Professional Documents
Culture Documents
Downloaded from https://www.cambridge.org/core. University of Cambridge, on 02 Feb 2020 at 21:41:16, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Cambridge Approaches to Language Contact
General Editor
SALIKOKO S. MUFWENE, University of Chicago
Editorial Board
robert chaudenson, Université d’Aix-en-Provence
braj kachru, University of Illinois at Urbana
raj mesthrie, University of Cape Town
lesley milroy, University of Michigan
shana poplack, University of Ottawa
michael silverstein, University of Chicago
Published titles
Salikoko Mufwene, The Ecology of Language Evolution
Michael Clyne, Dynamics of Language Contact
Bernd Heine and Tania Kuteva, Language Contact and Grammatical Change
Edgar W. Schneider, Postcolonial English
Virginia Yip and Stephen Matthews, The Bilingual Child
Bernd Heine and Derek Nurse (eds.), A Linguistic Geography of Africa
J. Clancy Clements, The Linguistic Legacy of Spanish and Portuguese
Umberto Ansaldo, Contact Languages
Jan Blommaert, The Sociolinguistics of Globalization
Carmen Silva-Corvalán, Bilingual Language Acquisition
Emanuel J. Drechsel, Language Contact in the Early Colonial Pacific
Lotfi Sayahi, Diglossia and Language Contact
Downloaded from https://www.cambridge.org/core. University of Cambridge, on 02 Feb 2020 at 21:41:16, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
The Emergence of Hybrid
Grammars
Language Contact and Change
Downloaded from https://www.cambridge.org/core. University of Cambridge, on 02 Feb 2020 at 21:41:16, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
University Printing House, Cambridge CB2 8BS, United Kingdom
www.cambridge.org
Information on this title: www.cambridge.org/9780521769983
© Enoch Oladé Aboh 2015
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2015
A catalog record for this publication is available from the British Library
Library of Congress Cataloging in Publication data
Aboh, Enoch Oladé, author.
The emergence of hybrid grammars : language contact and
change / Enoch Oladé Aboh.
pages cm. – (Cambridge approaches to language contact)
ISBN 978-0-521-76998-3 (hardback)
1. Interlanguage (Language learning) 2. Sublanguage. 3. Native
language. 4. Language acquisition – Age factors. I. Title.
P118.23.A36 2015
4170 .22–dc23
2014047368
ISBN 978-0-521-76998-3 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Downloaded from https://www.cambridge.org/core. University of Cambridge, on 02 Feb 2020 at 21:41:16, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Contents
1 Introduction 1
2 The agents of creole formation: geopolitics and cultural aspects
of the Slave Coast 16
3 The emergence of creoles: a review of some current hypotheses 60
4 Competition and selection 113
5 The role of vulnerable interfaces in language change: the case
of the D-system 171
6 The emergence of the clause left periphery 222
7 The emergence of serial verb constructions 269
8 Conclusions: some final remarks on hybrid grammars, the creole
prototype, and language acquisition and change 304
References 317
Author index 337
Language index 340
Subject index 343
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 09 Jan 2020 at 13:07:19, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Figures
vi
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 10 Jan 2020 at 06:08:20, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Maps
vii
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 10 Jan 2020 at 07:17:00, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 10 Jan 2020 at 07:17:00, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Foreword
salikoko s. mufwene
The present book represents the outcome of research that Enoch Oladé Aboh has
conducted on structures of especially his mother tongue Gungbe and on Haitian,
Sranan, and Saramaccan, over the past fifteen years, counting from the time
when I first met him at a conference on African linguistics at the University of
Cape Town, in January 2000. He has since then contributed, in paper after paper,
an enlightening comparative perspective that sheds informative light on how far
similarities obtain between the Gbe languages and these three creoles. This
subject matter has become central in the literature on the emergence of creoles
thanks to some relexificationists who claim that grammars of creoles are
patterned on those of their most important substrate languages.
In the case of these specific creoles, the field has been blessed to see some of
the relevant issues addressed by a linguist who is a native speaker of one of the
Gbe languages, who is impeccably well trained in the generative framework
used by prominent relexificationists such as Claire Lefebvre and John
Lumsden, and who has relentlessly investigated aspects of the grammars of
the relevant creoles based on not only the literature but also work with their
native speakers, as well as collaboration with other experts on the relevant
languages. So, this book brings to its readers the fruits of the application of the
comparative approach (not to be confused with the comparative method of
historical linguistics) to the letter. The reader will find in the following pages
plenty of data that raise issues about the Relexification Hypothesis applied to
Haitian, Sranan, and Saramaccan.
Contrary to claims that creoles form a typological class of their own, Aboh also
shows that the grammatical structures of these particular creoles are not identical,
not even in some respects that can be associated with their apparently most
conspicuous Gbe substrates. Some of the structures that distinguish them from
their lexifiers are not shared identically in the Gbe group either. Although this
does not mean that we should deny substrate influence on creoles’ structures, we
nevertheless should not continue thinking of structural similarities between
creoles and their dominant susbstrates in the traditional way, viz., that
particular meanings or structures from a particular language or group thereof
were transferred intact into the system of the emergent language (variety).
ix
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 07:13:12, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
x Foreword
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 07:13:12, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Foreword xi
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 07:13:12, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
xii Foreword
Aboh likewise mines the alleged simplicity of creoles not only by providing
counterevidence but also by asking whether a linguistic system should be
reduced to its morphosyntax, detached of its semantics. Is it accurate to
reduce morphosyntactic complexity/simplicity to what Michel DeGraff has
identified as “bit complexity,” which amounts to how many units (phonemic,
morphemic, and syntactic rules) a linguistic system is populated with? How
about the constraints that govern the usage of the relevant units and the ways in
which rules of their use interact with each other, either in feeding or bleeding
some of them? How about the impact of an overall simplicity of a language on
its expected capacity to convey information successfully, allowing easy
discrimination between different contents?
A great number of John McWhorter’s claims has also depended on his
assumption that creoles are young languages. Assuming that there is genetic
discontinuity between the lexifier and the relevant creoles, is complexity really
a function of how old a language is? Did modern “young languages” start from
scratch in the same ways that the hominine species evolved from having no
linguistic communication to producing languages, going through an embryonic
protolinguistic stage that involved short lists of denoting terms and minimal
grammar? Did the minds of the inventors of modern language varieties such as
creoles or even pidgins regress to the state of the mind of Homo erectus or early
Homo sapiens? There is no evidence in these language varieties or the
ecologies of their emergence suggesting that they may illustrate the
disputable position that “ontogeny recapitulates phylogeny.”
As a matter of fact, one should also ask whether, for instance, Gullah is
younger than other American English varieties and whether it is justified to
deny its ultimate ancestry in Old English, or perhaps the ancestors thereof.
From a genetic linguistics point of view, wouldn’t a more adequate account of
the differential evolution of English be one that does not too hastily invoke
overall simplification for some varieties simply because they have been
disfranchised as creoles for reasons that appear to be fundamentally social?
Besides, the morphological simplification of English does not appear to be an
exclusive peculiarity of the formation of creoles; it appears to have been the
general evolutionary trajectory of the language since the transition to Middle
English. Even if the producers of creoles have overextended the simplification
of the lexifier’s morphology, haven’t they also complexified some other aspects
of the grammars of the emergent language varieties? In the final analysis, don’t
things boil down to the typological realignment of some parameters of the
creoles’ grammars compared to their lexifier?
It is in the context of the above considerations that Aboh submits the
hypothesis of hybrid grammars. This notion is so much the more interesting
as it addresses the emergence of new language varieties from both the
ontogenetic and the communal perspectives, focusing respectively on
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 07:13:12, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Foreword xiii
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 07:13:12, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
xiv Foreword
shows that even a rule from a particular source language can be selected
partially; substrate grammatical patterns were not necessarily faithfully
reproduced in the emergent creoles and certainly not in identical ways from
one creole to another. This is as obvious with serial verb constructions as with
determiners and predicate fronting. Substratists should especially remember
that even the semantics was not faithfully reproduced, aside from the fact that
internal dynamics of the emergent systems generated some peculiarities that
can be traced to neither the lexifier nor the particular substrate languages
singled out in the present study.
This book reveals how complex the subject matter of the emergence of
creoles is and how germane it is to that of the formation of idiolects and
communal language creation in general. It is a demonstration of the useful
contribution that genetic creolistics can make to the study of language
acquisition and change, as well as to theoretical linguistics. Transcending the
traditionally narrow confines of discussions in creolistics, the book is both a
strong argument against creole exceptionalism and an invitation to keep away
from simplistic accounts. For those who have been complaining about lack of
data that support the contention that creoles have evolved like other natural
languages (bearing in mind that every case of language evolution is local,
constrained by ecological factors specific to the setting and time of its
emergence), there is no scarcity of data in this book. One is instead struck by
the modesty of the author and the soft tone of his claims. As I said in Mufwene
(2001), creolistics should not be a consumer discipline content with borrowing
ideas from other areas of linguistics (and other disciplines); it should also
contribute to understanding issues relevant to linguistics at large. Being more
detailed in the analyses of a wide range of constructions than is typical in
creolistics, The emergence of hybrid grammars has passed another milestone
toward that goal. Non-creolists that are typologically inclined will find it
compelling, while theoretically inclined creolists will appreciate the ways in
which the author makes the data relevant to formal syntax and historical
linguistics. I am so proud that Aboh chose Cambridge Approaches to
Language Contact to defend his positions and flattered that he invited me to
write a foreword for it.
s a l i k o k o s . mu f w e n e , University of Chicago
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 07:13:12, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Acknowledgments
I became familiar with some of the literature in creolistics in the mid 1990s
when I was writing my thesis. Back then, part of the generative literature on the
Gbe (and more generally Kwa) languages stemmed from work by Claire
Lefebvre and her team at UQAM. It also happened that most of this study
was meant to show substrate influence in creole genesis: the role of Gbe (Kwa)
languages in the emergence of the Caribbean creoles. While some of the
features or patterns identified in the creoles as Gbe/Kwa substrate influence
looked reasonable to me (e.g., predicate cleft), some others (e.g., the
directionality of certain functional items) appeared doubtful. In addition,
even the features that could be ascribed to the substrate languages never
seemed replicated identically in the creole. This observation led me to think
that the creoles I was reading about then might be telling us something more
fundamental about language acquisition in general: how learners weave
together different aspects of the heterogeneous inputs they are exposed to.
The chapters in this book present some of my thinking on this question over the
past decade.
Since the mid 1990s, I’ve had opportunities to talk to many people who
helped me think harder and reformulate the conclusions presented in this book.
I probably don’t recall everyone who took part in what turned out to be a
fantastic journey for me, but none of this could have been possible without the
friendship and support of Norval Smith, Pieter Muysken, Michel DeGraff, and
Salikoko Mufwene.
My thoughts actually started taking form when in 2001 Norval and Pieter
offered me a postdoc position in their NWO project: The TransAtlantic
Sprachbund. During the following ten years or so, I’ve have daily
discussions with Norval both in our offices over a coffee and on the train on
our way back home. Needless to say I learned a lot during these discussions, but
most importantly, I realized that history matters. Thanks, Norval. And thank
you, Pieter, for being always supportive and for always pointing me in the right
direction. My gratitude also goes to other members of the TransAtlantic
research team: Felix Ameka, Adrienne Bruyn, James Essegbey, and Margot
van den Berg.
xv
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 22:02:19, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
xvi Acknowledgments
It was also during this time that I met Michel DeGraff and Salikoko
Mufwene. I’m immensely indebted to both of them for their friendship and
support. Michel and Sali commented on almost all my writings related to
creolistics. Their numerous constructive comments, suggestions, and positive
criticisms throughout these years helped me refine the theory developed in this
book, broaden my research, and think harder about the big picture. I owe a great
deal of the literature cited in this work to them, and both gave me several
opportunities to present my work to broader audiences: Michel by inviting me
to teach a semester at MIT in spring 2008 and Sali by giving me the chance to
publish this book in this series. Sali’s role as a friend and editor of the
Cambridge Approaches to Language Contact series cannot be overestimated.
He has been supportive of this project from the very early phases. Despite his
busy schedule, he has always managed to find the time to critically evaluate the
manuscripts I sent him, starting from the book proposal up to the final product
now in the hands of the reader. It is not an overstatement to say that this book
could not have existed without his contribution. Kouzen and Grand-Frère, I
alone assume full responsibility for all remaining shortcomings and hope that
the end result does not let you down: ùn dó kpέ ná mì.
I’m immensely grateful to my sister Frieda Sessi Aboh for her numerous and
very constructive comments on earlier versions of Chapter 2. My gratitude also
goes to Umberto Ansaldo and Anne Zribi-Hertz for their comments and
suggestions on various aspects of the work presented here and for being very
supportive to me throughout these years. I feel equally indebted to my students,
my colleagues of the Amsterdam Center for Language and Communication,
and particularly to Kees Hengeveld for his friendship, for always being
supportive, and for making it possible for me to carry out my research in the
best of working conditions. I also thank Ekaterina Bobyleva for the numerous
constructive discussions we’ve had while she was writing her thesis under my
supervision, and Roland Pfau for the past gezellig moments and those to come.
Parts of Chapters 4, 5, and 6 were previously published as self-contained
essays, but much of this book took form in 2011–2012 when I was offered a
Netherlands Institute for Advanced Study in the Humanities and Social
Sciences (NIAS) fellowship for a year. I’m grateful to NIAS for believing in
this project and for giving me the chance to complete it. I’m particularly
indebted to my colleague Aafke Hulk, then the rector of NIAS. It is clear to
my mind that I could never have written this book if I had not been given the
opportunity to spend a whole year at NIAS, writing, meeting with people of
different backgrounds in the humanities, and contemplating the wonders of
nature in Wassenaar. Aafke, thanks for giving me this opportunity.
Finally, I owe the greatest debt to Anne, my wife, our son Fèmi and
daughters Aniola and Orê for always being there for me. Thanks for your
love. This book is dedicated to you.
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 07 Jan 2020 at 22:02:19, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Abbreviations
xvii
Downloaded from https://www.cambridge.org/core. The University of Iowa, on 18 Jan 2020 at 13:04:36, subject to the Cambridge Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
xviii List of abbreviations
Downloaded from https://www.cambridge.org/core. The University of Iowa, on 18 Jan 2020 at 13:04:36, subject to the Cambridge Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
1 Introduction
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
2 Introduction
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
Introduction 3
1
According to Chomsky (1986), an Internal-language, which is the object of linguistic study, is the
mental representation of the linguistic knowledge of a speaker. Thus, I-language is a mental
object, which contrasts with External language roughly characterized as the knowledge of
language use in a community, which includes performance.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
4 Introduction
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
Introduction 5
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
6 Introduction
Once we adopt this perspective, it becomes clear that the study of the
creation of change informs us about the evolution of language in general but
it also bears on issues of language acquisition. In understanding how learners
(whether L1/2L1 or L2) create new idiolectal varieties based on their individual
linguistic experiences, we hope to understand how learners access abstract
linguistic properties during the acquisition process. A crucial assumption that
this book builds on is that new varieties do not generally emerge as the result of
misanalysis (or imperfect replication), as is often assumed in theories relying
on imperfect SLA, but from ‘a certain analysis of abstract features of linguistic
entities’ (see, for instance, Klein and Perdue 1997). Though this linguistic
analysis may be deviant from the one generally accepted in the community or
the one a school teacher may expect learners to make in order to acquire
command of the target language (i.e., the normative form), it remains one of
the possible alternatives the learner has access to, based on UG and linguistic
experience. This in turn implies that in order to understand the driving forces of
language change we have to try to probe into the speaker’s knowledge at the
moment of the creation. That is, we have to probe into I-language. Obviously,
such endeavor can only be indirect, and in most cases speculative because we
often do not know when the change came into existence and diachronic studies
only inform us about the distances between two communal systems (which are
collective abstractions), for instance, between Middle English and Modern
English. While the enterprise may look completely hopeless, creole
languages might offer us an extraordinary opportunity to investigate the
creation of change, largely because of the shallow history and because we
have more information about the communal systems in contact and about the
ethnographies of the contacts than in traditional historical linguistics.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
1.1 Creoles as a test-bed 7
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
8 Introduction
interlanguages of an early stage.’ What this means is that creoles are instances
of interlanguages that got frozen before their inventors had the chance to reach
a more advanced stage in acquiring the target language. Here again, SLA
failure is assumed to be one of the driving engines of creolization.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
1.2 Speakers create variants not imperfect replicas 9
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
10 Introduction
see Smith 1987, Arends 1989), but differ with regard to their major superstrate
languages: French and English, respectively. According to Aboh (2006a), these
two creoles display specificity markers which function in a way very similar to
the Gbe languages. Therefore, both Haitian and Sranan display bare nouns
similarly to Gungbe as shown by the bracketed nouns in (6a–c):
(6) a. [Wosiyòl] manje kowosòl. [Haitian; DeGraff 1992: 105]
nightingale eat soursop
‘Nightingales eat soursop.’
b. Kofi, go na wowoyo go bai [bana] tya kon gi mi. [Sranan]
Kofi go LOC market go buy banana carry come give 1SG
‘Kofi, go to the market to buy me (plantain) banana(s).’
c. [Àxwèlé] nɔ̀ ɖù [gbàdó]. [Gungbe]
turtle-dove HAB Eat corn
‘A turtle-dove habitually eats corn.’
‘Turtle-doves habitually eat corn.’
Such nouns can be interpreted as definite, indefinite or generic
depending on context. Yet, Gungbe, Haitian, and Sranan display discourse
specificity markers (7):
(7) a. Wosiyòl manje [kowosòl a]. [Haitian; DeGraff 1992: 105]
Nightingale eat Soursop DET
‘Nightingales ate the soursop (in question).’
b. Kofi, teki [a bana] tya gi mi. [Sranan]
Kofi take DET banana carry give 1SG
‘Kofi, give me the (plantain) banana (in question).’
c. Àxwlé ɖù [gbàdó lɔ́ ] [Gungbe]
turtle-dove eat Corn DET
‘A turtle-dove ate the corn (in question).’
