provides a platform for the renovation of the identity of that group. It may well be that the linguistic activity of this minority is stimulated by musical genres such as calypso, Afro-pop and reggae, all of international renown, with whose messages its speakers can identify. It may not be unreasonable to say that the Afro-Limonese people are speakers of four codes: the creole, Standard English, Spanish, and the musical code! Their love for music pervades every minute of their life; it is apparent as they walk, because they dance and sing as they move; and it is otherwise constantly present, as they carry a boom-box and listen to music wherever they go. We will briefly present Limonese Creole within a socio-historical framework.⁵ Then the paper focuses on music as an identity factor, and on calypso as a musical genre.⁶ In the final section we speculate on the possible future of LC maintenance via the spread of Afro-Caribbean music.

2. The Limonese Creole of Costa Rica The African slaves who arrived in Costa Rica’s Central Valley in the 16th century eventually formed part of the indigenous and Spanish society of the time, and as a result their languages were lost. Meléndez and Duncan (1972: 23) point out that only a limited number of Africans went to Costa Rica at that time. Although no exact figures are available, Aguilar and Alfaro note: The analysis of the “Protocolos Coloniales de la Ciudad de Cartago” have allowed us to verify that the presence of the black element in the colonial capital was considerable and the increased [production] of the cacao plantations led to the legal and

The Limonese calypso as an identity marker illegal sale of slaves, in considerable quantities, because it was impossible not to acquire manpower to maintain the increasing cacao activity although, as we will see later, it was also essential in other agricultural and cattle-rearing activities and in different regions of the Province. (1997: 143, see also p. 183; our translation)

The Afro-Limonese of the second half of the 19th century arrived in Costa Rica, mostly from Jamaica but also from Trinidad, Barbados, and other places, to work on the construction of the railroad that would join San José, the capital, with Port Limón on the Caribbean coast. Once settled in Costa Rica, they were thoroughly manipulated by Minor Keith, a US engineer in charge of the enterprise. Later, when he started the United Fruit Company’s operations in the region, planting and exporting bananas, these immigrants were in the hands of Mamita Yunai ‘Mother United’ (Fallas 1941). They were encouraged by management to use their creole language and not to learn Spanish, so as not to open lines of communication with government officials. As a worker of the United Fruit Company observed at the time, “[o]nly God and Minor Keith know what is going on behind the scenes” (Putnam 1913: 102). This contingent of Afro-Limonese has remained culturally and linguistically different from the Spanish majority until very recently. They never went back to their homeland, and instead remained in Costa Rica permanently to work at the port or on the banana plantations, both properties of the United Fruit Company at that time. It is only now, after more than a century, that mixed marriages have become quite common as a result of Hispanic migration to the coast. The language of the Afro-Limonese, which I have called Limon(ese) Creole, comes from Jamaican Creole, the mother tongue of the majority of Afro-Caribbean immigrants.⁷ It is known by its speakers as /mekaytelyuw/ or /mekatelyuw/, which comes from Jamaican ‘make a tell you’ (Standard English ‘let me tell you’). There are 55,000 speakers of LC, who mostly live in the Province of Limón, although many have recently migrated to the Central Valley (Smith 1995: 341). Most of the AfroLimonese who live in the rural areas of the Province of Limón are likely to use basilectal forms of the creole, but those who have gone through elementary and secondary schooling in Port Limón, the capital of the province, are usually comfortable using mesolectal and acrolectal forms of the linguistic continuum. It stretches from Limonese Creole (LC) to Standard Limonese English (SLE) or Standard English (SE). Speakers of LC believe their language “has no grammar” and is “broken or flat English.” Since the language is only transmitted orally and there is scarcely any written literature in LC, speakers of /mekaytelyuw/ are “ashamed” of it and often avoid speaking it in public. This concept of “second class” language is often transferred to their own self-image, which then turns into a self-deprecating “second-class-citizen” impression, fostered by the apparent discrimination they are subjected to in Hispanic Costa Rica. Despite efforts of the central government to nationalize the region, the linguistic resistance of this minority manifests itself in the survival of its trilingual and tricultural identity. The majority of the /mekaytelyuw/ speakers are Costa Rican citizens

26


who speak Spanish, the creole, and Standard Limonese English. While Spanish is the official language of the country, and has undoubtedly affected the mother tongue of the Limonese people (Herzfeld 2002: 343–82), the loyalty of the Afro-Costa Ricans to their ethnic group is apparent in their use of the creole, in their food, in their games and sports, and in their music and songs.

3. Music as an identity factor Until quite recently, music was considered an autonomous experience unrelated to social issues (Stokes 1994: 1). Anthropologists maintained for a long time that music does not deal with the processes and events that make up “real life”, and therefore they condemned it to a precarious periphery.⁸ However, ethno-musicologists have argued that music is of central importance in indigenous societies (Seeger 1987, Turino 1989).⁹ It is now acknowledged that we can better understand the cosmological system of a society through music and dance. Accordingly, we have to rethink the function of music and dance as relevant forms of social communication. The social and cultural world has been modeled by modernity (by the industrial capitalist world, the nation-state), which today cannot be conceived without music. Music helps us articulate our understanding of other peoples, places, and times, and of ourselves in relation to others.¹⁰ Because music is socially significant, it provides the means by which we can establish and recognize different identities and the limits that separate one group from the “others”. Social actors use music in specific local situations to establish limits, and to maintain distinctions between “us and them”. Just as it is impossible to understand the characteristics of a people without knowing its language since the identity of a group must be seen from the “inside”, the same is true of musical practice. Groups are defined by their ability to articulate the differences between “them and us”, and as identities evolve socially, they cannot leave aside questions of colonialism, domination, and violence. Thus, when dominant groups find minorities that confront or challenge their interests, they suppress them one way or another. Music provides the principal means to handle the intense contradiction that exists socially between an egalitarian ethos and the leadership of factions. Trujillo, the dictator of the Dominican Republic (1939–1961), used merengue lyrics to publicize his political campaign; later when he was deposed, people kept the merengue, but dropped the forced lyrics. Peron used Argentine folklore to fabricate a false concept of populist identity, by making it compulsory for radio stations to play Argentine folklore on a daily basis. These are but two examples of the way in which music has been used to express relations of “domination or subordination” in Latin America. In the West Indies, Carnival celebrations have long been a vehicle for the expression and assertion of Creole and African-Caribbean sentiments dating back to the late 1700s. Modern calypso is tied to the popularization of Carnival. Calypso is “a product of diverse musical influences [. . .] including Creole song, procession

The Limonese calypso as an identity marker

song, neo-African song types and kalinda” (Manuel, Bilby and Largey 1995: 186). In Limón, Carnival, which commemorates Columbus’s landing in the New World (1492), takes place from October 12th, the “Day of the Cultures”, to the 14th. Rather than a religious celebration it is a symbolic public expression of the existence of the Afro-Costa Rican minority.¹¹ There are three accounts of how Carnival and calypso arrived in Limón. One account claims that it was brought over from St. Kitts by the workers hired to build the railroad at the end of the 19th century. They dramatized Bible stories (such as David and Goliath), masquerading on the streets, and they called the festivity St. Kitts rather than Carnival. Another explanation states that both Carnival and calypso were brought over from Panama by Albert King (a Limón-founding representative of Marcus Garvey’s United Negro Improvement Association, UNIA) in 1959 (Franklin Perry, p.c. October 20, 2002). Finally, Palmer (1986: 357) cites Selven Bryant who attributes the introduction of the festivity in Limón to the thousands of workers who had immigrated to Panama to work in the Canal Zone during WWII and returned to Costa Rica later in the 1940s. Be that as it may, Carnival was only a provincial celebration from its inception until the 1970s. It brought together the most famous calypsonians of the time — Edgar Hutchinson (a.k.a.“Olle” or “Pitún”), Walter Gavitt Ferguson (a.k.a. “Mister Gavitt”), and Francisco Flinger (a.k.a. “The Mighty Sparrow”), who was born in Grenada but went to Trinidad as a little boy and is now living in New York. They participated in lively calypso competitions. Carnival in Limón is the major annual event of the province. In the past, people who lived on the eastern coast of Costa Rica would travel on foot, by boat, by train or in the so-called “burrocarril” to join in the festivities. Today Limón’s Carnival is an occasion for thousands of expatriates to return and reestablish social bonds. They often arrive from New York, Miami, Los Angeles, and Panama on charter flights. Los Carnavales de Limón have turned into the most important tourist destination and commercial attraction of the region, even for domestic visitors (Palmer 1986: 357– 8). In addition to the local calypsonians, the event brings musicians from Panama, Puerto Rico, and the United States. Once considered the most outstanding activities of the Carnival, concerts and competitions between local calypsonians have today given way to the globalization of music and are therefore less prominent. The Carnaval de Limón presents parades in which percussive ensembles and dancers — comparsas — compete against each other with elaborate floats, costumes, and the use of clever, taunting songs. These ensembles usually include bongos, tom-toms, a bass drum, congas, and timbaletas, a pair of single-headed drums tuned a fourth apart and set side by side on a metal stand (Fernández 1998: 693). It is impossible to separate the musical aspects of the Carnival parades, machismo, and the musical supremacy of the minority group; they are vital symbols of their pride and their integrity as a community. Music is an invariably communal activity, that can join or segregate people, and the Carnivals of Limón really present the opportunity for the community to come together as a whole before the invasion of “the others”. These “others” may include the tourists who come from the outside, as

263


well as Hispanic Costa Ricans who may not be considered members of the group. Even if the Afro-Limonese have identified themselves with Costa Rica, music and dance are a convenient and appropriate way to support their challenge as a minority. Through calypso and reggae they can formulate their own social order, based on the moral values of their community. Even though calypso singers and players have not succeeded on the world stages the way Bob Marley has through reggae, they are very popular among Afro-Caribbeans everywhere. Traditionally, their lyrics expressed social commentary (criticism and praise), the inherited strands of kalinda (boasting), extemporaneous inventions and other genres. Lately, they have evolved along with society to include a wider range of themes, double meanings, political criticism, current events, different languages, and various musical styles (Winer 1993: 268).

4. Calypso as a musical genre The Caribbean musical style called calypso originated on the islands of Trinidad and Tobago. From there it spread throughout the West Indies. Beginning in the late eighteenth century, calypso evolved among Afro-Trinidadians out of a mixture of African and West Indian folk music. Calypso use in Carnival developed later: “By the latter half of the nineteenth century, the diverse genres [that coalesced to create calypso] came to be centered in pre-Lenten Carnival” (Manuel, Bilby and Largey 1995: 186). While many other islands in the West Indies have forms of calypso, it is in Trinidad and Tobago that calypso and its main venue, the Carnival, initially flourished. The early roots of calypso or kaiso (Winer 1993: 268) are difficult to trace. Scholars and historians of Trinidad have reconstructed some of the history of the genre, but still have not uncovered the origin of the word itself.¹² Some believe that it comes from the word cariso that first appeared in the 1780s, and was used to denote “some sort of satirical, extemporized Creole song” (Manuel, Bilby and Largey 1995: 186). These influences included French Creole songs, masquerade procession songs, neo-African genres like juba and bamboula, Venezuelan string-band music, British ballads, Creole songs from other West Indian islands, and the kalinda. Kalinda is working-class stick-fighting which includes singing, boasting and drum accompaniment, and is a Caribbean variation of the West African stick-fighting dance with musical accompaniment. The calypso tradition came to maturity in the annual Carnival celebration. In Trinidad, Carnival had begun as a French aristocratic festival in which the participants would dress in costumes and masks, and pay each other social calls. Island slave owners also allowed their slaves to celebrate in their own way, and by the mid 1800s the Africans had begun to dominate the event. This Carnival, referred to as jamette Carnival, combined loud street dancing with processions; camboulay processions (a recreation of a fire drill accompanied by drumming), stick-fighting,


masquerade troops, folkloric characters, and costumed stilt-walkers called Moko Jumbies. A point well attested by historians of calypso is that, at the turn of the twentieth century, the language of calypso in Trinidad changed from French Creole (the then popular vernacular language of identity) to the newly acquired English Creole. The linguistic switch did not change the role of calypso as a marker of Afro-Caribbean cultural identity. Twentieth-century Carnival can be divided into two main types: the mas (masquerade processions) and the calypso. The latter evolved into a “more elaborate, textoriented song performed for seated audiences in large tents erected for the occasion” (Winer 1993: 187). Calypso evolved quickly between 1900 and 1930 under the influence of commercialization which greatly affected the codification of styles. Almost all of its components were affected: “tamboo-bamboo” instrumentation (bamboo tubes struck with sticks) was replaced with string and brass bands; the call-and-response declamation style was replaced by solo singing; and the tunes employed were taken from a corpus of some fifty major-key calypso melodies. As mentioned earlier, by that time, the language used to sing them was an English-based creole. In the 1960s calypso had to fight for survival against the pressure of both rock and roll and reggae. A major development that affected calypso was the birth of soca. Its more vital composite pattern replaced the “customary bouncy, slightly ragged, but basically bland and generically Caribbean accompaniment pattern and, in a loosely standardized form, [it] has been the norm in most calypso since” (Winer 1993: 193). Soca is now used to refer to modern calypso and specifically to designate the dance music, in opposition to more traditional calypso, which maintains an emphasis on text. Soca was born in the international climate of the 1980s, and infused calypso with more danceable rhythms and pan-African texts. Beginning in the 1950s and 1960s, calypsonians from Panama and Trinidad, including Lord Cobra, Mighty Sparrow, and Lord Kitchener, became popular in Costa Rica (Fernández 1998: 691). In the recent past, during Carnival, Afro-Limonese calypsonians competed against each other for prizes sponsored by local radio stations and clubs. The highest prizes were awarded to the performers who exhibited the greatest skill in textural and musical improvisation. Today, the participation of calypsonians in the Carnival is not as relevant as it was in years past. This folk activity has succumbed, as many others of its kind, to the commercialization of public musical performances. Yet, it is remarkable that, in spite of the 1940s commercial exploitation of the calypso by the U.S record industry — which precipitated a change in the sound and lyrical content of the style — the “packaged” type of calypso did not totally replace the folk form. Calypsos continue to be written and performed in LC according to their original sound and structure in Costa Rica. It is true, however, that they are generally targeted towards a cultural microcosm, a specific audience that understands LC and expresses itself daily in some form of the linguistic continuum (LC-SLE). In spite of the strong competition waged by reggae among the Afro-Limonese youth, calypso is entrenched among them, since it had taken root in Limón long before the reggae style was even born.

265


The lyrics of Costa Rican calypsos today include a wide variety of topics such as immigrant history, daily life, political statements, or religious doctrine. Guitar, tenor banjo, or a vocal or percussion ensemble provide musical accompaniment for these calypsos. Calypso in many forms is currently quite lively in the musical scene of Costa Rica. Two of the well-known traditional Costa Rican calypsonians — Mr. Gavitt (mentioned above), and Herberth Glinton (a.k.a.“Lenky”) are still active today. Mr. Gavitt (Walter Ferguson), who lives in Cahuita, Limón Province, has recorded his first CD, Babylon, at 83 years of age in 2002. The CD was sold out as soon as it appeared. Ferguson’s songs “combine forms and contents that show the presence and evolution of the Afro-Costa Rican culture” (Monestel 2002: 4). As to Lenky — 68-years old in 2002, and a resident of Port Limón — he is a prolific calypso composer. As a musician, he played in the Carnival calypso groups of the 1960s, accompanying famous comparsas (dance groups) such as the Squelintans and the Langostas (Manuel Monestel, p.c. October 25, 2002). Recently, Manuel Monestel has recorded his music under the heading of “Don’t Stop the Carnaval”. At present, Limonese calypsos are being written and performed by several young groups both during and outside of Carnival time. While traditional calypsos continue to be written, others contain elements of a great number of other popular styles: they are influenced by Jamaican reggae and ska, rap, dancehall, hiphop, and even American rock and roll. Other new features of contemporary calypsos include the use of non-traditional instruments, such as European woodwind instruments (flutes) and modern electronic instruments, such as the synthesizer. All these changes may make Afro-Limonese calypsos very different from those originally brought to Limón. However, the result of such use of technology is not negative, since it allows the modern calypso to form part of the synchronic dynamic that flirts thematically with more contemporary rhythms, such as those mentioned above. In addition, several contemporary Costa Rican calypsonians retain many aspects of the original style.¹³

5. Calypso lyrics as facilitators of survival How can calypso music help the survival of LC? Similar to reggae in Jamaica, calypso has become the most effective messenger of the Afro-Limonese culture. Calypso lyrics often deal directly with its recognition and acceptance. The old and young generations of Afro-Limonese are immersed in this music, which transmits their message through their language. Through Calypso lyrics, feelings of anguish and pain as well as happiness and hope are shared and appreciated by the AfroLimonese in a manner that only people from the “inside” can experience. We included in this study five CDs and one cassette tape. Some of the CDs contained transcribed lyrics in their jackets, and other calypsos had been previously transcribed (Palmer 1986: 360–71). All the calypsos were recorded by young Costa Ricans — most of them Afro-Limonese.¹⁴


The songs that form part of the musical scene of Costa Rica today allude to a great variety of topics, including everyday life, the discrimination the Limonese minority feels subjected to, the celebration of Carnival, longing for a “real” homeland, confrontations with the local authorities; love for music and Anansi stories, and religious interpretation of events. The songs selected here are representative of the significance of calypsos as markers of identity for the Limonese “imported minority”. They address the essential worldview values held by the Afro-Limonese community, and include “In dis generachian”¹⁵,“Back to my country”,“Monilia”,“Everybody running to the Carnaval”,“Cabin in de water”, and “Tacuma and Anansi’s Party”. The numbers inserted in the comments on the lyrics refer to the specific lines to which these comments apply.

5.1 Socio-political commentary calypso “In dis generachian” was recorded by a group called Mekatelyu (CD title Comin nao), SE /Mekatelyuw/ ‘Coming now’ — the CD has no studio label.¹⁶ The lyrics express the frustration of the Afro-Limonese at the conditions in which they live. The calypso also emphasizes the antipathy felt by the Hispanic oligarchy towards the AfroCosta Rican minority. Furthermore, the song implores the friend (the understanding human being) who passes by to open his soul to its message of peace and not to engage in armed conflict (5). This calypso falls into the category of socio-political commentary since it expresses the despair of a marginalized group of people who live in an impoverished area, and who, though they do not criticize the government harshly, feel abandoned by its officials. It exemplifies calypso lyrics that have become a powerful vehicle for social protest and political consciousness, placing such issues at the very center of people’s concerns, thus appealing to the common man and woman. The lyrics are conveyed here first in a broad phonemic transcription of Limonese Creole — as it usually appears on the jacket of the CDs — followed by a Standard English interpretation of the lines. Code switching between Limonese Creole and Spanish is apparent in the refrain. “In dis generachian”, Refrain (no jacket) (1) In dis generachian, evritin na go so fayn[a] ‘In this generation, everything does not go so fine’ (2) Aal di gavermen jes wan tu mashop di kana ‘All the government just wants to (do is) destroy the (little) corner (of the world where we live)’ (3) Dis iz a tay dat wi neva let go ‘This is a knot that we will never be able to let go’

267


(4) Bikoz wi livin ina dis kayna wol; ‘Because we are living in this kind of world;’ (5) DIGO, fren, ¨QUE ESTÁ PASANDO? QUE LLEGUEN MIS PALABRAS¹⁷ ‘I say, friend, what is going on? (It is my hope) that my words (will) reach you’ (6) COMO PROYECTIL rayt down tu yor ALMA. ‘Like a projectile right down to your soul.’ (7) PAREN LA VIOLENCIA, DIGO, PAREN LAS ARMAS! ‘Stop the violence, I say! Give up (stop) the weapons!’ The fact that the refrain is expressed both in LC and Spanish and that there is code switching is a clear indicator that Afro-Limonese speakers are bilingual. However, it is obvious that LC is still very much part of their identity, or else Standard English would have been employed here. Calypso lyrics have helped preserve the use of mesolectal creole. One possible reason is that calypso is closely associated with Trinidad, mostly a mesolectal creole community, while reggae has been more instrumental in preserving the basilectal creole, more frequently spoken in Jamaica.

5.2 Identity calypso A group called Calypsos recorded Afro-Limonese Music of Costa Rica under the label Lyrichord. One of their calypsos is “Back to my country”, transcribed here from the jacket: “Back to my country” (8) I say mi neighbor want to kill me ‘I say my neighbor wants to kill me’ (9) Dis is only way and dis is very true; ‘This is (the) only way, and this is very true;’ (10) You better send me back to my country ‘You had better send me back to my country’ (11) Send me back to my native land ‘Send me back to my native land’ (12) Dat is the t’ing that I cyaant understand. ‘That is the thing that I cannot understand.’ When they arrived in Costa Rica in the 1870s to build the railroad that would join San José to Limón, Jamaican immigrants planned to return to Jamaica once they accumulated enough funds to support their families back home. However, they never managed to do so and, until 1948, they actually lived in limbo regarding the legal-

The Limonese calypso as an identity marker 269

ity of their immigrant status. It was then that José (“Pepe”) Figueres, after becoming president of Costa Rica, granted Costa Rican citizenship to the Afro-Caribbeans who had lived and worked in Costa Rica for over half a century. This action was long overdue, since many had given their life for their adopted country, fighting malaria and dysentery while building the railroad. Throughout their stay prior to that, the Afro-Limonese had been emotionally loyal to the British Queen, had worked for the United Fruit Company, and had moved comfortably along the speech continuum from LC to SLE while practicing Protestantism — among other religious cults, such as obeah and pocomía — but they did not speak Spanish. Thus they felt isolated from the dominant majority culture and sensed that Hispanic Costa Ricans did not trust them and ostensibly discriminated against them. After the 1950s the situation changed for the Afro-Limonese; they learned Spanish, and felt a greater sense of being “Costa Ricans” as a result of strong nationalistic sentiments that swept through Limón. Many even became Catholic. The lyrics of this calypso show, however, that after all these years, the pangs of considering themselves foreigners in their own land still create a gnawing emotional pain.“Back to my country” clearly indicates the feeling of being “persecuted” or “unwanted” (8). The song also sheds light on the desire to escape this reality, which they cannot understand (12) by going back to Jamaica — the dream that never came to be — where such discrimination would not occur (11). The lyrics of this calypso encapsulate this social group’s most essential values — the concept of ethnicity as defined by the members of the group themselves. Furthermore, the calypso is reflective of power relations, “us versus them”, and short of expressing effective symbols of vindication opts for suggesting the ultimate elimination of their political and economic subordination and exploitation by leaving the scene of their suffering.

5.3 Religious appeal calypso From a group called “Cantoamérica”, and their CD entitled Calypsonians, four songs are transcribed below: “Monilia”,“Everybody running to the Carnaval”,“Cabin in de water” [sic], and “Tacuma and Anansi’s Party”. This group was formed through personal associations developed at the University of Costa Rica in the 1980s, and it has since performed in Latin America, Europe, the United States, Mexico and Canada. When they are not on tour, they play for fund-raiser functions, and at a smoky bohemian alcove [. . .] in San Pedro. Usually a noisy bar, it simmers down as Monestel blends his soft voice with the syncopated rhythms of flute and vocals [. . .] The group is renowned for the interesting use of the timbal, bongos, maracas, güiro, claves, cowbells, conga drums, ocarinas and other native Costa Rican instruments. (Costa Rica Today, June 18, 1992)

In“20 años de Cantoamérica” (‘20 years of Cantoamérica’),Víctor Hugo Acuña writes in Ancora (the cultural supplement of the Costa Rican paper La Nación, October 8,


2000), that under the direction of Manuel Monestel, this group has linked the political and esthetic concerns of the new Latin American song with the sounds and rhythms of Caribbean music. Monestel has managed the emancipation of Costa Rican rhythms in a very special fashion by rescuing Limonese music, that is to say, the calypso, and Walter Ferguson’s compositions. It was a risky option since at the time when Monestel proposed to integrate the Afro-Limonese sound to the Costa Rican musical scene, Limón was just starting to become integrated into Costa Rica. By doing so, Monestel incorporated not only the music, but also the cultural identity of the Afro-Limonese to the Central Valley of Costa Rica (which considers itself white, Spanish speaking, and Catholic). Monestel, who is a sophisticated musician and well acquainted with Latin American and international music, has also superimposed the most recent developments of jazz, pop, and rock on the Afro-Caribbean rhythms. Thus, in these last twenty years, Monestel has substituted the sociocultural images of European Costa Rica for a multicultural society of three roots: the Amerindian, African and European. The first song transcribed below is “Monilia”, by Walter Ferguson. Its lyrics glance at the daily life of an immigrant’s history. They relate the devastation that Monilia roreri, an airborne fungus, wreaked on the region’s cacao plantations in the 1970s and 1980s (13), bringing about poverty and hunger (14). The calypso provides a religious interpretation of the events, and cautions audiences to seek repentance (23)– (28) (Fernández 1998: 692). The song refers to a woman whose name was Irene (17); she belonged to a well-to-do family (18), but the monilia plague affected their cacao plantation (19). Consequently, they lost everything and ended up by having to sell the property (20). The lesson to be learned from this experience, particularly by those who are not good Protestants and never read the Bible (25), is to become believers and get down on their knees to pray for sympathy from their Maker, or else they will hasten punishment (27, 28). The transcription used here comes from the jacket, which actually mixes SE words with LC syntax, and from Fernández (1998: 692). “Monilia” (by Walter Ferguson), Refrain: (13) Monilia come to stay ‘Monilia (you have) come to stay’ (14) And all you bring ES [sic] hungry belly ‘And all you bring is (a) hungry belly’ (15) You say you no goin’ no way ‘You say you (are) not going (to go) away’ (16) Til you bring me down to poverty. ‘Till you bring me down to poverty.’ (17) I knew a woman she name Irene, ‘I knew a woman who is called Irene,’


(18) She had a mighty family. ‘She had a well-to-do family.’ (19) Monilia plague the gal until she walkin’ lean ‘Monilia plagued the girl until she “walked lean” (lost her wealth)’ (20) She had to sell she property. ‘She had to sell her property.’ (21) Ladies and gents come listen to me ‘Ladies and gentlemen, come (and) listen to me’ (22) I want you all to understand ‘I want you all to understand’ (23) Monilia is a power from a high degree, ‘Monilia is a power from a High Source’ (24) And then it comes to kill off every man. ‘And then (therefore) it comes to kill everybody.’ (25) Who never read the Bible ‘Whoever (does) not read the Bible’ (26) I say they read it now contentedly ‘I say they read it happily now’ (27) Get down on they knees and they start to pray ‘Get down on their knees and they start to pray’ (28) Beg their master for sympathy. ‘(And) beg their Master for sympathy.’ Religion is a very important part of the lives and identity of the Afro-Limonese. Port Limón, the capital of the Province of Limón, can be considered truly cosmopolitan in as much as it has a remarkable number of churches of different denominations. On Sundays, in addition to going to church, families gather around the kitchen table and read the Bible. Sermons are now conveyed in LC, SE and even in Spanish, or a mixture thereof, catering to a diverse audience of believers (Herzfeld 2002: 47–8). Thus it is not surprising that the lyrics of this calypso make reference to an important aspect of Afro-Limonese spiritual life. Since the religious outlook of the community is universal — whether someone is a Catholic or a Methodist or an Anglican or a practicing member of a Pentecostal church — the dominant way of processing identity from this standpoint is that of non-conditional inclusiveness. Everyone who feels competent, be it linguistically, melodically, rhythmically or choreographically, may join in to share in this religious identity which is all pervading in the life of the Afro-Limonese. Thus this calypso’s lyrics would appeal to everyone who celebrates the power of the Supreme Being over human matters.