‘Turtle-doves ate the corn (in question).’
Aboh (2006a: 224) defines the combination of specificity and
definiteness in these languages as in (8) (see Chapter 5 for discussion):
(8) a. A specific definite noun phrase is strongly D(iscourse)-linked and represents a
unique referent assumed to be known to both speaker and hearer, to which the
speaker intends to refer.
b. A specific indefinite noun phrase need not be D-linked. It represents an
existing referent that the hearer may not know about, but which the speaker
has in mind and to which he/she intends to refer.
As is obvious from these examples, Haitian and Gungbe display the same
pattern in that the noun phrase precedes the specificity marker (viz, NP-det). In
Sranan, however, this marker precedes the noun phrase (viz., det-NP) even
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
1.2 Speakers create variants not imperfect replicas 11
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
12 Introduction
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
1.3 The structure of the book 13
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
14 Introduction
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
1.3 The structure of the book 15
While the discussion shows that we can account for creole formation by
investigating the recombination of syntactic features, it also indicates that such
recreated forms or structures often arise with local complexities: a
reorganization of the relevant linguistic module, such as when the creole
displays patterns which not only are absent in the source languages but also
exhibit ‘hybrid’ properties that could not have emerged independently from
these languages. For instance, Haitian does not exhibit consecutive serial verb
constructions of the Gbe type (Súrù ɖà làn ɖù. Lit. Suru cooked rice ate) even
though it displays serial verb constructions of transfer of possession that are
very similar to those found in Gbe and Kwa in general. These two types of
series are absent in French. Accordingly, Haitian appears to be more similar to
Gbe than it is to French in this respect. However, Haitian also displays
causative series involving the verb fèt (< faire) as V1. Such series are absent
in Gbe, contrary to French, where they are comparable to causatives involving
the verb faire ‘to make.’ In sum, Haitian developed verb serialization patterns
that are not isomorphic with the Gbe or French structures that contributed to
their development. In addition, because serialization in Haitian comes with its
own irregularities, this domain of Haitian grammar involves local complexity.
Given this observation and assuming that similar developments affect other
domains and sub-modules of Haitian grammar, one may reasonably wonder on
which ground various recent studies claim creoles to form a uniform language
type with simple grammars or even to involve the ‘simplest grammar of all’
(McWhorter 2001).
Chapter 8 concludes the book indicating that natural languages are hybrid
systems that emerge in the course of acquisition.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:17:25, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.002
2 The agents of creole formation: geopolitics
and cultural aspects of the Slave Coast
16
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 17
GHANA TOGO
Rep. of BENIN NIGERIA
TADO
EY
BOM
OYO
AG
ALLADA
BADAGRI/APA
PORTO-NOVO
KUTONOU/COTONOU
JAKIN/OFFRA
GLEHUE/WHYDAH
XWLAGAN/GREAT POPO
LITTLE POPO
LOME
KETA
ACCRA
1998, 2003; van den Berg 2007 for Suriname creoles, and Sylvain 1936; Singler
1996; Anglade 1998; Lefebvre 1998; Aboh 2006a; Aboh and DeGraff 2014,
forthcoming; Aboh and Smith 2015a) have shown that speakers of the Gbe
languages of the Kwa family contributed to the formation of these creoles.1
Section 2.2 briefly discusses population factors on the plantations in Suriname
and Haiti. Because there have been many well-informed studies on this topic, this
section only summarizes the historical and linguistic findings that are relevant for
the discussion in this book. Section 2.3 concludes the chapter.2
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
18 The agents of creole formation
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 19
2.1.1 The Aja communities on the Slave Coast: the case of Allada
Referring to the Slave Coast, Law (1994: 59) remarks:
This section of the West African coast east of the River Volta, comprising the western
half of the Bight of Benin, was so prominent in the supply of slaves for the Atlantic trade
that it became known to Europeans as the “Slave Coast”. The most powerful state (and
principal supplier of slaves for the Atlantic trade) in the region during the sixteenth and
seventeenth centuries was the kingdom of Allada.
3
This difference in the naming patterns of regions on the continents on both sides of the Atlantic
may be correlated with Mufwene’s (2001, 2005a) distinction between trade and settlement
colonies, the former in Africa and the latter in the Americas. They reflect differences not only
in the kinds of investments that Europeans had in the colonies but also in the attitudes they
developed toward them.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
20 The agents of creole formation
4
Pazzi (1979: 156) established this approximate date based on the fact that the Aja people were
never converted to Islam unlike most kingdoms in the environs of the Niger in early 1000 AD.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 21
EY
TADO
BOM
OYO
AG
ADA
ALL
N tse
PORTO-NOVO
JAKIN/OFFRA
BADAGRI/APA
KUTONOU/COTONOU
GLEHUE/WHYDAH
XWLAGAN/GREAT POPO
LITTLE POPO
LOME
KETA
Map 2.2 Map of migrations within and around the Allada Kingdom
about the rise of Allada but also about its institutions. Given that Allada was
developed enough to have political relations with the King of Beni, its
founders, that is, the people who left Tado (i.e., the Agàsú royal family),
must have arrived there several decades, or maybe a century, earlier. This
would set the foundation of this kingdom between the late fourteenth century
and the fifteenth century. That the King of Allada was able to establish regional
connections with other kings in the sixteenth century (e.g., the King of Beni)
and was then known to Europeans indicates that it was already a prosperous and
powerful kingdom. For instance, Father Jose de Najara, who visited this area in
1672, reports on an event that Pazzi (1979: 175) dated from around 1500
saying: “within the great number of wives that the King of Popo had there
was even a sister of the King of Allada, who for some reason refused to pay the
yearly tribute.” It is further mentioned in this report that the King of Allada
eventually defeated the King of Popo. The expansion of Allada, partially
illustrated by this report, continued throughout the sixteenth century. A note
by du Casse (1687) cited in Pazzi (1979: 176), reports that “the kingdoms of
Great Popo and Little Popo were both satellites of Allada . . . The same holds of
Whydah which depended on Allada” (my translation). These different reports
suggest that the Kingdom of Allada was organized around a central power that
ruled over a number of satellite or smaller kingdoms that had been annexed.
The exact size of Allada is difficult to estimate but some reports in the
seventeenth century (e.g., Olfert Dapper, a Dutch geographer) suggest that it
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
22 The agents of creole formation
TADO
GHANA TOGO Rep. of BENIN NIGERIA
Y
ME
GBO
OYO
A
ALLADA
PORTO-NOVO
BADAGRI/APA
KUTONOU/COTONOU
JAKIN/OFFRA
GLEHUE/WHYDAH
XWLAGAN/GREAT POPO
LITTLE POPO
LOME
KETA
stretched from Dànxòmὲ , to the north up to the coast on the Atlantic Ocean. It
shared a border with Oyo to the east and northeast (as indicated by the fact that
the port of Apa, in present day Badagri, in Nigeria, was under its control) and it
included the territories up to the defeated kingdoms of Great Popo to the west
(Pazzi 1979; Law 1997).5
A point shown on this map that is immediately relevant to this book is that
Allada once controlled all the major ports on the Slave Coast where the
Europeans were trading and sometimes settled: Apa/Badagri, Jakin, Offra,
Glehue/Whydah, and Xwlagan/Great Popo. While Jekin (also referred to as
Jakin, Jakkein, Jaquin) and Offra (sometimes referred to as Offer) seemed to
have been the primary ports of Allada, sources also occasionally mention
Whydah and Apa as controlled by Allada to whom the local governors
(Tògán ‘chief of country’) had to pay a yearly tribute. That such governors
5
Though the kingdom of Dànxòmὲ eventually became independent and even conquered Allada in
1724, I’m assuming here that it once belonged to the network of satellite kingdoms that were
under the rule of Allada and to which it had to pay tributes. For instance, a report dating from
1716 describes the ruler of Dànxòmὲ as follows “The Viceroy of Agoeme [i.e., Dahomey][in the]
province of Foin [Fon] in the dependent territories of the King of Ardres has thrown off the yoke,
he has made himself independent” (Law 1997: 110). This actually corresponds to oral tradition
where it is believed that Dànxòmὲ (Dahomey) was part of Allada until it became independent.
Oral history also suggests that prior to independence the Viceroy of Dànxòmὲ was appointed in
Allada where the rituals were held (see Pazzi 1979; Adande 1984; Law 1991, 1997 for
discussion).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 23
were under the rule of the King of Allada can be inferred from the fact that the
latter is referred to as “the Great King of Allada” in some historical documents.
This is presumably a translation of the local terminology such as the Gungbe
title Àxɔ́ sú àxɔ́ sú lέ tɔ̀ n ‘the King of kings’ (Pazzi 1979; Adande 1984; Law
1991, 1997). That Allada could have control over the coast can also be inferred
from the size of its army which amounted to 40,000–50,000 (Law 1997: 20).6
We can infer from these figures that Allada was a densely populated
kingdom. According to Law (1997: 20):
the Allada kingdom was generally regarded as relatively densely populated: Dapper in
the mid-seventeenth century, for example, describes it as having ‘towns’ [vleken, lit.
markets] and villages in great number; d’Elbée, traveling from Offra to the capital in
1670 found the land ‘peopled with many towns and villages’. The Dutch in the late
seventeenth century, it may be noted, thought Allada more densely populated than Benin
to the east.
The reports then continued with great details regarding the customs of the
inhabitants, their diets, cooking traditions, kitchen utensils, clothing, hygiene,
and funeral rites. The King is said to have two royal palaces in the capital,
which, according to Olfert Dapper as cited in Law (1997: 15), “were enclosed
by thick earth walls, and each was as large as the town of Monnikendam in
Holland” (see also Pazzi 1979 and Adande 1984). Likewise, Captain François
d’Elbée, who visited the city in 1670 as a representative of the French
Compagnie des Indes Occidentales, stayed in the royal palace, where he was
put up in an apartment reserved for foreigners. He reported that:
6
Other sources, e.g., d’Elbée (p. 381), estimate the King’s army at 500,000.
7
“C’est un païs plein de villages et bien situé pour le commerce, parce que c’est un terroir fecund
en millet, en vin de palme, en plantes et en fruits qui durent toutes l’année et qui leur sont d’un
grand secours, comme les injames, les batatates, les oranges, les citrons, les noix de coco etc.
C’est un païs de plaines et de vallées entrecoupé de grandes rivières et les chemins qui y mènent
sont grands et bien battus. On fait beaucoup de sel sur la côte d’Arder.” (The spelling is as in the
source document; see Adande 1984: 239.)
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
24 The agents of creole formation
this palace is very large and well constructed. It is divided into large courts and large
gardens. The building is all surrounded by porches which form several galleries that lead
to the various courts or gardens. It is a two-storey-house composed of big halls and
rooms including various furniture.8
As noted by Law (1997: 15), the existence of a storey-house in the Allada of the
seventeenth century where the King can put up some of his visitors is
particularly noteworthy as this is indicative not only of the wealth of the
Kingdom but also of its socio-cultural aspects and more generally of African
civilization prior to colonization. The description that Captain François d’Elbée
made of the royal palace also gives us indications about how to understand
Allada within the geopolitics and economies of the seventeenth century. Allada
was participating in international trade and the King had political and
commercial agreements with other kings in West Africa and overseas. We
know from the discussion in previous paragraphs that the King of Allada sent
ambassadors to Beni back in 1539. In addition, of the three kings that are
known to have ruled Allada, the first, Toxonu, sent an ambassador to the King
of Spain Philippe IV in 1658. Diplomatic relations between Allada and Spain
are also illustrated by the publication in 1658 of La Doctrina Christina by
Spanish missionaries who had settled there. This catechism represents one of
the oldest documents on a Gbe language that exists to this day and on which I
comment below.
The second King, Tézinon, received Captain François d’Elbée in the year of
1669–1670, as noted above. One objective of this visit was to re-establish and
strengthen commercial and diplomatic relations between the King of Allada
and the King of France Louis XIV. During Captain d’Elbée’s stay in Allada, an
incident between the newly arrived French and the Dutch factors who were
settled in Jekin for 25 years was brought to the King. In settling these matters,
the King Tézinon decided to send an ambassador, Mattheo Lopes (see
Figure 2.1), to the King of France Louis XIV. Mattheo Lopes who was one of
the King’s most appreciated translators embarked on La Concorde, the second
French ship sent to Allada the same year. Mattheo traveled to France with three
of his sons, three of his wives and six slaves. He is reported to have arrived in
France on December 3 of the same year and met with Louis XIV on December
19 (d’Elbée 1670: 308–315; Adande 1984).
These facts indicate that Allada was probably one of the most powerful
kingdoms in West Africa in the sixteenth and seventeenth centuries both in
terms of its army and institutions. As mentioned in previous paragraphs, the
8
“Ce lieu est grand et spacieux et assez bien bâti, il est divisé en grandes cours et grands jardins,
et le bâtiment est tout entouré de pilliers qui forment des galeries et porches pour se promener tant
du côté des cours que des jardins : il est élevé de deux étages où il y a divers appartements
fort grands et spacieux composés de salles, chambres et autres accompagnements.” (D’Elbée
1671: 418).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 25
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
26 The agents of creole formation
Local economy
The local economy was based on locally produced goods that were traded in the
kingdom and were also exported. Among such products, we find farm products
mentioned above by Olfert Drapper (e.g., fruits, millet, vegetables, etc.). While
part of these products is for local consumption, others such as salt and palm oil
were also used for exports. For instance, Pazzi (1979: 154) refers to a letter
dated from 1574 by G. M. Branco, a Portuguese merchant from São Tomé,
where he reports that “with our friend the King of Allada, who is close to Mina
[i.e., El Mina in present Rep. of Ghana] we trade for slaves, ivory, cotton cloth,
palm oil, and many vegetables such as yam and other foods. Each year we
import from that port one or two ships loaded of what is mentioned above” (my
translation).9
Clearly this text illustrates the two economic axes of Allada: slaves and
locally produced goods whether agricultural or manufactured. In addition, it
gives us an estimate of the quantity of such goods that is consistent with Olfert
Dapper’s description above that one finds a great number of vegetables and
other products there all year long. With regard to manufactured goods, Pazzi
(1979) further indicates that the economic strength of Allada also relates to its
craft industry involving artisans who were probably hired from neighboring
areas or kingdoms. A case in point is cotton cloths. Indeed, that Allada was
selling cotton cloths in the sixteenth century indicates that it was now
9
“Avec notre ami le roi d’Arda /Alada/ qui est près de Mina, nous envoyons traiter des esclaves
nègres, de l’ivoire, des pagnes de coton, de l’huile de palme et beaucoup de legumes comme
l’igname et autres aliments. Tous les ans il sort de ce port un ou deux navires de ce qui est indiqué
ci-dessus” (Pazzi 1979: 154, footnote 3). See also Law (1997: 92 and footnote 526).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 27
competing with Beni, which was then well known to Europeans for producing
cotton fabrics. One could think that the clothes bought in Allada might not have
been produced locally but were instead imported for subsequent export. While
this view may not be disregarded outright, some reports based on ships’
manifests in the seventeenth century cited in Pazzi (1979: 117) indicate the
following quantities: 588 pieces on a Dutch boat in 1645 and 5,000 pieces on an
English boat the following year. With regard to these cotton clothes, the English
Captain Philips remarks that, in Whydah another port further to the west, the
cotton clothes are made of strips of about 25 cm wide that are stitched together
in larger bands of three strips. The strips found in Allada were narrower. In
addition, Allada’s cotton strips also differed from those of Beni, which were
stitched together alternating between white and blue strips. Because the cotton
strips found in Allada were distinct from both those found in Whydah and
Benin, we can conclude that they were locally made (i.e., in the interior of the
Kingdom). We can infer from the Capitain’s remarks that Allada had local
weaving technology.10 Law (1997: 92) suggests that the cloth export from
Allada must have been somehow limited compared to that of Beni. However,
the relevant point here is the diversity of products and goods that could be
found on the coast of Allada which in turn could be indicative of the King’s
efforts to make Allada an attractive stop for European traders. Other
commodities sold to Europeans are cowries which served as exchange money
in West Africa during this period.
Given that these exports were mainly documented by Europeans trading on
the coast and involved in the slave trade, it is reasonable to assume that part of
these commodities were provisions for the slaves and the crews during the
middle passage. This would suggest that a significant part of Allada’s non-
human export actually fed into the slave trade. Thus, local economy, as defined
above, partially depended on the slave economy. This assumption is critical for
the argument put forth in this book as it relates to the origin of the Africans who
were exported to Haiti and Suriname in the seventeenth century.
Slave economy
As suggested by the quote from Law (1994) above, Allada was one of the main
suppliers of slaves for the Atlantic trade during the seventeenth and early
eighteenth centuries. We saw from the Portuguese merchant G. M. Branco’s
letter in previous paragraphs that the King of Allada was already trading slaves
back in 1574. Until 1635, the Portuguese monopolized the trade with Allada.
Comparing two Portuguese reports from 1607 and 1620 that showed estimates
of the value or volume of trade between São Tomé and ports in the Bight of
10
Actually, this has become a tradition in Benin, where one still finds traditional weavers in the
Royal Pallace of Agbómὲ, now a museum.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
28 The agents of creole formation
Biafra, Law (1994: 65) evaluates the number of slaves from Allada as about
667 slaves a year out of a total number of 1,167 slaves from the Slave Coast as a
whole. Though these numbers would grow dramatically when the trade was at
its peak (i.e., late seventeenth and eighteenth centuries), the Africans traded by
the Portuguese from Allada of the early seventeenth century represented
slightly more than fifty percent of the total numbers of slaves. As Law (1994)
remarks, this could be due to the fact that “although a small part of this trade
was in commodities other than slaves this is offset by the likelihood that (as was
generally the case later) slaves on the Slave Coast were cheaper than in
Angola.” This is indeed supported by the following extract of a letter by a
Dutch factor, I. Van Hoolwerf who arrived in Jeken (Allada’s port) in 1686.
According to him, “the slaves here are in good shape and cheap. As for our
living conditions here, we are in a period of peace with a relative abundance of
slaves” (Pazzi 1979: 64–65, my translation).11
The Portuguese monopoly over trade in the region was soon replaced by a
Dutch monopoly (1636–1653). Historical documents indicate that the Dutch
authorities from Brazil “envisaged imports of 1,500–1,600 slaves annually
from Allada, and a further 500–600 from Popo” (Law 1994: 66). While these
numbers were thought of as optimistic (Law 1994), they are nevertheless
important if we take into consideration the fact that Popo represented one of
the satellites of Allada. Adding these figures up, the estimates of slave out of the
Allada kingdom were about 2,000 to 2,200 annually. This estimate clearly
shows a sharp increase in the trade, which is four times the numbers reported in
previous paragraphs with regard to the Portuguese trade of the early
seventeenth century. Whether these objectives were achieved is unclear but
Law (1994: 66) indicates that the
WIC trade by van den Boogaart and Emmer suggests that Dutch purchases of slaves
from “Guinea” . . . averaged only about 1600 per year in 1637–44 and only once in
(1642) reached over 2000 in a single year. . . . Van den Boogaart and Emmer were able to
establish the provenance of around three-quarters of the total, of whom almost half
(48%) were from Allada.
According to Law (1994), the trade in Allada will then decline between 1649
and 1658 but peaked again in 1659–1667, when the Dutch had to compete with
other European nations also trading with Allada. The following dates show the
flow of different European trade settlements in Allada (Law 1997: 5–9).
– 1574–1636: The Portuguese monopoly.
– 1639: The Dutch West India Company (WIC) constructed a trade post in
Allada.
11
“Les esclaves ici sont bons et de bas prix. Quand aux conditions de vie dans cette region, nous
nous trouvons dans la paix avec une discrète abundance d’esclaves . . . ” (Pazzi 1979: 64–65).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 29
– 1658: Trade contacts between the King of Allada and the King of Spain.
– 1664: James Duke of York (future King James II) then Governor of the
Company of Royal Adventurers (CRA) wrote a letter to the King of Allada.
– 1670: The French West Indies Companies established a trade post in Allada.
The Kind of Allada sent an ambassador, Mattheo Lopez, to King of France
Louis XIV.
Another European company that is reported to have operated in Allada
during the same period is the Danish African Company. It is now clear that
not only were Allada and its main ports part of the world economy, it also seems
to have been a location favorable for trade where European nations competed.
This competition led to a dramatic increase in the number of slaves exported
from Allada. Of the 5,000 to 6,000 slaves the Dutch delivered to Spanish
America between 1668 and 1672, Law (1994: 73) reports that “it is clear that
a large proportion of these slaves were obtained from Allada.”12 This appears to
be consistent with the claims of François d’Elbée who visited Allada in 1670
and evaluated the exports from Allada at an average of 3,000 slaves per year.
Similarly, a Dutch source of 1670 indicates that “over and above the English
and French trade, the Dutch in Allada were purchasing every year . . . 2,500 to
3,000, sometimes even more slaves” (Law 1994: 75). In addition to the Dutch
trade, two French ships are reported to have commerced with Allada in 1670:
La Justice (captained by François d’Elbée) embarked 434 slaves, while La
Concorde, which arrived a few days before d’Elbée’s departure, embarked 563
slaves. Thus, Allada was exporting about 4,000 slaves a year. This incredibly
large number is consistent with another French source reported in Law (1994:
76) in which estimates of exports form Allada in 1671 were between 4,000 and
5,000 people.
Because these numbers sometimes include import estimates made as part of
the commercial plans of the European trading companies, they probably do not
reflect reality and should therefore be taken with caution (Curtin 1969). In an
effort to evaluate the overall slave export from Allada in the 1710s and its
impact on the local economy, Law (1997: 102) suggests that Allada’s highest
slave exports could be estimated at about 15,000 a year. Eltis (2011: 275) gives
even higher figures suggesting that the trade in Allada and its subordinated
ports (e.g., Whydah, Jakin) rose from about “1,000 a year in the 1650s to 8,000
in the 1680s, to a peak over 19,000 a year in the 1720s.”13 These impressive
numbers certainly fluctuated and varied according to various factors, including
12
Emmer (2005: 35), on the other hand, suggested a figure of 4,000 slaves delivered to Brazil per
year.
13
These figures can be put in the context of more recent figures by Eltis (2011: 272), which
indicate that in the years 1766 to 1776 the total volume of captives exported from Africa to the
Americas increased from about 12,000 to 80,000 a year: a clearly astonishing figure.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
30 The agents of creole formation
but not limited to war, climate, rain/drought, and epidemics. For instance, it is a
known fact that while wars may yield war prisoners to be sold as slaves, they
also often led to roadblocks whose primary impact was to prevent slave
caravans access to the coast. Similarly, several reports suggest that local
slave owners tended to sell their slaves in periods of droughts, when the costs
of maintaining a slave were high, rather than during advantageous seasons,
when the owners could use their slaves for their own purposes (e.g., farming),
to increase their income. A case in point is the following extract from a letter by
I. Van Hoolwerf dating from February 1688:
Since I arrived here, slave trade has been in steady progress, but these days, the trade is
slowing down because of the lack of wars in the interior and the abundance of grains this
year, which does not lead people to sell their slaves, contrary to last year during the
droughts. . . . The arrogance of the Fidalgo . . . A couple of months ago, he started a
war . . . as a consequence, the roads were completely blocked and we could no longer
receive slaves from the interior. (Pazzi 1979: 65–66, my translation).14
The fluctuations described by Hoolwerf affected not only the number of slaves
sold in Allada but also prices. In this regard, Law (1994: 78) reports that Dutch
accounts for slaves bought in Allada between 1638 and 1680 suggest an
average price of about 32 fl per head. English records of the 1660–1681
period, based on “prime costs i.e., cost in Europe of the goods exchanged for
slaves,” indicate an average price between £2.71 and £2.88. However, Law
(1997: 102) evaluates the price of the slave to be at 48.000 cowries
“conventionally equivalent to £15 in local value.” Assuming that Allada
slave export ranged between 15,000 and 19,000 slaves per year (Law 1997:
102 and Eltis 2011: 275, respectively), we can estimate the yearly gross income
of Allada to vary between £225,000 and £285,000, that is, between 720 and 912
million cowries.
Though these figures are speculative, we can infer from their magnitude that
various social classes in Allada must have benefited from the trade. Because of
its institutions and how the trade was organized locally, it appears that a large
portion of the income generated by trade in general and the slave trade in
particular was to the benefit of the King. This can be seen from his numerous
luxurious palaces described above. Other locals who benefited from this trade
were probably other state authorities (e.g., Ministers and Captains, see d’Elbée
1671: 438–440), the subordinated chiefs (i.e., the tò-gán) in the different ports
(e.g., Jakin, Whydah) and in the interior (e.g., Agbomey, capital of Dànxòmὲ )
14
“Jusqu’ici la traite a été en constant progress depuis mon arrive, mais en ces jours elle semble
ralentir un peu, à cause du manque de guerres à l’intérieur et de l’abondance des céréaux en cette
année, ce qui ne pousse pas, comme faisait la famine l’année passée, à vendre les esclaves . . .
L’insolence du Fidalgo . . . Il y a quelques mois, il se chargea d’une guerre sur lui seul . . . Les
chemins furent alors tellement bloqués que nous ne pouvions plus avoir d’esclaves” (Pazzi
1979: 64–65).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 31
2.1.2 Competition between Allada and its satellites and the fall of Allada
It is not difficult to imagine that Allada’s revenues from trade furthered
competition between the centralized power in the capital city of Allada and
the subordinated chieftaincies of the kingdom, including Dànxòmὲ and
15
What was happening then in Allada is quite similar to current situations in many developing
countries where a central power in alliance with some multinationals monopolize the resources
of the country at the expense of local populations. In the fifteenth to eighteenth centuries, the
resources then were manual labor: the slaves.
16
“Quoi que ce Prince entende fort bien le Portugais, ou pour mieux dire, l’Espagnol corrompu,
tenant de langue Franque, il ne laisse pas de se faire tout interpreter par ses deux interprètes,
Mattheo et Fransisco” (d’Elbée 1670: 405).
17
“La langue Portugaise corrompue s’y est conservée jusqu’à present, et produit un jargon ou
langue franque que Presque tout le peuple entend, et parle de sorte que ceux qui savent le
Portugais n’ont pas besoin d’interprète dans cet État” (Labouret and Rivet 1929: 20).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
32 The agents of creole formation
Whydah, and neighboring kingdoms such as Beni. Indeed, between the late
seventeenth and early eighteenth centuries, the Kingdom of Allada had to
contend with internal pressures from its various chieftaincies demanding
independence and strife from the ever-expanding Fon Kingdom of Dànxòmὲ
north to the Ko (see Maps 2.2 and 2.3).
According to Pazzi (1979: 195) the Kingdom of Dànxòmὲ emerged in the
early 1600s as an offshoot of the Allada royal family, after a family dispute over
the succession of the late King Kɔkpon. The royal family split into three
different groups: one remained in Allada representing the central power,
while the other two migrated (south)east and north, respectively. The north
bound group founded Dànxòmὲ. Given such circumstances, it is easy to assume
that it has always been a challenger to the central power in Allada, which it
eventually vanquished and conquered on March 30, 1724 (Law 1997: 114; see
also Pazzi 1979; Adande 1984). The decline and fall of Allada have been
largely attributed to internal divisions, presumably motivated by the need of
the King to maintain military power and control over an economy fuelled
largely by the triangular slave trade. Allada suffered a severe defeat against
Oyo in 1698, but the reasons for this war seemed to be “an appeal to the Oyo by
the subject of the King of Allada for protection against the misgovernment of
his viceroys” (Law 1997: 111). This happened shortly after the Fon from
Dànxòmὲ asserted their independence (see note 5) “The Viceroy of Agoeme
[i.e., Agbómὲ ] [in the] province of Foin [Fon] in the dependent territories of the
King of Ardres [Allada] has thrown off the yoke, he has made himself
independent” (Law 1997: 110). The following two reports from different
visitors cited in Law (1997) illustrate this period of instability:
Arda [Allada] was once a powerful and populous nation . . . but having afterwards been
weakened by the revolts of the Quitta [Keta] . . ., and the Popoe, Whydah, Appee [Ekpe],
Bidagry, and Lagos, it became incapable of opposing arms of Trudo. (Law 1997: 111,
emphasis is mine)
Trudo, commonly known as Agaja was then the king of Dànxòmὲ. Quite
remarkably, all the names in italics in this quote referred to slave ports once
controlled by Allada and around which big cities emerged (see Map 1 and
Pazzi 1979, Law 2011). This gives further credence to the idea that control
over the economy played an important role in the crisis that Allada went
through during the late seventeenth and early eighteenth centuries. As a
result, the fall of the Allada capital in 1724 is the epilogue to a long period
of tension involving successive wars which considerably weakened the
Kingdom. As Law states:
The Kingdom of Arda [Allada] was reduced, both in extent and power, by the revolt and
separation of several provinces, before it was attacked by the Dahomans, under Guadja
Trudo. (Law 1997: 112)
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 33
1680 1690 1700 1710 1720 1730 1740 1750 1660 1670 1780 1790
1689 1699 1709 1719 1729 1739 1749 1759 1669 1779 1789 1803
It is not clear to what extent this critical period and the subsequent fall of
Allada affected the slave trade generally. For instance, the English factor
Bulfinch Lamb, who was made prisoner during the conquest of Allada by the
Fon of Dànxòmὲ, reports that “there were more than eight thousand enslaved
prisoners of war” (cited in Pazzi 1979: 246). There is little doubt that these
prisoners were sold to European factors trading on the coast. These facts appear
compatible with the figures in Table 2.1 taken from Arends’ (1995b: 243)
estimations of Dutch slave imports to Suriname. This table suggests that while
the trade apparently flourished until 1740 with a peak period between 1700 and
1719, almost no slaves were bought on the Slave Coast (SC) until 1740 when
the Gold Coast (GC) became the major source of slaves.
The peak in slave numbers between 1700 and 1719 is important because it is
during this period that Allada had to face internal divisions often leading to
wars and roadblocks. In addition, the political crisis within Allada could have
forced the trade to shift to the Gold Coast, thus allowing some of the slaves of
Allada to be sold on the Gold Coast, which accounts for the flourishing of the
trade there by 1729 (see Pazzi 1979; Law 1994, 1997). With this in mind, we
can now ask: How structured was the supply of slaves and where did they
originate?
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
34 The agents of creole formation
ventures. Indeed, there must have been slave raiders as well as smugglers
selling slaves through the different ports of Allada. The following letter from
W. Bosman, a trader of the Dutch West India Company (WIC) cited in Pazzi
(1979: 67, my translation) is illustrative of such slave raiders.
A description of Little Popo, it is an infertile land: its inhabitants, some of whom were
chased away from Accra, and are good soldiers, their war with those of Offra /Jaken/ and
Fida /Whydah/ and what happened afterward. They too live on loot and slave
trafficking.18
Given the ‘illegal’ practices of the Dutch WIC on the West African coast
(Pazzi 1979; Law 1991, 1994, 1997), we can infer from this quote that a
fraction of the overall exported Africans from Allada and its related ports
was provided by slave raiders who were not necessarily under the control of
the central power in Allada. However, historical documents on this kingdom
suggest that much of the trade was strictly controlled (if not organized) by the
State. Because of the State central role in the trade, there exist some indications
about the origins of the slaves. Given the tensed political situation described
above and the Fon expansion that led to the fall of Allada, it is obvious that
these troubled times must have yielded a lot of enslaved prisoners. For instance,
Law (1994: 83) asserts that “in general, the evidence suggests that the principal
source of slaves for export was capture in warfare.” Because Allada was often
at war with its own subordinate states, many of which derived from the
ancestral royal town of Tado, we can assume the following:
– Many of the slaves sold on the coast came from further inland, as a con-
sequence of internal warfare.
– The slaves must have been primarily from the same Aja-Tado culture.
Earlier documents cited by Pazzi (1979) and Law (1994, 1997) suggest that
the slaves traded in Allada came from two zones: Yoruba lands (probably Oyo)
and Dànxòmὲ . As one reads in Law (1994: 83),
In the early Dutch trade, in the 1630s or early 1640s, it was reported that many slaves
were supplied for sale at Allada from the kingdom of “Lukumi” or Yoruba or even
further in the interior.
Olfert Dapper, who reported about Allada of the mid seventeenth century,
also suggests that “the slaves sold in Allada were brought from the kingdom of
Ulkami, i.e., Lukumi” (Law 1997: 101). Other documents indicate that the
slaves came from, or through, Dànxòmὲ. Indeed, a document dating from
March 1680 and cited by Law (1994: 83) suggests that “the greatest part of
18
“Une description de Petit Popo, c’est un pays infertile. De ses habitants, dont certains ont été
chassés d’Accra, et sont de bons soldats. Leur guerre avec ceux d’Offra et Fida, ce qui s’est
passé après. Eux aussi vivent de butin et de traffic des esclaves” (Pazzi 1979: 64).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 35
the slaves sold in Allada came from Dahomey.” Whether Dànxòmὲ , which was
subordinated to Allada until 1724, directly supplied the slaves to the coast or
allowed slave caravans from the north to cross its lands and reach the coast is
not clear. This is so even though Law (1997: 101–102) reports that “an increase
in the price of slaves at Whydah in 1688 was explained by reference to certain
differences with the King of Fouin [Fon], which had interrupted the supply
from the interior.” The question also arises with regard to other satellites of
Allada such as Whydah and Jakin for which it is not clear whether they were
direct providers of slaves. Though more research is needed here, the geopolitics
of the time which favored internal competition between the satellite states and
allowed various alliances supports the hypothesis that many of the slave
cohorts must have come from the interior of Allada, within the frontiers of
the kingdom (see Map 2.3), as a result of warfare. Various documents (and
citations) suggest that selling people from the interior must have been a
common practice throughout the seventeenth and the eighteenth centuries.
Law (1997: 104) reports that according to William Bosman, in the 1690s,
“most of the slaves that are offered to us are prisoners of war, which are sold
by the victors as their booty.” The author further mentions a document dating
from the 1720s where it is reported that:
The slaves exported through Whydah were captured in wars close to the coast, alleging
that, if the king of Whydah could not get enough slaves by “fair agreements with his
country neighbours”, he marches an army, and depopulates, and observing He and the
King of Ardra adjoining, commit great depredations inland. (Law 1997: 104–105).
Such practices must have continued after the Fon conquered Allada in 1724.
Bulfinch Lamb, the English trader who was in Allada when it was conquered by
the Fon in 1724 indicates that there were about 8000 prisoners (Pazzi 1979:
247). Clearly, these were soldiers of the defeated kingdom or inhabitants of the
capital city of Allada. As Pazzi (1979: 257) further explains, Agaja (i.e.,
Trudo), who conquered Allada, suffered a severe defeat by the Oyo in the
1730s and had to pay a yearly tribute to Oyo. Pazzi argues that such a payment
arrangement was possible thanks to the slave trade in Whydah (then under the
control of Dànxòmὲ). Raids on the Maxi people living north of Agbomey, the
capital city of Dànxòmὲ, were instrumental in keeping the trade going.
All these facts suggest that a significant portion of the slaves sold through the
ports of Allada came from within the Kingdom. With regard to the Slave Coast,
Arends (1995b: 249) argues that the maximum distance between the areas
where the enslaved Africans were captured (or bought) and the location where
they embarked is about 200 km. This puts the zone of ransack within the
frontiers of Dànxòmὲ, a former satellite of Allada.