27

272

Anita Herzfeld and David Moskowitz

5.4 Carnival calypso The same group recorded “Everybody running to the Carnaval”, a Walter (“Mr. Gavitt”) Ferguson song. Mr. Gavitt recorded his calypso “Monilia” in an album produced by the Costa Rican Ministry of Culture, Youth, and Sports. Folkway Records of the US also produced a long play of his original songs entitled “Calypsos of Costa Rica”. The transcriptions presented here are taken from Palmer (1986: 360) rather than from the jacket, because they more accurately reflect the use of the creole. The SE translation is based on the Spanish translation offered by Palmer who acknowledges the contributions of Quince Duncan and Manuel Monestel in producing it (Palmer 1986: 360). “Everybody running to the Carnaval” (by Walter Ferguson) (29) Everybody running to de Carnival, I say ‘Everybody is running to the Carnival, I say’ (30) Sing de song ‘To sing the songs’ (31) Everybody running the spree on Carnival Day ‘Everybody runs to have a good time on Carnival Day’ (32) Old an’ de young, de great an’ de small ‘The old and the young, the big (old) and the small (young)’ (33) Everybody looking to de Carnival ‘Everybody is looking for the Carnival’s (happiness)’ (34) Everybody running to de Carnival all day ‘Everybody is running to the Carnival all day (long) ‘ (35) One-shirt Willie and he mother-in-law May May ‘Poor Willie and his mother-in-law May-May’ (36) See dem pushing through de crowd on Carnival Day? ‘See how they are pushing through the crowd on Carnival Day?’ (37) Nuttin’to eat and nuttin to wear ‘They have no money to eat and nothing to wear’ (38) But on Carnival, dey are bound to be dere ‘But on Carnival (day), they are bound to be there’ (39) Everybody running to de Carnival, I say ‘Everybody is running to the Carnival, I say’ (40) Twis-eye Maryann, she went along with a policeman ‘Cross-eyed Mary Ann, she made up with a policeman’


(41) Make me to understand she running to take a Calypsonian. ‘She pretended that she was looking for a Calypsonian.’ (42) (You hear de number!) ‘Put some rhythm!’ The Carnival theme is very appropriate for the lyrics of calypso since it has been the calypso’s “home” in Limón from the very start. The style and the atmosphere depicted lend themselves to the composer’s sense of humor. In the words of Mr. Gavitt himself: In the calypso it does not matter if the singer has a poor voice; what is important are the lyrics. When [he] starts singing everybody listens and laughs. That is the touch of the calypso: it wishes for the people to have a good time with the lyrics and the rhymes. (Palmer 1986: 360)

Mr. Gavitt does not play in public any longer, but he cannot stop composing since all events that take place around his town turn out to be good subjects for his music. He has listened to calypso music since his childhood because a Trinidanian who lived in Cahuita (where Mr. Gavitt grew up) entertained his neighbors with calypsos. Mr. Gavitt started composing calypsos in the 1960s and has not stopped doing so since then. He was introduced to music by his aunt in Port Limón, where he played the organ in church. Later he often borrowed a guitar at the “Turk” Tabash’s bar in Cahuita, to entertain the customers, but his first owned instrument was a dulzaína (a flageolet), given to him by his brother. He taught himself to play the clarinet, and from then on he joined a group called “Miserable”, with whom he played Caribbean rhythms. It was then that he engaged in musical duels with well-known calypsonians of his time, such as Olle and Rají (Monestel 2002: 4). Carnival expresses, at various times, national, cultural, and ethnic identities. Thus the Carnival in Limón is both a release valve for the oppression of everyday life in an ethnic and class-stratified society and a time of inversion, where those at the bottom seize the moment to show their rebellion at injustice. Carnival is an amalgam of grass roots and high culture, anarchy and unity, creativity and tradition, and most of all it is a space of identity formation and confirmation. As this calypso states, everybody wishes to participate in the Carnival (29–34), the poor and the well to do (35–9), everybody is looking for a moment of shared revelry (40–2). It is a celebration that is accessible to a broad range of people — one that brings together not just the Afro-Limonese but the entire community and even outsiders. However, the Carnival in Limón is associated with the Creole people, because the calypso is part of their musical identity, as is their dancing and drum-playing, their colorful Caribbean clothing and their games, all within the frame of exotic vegetation, the brilliant sun and warm climate, and the blue sea.

273

274


5.5 Conflict calypso The next song transcribed is based on an anecdote which allows Mr. Gavitt to comment on people’s changing nature (43), while he acknowledges, tongue in cheek, that young people are getting “smarter” (44). Bato, the man referenced in this song, built a hut on stilts by the seaside (46), on a lot that belonged to the town, now part of Cahuita’s National Park (51). No one knows why he did it — the song claims he must have been inspired by “the devil and his daughter!” (70). But the result was a huge confrontation between the National Park’s administrator and the Cahuita people represented by Bato; both sides claimed that their rights had been infringed upon (58–66). Bato tries to defend his rights by claiming he is a Costa Rican — covertly alleging that he should therefore not be excluded on account of his belonging to a different ethnic group (64). Needless to say, the National Park won. This theme is typical of Ferguson’s lyrics, since through real or fictitious events, neighbors or invented characters, he offered in his calypsos an honest account of his people. Whether it is the calypsonian who was forsaken by his lover, or the musician who was traded for a high fidelity record player or a television set, or the student who was misunderstood by his teacher, all of these themes appear in his songs; some are treated with humor, others with irony, but they are all conveyed with great sincerity (Monestel 2002: 4). Entering a society that saw them only as black, regardless of their accomplishments and service to the country, Afro-Limonese live, work and struggle to succeed in an adverse milieu. In this case, however, Bato has infringed the laws of the country by building his hut in a prohibited area, therefore the calypso claims that he must have been led to do so by some “evil spirits”. The lyrics show the calypsonian’s sympathy for an amusing quarrel in which one of his people participated, but it ultimately condemns his transgressing the law. “Cabin in de water” (by Walter Ferguson) (43) This younger generation ‘The young generation’ (44) Every day de people getting smarter ‘Its people are getting smarter every day’ (45) Now they make me to understand ‘They have already told me’ (46) Bato build a cabin in de water ‘That Bato built a hut on the sea water’ (47) Oh, de cabin in de water ‘Oh, the cabin in the water’ (48) Mister Bato was de author ‘Mister Bato was the author’


(49) I knew he was a diver, but I never knew ‘I knew he was a diver, but I didn’t know’ (50) De bugger was a builder ‘The guy was a builder’ (51) The mistress of the National Park ‘(To) the lady [manager] of the National Park’ (52) Bato told her it was a rumor ‘Bato said that “it was just a rumor”’ (53) She decided to take a walk ‘ [But] she decided to take a walk’ (54) Lo and behold: a cabin in de water ‘Lo and behold, [there was] a cabin in the water!’ (55) The Mistress was getting hot ‘The lady got excited’ (56) When she saw de cabin in de sea ‘When she saw the cabin in the sea’ (57) Jumping like red beans boiling in pot ‘ [And she was] upset, like red beans boiling in a pot’ (58) And she tell him must pull it down immediately ‘And she told him (he had) to pull it down immediately’ (59) The building was quite erect ‘The building was (quite) perfectly built’ (60) Imagine, it was standing in the sea! ‘Imagine, it was standing in the sea!’ (61) The lady called him an architect ‘ “Of course”, she said, “you are an architect”’ (62) “But you going to pull it down immediately!” ‘ “But you are going to pull it down immediately!”’ (63) Now they come to a big dispute ‘ [As a result] there was a huge confrontation’ (64) Bato said, “Me born in Costa Rica” ‘Bato said, “I was born in Costa Rica (= I am Costa Rican)!”’ (65) “You could a born in Ethiopia” ‘ “Even if you had been born in Ethiopia”’

275


(66) “Me no want no cabin in de water!” ‘ “I don’t want a (any) cabin in the water!”’ (67) Kiaky Brown was telling me ‘Kiaky Brown told me [this story]’ (68) About the cabin in de water ‘About the cabin in the water’ (69) “Bato build something in de sea ‘ “Bato built some thing [this cabin] in the sea”’ (70) “Must be build it with the devil and he daughter . . .” ‘ “He must have built it with the devil and his daughter. . .”’

5.6 Anansi calypso The last calypso transcribed here is “Tacuma and Anansi’s Party”, also by Mr. Gavitt. The West Indies have a rich folk tradition which is very much alive in Annancy (also spelled Anansi) stories. They are fables based on the cunning and shrewdness of animals that share certain characteristics with humans. Mr. Gavitt has captured here the humor of the legendary characters, both in the voices and the sounds attributed to the animals as well as in the action described by the lyrics. He invented a musical group made up of animals called Anansi, Tacuma, Mono (Monkey), Burro (Donkey) (73, 85–8), and he imitated the violin sounds produced by Anansi and Breda (Brother) Mono’s voice fi fi a fa, fi fi a fa (79). When Mr. Gavitt started playing the clarinet with his imaginary musical animals, the borders between fantasy and reality disappeared (93). The chorus (lu lu lu) (94–5) imitated the sound of the clarinet; thus, the listener was supposed to construct a world inhabited by real and fictitious characters (Palmer 1986: 369). “Tacuma and Anansi’s Party” (71) I was invited to a party ‘I was invited to a party’ (72) And I was glad for de festival ‘And I was glad (to attend) the festival’ (73) It was Tacuma and Anansi ‘With Tacuma and Anansi’ (74) You must imagine how the people was liberal! ‘You can imagine how generous the people were!’ (75) They kindly gave me a seat ‘They kindly offered me a seat’


(76) While they practiced a melody ‘While they were rehearsing a melody’ (77) The music was soundin’ sweet ‘The music sounded great’ (78) They were singin’ “gal you want fi come kill me” ‘They were singing “Woman you want to kill me!” ‘ (79) Fi fi a fa, fi fi a fa (80) “Gal you wan fi come kill me” (repetition) ‘ “Woman, you want to kill me” (repeated)’ (81) Monkey marry to Baboon daughter ‘The Monkey married the Baboon’s daughter’ (82) “Gal you want fi come kill me” ‘ “Woman you want to kill me”’ (83) Catch him little sister and dash her in the water ‘He caught his little sister and threw her into the water’ (84) “Gal you want fi come kill me” ‘ “Woman you want to kill me”’ (85) Anansi playin’ de fiddle ‘Anansi played the violin’ (86) Tacuma beatin’ the drum ‘Tacuma was beating the drum’ (87) Breda Donkey givin’ a riddle ‘Brother Donkey was saying riddles’ (88) Breda monkey drinkin’ the rum ‘Brother Monkey was drinking the rum’ (89) (Hear them bawling) ‘(Hear them bawling)’ (90) See me how me jumping out me seat ‘See how I am jumping out of my seat’ (91) And I running home like a jet ‘And (how) I ran home like a jet’ (92) For the music was sounding sweet ‘Since the music sounded so good’ (93) I went and bring out my clarinet (Hear me playing) ‘I went and brought out my clarinet (Hear me playing)’

277

278


(94) Lu lu lu lu lu lu lu (95) Lu lu lu lu lu lu (96) Now the crowd was getting bigger ‘Then the crowd was getting bigger (closer) ‘ (97) They come to join the festivity ‘They joined the festival’ (98) Monkey into his liquor ‘(Brother) Monkey was already drunk’ (99) He start to shout out his melody (and he shout) ‘(And) he started to shout his song (and he shouted)’ Anansi stories are an important element of the Afro-Limonese folklore. While children in Western-European cultures are brought up listening to their parents’ readings of the Grimm brothers stories, in which the prince marries the beautiful girl to live happily ever after, children in the Caribbean are told Anansi stories, in which conniving animal characters get punished for their wrong-doings. Children are likely to be “hardened” by different forms of teasing, so as not to allow them to show signs of weakness, and in preparation for a future of struggle and discrimination. Teasing, boasting and insulting are essential forms of speech for people to defend themselves against all forms of attacks or criticism. These speech conventions are meaningful not only in that they mediate verbal expressions, but also in that they express larger meanings, such as the shared ethnic identity of the society that uses them.

6. Concluding note Lyrics of calypsos, such as those presented here, play a critical role in the development of a consciousness of negritude, the concept that Aimé Césaire propounded for the Creole people of the Caribbean. Even though that message came from the outside, the voice found an echo in people’s reality in Limón, allowing them to formulate their own specific expression of ethnic identity. The great contribution of the calypsos is that the messages of the lyrics (which are much more effectively transmitted to the audience than through instrumentation only) express an alternative worldview, meaningful to the Afro-Limonese. For that reason, we believe Calypso lyrics could help modify the attitude the Limonese have of themselves and their language, and eventually could articulate a more positive construction of their changing identity. Although reggae and Afro-pop are also very popular today and contain similar socially-conscious lyrics, they are recent musical developments, and are not ethnically and historically as closely connected to the Afro-Limonese culture as are calypsos.


Language is often seen as the most critical characteristic of identity for a group. While many scholars (Fishman 1989, Giles and Saint-Jacques 1979, Gumperz 1982) consider the possession of a given language of particular relevance — almost essential — to the maintenance of group identity, others (Edwards 1985) claim that it is important not to lose sight of its non-unique status as a marker. Of all the powerful elements of group identity (ethnicity, nationalism and the relationship between them), “the most important ingredients are the subjective sense of groupness and the continuation of group boundaries” Edwards (1985: 22). He considers that these two properties are indeed related, but since certain aspects of group culture are always subject to change, the continuing identity must depend upon elements that transcend any purely objective markers. Music, on the other hand, is also socially meaningful in terms of identity, largely because it provides means by which people recognize identities and places, and the boundaries that separate them (Stokes 1994: 5). Musical performance, as well as listening [and other related activities] provide the means by which ethnicities and identities are constructed and mobilized. People can equally use music to locate themselves in quite idiosyncratic and plural ways. A private collection of records, tapes or CDs, for example, articulates a number of highly idiosyncratic sets of places and boundaries. A moment’s reflection on our own musical practices brings home to us the sheer profusion of identities and selves that we possess. (Stokes 1994: 4)

Furthermore, although we usually characterize “black” music in the Americas in terms of its African heritage, and even though certain playing techniques and features of style may indeed carry significant markers of historical referents, they are not the only means that define the essence of the relationship of music to black ethnicity. In contemporary Afro-American communities in the Caribbean and South America, some aspects of cultural expression may fulfill a vital purpose of identity (Béhague 1994: vii). Thus it is precisely the performance of calypsos and their lyrics that are part of a larger social process that expresses resistance to subordination and exploitation (as seen by the calypso lyrics analyzed), while they also act as a resource in the creation and affirmation of the social identity of the Afro-Limonese. One can speculate that the importance in Costa Rica of Standard English as the prestige language through international television, the popularity of calypso, reggae and the Afro-pop culture usually expressed in creole, and the teaching of English as a second language in the public schools of the country could possibly give LC a new lease on life (Winkler p. c. Oct 1998). As long as calypsonians transmit a shared message through their lyrics, they will express their sense of self to the society in which they live.

Notes 1. The authors are very grateful to the reviewers and the editors for their meaningful and useful comments. We would also like to acknowledge with thanks the suggestions received

279

280 Anita Herzfeld and David Moskowitz from Manuel Monestel, Franklin Perry, and Gail Krotky, as well as the library support given by Emily Bergers. 2. According to a study of genetic markers conducted at the University of Costa Rica by Dr. Ramiro Barrantes in 1992, the Caucasoid element varies between 40% and 60%; the Negroid component between 10% and 40%, and the Amerindian between 20% and 30% (Bozzoli 1993: 2). 3. Indigenous languages spoken in Costa Rica include: Bribri, Cabécar, Maleku or Guatuso, Térraba, and Guaymí. 4. In Costa Rica, the Ministry of Education has established programs to teach literacy in the indigenous languages, especially in the area of Talamanca. However, there is still a great need for more government initiatives to implement a satisfactory program of bilingual bicultural education across the nation (Bozzoli and Guevara 2002: 42–3). Quality bilingual education is generally lacking all over South and Central America (Herzfeld 2000). 5. For a more thorough treatment of the language and its social correlates, see Herzfeld (2002). 6. Although calypso has been studied in Trinidad, Limonese calypso lyrics have not been given systematic consideration as an important cultural message. 7. I (Herzfeld) was the first linguist (1975) to call this English-based creole “Limón or Limonese Creole” rather than /mekaytelyuw/, the popular name used by its speakers. There is now a substantial bibliography of work on Limonese Creole (Herzfeld 2002). 8. Ethnomusicology as a field of study has not always been incorporated to music departments in this country. The dispute between ethno-musicologists and anthropologists brought about a reaction which gave music exaggerated importance. See, for instance, the defense that Lévi-Strauss (1983: 23) made in The Raw and the Cooked, in which he supports the idea that the essence of mythological thinking can be found in music. He illustrates and compares the principles of symmetry, inversion, equivalence, and homology, which constitute the structure of mythical thinking, with the baroque repertoire and the techniques of classical composition (Stokes 1994: 2). 9. Seeger, who worked with the Suyá of Xingu in Brazil, described his monograph as “musical anthropology”. 10. The classical music of my formative years and the tango of my adolescence constitute basic elements of my (Herzfeld) own identity. They stand for the place where I was born, the milieu in which I grew up and acquired the experiences that marked my personality. 11. On the 25th of September of 2002, Limón celebrated the 500th anniversary of the date on which Columbus actually landed in Cariari, Limón, in 1502. 12. The history, development, and psychological, sociological and literary nature of calypso are discussed by various authors (Cowley 1996, Elder 1966, Manuel, Bilby and Largey 1995, Quevedo 1983, Regis 1999, Rohlehr 1970, and Warner 1982). 13. In the United States Decca and RCA Victor both recorded and distributed island calypso artists from the 1930s to the 1950s. The rise of the Jamaican-American Harry Belafonte in the 1950s did much to popularize calypso. 14. They were recorded in the studios of the capital, San José. However, when Mr. Gavitt or Lenky is to be recorded, the equipment has to be taken over to Limón or Cahuita, as the case may be, because the artists would probably not travel to the capital (M. Monestel, p.c. Octo-

The Limonese calypso as an identity marker ber 25, 2002). In effect, that is exactly what happened with the most recent recording of Mr. Gavitt’s Babylon. The room of his cabin on the beach of Cahuita, where the CD was recorded, had to be lined with mattresses, the temperature had to be increased to isolate his voice and his guitar, while someone had to quiet the dogs’ barking (feeding them sausages!) and the parrots’ screaming outside (Ross 2002: 4). 15. When lyrics have been transferred directly from the CD jacket, the spelling adopted here is the one used by the recording company, as it is their interpretation of how Limonese Creole sounds. Never having been exposed to phonemic spelling, they adapt Standard English orthography to creole syntax and pronunciation. For a partial listing of the calypsos recorded in Costa Rica, see the Discography. 16. Reproductions of some lyrics are kept to short excerpts to adhere to the 10 % “fair use” statute. We have copyright release from Manuel Monestel of Kaiso Music; thus quotes from his recordings are considerably longer. 17. Spanish items are given in small capital letters.

References Acuña, V. H. Ancora. Suplemento Cultural de La Nación. Octubre 8, 2000 Aguilar, O. and I. Alfaro. 1997. La esclavitud negra en Costa Rica: origen de la oligarquía económica y política nacional. San José, Costa Rica: Progreso Editorial. Béhague, G. H. (ed.). 1994. Music and Black Ethnicity: The Caribbean and South America. Miami: North-South Center Press, University of Miami. Bozzoli, M. E. 1993. “La población costarricense: diversidad, tolerancia y discriminación”.Ms. Bozzoli, M. E. and M. Guevara. 2002. Los indígenas costarricenses en el siglo xxi: algunas perspectivas para la acción. San José, Costa Rica: Editorial EUNED. Costa Rica Today, June 18, 1992. Cowley, J. 1996. Carnival Canboulay and Calypso: Traditions in the Making. Cambridge: Cambridge University Press. Edwards, J. 1985. Language, Society and Identity. Oxford: Basil Blackwell. Elder, J. D. 1966. “Evolution of the traditional Calypso of Trinidad and Tobago: A sociohistorical analysis of song change”. PhD. Thesis. Philadelphia: University of Pennsylvania. Fallas, C. L. 1941. Mamita Yunai. San José: Editorial Soley y Valverde. Fernández, C. A. 1998. “Costa Rica. South America, Mexico, Central America and the Caribbean.” In The Garland Encyclopedia of World Music, D. A. Olsen and D. E. Sheehy (eds.), 2: 680–705. New York: Garland Publishing, Inc. Fishman, J. A. 1989. Language and Ethnicity in Minority Sociolinguistic Perspective. Clevedon, PA: Multilingual Matters, Ltd. Giles, H. and B. Saint Jacques (eds.). 1979. Language and Ethnic Relations. Oxford: Pergamon Press. Gumperz, J. J. (ed.). 1982. Language and Social Identity. Cambridge: Cambridge University Press. Herzfeld, A. 2000. “Integración o pluralismo: La educación intercultural bilingüe de los Maya-hablantes.” In Actas del XI Congreso Internacional de la Asociación de Lingüística y Filología de la América Latina. Universidad de las Palmas de Gran Canaria, J. A. Samper and M. Troya (eds.), 1745–1753. Las Palmas: Librería Nogal.

28

282

Anita Herzfeld and David Moskowitz Herzfeld, A. 2002. /mekaytelyuw/: La lengua criolla. San José: Editorial de la Universidad de Costa Rica. Lévy-Strauss, C. 1983. The Raw and the Cooked. Harmondsworth: Penguin. Manuel, P., K. Bilby and M. Largey. 1995. Caribbean Currents: Caribbean Music from Rumba to Reggae. Philadelphia: Temple University Press. Meléndez, C. and Q. Duncan. 1972. El negro en Costa Rica. San José: Editorial Costa Rica. Monestel, M. 2002. La Nación. Revista Ancora. 24 de noviembre de 2002, 4. Palmer, P. 1986. Wa’apin man: La historia de la costa talamanqueña de Costa Rica según sus protagonistas. San José: Instituto del Libro. Putnam, G. P. 1913. The Southland of North America: Rambles and Observations in Central America during the Year 1912. New York: G. P. Putnam’s Sons. Quevedo, R. (Atilla the Hun). 1983. Atilla’s Kaiso: A Short History of Trinidad Calypso, Errol, Hill (ed.). St. Augustine, Trinidad: University of the West Indies. Regis, L. 1999. The Political Calypso: True Opposition in Trinidad and Tobago 1962–1987. Gainesville, FL: The University Press of Florida. Rohlehr, G. 1970. “Sparrow and the language of calypso.” Savacou 2: 87–99. Ross, Y. 2002. La Nación. Revista Ancora. 24 de noviembre de 2002, 4. Seeger, A. 1987. Why Suyá Sing: A Musical Anthropology of an Amazonian People. Cambridge: Cambridge University Press. Smith, N. 1995. “An annotated list of creoles, pidgins, and mixed languages.” In Pidgins and Creoles: an Introduction, J. Arends, P. Muysken, and N. Smith (eds.), 331–75. Amsterdam: John Benjamins. Stokes, M. (ed.). 1994. Ethnicity, Identity and Music: The Musical Construction of Place. Oxford: Berg. Turino, T. 1989. “The coherence of social style and musical creation among the Aymara in southern Peru.” Ethnomusicology 33: 1–30. Warner, K.1982. Kaiso! The Trinidad Calypso. Washington. D. C.: Three Continents Press. Winer, L. 1993. Trinidad and Tobago. Amsterdam: John Benjamins.

Discography Mekatelyu. Comin nao. Johnnyman, Gabo, Goyo, and Pablito, ACAM 1999, 2001. Compact Disc. Contains : Mianmitelden Amarily love reggae Awanano c 25. Farafa, Oh Jah jah La Pirula Lirica In dis generachian Compas Cantoamerica. Calypsonians. Manuel Monestel. Roberto Huertas, Marcial Flores, Carlin Castro, Sergio Chavarria, Marvin Brenes, Dennis Fallas, Ricardo Alvarado, 1997. Compact Disc. Contains : Calaloo (W. Ferguson) The first time I saw Limon Cabin in the water (W. Ferguson) Puerto Viejo Tacuma and Anansi’s Party Land Lady (W.Ferguson) Monilia (Walter Ferguson) On Carnaval Day Everybody running to the Carnaval (W. Ferguson)

The Limonese calypso as an identity marker Cantoamerica. Aribarumba. Same group as above, n.d. Compact disc. Contains : On Carnaval day (W. Ferguson) Puerto Viejo A una paloma Til the end of time The first time I saw Limon Chano Pozo Viejos Tambores ¿Quién te escucha? Monilia (W. Ferguson) Habitante eterno de la tierra Afrolimon Various artists. Calypsos: Afro-Limonese Music of Costa Rica. Lyricord, 1988. Cassette. Contains: Primer ritmo de comparsa Day by day Trembling Calaloo Doctor Bombodee Carnival Day Yuca and plantin My teacher never like me Mother’s love Back to my country G. O. O. D. Monilia Calaloo Banton and Ghetto. Contradicciones. P. R. C. Crew, DIDECA, 1999. Contains : Tarde pero seguro Tu arca Ven girls Man a shotter No entiendo Pier gial A we Date a respetar Manzana prohibida Armagedon Paren la violencia Various artists. Afro-Costa Rican Folks; Caribbean Calipso [sic] Fight Against the Crack. Oscar Sequeira Higgs, Luis Alberto McKay, Rafael Espinoza Zapata, Moises Serrano Rivera, Ernesto Brown Brown, Carlos “Pipo” Chaves, Lalo Gabbiano. Producciones Sibo, n.d. Compact disc. Contains : Running away from the crack La lluvia Contradiction in history A Costa Rica con amor Carnaval in Port Limon Down by the river Pig tail Emoción estática The rule of the tyranns Por los canales del Tortuguero Congo Malata Portete Various artists. Calipsos Limonenses. Oficina de Extensión Cultural (Universidad Estatal a Distancia) Centro de Investigación y Conservación del Patrimonio Cultural (Ministerio de Cultura, Juventud y Deportes). Sony Music, Central America, n.d. Cassette. Contains : Side A: Roberto Kirlew (Buda) Carnaval day Playa Cahuita El zopilote Chinese baby

Buda y su charanga: Day by day Joseph Darking (Tun): Zancudo Natalie oh

283

284 Anita Herzfeld and David Moskowitz Side B: Edgar Hutchinson (Pitún): En Limon Seventy two weeds On top of the moon Yuca and plantain Walter Gavitt Ferguson: Cabin in the water Mi rumba (Festival) Everybody running to the Carnaval

CHAPTER 13

The speech event kuutu in the Eastern Maroon community * Bettina Migge

Johann Wolfgang Goethe-Universität

Sociolinguistic analyses of creoles are generally restricted to morphosyntactic aspects, drawing their data from vernacular speech in informal interviews. While this approach has undoubtedly contributed to a better understanding of the grammatical competence of creole speakers, it has provided relatively little insight into their communicative competence. The present study follows the “ethnography of speaking” approach (Hymes 1972) to investigate the social and linguistic properties of an important formal event in the Eastern Maroon community, the kuutu ‘council meeting’. The data underlying this study were collected among the Pamaka maroons. My investigation shows that the kuutu event is characterized by structured social and linguistic practices that provide important face-saving strategies, and create an aura of dignity, importance and respectability. The social practices described here include participation privileges, and procedures for organizing and holding a kuutu. Concomitant linguistic practices include turn-taking procedures as well as lexical and pragmatic choices. The social conventions and speech acts described here are primarily associated with titled persons and elders, who are the sole active participants in a kuutu. The analysis provided also suggests that speech genre analysis offers important insights into the nature of linguistic varieties and the social meanings they index.

1. Introduction Sociolinguistic research on creoles generally relies on speech data collected in relatively informal social settings. This methodological approach can be traced back to quantitative sociolinguistics that identified the vernacular as the most representative speech genre (Labov 1972, Milroy 1987).¹ Participants in such settings are assumed to employ their most “natural” speaking style, since they are thought to pay little or no attention to overt societal speech norms. However, anthropological research has provided abundant evidence that all (speech) behavior is subject to social monitoring and thus equally “natural”. In both

286 Bettina Migge

formal and informal interactions people select their social and linguistic practices based on various social considerations, such as situational appropriateness (the practices characteristic of a specific setting), the goals of the interaction, and the social identities they want to project (Eckert 2001). The main difference between formal and informal settings is that each involves a different set of social and linguistic norms. Participants in formal contexts target different types of social and linguistic practices than those who engage in informal contexts. An exclusive focus on only one kind of social setting and speech style thus appears to be arbitrary, and obviates a full understanding of the range of linguistic practices available within a community and cross-linguistically. In addition, sociolinguistic research often investigates structural and macro-social constraints that condition the variation between the local standard language (typically the superstrate) and the creole. Analyses that focus exclusively on morphosyntactic features, such as the copula or pronouns, pay little attention to the social meanings users attach to individual variants — with the exception of Sidnell (1999) — and to the features of particular speech events — with the exception of Abrahams (1983), Rickford (1986), and Sidnell (2000). Because of this predominantly one-sided approach, a significant amount of research remains to be done on (1) the range of discourse practices in creole-speaking communities, (2) the social and linguistic characteristics of such practices, (3) their social meanings and pragmatic functions, and (4) their role in the construction of social identities. A thorough understanding of the issues listed above is, however, essential to a comprehensive analysis of the linguistic makeup of Creole societies, the norms governing interactions, the nature of social identities, and the social (political and legal) organization of these communities. The goal of this paper is to investigate one important formal speech event in the Eastern Maroon (EM) communities of Suriname and French Guiana (Guyane), the kuutu ‘political meeting’ or ‘council’. More specifically, I will focus on the linguistic practices that characterize this type of event and on their related social meanings and functions. My observations indicate that the kuutu linguistic practices differ considerably from informal speech events in the EM community. Furthermore, they appear to be an exclusive feature of the maroon societies of Suriname. Other Creole communities in this region model their formal practices according to the norms of the local standard European language (Abrahams 1983: 90; Garrett 2000). The social and linguistic practices characteristic of kuutu events are subject to considerable structuring. Their function is to save the participants’ face, and to impart an aura of social respectability and responsibility to its users. Kuutu events also highlight the social significance of the issues being discussed. The paper is organized as follows: Section 2 briefly presents the EM communities, their linguistic background, and the speech data and the method of data collection. Section 3 investigates the kuutu setting, using Hymes’ (1972) “ethnography of speaking” framework. Section 4 focuses on the linguistic practices and norms

The speech event kuutu in the Eastern Maroon community

that apply in a kuutu. The final section (§5) summarizes my findings and discusses their implications.