The geopolitics of the Allada Kingdom therefore suggests that a significant
part of the slaves sold through its ports were of Aja origin. Various factors
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
36 The agents of creole formation
support this view. For instance, Captain François d’Elbée who visited Allada in
1670, reported that:
These slaves are of several sorts, some are prisoners of war, others contributed by
neighbouring kingdoms, and these are foreigners; in addition to which there are the
slaves born in his country, and those who are so through having incurred legal
punishments . . . there are also some of them who are condemned to slavery for failing
to pay their debts, and whom their creditors have sold for their payments. (cited in Law
1997: 104)
Though not explicitly stated in this quote, it appears from previous discussion
and the history of Allada that the prisoners of war must be of Aja descent owing
either to the Fon expansion or to internal conflicts between the different
chieftaincies or between the chieftaincies and the central power. Likewise,
Captain François d’Elbée reported that the Queen of Allada sold him eight of
her co-wives as slaves. These women were presumably of Aja descent. The
idea that a significant number of slaves sold on the markets of Allada must have
been of Aja descent is compatible with the following document cited in Pazzi
(1979: 66, my translation) which mentions the ethnicities of the slaves:
Van Hoolwerff had wounded three blacks, but not to death. Then those of Arda, after
surveilling him for seven or eight days attacked him and took him to the Fidalgo who
had his hands and feet tied and removed all the Arda slaves and Mina slaves from the
factory.19
While the Arda (Allada) people directly relate to the Aja-Tado migration, the
Mina include both Fanti-Anε from El Mina, and the Ga from Accra. These
people migrated to the Aja-Tado area where they merged into a single ethnic
group with its own language currently referred to as Gengbe. It is noteworthy
that the author of this report was aware of the ethnicity of the slaves in his
factory. Contrary to common beliefs, this is partial evidence that European
factors on the coast had good knowledge of local customs and the different
ethnicities they dealt with. Assuming that language is a reliable ethnicity
marker, we can infer that some of these factors also knew the local
languages, a fact that I will return to later.
Focusing on the ethnicity of the slaves, the structure of the trade as imposed
by the King of Allada represents an important factor which indirectly restricted
the slave supply to what we could refer to as a ‘local reservoir’ (i.e., mainly
people of Aja descent). Indeed, one reason why the Slave Coast provided a far
greater number of slaves (i.e., through Allada and its ports) than the Gold
Coast, for instance, has to do with its institutions and the authority of the State,
19
“Van Hoolwerff avait blesse trois noirs, mais non à mort. Alors ceux d’Arda, après l’avoir
surveillé pendant sept ou huit jours, l’ont attaqué et emmené chez le Fidalgo, qui le fit lier mains
et pieds et fit enlever tous les esclaves Arda et Myna qui étaient à la factorerie” (Pazzi 1979: 66).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 37
which Eltis (2011: 274) argues, is “necessary not so much for enslavement as
for providing orderly markets and debt-collection institutions at the point
where the enslaved Africans became the property of Europeans.” In the case
of Allada, the slave transactions were organized on the ports only after proper
authorization from the King. Law (1997) provides ample evidence of the royal
control over this trade. As discussed there, historical documents indicate that
trade in Allada was possible on the sole condition that the Europeans factors
first traded with the King or his courtiers. In this regard, a Dutch report
mentioned that “you cannot conduct trade with anyone else before first
meeting the King” (Law 1997: 95). Other European reports (e.g., Captain
François d’Elbée) suggest that traders were required to first purchase their
slaves from the King and his entourage (including local chiefs) before trading
freely on the coast. This is clearly illustrated by the following report cited in
Law (1997: 100):
The King of Ardrah has obliged the Portugueze at Jacquin to pay him 35 sletias per slave
for custom, whereas it used to be but 17, he has likewise made Captain Heartsease buy
20 slaves off him (which are good for nothing) before he can mark one off anybody else,
and demands all the booges [i.e., cowries] for himself. But what’s worse talks of
obligeing him and all Captains that shall come there for the future to go to Ardrah and
there stay and make their trade.
It appears from this document that not only did the King require to be the first
contact of the Europeans, but he also levied taxes on the slaves. It is likely that a
similar tax applied to the African sellers as well (Law 1997: 97). Given that the
capital city of Allada is about 50 km inland, the requirement that Europeans had
to trade first with the King also implies that they first had to travel inland before
coming back on the coast to trade. This is possible because the King and his
entourage controlled the roads to the interior and could effectively block slave
supplies from the interior in period of crisis. (See e.g., Hoolwerf’s letter cited
above.) We can therefore conclude that the State of Allada monitored both the
trade on the coast and the supply of slaves. Consequently, this Kingdom more
than any other on the Slave Coast “provided an ordered environment and a set
of rules for carrying on business” (Eltis 2011: 275). A report from the 1720s
describes the subjects of the King of Allada in the following terms: “all of them
obsequious slaves, who dare not sell or buy anything without license” (Law
1997: 95).
I submit that an indirect consequence of this “ordered environment” is that
slave traders, Africans and Europeans alike, could not afford to fetch their
slaves beyond the limits of the Kingdom. Recall that the Kingdom of Allada is
sandwiched between Oyo and its slave ports to the east and Ƞɔtse to the west,
where the Gold Coast lies. Both zones are heavily involved in slave trade.
North of Allada was the competitor and later conqueror Kingdom of Dànxòmὲ.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
38 The agents of creole formation
Beyond Dànxòmὲ there were various Muslim kingdoms, which were less
favorable to slave trade (Law 2011). Given this geo-political situation,
organizing slave caravans without a “license” from outside the limits set by
the Kingdom exposed traders to severe sanctions from the King,
notwithstanding any other perils they might have faced en route. But even if
the traders could possibly still organize slave caravans, such an economic
enterprise would not have been appealing. It is unlikely that Allada was the
only kingdom in the region which laid taxes on slave trade and controlled
the slave routes. Accordingly, bringing back slaves from beyond the borders of
the Kingdom would have unnecessarily increased the risks and costs of
transporting the human cargoes. Prices would have become too high and the
commerce less lucrative.
According to Law (1994, 1997), the slave trade flourished in Allada partly
because prices were lower there than in competing neighboring kingdoms. My
contention is therefore that a significant portion of the people sold in Allada
were of Aja descent. In addition to warfare and political instability in the
interior, economic factors (e.g., the cost of a slave) contributed to making
unfortunate locals the prime targets and victims of the trade.
2.1.4 The origins of the slaves: some notes on their cultures and languages
It is important to remember that, compared to other regions in West Africa
involved in the slave trade (e.g., Gold Coast), “more people left the Bight of
Benin than any other region except West Central Africa” (Eltis 2011: 275).
Given the prominent role of Allada in this trade, this would mean that a
significant number of the slaves taken from the Bight of Benin were of Aja
descent, as noted above. For instance, Eltis (2011: 276) claims that “the
majority of slaves before 1725 would have been Gbe speakers.” The author
further suggests that the slaves embarked in Allada after this period were
probably “a mix of Gbe-speaking and Yoruba people” (Etis 2011: 276). In
the context of our discussion, this is an important finding. Even though West
Central Africa appears to have been the primary supplier of slaves throughout
the slave trade period (viz., during the seventeenth and eighteenth centuries),
the burst of Allada in this trade, the astonishing growth of slave exports (e.g.,
from 1,000 to 19,000 according to Eltis 2011), and the structure of its slave
economy suggests that large numbers of Aja people must have been shipped
together to the Americas. This conclusion is contrary to the common
assumption that the slaves came from diverse places in Africa and were
dispatched to diverse places in the Americas. A closer look at the political
and economic structures underlying the slave trade in different West African
locations (viz., Upper Guinea Coast, Gold Coast, Bight of Benin/Slave Coast,
Bight of Biafra, and West Central Africa) suggests otherwise. In the case at
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 39
20
This of course need not mean that under other ecological circumstances, the enslaved Africans
from West Central Bantu would not have a greater impact than the Kwa as well as the Aja
descent people (see, for instance, Thornton and Heywood 2007). It is precisely because of such
variations depending on the ecology of the colony that more work is needed in order to
determine the contributions of the different ethnic groups that came in contact in the Americas.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
40 The agents of creole formation
communities of the Slave Coast such as the Yoruba also commonly referred to
as the Nagos.
Some cultural traits of the Aja In previous sections, I argue that people of Aja
descent originally migrated from Oyo, a Yoruba land (Pazzi 1979, 1984, and
much related work). We may assume that there has been sustained contact
between these two ethnic groups. Indeed, various cultural traits are shared by
these communities, for instance, the Aja cult of Afa, which is a variant of the Ifa
cult among the Yoruba. As reported by Pazzi (1979: 50), the bokɔ-jìsá tradition
considers Ifε to be the spiritual center of the Afa as well as the place where
spirits will return after death, thus directly relating the Aja to Oyo. The Aja and
the Yoruba share a weekly calendar of four days for rituals to the ancestors as
well as for agricultural purposes, although the Aja have also adopted an
additional seven-day calendar for celebrating birth days. Finally, both the Aja
and the Yoruba practice circumcision (Pazzi 1979). All these traits point to a
common cultural heritage from Oyo, which probably also involves a common
linguistic heritage.
Other cultural traits of the Aja include the cults of Vòdú, of the royal
python Dàngbúi and of the related python Dàn, and of lightning Sò. These
cults (and variants thereof) are found across communities of Aja descent,
many of which have distinctive facial scarifications often identifying them
with their religion. Yet, another cultural aspect that is shared by people of
this ethnicity (and that was maintained in the Caribbean) is their common
naming system described below. As is well known in creolistics, children
born in this cultural area systematically get a day name, which varies
according to gender.
Female Male
It is clear from this brief description that although the Aja share some core
cultural practices that make their ethnic group relatively homogeneous and
distinct from other groups, they also share significant cultural practices with
other Kwa people (e.g., Yoruba). In the context of this book, it seems to be
justified to assume that such cultural homogeneity within the Aja group and
kinship with other Kwa people must have influenced significantly the
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 41
emergent creole cultures and languages. Such cultural and linguistic affinities
must have motivated these enslaved Africans to interact whenever possible
with one another and learn from one another while adjusting to their new
societies. An aspect of such interaction that is important in our study regards
language.
Some Aspects of Aja languages People of Aja descent speak the Gbe
languages. Speakers of these languages currently live in West Africa, in the
southern part of the Volta region in Ghana, the southern part of Togo, the
southern part of Benin, and in different localities of Ogun State and Lagos State
in Nigeria. Capo’s (1991) study on comparative phonology suggests that the
Gbe languages form a continuum from Nigeria to Ghana and involve five main
branches: Ewe, Gen, Aja, Fon, and Phla-Phéra, as illustrated below (Capo
1991: 15):
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
42 The agents of creole formation
short random list (adapted from Kluge 2000) below. For the purpose of
clarity, tone specifications as well as nasality were not included.21
Because the Gbe languages are isolating, they lack the typical noun class
systems found in Bantu. As a result, the roots listed in Table 2.2 are usually found
as is (i.e., with no other morphological modification) in sentences. While nouns
may involve more than one syllable, verbs are generally monosyllabic, except
when they derive from reduplication. In this regard, we observe that contrary to
the verbs ‘eat,’ ‘drink,’ and ‘pull,’ which are all monosyllabic and usually involve
the same roots in these languages, the verb ‘push’ is disyllabic: it derived from
reduplication of the verb roots tu/cui or sε which can be roughly translated as
‘displace.’ That all the varieties resort to the same word-formation process (viz.,
reduplication) in very specific cases like this one is indicative of the tight
relationship between these languages. In her lexical comparison, Kluge (2005:
22) concluded that lexical similarity of the entire Gbe cluster ranges between 64
percent and 73 percent, depending on the method of calculation.
The same pattern is observed in syntax. Here again, the Gbe languages show
very similar patterns summarized below using word order patterns in both the
noun phrase and the clause. Other aspects are detailed in Aboh (2004a) and
Aboh and Essegbey (2010). The data presented here are primarily based on
Gungbe, but when necessary I draw on data from other Gbe languages as well.
21
For a more representative list, the reader is referred to Kluge (2000, 2005). Kluge (2000)
provides a list of a hundred words based on the Swadesh list.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 43
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
44 The agents of creole formation
While these two options are found across Gbe, the languages differ regarding
whether the construction is sensitive to features such as [human], [animate],
[alienable], or whether the possessive markers are overtly realized or not.
Finally, most Gbe languages involve both a set of prepositions and
postnominal locative elements. The combination of these two leads to
complex locative expressions such as in (5), discussed in Ameka (2003) and
Aboh (2005a, 2010a, 2010b):
(5) Áyɔ̀ zé sìn sɔ́ n távò jí. [Gungbe]
Ayo take water PREP table ON/ABOVE
‘Ayo took (the) water from the table.’
With regard to the clause, the Gbe languages are generally assumed to be of
the SVO type. Indeed a simple declarative sentence in Gungbe can be
illustrated as in (6a), in which the aspect marker precedes the verb. Note,
however, from example (6b) that the equivalent sentence in Gengbe displays
a post-verbal aspect marker. Put together, we get the variation in (6c) which
again derives from language specificities as to which aspect marker may/must
precede and which may/must follow, as discussed in Aboh (2003a, 2004a,
2009b) and Aboh and Essegbey (2010):
(6) a. Mí nɔ̀ xɔ̀ kòkló. [Gungbe]
1PL HAB buy chicken
‘We usually bought/used to buy chicken(s).’
b. Mí plé-nà kòkló. [Gengbe]
1PL buy-HAB chicken
‘We usually bought/ used to buy chicken(s).’
c. Subject>(Aspect)>Verb>(Aspect)>Object>Adjunct
Another context where one finds variation across Gbe is the expression of
negation. As discussed in Aboh (2004a, 2010c), one finds three basic patterns
across Gbe: preverbal negation as in Gungbe (7a), sentence-final negation as in
Fongbe (7b), and simultaneous preverbal and sentence-final negation as in
Gengbe (7c).
(7) a. Mí má xɔ̀ kòkló. [Gungbe]
1PL NEG buy chicken
‘We did not buy any chicken.’
b. Mí xɔ̀ kòkló á. [Fongbe]
1PL buy chicken NEG
‘We did not buy any chicken.’
c. Mí mú xɔ̀ kòkló ò. [Gengbe]
1PL NEG buy chicken NEG
‘We did not buy any chicken.’
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 45
With regard to word order variation, it is important to observe that all Gbe
languages also display SOV order in very specific contexts. A case in point is
the expression of progressive aspect which requires the object to precede the
main verb. In addition, such OV constructions typically include a post-verbal
particle (Aboh 2003a, 2004a, 2009b; Aboh and Essegbey 2010).
(8) Mí ɖò kòkló xɔ̀ wὲ [Fongbe]
1PL ASP chicken buy PCL
‘We are buying (a) chicken.’
Such OV orders can be related to serial verb constructions (Aboh 2003a,
2009b), which are found across Gbe. An example of the instrument serials is
given in (9).
(9) Mí sà kòkló ɖòkpó ɖù sɔ̀ [Gungbe]
1PL sell chicken one eat yesterday
‘We sold one chicken yesterday.’
Word order variation also arises in interrogative (10a), focalized (10b), and
topicalized (10b) sentences:
(10) a. Étέ wὲ mí xɔ̀ ? [Gungbe]
What FOC 1PL buy
‘What did we buy?’
b. Kòkló wὲ mí xɔ̀ .
chicken FOC 1PL buy
‘We bought a chicken.’ (It’s a chicken that we bought.)
c. Kòkló lɔ́ yà mí xɔ̀ ὲ.
chicken DET TOP 1PL buy 3SG
‘As for the chicken, we bought it.’ (The chicken, we bought it)
In all these cases, the relevant category must front to the left edge of the
sentence, where it is typically bounded with a specific interrogative, focus, or
topic marker. As Aboh (2004a, 2004b, forthcoming) shows, these markers have
a very specific syntax that impacts discourse semantics in Gbe. For instance,
most Gbe languages exhibit verb focus constructions, known in creolistics and
generative syntax as ‘Verb/Predicate Clefting,’ in which the verb in the focus
position is repeated in its regular position in the rest of the sentence, as
illustrated by the Gungbe example in (11a). The Ewegbe example in (11b)
indicates that in some of these languages, the fronted verb must be reduplicated
(Ameka 1992, 2010).
(11) a. Xò wὲ yè xò è. [Gungbe]
beat FOC 3PL beat 3SG
‘They beat him/her.’
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
46 The agents of creole formation
b. ɸo-ɸo é wò ɸo é. [Ewegbe]
RED-beat FOC 3SG beat 3SG
‘S/he beat him/her.’ [Adapted from Ewegbe (Kwa); Ameka 1992: 12]
The properties presented here are summarized in the following table
showing word order patterns in Gbe.
Two questions arise from this brief description of the Gbe languages: (i) How
did these languages relate to the larger family of the Kwa languages of West
Africa? And (ii) How do the synchronic aspects of these languages as described
in Table 2.3 relate to Gbe varieties of the seventeenth and eighteenth centuries?
Gbe in the larger context of Kwa languages The first question is easy to
answer because the Gbe languages display morphosyntactic patterns that are
largely shared by other Kwa languages. Most Kwa languages have isolating
morphosyntax, although some languages show some degree of agglutination.
In any case, almost all the syntactic patterns in Table 2.3 are found across Kwa
languages (Aboh and Essegbey 2010). Kwa languages typically display focus
constructions of the type described in Gungbe, in which non-verbal elements
must be fronted to some clause peripheral position leaving a gap inside the
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 47
clause (as in 10b). Such focus constructions play a role in the formation of wh-
questions which show the same pattern (10a). On the other hand, verbal focus
typically results in a doubling strategy where the focused verb is fronted but
another token is found in its canonical position (11). These properties are not
unique to Kwa (see Koopman 1984; Mufwene 1987; Manfredi 1993; Aboh
2006c, 2007a, forthcoming; Aboh, Hartmann, and Zimmermann 2007, Aboh
and Dyakonova 2009), but they are found robustly in these languages.
Serial verb constructions are yet another salient peculiarity of this language
family. The same holds true of the periphrastic TMA marking and its
implications for word order (Aboh and Essegbey 2010). Indeed, a significant
number of Kwa languages of the Bight of Benin display a VO versus OV
alternation that is sensitive to aspect licensing (Déchaine and Manfredi 1997).
All these points lead to the conclusion that, morphosyntactically, Kwa
languages form a typologically homogeneous group. This means that an
enslaved African of Aja descent speaks a language that is typologically very
close to the languages of the enslaved people of Nago or Edo descent.
Accordingly, though it is often assumed that the slaves were of too diverse
origins, this might not be true when considering their linguistic profile and the
typological relations between the languages. Based on this, and taking into
account the geopolitics of the Slave Coast, it appears that while it might be
difficult for an enslaved person of Aja, Nago, or Edo descent to acquire a
Central Bantu language readily, this must not have been the case when such
enslaved people were confronted with a neighboring Kwa language.
Indeed, traditional considerations of linguistic diversity based on the vast
and diverse area from which the slaves were imported do not do justice to the
typological homogeneity of some subareas (and to the spread of some areal
features in sub-Saharan Africa). Such structural kinship suggests that in a
number of ways there were larger proportions of Africans sharing structural
properties than suggested by geography-based demographics alone. Also, it is
not excluded that the Kwa languages shared some of the same properties with
Bantu languages, such as the position of demonstratives cum determiners and
some nominal modifiers, verb compounding or light verb constructions which
appear to be similar to Kwa serial verb constructions. This, is an important
factor to keep in mind when it comes to the intra group communicative
strategies that the enslaved Africans might have adopted as well as the
learning strategies they might have developed in their attempts to acquire
European languages.
Indeed, the above specifics about how and where the slave trade was
conducted in West Africa suggest that the impact of speakers of Kwa
languages on the emergence of Haitian, Sranan, and Saramaccan could be
explained by the fact that the enslaved Kwa speakers might have entertained
the same learning hypotheses while acquiring the European colonial
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
48 The agents of creole formation
Modern Gbe versus ‘Old Gbe’ The high rate of lexical similarities among
current Gbe languages and their typological closeness to other Kwa languages
helps us address the vexing issue of the relation between what we can refer to as
a Modern Gbe language, say Gungbe, and its ancestor spoken in Allada in the
seventeenth century. Indeed, we can infer from the current strong similarities
among Gbe and between Gbe and other Kwa languages that whatever changes
they may have undergone over the past three hundred years or so are probably
not significant enough to dispute the assumption that they were just as related
during the time of the slave trade. In this regard, we have access to two old
documents only.
Using La Doctrina Christina (1658) and La Grammaire Abrégée (1730)
as references and framework, let’s consider certain aspects of seventeenth-
and eighteenth-century Gbe. La Doctrina Christina is a catechism published
by Spanish priests sent to Allada in the context of cooperation between the
King of Allada and the King of Spain (see Section 2.1). Because of the
obscure transcription system used by the Spanish monks to transcribe this
Gbe language, it is not yet possible to reconstruct all the lexical forms that
appear in this document. Yet, a cursory look reveals strong similarities
between that variety and Modern Gbe languages. For instance, three
words in Table 2.2 are transcribed in the document. In my presentation
below, the first line is the Spanish text followed by the Gbe translation as
written by the missionaries including the relevant lexical item in boldface.
The third line provides the same item in Gungbe preceding its translation in
parentheses.
(12) a. Buelue tus ojos misericordiosos [La Salve]
Dio nucu-bo-mi nunc-mito
nùkún (eye: Gungbe)
b. Los tres primeros pertenecen al honor de Dios [Los Mandamientos de la Ley
aton juegue teegue nudajoguana Vodu de Dios]
àtɔ̀ n (three: Gungbe)
c. La primera dar de comer al hãbrieto [Las obras de Misericordia]
Uguegue na-nu-ie-ru jobono
ɖù (eat: Gungbe)
Other lexical items found in this text include dasi ‘ɖàsìn: tears’, suru ‘súnù:
male person’; nubo ‘núgbó: true/truth’ as shown in E Vodu suru nubo the
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 49
translation of the Spanish text Dios y hombre verdadero.22 We can see from the
Gungbe equivalents of these words that they haven’t changed much. In a
related vein, a close look at the translations in (12) provides us with
significant information about the language. In the translation of (12a) we see
the italicized words bo, which corresponds to Gungbe verb bɔ̀ meaning ‘to
help’ or ‘to forgive.’ Next we see the first-person plural weak pronoun mi (mí in
Gungbe) as well as its possessive equivalent mito (mítɔ̀ n in Gungbe). The latter
is even more informative as it shows that in both seventeenth-century and
Modern Gbe, possessive pronouns are formed by combining a weak pronoun
and the possessive marker illustrated in (4b).
The translation of example (12c) is also interesting. Here we see the lexical
verb na (ná ‘to give’ in Gungbe) in a serial verb construction including the verb
ru (ɖù ‘eat’ in Gungbe). In this construction, the so-called ‘shared object’ is
nu-ie (nú-è, /thing this/ ‘this/the thing’). As we can see in these sequences the
noun precedes the determiner, just as it does in Modern Gbe. Finally, another
aspect of Gbe that stands out in the manuscript involves the use of a focus
marker in (13a–b). Though the transcription is not always systematic, the fact
that the Spanish authors transcribed the focus marker è with an accent indicates
that they understood its function as marker of emphasis.
(13) a. El tercero creer que es Hijo [Los Articulos de la Fè]
b. atoe un-dique Vi-to-è
three 1SG-believe child-POSS-FOC
‘the third is THE CHILD’
The examples in (14a–b) indicate that the focus marker attaches to wh-
phrases in wh-questions:
(14) a. Quien es Dios? [Sobre la Dotrina Christiana]
Menu-e Vodu?
who-FOC Vodu
‘Who is Vodu?’
b. Como es Dios?
Anuniqu-e Vodu?
how-FOC Vodu
‘How is Vodu?’23
22
It is interesting to realize that in these early texts, the term vodu ‘voodoo’ was used as translation
for God. This is evidence that this term acquired its negative connotations later only when
missionaries on the Slave Coast or in Haiti decided to make a sharp distinction between their
‘God,’ the warrant of their “mission civilisatrice,” and the ‘god’ of the “primitive” Africans to
be brought to light.
23
This wh-phrase here could be alternatively segmented as ‘anuni-que’, where ‘que’ represents
the focus marker.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
50 The agents of creole formation
Further work is needed to establish precisely which Gbe variety served for
the translation of La Doctrina Christina, but the facts discussed here show
that significant grammatical aspects of Gbe have not changed since the
seventeenth century. The second historical document at our disposal, La
Grammaire Abrégée (GA), further confirms this conclusion. This annex to
Voyage du Chevalier des Marchais, reproduced in 1730, represents a
pedagogical grammar. This can be determined from the description
provided by the author and translated here:24
Grammaire abrégée, an interview in French and in the languages of the Negroes of
Whydah, it is very useful for those who trade Negroes in that Kingdom, and for ship
surgeons who can communicate with the Negroes when they are sick. It includes what
can serve as a small dictionary. (Cited in Labouret and Rivet 1929: 31)
24
“Grammaire abrégée, on entretien en Langue Françoise et celles des Negres de Juda, très-utile à
ceux qui font le commerce des Noirs dans ce Royaume, et pour les Chirurgiens des Vaisseaux
pour interroger les Noirs lorsqu’ils sont malades. Ce qui peut servir pour composer un petit
Dictionnaire” (Labouret and Rivet 1929: 31).
25
Note that anam packs together the sequence wá ná mì in (modern) Fongbe/Gungbe. In normal
speech, however, the -i of the pronoun mi is dropped leading to the transcription found in the
Grammaire abrégée.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.1 Geopolitics on the Slave Coast: the case of Allada 51
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
52 The agents of creole formation
some extent the effect of the particular approximations of the target lan-
guage by speakers of typologically related languages at a significant early
stage of the emergence of the new vernacular.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.2 Plantation settlement colonies: Suriname and Haiti 53
Gengbe (literally, ‘you awake?’) do not have this literal meaning but rather
‘good morning/day.’ Finally, when we asked our consultants about the
genealogy of their clan, my fellow researcher, James Essegbey (from Ghana)
and I were astonished to hear that they identified themselves with three
different groups: Ji-vi-Allada ‘born-child-Allada’; Ji-vi-f ɔ̀ n ‘born-child-fɔ̀ n’;
and Ji-vi-xwèɖà ‘born-child-xwèɖà’, that is, three Gbe ethnic groups: Àlàdá,
Fɔ̀ n, and Xwèɖà related to three Kingdoms or chieftancies discussed in this
book: Allada, Dànxòmὲ , and Xwèɖà (Whyddah), respectively. Another related
clan we were told, lived in a nearby village called Dànxòmὲ . As we recounted
what we knew from oral history about our own origins as Aja people, their
expansion, and how the communities they mentioned related to each other, we
were thereafter referred to as Aja-nengre.
While all these observations aroused in me the feeling of a deep (sometime
painful) ancestral link with the people, there were also many aspects of their
culture which were completely foreign to me. A case in point is Saramaccan
woodcarving representing astonishingly geometrical and highly abstract
forms from a Gbe perspective. Indeed, Gbe traditional designs, paintings,
woodcarvings, and copper sculptures are generally naturalistic. So I wasn’t at
home after all! I’m not sure how to evaluate this anecdote from a historical
perspective, but two things seem clear to me:
(i) The Saramaccan people were not just New World’s cousins of the Gbe or
Aja. Though there are similarities, the Saramaccan certainly developed a
new culture integrating elements of their new ecology.
(ii) The Africans enslaved in Suriname included people of Aja origin. Some of
these contributed to the maroon society of the Saramaccan; and they seem
to have been numerically significant enough to have contributed a large
numbers of traits to the new Saramaccan culture.
Point (i) has been documented extensively by Price (1973, 1975, 1976, 1983,
1996, 2002, 2008), Price and Price (1991, 1999, 2003), and other scholars. I
will not discuss it here. Likewise, point (ii) has been documented in work by
Postma as reported in Arends (1989, 1995b), as well as by Smith (1987, 1996,
1999, 2009), who has focused on both demographic and linguistic factors. For
instance, according to Smith, Saramaccan has not only retained Gbe lexical
items but also selected functional items such as question words and a focus
marker from a variety of Fongbe. This finding has been corroborated by Aboh
(2006a, 2006b, 2007a), Migge (1998, 2003), Migge and Winford (2007, 2009),
and Winford (2007), among others.
Here, I am particularly interested in whether a Gbe (or any other African
language e.g., Kikongo) could have been retained within the Saramaccan
society? If so, why were the African languages eventually displaced by the
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
54 The agents of creole formation
emerging creole (Mufwene (2005a, 2005b, 2008)? I will come back to these
questions throughout the book.
60
50
40
Kwa
30
Bantu
20
10
0
1664 1680 1690
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.2 Plantation settlement colonies: Suriname and Haiti 55
change happens when the French colonies shifted from “crops such as tobacco
and cotton to sugar” (Singler 1996: 206), that is, a time when we can assume a
transition from small-scale plantations, called habitations in colonial French
and translated as homesteads in English, to large plantations, which required
much larger slave populations and with which the emergence of creoles has
been associated. This appears to have been a crucial factor still consistent with
Mufwene’s (1996, 2001) Founder Principle, according to which the founder
populations during the emergence of creoles were more likely to influence their
structural properties than late comers.
Debien (1974) provides specific figures on the population of different
regions of Saint-Domingue in the eighteenth century which indicate that the
colony included a significant number of both Congo and Kwa people. Below, I
report the figures as presented in Debien (1974) for the ethnic groups
originating from West Central Africa (i.e., Congo) and those from the Slave
Coast (i.e., Nago, Arada, Ibo, Mines, Adia/Adja, and Gold Coast). The figures
under the label “Other” include ethnic groups from other geographical areas.26
With regard to the South of Saint-Domingue, Debien (1974: 59) writes:
the Congos were not particularly numerous on the lists we consulted as of 1767. The
Ibos were then of about the same proportion, while there were a few Nagos and Aradas.
But soon after, the Congos were almost as many as the Creoles, while the Ibos were less
but the number of Aradas and Nagos increased.
The author further indicates that of the slaves whose origin was known, “the
Congos were 488, the Ibos 144, the Aradas 216 and the Nagos 203.” At this
stage, it is important to remember that the Aradas and Nagos share several
ethnolinguistic traits that justifies grouping them together (see Section 2.1.4).
This would mean that the community described by Debien in the South of
Saint-Domingue included two major groups: the Congos and the Kwa (i.e.,
Arada-Nago).
Regarding the northern part of the island, Debien (1974: 65) reports the
following figures dating from 1760 to 1785. The whole slave population
included 3,568 people. Of the 1,399 known origins, Debien provides the
following figures: Congos 633; Nagos 168; Arada 120; Ibos 55; Mines 43;
Adias/Aja 15; Gold Coast 14; Other 351.
Finally, Debien (1974: 67) presents the following figures dating from 1796
and 1797 for the whole island. The slave population included 14,167 people
including 7,445 Creoles, 391 Mixed and 63 whose origin was unknown. Of the
remaining 6,268 slaves of African origin, the author provides us with the
26
The records consulted by Debien listed such groups as: Aguias, Alemonans, Aras, Bambaras,
Bandias, Barbas, Bibis, Bobos, Cangas, Cap-Laous, Cotocolis, Coromantis/Cramentis,
Dambouans, Dias, Foëdas, Haoussas, Misérables, Mallés, Mandingues, Mocos,
Mondongues, Mozambiques, Poulars, Sénégals, Sosos, Tacouas, Tapas, Thiambas, Timbouts.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
56 The agents of creole formation
Congo
1800
1600 Nago
1400 Arada
1200
1000 Ibos
800
600 Mines
400 Aja
200
0 Gold Coast
South 1757 North 1760 Saint-
Other
1791 1785 Domingue
1796 and
1797
following figures for the ‘major’ ethnic groups including 4,792 people: Congos
1,651; Nagos 736; Aradas 544; Ibos 519; Mines 44; Adias/Aja 25; Other 1273.
Though the ethnic groups labeled “other” represent a sizable number, it appears
that these enslaved Africans have rather diverse origins (see note 26, which
gives details about these ethnic groups). Focusing therefore on Congos, Nagos,
Arada, Ibos, Mines, and Aja, the figure presented here can be put together as in
Figure 2.3.
What these figures show is that the Congo group remained dominant through
the second half of the eighteenth century. Similarly, the group labeled as
“Other” increased significantly. However, it is remarkable that the Kwa
group including the Nagos, Aradas, Ibos, Mines, Aja, and Gold Coast
represented the most significant group. Merging the figures of these
communities together, we obtain the following results: 573 Kwa in the south
(1757–1791), 415 Kwa in the north (1760–1785), and 1868 Kwa for the whole
island (1796–1797). Put together, we get the graph in Figure 2.4 which gives us
the contrast between the Congo, the Kwa, and Other ethnic groups during this
period.
This regrouping shows that the Kwa group was not only significant
throughout the second half of the eighteenth century but this population went
through an extraordinary growth to the point where Kwa people were more
numerous than the Congo in 1797. According to Debien (1974), the creole (i.e.,
the locally born) slaves were always in the majority. Put together, these
demographic factors point to three main groups which might have played a
significant role in structuring the emerging Haitian culture: the Creoles, the
Congo – who could have been more ethnically diverse than it looks on the
surface as they were captured further inland than on the Slave Coast (see
Section 2.1) – and the Kwa who came in compact and homogeneous
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.2 Plantation settlement colonies: Suriname and Haiti 57
2000
1800
1600
1400 Congo
1200
1000 Kwa
800 Other
600
400
200
0
South 1757 1791 North 1760 1785 Saint-Domingue
1796 and 1797
Figure 2.4 Contrasting the Congo and the Kwa in eighteenth-century Saint-
Domingue (adapted from Debien 1974)
27
This need not mean that the Congo were speakers of typologically different languages. To the
contrary, the languages spoken by these slaves must have been relatively close typologically. As
noted earlier, aspects of these languages (e.g., positioning of nominal modifiers, verb com-
pounds) are also similar to certain Kwa patterns. Put together, these groups are more likely to
converge in their approximations of the target language. Where these groups differ however, is
their internal dynamics and to what extent that dynamics affect the emerging creole culture. The
fact that the Kwa arrived in more compact groups seems to have favored them in establishing
some of their cultural and linguistic traits.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
58 The agents of creole formation
2.3 Conclusion
This chapter shows that, during the seventeenth and eighteenth centuries, the
Kingdom of Allada, on the Slave Coast, played an important role in the
triangular trade. In this period, the slave export in Allada peaked to an
average ranging between 15,000 and 19,000 slaves a year (e.g., Law 1991,
1994, 1997; Eltis 2011). Our findings indicate that socio-political and
economic factors in the Bight of Benin in the seventeenth and eighteenth
28
This dictionary was kindly brought to my attention by Michel DeGraff to whom I’m very
grateful.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
2.3 Conclusion 59
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:26:45, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.003
3 The emergence of creoles: a review of some
current hypotheses
Kwa speakers are generally assumed to have played a central role in the
formation of various creoles in the Caribbean. The conclusions of Chapter 2
suggest that in the case of Suriname and Haiti we can even be more specific and
single out slaves of Aja descent (i.e., speakers of Gbe languages), as critical
agents in the emergence of their respective creoles. While the speakers of Gbe
languages were numerically certainly not the dominant group throughout the
period of slavery, their contribution to the emerging culture in these colonies
may have been boosted by the fact that they shared a fairly homogeneous
cultural background. Chapter 2 shows, indeed, that the geopolitics of the
Slave Coast probably favored the enslavement of closely related ethnic
groups. The question then is how to integrate these findings in a theory of
language acquisition, language creation, and language change that can both
account for the emergence of the creoles in Haiti and in Suriname and provide
new insights into common phenomena of diachronic change.