2. The Eastern Maroon community The EM community is made up of three autonomous groups, the Aluku or Boni, the Ndyuka or Okanisi, and the Pamaka. These settlements were founded between 1710 and 1760 by slaves who had fled the plantations of Suriname (Hoogbergen 1990). Maroon villages are located in the interior of the tropical rain forest of Suriname and Guyane along the Marowijne river (in the case of the Aluku, Ndyuka and Pamaka), and its tributaries, the Lawa river (Aluku) and the Tapanahoni river (Ndyuka). Since the civil war in Suriname and the country’s economic decline in the late 1980s and 1990s, a significant number of EMs have moved to the urban zones of Guyane, particularly to the town of St. Laurent du Maroni. This city presently has the largest urban EM settlement in the region. The mother tongue and primary community language of the EM communities is the Eastern Maroon Creole (EMC). It is classified as an English-lexified creole, since roughly 77% of its basic vocabulary is derived from English. Significantly smaller numbers of lexical items come from Portuguese, Dutch, local Amerindian languages, and African languages (Huttar 1985, Smith 1987). EMC is said to be a conservative or radical creole because its structural system has been subjected to limited influence from its superstrate English, or other European languages spoken in the region, such as Dutch and French. There are three major EMC varieties that coincide with the three existing EM groups (Aluku, Ndyuka or Okanisi, and Pamaka). These originally developed on the plantations of Suriname between 1680 and 1720 (Migge 2002). The three varieties are mutually intelligible, but differ slightly on the phonological level, as well as the lexical. For example, Ndyuka uses long vowels in some contexts, whereas the other two varieties use short vowels (cf. bataa vs. bata ‘bottle’). On the lexical level, the ability marker is poi in Ndyuka and man in Aluku and Pamaka.² Two relevant publications provide linguistic and socio-cultural information on the Ndyuka community. One is a comprehensive grammatical description of Ndyuka (Huttar and Huttar 1994), and the other is a basic dictionary (Shanks 2000).

2.1 Data collection This study is based primarily on observations and recordings of various kuutu that took place in the Pamaka community (Map 1, next page) between 1995 and 2001. Most of the meetings took place in the village of the gaaman ‘paramount chief ’ of the Pamaka, Langa Tabiki. Some kuutu events were held in the Pamaka villages of Boni Doo, Nason, Lokaloka, and Badaa Tabiki, and in the French town of St. Laurent du Maroni.³ Topics debated in the kuutu events included mainly local matters, such

287

288 Bettina Migge

Atlantic Ocean

SURINAME

French Guyana

Guyana

Pamaka

Brazil

Map 1. Approximate area of the Pamaka

as intra-familial conflicts and disputes in the local gold-mining areas, or the organization and initiation of important ceremonies such as puu baaka ‘end of mourning rituals’. However, external issues, such as the status of relations between the Pamaka and the national government of Suriname, also figured in the kuutu debates. Most of the data provided in the present study are taken from a kuutu that was held in March 2001 in the village of Langa Tabiki. This meeting dealt with a long-standing problem between the kabiten ‘lineage head’ of Badaa Tabiki, and the members of the Mma Sanna bee ‘extended family founded by a woman called Mma Sanna’. Essentially, they were accusing their kabiten of disregarding his obligations, i.e. looking after his people and their territory, and of making undue use of his social powers. According to them, he was attempting to bring a gold-mining concession opened up by a family member under his sole control. In so doing, he intended to force family members who worked there off the concession, in order to install Brazilian and Surinamese-Chinese small-scale mining enterprises that would then have to channel a percentage of their benefits into his own pockets. The recordings were either made by myself, or by a native field worker. They were subsequently transcribed by the field worker or by another native speaker of the EMC. Some of the information on which this study is based was obtained in the course of informal discussions with the fieldworker and various members of the EM community.

3. The social characteristics of a kuutu 3.1 Location, physical makeup, and participants The word kuutu is probably derived from Kikongo kútu ‘ear’ (cf. the related Kikongo kúta ‘to be assembled, to assemble, to form a reunion, etc.’ (Laman 1964: 342–3).⁴ In

The speech event kuutu in the Eastern Maroon community 289

the EMC kuutu is used to mean both ‘to complain’, and ‘to debate, to argue’, as in: (1) Hii dey, a e kuutu den swagi f ’ en. whole day he PROG complain the-PL in-laws for-him ‘He is always complaining about his in-laws.’ (2) Den kuutu a toli fu a gowtu busi. they debate the-SG story for the-SG gold forest ‘They debated the problems in the gold-mining areas.’ It is to the latter sense,‘debate’, that EMs trace the meaning of the noun kuutu ‘a meeting in which disputes are discussed and arbitrated’. It is one of the most important formal speech events in the community. There are two broad types of kuutu: (i) Kuutu fu konde sani ‘village council or meeting’ whose function is to organize upcoming community-wide events (e.g. burials, end of mourning period etc.). This type of kuutu addresses matters such as the date and location of events, people’s contributions in kind, money and labor, rules of behavior, and the initiation or legalization of particular ceremonies. (ii) Kuutu fu seeka sani ‘meeting for arbitrating problems, court of justice’, whose function is to preserve or re-establish social harmony whenever the need arises. This type of kuutu is organized to arbitrate serious disputes between relatives, or to investigate a crime. It thus provides a forum for the opposing parties to present and defend their contradictory views, and to try to resolve them in an impartial environment. The present research deals primarily with this kind of kuutu. Kuutu are held in a local kuutu osu ‘debating house’, which is a rectangular open-roof structure. Kuutu osu built by the Surinamese government are made of stone, and are closed off at one of the short sides. Their use is generally restricted to kuutu that involve representatives of the national government. Kuutu events are also frequently held in other public or semi-public locations of the village, such as the ground floor of the house on stilts of a titled person, or in the area in which the faaka tiki ‘shrine of the ancestors’, and the kii osu ‘mortuary’ are located. The speaker participants and the audience bring their own seating, either regular chairs (sutuu), or local stools (bangi). No strict rules of clothing apply in a kuutu: People wear their regular clothes. Titled persons bring a token of their official uniform, such as their hats, in the case of a kabiten ‘(sub)lineage head’. There are two broad groups of participants in a kuutu, speaker participants and audience. Speaker participants are titled persons and elders.⁵ Titled persons are the official members of an EM government (lanti), including the gaaman ‘paramount chief ’, the ede kabiten ‘the most powerful lineage heads’, the (kowonu) kabiten ‘regular (sub)lineage heads’, and the basia ‘assistant to the gaaman and kabiten’.⁶ The

290 Bettina Migge

beneficiaries of these positions are selected from among the sisters’ (male) children (sisa pikin) of their predecessor, i.e., through matrilineal inheritance. They are generally older people, but in recent years, the number of kabiten who are in their thirties and forties has increased considerably because the Surinamese government refuses to recognize elderly kabiten. Titled persons can be male or female, but they are assigned different tasks. Female basia, in accordance with prevailing gender norms, are primarily responsible for organizing the upkeep of the public area of the village and the preparation of food during official events. A male basia serves as a ritual interlocutor (pikiman) in meetings, disseminates messages (bali basia), and assists the kabiten in all his activities. Due to pressure from the government of Suriname, most (sub)lineages recently created umanpikin kabiten — ‘ female (sub)lineage head’ positions — in order to increase women’s participation in the political decision-making process. Because of the novelty of this office, women’s positions are subordinate to the positions held by men. From the maroon perspective, the positions with the longest tradition (e.g., those from the lowe tin ‘time of flight’) have the greatest social prestige, and provide the greatest sociopolitical power. Umanpikin kabiten deal with women’s issues, and substitute for their male counterparts in their absence. For example, they may arbitrate small-scale problems, and receive members of the national government. Most elders are responsible (male) persons above 40 or 45 years of age. Individuals over 60, however, have the greatest social power. Female elders generally do not play an active role in public sociopolitical events. Titled persons and elders jointly manage the social, judicial and political affairs of the community. They give advice in times of crisis, arbitrate disputes (taki, kuutu), marry people (akisi uman ‘ask woman’), and organize ritual ceremonies such as burials (beli, olo), end of mourning rituals (puu baaka ‘remove mourning’), and spirit appeasement rituals (kunu pee ‘avenging spirit ritual’). They also represent the community or its matrilineal (sub)lineages (bee). In a kuutu, they carry out the discussion for the disputing parties. If titled persons or elders are involved in a dispute, they may represent themselves. However, their arguments are more persuasive if other titled members or elders accompany them. If they are on their own, they are less likely to be taken seriously since they failed to mobilize supporters. Untitled persons, and younger women and men, including the people involved in the dispute, do not generally take an active part in a kuutu. Their case is presented, discussed, and arbitrated on their behalf by the titled persons and elders of their bee. They are part of the audience but may be called upon to provide ‘expert testimony’. They can only indirectly influence the decision-making process by petitioning the active participants prior to the meeting, or by clapping or shouting out the active participants during the event. In a kuutu, the speaker participants usually sit apart from the audience and the most important titled members and elders (e.g., the gaaman, the ede kabiten, the most senior elders) sit separately from those of intermediate status (e.g. kowonu kabiten, other elders). All participants are arranged either in a full or in a half circle.


The speaker participants sit closely together in the center of the half circle while all others sit behind them, or on the fringes. In the case of a full circle, the active participants cluster in one area, and the audience in another. Some passive participants may sit behind the people of the inner circle. In kuutu fu konde sani ‘village council’, the presiding members face the audience.

3.2 The overall structure of a kuutu A complex sequence of semi-formal negotiations called taki (< talk) typically precedes a kuutu. The function of this preamble is to facilitate mediation between the disputing parties. The disputants, or their close kin, make a formal complaint to an elder, generally the tii ‘maternal uncle’ or gaan tii ‘mother’s maternal uncle’. The elders discuss the problem with the disputants, and attempt to bring the disputing parties together in an effort to resolve the problem locally and quickly. If it cannot be resolved at this level, the kabiten of the bee organize a kuutu in order to bring the issue in front of the lanti ‘government, general public’. If one or several local kuutu cannot resolve the problem, the matter is brought before the government of the entire group, e.g. the Pamaka. A kuutu may be called from one day to the next if an urgent matter arises. However, people are generally informed a couple of days ahead of time by messengers, by radio (on the coast) and/or by word of mouth. The day of the kuutu, a male basia (‘assistant to the gaaman and kabiten’) formally announces the time of the event. A kuutu can take place at any time but important community-wide ceremonies (e.g. beli, puu baaka, etc.) that last several days provide primary occasions for holding a kuutu, since most of the important people of a bee and from other bees gather on such occasions. They use the time between events linked to the ceremony to work on pressing community issues. Kuutu fu seeka sani ‘meetings for arbitrating problems, court of justice’ have the following overall structure: (i) Pre-event 1: People arrive at the venue, exchange (formal) greetings with the people already present (if they meet them for the first time that day), look for a suitable place to sit, and engage in small talk until the beginning of the kuutu.⁷ (ii) Pre-event 2: Once a sufficient number of the expected participants is present, important current issues, such as the outcome of a previous kuutu, are briefly discussed. The situation determines whether or not pre-event 2 will take place, and if so, what will be its actual duration. (iii) Opening of the kuutu: The highest-ranking members of the two disputing parties introduce the topic under discussion, and talk about the desired outcome of the kuutu. Each spokesperson presents his party’s grievance. With each introduction the problem becomes more narrowly defined, and clearly stated. Initially, figures of speech are used to allude to the issues under discussion (see Section 4. 4), but as the meeting progresses, the motives for disputes are overtly

29

292 Bettina Migge

acknowledged. In the case of issues that may lead to heated confrontations, participants are reminded of the rules of conduct that apply in a kuutu. (iv) Discussion: This is the longest part. The representatives of the two disputing parties present their case, and respond to the points of view and evidence provided by the other party. The highest-ranking members have the privilege to speak first. If the discussion threatens to become offensive, the presiding members or other participants remind the assembly of the rules of conduct. (v) Seclusion: Once all the views have been presented and discussed, or if some previously unknown issue arises in the course of the discussion, the members of each party usually go into seclusion (beni/go a se) to discuss possible solutions, positions and strategies in private, before presenting them in public. (vi) End: If the kuutu succeeds, the highest-ranking members make final resolutions. They rule on the outcome of the dispute, and make recommendations to prevent further problems. If the arbitration fails, a procedure for continuing the process is proposed.

3.3 Rules of conduct The most important behavioral norm in a kuutu is that participants have to be sensitive to positive and negative face. Positive face refers to (positive) self-image and personality, and to recognition and appreciation by others. According to Brown and Levinson (1989: 62), based on Goffman (1967), negative face refers to the necessity to avoid imposition on a person’s time, personal space, and freedom of action. In the context of the kuutu, face-saving protocols require that participants avoid undue direct confrontation, accusation, offence, criticism and demands involving fellow participants. Moreover, they are expected to properly present the issues at stake, and to work towards a solution that is acceptable to all parties concerned. Another important rule of conduct involves the acceptance of existing social hierarchies. People of lower social status should submit to those having higher social status. The former should pay attention to the latter’s point of view, heed their advice, and accept their decisions. Finally, people in a kuutu are expected to act in a calm and reserved manner. They should abstain from open expressions of emotions such as anger, (un)happiness, and must behave saka fasi ‘with humility’. They should not flaunt their knowledge, abilities and status, or abuse their social power.

4. Linguistic practices in a kuutu The style of speaking used in a kuutu is generally referred to as lesipeki taki/fasi ‘respectful speech’. It is employed in all kinds of socially important formal events (e.g., kuutu, taki, libations, appeasement ceremonies, or official requests in the presence of elders and people of high social status, such as titled persons and elders).


Lesipeki taki/fasi is distinct from kowonu taki and wakaman taki in terms of discursive practices (see below) and native categories. Kowonu taki/fasi ‘everyday speech’ is employed in typical everyday in-group settings among people of equal social status, or when talking to people of lower social status. Such settings include speech accompanying work activities, and informal chatting. Various types of code mixing, particularly when involving the coastal creole Sranan Tongo, is often referred to by such terms as wakaman taki/fasi ‘traveler’s speech’, or yunkuman taki ‘young man’s speech’. This speech style is closely associated with young men and street life. Unlike these two speech styles, lesipeki taki/fasi conforms to the rules of conduct discussed in Section 3.3, as it exhibits the four properties of formal language use identified by Irvine (1979: 776–9): (i) Increased code-structuring “concerns the addition of extra rules or conventions to the codes that organize behavior in a social setting” (Irvine 1979: 776). With respect to language, this may involve employing intonational (e.g. meter, or pitch contours), phonetic, lexical, morphosyntactic, and/or discursive features that set this variety apart from other ways of speaking. (ii) (Increased) code consistency. “At many different levels of linguistic organization [. . .] speakers select from among alternatives that have contrasting social significance” (Irvine 1979: 777). There are local rules of co-occurrence that stipulate the extent to which the choices have to be consistent. In formal settings “consistency of choices (in terms of their social significance) seems to be greater than in ordinary conversations, where speakers may be able to recombine variants to achieve special effects” (Irvine 1979: 777). (iii) Invoking of positional identities. “Formal occasions invoke positional and public, rather than personal, identities [. . .]. Public, positional identities are part of a structured set likely to be labeled and widely recognized in society [. . .]. Personal identities, on the other hand, are individualized and depend more on the particular history of an individual’s interactions” (Irvine 1979: 778). (iv) Emergence of a central focus. “[. . .] a main focus of attention — a dominant mutual engagement that encompasses all persons present (see Goffman 1963: 164) — is differentiated from side involvements” (Irvine 1979: 779). There are four features that saliently distinguish language use in a kuutu from everyday verbal interactions: (1) The nature of turns and turn-taking, (2) vocabulary and language choices, (3) address forms and (4) the use of figures of speech.⁸ These features are discussed below.

4.1 Turns and turn-taking in a kuutu The nature of turns and turn-taking in a kuutu is subject to considerable structuring. A speaker (takiman, TM) is supposed to select a new turn, either by asking the pikiman ‘ritual responder’ (PM) for permission to speak (3a), or by calling the atten-

293

294 Bettina Migge

tion of the pikiman (3b). However, people of high social rank, such as the gaaman or high-ranking kabiten, also claim turns without previously securing the pikiman’s permission (3c). TM: We, mi wani begi piimisi. ‘Well, I want to ask for permission to speak.’ PM: Ya, dda. ‘Yes, elder.’ TM: Mi wani taki wan sani. ‘I’d like to say something.’ PM: Ya, dda. ‘Yes, elder.’ TM: Mi sa taki? ‘May I speak?’ PM: Dda, kula kon. ‘Elder, present your matter.’ b. TM: We, basia. ‘Well, assistant.’ PM: Abii, kabiten. ‘Yes, kabiten.’ TM: Da a de so. ‘Then it is like that.’ PM: A so a de, kabiten. ‘That’s right, kabiten.’ c. Lanti:⁹ Iya, u yee, baa. ‘Yes, we heard.’ [end of turn] TM: A moy so. ‘It’s nice like that.’ PM: Eeya, dda. ‘Yes, elder.’ TM: A moy so yee! ‘It’s nice like that!’ PM: Ya, dda. ‘Yes, elder.’

(3) a.

The end of a turn is formally announced by the takiman using one of the phrases in (4). (4) a. (Iya,) u boo de, (papa). ‘(Yes), we breathe there (elder).’ b. Kwolon/gbolon! ‘It’s finished.’ [idiophone] c. (Gbolon/kwolon), da u tapu de (fosi). ‘It’s finished, so we stop here (for now).’ The pikiman ‘ritual responder’ then confirms the end of the turn by asking the other participants if they have heard and understood (yee) the speech. The latter generally respond affirmatively. The pikiman usually addresses all the participants. In addition, he may also single out the people who were addressed by the takiman’s speech (5). Only after the turn has been officially ended as described above is it permissible to begin a new turn (3). Kwolon! ‘It’s finished!’ Lanti, u án yee, papa? ‘Public, didn’t you hear and understand, elders?’ Lanti:¹⁰ Ya, u yee, aiyoo! ‘Yes, we heard and understood!’ PM: Baala S., u yee? ‘Mr. S., did you hear and understand?’ Baa S.: Papa, u yee ‘Elder, I heard and understood.’ PM: Baala B., i yee. ‘Mr. B., did you hear and understand?’ Baa B.: U yee, papa. ‘I heard and understood, elder.’

(5) TM: PM:


Once a takiman has obtained the right to speak, he generally comments on the form and content of the previous speaker’s turn, particularly if the previous speaker has a high social rank. In order to respect the previous speaker’s face, his comments are generally positive (6). (6) a. b. c. d.

A moy so yee! A taki moy fu toko. I taki moy. Eee, u gi gaaman gaantangi fu a taki

‘It’s nice like that!’ ‘The speech is/was very nice.’ ‘You spoke well.’ ‘We thank the gaaman for his speech.’

In some cases, especially if the current speaker has a high social rank and disagrees with the previous speaker, he directly responds to the turn (3c). However, he usually prefaces his speech with a hedge, i.e., he apologizes in advance for the possible offence that his speech may cause. (7) TM: Mi e gi piimisi. PM: Leti so, dda. TM: A mi anga baala B. ná abi taki ete! PM: A so a de. TM: A no baala B. kali mi a ini a kuutu. PM: Nono, dda! TM: Dda A., a án mu holi a taki a baka.

‘Excuse (my rude words).’ ‘Right, elder.’ ‘Me and Mr. B. do not have anything to discuss as of yet!’ ‘It’s like that.’ ‘It wasn’t Mr. B. who called me for the meeting.’ ‘No, elder!’ ‘Elder A., he should not distract from the topic.’

During a turn, only the ritual responder, pikiman, responds at regular intervals (piki taki) to the takiman. The pikiman is selected among the basia, or male elders, at the beginning of a kuutu, and generally holds this function throughout the kuutu. His responses (piki) come from a conventionalized set of short phrases that essentially signal agreement with the takiman (8). The same phrases are reapplied many times, but it is considered bad piki if he does not regularly alternate between the available options. Repetition is said to create monotony and to distract from the takiman’s speech. (N)a so a de (dda/papa); na so; a de so (dda/ppa). b. Ya, dda/papa; iya, dda/papa. c. Leti so (dda/papa); a leti so (dda/papa). d. Aiyoo (dda/papa); eeya/eiya ai, baa (dda/papa). e. Kwetikweti (dda/papa); nono (baa dda/papa).

(8) a.

‘It’s like that.’ ‘Yes, elder.’ ‘Right.’ (dda/papa); ‘Yes.’ ‘No/not at all.’

If the takiman asks a (rhetorical) question, the pikiman has to either answer it (9a) if the answer is obvious, or he simply repeats the question (9b).

295

296 Bettina Migge

TM: [. . .] I e si en boo fu ati? A sani ya, i e si en? PM: No, no. b. TM: I sabi san pasa? PM: San pasa?

(9) a.

Do you see his heart? This thing, do you see it? ‘No, no.’ ‘Do you know what happened?’ What happened?’

The piki-style requires a skilled interplay between the takiman and pikiman. In this dialogue, the takiman typically leaves short pauses or intervals after what he considers to be semantically significant units. The work of the pikiman then is to anticipate these pauses and to insert the piki so that it highlights or underscores the vital parts of the message (10). If the takiman talks too quickly and does not provide intervals, or if the pikiman does not properly follow the takiman’s speech, the piki-style cannot be realized. (10) TM: Da i si a sonu, . . . PM: Na so. TM: anga a kunu. PM: Na tuu. TM: Da te a kon gaandi . . ., PM: Na leti so. TM a án man moo. PM: Na tuu. TM: Da a obo ana kon a mindi. PM: Na tuu. TM: kon suku wan peesi.

‘So you see the sin, . . .’ ‘Right.’ ‘and the avenging spirit.’ ‘It’s true.’ ‘Then when it becomes old, . . .’ ‘Right’ ‘it gets tired.’ ‘It’s true.’ ‘Then it raises its hand and comes back among the family,’ ‘It’s true.’ ‘to look for a place (to rest).’

The pikiman has two main functions. First, he guides or encourages the takiman’s speech by providing him with input that supports his positive face. Second, by being the only permissible interlocutor during the takiman’s speech, he shields the takiman from distracting or disruptive comments from the other participants, i.e., he protects his negative face. He also highlights the role of the takiman, whose speech thus becomes the sole focus of attention. It is generally not permissible for the other participants to overlap or interrupt the current speaker. However, there are two situations in which interruptions are tolerated: first, if the content or form of the speech becomes disrespectful or offensive (11), and second, when a higher-ranking person considers it necessary to speak immediately at the beginning of the turn. These interruptions are generally prefaced with apologies to protect the current speaker’s face, and to avoid being accused of breaking the rules, i.e. to project a respectful self-image. (11) TM1: [Talking about all the bad things the ede kabiten did] Efu a so a be taagi mi ‘If he had told me that we (the bee) taki u anga en, a ná and he (the kabiten) are not the be a wan mi no be o same, I would not have inaugurated him.’ go de.


PM: A so a de. TM1: Da a kabiten . . . TM2: Eee, a ná enke mi e koti. TM3 (to TM2): No, no, no. TM2: A, a pisi fu a taki de, ma a u mu taki en ma . . .

‘Right’ ‘Then the kabiten. . .’ ‘It’s not like I’m interrupting.’ ‘No, no, no.’ (= Don’t interrupt) ‘That part of the speech, we have to discuss it, but (in that way?).’

Although the above-described format clearly prevails during a kuutu, two participants may engage in direct exchanges (12). These exchanges are usually of a combative nature, and constitute a violation of the norm that stipulates that participants should behave in a reserved manner and abstain from direct confrontation. They are quickly discouraged (TM3). (12) [The basia (TM1) had just openly accused the kabiten (TM2) of forcing his people off the bee’s gold-mining concession in order to install outsiders from whom he can collect a percentage of their revenue. TM1 now tells TM2 that his actions will have consequences. TM2 interprets this to mean that TM1 is threatening to use physical violence against him]. TM1: U án be poti en de fu toon fetiman. PM: Leti so. TM1: A taanga fu feti, ma koni de tu. PM: Leti so, dda. TM1: I yee mi? TM2: Koni de! TM1: Koni de! TM2: Fi i feti anga mi! TM1: Koni de fu feti. TM3: Yee, yee! TM3: Efu i naki a sama, . . . TM3: I sa go wasi baka, . . . PM: Eeya. TM3: Ma efi i taagi a sama wan taki di án bun, da a palaki fika de.

‘We did not make him kabiten so he becomes a trouble maker.’ ‘Right.’ ‘He’s hard to fight, but there’s help.’¹¹ ‘Right, elder.’ ‘Do you hear and understand me?’ ‘There is help!’ [lit.= knowledge exists] ‘There is help!’ ‘For you to fight with me!’ ‘There is help to fight.’ ‘Listen, listen’ . . . ‘If you hit someone, . . . ‘You may go and wash it off, . . .’ ‘Yes.’ ‘But if you tell a person something that isn’t good, then it gets stuck there.’

4.2 Language and word choices in a kuutu Besides respecting well-defined turn-taking rules, participants in a kuutu are expected to select appropriate lexical items (abi lesipeki/kiyo ‘code-consistency’). It is essential to avoid taboo and common or vulgar terms, such as those related to sex

297

298 Bettina Migge Table 1. Respectful expressions in the Eastern Maroon Creole Respectful expression

Common expression

English gloss

tia sama, boliman dda sama, goniman koo sama,sama fu sileti konde waka anga wan uman, mankeli/piimisi anga wan uman de anga sikin, de tu fasi, a lasi mun de ne en futu, kii dia, go a se man peesi, taku pisi uman peesi gadu yeepi a uman saanti poti wan sani gi wan sama, de anga faya a baka abi wan sidon, sidon makandi

uman, folou, frou man, masra dede sama soki

‘wife, woman’ ‘husband, man’ ‘ancestors’ ‘have sex’

de anga bee go a doo pipi popoi meki kosi, gaan mofu koloi, wisi kuutu

‘be pregnant’ ‘menstruate’ ‘penis’ ‘vagina’ ‘give birth’ ‘curse’ ‘bewitch’ ‘meeting’

and witchcraft (wisi), and to use instead a specific set of terms denoting respect.¹² Table 1 presents a non-exhaustive list of such respectful terms and their common counterparts. People also tend to avoid direct reference to topics that may threaten participants’ face, such as direct criticisms, and accusations or restriction of their personal freedom. They generally allude to them using figures of speech (see Section 4.4). However, if it becomes necessary to openly discuss such matters, they are typically prefaced with abundant apologies (13) to mitigate the offensive force of the words. In example (13), only the takiman’s words are provided; the responses of the pikiman are omitted for reasons of space. (13) – Mi begi piimisi dii toon, papa J.! – So! Mi begi piimisi. – Na omen toon. – Den sama di sidon ya na mi tata sidon ya ye. – Da mi begi piimisi fu a gaaman bika mi tata. – Fu mi mofu án mu misi gaanenge. – A fu gaanenge, mi e begi a piimisi. – Mi begi a piimisi fu lespeki . . . – Ma mi taki kabiten A., . . . – Baala A. [kabiten A.] gi mi bigi sen.

‘I ask for forgiveness three times, elder J.!’ ‘Okay! I ask for forgiveness.’ ‘So many times.’ ‘The people who are here, it’s my elder who sits here.’ ‘I ask forgiveness of the gaaman since he is my father. ‘So that I am not going to say rude things.’ ‘It’s for rudeness I ask for forgiveness.’ ‘I ask for forgiveness out of respect . . .’ ‘But I say kabiten A., . . .’ ‘Mr. A.[ kabiten A] he greatly embarrassed me.’

The speech event kuutu in the Eastern Maroon community 299

Respectful speech also excludes vulgar language use, which in the EM community is generally considered to derive from the coastal creole Sranan Tongo, the language associated with street life and “cultural deprivation”. Participants in a kuutu thus tend to refrain from using Sranan Tongo-related terms, which evoke nonprestigious identities and values considered inappropriate and offensive in formal settings. Expressions from Dutch or French — the official languages of Suriname and Guyane, respectively — or another maroon creole such as Saamaka, do not carry such negative connotations. While the two European languages are associated with external social power and learning, maroon languages like Saamaka are associated with maroon power and traditions. However, code-switching involving any of these languages or other varieties of the EMC is minimal since undue use of varieties other than the local EMC variety is perceived as an attempt to disassociate oneself from the local community.

4.3 Address forms In a kuutu, people’s positional identities such as their social status, standing and function within the community play a very important role. They determine a person’s speaking rights and overall power. In order to show the utmost respect and overtly acknowledge status differences and social positions, participants in a kuutu always use official titles when addressing or referring to others. These titles mark official rather than personal identities, and thus carry high social prestige. In the EMC, there are two types of titles: titles of courtesy and function titles. Titles of courtesy codify differences in gender, social status, and age. Traditional function titles are not gender differentiated, whereas terms referring to modern professional titles are often gender-sensitive, as in the source languages (Dutch and French). Table 2 (next page) provides a non-exhaustive list of EM function titles and titles of courtesy. These official titles are combined with the person’s public or official EM name, rather than their European first name or their last name. The official EM name is most closely associated with the bearer’s self and carries the greatest respect.¹³ When participants in a kuutu refer to themselves, they tend to use “diminutive” terms (14) since it is considered rude to openly display one’s status or power. (14) a. Mi na wan pikin fesiman fu a bee. ‘I am a little leader of the lineage.’ b. Mi a pikin man. ‘I am only an unimportant person.’ c. Mi a pikinnenge. ‘I am but a child.’

4.4 Figures of speech Another linguistic property that is characteristic of language use in a kuutu is the use of figures of speech, locally known as nongo. These are deemed “a central repository of moral wisdom and values” (Price and Price 1999: 241). Figures of speech permit the expression of confrontational and offensive issues, as well as criticism,

300 Bettina Migge

Table 2. Common titles of address in the EMC [For details see Migge (2001)] Female

Male

Titles of courtesy Gloss

Titles of courtesy Gloss

Sa + Name

‘Ms for woman of lower social status/age’ ‘Ms for women of intermediate status/age’ ‘Ms for women of high social status/age’

Ba + Name Baala (+ Name) Tii/Tiu (+Name) Dda, p(a)pa (+Name)

‘Mr. (low social status/ age)’ ‘Mr. (intermediate status/age)’ ‘Mr. (high social status/ age)’

Gloss

Function titles

Gloss

Gaaman same as female same as female

‘head of EM community’

Mesta (+Name) Met (+Name) Opasi (+Name) same as female

‘male teacher’

Tia (+Name) Mma (+Name) Function titles

— — Kabiten (+Name) ‘(sub)lineage head’ Basia (+Name) ‘assistant to kabiten or gaaman’ Yefrow (+Name) ‘female teacher’ Met(r)es (+Name) Siste (+Name) ‘female nurse’ Data (+Name) ‘medical personnel’

‘male nurse’

accusation, and so forth, but they do so in a veiled and relatively non-offensive manner. Maroon cultures have a rich set of such figures of speech, which are obligatorily employed to save kuutu participants’ face. Code-structuring and code-consistency provide strategies to avoid face threatening speech acts. The same code conventions also project the image of a good takiman ‘speaker (in a formal event)’. This mechanism “represents the ultimate achievement in the verbal arts” (Price and Price 1999: 241). Without appropriate use of nongos, a speaker cannot be recognized as a dignified and important person. Nongo is essentially a cover term for different categories of rhetorical figures of speech.¹⁴ Some nongos are proverbs, that is, shorter or longer sayings that embody general truths.¹⁵ For reasons of space, I omit the responses of the pikiman in the longer examples. Ala pii tifi a ná lafu. ‘Things aren’t always as they seem.’ [lit. = all bearing of teeth isn’t laughter.] b. [Introducing the topic of a kuutu that deals with the nature of the leadership of the gold mining area, i.e. whether one or two persons should be in charge] – Fa a de, wan boto ná ‘The way it is, a boat does not abi tu (si)tiiman. have two leaders.’ – A wan boto, a ná a fa a bigi. ‘A boat, it’s not about its size.’

(15) a.


– Luku fa a masini e daai de. – A wan kodo man e tii en. – Ma a omen man e de na a boto ede. – Anga kodeli – Ma i si pe a man de a lasi de? – Ne en tii en.

‘Look how the motor turns/runs.’ ‘It’s only one person who steers. ‘But many people are in the front part of the boat.’ ‘In an ordered fashion’ ‘But you see where the man is in the back there?’ ‘It’s he who steers it.’

Some nongos resemble metaphors. They compare people and their actions or situations to objects or plants (16). (16) [Talking about the continuation of a controversial arbitration process that was temporarily suspended by one of the disputing parties involved] – Mi no man taki mi no o de. ‘I cannot say I won’t take part a ini. (in the kuutu).’ – Bika mi na napi. ‘Because I am a napi (a type of ground vegetable).’ – Pe den e paandi mi, mi ná e ‘I don’t root where they plant me, beli de, mi e gwe te so go beli. I go over there to root.’ – Da kande te den o kaba ‘Maybe when they are ready fu kon a gaaman. to come to the gaaman.’ – Da mi a napi, di mi án beli ‘Since I do not root there, de, da ná mi sa du. then I may not participate.’ – Bika na a bun pisi doti ‘Because I did not find the mi án feni de. good planting ground there.’ – Mi a wakaman. ‘I am a traveler.’ Some nongo are fables or folktales locally called mato. (17) [The leaders of the Maroon and Amerindian governments had unanimously decided not to sell part of their rainforest to Asian wood logging companies. The gaaman of the Pamaka is now telling the other members of his government that he told the other gaaman to be honest and determined]¹⁶ – Soo! Da tu biya be de, ‘Okay! There were two young men, den be abi mma. they had a mother.’ – Ma den mma gaandi. ‘But their mother was old.’ – Neen den taki, biya! ‘Then they said: well, friend, I sa san u o du? you know what we’ll do?’ – U mma e muliki, ‘Our mother is a strain, u ná e man waka fa u wani, we cannot do what we want to, eeye, a mma gaandi, ala ten yes, mother is old, all the time u mu e solugu u mma we have to take care of our mother, namo, i sa san u o du? you know what we’ll do?’

30

302

Bettina Migge

– Kon u tja a mma go kii. – U o tja en go na a gaan sabana ini anda, da u gi en faya boon. – Den taki, ai! I taki bun, biya ma i sa san u o du? – U o luku wan dey moo. – Neen di a dey doo, – Da wan fu den, a komoto, a go na a sabana pe den o poti na a sabana, pe den o poti den mma fu boon. – Neen a diki wan olo, [. . .] – Neen di a kaba diki a olo, . . . – mooy, neen a taa wan kon. A taki, we baya, we, u taki wan sani ma i ná e piki!, Neen a taki, we, a yuu doo, a dey di i wani, da i meki u go. Neen a taki, a bun, da u o go! – Den poti dey, te a kaba, den diki den mma poti a baka. – Neen den diki poti,neen den go, moy silli, neen den poti den mma, neen den gi a sabana faya. – Da a wan fu den di diki a olo, a taagi en taki: “Mma, luku wan olo ya. Te a faya e waka e kon, u o gwe a wan se.” – Neen den gi a sabana faya. – Di den boon te a kaba mooyn, den djombo go piisii. Oh!! Biya, a mma, luku ya a dede, a dede. – Den de, den de te, dey e pasa. Wan fu den taki, oh biya, mi e si wan sama a ini i osu ini de, sama de ape?

‘Come, let’s go and kill mother.’ ‘We’ll bring her to the big savanna and burn her there.’ ‘They say: You spoke well friend, but you know what we’ll do?’ ‘We’ll wait another day.’ ‘Then when the day came,’ ‘Then one of them, he went to the savanna where they wanted to burn the mother.’ ‘Then he dug a hole,’ ‘Then when he had finished digging the hole, . . .’ ‘nicely, then the other one came. He said, well my friend, well, we talk about it, but you don’t reply. Then he said, well, the time is now, the day you want let’s go. Then he said, okay, then let’s go.’ ‘They set the day and lift their mother on their back.’ ‘They lift her on their back and went nicely, they set down their mother and set the savanna on fire.’ ‘Then the one who dug the hole told her: “Mother, look there is a hole. When the fire comes towards you, go to one side.” ‘Then they set the savanna on fire.’ ‘When they had finished burning, they jumped to celebrate. My friend, look, at the mother, she is dead.’ ‘After some time, one of them said: Oh, my friend, I see someone in your house, who is there?’


– A go luku, a taki, biya, na i mma de ape? A taki, eeye! – A taki, aah! We, u be taki, u o kii u mma? – Ma mi án be fii fu kii en. Da na so a waka. – Neen a taki, iya. Neen mi taki, da na Anainsi anga Sensen , a den tu man de. – Anansi anga Sensen. – Ma Anainsi koni, ma Sensen puu en mma. – Neen mi taagi den taki, u án mu pee Anainsi anga Sensen. – Te u taki u o du wan sani, – U e du fu leti, u e du fu leti. – Ma a án mu kon fu Anainsi anga Sensen.

‘He went to look, he said, friend, is that your mother? He affirmed. He said: Didn’t we say we’ll kill her?’ ‘But I did not feel like killing her, that’s all.’ ‘He affirmed. I said, it’s like Anainsi and Sensen, these two men are like them.’ ‘Anainsi and Sensen.’ ‘Anainsi is intelligent, but Sensen helped his mother.’ ‘I told them [the other gaaman] we should not play Anainsi and Sensen.’ ‘When we say we’ll do something,’ ‘We aim to do it right.’ ‘It should not turn into Anainsi and Sensen.’

5. Conclusion In this paper I discussed in some detail the social and linguistic properties of kuutu events among the Eastern Maroons of Suriname and Guyane. My discussion suggests that the kuutu is a well-structured social event. Relatively fixed procedures determine the organization of a kuutu, and participation in this kind of event is based on social rank. Active participation is restricted to people who hold high social status, and high ranking dignitaries are granted speaking rights prior to low(er) ranking elders and titled persons. Linguistic form is highly regulated as well. There is a relatively fixed procedure for requesting and granting turns, a fixed internal structure of turns, and a special lexicon (words, expressions, names/titles) associated with respectful speech. This structuring of social and linguistic practices in the conduct of a kuutu serves to save face, and to highlight the dignity and importance of the participants, and of the topics they discuss. Social and linguistic practices also play an important role in the construction of local social identities (Irvine 2001). Participation privileges in a kuutu construct two broad, hierarchically ordered social groups that are indexed with different social attributes. One group consists of male titled members and elders. Since they regularly engage in formal events and in the social and linguistic practices that characterize them, these practices and their social meanings are mapped onto this group of people. In the EM ideology, titled persons and elders are consequently said

303

304 Bettina Migge

to be powerful, dignified, respectful and verbally skilled. The other group includes all the untitled young women and men who are generally not allowed to participate actively in prestigious and formal community events. Because of their non-participation untitled members are assumed to be sociopolitically inexperienced, undignified, and lacking in respect and vital social and verbal skills. This discussion suggests that a speech genre analysis that focuses on situated social events, i.e., events that are part of the community repertoire, is essential to a full understanding of the sociolinguistic makeup of creole-speaking communities. A discourse-based analysis provides a better insight into the nature of the linguistic repertoire of creole speakers.

Notes * I thank two anonymous reviewers, Jack Sidnell, and the editors, G. Escure and A. Schwegler, for valuable comments and criticisms. The research was supported by the IRD Cayenne (Laboratoire linguistique) and the Délégation générale à la langue française under the project Langues, pratiques et ressources linguistiques en Guyane. 1. As observed by Sidnell (p.c., June 2003), this methodology also has its origins in relatively unchallenged ideas about creole languages, as these creoles are typically characterized as (the most) vernacular, natural, and authentic speech forms (of a community). The aim of this paper is to challenge these ideas. 2. See Bilby (2002) for a discussion of the differences between Aluku and Ndyuka. Goury and Migge (2003) discusses the differences between Aluku, Ndyuka and Pamaka. 3. Of the roughly fifteen kuutu events I observed, eight were tape recorded. 4. I thank Armin Schwegler for suggesting this etymological origin of kuutu. Bakongo men and women (speakers of Kikongo) were almost certainly among the early and later contingents of Suriname slaves (Arends 1995). In other parts of the Americas, Kikongo kútu has also been preserved. Thus, in Cuba kutu (or plural makutu) is used in the Palo Monte ritual code to this day, and with a variety of meanings, including ‘ear(s)’. For an introduction to the Palo Monte ritual language, see Fuentes Guerra (1996, 2002), Schwegler (2002), and Valdés Acosta (2002). 5. For information on the makeup of the Pamaka community, see also Lenoir (1973: 18–50). 6. The word lanti is used to refer to both, the government of a people (e.g., Pamaka lanti ‘the leaders of the Pamaka community’), and the leaders of a subsection of the community (e.g., Mma Sanna bee lanti ‘the leaders of the Mma Sanna sublineage’). Moreover, lanti is used to refer to the general public “those [people] who can in a judicial sense serve as a [sic] potential witness” (Shanks 2000: 111) or any of its representatives, namely, any person attending a kuutu. 7. For a list of respectful or formal greetings in the EMC, see Goury and Migge (2003). 8. See also Huttar and Huttar (1988) for a discussion of a staged formal event as language play. 9. This is a chorus reply of all the members present at the meeting.

The speech event kuutu in the Eastern Maroon community 10. This is also a chorus reply of all the members present at the meeting. 11. A taanga fu feti refers to the fact that it is extremely difficult to remove a kabiten from his position since it is believed that the person replacing him would be killed by his predecessor’s avenging spirit. Koni de means that there are also other ways of dealing with the situation, such as to remove him and to leave the position vacant until his death. 12. Note that words such as uman, frou ‘woman’ have sexual connotations while boliman (lit. ‘cook’) and tia sama (lit. ‘aunt person’) do not. In a similar vein, terms such as koo sama ‘lit. cold/quiet person’ and sama fu sileti konde ‘person of the supernatural world’ suggest that deceased persons have simply moved on to another physical state but are still (active) members of the community — they give advice to its living members. In contrast, dede sama ‘dead person’ denotes a person that has been lost to the community. 13. See also Price and Price (1972) for a discussion of naming practices among the Saamaka. 14. Price and Price (1999: 240–1) mentions that rather than fully presenting the figures of speech (proverbs, folktales), skilled speakers prefer to make cryptic allusions to them. They expect their interlocutors/audience to complete them mentally for themselves. While I have often encountered such allusions, the data used for this study did not, to my knowledge, include any such ellipsis. 15. Koanting and Velanti (2000) provides 350 short proverbs. 16. An anonymous reviewer notes that it is a very well known Anansi story. It is often told in other contexts such as wakes and related death rites.

References Abrahams, R. 1983. The Man-of-words-in the West Indies: Performance and the Emergence of Creole Culture. Baltimore: The John Hopkins University Press. Arends, J. 1995. “Demographic factors in the formation of Sranan.” In The Early Stages of Creolization, J. Arends (ed.), 233–85. Amsterdam: John Benjamins. Bilby, K. 2002. “L’aluku: un créole surinamien en territoire français.” Amerindia 26/27: 279– 92. Brown, P. and A. Levinson. 1989. Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press. Eckert, P. and J. R. Rickford (eds.). 2001. Style and Sociolinguistic Variation. Cambridge: Cambridge University Press. Eckert, P. 2001. “Style and social meaning.” In Eckert and Rickford, 119–26. Fuentes Guerra, J. 1996. Raíces bantu en la Regla de Palo Monte. Cienfuegos: Ediciones Mecenas. Fuentes Guerra, J. 2002. Nzila ya mpika (la ruta del esclavo). Una aproximación lingüística. Cienfuegos: Ediciones Mecenas. Garrett, P. 2000. “ ‘High’ Kwéyòl: The emergence of a formal creole register in St. Lucia.” In Language Change and Contact in Pidgins and Creoles, J. H. McWhorter (ed.), 63–101. Amsterdam: John Benjamins. Goffman, E. 1963. Behavior in Public Places. New York: Free Press. Goffman, E. 1967. Interactional Ritual: Essays on Face to Face Behavior. New York: Harper and Row.

305

306 Bettina Migge Goury L. and B. Migge 2003. Grammaire du nengee: Introduction aux langues aluku, ndyuka et pamaka. Paris: Collection Didactique. Editions IRD. Hoogbergen, W. 1990. “The history of the Suriname maroons.” In Resistance and Rebellion in Suriname: Old and New, G. Barna-Shute (ed.), 65–102. Williamsburg: The College of William and Mary. Huttar, G. L. 1985. “Sources of Ndyuka African vocabulary.” De Nieuwe West-Indische Gids 59: 45–71. Huttar, G. L. and M. L. Huttar. 1988. “A humorous Paramaccan text.” Southwest Journal of Linguistics 8: 34–50. Huttar, G. L. and M. L. Huttar. 1994. Ndyuka. London: Routledge. Hymes, D. 1972. “Models of the interaction of language and social life.” In The Ethnography of Communication, J. J. Gumperz and D. Hymes (eds.), 35–71. New York: Holt, Rinehart and Winston. Irvine, J. 1979. “Formality and informality in communicative events.” American Anthropologist 81: 773–90. Irvine, J. 2001. “Style as distinctiveness: The culture and ideology of linguistic differentiation.” In Eckert and Rickford, 21–43. Koanting, E. D. and C. Velanti 2000. I de, i mu abi! Dii ondoo feifitenti Nongo Taki fu Okanisi. Paramaribo, Suriname: SIL. Labov, W. 1972. Sociolinguistic Patterns. Philadelphia: Pennsylvania University Press. Laman, K. E. 1964 (1936). Dictionnaire kikongo-français (2 vols). Ridgewood, NJ: The Gregg Press. Lenoir, J. D. 1973. The Paramacca Maroons: A Study in Religious Acculturation. Ph.D. Thesis. New York: New School for Social Research. Migge, B. 2001. “Communicating gender in the Eastern Maroon Creole.” In Gender Across Languages (vol. 1), M. Hellinger and H. Bußmann (eds.), 85–104. Amsterdam: John Benjamins. Migge, B. 2002. “The origin of the copulas (d/n)a and de in the Eastern Maroon Creole.” Diachronica 19: 83–136. Milroy, L. 1987. Observing and Analyzing Natural Language. Oxford: Blackwell. Price, R. and S. Price. 1972. “Saramaka onomastics: An Afro-American naming system.” Ethnology 4: 341–67. Price, S. and R. Price. 1999. Maroon Arts: Cultural Vitality in the African Diaspora. Boston: Beacon Press. Rickford, J. 1986. “Concord and contrast in the characterization of the speech community.” Sheffield Working Papers in Linguistics 3. Schwegler, A. 2002. “El vocabulario (ritual) bantú de Cuba.” In La Romania americana. Procesos lingüísticos en situaciones de contacto, N. Díaz, R. Ludwig and S. Pfänder (eds.), 97–194. Frankfurt: Vervuert Verlag/Madrid: Iberoamericana. Shanks, L. 2000. A Buku fu Okanisi anga Ingiisi Wowtu. Aukan-English Dictionary and English-Aukan Index. Paramaribo, Suriname: SIL. Sidnell, J. 1999. “Gender and pronominal variation in an Indo-Guyanese creole-speaking community.” Language in Society 28: 367–99. Sidnell, J. 2000. “Primus inter pares: Story-telling and male-peer groups in an Indo Guyanese Rumshop.” American Ethnologist 27: 72–99. Smith, N. 1987. The Genesis of the Creole Languages of Surinam. Ph.D. Thesis. Universiteit van Amsterdam. Valdés Acosta, G. 2002. Los remanentes de las lenguas bantúes en Cuba. Havana: Fundación Fernando Ortiz.

CHAPTER 14

Reflexivity in French-based creoles Katrin Mutz

Universität des Saarlandes

Several studies have investigated the structural or functional restrictions underlying reflexivity in creole languages (cf. Carden 1993, Carden and Stewart 1988, Corne 1988, 1989, Heine 2001, Muysken and Smith 1995). However the diverse reflexive constructions that occur in specific creoles have yet to be fully investigated. My analysis, based on data drawn from several “French-based” creole languages, demonstrates that the choice of the “appropriate” reflexive construction from among a range of options is constrained by various converging aspects, i.e., (1) the semantics and the valency-structure of the verb, (2) the register used (e.g., spoken versus written), (3) lexicalization matters, and (4) discourse context. In this study I will present a “functional paradigm” of the creole reflexive elements: in (basilectal) creoles one does not find elements whose function is solely to mark reflexivity (instead, these elements also mark possessive, emphasis, and so on). Creole reflexive markers also differ in that they are less grammaticalized (and therefore typically more transparent) than those of their lexifiers (thus French se, for instance, is non-transparent, and also has functions other than simply “reflexivity”; e.g., anticausativity).

1. Introduction Substantial work on synchronic and diachronic aspects of reflexivity in (Frenchbased) creoles has been published in recent years (e.g., Carden 1993, Carden and Stewart 1988, Corne 1988, 1989, Déchaine and Manfredi 1994, Heine 2001, Kriegel 1996, Mufwene 2000, Muysken and Smith 1995). However, there is, to my knowledge, no comprehensive analysis of the structural or functional constraints that operate synchronically to produce the diverse reflexive constructions found in Frenchbased creoles. This study first offers a short introduction to the concept of reflexivity and how it is treated in the literature. Section 2 then presents the various forms of reflexive constructions in creole languages. Section 2.1 examines the different construction types of reflexivity, selected exemplarily from French-based creoles. Section 2.2 follows with an account of various parameters that seem to play a role in the selection of appropriate reflexive strategies. The concluding section sketches the functional domain of the elements used in the reflexive construction types analyzed in this paper.

308 Katrin Mutz

1.1 Reflexivity Reflexivity is one of the most widely discussed phenomena in linguistic literature, with seminal work done by Faltz (1985) and Geniušiene (1987). What is to be understood by the notion of reflexivity is not as obvious as it seems at first glance. Different frameworks offer diverging definitions, depending on whether they focus on syntactic or on semantic aspects (cf. Faltz 1985: 14, Geniušiene 1987: 58, Grimshaw 1990: 154, Kemmer 1994: 207, Mohanan and Mohanan 1998: 173). In this paper, instances of semantic reflexivity will be discussed, i.e., those verbal (or predicative) constructions that refer to an “action” in which the entity in the role of the agent/actor/experiencer is co-referential with the entity which typically assumes the role of the undergoer/patient/theme or, presumably more seldom, the role of the beneficiary/recipient. Reflexivity then means the merging of two thematic roles. In Grimshaw’s approach (1982, 1990) the syntactic consequence of this merging process is the elimination (or suppression) of a syntactic argument relative to the non-reflexive construction. On syntactic grounds reflexivization is thus understood as a process of intransitivization. In a more traditional approach, reflexive elements can be treated as complements of the verb, along the lines of direct or indirect objects (see Kriegel 1996: 47–50). As will become evident in Section 2, the line between reflexive and non-reflexive actions is not easy to draw since semantic reflexivity does not always correspond to formally marked reflexivity and vice versa; a given action, like dressing, might be expressed in one language by a reflexive construction (e.g., French s’habiller), in another language it might be constructed non-reflexively (e.g., English to dress or to get dressed). For the latter case the term “inherent reflexive” is often used. The literature on creoles frequently claims that semantically inherent reflexives are not formally marked. For that reason, cases like Fr. se souvenir ‘to remember, to recall’, s’en aller ‘to leave, to go’ should not be found in creole languages.¹ An example of such a claim is given in Carden and Stewart: We suggest that French “pronominal verbs” normally appear as Mau[ritian] intransitives just in case the French reflexive is semantically empty [. . .] (Carden and Stewart 1989: 80)

My main concern will be with reflexive constructions that have an iconic relationship between semantics and formal expression (cf. Mufwene 2000). This paper will, however, also take into account those cases in which the symmetry between form and function is blurred.

2. Reflexivity in (French-based) creole languages Reflexives tend to be innovative in creoles in that they are not directly transferred from input languages but have arisen anew during creolization (cf. Carden and Stewart 1988):

Reflexivity in French-based creoles 309 Leur disparition [i.e., la disparition des morphèmes réfléchis] pendant la créolisation a causé un vide fonctionnel et a ainsi permis une restructuration complète d’un système peu transparent.² (Kriegel 2000: 73)

This formation of a new “reflexive system” in creoles should come as no surprise since reflexivity, though corresponding — in the optimal case — to some extralinguistic reality, is a feature of the grammatical information of a verb, and is thus, like nearly all grammatical elements, lost and recreated during creolization.³ There are different opinions in the literature on how the reflexive constructions have emerged. Whereas Corne (1989) advocates influence of the superstrate, Lefebvre (1998) and Carden (1993) argue for substrate origins. Heine (2001) favors a language-internal grammaticalization, whereas Carden and Stewart (1988) and Muysken and Smith (1995) tie the origins of the reflexive constructions to universal aspects of internalized grammar.

2.1 Construction types Heine (2001) and Muysken and Smith (1995) offer a general account of the different reflexive devices found in creole languages. They note the following strategies (and combinations thereof):⁴ a. Personal pronoun (used for all three persons), e.g., in Seselwa (Seychellois), Guadeloupean Creole. b. Personal pronoun (used for first and second person), e.g., in Southern Haitian.⁵ c. Reflexive marker (exclusively used for the third person), e.g., in Negerhollands. d. Bare noun stem (involving a “body-(part-)”noun), e.g., in Cape Verdian. e. Personal pronoun + “intensifier”, e.g., in Morisyen (Mauritian), Louisiana creole. f. Possessive attribute + “intensifier”, e.g., in Papiamentu. g. Possessive noun phrase (possessive pronoun and “body-noun”), e.g., in Haitian, Seselwa. h. Personal pronoun + noun, e.g., in Saramaccan. i. Personal pronoun + “intensifier” + noun, e.g., in Saramaccan. j. Zero marker, e.g., in Guadeloupean Creole, Morisyen.⁶ In French-based creoles, construction types a., b., e., g. and j. are commonly used to refer to reflexive actions. Examples (1)–(10) illustrate these strategies by drawing on a small corpus⁷ of oral and written creole texts⁸ from Haiti, Guadeloupe, Mauritius, and the Seychelles islands. According to Faltz (1985: 32), the strategy of using the personal pronoun for the construction of reflexivity is a rather uncommon solution in the languages of the world, and is typologically marked (cf. Carden and Stewart 1989: 85). In creole languages the pattern of VERB + PERSONAL PRONOUN shown in (1)–(3) is very productive:⁹

30

Katrin Mutz

(1) Haitian Ti-fi-a fin abiyé avèk vwal é âswit mwê abiyé-m. little-girl-DET ASP dress with veil and then 1SG dress-1SG ‘The girl finished dressing, with a veil, and then I dressed myself.’ (Hall 1953: 102) (2) Seselwa Sûgula i koz li tu sel. Sûgula 3SG talk 3SG all alone ‘Sûgula talked to himself.’ (Bollée 1977: 194) (3) Guadeloupean Ou ké maské’w dèyè gro pyé figyé-la sa. 2SG TEMP hide-2SG behind big fig-tree-DET DEM ‘You will hide behind that big fig-tree.’ (Rutil 1981: 162) In (1)-(3) above, the pronoun used to express reflexivity has the same form and position as the direct object personal pronoun. Depending on the creole language, the pronoun is formally identical or non-identical to the personal pronoun in subject position. When the pronoun refers to the 3rd person, ambiguity can arise because it can be read as a direct object or, alternatively, a reflexive pronoun. In some languages or dialects (e.g., Southern Haitian Creole), this ambiguity has been resolved: a pronoun can be used reflexively only in the first and second person, whereas in the third person another reflexive strategy must be used (Carden and Stewart 1988: 18). A second reflexive strategy is given in examples (4) and (5), where the intensifier (INTS) mem (< Fr. même ‘self ’) is combined with a personal pronoun (e.g., mo < Fr. moi ‘I’): (4) Morisyen Mo kontan mo finn donn momem enn kado pou mo laniverser. 1SG like 1SG ASP give 1SG-INTS DET present for 1SG birthday ‘I like it that I have given myself a present for my birthday.’ (www 1) (5) Morisyen Kouma enn dimoun ki pe koz ar limem. like a person REL ASP talk with 3SG-INTS ‘Like a person that talks to oneself.’ (www 2) Mem might originally have been introduced to resolve ambiguity within the reflexive construction when the pronoun referred to the third person. In the literature on Morisyen and Seselwa, the element mem is claimed to be optional in the reflexive construction, at least with some verbs (Corne 1988: 71, Kriegel 1996: 141). Conversely, in Louisiana Creole, mem appears to be fully grammaticalized, as it is an integral, obligatory part of the reflexive construction (PERSONAL PRONOUN + mem). Faltz (1985: 34) calls this strategy an “adjunct secondary reflexive” or a “fused adjunct reflexive”. This kind of patterning can be found not only in creole languages (e.g., in Saramaccan) but also, for instance, in English or in Dutch.