The study of creole languages over the past thirty years or so has revolved
around this question, about which various genetic hypotheses have been
proposed. Three schools of thought are often referred to in the literature:
superstrativism, substrativism, and universalism. These schools differ
fundamentally regarding the particular mechanisms that produced a
particular Creole or Creole-type, though they all build on the central idea
that unsuccessful or incomplete L2 learning is the key to understanding the
so-called creolization process. While all theories of language change
presuppose language acquisition (either L1 acquisition or L2 acquisition),
noteworthy here is the fact that these genetic hypotheses presuppose failure
or interruption of acquisition: for one of these reasons, the enslaved
Africans could not acquire the target language completely. Proponents of
these hypotheses have evoked various reasons to explain this failure or
interruption (e.g., resistance to the oppressor or expressive need) but one
motive that is recurrent in the literature is the large discrepancy between the
European population providing the linguistic model and the enslaved
African population targeting the colonial language. The received wisdom
therefore is that in the case of creolization there was a break in language
60
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
The emergence of creoles 61
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
62 The emergence of creoles
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.1 Creoles as distant varieties of their lexifiers 63
they were spoken on the colonies. Chaudenson (2003: 448), for instance,
argues that French creoles result from “the unguided appropriation of
approximate varieties of [the colonial] French koiné.” Creole languages
emerged as a consequence of imperfectly acquired varieties of European
languages, hence the idea of ‘approximation.’ This theory presupposes that
even though creoles may differ from the target European varieties in various
respects, they share basic and fundamental similarities with these languages.
Positing a “français zéro,” which includes the total range of variation in
the French varieties spoken by native French speakers, Chaudenson (2003)
argues that French creoles developed from successive restructuring stages
corresponding to ‘approximations of approximations,’ which crystalize
variation in tendencies that are actually observable in varieties of non-
standard French. In order to accommodate the fact that French creoles are not
isomorphic with non-standard French varieties, Chaudenson (2003: 185)
further suggests that a possible distinction between non-standard French and
a French-based creole could be that the latter includes restructured patterns that
do not belong to “français zéro.” That is, French-based creoles developed new
patterns outside of the range of variation observable in “français zéro.” Such
exogenous patterns could be due to “self-regulating processes” that apply as the
language evolves or, when possible, to substrate transfer. Substrate transfer,
however, is subject to the following two principles:
(1) The principle of superstrate as gatekeeper
The superstrate (i.e., the target language) must provide favorable conditions
or contexts for substrate patterns to be integrated.
According to Chaudenson (2003: 199), this would mean that “les substrats
ne peuvent pas passer en force,” meaning that the substrate patterns cannot
force their way into the creole unless the grammar of the superstrate language
allows it. This view is compatible with the second principle: congruence.
(2) The principle of congruence
Those morphosyntactic patterns that are (partly) shared by the superstrate
and the substrate languages are more likely to be selected into the emerging
creole than conflicting patterns.
According to Chaudenson, the combination of these two principles can
explain the variation observed within the nominal domain in French-based
creoles and summarized in Table 3.1.
As this table shows, definiteness markers and demonstratives do not exhibit
the same distribution across French-based creoles. In Mauritian for instance,
the demonstrative precedes the noun while the definite marker follows it.
In Antillean creoles, the demonstrative follows the noun but precedes the
definite marker. Guadeloupian displays an alternative order, in which the
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
64 The emergence of creoles
1
See also Bobyleva (2013) for a comprehensive study of the nominal domain and the different
kinds of variation across creoles.
ce . . . (là), ce . . . (ci) ste . . . (la) . . . la (Quebec French and sa; sa . . . la; . . . la (Seychelle
Missouri French) and Mauritian)
cet . . . (là); cet . . . (ci) le/la . . . la; . . . la (Quebec French sa . . . a (Guyanese)
and Missouri French)
ce . . . (là); ce . . . (ci) ce . . . là, ce . . . ça (Louisiana . . . sa-la (New Louisiana)
French)
ces . . . (là); ces . . . (ci) ces . . . la; ces . . . ça (Louisiana . . . sa-a, sila-a (Haitian)
Acadian French)
ces . . . (là); ces . . . (ci) ces . . . ça là (Louisiana Acadian . . . la (most early Caribbean
French) creoles)
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.1 Creoles as distant varieties of their lexifiers 65
appear convincing in this case, it cannot account alone for all the relevant
morphosyntactic properties found in the creoles. Let’s consider the following
sentences in which the verb is focused by doubling. Such constructions are
found in many Caribbean creoles, including Saramaccan and Haitian.
(3) a. Si Kofi *(si) di mujee. [Saramaccan]
see Kofi see DET woman
‘Kofi SAW the woman.’
b. Se mache Bouki te mache, li pa te kouri [Haitian]
FOC walk Bouki TNS Walk 3SG NEG TNS run
‘Bouki had WALKED, not run.’ [DeGraff 2007: 113]
These examples, to which I return in Chapter 6, correspond to what,
according to Chaudenson (2003: 222), should be considered as cases of
substrate transfer, because they display patterns that are absent from the
lexifiers (e.g., French and English) but are attested in the substrates. As
shown in (4a) and (4b) verb focus with raising is impossible in French and
English (Glaude and Zribi-Hertz 2012), but not in Gungbe (Gbe), where such
constructions are grammatical (4c).1
(4) a. *C’est volé Bokassa a volé l’argent de l’Etat.
b. *It is stole Bokassa stole the State money.
c. Fìn Sàgbó fìn kwέ mítɔ̀ n. [Gungbe]
Steal Sagbo Steal money 1PL.POSS
‘Sagbo STOLE our money.’
Because Chaudenson does not discuss verb focus with doubling specifically,
it is not clear to me how one would account for the development of such
innovative patterns in his framework. Indeed, constructions such as in (3)
violate the two principles articulated in (1) and (2), because the superstrate
1
Note, however, that verb focus constructions with doubling are attested in Romance. Below are
some examples from Spanish and Portuguese:
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
66 The emergence of creoles
language in this case does not display verb focus with doubling. Accordingly,
they cannot provide a favorable context for such constructions to be
restructured, nor can they act as reinforcers in a context of congruence. Yet,
we can see from the Haitian example that verb raising with doubling is
combined with the French clefting device c’est ‘it is’ which is realized in the
creole as se. Comparing the examples in (3), (4), and (5), what we see is a
construction that clearly combines properties of verb focus in the Gbe (or Kwa)
languages with ingredients of focus in French (Romance), where the
highlighter c’est, realized as se, plays an important role (Glaude and Zribi-
Hertz 2012).
It appears that a strict superstratist hypothesis is inadequate. Indeed, while
the two language types are congruent regarding the semantics of focus
constructions, the examples in (3), (4), and (5) indicate that they display
divergent syntactic properties. French requires c’est ‘it is’, typical of a cleft
structure but this device is absent in Gbe. On the other hand, Gbe requires verb
fronting with doubling, a strategy excluded by French syntax.
Things get worse when we take a closer look at cases, such as illustrated in
Tables 3.1 and 3.2, which have been claimed to support the superstrate
influence hypothesis. As I argued above, the position of the definite
determiner (viz., pre-nominal versus post-nominal), can be accounted for on
the basis of the range of variation attested in “français zéro.” However, this
view does not extend easily to number markers. The latter consistently occur
pre-nominally in (non-standard) French varieties, but surface both pre-
nominally and post-nominally in French-based creoles. We must explain
where this asymmetry comes from.
A possible answer is that the number markers developed subsequent to the
definite determiner, which indirectly determines its position. But this does not
appear to be the right explanation since these two elements do not always
cluster on the same side of the head noun and may actually occur separately on
both sides. In addition, the number marker appears to precede or follow the
demonstrative and the definite marker across the creoles. This is the case, for
instance, in Mauritian and Antillean as indicated by the two rows in grey in
Table 3.1. The relevant patterns are repeated in (5).
(5) Definite marker vs. Demonstrative Number maker vs. Demonstrative
Mauritian sa N la sa ban N la
DEM N DEF DEM PL N DEF
Antillean N sa (l)a se N sa (l)a
N DEM DEF PL N DEM DEF
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.1 Creoles as distant varieties of their lexifiers 67
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
68 The emergence of creoles
(i) The Haitian number marker is specified for definiteness because of its
pronominal origin, as well as its usage in topic constructions where it
resumes a definite topic antecedent.
(ii) The Haitian definite determiner is specified for discourse-anaphora
because it developed from the French nominal demonstrative reinforcer
là, which encodes discourse deixis.
However, this hypothesis faces a number of empirical problems and
has theoretical implications which are inconsistent with theories of clause-
periphery constructions involving topicalization and focalization in particular.
An empirical problem with Chaudenson’s view is that the DPs including ïo/yo
are restricted to the subject position as in (7a), though the same constructions
can also occur in object position, as in (7b):3
(7) a. Zozo la yo va bientôt volé. [Ducœurjoly 1803: 336]
bird DET PL will soon fly
‘These/the birds will soon fly away.’
b. Mo voir zhomme la yo.
1SG see man DET PL
‘I saw the/these men.’
It is not clear to me how Chaudenson’s proposal can account for cases like
(7b) where a noun phrase marked by yo is in object position. One could argue
that the object position is a generalization from the pattern in the subject
position. Indeed, Chaudenson’s main point is that creoles did not literally
replicate the lexifier’s system but have extended the application of some of
the strategies that were allowed in “français zéro.”
However, the analysis proposed for the N-eux sequences in subject positions
is not without problems. Indeed, Chaudenson’s scenario according to which the
French resumptive strong pronoun eux in (6d) develops into a DP-internal
number marker ïo/yo in Haitian Creole (6a, 7a) implies that the Africans
learning French did not figure out that in a context like (6d) the resumptive
pronoun eux and the preceding noun phrase tous ces prêtres juifs-là do not form
a constituent and therefore should not necessarily be adjacent, as shown below
in (10). Under current syntactic theories as discussed in Chapter 6 and
references cited therein, a topic sentence like (6d) can be alternatively
represented as in (8a) or (8b) where the topic and the resumptive pronoun
belong to two different syntactic domains.
(8) a. . . . [TOPICP tous ces prêtres juifs-là [FOCP . . . . [IP eux complotèrent]]]
b. . . . [TOPICP tous ces prêtres juifs-là [FOCP eux [IP teux complotèrent]]]
3
I am immensely grateful to Michel DeGraff for sharing all these references with me.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.1 Creoles as distant varieties of their lexifiers 69
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
70 The emergence of creoles
spec Top¢
DP1
Top IP
Tous ces prêtres
juifs-là spec
DP2 complotèrent
eux
The hypothesis that the creators of Haitian have conflated DP1 and DP2 as
illustrated in (12) suggests that they failed to properly analyze topic
constructions as involving a topic (i.e., the antecedent) that is being resumed
by a tonic pronoun which itself can sometimes be focused. Such a conclusion
contradicts Chaudenson’s own view that the creole speakers did acquire the
koinés they were exposed to. Indeed, Haitian Creole does distinguish between
topic and focus constituents, as shown by (13):4
(13) [Tout prèt jwif sa yo] (ki la a), se [yo] ki te fè konplo
all priest Jew DEM PL who there DET FOC 3PL COMP TNS make plot
‘All these Jew priests (who are there), it’s them who plotted.’
In this example, the topic Tout prèt jwif sa yo in sentence-initial position and
the focused resumptive subject pronoun yo are separated by the intervening
relative clause ki la a ‘who are there.’
In addition, Chaudenson’s scenario implies that these speakers also failed to
make the right generalization about strong pronouns as a constituent with
internal structure similar to that of a noun phrase. Instead, the pronoun eux is
here wrongly analyzed as a phrasal clitic (probably a head of some sort) that can
attach to any preceding phonological phrase. One can defend such a scenario,
because it recalls certain Haitian forms such as lapolis ‘the-police,’ in which
the determiner is fused to the noun. It is often assumed or implied that such
forms emerged from an incorrect segmentation of the French noun phrase
involving a determiner followed by the noun. Accordingly, the representation
in (12) could be just another example of this confusion.
This hypothesis is however highly improbable. First, examples such as lapolis
sometimes form minimal pairs. In the case at hand, lapolis ‘police station’ is
4
This Haitian translation of sentence (9a) was kindly suggested to me by Michel DeGraff
(p.c. July 5, 2012).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.1 Creoles as distant varieties of their lexifiers 71
opposed to polis ‘police officer’, as explained by, for instance, Tagète and Urciolo
(1993). Thus, what at first sight looks like a mere confusion by the creators of
Haitian turns out to be a more subtle property of the lexicon where the presence of
the fused determiner is related to meaning distinction. The phenomenon is not
general and some fused forms such as nonm (< un homme) and zanfan (les enfants)
in Lesser Antillean Creoles do not involve minimal pairs. Given this variation, one
cannot immediately conclude that the creators of these forms were all confused by
the French determiner system. Actually, looking at the minimal pairs closely, it
appears that the creators of these forms acquired their syntax as well as (part of)
their semantics, though the process seems to have involved reanalysis. There is not
much work on the contrast lapolis versus polis exemplified in Haitian Creole.
However, if it turns out that there is a general tendency for words embedding the
determiner to denote generic referent (e.g., an institution, a kind) while words
lacking the determiner are primarily referential (e.g., pointing to individuals or
specific entities in discourse), then this contrast may relate to:
(i) structural properties of the noun phrase (i.e., DP): the two forms are not just
contrastive lexemes of the Haitian lexicon but rather involve different
structural make-up;
(ii) subtle semantics of French definite determiners.
Indeed, example (14) shows that French determiners can encode both
definiteness and generic reference. In this sentence, the noun phrase le vélo
‘the bicycle’ can receive generic reading (i.e., cycling), as in (14a), or a definite
reading (i.e., the bicycle being discussed), as in (14b):
(14) Tu sais quoi? Mon fils aime bien le vélo.
you know what? My son loves well the bicycle.
a. You know what? My boy loves bicycles in general (i.e., he likes biking).
b. You know what? My boy loves this bicycle (i.e., the one we are talking about
now).
What emerges clearly here is that the definite determiner allows usages
where it denotes generic entities. This function, it seems to me could have
well triggered the Haitian contrast between pairs like lapolis versus polis. More
study is needed before we confirm this analysis. But if we allow ourselves to
entertain this hypothesis, then we are led to conclude that the Haitian speakers
did have access to abstract properties of the noun phrase in French.
This conclusion is actually compatible with example (6c), repeated here for
convenience:
(6) c. Tous pères jouifs la ïo tous faire complot.
all father Jews DEM PL ALL make plot
‘All those Jew priests, they plotted together . . .’
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
72 The emergence of creoles
5
See also work by Deprez, Sleeman, and Guella (2011), who studied the acquisition of French
determiners by L2 Dutch and Arabic learners and showed that even though these learners
produce non-target-like sequences, they seemed to have acquired the basic semantic properties
of these determiners (e.g., their obligatoriness in argument position).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.2 Substrate influence as the main factor 73
cited there) and it does not function as a third person plural pronoun (the typical
source for nominal plural markers in several Atlantic creoles). Similarly,
Guadeloupian se arguably developed from the French demonstrative plural
ces ‘these/those’ or possibly the possessive pronoun ses (his, her plural) and
does not function as third person plural pronoun either (Déprez 2004, 2007).
Quite remarkably, the plural marker in Guadeloupian precedes the noun but the
latter precedes the definite marker which precedes the demonstrative (se N la
sa): an order that is completely alien to non-standard French varieties. The
plural markers in French-based creoles show varying syntactic properties
which do not always correlate with the syntactic properties of their etyma in
non-standard French. Thus, even though the morphological form of these
markers clearly shows that they evolved from French it is not always
possible to explain their syntactic behavior only on the basis of the syntax of
their French etyma.
I conclude that a theory of the development of creoles that only argues for
a successive restructuring process that takes the creoles away from their
lexifier without any other complementary explanation cannot account for all
the relevant facets of the evolution. For instance, our observations on number
marking in French creoles indicate that we need a more articulated theory
that takes into account the language contact ecology of the plantation, by
investigating the (possible) contribution of competing languages or varieties
present in the colony during the critical period of the emergence of the
creole. In addition, it appears that, even though principles (1) and (2)
might operate in certain cases, they do not seem to be major driving forces
in the development of the creole. Our conclusions about superstrativism
extend to their opposing alternatives, that is, theories of creolization that
only put forward the influence of the substrate languages spoken by the
slaves. In the following section, I discuss one extreme example of such
theories.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
74 The emergence of creoles
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.2 Substrate influence as the main factor 75
LA LB
/PHON/A /PHON/B
SYNA SYNB
SUBA SUBB
SEMA SEMB
SELA SELB
/PHON/A¢ /PHON/B¢
SYNA SYNB¢
RELEXIFICATION SUBA SUBB¢ TRANSLEXIFICATION
SEMA SEMB¢
SELA SELB¢
Receptor language
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
76 The emergence of creoles
properties such as those listed in (15). As the author argues, “this is because . . .
relexifiers either do not have access to this information or, if they do, they do
not use it in creating the new lexical entry (p. 17).” Thus, Lefebvre (1998: 16)
proposes the following description for relexification:
[phonology]j’or [Ø]
[semantic feature]k
[syntactic feature]n
Looking at the new lexical entry, we realize that this lexical item retains all
abstract properties in the “original lexical entry” (i.e., in the speaker’s native
tongue). Even the phonetic strings are parsed on the model of the learner’s
native grammar. In this representation, the newly created item is fundamentally
different from its source in the lexifier language because it is strictly
constrained by the native grammar of the speaker. Accordingly, the noun
phrases in (17), which contrast Haitian Creole with Gungbe, could represent
a perfect illustration of relexification.