Reflexivity in French-based creoles

The marking of reflexivity by a complex device (BODY (PART) NOUN + preposed or postposed POSSESSIVE PRONOUN) is often invoked as a prototypical feature of creole languages. In actuality, this strategy (called “head reflexive” by Faltz 1985: 29) is cross-linguistically very common, even in non-creole languages (cf. König 2001).¹⁰ (6) Seselwa I n koriz so lekor. 3SG ASP correct 3SG body ‘He has corrected himself.’ (Corne 1988: 75) (7) Haitian Jafrikayiti dekri tèt li konsa [. . .] Jafrikayiti describe head 3SG like-this ‘Jafrikayiti describes himself in the following way [. . .]’ (www 3) Another reflexive construction is treated in the literature as being “zero-marked” or “inherently lexically marked” within the intransitive verb stem (Carden and Stewart 1989: 80, Corne 1988, Heine 2001: 9), as in examples (8) and (9): (8) Haitian Mâmâ-m lévé Ø. mum-1SG get up REFL English: ‘My mom got up.’ French: ‘Ma maman s’est levée.’ (Hall 1953: 78) (9) Guadeloupean Mwen ka touné Ø an légliz é mwen ké touné’w an labé. 1SG ASP change REFL in church and 1SG TEMP change-2SG in priest English: ‘I will turn myself into a church, and I will turn you into a priest.’ French: ‘Je vais me transformer en église et je vais te transformer en prêtre [. . .]’ (Rutil 1981: 199) Table 1. Summary of reflexive patterns (French creoles) Sey Pronoun Pronoun + mem “Body” + Poss Ø-marker a

Mau

Hait W

S

N

x (x) X

X X x

— — X tèt

x — X tèt

x

x

X — X tèt kò

Gua

Lou

X — x

x X (x)

x

Notes : X: primary strategy; x: peripheral strategy; (x): very marginal strategy; — : strategy not attested a For Haitian Creole and Louisiana Creole no data are available about the productivity of “zeromarking” in the local varieties.

3

32

Katrin Mutz

The various reflexive patterns in the creoles under investigation and the productivity of these patterns as reported in the literature are schematized in Table 1. The picture that emerges in the literature on creole reflexives can be summarized as follows: • Unlike French (se, soi) or German (sich), creoles have no monomorphemic, nontransparent markers of reflexivity. • Creole languages exhibit rich formal variation as regards the expression of semantic reflexivity. In one and the same language there are often both simple and complex strategies, even with respect to the same verb as exemplified in (10): (10) Guadeloupean a. i pa ka vwè ola, La Florantin é Ti-Jan pé ay kaché Ø. 3SG NEG ASP see where La Florantin and Ti-Jan can go hide REFL ‘He could not figure out where Florantine and Ti-Jean could have hidden.’ (Rutil 1981: 199) b. i alé kaché’ y adan on pyès kann yo té tini. 3SG go hide 3SG in DET plantation sugar cane 3PL TEMP own ‘He hid in a sugar cane plantation which they owned.’ (Rutil 1981: 233) • Since the elements used in the reflexive construction are polyfunctional (e.g., direct object reading, or reflexive reading of the pronoun, see section 3), ambiguity can arise (cf. Adone 1994). Even though substantial work has been done on synchronic and diachronic aspects of reflexivity, the factors that must be accounted for to explain the distribution of the various construction types have not yet been clarified. Seminal work carried out by Corne (1988) points to underlying semantic selection restrictions governing the distribution of the different strategies. Another issue, as yet undeveloped, concerns the productivity of the respective construction options in each language.¹¹ Section 2.2 addresses the first question by examining the parameters or factors potentially responsible for the occurrence restrictions of the reflexive strategies.

2.2 Distribution and constraints Current research offers but limited discussion of the distribution of the reflexive constructions in creole languages. In studies addressing reflexivization only a small number of creoles (especially Morisyen and Haitian Creole) are included, and explanations proffered typically refer to a single parameter, e.g., the semantic class of the verbs involved. Integrating earlier analyses found in the literature, I will present here several factors that must be included in any explanation of the distribution of the various reflexive constructions. The selection (from among the different options available) of the “appropriate” reflexive construction seems to depend on the convergence of these (and other) factors.


Syntactic function and semantic role Within a perspective which identifies the reflexive element as an argument of the verb, one can look for correlations between the construction type and the syntactic function of the reflexive element, i.e., its function as a direct object, an indirect object, or as a prepositional complement. A first look at the data of our corpus confirms Carden’s results (1993) concerning Seselwa lekor (< Fr. le corps ‘the body’). Lekor is mainly used as a direct object, and never appears after prepositions. Corne’s claims (1988) that the Morisyen reflexive construction with PRONOUN + mem typically appears after a preposition and that the pronoun in the “simple pronoun construction” typically has the semantic role of a beneficiary or recipient are both falsified by the data of the corpus examined, as shown in examples (11) and (12): (11) Morisyen Kan Indra ti kit Moris li ti promet limem ki zame li when Indra TEMP leave Mau. 3SG TEMP promise 3SG-INTS REL never 3SG pa ti pou remet lipie dan sa pei la. NEG TEMP TEMP return foot in DEM country DET ‘When Indra left Mauritius, she promised herself that she would never ever come back to that country again.’ (www 4) (12) Morisyen Mo fin kup mua ek en kuto. 1SG ASP cut 1SG with a knife ‘I cut myself with a knife.’ (Papen 1978: 397) In (12) the pronoun mua does not assume the semantic role of the recipient, but has the role of the patient of the action, and in (11) the element limem does not follow a preposition. The correlation in Morisyen between the use of the reflexive construction involving the pronoun plus identifier mem, and its appearance after a preposition contradicts Corne’s claim: the corpus under consideration does not include any post-prepositional reflexive construction (other than PRONOUN + mem), and the PRONOUN + mem construction appears quite often without a preceding preposition. Haitian and Guadeloupean Creole do not seem to reveal any correlation between the syntactic function or the semantic role of the reflexive element and the construction chosen. Complex predicates A first analysis of the corpus suggests that in the creole languages the respective primary reflexive strategy (see Table 1) is chosen to indicate reflexivity with complex predicates (examples (13) and (14) illustrate this).

33

34

Katrin Mutz

(13) Haitian Responsab jistis ak lapè te kontinye poze tèt li kesyon [. . .] responsible justice and peace TEMP continue put head 3SG question ‘The Minister of Justice and Peace continued to ask himself the question [. . .].’ (www 5) (14) Seselwa Lenstiti i war li en obligasyon pour sansibiliz dimoun lo nou institut 3SG see 3SG in obligation to sensitize people on 1PL lalang. language ‘The institute sees itself obliged to sensitize the people to our language.’ (Lavwa Kreol Spesyal, Oct. 1999, cit. Kriegel 2000: 71) This hypothesis should be tested in the context of additional data. Semantic features of the antecedent In the unmarked case, the antecedent (typically the subject) of a reflexive construction has the semantic role of the agent, and bears its prototypical features, in the sense of Dowty (1989). That is, the antecedent is typically a human entity which controls the action. In the corpus under inspection only a very small number of examples can be found in which the antecedent of the reflexive element is [−human]. Example (15) is one of these. (15) Haitian Pwogram lan detwi prop tèt li. program DET destroy own head 3SG ‘The program destroys itself.’ (Haitian informant) Via a metonymic transfer the antecedent can often be re-identified as a human entity (16): (16) Haitian Dêmokratia te vle di: pèp la pou gouvène tèt li. democracy TEMP want say people DET can govern head 3SG ‘Democracy meant: the people can govern themselves.’ (www 6) Syntactic function of the antecedent Various studies (Déchaine and Manfredi 1994, Faltz 1985, König and Siemund 2000) have observed that there seems to be a general tendency of correlation between complex reflexive constructions, local binding and non-subject-orientedness versus simple reflexive constructions, long-distance binding and subject-orientedness. Carden and Stewart (1988) analyze the binding and domain properties of the tèt li-construction of the Haitian Creole of Port-au-Prince and observe two conditions for its appearance: the subject condition [+SC], i.e., the reflexive must refer to a subject, and the clausemate condition [+CM] (as in (17a)), i.e., the reflexive must have its antecedent in its own clause (17b).


(17) Haitian a. Emili pale (ak) Robèj de tèt- lii. Emil speak with Robert of head 3SG ‘Emili talks to Robert about himselfi.’ (Carden and Stewart 1988: 11) versus b. Emili pale (ak) Robèj de lii,j,k.. Emil speak with Robert of 3SG ‘Emili talks to Robertj about himj,k/himselfi.’ (Carden and Stewart 1988: 11) As to the reflexive constructions of the creole languages of Mauritius, Guadeloupe and the Seychelles, the domain properties of the different construction types have yet to be investigated. Mood (imperative) At least for the creole of Guadeloupe, the corpus suggests that reflexivity in imperative sentences is preferentially expressed through the pronoun stategy: (18) Guadeloupean a. Pa okipe’w! not worry-2SG ‘Don’t worry!’ (Rutil 1981: 61) b. Maman lévé’w! Mum get up-2SG ‘Mum, get up!’ (Rutil 1981: 61) c. Size’w la! sit-2SG there ‘Sit down right up there!’ (Rutil 1981: 136) In the respective declarative sentences with these predicates, though, reflexivity is typically “zero-marked”:¹² (19) Guadeloupean Manman-la lévé-Ø é alé sizé-Ø dèwô- la. mum-DET get up-REFL and go sit-REFL outside there ‘The mother got up and sat down outside.’ (Rutil 1981: 146) Even predicates that in declarative sentences appear as intransitives are represented in imperative sentences by a reflexive pronoun: (20) Guadeloupean a. Manman vini’w! mum come-2SG ‘Mum, come here!’ (Rutil 1981: 163)

35

36

Katrin Mutz

b. Rantré’w vit! come home-2SG quickly ‘Come home soon!’ (Rutil 1981: 145) This latter construction recalls Corne’s “pronominal imperative” (1988: 109), described for the creoles of Réunion and the Seychelles: (21) Reunionese Asper au! wait 2SG ‘Wait!’ Seselwa Reste u dan u pti kwen! stay 2SG in 2SG little corner ‘Stay in your little corner!’ As far as other creole languages are concerned, no such claim about “pronominal imperatives” can be made yet due to lack of data. Register Socio-pragmatic variables such as the type of discourse, the speech situation (formal, informal), and diastratic differences between the interlocutors might play a role in the selection of a reflexive construction within a given context. Register might also be decisive (Carden and Stewart 1988: note 35).¹³ The written register demands regularization and standardization to a greater extent than does speech, as Ludwig (1996) has demonstrated. Since it has, in general, a more “grammar-like” status than the merely spoken register that has more of a “pragmatic-like” status, we expect to find in the written data of our corpus a rule-governed distribution of the reflexive constructions based on syntactic or semantic grounds, and a lesser degree of variation. Following the above assumptions on the status of spoken discourse vs. written discourse we expect to find more covert,“less marked”, reflexive constructions (zero marking, pronouns) in oral speech, and more overt and more complex markings of reflexivity (e.g., PRONOUN + mem, possessive construction) in written discourse. Yet, contrary to our assumption, the first findings suggest that in the oral register¹⁴ redundant marking of reflexivity is highly productive since we find many instances of overt marking of reflexivity with so-called semantically inherent reflexives: (22) Guadeloupean a. Yo té ka abiyé-yo an pay. 3PL TEMP ASP dress-3PL in straw ‘They dressed with straw.’ b. Yo vlopé-yo konsa évé sa. 3PL wrap-3PL like this with this ‘Like this they wrapped themselves up with this.’ (Ludwig, Telchid and Bruneau-Ludwig 2001: 84)


The semantic category of the predicate Research on the semantics of the predicates involved in a reflexive construction in Morisyen and Seselwa (Corne 1988, Kriegel 1996) has shown that selectional restrictions govern the distribution of the different reflexivization devices. Corne (1988: 71–5) establishes for Morisyen the following classes of “verbs” that occur within different reflexive constructions: a. Verbs of class I: selecting pronoun (+/- mem), e.g., aranz, esprime, gete, konye, esplike, envite, korize, kupe, large, mete, respekte, tuye, lave, pandi, sulaze, zete. b. Verbs of class II: with or without a pronoun, e.g., abitye, alonze, asize, bese, debruye, deside, grate, abiye, amize, ampese, benye, degaze, dezabiye, kasyet, leve, prepare, sape, sove, penye, raze, santi. c. Verbs referring to physical actions (and some emotion verbs), selecting the possessive construction with lekor. Corne does not point out the (semantic) features that the predicates in class I or, alternately, class II have in common to justify independently the classification of the predicates in two separate categories, i.e., apart from their selecting different reflexivization constructions. A number of predicates enumerated in class I or class II refer to physical actions which, following Corne, should be represented by a possessive construction involving a “body noun”. For a better understanding of the correlation between the semantics of the predicates and the reflexive constructions in which they appear, we refer to the analyses of Kemmer (1994), König (2001) and König and Siemund (2000). König and Siemund (2000: 61) distinguish between verbs referring to typically non-other-directed situations like dressing, shaving, protecting, and so on, and predicates which designate situations which are typically other-directed like hurting, beating, talking etc. (Table 2). The problem with this classification is that the terms “typically other-directed” and “typically non-other-directed” are not clearly defined, and open to varying interpretation. For example, the situation of being proud (or being ashamed) of something, classified by König and Siemund as describing a typically non-other-directed state of affairs, refers to a kind of emotion, which itself is categorized by König and Siemund as an other-directed situation. The notion of “directedness” is not entirely Table 2. Situation types Non-other-directed situations a

Other-directed situations b

Grooming Preparing, protecting Defending, liberating, preparing Be proud/ashamed of

Violent actions (killing, destroying) Emotions (love, hate) Communicating Be jealous of/angry with/pleased with

a b

Called “inherent reflexives” by Muysken (1993: 292) Called “transitives” by Muysken (1993: 292)

37

38

Katrin Mutz Table 3. Correlation between situation type and reflexive construction

Non-other-directed situations

Other-directed situations

Verbal stratgey No/optional anaphor Simple/weak/SE anaphor Single intensifier

Nominal strategy Obligatory anaphor Complex/strong/SELF anaphor Double intensifier

Source : König and Siemund (2000: 63)

clear either. Is “directedness” intended to refer to the patient, to the recipient or to the beneficiary of a given situation? And what about an action described by a ditransitive verb involving three arguments like to sell: John sells the book to Paul? Notwithstanding this uncertain classification, there are nevertheless correlations linking the two classes and the reflexive constructions in which they appear (see Table 3). The more complex strategy tends to be used for the more remarkable (i.e. conventionally other-directed) situation; the less complex strategy tends to be used for inherently reflexive verbs and for conventionally non-other directed [sic] situations. (König 2001: 758)

The analysis offered in König (2001) and König and Siemund (2000) can be combined with the findings of Kemmer (1994), who examined the so-called “body action middles” in the languages of the world and detected that these predicates are not to be found in complex constructions (due to language economy, see Haiman 1985): a. b. c. d.

Grooming actions (e.g., bath, dress, wash, shave); Change in body posture (e.g., stand up, sit down, lie down); Non-translational motion (e.g., stretch, turn, shake); Translational motion (e.g., fly, climb, walk, arrive).

These actions are non-other-directed, in the sense of König and Siemund, i.e., they cannot be decomposed into “parts of acting and acted-on entities” (Kemmer 1994: 198). Applying the results of König and Siemund and Kemmer to the reflexive constructions in the French-based creoles under investigation, we obtain the clines illustrated in Table 4. The clines in Table 4 predict that predicates referring to non-other directed situations are typically used within less complex reflexive constructions (e.g., zero marking) than predicates referring to other-directed situations. The schema in Table 5 (pp. 322–3) lists 150 predicates (with reflexive constructions), which are correlated with König and Siemund’s (K/S) and Kemmer’s (K) analyses. Those data are drawn from my creole corpus of Haitian, Guadeloupean, Morisyen and Seselwa.¹⁵

Reflexivity in French-based creoles Table 4. Clines of reflexive constructions

Less complex < more complex Non-other-directed situations/inherent reflexives/body-action middles < otherdirected situations Zero marking < personal pronoun < personal pronoun+ mem/possessive pronoun + tèt/lekor etc.a a

The two constructions seem to have the same degree of complexity, with both constructions consisting of two elements.

About 70% of the data conform to the analysis provided by König and Siemund and Kemmer. The remaining 30%, which does not correspond to their analyses, might be explained by the advanced stage of grammaticalization of the respective reflexive markers: ‘[. . .] there is almost always [at least] one stage-II form in a given creole, that is, a fully grammaticalized reflexive.’ (Heine 2001: 36)¹⁶ There might well be other semantic restrictions at work (if not one of the factors mentioned in the preceding sections), such as the ones mentioned by Corne (1988) or Kriegel (1996). These authors found that, in Morisyen (as in Papiamentu, see Muysken 1993), the reflexive strategy associated with predicates that refer to physical actions is the possessive construction involving a body-noun. However, in the corpus of Morisyen underlying this article, a number of examples with a lekor-construction can be found in which “physicalness” can only be ascribed to the predicate if metonymic processes are assumed (see examples (23) and (24)). (23) Morisyen Parski Sir Ralph ti bizen cheke sipa Ganga ti pe bien nouri because Sir Ralph TEMP need check if-not Ganga TEMP ASP well eat so lekor. 3SG body ‘Because Sir Ralph had to ensure if Ganga had eaten enough.’ (www 7) (24) Morisyen Mo sir mo pou kapav defann mo lekor kont agresion Blob. 1SG sure 1SG FUT can defend 1SG body against attack Blob ‘I am sure I can defend myself against Bob’s attack.’ The claim of correlation between the reflexive “possessive construction” and “physical” predicates made by Corne and Kriegel for Morisyen is not supported for the creole of Guadeloupe. The possessive construction with the body-noun kò is used in our corpus of Guadeloupean with the following predicates to express reflexivity: fèmé (1×), viré (2×), kouvè (1×), paré (1×), mété (2×), senti (1×), oublié (1×), fou (1×), fouté (1×), sové (3×), débouyé (1×). Only some of these predicates refer to physical actions.

39

320 Katrin Mutz Table 5. Analysis of reflexive patterns (French creole corpus) Haitian Personal pronoun

Non-other- lave, raze, santi, abiye, Zero-marking is less complex directed amize, maske, reziye and hence would have been more consistent with K/S and K. Otherdirected

wè, blese, ede, pale de, The possessive constr. is more complex trouve and hence would have been more consistent with K/S and K.

Possessive Non-other- kache, leve, desann, constr. with tèt directed defann, fyè de, vante Otherdirected

Consistent with K/S and K (the most mete, dekri, swen, met, bay, respekte, complex reflexive strategy is used). gouvène, poze kesyon, rann, touye, pann, detwi, fè lapwòpte, blese, van, pale ak

Possessive Non-other- kache, wete, vire, constr. with kò directed poze, fatige Otherdirected Zero marker

Not consistent with K/S and K (a less marked reflexive strategy exists).

blese, detwi, mete, kache, blie, souke, lage, raje

Non-other- leve directed

Not consistent with K/S and K (a less marked reflexive strategy exists). Consistent with K/S and K (the most complex reflexive strategy is used). Consistent with K/S and K (the least complex reflexive strategy is used).

Guadeloupean Personal pronoun

Non-other- kaché, kamouflé, Zero-marking is less complex directed bengné, abiyé, pengné, and hence would have been more maské, okipé consistent with K/S and K. Otherdirected

touvé, mété, gadé, démaré, swiyé, fòmé

Possessive Non-other- viré, pronmlé, constr. with kò directed débouyé, fouté, mété a ri, senti, oublié Otherdirected Zero marker


fèmé, sové, senti, paré, Consistent with K/S and K (the Most kouvé complex reflexive strategy is used).

Non-other- touné, kaché, goumin Consistent with K/S and K (the least directed complex reflexive strategy is used).

Morisyen Personal pronoun

The possessive constr. and/or the constr. with mem are more complex and hence would have been more consistent with K/S and K.

Non-other- Not attested in the directed corpus

Reflexivity in French-based creoles Otherdirected

inform, santi, okipe, lev, servi, diman, blese, dominn

Personal pron. Non-other- defann + mem directed Otherdirected

Possessive constr. with lekor

Zero marker


envit, koz lor, pandi, Consistent with K/S and K (the most zet, ouver ant, trouv, complex reflexive strategy is used). donn, koz ar, servi, ferm, dakor ar, dir, dir ar, apel, zis ar, promet, pran, konfians dan, remet, fors, repran lor, ouver ant

Non-other- defann, amiz, relax directed Otherdirected


pandi, grat, zet, touy, kal, lev, nouri, tranp, rann

not consistent with K/S and K (a less marked reflexive strategy exists). Consistent with K/S and K (the most complex reflexive strategy is used).

Non-other- rapel, amize, leve, directed asize

Consistent with K/S and K (the least complex reflexive strategy is used).

Non-other- prepar, maske, bey, directed devlop

Zero-marking is less complex and hence would have been more consistent with K/S and K.

Seselwa Personal Pronoun

Otherdirected

kondwir, war, met lor, gay


Personal pron. Non-other- Not attested in the + mem directed corpus dir

Consistent with K/S and K (the most complex reflexive strategy is used).

Non-other- miltipliy, eksprim directed


Otherirected Possessive constr. with lekor

koriz, evit, fra

Consistent with K/S and K (the most complex reflexive strategy is used).

Non-other- leve, vire, asiz directed

Consistent with K/S and K (the least complex reflexive strategy is used).

Otherdirected Zero marker

32

322

Katrin Mutz

Other factors, presumably decisive for the selection and distribution of the various reflexive constructions in (French-based) creoles, like the role of aspect and aktionsart of the predicate, prosodic restrictions, the discourse prominence of the antecedent (topic, focus), or other discourse organizing principles (Muysken and Smith 1995: 272) which await thorough analysis, can only be mentioned in passing here.

3. The “functional domains” of the elements involved in the reflexive constructions The strong variation between the reflexive markers, determined as it is by a number of different factors (see the preceding Section 2), seems to suggest that no reflexive marker in the creole languages has acquired a “fully” grammaticalized status yet. In languages with a “fully” grammaticalized reflexive marker, like French or German, the grammaticalized status shows up not only in the (syntactically specified) distribution of the respective marker, but also in its further functions of “deriving intransitivity”, e.g., Fr. se as a marker of anticausativity (La porte s’ouvre ‘The door opens’) or passivity (La Tour Eiffel se voit de loin ‘The Eiffel Tower can be seen from far away’). A look at the elements involved in the reflexive constructions shows that in the (basilectal) creoles there is no element whose “genuine” or only function is to mark reflexivity. Besides their reflexive meaning, the elements used in the reflexive construction convey at least one other function. But these non-reflexive functions are not due to further grammaticalization, as in the case of French, but represent, so to speak, the “precursors” in the grammaticalization process towards markers of reflexivity.¹⁷ In the following section I present the range of these non-reflexive functions. Personal pronouns The use of the personal pronouns as reflexive markers is derived from the original function of the personal pronouns as direct objects (Carden and Stewart 1988). In some creole languages this form is identical to the personal pronoun in subject position. In many creoles (especially in the Atlantic ones), a further function of the personal pronoun is that of a possessive marker.¹⁸ Example (25) illustrates the use of the personal pronoun mo (< Fr. moi) as a subject pronoun, an object pronoun, and a possessive pronoun. (25) Guyanese Mo di yé yé lésé mo wè si mo ka djéri mo kò pas sa 1SG say 3PL 3PL let 1SG see if 1SG ASP cure 1SG body because DEM simen-an mo byen. i di mo enben. lò to ké pi mal to week-DET 1SG well 3SG say 1SG well when 2SG heart more bad 2SG


ké vin koté mo. mo di non m’ pa ké vin koté to. TEMP come place 1SG 1SG say no 1SG NEG TEMP come place 2SG ‘I told them: let me see if I get cured, because, this week, I feel fine. He told me: okay, if you feel worse, you will come to me. I said: No, I won’t come to you.’ (Pfänder 2000: 106f) In some creole languages, e.g., in Haitian, the pronoun conveys reciprocal meaning with plural referents (Corne 1988: 77). In this use ambiguity can arise between the direct object, the reflexive and the reciprocal meaning (26). (26) Haitian Yo wè yo. 3PL see 3PL ‘They saw them(selves)/each other.’ (Déchaine and Manfredi 1994: 201) Other creoles have genuine reciprocal markers, e.g., in Seselwa kamarad (< Fr. camarade ‘companion’) (see Bollée 1977: 66), and in Louisiana creole en-a-lot (< Fr. l’un l’autre ‘one another, mutually’) (see Neumann 1985: 262). Body (-part)-nouns The nouns involved in a reflexive possessive construction refer, depending on the language, to an essential body-part like in Haitian (tèt ‘head’), or to the body as a whole, as in Seselwa and Morisyen (lekor) and the creoles of Haitian and Guadeloupe (kò). The referential use of these nouns is still highly productive in the respective languages (in some contexts ambiguity can arise between the grammatical reflexive reading and the referential reading): (27) Morisyen So lekor pe tranble, so ledan klake, so fron kouver ar 3SG body ASP tremble 3SG tooth chatter 3SG fronthead covered with sier-froid. sweat-cold ‘His body trembled, his teeth chattered, his forehead was covered with cold sweat.’ (www 8) (28) Haitian Ou wè tèt ou. 2SG see head 2SG ‘You saw yourself; you saw your head.’ (Déchaine and Manfredi 1994: 204) Carden and Stewart (1988: 20) report for Northern Haitian another function of tèt, namely an emphatic use:

323

324

Katrin Mutz

(29) Northern Haitian Mwên mâde Mari ki mûn te kraze plat-la. Li êdike 1SG ask Marie REL people TEMP break plate-DET 3SG indicate tèt-Emil. head Emil ‘I asked Marie who had broken the plate. She said it was Emil.’ Mem Although a part of reflexive constructions in a number of creole languages (but not, e.g., in Haitian), mem alone cannot express reflexivity. Used “alone”, mem functions as an intensifier¹⁹ (König 2001, König and Siemund 2000), that can “intensify” a noun, a pronoun, an adjective, a verb or a whole sentence, as shown in (30)–(32): (30) Louisiana creole Sa se detestab mem! this is abominable INTS ‘That’s really abominable!’ (Neumann 1985: 341) (31) Haitian Emil blese li mem. Emil bless 3SG INTS ‘Emil himself got hurt; Emil hurt him.’ (Carden and Stewart 1988: 53) (32) Seselwa Napa person [. . .] zot pa war narjê mem. there-has-NEG person [. . .] 3PL NEG see nothing INTS ‘There was nobody . . . they didn’t see anything at all.’ (Bollée 1977: 142) In Morisyen and Seselwa, mem is said to convey a trace of an intensifier meaning even in its use in reflexive constructions, serving thereby to disambiguate the sentence when used with the third person (Corne 1988: 72): (33) Morisyen a. Lii fin pandi lii,j . 3SG ASP hang 3SG ‘He hanged himself; he hanged him.’ b. Lii fin pandi limemi. 3SG ASP hang 3SG-INTS ‘He hanged himself.’ In the creole of St. Louis (Louisiana), the original intensifier meaning of mem is completely “bleached”. Mem has become an obligatory part of the reflexive construction.²⁰ In all creole languages investigated, mem has the function of an identifier:


(34) Guadeloupean Ou we tout moun pa ka ni mem pense. Tout moun pa pense 2SG see all people NEG ASP have same idea all people NEG idea mem jan. same manner ‘You see that people don’t have the same ideas. They don’t think in the same manner.’ (www 9)

4. Conclusions This overview of the types of reflexivity construction in some French-based creoles has led me to delineate the factors that need to be accounted for in an analysis of the (distributional) restrictions on the different reflexive constructions. The functioning of these factors and their hierarchization in governing the distribution of the various reflexive strategies in each creole will have to be investigated in future research. In the last section, an outline of the “functional paradigm” of the elements involved in the reflexive constructions demonstrates that these elements are still tied to their original functional domain, although they have begun to move in the direction of (not yet fully) grammaticalized reflexive markers. Future research will have to consider the problems and issues raised in this discussion in order to gain a better understanding of the reflexive system of the Frenchbased creole languages.

Notes 1. In French-based creoles, reflexive constructions like Fr. se souvenir or s’en aller are generally explained away as acrolect constructions. 2. ‘Their disappearance [i.e., the disappearance during creolization of reflexive morphemes] has created a functional gap and has thus allowed the complete restructuring of a mostly non-transparent system.’ 3. Still, reflexivity is a marked grammatical category, as observed by LaPolla (1995). Reflexivity is not an essential grammatical device. 4. The creoles mentioned are to be understood as mere examples of the different construction types. This does not mean that the respective strategy is the only or the primary strategy in the languages mentioned. 5. With Emil blese li in Southern Haitian meaning “Emil hurts him” and not “Emil hurts himself ” (cf. Carden and Stewart 1988: 18). 6. In general, one has to be cautious when addressing the delicate notion of “zero-marking”. But the overt marking of reflexivity in imperative sentences with predicates which in other contexts show no reflexive marker (see §2.2) supports the validity of the claim that some creole languages use “zero-marking” to express reflexivity.

325

326

Katrin Mutz 7. Haitian: 75 tokens/48 types; Seselwa: 103 tokens/54 types; Morisyen: 63 tokens/40 types; Guadeloupean: 209 tokens/73 types. 8. This corpus is based on Ludwig, Telchid and Bruneau-Ludwig (2001) as well as on narratives given in Rutil (1981). I also use examples and texts found in the linguistic literature and on the Internet (see references). 9. The orthography given here is that found in the source materials. Throughout this article, English translations of creole examples are mine, except for (17), (21), (26), (29), (31), and (33). 10. Diachronically the emergence of this kind of structure can be described by metonymic transfer in a grammaticalization process (Heine 2001). 11. For divergent opinions on the productivity of the personal pronoun strategy in Morisyen, see Corne (1988) and Adone (1994). 12. See note 7. 13. But Kriegel (1996: 146) notes this: “Der Parameter Mündlichkeit und Schriftlichkeit spielt bei der Markierung von echter Reflexivität keine zentrale Rolle. Es kann lediglich eine Tendenz beobachtet werden, nach der in nähesprachlichen Diskursen häufig ganz auf morphologische Markierung verzichtet wird” (‘The parameters “orality” and “written discourse” do not play a central role in the marking of “true reflexivity”. One can merely observe the tendency that, in less formal or more personal discourse, morphological marking is frequently absent’ [my translation]). 14. In written speech, acrolectal influence is to be taken into account. 15. Due to the imprecision of the term “other-directedness” it is not always clear whether an action or a situation should be described as other-oriented or non-other-oriented. 16. But see section 3, where the grammaticalized status of the primary strategy is scrutinized. 17. Heine states: ‘Older reflexive forms may acquire further grammatical uses as reciprocal, middle, anticausative, and passive markers [. . .] While some creole reflexives have in fact assumed reciprocal and to a limited extent also middle uses, we are not aware of any creole reflexive that has given rise to an anticausative or a passive construction. In short: The time available was too short for creoles to develop verbal or inflectional reflexives, or to undergo further grammaticalization processes.’ (2001: 44) and ‘We also noted that we have found no creole where personal pronouns have been grammaticalized to the extent that they serve no purpose other than expressing reflexivity’ (2001: 32). 18. There are at least two possible analyses of this state of affairs: the common view is that the pronoun is described as being polyfunctional (the reference of the pronoun is the same in all four uses, only its grammatical relation or function changes); another analysis is to claim the existence of four homonymic pronouns: 1. a personal pronoun in subject position, 2. a personal pronoun in object position, 3. a possessive pronoun, 4. a reflexive pronoun. This second analysis runs counter to the creole language system which in different domains and at different levels is characterized by a tendency towards “unspecificity” concerning function or category (e.g., no clear-cut boundaries between syntactic categories). 19. In his description of the reflexive system of Morisyen, Corne (1988: 78) presents a sentence in which li mem does not convey reflexive meaning, as Corne claims, but the emphatic, intensifying meaning mentioned here: Personne pas conne l’ouvraze Bon Dié plis bien qui li même ‘Nobody knows God’s work better than (God) Himself ’.

Reflexivity in French-based creoles 20. The analysis of the creole reflexive constructions given here seems to contradict the following generalization made by König: In languages which draw a clear distinction between reflexive pronouns and intensifiers [. . .], intensifiers can be adjoined to both (referential) NPs and pronouns without restrictions, but they cannot occur in argument positions without an accompanying focus [. . .]. Languages which lack a clear formal distinction between intensifiers and reflexive anaphors may also allow intensifiers in argument positions without accompanying pronouns [i.e. in subject-position] (2001: 752)

References Adone, D. 1994. “Creolization and language change in Mauritian Creole.” In Creolization and Language Change, I. Plag and D. Adone (eds.), 23–43. Tübingen: Niemeyer. Bollée, A. 1977. Le créole français des Seychelles. Tübingen: Niemeyer. Carden, G. 1993. “The Mauritian creole lekor reflexive: Substrate influence on the target-location parameter.” In Atlantic Meets Pacific. A Global View of Pidginization and Creolization. Selected Papers from the Society for Pidgin and Creole Linguistics, F. Byrne and J. Holm (eds.), 105–17. Amsterdam: John Benjamins. Carden, G. and W. A. Stewart. 1988. “Binding theory, bioprogram, and creolization: Evidence from Haitian Creole.” Journal of Pidgin and Creole Languages 3: 1–67. Carden, G. and W. A. Stewart. 1989. Mauritian Creole reflexives: A reply to Corne.” In Journal of Pidgin and Creole Languages 4: 65–101. Corne, C. 1988. “Mauritian Creole reflexives.” Journal of Pidgin and Creole Languages 3: 69– 94. Corne, C. 1989.“On French influence in the development of creole reflexive patterns.” Journal of Pidgin and Creole Languages 4: 103–15. Déchaine, R.-M. and V. Manfredi. 1994. “Binding domains in Haitian.” Natural Language and Linguistic Theory 12: 203–57. Dowty, D. 1989. “On the semantic content of the notion ‘thematic role’.” In Property Theory, Type Theory and Natural Language Semantics, G. Chierchia, B. Partee and R. Turner (eds.), 69–130. Dordrecht: D. Reidel. Faltz, L. M. 1985. Reflexivization: A Study in Universal Syntax. Dordrecht: Foris. Frajzyngier, Z. and T. S. Curl, (eds.). 2000. Reflexives. Forms and Functions. Amsterdam: John Benjamins. Frajzyngier, Z. 2000. “Domains of point of view and coreferentiality.” In Frajzyngier and Curl, 123–52. Geniušiene, E. 1987. The Typology of Reflexives. Berlin: Mouton de Gruyter. Grimshaw, J. 1982. “On the lexical representation of Romance reflexive clitics.” In The Mental Representation of Grammatical Relations, J. Bresnan (ed.), 87–148. Cambridge, MA: MIT Publications. Grimshaw, J. 1990. Argument Structure. Cambridge, MA: MIT Publications. Haiman, J. 1985. Natural Syntax: Iconicity and Erosion. Cambridge: Cambridge University Press. Hall, R. A. 1953. Haitian Creole. Grammar, Texts, Vocabulary. Philadelphia: American Folklore Society. Heine, B. 2001. Accounting for creole reflexive forms. Preliminary version (http:// www.pca. uni-siegen.de/list/num.html).

327

328

Katrin Mutz Kemmer, S. 1994. “Middle voice, transitivity, and the elaboration of events.” In Voice. Form and Function, B. Fox and P. Hopper (eds.), 179–230. Amsterdam: John Benjamins. König, E. 2001. “Intensifiers and reflexive pronouns.” In Sprachtyplogie und sprachliche Universalien. Ein internationales Handbuch, M. Haspelmath et al. (eds), 747–60. Berlin: Wilhelm de Gruyter. König, E. and P. Siemund, 2000. “Intensifiers and reflexives: a typological perspective.” In Reflexives. Forms and Functions, Frajzyngier and Curl,:41–74. Kriegel, S. 1996. Diathesen im Mauritius- und Seychellenkreol. Tübingen: Narr. Kriegel, S. 2000. “Distribution fonctionnelle des morphèmes réfléchis en créole mauricien et seychellois.” Etudes Créoles 23: 66–78. LaPolla, R. J. 1995. “On the utility of the concepts of markedness and prototypes in the development of morphological systems.” The Bulletin of the Institute of History and Philology 66: 1149–1185. Lefebvre, C. 1998. Creole Genesis and the Acquisition of Grammar. The Case of Haitian Creole. Cambridge: Cambridge University Press. Ludwig, R. 1996. Kreolsprachen zwischen Mündlichkeit und Schriftlichkeit. Tübingen: Narr. Ludwig, R., S. Telchid, F. Bruneau-Ludwig. 2001. Corpus créole. Textes oraux dominicains, guadeloupéens, guyanais, haïtiens, mauriciens et seychellois. Enregistrements, transcriptions et traductions. Hamburg: Buske. Mohanan, K. P. and T. Mohanan. 1998. “Strong and weak projections: Lexical reflexives and reciprocals.” In The Projection of Arguments. Lexical and Compositional Factors, M. Butt and W. Geuder (eds.), 165–94. Stanford: CSLI Publications. Mufwene, S. 2000. “La fonction et les formes réfléchies dans le mauricien et le haïtien.” Langages 138: 114–27. Muysken, P. 1993. “Reflexes of Ibero-Romance reflexive clitic + verb combinations in Papiamentu: thematic grids and grammatical relations.” In Focus and Grammatical Relations in the Creole Languages, F. Byrne and D. Winford (eds.), 285–301. Amsterdam: John Benjamins. Muysken, P. and N. Smith. 1995.“Reflexives.” In Pidgins and Creoles. An Introduction, J. Arends et al. (eds), 271–88. Amsterdam: John Benjamins. Neumann, I. 1985. Le créole de Breaux Bridge, Louisiane. Étude morphosyntaxique. Textes. Vocabulaire. Hamburg: Buske. Papen, R. A. 1978. The French-based Creoles of the Indian Ocean: An Analysis and Comparison. Ph.D Thesis, San Diego: University of California. Ann Arbor: MI University Microfilms. Pfänder, S. 2000. “Le créole guyanais: témoin d’une phase plus ancienne de la créolisation?” Etudes Créoles 23: 101–16.

Sources Fam Lite 6 http://pages.intnet.mu/develog/proz/zistoirlong/kalipa.htm (www 1, 14–2–2003) http://pages.intnet.mu/develog/tizistoir/ernestotm (www 2, 14–2–2003) http://www.i-port.net/sd-in-j/author.htm (www 3, 14–2–2003) http://pages.intnet.mu/develog/tizistoir/glasgow.htm (www 4, 14–2–2003) http://www.haiti-progress.com/2000/sm000927/XCREOLE1.htm (www 5, 14–2–2003) http://www.haiti-progress.com/1999/sm991124/Xperpsec.htm (www 6, 14–2–2003)

Reflexivity in French-based creoles http://pages.intnet.mu/develog/proz/zistoirlong/jamouna.htm (www 7, 14–2–2003) http://pages.intnet.mu/develog/proz/klekle.htm (www 8, 14–2–2003) http://www.zoukarchive.com/book/December2000.shtml (www 9, 14–2–2003)

329

CHAPTER 15

The role of style and identity in the development of Hawaiian Creole* Sarah J. Roberts

Stanford University

According to available evidence, Hawai’i Creole English (HCE) emerged as a language increasingly distinct from Hawai’i Pidgin English (HPE) and Standard American English (SAE) between the 1900s and 1930s, and its development predominantly involved local-born children and adults. In this paper I consider how linguistic ideology and group identity factors, which bear more directly on social motivation, may have influenced the development of creole “continua” — dimensions of linguistic variation which converge with and diverge from the idealized speech of in-group and out-group identity categories. The linguistic elaboration attested in the development of HCE is thus viewed as essentially stylistic. Modernist ideologies promoted the shift from ancestral languages (ALs) to HCE. Evidence of HCE’s adoption of identity-marking functions from ALs may be found in attested linguistic attitudes pertaining to HCE and SAE.

1. Introduction Style, or situationally dependent intraspeaker variation, has often been a neglected area of creole research. Most studies focus on interspeaker variation and examine only the single dimension of variation between the basilectal creole and the colonial standard variety, often referred to as the post-creole continuum (Bickerton 1975, DeCamp 1971, Rickford 1987). In accord with the “life-cycle” notion of pidgin/creole genesis (cf. Hall 1966), the post-creole continuum was once commonly thought to postdate the transition of a contact language from pidgin to creole, arising through the decreolization of a relatively homogenous basilect. But many now recognize that the heterogeneity attributed to decreolization may exist early in the creole formation process and the basilect itself may not be the earliest form of the creole (Carrington 1992, Mufwene 1991, 1994, Winford 2000). At the same time, creole development has been linked to style as it involves an expansion of expressive resources and new situational uses as the creole takes over social functions formerly held by ancestral languages (Mühlhäusler 1980, 1997, Valdman 1977).

332

Sarah J. Roberts

In this paper, I will draw on recent sociolinguistic work on identity and style, as well as evidence from the development of Hawaiian Creole, to propose an approach that directly relates social meaning to creole development via stylistic variation. Thomason and Kaufman (1988), Kouwenberg (1992), Meyerhoff and Niedzielski (1994), and others have stressed the importance of linguistic negotiation in the formation of contact varieties. As Bell (1997) points out, such processes of accommodation underlie style. One recent model of creole development, proposed by Siegel (1997) and drawing on LePage and Tabouret-Keller (1985), claims that an individual creole develops coherence from a heterogeneous assortment of variants through leveling in a context of social unification. This convergence involves the formation of a shared group identity — which arises through differentiation from other identities, and the convergent variety in turn is divergent from other previously-existing varieties. Identity has the same role in stylistic divergence in non-creole languages (Bell 1997, Irvine 2001) and creole development may itself draw on synchronic style-shifting.

2. Basilect formation as linguistic divergence Sociolinguistic inquiry into diachronic change in dialectal English has highlighted the processes of divergence and convergence and the role of identity in motivating differential use of linguistic variables (Auer and di Luzio 1988, Esch and Jones 2001, Labov 1963). Historical and contemporary evidence of change in dialect distance between North American White Vernacular Englishes (WVEs) and African-American Vernacular English (AAVE) has especially inspired debate on linguistic divergence and convergence (Bailey and Maynor 1989, Fasold 1981, Rickford 1992). Varying concepts of linguistic divergence have been expressed within creolistics. Chaudenson (1992) and Mufwene (2001: 50–4) both view creoles as emerging via gradual drift from colonial varieties in a social context of decreasing access to native speakers, a process Mufwene terms “basilectualization”.¹ Other approaches stress the role of pidginization in eliminating superstratal features and innovating new structures, while creole formation itself has sometimes been treated as involving a structural elaboration and expansion of pidgin grammar (Bickerton 1981, Mühlhäusler 1980: 22, Valdman 1977). The case of Hawai’i Creole English (HCE) is particularly interesting, as it emerged relatively recently in a rather literate colonial society. This circumstance has led to the documentation of all the developmental stages of HCE from its very inception (Roberts 1998: 5–6). Documentary evidence shows that HCE began to split off from an earlier form of pidgin English, Hawai’i Pidgin English (HPE), roughly one hundred years ago, and by the 1930s it was recognized as a distinct language by contemporary observers. Reinecke, who in the 1930s investigated the origins and development of HCE, wrote that it was around 1900 that “the present dialect of what may be called ‘the

The role of style and identity in the development of Hawaiian Creole

second generation’ — and the third as well — and those influenced by them, then began to take form. . . . It is decidedly different from the English spoken by the immigrants who came to Hawaii as adults” (Reinecke to Elsie Wilcox, 8 December 1932, p. 1; cited in Roberts 2000: 286). He also claimed that the 19th-century pidgin of foreign-born laborers and Hawaiians “was the basis from which the children of the immigrants and natives, educated in the English language schools, derived a form of English also commonly called pidgin English but more adequate and more refined than the makeshift speech of immigrants” (1969 [1935]: 194). Kaapu (1937: 95–6) expressed a similar view around the same time: That the pupils of the public schools now speak a language very different from the pidgin used by the immigrant forbears of most of them, yet in varying degrees having remnants of the vocabulary and syntax of pidgin is quite evident to the careful observer.

Even students at Maui High School adopted the term “Jargonese” in 1938 to distinguish their language from that of the immigrant “first generation” (Hawaii Educational Review, January 1939: 139). All this indicates that by the 1930s, a distinct variety of HCE had emerged, associated in the popular imagination with the local-born population, and distinguished linguistically and ideologically from the speech of immigrants. Linguistic evidence from old pidgin/creole texts also attests a diverging trajectory of syntactic and lexical change in HCE between 1900 and the 1930s. The earliest texts attributed to the locally born and/or children (between 1890 and 1910) are little different from other contemporary texts of HPE. But by the 1910s, certain low-frequency items present in early HPE, such as past tense been and the infinitive marker for, had become much more frequent in texts attributed to the locally born, especially children (Roberts 1998, 1999). In the 1910s, been was almost an obligatory feature in texts attributed to children, suggesting that it had become a stereotype, a shibboleth, of the speech of children — an exaggeration, surely, of linguistic practice but no doubt reflecting an actual linguistic divergence. Then, new features appeared in non-immigrant texts in the 1910s and 1920s which were either wholly absent or exceptional in texts attributed to immigrants. Examples from the 1910s include the finite complementizer for (see Roberts 1998: 26–32 for discussion), the NP-introducing copula dass, dasswhy ‘because,’ and the adjectival intensifier some. In the 1920s, novel features include the progressive marker stay (see Roberts 1999: 59–60 for discussion), boy as a clause-final discourse marker, that kind as a filler noun phrase, wen as past or perfective marker, hybolic ‘difficult aspects of a language, highbrow use of language’ as a metalinguistic term, and slang terms such as zaza kid ‘mischievous youth’ and shaka ‘wonderful, excellent.’ By the 1930s, the first instances of tense/aspect marker combinations (such as been stay for past progressive) appeared in texts attributed to the locally born (Roberts 1999: 60–3). These innovations represent structural and lexical changes which contributed greater stylistic and functional flexibility: the integration of auxiliaries into a more

333

334

Sarah J. Roberts

nuanced tense/aspect system, phonological erosion (been becoming wen, dasswhy becoming asswhy),² discourse markers, metalinguistic terms, and slang. At the same time, certain older HPE features decreased in use over time in localborn texts. All same ‘same as, thus’ was replaced by just like and like that, the use of me as first person subject pronoun became very rare by the 1930s, stay mostly replaced stop as locative verb, if displaced suppose as conditional marker, plenty mostly displaced too much as intensifier and quantifier, and for what and for why mostly displaced what for ‘why’. Later texts of local-born speech have fewer “archaic” features and more “HCE-exclusive” features. Finally, fairly recent HPE features (such as SOV word order, rather rare in late 19th-century HPE but frequent in plantation varieties from the 1910s onward) were not adopted into the speech of the locally born. These data indicate that HCE developed in part as a divergence from the earlier HPE spoken by immigrants and older Hawaiians, and that the HCE speech community largely consisted of local-born children and adults. The following samples of “pidgin” attributed to local-born children illustrate these changes: Phase I (minimal divergence from HPE) (1) “Me go look stars — go look stars.” “Too much work — too warm — some more tomorrow — no use — mahope.” “Me lend!” “Me too lend!” (1891; spoken by part-Hawaiian children at school, Makapela, Hawai’i) (2) “You funny kind, pupule haole [crazy white person], I no got tail.” “I no like you, you too much cheeky haole, nobody like you.” (1901; spoken by a Hawaiian girl, Honolulu, O’ahu) Phase II (moderate divergence) (1) “The Indian he got one gun, he been shoot one cowboy.” “Oh Tony, where you been get?” “Oh Tony, suppose policeman come, what you do?” “Suppose you no like be girl you no need play.” (1915; spoken by local-born children at play, O’ahu) (2) “Where you stop this time?” “She no stay, she stay hospital.” “This time he stay coming.” “He been tell he like one hoe.” “John he get one bicycle.” “Misao been blow my face.” “He get too much money.” “Suppose you no go, you sorry.” “You been eat lunch already?” (1921; spoken by local-born children) Phase III (maximal divergence) (1) “Oh yea, that guy been go Maui High befo’ yea?” “Naw, that guy stay going on the big kind small boat now to India. You tink zaza same kid been stay Kahului befo’, yea?” (1924; conversation between Maui High School students watching a soccer game at school)


(2) “You no going get candy.” “The boys stay come on the bus.” “He get two pairs shoes.” “She been stay wearing one ribbon.” “The girl been get dirty nails.” “You get plenty junks.” “I been stay play with him yeah, and he hit me you know.” “The two twins been stay in school.” “I been see plenty pictures.” (1937; spoken by schoolchildren, Maui) This suggests that immigrant adults contributed little to the development of HCE.³ This is seemingly at odds with the recent influential view that adults — not children — are responsible for creole development (see especially Mufwene 2001: 131).⁴ The important contrast however is not adults vs. children but locally born vs. foreign born, and the conflict arises in part from a rigid division between children and adults. Demographic evidence shows that the local-born population increased fairly rapidly after massive immigration commenced in the 1880s. This situation differs considerably from what is known about other creologenic societies, particularly those in the Atlantic where native-born populations were slow to grow (Roberts 2000). In Hawai’i, the local-born population was largely under the age of twenty in the 1900s and 1910s when most of the primary structural developments began to transpire. But at that time, children mostly learned “pidgin” in school, from play friends, or older siblings — not in early childhood. The exceptions were third-generation children (grandchildren of immigrants, i.e., sansei in the case of Japanese) who acquired “pidgin” in the home as a native language from local-born parents. Third-generation Chinese and Portuguese had indeed arisen by the 1910s. So adults did play an important early role in this process, as well as teens and children in middle childhood (including those likely still within the critical period). The divergence that contributed to the formation of basilectal HCE was not just limited to HPE but also involved a divergence from WVEs, and specifically the versions of American Standard English (ASE) taught in the schools.⁵ Reinecke noted that basilectal HCE distinguished itself from both HPE and ASE via different continua of variation: Because the colonial dialect [i.e. HCE] forms part of the same speech continuum with the creole dialect [i.e. HPE], it is difficult to single out any speech element and say, “This is ‘pidgin,’” or, “That is dialect.” The aggregate of the speech elements, not any single element, is what matters in drawing the distinction between the two forms of communication. . . . While the dialect differs as markedly from the Standard English as the creole dialect does, the two unorthodox forms of speech differ just as markedly from each other. (Reinecke 1969 [1935]: 160–1)

Linguistic data suggest a very similar picture. Table 1 displays variation of lexical and syntactic features in speech attributed to immigrants and local-born speakers in pseudonymous letters and creole plays. Since these represent fictional characters or writers, the speakers are labeled as personae. Persona A is portrayed as a Chinese HPE-speaking immigrant. His speech differs markedly from the other personae. No preverbal tense/aspect markers occur, and his distinctive use of suppose, me, stop,

335

336

Sarah J. Roberts

and too much is not shared by the other personae. Interestingly, these features do occur commonly in local-born texts up to the early 1920s (see examples 1–4 above), but not the 1930s, which is the decade mostly represented in Table 1. Other HPE features however were shared to varying degrees by local-born personae — such as possessive got by Personae B, C, and D (but lacked altogether by E), habitual all time was shared by Persona B, and what for by Persona E. Features convergent with ASE also varied by persona. As regards the use of stay for location and progressive, B is closest to ASE (as well as most other varieties of WVEs spoken in Hawai’i) by lacking stay altogether, where we find VERB -ing for progressive, was VERB -ing for past progressive, and ees ‘is’ for location. Persona C uses stay only for location, so we find VERB -ing for progressive and past proTable 1. Variation of linguistic features in speech attributed to immigrants and local-born speakers Personae Features

A

B

C

D

E

Past tense marking

ADV

ADV, been

ADV, wan

ADV, been, wan

ADV, been

Future tense marking

ADV

ADV, going

ADV, goin

ADV, gon

ADV, goin

Progressive marking

repetition of Ving verb

Ving

stay V

stay Ving

was Ving

Ving

was V

been stay V(ing)

Past progres- repetition of sive verb Habitual marking

all time

eri time,

every time

stay V, all time, stay V, eri time errytime

Conditional

suppose

eef

if

if

eef

1st singular

me

I, me I

I

I, me I

I, me I

Location

stop

ees

stay

stay

stay

Possession

got

have, got, get

get, have, got

have, got

get, have

“Many, alot”

too much,

plenty

planty

planty

plenty

‘Why’

what for, wasamata

for what

for why

for why

why, wat fo

“Same as, thus”

all same

just like, like, like dat

same like

just like, like, like dat, like dat

jes like, like,

Personae : A = elderly Chinese immigrant (Bond 1936), B = Portuguese local-born adult (Anonymous 1932a), C = Hawaiian and Chinese local-born girls in mutual conversation (Chun 1936), D = local-born high school senior (Anonymous 1932b), E = middle-aged Hawaiian adult (Anonymous 1946).


gressive. Persona D uses stay for both progressive and locative but was VERB occurs for past progressive, convergent with ASE in the use of was but lacking aspectual inflection. Persona E is most basilectal or divergent from ASE in its use of stay for location, stay for progressive, and the tense/aspect combination been stay for past progressive. Tense/aspect combinations are among the most radical features in basilectal HCE. Such evidence suggests a multidimensional continuum of variation, one axis diverging from HPE and another diverging from ASE. Variation cannot be described simply in terms of a continuum between basilectal HCE and ASE since speakers also varied in the extent to which their speech diverged from HPE. The diachronic picture is that divergence from HPE and ASE increased over time, expanding the overall continua in the direction of the basilect. Convergence with ASE would have also increased diachronically, as exposure to ASE and its acquisition and use increased over time in the English language schools, employment, and other institutionally based contexts. The post-creole continuum between ASE and HCE did not emerge via decreolization; mesolectal and, especially, acrolectal varieties of HCE instead represent developments in the creole formation process concurrent with the development of the basilect (cf. Alleyne 1971).⁶

3. Linguistic divergence and stylistic variation Linguistic divergence underlies recent sociolinguistic approaches to style. Bell (1997) regards style as interpersonal audience design, shaped by convergent and divergent relational forces arising from the situational context. The notion of audience design is rooted in speech accommodation theory (cf. Street and Giles 1982, Giles and Coupland 1991) which posits convergence and divergence as basic accommodation strategies deployed by speakers to indicate identification with, or dissociation from, the speech of others. These strategies are rooted in psychological desires for affiliation and social interaction. Bell (1997: 244) characterizes the production of style in the following manner: 1. 2. 3. 4.

Group has its own identity, evaluated by self and others. Group differentiates its language from others’: “social” or interspeaker variation. Group’s language is evaluated by self and others: linguistic evaluation. Others shift relative to group’s language: “style” or intra-speaker variation.