(17) a. Krab mwen sa a yo [Haitian, Lefebvre 1998: 78]
b. àgásá cè éhè lɔ́ lέ [Gungbe]
crab 1SG-POSS DEM DET PL
‘these crabs of mine’
c. N>Poss>Dem>Det>Pl
As the sequence in (17c) indicates, Haitian Creole and Gungbe appear to
manifest the same word order in the noun phrase: the possessive pronoun, the
demonstrative, the determiner, and the number marker all occur post-
nominally, in this very order. Note also from these examples that both
Gungbe and Haitian Creole lack number inflections and gender specifications
on the head noun and the demonstrative.
In French, the determiner, the demonstrative, and the possessive pronoun are
pre-nominal and they are specified for gender and number (18a–c; 18a’–c’).
(18) a. le(s) crabe(s) a’ La table
DET.MASC.PL crab DET.FEM
table
b. ce(s) crabe(s) b’ Cette table
DEM.PL crab DEM table
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.2 Substrate influence as the main factor 77
6
For a detailed discussion of this topic, see Aboh and DeGraff (2014).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
78 The emergence of creoles
On the other hand, the parallels between Haitian Creole and the Gbe
languages as opposed to French appear to be readily compatible with the
Relexification Hypothesis. Acording to Lefebvre (1998: 78) the Haitian
“[+definite] determiner, the plural marker and the [+deictic] terms have
been created through relexification” whereby relexification is understood as
“a mental process that builds new lexical entries by copying the lexical entries
of an established lexicon and replacing their phonological representations with
representations derived from another language.” While this analysis might look
convincing at first glance, it comes with a number of drawbacks, which point to
both internal theoretical inconsistencies and empirical inadequacies. I discuss a
few here, though the reader should consult DeGraff (2002) for a very detailed
critical evaluation.
Regarding the conceptual inadequacy, an important point that is often
overlooked in the application of the Relexification Hypothesis to creole
languages is the fact that Muysken’s (1981a) model was designed to account
for the emergence of a new language, Media Lengua, within a bilingual
community, in which only two major languages were in contact. As Muysken
(1981a, 1997) makes clear, this contact language is not used for communication
outside the Quechua community (i.e., with other ethnic groups). The author
then argues that this variety came into existence probably due to a need within
the community to create a new identity rather than the sudden need to create a
makeshift language that would allow speakers of different linguistic
background to communicate (but see Shappeck 2011 for an alternative).
Muysken (1997: 376) also explains that “many Media Lengua speakers also
speak fluent Spanish.” The author further demonstrates that none of the social,
cultural, and political factors that are often evoked to account for the emergence
of pidgins and creoles applies to the emergence of Media Lengua.
So, this is a language variety that resulted from the contact of only two
languages, Quechua and Spanish, both of which the speakers spoke fluently.
This sociolinguistic setting makes perfect sense when one considers Muysken’s
(1981a) relexification model as a process that primarily involves abstract
lexical properties in the languages in contact. Therefore it is by definition a
mental process that is available to bilinguals or at least to speakers who have
acquired pieces of knowledge of abstract properties of a language that they are
learning or that they are confronted with in their daily lives. Crucially,
relexification so defined is not a theoretical model that can be applied to the
African second language learners of French in Haiti, who are assumed by
Lefebvre (1998) and (Plag 2011), to have no intuition whatsoever about the
target language.
In this regard, let us recall that most theories of creolization heavily rely on
the idea that these languages emerged as a result of restricted quantitative and
qualitative access to the target language. In the plantation context as envisaged
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.2 Substrate influence as the main factor 79
7
This view, which is shared widely within the creolist community, is actually misleading. A
cursory search in the following database www.marronnage.info/en/index.html reveals an
astonishing number of advertisements describing (runaway) slaves as speaking the colonial
language fluently. It is also important in the context of this discussion to realize that a
significant number of these slaves were multilingual. A case in point is the following
advertisement in Haiti “1767-02-25 – Francisque, étampé sur la joue droite V, parlant
français, espagnole & anglois, marqué de petite vérole, ayant le nez écrasé, trapu & de
moyenne taille, maron depuis la fin de janvier dernier. Ceux qui le reconnoîtront, sont priés
de le faire arrêter & d’en donner avis à M. Durse aîné, Capitaine de Navire, au Port-au-Prince,
ou à M. Laville, Négociant au Cap. Il y aura récompenses.” I thank Michel DeGraff for
making this gold mine available to me.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
80 The emergence of creoles
d. mɔ̀ pὲ ɖòkpó
priest one
‘one priest’
Haitian, on the other hand, does not have a proper indefinite marker in post-
nominal position. Instead, this language uses the numeral yon ‘one,’ which in
Romance and Germanic languages can also function as a specificity marker. In
addition, the Haitian numeral occurs in pre-nominal position, like other
numerals in this language.
(23) yon jwèt ti-moun lan [DeGraff 1992: 111]
One toy child DET
‘a toy of the child’
The contrast between Haitian and Gungbe suggests Table 3.3, which shows
the determiner-like elements in Gungbe and Haitian.
It appears from this table that where Gungbe has two dedicated morphemes
for encoding specificity and definiteness (in addition to numerals), Haitian has
only one. Clearly this is a problem because it suggests that the Gbe speakers
ended up not relexifying part of their grammar. In response to this issue,
Lefebvre (1998: 78) claims that “the substratum languages’ so-called
indefinite determiner was abandoned. Haitian developed its own indefinite
form.”
While one may want to hold this view, the question is why this strategy was
abandoned. Why, according to the relexificationist hypothesis, would speakers
sometimes fail to relexify a significant property of their native grammar? And if
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.2 Substrate influence as the main factor 81
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
82 The emergence of creoles
other nominal modifiers. The paradox is even more puzzling because in the
case of (17) what is retained is a grammatical item, while regarding (24) we
are dealing with lexical items which, as we know from studies of language
contact, can be borrowed in a language without necessarily affecting the
grammar of the recipient language. For example, the French adjective petit
‘small’ occurs pre-nominally when used attributively as in j’ai appelé un petit
enfant ‘I have called a small child.’ Yet, in Gungbe, one must say ùn yrɔ́ ví
petit ɖé ‘I call child small ind: I called a small child,’ with the borrowed
adjective petit following the noun. The alternative order *ùn yrɔ́ petit ví ɖé
‘I call small child ind: I called a small child’ is unacceptable. If relexification,
as defined by Lefebvre (1998), were the only process at work in the formation
of Haitian Creole, one would expect the Gbe speakers learning French in
seventeenth-century Haiti to produce the Gbe-type Noun-Adjective order
systematically, which they did not.
More generally, these facts show that Lefebvre’s (1998) style of analysis
must assume that the relexifiers were able to identify the relevant directionality
properties in both the superstrate and the substrate languages. But, as DeGraff
(2002: 359) already pointed out, “the determination of the relevant substrate-
superstrate overlap presupposes, at the very least, the knowledge necessary to
segment the relevant superstrate forms from spoken utterances, plus a fair
amount of abstract analysis.” That is, the Haitian relexifiers could not have
identified the directionality parameter associated with these nominal modifiers
(e.g., adjective, numeral, and quantifier) unless they were able to
identify their functions and their combinatorial properties according to the
morphosyntax of the noun phrase in French. This brings us back to criticisms
already formulated in the preceding discussion, viz., that Lefebvre’s (1998)
version of relexification contains a serious contradiction that cannot be
resolved in any principled way.
Because my criticisms here have focused on the version of relexification
proposed by Lefebvre (1998), the interested reader may want to know whether
Muysken’s (1981a) original formulation could offer a viable alternative to
Lefebvre (1998). The answer, it seems to me, is yes provided we accept that
some Africans were bilinguals and could speak the lexifier language fluently
(as I explain in Chapter 4). These speakers may unconsciously engage in
relexification, but the question we ought to ask is what impact such relexified
forms may have on the population level where other probably less proficient
African speakers produced equally competitive variants. There are probably
traces of relexification or similar mental processes in contact languages in
general (e.g., Schwartz and Sprouse 1996), and in creoles as well. However,
what matters for our discussion is that we cannot maintain (as is done in
Lefebvre 1998 and much related work) that relexification is the sole driving
force behind the development of creoles. Not only is such a view incompatible
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.3 The Language Bioprogram 83
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
84 The emergence of creoles
grammatical structure then yielding the macaronic jargon (viz., the pidgin).
Accordingly, there was a break of transmission between the homestead phase
and the plantation phase which in turn involves two interrelated stages:
(i) The pidgin phase: Inaccessibility to L2 and absence of a common lingua
franca within the slave community led to an extreme disintegration of the
target producing a macaronic jargon.
(ii) The creole phase: The pidgin is the ground zero of language degeneration.
It cannot serve as input for L1 acquisition. Consequently, children born on
the plantation during this period developed a new language relying exclu-
sively on their language instinct: the language bioprogram.
This theory implies that Hall’s (1962) pidgin-to-creole life cycle results
from two interrelated processes: language disintegration which leads to
pidginization, and linguistic reconstruction which is characterized by the
creation ab ovo of a new language by children. This makes clear why Bickerton
(1988a, 1988b, 1999) concludes that creoles are instances of catastrophic language
change with fundamentally different properties from older human languages that
are ‘normally’ transmitted from generation to generation. After analyzing the
language of trained apes, children under two, adults who were deprived of
language in their early years, and pidgins, Bickerton (1990a: 122) concludes that
the evidence just surveyed gives grounds for supposing that there is a mode of linguistic
expression that is quite separate from normal human language and is shared by four
classes of speakers: trained apes, children under two, adults who have been deprived of
language in early years, and speakers of pidgin. Since this mode emerged spontaneously
in the three human classes; since the second class includes all members of our species in
their earliest years; and since the fourth class potentially includes any person at any time,
we may regard the mode as a species characteristic.
8
This longstanding (but unmotivated) view is embraced by McWhorter (2001) in his article “The
world’s simplest grammars are creole grammars” and in related studies, where the label creole
stands for a cluster of linguistic traits defining a typological class (of simple languages),
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.3 The Language Bioprogram 85
(i) Non-basic morphemes: if lost, will not be reconstituted (e.g., most pre-
positions, derivational morphology).
(ii) Basic morphemes: if lost, must be reconstituted (e.g., TMA, locative
adpositions, articles).
Because creoles emerged from the conjunction of (i) and (ii), what appears to
be cross-creole resemblance is considered to be direct expressions of the human
capacity for language, what he calls the “Language Bioprogram” (Bickerton
1984). Features that are cited the most to illustrate cross-creole similarities
include, among others:
– Rigid SVO word order.
– Lack of inflectional morphology, which correlates with the use of free
morphemes to encode tense, mood, aspect specifications in the fixed order
Tense>Mood>Aspect (TMA).
– Near-absence of prepositions which is compensated by the use of serial
verbs.
– Absence of non-transparent wh-words: Many creoles display bi-partite wh-
phrases consisting of a wh- or question-word and a second term encoding
location, time, cause, person.
– The use of numeral ‘one’ as indefinite determiner and the development of the
definite determiner from a distal demonstrative derived from the superstrate.
– The use of third person plural pronoun as a nominal number marker.
As Bickerton makes clear, the fact that these properties sometimes cluster
together in creole languages does not necessarily imply that they are all direct
expressions of the Language Bioprogram. Some, for instance the usage of bi-
partite wh-words and the fixed order of TMA markers, may be specific to the
Language Bioprogram. Others, however, (e.g., the use of third person plural
pronoun to mark number) could relate to the individual history of the language
(e.g., the types of languages in contact).
If creoles were indeed the direct expression of the Language Bioprogram,
then these languages would present us with a special case of first language
acquisition in a situation of inadequate input. This way, the study of these
languages would inform us on the internal structure of what the Language
Bioprogram might be and how it might relate to other human cognitive
capacities. There have been several critical evaluations of Bickerton’s
Language Bioprogram, and the creole typology that it suggests. I will not
discuss these issues here and the reader is referred to Muysken and Smith
(1986), Newmeyer (1988), Thomason and Kaufman (1988), DeGraff (1999,
2003), Mufwene (1991, 1999, 2001, 2008), and the references cited therein.
presumably a cluster that is closer to UG (but see Muysken 1988; DeGraff 1999, 2001a, 2001b,
2003; Mufwene 2001, 2008 for a critique of this position).
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
86 The emergence of creoles
For the sake of this discussion, I only draw the reader’s attention to a few
theoretical and empirical inconsistencies of the Language Bioprogram
hypothesis, which suffice to show that it is inadequate in many respects.
These inadequacies in turn help better contextualize the proposal I develop in
this book.
Recall from previous discussion that Bickerton (1988a) hypothesizes that
creoles developed backward from the conjunction of language disintegration
(i.e., loss of grammatical properties leading to pidginization) and language
creation (i.e., reconstitution of a full-fledged grammar). Bickerton (1988a: 278)
further concludes that morphemes lost in the pidgin phase are of two types,
which are listed under (26) and (27):
(26) Non-basic morphemes: Morphemes that, if lost, will not be reconstituted
a. Grammatical gender
b. Number agreement
c. Bound verbal morphology, (e.g., conjugation forms)
d. Derivational morphology
e. Pronominal case and gender forms
f. Most prepositions
(27) Basic morphemes: Morphemes that, if lost, must be reconstituted
a. Articles
b. Tense/aspect/modality forms
c. Question words
d. A pluralizer
e. Pronouns for all persons and number
f. A general locative preposition
h. An irrealis complementizer
i. A relativizing particle
j. reflexives and reciprocals
A question that immediately comes to mind is what underlies the
distinction between (26) and (27). One explanation that Bickerton
provides is that while the properties in (27) are found across creole
languages, those in (26) are usually absent in the creoles he studied even if
the same properties are present in the source languages. According to
Bickerton, the properties in (27) represent the core (or unmarked)
properties of syntax. As he puts it:
presumably, the single universal syntax requires some minimal set of grammatical
morphemes for the discharge of the functions that it stipulates. If these are lost,
replacements are recruited from a limited set of lexical items. The fact that, across
creoles, similar lexical items with similar properties are recruited suggests that
there must be markedness in the inventory of (possible) lexical properties and that
creoles select unmarked options – but this possibility requires more extensive study.
(1988a: 279)
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.3 The Language Bioprogram 87
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
88 The emergence of creoles
(28) a. Bouki achte yon bèl machin. Men, machin tonbe bay Bouki pwoblèm.
B. buy a nice car but car fall give Bouki trouble
‘Bouki bought a nice car, but the car started to give trouble to Bouki.’
b. Chwal la te lage. Asefi Te al chache l toupatou . . .
horse DET ANT escape Asefi ANT go look.for 3SG everywhere
Men, chwal gen tan tounen nan poto pandan Asefi te pati.
but horse have time return in pole while Asefi ANT leave
‘The horse had escaped. Asefi had gone to look for it all over, but the horse
already returned to the pole while Asefi was gone.’
In comparison to articles in Romance and Germanic one expects definite
articles in creoles to occur upon second mention of a newly introduced referent
(as in 28a). Example (28b) further shows that anaphoric use of bare nouns is not
limited to newly introduced referents. In this sentence, the second mention of
the referent chwal is a bare noun even though the previously mentioned antecedent
is already a specific definite involving the determiner la comparable to articles in
Romance and Germanic. In these examples both the bare nouns machin (28a) and
chwal (28b) denote specific definite referents, yet they do not include a determiner.
Such bare nouns are not found in French, the lexifier of Haitian Creole.
Given that these languages, like many others (e.g., Kwa) do not necessarily
need articles to encode definiteness, the expected unmarked option for these
languages should be not to develop articles at all (contrary to evidence).
Therefore, one wonders why creoles, which, according to Bickerton, selected
unmarked specifications of the Language Bioprogram, would develop
redundant mechanisms for encoding definiteness: by means of bare nouns
and articles. Instead, it seems that the creoles develop these article-like
elements because they are found in the source languages where they perform
various discourse functions including marking topicality (cf. Aboh 2004c;
Aboh and DeGraff 2014).
A similar question can be raised with regard to (27c), (27e), (27f), (27i), and
(27j) for which it is not clear why they should be basic to language. With regard
to (27c), Aboh and Pfau (2011), for instance, show on the basis of signed and
spoken languages that question words are not necessary for asking questions.
This study shows that not all languages have question words of the type found
in Romance, Germanic, or Creole languages (i.e., the so-called wh-phrases).
A case in point is Indian Sign Language where content questions involve a
clause-final question particle that scopes over an associate phrase inside the
clause. In appropriate context, the associate may be left unpronounced leaving
the question particle only in clause-final position (see also Aboh, Pfau, and
Zeshan 2005). It appears from this study that what is required in Indian Sign
Language content questions is a sentential question particle that encodes the
interrogative force.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.3 The Language Bioprogram 89
Haitian provides us with a very interesting case against (27e), because it does
not display distinct forms for all persons and numbers. Indeed, the Haitian form
nou stands for both first and second-person plural. Notice that this case appears
more complex than the situation with the English form you which can be
considered a unique morphological form for second person, lacking a number
distinction, or the French forms il/ils versus elle/elles, for which the singular
and plural are homophonous except in contexts where liaison applies. This can
be illustrated with il/elle a ‘he/she has’ vs. ils/elles ont ‘they have’ in contrast
with il/elle mange ‘he/she eats’ vs. ils/elles mangent ‘they eat,’ which are all
pronounced the same way, the gender distinction notwithstanding. In Haitian
what gets blurred is the person distinction, a potentially more basic feature, if
one agrees with Bickerton.
Stipulation (27f) also appears to be unevenly distributed across languages. In
the Gbe languages, for instance, some locative expressions as in Súrù yì [xwé
gúdò] (lit. ‘Suru went house back’) only involve a complex noun phrase
consisting of the ground and its parts. No other general preposition is needed.
Examples like this clearly suggest that the overt grammatical marking of
location is not necessary for the expression of location. The same point arises
with property (27i). We can indeed imagine languages without relative clauses,
and therefore without relativizers, which is apparently true of some
Austronesian languages (e.g., Comrie 1989; Lyle Campbell 1996).
Finally, the Gbe languages present yet another counterevidence to (27j),
because they do not formally distinguish between reciprocals and reflexives in
all contexts. Accordingly, the following sentence is ambiguous between the
two readings in (29a) and (29b).