Identity therefore serves as a basis of social and linguistic differentiation, and evaluation links social attitudes toward groups’ patterns of variation. These attitudes then filter down to the level of intraspeaker variation. Bell states:“Style derives its meaning from the association of linguistic factors with particular social groups” (1997: 243). Bell’s model takes as its starting point predefined “groups” which already possess their own identities. It accepts uncritically the notion, questioned by Butler (1990), Cameron (1998), Barrett (2001), and many others, that identity is a pre-

337

338

Sarah J. Roberts

determined and stable fact instead of a construct constituted through social and linguistic practice. Identity cannot be separated from the social performances that create and reproduce social meaning (Eckert 2000). In the case of HCE, this fact is especially clear since the identities commonly associated with the language — such as “Local” identity in contemporary Hawai’i (Leong 1997, Hormann 1960, Masuda 1993, Okamura 1980, Tamura 1996, Yamamoto 1979) — developed with the creole, and did not precede the formation of local-born communities. The approaches to style by Irvine (2001) and Gal and Irvine (1995) posit that identity and style are co-constructed and regard identity categories as constituted semiotically via the situational use of linguistic variables. Style may be viewed as a production in which an aggregate of linguistic variables work together to index a particular social meaning, resulting in both linguistic and social differentiation. Instead of treating stylistic variation as merely reflective of one’s social address or identity, they view style as the linguistic means through which identity is produced in discourse (cf. Podesva, Roberts and Campbell-Kibler 2001: 178–81). Irvine mentions that distinctiveness structures both identity and style: “Whatever ‘styles’ are, in language and elsewhere, they are part of a system of distinction, in which a style contrasts with other possible styles and the social meaning signified by the style contrasts with other social meanings” (2001: 22). It is crucial to understand that this process occurs at an ideological level. Mufwene (2001: 81–6) rightly points out that there is no such thing as a monolithic ASE; rather, creoles exist alongside a set of varying regional and social varieties (especially British and North American WVEs). Similarly HPE itself was not a homogeneous language but had its own regional, ethnic, and individual heterogeneity. Irvine (2001: 33–4) points out that linguistic divergence involves the semiotic process of erasure, which reduces speakers to ideologically defined groups and reduces idiolectal variation to homogeneous, bounded languages. In linguistic practice, speakers probably style-shifted with these constructs in mind — thereby reducing the range of WVEs to “good English”. The role of such constructs in shaping stylistic variation does not mitigate the fact that the speech situation was actually far more complex. Ochs (1991) provides an explicit ideological model of how linguistic variables come to express social meanings. Some features, such as pronouns he and she with respect to gender, directly index identity categories. But most only indirectly index identity categories through intermediate pragmatic meanings relevant to the immediate situational context. A familiar example is/-in/vs./-ing/variation (as in walkin’ vs. walking), which produces a stance of informality in the immediate conversational context but which also indexes working-class identity and is meaningful ideologically in contrasting a friendly, close-knit working class group against a more institutionally based middle class. To use another example, a speaker of a style containing a high degree of variables indexing a particular gender identity is not necessarily intending to communicate “I’m a middle-class white female” but her style has the effect of constructing her in such a manner. In Ochs’ view, identity


categories are constituted through such indirect relations and derive their semiotic meaning through linguistic practice. This approach to pragmatic indexicality may serve as an explicit model of how identity is performed linguistically. My proposal in this paper is that the linguistic divergence that diachronically resulted in the differentiation of HCE was synchronically realized as situational styleshifting. Situational contexts which favored the use of highly divergent versions of HCE were those that also involved ideological dissociation from the speech of speakers of WVEs and HPE, while contexts which favored the use of highly convergent versions of HCE were those that involved maximal accommodation to the linguistic practice of said groups. Speakers would have varied in their access to, and involvement in, the individual situations and communities of practice that place norms on language use (cf. Eckert and McConnell-Ginet 1992). This would have then resulted in variation in the kinds of styles controlled by speakers, producing the creole continua of interspeaker variation. The web of identity invested in the range of styles used in creole communities had a dynamic effect in constraining the consolidation of features (as judged in linguistic evaluation) into a single recognized variety, and thus shaped the drift into divergence.

4. Linguistic ideology and identity in Territorial Hawai’i As discussed above, the linguistic performance of identity is a crucial social function of language and as such it played an important role in creole formation. The 1900s and 1910s were a period of substantial language shift from ancestral languages (ALs) to HCE, particularly for Hawaiian, Portuguese and Chinese locally born. Through the course of language shift, HCE became increasingly invested with the identity-marking functions of language, and thus was central to the construction of local-based identities. So what identity categories were salient at the time HCE emerged, and what relation did they have with linguistic practice? One very important identity project in the 1910s was the negotiation of ethnic identity with hegemonic “American” identity, which was fairly new at the time — as Hawai’i was annexed to the United States in 1898 and became a Territory only in 1900. The process of language shift was motivated a great deal by the adoption of colonial political identities. In the preceding period (from 1880 to about 1900), there was widespread multilingualism in the local-born group. Local-born children readily learned other ALs, and code-switching was very frequent (Roberts 2000). Pidgin had not yet emerged as the common medium of peer-based communication for local-born children. But there was a key shift in linguistic ideology around 1900–ALs increasingly marked speakers as foreign, non-American, non-citizens, non-local, while “Pidgin”, as the accessible form of English in the community, was identified as the language of America, and was thus valued over ALs. ALs soon became stigmatized in in-group discourse. Remarks in life histories written by local-born teens born between 1902

339

340 Sarah J. Roberts

and 1913 discuss this situation.⁷ One Portuguese girl expressed a highly negative evaluation of her AL and gave a political justification for her attitude: Sometimes my mother talks to me in Portuguese language. I hate this. I have told her many a many times not to because this is an American country, not a Portuguese country. (Life History, HJ-88 [Portuguese])

Another writer mentioned what happened in 1913 when she tried to use her AL at school: When I first entered school I used to speak Chinese to the girls whom I knew. The Hawaiian girls used to tease me,“Eh, you talk pake [Chinese], no shame.” From then on I became very bashful and never spoke much Chinese. (Life History, N-58 [Chinese girl, born 1907])

This situation stands in stark contrast to that in the 1880s and 1890s when ALs were freely used in school. Norms were forming — ALs were fine at home and in ethnic contexts, but not in peer discourse. ALs were viewed as un-American and were increasingly devalued in peer and institutional contexts where American identity was stressed. The same writer also noted that ridicule extended to other cultural practices, including the wearing of clothes: I started to attend the Halawa school when I was about six years of age. I wore Chinese clothes at times. The children used to tease me “Hey, you wear pake pants” so I discarded my Chinese clothes. Mother could not make me wear them.

The use of ALs in peer contexts persisted most often among Japanese locally born, who were highly conservative in retaining ancestral practices (Roberts 2000). They comprised about half of the local-born population by the late 1910s, and thus faced less pressure to accommodate to the other groups. Their differing linguistic practice and network relations reified the Japanese as a distinct group, and individuals from other ethnic groups often evaluated them as too “cliquish”: One nationality that I could not get along with was the Japanese. In school I tried to go with Japanese girls for awhile but whenever they got together they spoke in their own language and I did not know what they were talking about. (Life History, N-26 [Chinese girl, born 1907]) I haven’t got any use for Japanese because most of them are hard-boiled and always talk about you in their own language. (Life History, MK-144 [Russian local-born child, born ca. 1910]) They would talk in their native tongue. It made me feel as though they were talking about me. I tried to keep away as much as possible but at times I’d think it over and go back and make friends with them. (Life History, MK-103, [Hawaiian boy, born ca. 1909])

Over time local-born children worked out individual solutions in the negotiation of ancestral practices and abandoned ALs in peer discourse through the adoption of American identities:

The role of style and identity in the development of Hawaiian Creole I held my head high for being a Chinese. To carry this attitude further, I refused to be dressed in American dresses until I was in the grammar grades when I realized the importance of assimilation and Americanization, then I discarded my native blouses and pantaloons. . . . Although I persisted to be “Chinesey,” I gradually allowed myself to associate with classmates that were other than Chinese and in this way was I able to speak English with more ease and fluency. (Life History, H-23 [Chinese girl, born ca. 1903]) The girls go to extremes: they paint, powder and bob their hair, frizzing it up, all just to look like their American schoolmates. They belittle the customs and traditions of their fathers and adopt these so-called American customs in an effort to appear the “Real thing”. . . .They have a wonderful opportunity to profit by their parents speaking other languages, but they seem to take pride in knowing how to speak just English. (Life History, H-25 [Anglo-American girl, born ca. 1906])

Within ethnic communities, the linguistic and social practice of local-born children also often constructed them as distinct from the foreign born: He [the writer] condemned his people for evading the Chinese customs and following these INDECENT Occidental practices. He condemned the numerous debaucheries and downfalls of young girls and boys. He attributed these as results of their adoption of the foreign customs and practices. (Life History, H-36 [foreignborn Chinese boy, born 1903 in Nam Long, Canton]) I often regret that I did not go to Chinese language schools for the people of the older generation often call us “Hawaiian Pigs,” because we did not know the Chinese language and customs. However I have often wondered why they came here if they wanted to stick to their old customs. (Life History, N-26 [Chinese girl, born 1907]) According to racial classification we are Japanese and yet the people of Japan are rather inclined to look down upon us merely “Hawaiian-borns” — just the descendents of immigrants. Even the older Japanese residents here comment on our action as “typical of the Hawaiian-born” when it is not according to their view and “exceptional for a Hawaiian-born” if it were otherwise. (Life History, N-21 [Japanese girl, born in 1907])

However, the shift from ancestral cultural practices was mediated by an awareness that “American” practices were also Haole, or “white”, and the social appropriateness of specific practices were under constant debate and contestation. This extended far beyond language to all the other practices that indexed identity. The examples above suggest that Western clothing was a fairly safe means of expressing American identity, though many foreign-born parents objected to the wearing of “balloon pants” or “sailor mokus” by local-born boys. One classic example of a problematic practice in the 1920s was hair bobbing which lay at an intersection of many identities, including gender (girls were sometimes teased for having the “boy-cut”), urban identity (long-haired girls new to Honolulu schools were sometimes teased for their braids, while certain bobbed girls from Honolulu new to rural schools could barely wait for their hair to grow

34

342

Sarah J. Roberts

out), and ethnicity (foreign-born Japanese and Chinese parents often viewed hair bobbing as a denial of ethnic heritage). It was also problematic for invoking white racial identity in addition to American colonial identity and modernity. In 1924, a white schoolteacher in Kealakekua, Hawai’i, tried to convince one of her Japanese students to bob her hair for comfort’s sake but the girl steadfastly refused and finally explained: “If I cut, when I come in school, everybody laugh. If American girl, nice; but Japanese girl, f-a-a-a-h-ny” (Honolulu Advertiser, 2 November 1924: 2). ASE, the language of the schools and white colonial culture, ideologically indexed Haole identity. Its use⁸ in in-group contexts was frequently censured by peers. This was because it projected the white vs. non-white racial division within otherwise egalitarian social relations among locally born (the semiotic process of recursivity in linguistic ideology, cf. Gal and Irvine 1995). One girl vividly wrote: When I was in Central Grammar I know we were taught to speak correct English both in and outside of school but when I returned to Maui, my old friends made fun of my English which embarrassed me. They said I was “stuck up” and “you think you Haole” so I had to use the pigeon English and I am having the hardest time breaking that habit. (Life History, LJ-81 [local-born girl, born ca. 1912])

Another student wrote in the early 1930s: When we students use good English we are sure that we will be called “haoles.” And of course we hate to be called “haoles” because of many reasons. (Reinecke 1969 [1935]: 213–14)

This type of name-calling was further racialized in taunts such as “black Haoles” and “sunburned Haoles”. Other taunts invoking the power asymmetry in ASE include a family of descriptors beginning with “high”: “hi-hat”, “high-brow”, “hi-toned”, and “hybolic”. “Hybolic” originally signified aspects of a language difficult to learn and was often used in this sense (typically in reference to ALs) in the corpus of life histories written in 1927, and frequently in the full form “hyperbolic”, but a few writers also used it in the sense of “highbrow use of English”, and one writer creatively produced the blend “hybrowlic English” to refer to ASE. Another early use of the term jokingly associated difficulty in speaking ASE with difficulties in acting “poo-bah” or “highbrow”: John Ventura [freshman at Lahainaluna High School], overheard discussing the difficulties of acting Poo-Bah: “Ah — I no can say. Too much hybolic. Tongue no can huli [turn].” (Lama Hawaii, 1927, p. 83; annual for Lahainaluna High School)

In the 1900s and 1910s, very few locally born had some command of ASE (aside from passive knowledge), and “pidgin” was used by students in the classroom with teachers and peers. It was not until the 1920s and 1930s that a good proportion of creole speakers had some competence in ASE. But by the 1910s ASE was already linked to Haole identity and stigmatized in peer-centered discourse. There was also a shift in institutional attitude towards HCE during this period. From the 1890s to the early 1910s, teachers and educators were fairly tolerant of a


pupil’s use of “pidgin”, since the emphasis at the time was to get students to speak English instead of ALs at school. But by the mid-1910s, a version of the Standard Language Ideology described by Lippi-Green (1997) began to assert itself aggressively in the Territory. Local-born teachers whose speech fell short of departmental standards were replaced (particularly towards the end of World War I) by white mainland teachers ignorant of Island English and inclined to view it as a parasitic aberration of ASE and in need of eradication. The increasingly confrontational attitude of the schools toward HCE in the 1920s reinforced the association of ASE with institutional and Haole identity. An irony in this development is that the hegemonic “American” identity, which originally promoted the shift to HCE, was now turned against it. Another development in the 1920s that further solidified the indexing of ASE with Haole identity was the inception of “English standard” schools in 1924. These schools were designed as “pidgin-free” zones of education for white children and local-born students competent in ASE. This resulted in a quasi-segregation of pupils, with white children and most other local-born children engaging in distinct worlds of social interaction, defined by one’s competence of ASE and non-use of HCE (Sato 1985, Tamura 1996). There were other social meanings linked to ASE and HCE. Throughout the 1930s and possibly as early as the 1920s, local-born male speakers of ASE were often teased as “sissies”. This suggests that HCE had become a resource for performing masculine identity. The writer of a letter to the editor commented in the 1940s: Kamehameha [School] boys speak pidgin English on the campus and in their dormitories. To speak correct English is “sissy,” they say. (Honolulu Advertiser, 9 May 1943, p. 24)

It is unclear whether gender differences existed in the degree to which HCE diverged from ASE in in-group discourse, but it may have been the case that localborn women made a greater use of ASE-linked variables than men. The following comment by a local-born girl during World War II illustrates how ASE and HCE were relevant to gender identity: If we girls speak good English ourselves and show disapproval of pidgin English, we will be exerting social pressure on the boys. I’m sure that this pressure will overcome the stupid pressure of the unruly boys who insist that speakers of good English are sissies. Is there any boy here who is so tough that he would dare to call General MacArthur who speaks very good English a sissy? If General MacArthur is a sissy, I’m sure we all would want to be sissies. (Miss Flora Okura, Grade 9, Waimea High and Elementary School, Hawaii Educational Review, May 1943: 268)

What makes this remark particularly interesting is that it reveals an intersection between ethnic and gender identity — Okura’s argument erases General MacArthur’s ethnic and out-group status, and ignores the equally prevalent linkage between ASE and Haole identity in local-born communities. The local-born population in the 1910s and afterward faced several different linguistic marketplaces with varying norms on language use: the ethnic-centered

343

344 Sarah J. Roberts

domain of home and ethnic institutions, the pan-ethnic and accommodating context of peer-centered interaction in peer groups and at school, and the institutionally based Americanizing context of the school classroom and white culture. Individual speakers would have engaged in multiple communities of practice, and controlled a stylistic range of lects adapted to the interactional norms of each situation. Styles convergent with ASE were useful in accommodating to the speech of whites in school and elsewhere, but they were also a resource for performing Haole identity within local-born communities. Portuguese children in particular occupied a contested position in conventional identity categories in Territorial Hawai’i: Haoles were otherwise classified as whites, Portuguese were considered white, but Portuguese were often not considered to be Haole — a term often reserved for those of Anglo-Saxon ancestry (cf. M-23, M-34, M-50, MK-31 and others). The use of styles approximating ASE in peer-centered contexts would have constituted one means of laying claim to this contested identity, as well as affecting a “dis-identification” with other non-white local-born speakers. The use of styles divergent from ASE, on the other hand, would have also allowed other Haole individuals to interact with non-white locally born on more of an equal footing.⁹ The foregoing indicates that the basilect did not develop in an evaluative vacuum. Although language-internal factors were doubtless responsible for most of the structural developments in HCE, the directionality in which the basilect diverged from ASE and WVEs reflected language ideologies which differentiated local-born speakers from foreign-born immigrants and Haoles, and enabled an indexing of ALs/HPE with non-American identity, and ASE with Haole identity. Although a convergent, unified local-born speech community was then under formation, an identity identical to that of the present-day “Local” was not yet articulated (see also Okamura 1980).

5. Conclusion In summary, the process of differentiation that produced the multidimensional continua of HCE was fundamentally stylistic — as a pidgin originally used in a utilitarian, limited domain became employed in an increasing array of new social situations. Although individual intraspeaker variation is not commonly attested in early sources, the role of style is implicated in the patterns of divergence and convergence. Styles convergent ideologically with ASE were necessary in formal and institutional contexts such as school and in white-collar employment for some, while styles convergent with HPE were needed for discourse with immigrants and HPE-speakers.¹⁰ The development of the basilect in directions away from both ASE and the HPE of immigrants provided greater resources for performing identities that were distinguished both from ancestral culture and Haole culture. The eventual result was a language that was ideologically bounded and linguistically diffuse, intimately associated with modern “local” culture in Hawai’i.


Notes * This paper was presented at the Society for Pidgin and Creole Linguistics Annual Meeting in San Francisco, California, in January 2002. Previous drafts of this paper were previewed and commented on by Mary Bucholtz, John Rickford, Jeff Siegel, Elizabeth Traugott, and two anonymous referees. I would also like to thank the Style, Language, and Ideology Collaborative at Stanford for discussions on this material. Any shortcomings are my own. 1. Mufwene defines basilectualization as “the process by which a language variety diverges structurally toward a pole at an extreme from that of its lexifier” (2001: 209). According to Mufwene, basilectualization, as a contributing process to language speciation, is not limited to creoles but should be regarded as one general consequence of language contact. 2. The form wen is commonly regarded as a lenited variant of been (Bickerton 1980: 7, Carr 1972: 156, Holm 1989: 525, Roberts 1998: 18, 35),“the combination of low stress and the more rapid speech of the native speaker leads very naturally to weakening of this kind” (Bickerton 1977: 12). Similar developments occurred in Caribbean Creole English, where the form wen is found alongside Jamaican CE ben, en and Providencia CE men (Cassidy 1961: 40, Hancock 1987: 292, 300, Holm 1989: 407). However, not all have regarded wen as merely a weakened variant of been. Sato believed wen developed independent of been, possibly derived partially from English went (Romaine 1988: 54). The form went with a final obstruent has been attested in HCE speech, but Mühlhäusler regards it as a “borrowing from the superimposed English” which “illustrates the fact that Creoles exhibit a stylistic distinction” (1997: 71) between the basilect and acrolect. The data presented in this paper suggest that wen emerged fairly abruptly by the late 1920s, if not somewhat earlier, which is appropriate for the speed of a lexical innovation. While English went presumably exerted influence during decreolization, the superstratal form may have also played a role in the earlier development of wen from been. 3. In theory the changes in HCE between 1900 and 1930 could result from the influence of 20th-century Japanese and Filipino adult immigrants, but this is almost certainly not the case. Although Bickerton found that such immigrants were “heavily influenced by nativelanguage grammar” (1984: 211), the features they used were not typical of HCE. As noted in this paper, SOV word order is one Japanese-influenced feature that was very common in 20th-century HPE but nearly always absent in HCE. Kaapu (1937) found very little evidence of Japanese influence in HCE. Reinecke also noted that “the Hawaiians, Chinese, and Portuguese between them had pretty well fixed the form of the ‘pidgin’ spoken on plantations” while the Japanese, Filipinos, Koreans, and other later immigrants “added little to it” (1969: 93–4). Siegel (2000) similarly found considerable influence from Chinese and Portuguese in HCE, particularly in the case of the finite complementizer for, progressive stay, and auxiliary combinations — features identified in this paper as divergent from earlier HPE. It is probably the case that HCE was more divergent from the speech of Japanese and Filipino immigrants than of earlier Chinese and Portuguese arrivals. Bickerton’s comparison of HCE with the HPE of 20th-century Japanese and Filipinos probably made the former seem more like an ex nihilo creation than it really was. 4. Mufwene asserts that “creoles could not have possibly been invented by children” (2001: 131). Others have also expressed doubt that children play an important role in creole formation. Singler (1992, 1995: 220–1) notes that in several instances children could not have had a significant effect on the stabilization and expansion of creole grammar. Jourdan and Keesing (1997: 404) show that native speakers did not play a significant role in the

345

346 Sarah J. Roberts grammatical expansion of Solomons Pijin. And Samarin (1997: 186–99) doubts that the primary “agents of change” in the case of Sango and Tok Pisin were children. Parkvall suspects that “the number of children itself does not directly affect the degree of restructuring more than marginally” (2000: 197), and Baker similarly feels that the nativization of pidgins and creoles by children “had a negligible effect on speed of development” (1995: 17), though he acknowledges HCE as a possible exception to this pattern. 5. Divergence describes a situation involving two diverging varieties that were closer together at one time. At the time HCE first emerged, HPE was already highly divergent from WVEs and ASE. This divergence thus was a process that began with the formation of HPE and continued in the creole. The emergence of tense/aspect auxiliary combinations and finite for-complementation, for instance, added to the distance between basilectal HCE and WVEs. However, as access to WVEs increased over time, the creole also underwent convergence with more acrolectal registers. 6. This view does not at all exclude the possibility that decreolization occurs as a trajectory of change in the modern variety of HCE. There are some indications that HCE has been losing basilectal features over time (Romaine 1988: 294–5, Sato 1991). 7. These life histories were collected in 1926 and 1927 by sociologist William C. Smith who distributed a research questionnaire in a number of junior high and high schools throughout Hawai’i (Roberts 2000: 276–7). Approximately 1,550 life histories are reproduced in 17 rolls of microfilm available through the University of Oregon Library. The numbering system used in this paper was Smith’s own. HJ-88 for instance refers first to the school (Honoka’a Junior High) and second to the student number. 8. Or the use of a style of HCE which veered too close to ASE, as judged through linguistic evaluation. 9. Sometimes this approach backfired. Reinecke, discussing attempts by Haoles “to attain intimacy with those who naturally speak the dialect” noted that if done “naturally and with discretion, it is approved by the users of dialect” (1969: 180), but if done without tact such attempts can be quite condescending to creole speakers. 10. Local-born informants interviewed by William Wilson described their use of HPE with immigrant speakers in their communities. The following exchange between Wilson and Mary K. Puku’i is illustrative: “Q. So if you went to one of these fish sellers and you said, O, I want that fish. How would you say that? A. Mi dis makemake. If it was Japanese [I would say] mi dis makemake. Q. You change if you’re talking to Japanese? A. Yes. Mi dis makemake. Dis no raiki, mi raik dat, dat wan.” (Wilson 1976: 95–6). This example includes the use of SOV word order (a feature from the Japanese substrate) and the subject pronoun mi, both features atypical of HCE. Local-born speakers thus accommodated to the pidgin of the foreign born according to the situation.

References Alleyne, M. C. 1971. “Acculturation and the cultural matrix of creolization.” In Pidginization and Creolization of Languages, Dell Hymes (ed.), 169–86. Cambridge: Cambridge University Press. Anonymous. 1932a. “Confessions of Joe Manuel of the Raddio Patrol.” Weekly column in the Honolulu Star-Bulletin, October 8–December 31, 1932.

The role of style and identity in the development of Hawaiian Creole Anonymous. 1932b. “My Pal Groucho.” Weekly column in Ka Punahou, September 20–December 15, 1932. Anonymous. 1946. “Hukilepo Joe Says.” Weekly column in Hilo Tribune-Herald, January 5– December 29, 1946. Arends, J. (ed.). 1995. The Early Stages of Creolization. Amsterdam: John Benjamins. Auer, P. and A. di Luzio (eds.). 1988. Variation and Convergence: Studies in Social Dialectology. Berlin: Mouton de Gruyter. Bailey, G. and N. Maynor. 1989. “The divergence controversy.” American Speech 64: 12–39. Baker, P. 1995. “Some developmental inferences from the historical studies of pidgins and creoles.” In Arends, 1–24. Barrett, R. 2001. “Is queer theory important for sociolinguistic theory?” In Campbell-Kibler Podesva, Roberts and Wong (eds.), 25–43. Bell, A. 1997. “Language style as audience design.” In Sociolinguistics: A Reader and Coursebook, N. Coupland and A. Jaworski (eds.), 240–50. New York: St. Martin’s Press. Bickerton, D. 1975. Dynamics of a Creole System. Cambridge: Cambridge University Press. Bickerton, D. 1977. Change and Variation in Hawaiian English, Vol. 2: Creole Syntax. Honolulu: University of Hawai’i: Social Science and Research Institute. Bickerton, D. 1980. “Creolization, linguistic universals, natural semantax, and the brain.” In Issues in English Creoles: Papers from the 1975 Hawaii Conference, R. R. Day (ed.), 1–18. Heidelberg: Groos. Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma. Bickerton, D. 1984. “The language bioprogram hypothesis.” The Behavioral and Brain Sciences 7: 173–221. Bond, K. L. 1936.“We’ll go see the world.” College Plays, Vol. 1. Honolulu: English Department, University of Hawai’i. Butler, J. 1990. Gender Trouble: Feminism and the Subversion of Identity. New York: Routledge. Cameron, D. 1998. “Performing gender identity: Young men’s talk and the construction of heterosexual masculinity.” In Language and Gender: A Reader, J. Coates (ed.), 270–84. Oxford: Blackwell. Campbell-Kibler, K., R. J. Podesva, S. J. Roberts and A. Wong (eds.). 2001. Language and Sexuality: Contesting Meaning in Theory and Practice. Stanford: CSLI Publications. Carr, E. B. 1972. Da Kine Talk: From Pidgin to Standard English in Hawaii. Honolulu: University of Hawai‘i Press. Carrington, L. D. 1992. “Images of creole space.” Journal of Pidgin and Creole Languages 7: 93–9. Cassidy, F. G. 1961. Jamaica Talk: Three Hundred Years of the English Language in Jamaica. London: Macmillan. Chaudenson, R. 1992. Des îles, des hommes, des langues: Essais sur la créolisation linguistique et culturelle. Paris: L’Hartmann. Chun, W. C. 1936. “For you a lei.” College plays, vol. 1. Honolulu: English Department, University of Hawai‘i. DeCamp, David. 1971. “Toward a generative analysis of a post-creole speech continuum.” In Pidginization and Creolization of Languages, D. Hymes (ed.), 349–70. Cambridge: Cambridge University Press. Eckert, P. 2000. Linguistic Variation as Social Practice: The Linguistic Construction of Identity in Belten High. Malden, MA: Blackwell.

347

348

Sarah J. Roberts Eckert, P. and S. McConell-Ginet. 1992. “Think practically, look locally: Language and gender as a community-based practice.” Annual Review of Anthropology 21: 461–90. Esch, E. and M. Jones (eds.). 2001. Patterns of Linguistic Convergence and Divergence. Berlin: Mouton. Fasold, R.W. 1981. “The relationship between black and white speech in the South.” American Speech 56: 163–89. Gal, S. and J. T. Irvine. 1995. “The boundaries of languages and disciplines: How ideologies construct difference.” Social Research 62: 967–1001. Giles, H. and N. Coupland 1991. Language: Contexts and Consequences. Keynes: Open University Press. Hall, R. A. 1966. Pidgin and Creole Languages. Ithaca: Cornell University Press. Hancock, I. 1987. “A preliminary classification of the Anglophone Atlantic creoles, with syntactic data from thirty-three representative dialects.” In Pidgin and Creole Languages: Essays in Memory of John E. Reinecke, G. G. Gilbert (ed.), 264–333. Honolulu: University of Hawai‘i Press. Holm, J. 1989. Pidgins and Creoles (Vol. 2). Cambridge: Cambridge University Press. Hormann, B. L. 1960. “Hawaii’s linguistic situation: A sociological interpretation in the new key.” Social Process in Hawaii 24: 6–31. Irvine, J. 2001. “‘ Style’ as distinctiveness: The culture and ideology of linguistic differentiation.” In Style and Sociolinguistic Variation, P. Eckert and J. Rickford (eds.), 21–43. Cambridge: Cambridge University Press. Jourdan, C. and R. Keesing 1997.“From Fisin to Pijin: Creolization in process in the Solomon Islands.” Language in Society 26: 401–18. Kaapu, M. K. 1937. A Study of the Influence of Japanese Syntax and Idiom upon the Spoken and Written English of a Group of Ninth Grade Pupils. Honolulu: University of Hawai‘i M. A. Thesis. Kouwenberg, S. 1992.“From OV to VO: Linguistic negotiation in the development of Berbice Dutch Creole.” Lingua 88: 263–99. Kubo, L. M. 1997. Reading and Writing Local: A Politics of Community. PhD Thesis. Honolulu: University of Hawai’i. Labov, W. 1963. “The social motivation of sound change.” Word 19: 273–309. Leong, C. S. 1997. “You Local or What?” An Exploration of Identity in Hawaii. PhD Thesis. Cincinnati: Graduate School of the Union Institute. Le Page, R. B. and A. Tabouret-Keller. 1985. Acts of Identity: Creole-based Approaches to Language and Identity. Cambridge: Cambridge University Press. Lippi-Green, R. 1997. English with an Accent. London: Routledge. Masuda, B. 1993. The Divided Local: A Personal Foray into the Problematics of Hawai‘i Poetic Identity. Senior Honors Thesis. Honolulu: University of Hawai‘i. Meyerhoff, M. and N. Niedzielski. 1994. “Resistance to creolization: An interpersonal and intergroup account.” Language and Communication 14: 313–30. Mufwene, S. 1991. “Is Gullah decreolizing? A comparison of a speech sample of the 1930’s with a speech sample of the 1980’s.” In The Emergence of Black English: Text and Commentary, G. Bailey, N. Maynor and P. Cukor-Avila (eds.), 213–30. Amsterdam: John Benjamins. Mufwene, S. 1994. “On decreolization: The case of Gullah.” In Language and the Social Construction of Identity in Creole Situations, M. Morgan (ed.), 63–99. Los Angeles: Center for Afro-American Studies.

The role of style and identity in the development of Hawaiian Creole 349 Mufwene, S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Mühlhäusler, P. 1980. “Structural expansion and the process of creolization.” In Theoretical Orientations in Creole Studies, A. Valdman and A. Highfield (eds.), 19–55. New York: Academic Press. Mühlhäusler, P. 1997. Pidgin and Creole Linguistics. London: University of Westminster Press. Neumann-Holzschuh, I. and E.W. Schneider (eds.). 2000. Degrees of Restructuring in Creole Languages. Amsterdam: John Benjamins. Ochs, E. 1991. “Indexing gender.” In Rethinking Context, A. Duranti and C. Goodwin (eds.), 335–58. Cambridge: Cambridge University Press. Okamura, J.Y. 1980. “Aloha kanaka me ke aloha ‘aina: Local culture and society in Hawaii.” Amerasia 7: 119–37. Parkvall, M. 2000. “Reassessing the role of demographics in language restructuring.” In Neumann-Holzschuh and Schneider, 185–213. Podesva, R., S. J. Roberts and K. Campbell-Kibler. 2001. “Sharing resources and indexing meanings in the production of gay styles.” In Campbell-Kibler, Podesva, Roberts and Wong (eds.), 175–89. Reinecke, J. 1969. Language and Dialect in Hawai’i: A Sociolinguistic History to 1935. Honolulu: University of Hawai‘i Press. Rickford, J. 1987. Dimensions of a Creole Continuum: History, Text, and Linguistic Analysis of Guyanese Creole. Stanford: Stanford University Press. Rickford, J. 1992. “Grammatical variation and divergence in Vernacular Black English.” In Internal and External Factors in Syntactic Change, M. Gerritsen, and D. Stein (eds.), 175– 200. Berlin: Mouton. Roberts, S. J. 1998. “The genesis of Hawaiian creole and diffusion.” Language 74: 1–39. Roberts, S. J. 1999. “The TMA system of Hawaiian Creole and diffusion.” In Creole Genesis, Attitudes, and Discourse: Studies Celebrating Charlene J. Sato, J. Rickford and S. Romaine (eds.), 45–70. Amsterdam: Benjamins. Roberts, S. J. 2000. “Nativization and the genesis of Hawaiian Creole.” In Current Issues in Creole Studies, J. McWhorter (ed.), 259–302. Amsterdam: John Benjamins. Romaine, S. 1988. Pidgin and Creole Languages. London: Longman. Samarin, W. J. 1997. “The creolization of pidgin morphophonology.” In The Structure and Status of Pidgins and Creoles, A. K. Spears and D. Winford (eds.), 175–216. Amsterdam: John Benjamins. Sato, C. J. 1985. “Linguistic inequality in Hawaii: The post-creole dilemma.” In Language of Inequality, N. Wolfson and J. Manes, 256–72. Berlin: Mouton. Sato, C. J. 1991. “Language change in a creole medium: Decreolization?” University of Hawai‘i Working Papers in ESL 10: 127–47. Siegel, J. 1997. “Mixing, leveling, and pidgin/creole development.” In The Structure and Status of Pidgins and Creoles, A. K. Spears and D. Winford (eds.), 111–49. Amsterdam: John Benjamins. Siegel, J. 2000. “Substrate influence in Hawai‘i Creole English.” Language in Society 29 (2): 197–236. Singler, J.V. 1992. “Nativization and pidgin/creole genesis: A reply to Bickerton.” Journal of Pidgin and Creole Languages 7: 319–33.

350

Sarah J. Roberts Singler, J.V. 1995. “The demographics of creole genesis in the Caribbean: A comparison of Martinique and Haiti.” In The Early Stages of Creolization, J. Arends (ed.), 203–32. Street, R. L. and H. Giles. 1982.“Speech Accommodation Theory: A social cognitive approach to language and speech behaviour.” In Social Cognition and Communication, M. E. Roloff and C. R. Berger (eds.), 193–226. Beverley Hills, CA: Sage. Tamura, E. H. 1996. “Power, status, and Hawai’i Creole English: An example of linguistic intolerance in American history.” Pacific Historical Review 65: 431–54. Thomason, S. G. and T. Kaufman 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Valdman, A. 1977. “Creolization: Elaboration in the development of Creole French dialects.” In Pidgin and Creole Linguistics, A. Valdman (ed.), 155–89. Bloomington: Indiana University Press. Wilson, W. 1976. Unpublished volume of fieldwork on Pidgin Hawaiian. Winford, D. 2000.“‘ Intermediate’ creoles and degrees of change in creole formation: The case of Bajan.” In Neumann-Holzschuh and Schneider (eds.), 215–46. Yamamoto, E. 1979. “The significance of the local.” Social Process in Hawaii 27: 101–15.

Index AAVE (African American Vernacular English) 67–8, 76, 82, 84, 332 Abrahams 286 acrolect 147, 161–3, 184, 187, 230, 325, 345 Adam 43–5, 57, 63 address forms 293, 299 Adone 189, 206, 312, 326 Adone 189, 206, 207, 312, 326, 327 adstrate 1, 2, 8, 12, 14–16 adverb 12, 13, 26, 137, 141 adverbs (reduplicated) 12, 14, 16 African American Vernacular English 67, 68 Afrikaans 193 Afro-Caribbean 39, 259–61, 265, 270 agreement 9, 28, 81, 82, 97, 98, 102–5, 110, 116, 117, 119, 120, 122, 124, 140–4, 150, 151, 155, 190, 204, 206, 295 Aikhenvald 35, 39, 43–6, 62 Alleyne 72, 223, 235, 337, 346 Aluku 287, 304 Amazonian languages 44, 45 Anansi 267, 276–8, 303, 305 Andersen 132, 133, 211, 217, 218 Angola 98, 101, 102, 121, 125, 126 Ansaldo 1, 5, 6 antecedent 314, 322 anticausativity 307, 322 Arawak 35, 36, 39, 43–53, 60, 62, 63 Arends 21, 22, 27–31, 33,128, 130, 304 aspect 1, 21, 51, 55, 56, 60, 134, 139, 140, 142, 143, 148–52, 156, 157, 161, 213–15, 217, 238, 240–2, 250, 251, 271, 322, 333–5, 337, 346 Austronesian languages 10 auxiliary 26–8, 56, 59, 82, 141, 142, 151, 155, 212, 217, 345, 346 Bailey 75, 85, 332 Baker 28, 68, 84, 155, 205, 346 Bakker 28, 130–2 Baptista 14, 17, 120 basilect 147, 161–3, 166, 169, 170, 187, 331, 332, 337, 344, 345 Batalha 4, 6, 8, 9, 17 Baxter 1, 8, 97, 102, 120 Bay Islands of Honduras 37, 61 Belize 35–9, 41, 42, 44, 46, 53–9, 61–3 Bell 332, 337 Berbice Dutch 193 Bickerton 28, 32, 105, 131, 193, 212, 218, 221–3, 238–42, 245, 246, 250–1, 331, 332, 345, 349

Black loyalists 70, 71, 74 Boni 287 Book of Negroes 70, 72–4, 76, 85, 86 Breton 37, 43, 45, 57, 62 Bruyn 27, 130, 191, 192, 198, 205, 207 calypso lyrics 259, 260, 266–8, 278–80 Cameroonian Pidgin English 223 Cantonese 1–4, 8, 9, 13–18, 144 Cape Verde 6, 14, 98, 101, 102, 121 Carden 307–11, 314–16, 322–5, 327 Carib 35–8, 43–9, 51, 61–3, 73 Carnival 262–7, 272, 273, 281, 283 Carrington 237, 238, 240, 241, 246, 250–1, 331 case 11, 15, 16, 18, 28, 29, 32, 35, 47, 53, 55, 56, 59, 63, 72, 79, 85, 87, 100–2, 111, 119, 122, 124, 126, 128, 132, 134, 137, 142, 145, 150, 153, 158, 170, 172, 174, 177, 178, 187, 189– 92, 202, 204–6, 212, 214, 215, 222, 223, 238, 239, 242–6, 249–51, 274, 280, 287, 289–92, 308, 309, 314, 322, 332, 335, 338, 343, 345, 346 Cassidy 84, 144, 345 Cayman Islands 37, 39 Chaudenson 6, 18, 196, 207, 332, 347 Clarkson 67, 68, 70–2, 77–8 Clements 5, 7, 10–12, 15, 17 code mixing 293 (see also mixing) Coelho 37, 40 Comrie 200, 241, 244, 248, 251 conclusive passive voice 212 consonant clusters 48, 163, 171–5, 181, 184–7 copula 9, 21, 24, 27, 28, 30–2, 141, 151, 152, 156, 217, 241, 242, 251, 252, 286, 333 Corne 155, 193, 194, 307, 309–12, 317, 319, 323, 324, 326 Costa Rica 259–61, 263–70, 275, 279–81 court records 21–4, 28–30, 32 Couto 6 creole genesis 127, 128, 134, 136, 153, 331 creolization 9, 35, 36, 60, 61, 75–8, 83, 85, 130–2, 154, 198, 205, 308, 309, 325 Crystal 43 CV production 169 DeCamp 147, 331 definiteness 124, 201 depidginization 105 Derbyshire 45 descriptive clause 242, 243

352

Index diaspora 35, 67, 69, 83 discourse 44, 49, 52, 53, 128, 138, 139, 141, 144, 159, 239, 248, 249, 251, 286, 304, 307, 316, 322, 326, 333, 334, 338–40, 342–4 Dixon 44, 45, 62–3 Dorian 55, 60, 153 dos Santos Ferreira 5, 7, 11–15 Dressler 60, 213 Dutch 5, 22, 28, 33, 139, 154, 161, 163, 164, 168, 169, 171, 175, 177, 185, 191, 193, 196, 209, 210, 212–16, 218, 287, 299, 310 Eastern Maroon Creole 287, 298 Eckert 286, 338, 339 Eckkrammer 209, 211, 213 endangered language 35, 44 English 3, 22, 27, 28, 30, 33–5, 39–42, 44, 46–8, 53–5, 58, 61, 62, 67, 68, 70, 71, 77, 78, 81–5, 127, 129, 130, 132, 136–45, 147–50, 152–8, 161–4, 166, 168, 169, 171, 175, 185, 186, 189–91, 193, 194, 196–201, 204–6, 212, 216, 221–35, 238, 239, 249, 260–2, 265, 267, 268, 279–81, 287, 298, 306, 308, 310, 311, 326, 331–3, 335, 337, 339, 341–3, 345 English-based creole 125, 129, 130, 136, 140, 141, 145, 149, 154, 156, 221, 265, 280 equative clause 241, 242 Escure 17, 35, 38, 39, 42, 56, 58, 250, 304 ethnic identity 152, 278, 339 evidential particles 45, 54 faithfulness constraints 164, 165, 177 Fasold 332 figures of speech 291, 293, 298–300, 305 Frank 237, 249, 251 French 35–7, 40, 43, 45–54, 61–3, 98, 138, 139, 154, 189, 192, 196, 198, 201, 206, 212, 223, 237, 248, 264, 265, 286–8, 299, 307–9, 311, 312, 318, 320, 322, 325 functional domain 307, 322, 325 future tense 56, 57, 217, 240, 241, 336 Garifuna 35–50, 53–63 Garinagu 37, 39–42, 44, 53, 54, 56, 57, 61, 62 gender 9, 28, 45, 51, 52, 55, 61, 102, 109, 120, 122, 290, 299, 338, 341, 343 Goa 1–3, 5 grammatical gender 45, 61 grammaticalization 28, 36, 49, 59, 309, 319, 322, 326 Guadeloupean (Creole) 248, 310–13, 315–20, 326 Guatemala 35–8, 41, 42, 61 Guinea-Bissau Kryol 193 Gullah 67–9, 71, 74, 75, 77, 78, 83–5 Guyanais 193 Guyanese (Creole) 223, 231, 238, 245, 322

Haitian Creole 154, 193, 223, 238, 240, 310–12, 314, 327 Hall 71, 222, 223, 235, 310, 311, 331 Hancock 5, 9, 71, 84, 345 Herzfeld 61, 259, 262, 271, 280 Hokkien 1, 2, 5, 8, 9, 17 Holm 5–10, 12, 130, 131, 136, 138, 141, 154, 345 Honduras 35–41, 43, 44, 53–9, 61 Huber 67, 68, 70, 85 Hymes 285 identity 31, 36, 40, 71, 72, 79, 144, 152, 154, 259– 62, 265, 267, 268, 270, 271, 273, 278–80, 331, 332, 337–44 identity marker 259 indentured labor 98, 119 interlanguage 102, 106, 111, 118, 127, 129, 132, 133, 137, 139, 153, 154 interspeaker variation 331, 337, 339 irrealis 54–6, 149, 240 Irvine 293, 303, 332, 338, 342 Ivens Ferraz 6, 7, 12 Kaufman 128, 130, 131, 332 Kikongo 8, 47, 150, 153, 288, 304 Kouwenberg 212, 219, 332 Krio 67–71, 78, 84, 221–35 kuutu 285–93, 295, 297–301, 303, 304 Labov 285, 306, 332, 348 language attrition 36, 47, 55, 60 Lenz 210, 212, 217 lexical tone 222 lexicalization 307 Limonese Creole 259–61, 267, 280, 281 Limonese English 261, 263 linguistic convergence 348 linguistic divergence 332, 333, 337–9 linguistic ideology 331, 339, 342 Lipski 129, 156 Loanwords 42, 46–8, 50, 53, 54, 61, 62 Lorenzino 120, 121 Louisiana Creole 193, 309–11, 323, 324 Loyalists 70, 71, 73, 74 Macanese 1–11, 13–19 Macao 1–8, 10, 17–18, 102 Maduro 215, 216, 218 Maipuran 35, 43 Malacca 1–8, 10, 17 Malayo-Portuguese 1, 4–9 maroons 23, 37, 48, 70, 71, 84, 156, 285, 303 Matthews 1, 5, 6, 13, 15, 18, 155 Maurer 209, 212 Meade 161, 186–7 mekaytelyuw 261, 280 mesolect 162, 163, 188, 230

Index Mestizo 40, 42, 259 Migge 285, 287, 300, 304 Milroy 285 mixing 6, 118, 293 modality 55, 149, 238 Montgomery 69, 77, 79–84 Mosquito Shore 37 Mozambique 98, 101, 102 Mufwene 84, 129, 307, 308, 328, 331, 332, 335, 338, 345, 349 Muller 209, 216, multilingualism 37, 128, 216, 259, 339 musical genre 259, 260, 264 Mutz 307 Muysken 128, 131, 191, 192, 198, 205, 207, 236, 307, 309, 317, 319, 322 Myers-Scotton 118, 124 Naro 119 national identity 259 Ndyuka 287, 304 negation 26, 102, 127, 137, 141 Negerhollands 191–3, 205, 309 Nicaragua 37–40 Noble 45 nominalizer 55 nonpunctual 240, 242, 246, 251 nonstative 237–40, 243–7, 249–51 Nova Scotia 67, 70–4, 78 Nova Scotians 67, 69, 71, 72, 75–9, 82–4 NP accessibility hierarchy 200, 201 number agreement 102, 104, 122, 151 number (and person prefixes) 61 number (distinction) 9, 45, 102, 109, 143, 151 number (French loanwords) 50, 63 number (sufffixing) 61 numeral 15, 16, 50, 63 Okanisi 287 onsetless syllable 163 Palenquero 127, 130, 135, 150–3, 156, 193, 217 Pamaka 285, 287, 288, 291, 301, 304 Panama 37, 38, 263, 265 Papia Kristang 5, 6, 9 passive voice 156, 209–13, 215–17 past tense 29, 35, 56, 58, 156, 162, 240, 333, 336 Patrick 142, 147, 148, 158, 163 perfective 5, 51, 52, 120, 147, 333 Perl 21, 22 pidgin 5–7, 10, 15, 1722, 28, 61, 68, 84, 129–33, 136, 138, 142, 144, 149, 150, 153, 154, 223, 236, 331–3, 339, 343–6 pidginization 36, 43, 59–61, 77, 82, 83, 132, 133, 332 pitch-accent languages (or creoles) 221–5, 229, 235

Portuguese 1–10, 12, 13, 15–18, 97–106, 108–11, 114, 117–24, 138, 142, 144, 192, 193, 287, 335, 336, 339, 340, 344, 345 Portuguese(-based) creole 1, 2, 4, 5, 7, 120 possessive 9, 45, 46, 51, 52, 141, 307, 309, 311, 316, 317, 319–23, 326, 336 pragmatic function 28 Príncipe 125 Principense 193 pro-drop 26, 138, 151 processability hierarchy 127, 129 progressive 151, 156, 212, 238, 239, 244, 247, 248, 333, 336, 337, 345 punctual 251 question formation 127, 136, 137, 141, 142, 152 reduplication 1, 2, 6–8, 10–16, 32 reflexive construction 307, 308, 310–14, 316–18, 322, 324 reflexivity 307–13, 315, 316, 319, 322, 324, 325 reggae 260, 264–6, 268, 278, 279 register 133, 136, 137, 140, 143, 144, 147, 211, 212, 216–18, 305, 307, 316, 346 Reinecke 87, 332, 333, 335, 342, 345, 346 Rickford 84, 286, 331, 332, 345 Roberts 22, 155, 156, 331–3, 335, 338–40, 345, 346 Romaine 131, 209, 345, 346 saliency 97, 103, 105, 106, 108, 110, 111, 120 Sankoff 122 Santos Ferreira 5, 7, 11, 1–15 Schmidt 55 Schneider 34, 75, 84, 222, 223 Schumann 21, 22, 27, 29–31, 132, 133 Schwegler 8, 61, 135, 150, 152, 154, 156, 250, 304 Seselwa (Seychellois) 309 Seuren 29, 160, 193, 221, 236, 242, 251 Sidnell 286, 304 Siegel 128, 145, 154, 332, 345 Sierra Leone 67–75, 77–80, 83–5, 98, 221, 224–6, 234 Singler 61, 120, 130, 160, 252, 345 Smith 5, 22, 27, 128, 131, 165, 261, 287, 307, 309, 322 soca 265 Spanish (language) 35–7, 39–42, 46–8, 52–5, 60, 62, 127, 129, 135, 138, 142, 143, 150–3, 212– 16, 259–62, 267–72 Spanish (lexicon) creole 127, 130, 150–3 Spears 239, 240, 248, 249 speech event 285, 286 Sri Lanka Portuguese 193 stative 13, 237–51 stress 62, 103, 106–8, 110, 122, 151, 215, 221–5, 228–36, 345

353

354

Index style (language) 78, 215, 273, 285, 286, 292, 293, 296, 331, 332, 337, 338, 344–6 style (musical) 264–6, 279 substrate 1, 2, 6, 8, 10, 14, 27, 28, 35, 43, 119, 128, 129, 131–3, 136, 142, 150, 152, 154, 155, 221, 309 superstrate 43, 110, 129, 131, 132, 134–6, 145, 150, 155, 212, 286, 287, 309 Surinam 30, 68 syllabic consonants 171–5, 180, 182, 185 syllable 5, 10, 62, 106–11, 118, 151, 161–4, 168– 77, 180, 182, 185–8, 215, 222–5, 227–33, 235 target language 127–9, 131, 135–8, 153 Taylor 35, 37, 43, 46, 50–3, 56, 57, 62 tense 9, 19, 21, 29, 35, 55–8, 139–43, 145, 147–53, 155–6, 162, 163, 213, 217, 237, 238, 240, 241, 245, 249, 250, 333–7, 346 Thomason 128, 130, 131, 160, 332, 350 tico 259 TMA 8, 9, 24, 28, 45, 55, 61, 125, 126, 139, 140, 237–50 Tok Pisin 130, 132, 136, 144, 148, 149, 191–3, 346

tone 18, 80, 217, 218, 221–35, 342 tone harmony 222 tone marking 221, 224, 225, 228, 229 Tonga 97–102, 106, 108, 113, 114, 116–24 Tonga Portuguese 97, 99, 100, 102, 114, 117–20, 122–4 Trinidad 252, 261, 263–5, 268, 280 turn-taking 285, 293, 297 Umbundu 97, 101, 102, 106, 109–11, 118, 119, 121, 123, 124 Valdman 132, 331, 332 van den Berg 21, 22, 29 VARBRUL 97, 103–6, 110, 114–17, 122, 123 well-formedness constraints 164, 165 Winer 264, 265 Winford 27, 34, 75, 77, 82, 84, 239, 240, 249, 331 Wolfram 106, 156 Wunderlich 203, 206 Wurm 145 Yip 13, 15, 18, 155

In the series Creole Language Library the following titles have been published thus far or are scheduled for publication: 1 2 3 4 5 6 7

8 9 10 11 12

13 14 15 16 17 18 19

20 21 22 23 24 25 26 27

MUYSKEN, Pieter and Norval SMITH (eds.): Substrata versus Universals in Creole Genesis. Papers from the Amsterdam Creole Workshop, April 1985. 1986. vii, 311 pp. SEBBA, Mark: The Syntax of Serial Verbs. An investigation into serialisation in Sranan and other languages. 1987. xv, 218 pp. BYRNE, Francis: Grammatical Relations in a Radical Creole. Verb Complementation in Saramaccan. With a foreword by Derek Bickerton. 1987. xiv, 293 pp. LIPSKI, John M.: The Speech of the Negros Congos in Panama. 1989. vii, 159 pp. JACKSON, Kenneth David: Sing Without Shame. Oral traditions in Indo-Portuguese Creole verse. 1990. xxiv, 257 pp. SINGLER, John Victor (ed.): Pidgin and Creole Tense/Mood/Aspect Systems. 1990. xvi, 240 pp. FABIAN, Johannes (ed.): History from Below. The “Vocabulary of Elisabethville” by André Yav: Text, Translations and Interpretive Essay. With the assistance of Kalundi Mango. With linguistic notes by W. Schicho. 1990. vii, 236 pp. BAILEY, Guy, Natalie MAYNOR and Patricia CUKOR-AVILA (eds.): The Emergence of Black English. Text and commentary. 1991. x, 352 pp. BYRNE, Francis and Thom HUEBNER (eds.): Development and Structures of Creole Languages. Essays in honor of Derek Bickerton. 1991. x, 222 pp. WINFORD, Donald: Predication in Caribbean English Creoles. 1993. viii, 419 pp. BYRNE, Francis and John HOLM (eds.): Atlantic Meets Pacific. A global view of pidginization and creolization. 1992. ix, 465 pp. BYRNE, Francis and Donald WINFORD (eds.): Focus and Grammatical Relations in Creole Languages. Papers from the University of Chicago Conference on Focus and Grammatical Relations in Creole Languages. 1993. xvi, 329 pp. ARENDS, Jacques (ed.): The Early Stages of Creolization. 1996. xvi, 297 pp. KIHM, Alain: Kriyol Syntax. The Portuguese-based Creole language of Guinea-Bissau. 1994. xii, 310 pp. ARENDS, Jacques, Pieter MUYSKEN and Norval SMITH (eds.): Pidgins and Creoles. An introduction. 1994. xv, 412 pp. CLEMENTS, J. Clancy: The Genesis of a Language. The formation and development of Korlai Portuguese. 1996. xviii, 282 pp. THOMASON, Sarah G. (ed.): Contact Languages. A wider perspective. 1997. xiii, 506 pp. ESCURE, Geneviève: Creole and Dialect Continua. Standard acquisition processes in Belize and China (PRC). 1997. x, 307 pp. SPEARS, Arthur K. and Donald WINFORD (eds.): The Structure and Status of Pidgins and Creoles. Including selected papers from meetings of the Society for Pidgin and Creole linguistics. 1997. viii, 461 pp. RICKFORD, John R. and Suzanne ROMAINE (eds.): Creole Genesis, Attitudes and Discourse. Studies celebrating Charlene J. Sato. 1999. viii, 418 pp. McWHORTER, John (ed.): Language Change and Language Contact in Pidgins and Creoles. 2000. viii, 503 pp. NEUMANN-HOLZSCHUH, Ingrid and Edgar W. SCHNEIDER (eds.): Degrees of Restructuring in Creole Languages. 2001. iv, 492 pp. SMITH, Norval and Tonjes VEENSTRA (eds.): Creolization and Contact. 2001. vi, 323 pp. MÜHLEISEN, Susanne: Creole Discourse. Exploring prestige formation and change across Caribbean English-lexicon Creoles. 2002. xiv, 332 pp. MIGGE, Bettina: Creole Formation as Language Contact. The case of the Suriname Creoles. 2003. xii, 151 pp. MOUS, Maarten: The Making of a Mixed Language. The case of Ma’a/Mbugu. 2003. xx, 322 pp. ESCURE, Geneviève and Armin SCHWEGLER (eds.): Creoles, Contact, and Language Change. Linguistic and social implications. 2004. x, 354 pp.

Creoles, Contact, and Language Change: Linguistics and Social Implications (Creole Language Library)

Creolization and Contact (Creole Language Library)

Creole Formation as Language Contact: The Case of the Suriname Creoles (Creole Language Library)

Structure And Variation in Language Contact (Creole Language Library)

Creole Discourse: Exploring Prestige Formation and Change Across Caribbean - English-lexicon Creoles (Creole Language Library)

Mbugu (Creole Language Library)

Language Contact and Grammatical Change (Cambridge Approaches to Language Contact)

Language Contact, Creolization, and Genetic Linguistics

Language Planning and Social Change

Language Contact, Creolization, and Genetic Linguistics

Contact Languages: A Wider Perspective (Creole Language Library)

Contact languages: a wider perspective (Creole Language Library)

Language Contact and Bilingualism

Language Complexity: Typology, contact, change

Language Contact

Language Contact

Language Contact

Constructions and Language Change

Creoles, their Substrates, and Language Typology (Typological Studies in Language)

Creoles, their Substrates, and Language Typology

Language Contact (Cambridge Textbooks in Linguistics)

Language Change (Language Workbooks)

Language and Social Relations

Language Complexity: Typology, contact, change (Studies in Language Companion Series)

Language and Social Identity

Language and social behaviour

Cognitive Linguistics and Language Teaching

Language and Power (Language in Social Life)

Kriyol Syntax: The Portuguese-based Creole Language of Guinea-bissau (Creole Language Library)

Kriyol syntax: the Portuguese-based Creole language of Guinea-Bissau (Creole Language Library)

Synchronic and Diachronic Perspectives on Contact Languages (Creole Language Library, Volume 32)

Creoles, Contact, and Language Change: Linguistics and Social Implications (Creole Language Library)

Creolization and Contact (Creole Language Library)

Creole Formation as Language Contact: The Case of the Suriname Creoles (Creole Language Library)

Structure And Variation in Language Contact (Creole Language Library)

Creole Discourse: Exploring Prestige Formation and Change Across Caribbean - English-lexicon Creoles (Creole Language Library)

Mbugu (Creole Language Library)

Language Contact and Grammatical Change (Cambridge Approaches to Language Contact)

Language Contact, Creolization, and Genetic Linguistics

Language Planning and Social Change

Language Contact, Creolization, and Genetic Linguistics

Contact Languages: A Wider Perspective (Creole Language Library)

Contact languages: a wider perspective (Creole Language Library)

Language Contact and Bilingualism

Language Complexity: Typology, contact, change

Language Contact

Language Contact

Language Contact

Constructions and Language Change

Creoles, their Substrates, and Language Typology (Typological Studies in Language)

Creoles, their Substrates, and Language Typology

Language Contact (Cambridge Textbooks in Linguistics)

Language Change (Language Workbooks)

Language and Social Relations

Language Complexity: Typology, contact, change (Studies in Language Companion Series)

Language and Social Identity

Language and social behaviour

Cognitive Linguistics and Language Teaching

Language and Power (Language in Social Life)

Kriyol Syntax: The Portuguese-based Creole Language of Guinea-bissau (Creole Language Library)

Kriyol syntax: the Portuguese-based Creole language of Guinea-Bissau (Creole Language Library)

Synchronic and Diachronic Perspectives on Contact Languages (Creole Language Library, Volume 32)

Recommend Documents