(29) Súrù kpó Àsíbá kpó mɔ̀ n yé-ɖè
Suru COORD Asiba COORD See 3PL-ɖè
a. ‘Suru and Asiba saw themselves (e.g., in a mirror).’
b. ‘Suru and Asiba saw each other (e.g., at a wedding).’
So, all in all, the only property that seems to hold across languages is
that in (27b): all languages display expressions of tense, aspect, and
modality. In creole studies, certain authors (including Bickerton) insist on
the fact that creole languages might be exceptional in exhibiting the order
[Tense–Mood–Aspect]. Unfortunately, such claims are made in total
ignorance of the typological literature. Indeed, it has been shown since the
1980s in creolistics (Muysken 1981b) and both in functionalist and
generativist literature (e.g., Foley and Van Valin 1984; Bybee 1985;
Hengeveld 1989; Pollock 1989; Cinque 1999) that the more complete
order Mood[proposition-oriented]–Tense–Mood[participant-oriented]–Aspect is
a universal tendency. So it can’t help to single this sequence out as a
typical property of creole languages as imposed by the Language
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
90 The emergence of creoles
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.3 The Language Bioprogram 91
if the driving force behind creolization is simply the extreme rarity of available
grammatical morphemes, and if the characteristic phenomena of creoles arise largely
as a result of compensatory mechanisms, we would expect to find that in normal
acquisition many creole-like features would make only sporadic and fleeting appear-
ances, if indeed they appeared at all.
Once again, the ultimate explanation falls back on restricted input, but this
leads to a contradiction, given that many subtle features of the target languages
found their ways into the creoles. For instance, the discussion on Haitian
adjectives in previous paragraphs shows clearly that Haitian speakers
acquired the morphosyntax and semantics of French adjectives. Accordingly,
it cannot be the case that these learners found themselves in situations of
“extreme rarity of available grammatical morphemes.”9
I will return to the issue of the input in subsequent chapters. With regard to
the present discussion, it seems to me important to show that the thesis, central
to the Language Bioprogram hypothesis, viz., that creole languages display
systematic similarities because they express unmarked (or default) strategies
made available by the human language capacity, is disputed by many empirical
facts. These inadequacies add to the conceptual flaws discussed in previous
paragraphs. As already observed by Muysken (1988), adpositions represent
obvious counterevidence to the disintegration–reconstitution view developed
in Bickerton’s work. Indeed, Van den Berg (2007) presents ample data
suggesting that adpositions never got lost in Sranan even though they were
reanalyzed. As is evident from Table 3.4, eighteenth-century Sranan already
9
Alleyne (1971) also pointed out that the presence in creoles of forms corresponding to inflected
items in the lexifiers (e.g., ded ‘die, dead’ and broko ‘break’ in English creoles) suggests that the
varieties to which the producers of creoles were exposed were not morphologically as impover-
ished as has traditionally been claimed.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
92 The emergence of creoles
10
To the best of my knowledge, Bickerton never discusses this asymmetry in his work.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 93
Dem-(PL)-Adj-N-(PL) Bahamian
(PL)-Dem-(PL)-(Adj)-N-(Adj) Kriyol, Tok Pisin, Seychelles, Mauritian
Dem/PL-Adj-N-PL-Dem Berbice Dutch, Sranan
(PL)-(Adj)-N-(PL)-(Adj)-(PL)-Dem Fa d’Ambu, Nubi
(Adj)-N-(Adj)-(Dem)-(PL)-(Dem) Haitian, Papiamentu
(PL)-Adj-N-Dem Sango
(Dem)-PL-N-Adj-(Dem) Lingala
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
94 The emergence of creoles
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 95
(30) S
NPsubj VP
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
96 The emergence of creoles
Processing
Stage procedure L2 processing Morphology Syntax
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 97
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
98 The emergence of creoles
This quote not only shows the conceptual flaws of this theory but also its
methodological shortcomings. Because I discussed some of the conceptual
drawbacks in previous paragraphs, I now point the reader only to the fact
that while the title of Plag (2008a, 2008b, 2009a, 2009b), “creole as
interlanguages,” suggests that the interlanguage hypothesis accounts for all
aspects of creoles, what the author actually means is that it accounts for some
aspects only. The question then is how to account for those aspects of creoles
which do not fall under the interlanguage generalization (e.g., the emergence of
TMA systems, see Plag 2008a, footnote 2). And concerning this, one should
logically ask whether the solutions or analyses that account for these other
aspects cannot also extend to those that the interlanguage hypothesis is claimed
to cover, thus making the whole interlanguage hypothesis redundant and
therefore undesirable.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 99
These questions bring me to the methodological point. The view the author
adopts in his series of columns in the Journal of Pidgin and Creole Languages
is that creoles arose as a consequence of radical SLA failure. This is so even
though there exists no clear evaluation of what that failure consists of and how
it relates to the target. This in turn raises the question of the target on the
plantation and how to evaluate such target a posteriori (see Chaudenson 2001,
2003, and much related work). Yet, the rationale entertained by this author is
that because we know that creoles emerged from SLA failure and because SLA
generally involves the development of interlanguages, we can deduce current
creole properties from those putative interlanguages that each African could
have developed on the plantation while trying desperately to acquire an
inaccessible target. But if the target language is inaccessible, as claimed by
some popular theories of creolization, and assumed by Plag, then it is pointless
to speak of an interlanguage, because there is no point of reference. Yet, Plag
compares modern creoles, outcomes of the interlanguages developed by the
learning populations on the plantations, to various instances of interlanguages
described in the SLA literature. This is not unproblematic though.
First, interlanguages are not full-fledged systems that can be taken wholesale
as entities of linguistic comparison. Interlanguages are transitional stages that
learners go through in the course of acquisition. So when we refer to
interlanguages, we actually refer to different stages of acquisition. These
stages are not necessarily arranged linearly (contrary to what is presented in
Table 3.6) and, more importantly, not all L2 learners necessarily go through
exactly the same stages. Because of this, comparing a creole to a learner’s
interlanguage amounts to comparing what DeGraff (1999) refers to as an
E-creole (the creole spoken at a community level) to a developing I-language
(i.e., the interlanguage that a learner X may produce at a given point in time
during L2 acquisition). The problem here is that interlanguages are unstable
systems by definition and, as such, they cannot be compared in a holistic way to
a communal linguistic system, which is variable inter-individually. This is so
even though it is perfectly possible to compare sub-modules of the developing
interlanguage to those of the target in order to evaluate or describe the learner’s
acquisition path and the possible learning hypotheses she might be
entertaining.
Second, studies of SLA have shown that target-deviant forms that learners
may develop in the course of acquisition are allowed by UG in the sense that
they are often found in some languages of the world, though these are (or might
be) absent in the target.12 Given this, that some supposedly creole structure may
resemble some interlanguage which in turn is found in some language of the
12
This should not be surprising given that these target-deviant structures are human linguistic
creations, which, just as natural languages, stem from the human language capacity.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
100 The emergence of creoles
NPsubj VP
Det N
Agreement between
Peter.3sg and V reflects a dog Matching between the article and the noun
S-procedure: Inter-phrasal reflects phrasal procedure, that is, Intra-
Information exchange phrasal information exchange.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 101
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
102 The emergence of creoles
13
Unfortunately, Plag’s columns include misrepresentations of Booij’s (2005) careful and detailed
discussion of the role of inflection and the distinction between contextual and inherent
morphologies. Booij’s analysis contradicts Plag’s claims on numerous counts.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 103
(34) DPA
DA FPA
les
APA NPA
tres hopitaux
grands
As the index ‘a’ (agreement) shows, agreement in (34) arises between the
NP that combines with the AP to form FP. The latter agrees with the D it
combines with to form DP. In current generative terms, adjectival phrases are
assumed to realize a specifier position. With this in mind, we can further say
that the agreement in the noun phrase in (32) arises as a consequence of a local
relation between the noun head and the adjectival modifier in its specifier, while
agreement between D and FP arises from a local relation between a head and its
complement. Given this very simple description, what first looks like a simple
matching process (as assumed in Plag 2008a, contrary to Booij 2005) is
actually a succession of locally construed agreement relations that involve a
head and a phrase in its specifier or a head and a phrase in its complement
domain. But because the relation is always between the head of a phrase and
that of another phrase, it necessarily involves inter-phrasal information
exchange regardless of whether these phrases belong together in a larger
phrase that could be a noun phrase, a relative clause, or a sentence. As a
matter of fact, note for instance that French (35a) and other Romance
languages display subject–verb agreement instantiating a local relation
between the auxiliary verb under T, the head of a tense phrase, and the noun
phrase in [spec TP] as illustrated in (35b):
(35) a. Les travailleurs ont manifesté dans la ville
the-PL worker-PL have.3PL demonstrated in the city
‘Workers have demonstrated in the city.’
b. TP
spec T’
DP
T VP
les travailleurs ont
manifeste dans la ville
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
104 The emergence of creoles
spec T¢
DP
T VP
Spec V¢
V DP
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 105
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
106 The emergence of creoles
Under this partial representation, the agreement between the relativized DP1
dee fisi and the relativizer dee inside the clause CP is a clear example of
contextual agreement signaled by contextual morphology (or inflection) on
the relativizer.14 Taking the description a step further, we can informally
describe the di/dee alternation in Saramaccan as in (41), with the change
being sensitive to the context in which the relativizer occurs, that is, in the
context of singular versus plural relativized noun head.
(41) REL[SG]= d + INFL[SG] ➔ d-i
REL[PL] = d+ INFL[PL] ➔d-ee
In a sense, the Saramaccan di/dee variation reminds us of French que/qui
alternation as in (42a–b), where the complementizer qui indicates subject
relatives as opposed to que, which marks object relatives. According to Rizzi
(1996), the que/qui alternation involves agreement. Note that in Standard
English the agreement occurs in the wh-phrase.
(42) a. L’homme que j’ai invité <l’homme>
the.man that.OBJ 1SG.have invited
‘The man whom I invited.’
b. L’homme qui a invité Jean
the.man that.SUBJ has invited John
‘The man who invited John.’
Complementizer agreement is rather common and one finds the direct
equivalents of the Saramaccan di/dee alternation in Gengbe and Ewegbe
(Western Gbe), where the relativizer is sensitive to number agreement.
Examples (43a-b) are from Gengbe:
(43) a. [[Axɔ́ má] [CP kè lè kplɔ̌ á jí á]] nyɔ́ . [Kangni 1989: 26]
Book REL BE.located table DET on DET be.good
‘The book that is on the table is good.’
b. [[Axɔ́ má] [CP kè-wó lè kplɔ̌ á jí á]] nyɔ́ .
book REL-PL BE.located table DET on DET good
‘The books that are on the table are good.’
Similar examples are also found in South Hollandic Dutch (Germanic),
Zwart (2006: 68):
(44) . . . Jonge-s die-e werk-e wil-le. [South Hollandic Dutch]
boy-PL REL-PL work-INF want-PL
‘ . . . Boys that want to work.’
14
See DeGraff (2009: 950–952) for a discussion of other examples of inter- and cross-clausal
dependencies that further illustrate the inconsistency of Plag’s analysis.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 107
These facts clearly show that, like non-creole languages, creoles exhibit
contextual inflectional morphemes that participate in agreement phenomena.
The fact that similar contextual inflectional morphemes are not typically found
in the verbal domain in creole languages need not be attributed to a putative
interlanguage origin, no more than the same cannot be claimed to account for
the extreme paucity of English verbal inflections (compared to other Germanic
languages) or for the total lack of verbal inflections in Gbe and Sinitic. Not only
do the facts presented here contradict Plag’s (2008a) claims about creoles in
general, they also show that Saramaccan speakers can select the right noun
phrase that agrees in number with the equivalent relativizer. This indicates that
their language does not reflect fossilization of some early-stage interlanguage.
If anything, these constructions would correspond to advanced stages, 3 or 4 in
Plag’s (2008a) scheme.15
A final empirical fact against the interlanguage view comes from
work on SLA which Plag (2008a, 2008b, 2009a, 2009b) ignored in his
columns, though this study is very relevant to his hypothesis. Klein and
Perdue (1997) conducted a longitudinal cross-linguistic study of second
language acquisition outside the classroom. The study involved a total of
40 Punjabi, Italian, Turkish Arabic, Spanish, and Finnish adult learners
of (varieties of) Dutch, English, French, German, and Swedish and lasted
30 months, three times longer than John Schumann’s (1978) study on
interlanguages. They observed that all the learners developed a
communicative system that:
– seemed to be determined by the interaction of a small number of organiza-
tional principles,
– was largely (though not totally) independent of the specifics of source and
target language organization,
– was simple, versatile and highly efficient for most communicative purposes.
(p. 303)
The authors refer to this system as the Basic Variety (BV). They further
observed that for about one third of the learners investigated, acquisition ended
on this structural level, some minor variation aside, they only increased their
lexical repertoire and learnt to make more fluent use of the BV (Klein and
Perdue 1997: 303).
With regard to the structural properties of the BV Klein and Perdue (1997:
332) give the following description:
15
One cannot resort to plain substrate influence or full access to account for this phenomenon
since the Saramaccan relative clause in (36) is structurally different from the Gbe relative
clauses in (42). See Aboh (2006a) for a discussion of the structure of the DP and relative clauses
in Gbe and Saramaccan.
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
108 The emergence of creoles
16
Klein and Perdue further argue that
One way the learner has of improving his or her expressive power is simply to enrich
vocabulary, especially by adding temporal adverbials, and to perfect technique on this instru-
ment. And about one-third of the 40 learners whose acquisition was investigated do exactly this:
they do not go beyond the BV, but they steadily improve it in these two respects.. . . Speakers of
the BV can say what they want to say about temporal relations – not what the structure of the
language forces them to say. (p. 323)
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.4 Plag’s interlanguage hypothesis 109
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
110 The emergence of creoles
3.5 Conclusion
This chapter has shown that monolithic theories that account for creole genesis
exclusively by invoking inheritance from the superstrate, substrate influence,
language universals, or fossilization of some early interlanguage stage all fail
both to explain the relevant properties of creole languages and to relate the
emergence of creoles to more relevant aspects of language acquisition. In a
way, what the theories discussed here aim at is a description of the type of
languages that learners may develop when they fail to acquire the target
language owing putatively to a break in its transmission, a restricted access
to it, a disrupted unguided learning, or, worse, possible psychological
impairment (as may be inferred from the interlanguage hypothesis). Because
these theories are formulated in holistic terms, they try to account for the so-
called creole system as a whole and therefore miss the important subtle
variations that may be observed across or within the modules of a language,
e.g., between morphology and syntax, or between DP and CP. Such variations
may be more indicative of how the different ingredients of the languages
in contact were combined into a coherent grammatical system: the speaker’s
I-language (or I-creole).
A point that is particularly damaging to the SLA failure hypothesis or any
variant thereof is that it hypothesizes that a stable contact language like a creole
can emerge only if:
(i) the population of learners is highly heterogeneous (as is supposed to have
been the case with the enslaved Africans), and
(ii) these learners have limited or no access to an adequate linguistic model
representing the target language.
A problem with this view is that not all contact languages that are regarded
as creoles or share superficial structural similarities with them arose in such
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
3.5 Conclusion 111
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
112 The emergence of creoles
Downloaded from http:/www.cambridge.org/core. University of Florida, on 03 Jan 2017 at 17:31:39, subject to the Cambridge Core terms of use,
available at http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139024167.004
4 Competition and selection
A question we were left with in Chapter 3 is why the mutineers and the
Polynesians stranded on Pitcairn Island developed Pitkern? Given the size of
this island and the structure of the clan it would have been possible for the
Polynesians to acquire the English vernacular, probably a koiné, spoken by the
mutineers, just as the latter could have shifted to Tubaian. Even more puzzling
is the fact that even though almost all the English speakers (including other
males on the island) had died within the first ten years of settlement – leaving
aging Sir John Adams as the only surviving linguistic model for English – the
remaining six women, mothering the twenty-three children, and who then
represented the absolute majority on the island, did not shift back to Tubaian.
Instead, this apparently tight community developed a new language from the
mutineers’ English vernacular. None of the sociological scenarios often evoked
in creolistics to uphold the exceptionalist view of creole development can
account for this case. While some authors tend to ignore the development of
Pitkern and similar cases, Thomason and Kaufman (1988: 149) argue on the
basis of Pitkern that “the evidence for the traditional view that a full-fledged
pidgin language, relexified or not, was the major direct source for every
plantation creole (not to mention Pitcairnese) strikes us, as it struck others,
as thin.”
Indeed, it appears that the development of Pitkern is comparable to similar
situations in the Caribbean where the Africans maintained the emerging creole
even though geopolitics and population factors allowed them to make a
different choice: for instance, to switch (back) to an African language of their
choice. A case in point is Saramaccan which continued to develop after
the Africans had escaped the plantations. As explained in Chapter 2, the
Saramaccans kept very many aspects of Gbe culture (e.g., the structure of the
clan and its chieftaincy, religious practices, and culinary practices), as well as
many basic vocabulary items. These are clear evidence that there must have
been enough Gbe speakers around to pass on this legacy to the Saramaccan
This chapter builds on Aboh (2009a), the results of which are presented again here and updated
according to new findings.
113
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
114 Competition and selection
language and society. The same could be said of Haiti after Independence. The
discussion in Chapter 2 shows that the plantation colonies of Suriname and
Haiti hosted significant groups of people of Aja descent who could have
developed an intra-group Gbe vernacular, which could have enabled them to
maintain a more obvious cultural link with home. In both cases, the leaders
of these communities clearly rebelled against the colonial power and one
could imagine that this rebellion could have translated into the adoption or
development of a new language different from the language of their oppressors.
But this apparently did not happen. Just as in the case of Pitkern, the runaway
slaves, and thus independent colonies, maintained (and further developed) a
language born in the early stages of the settlement.
Because of the important role the founder populations play in the development
and maintenance of the emerging community language, it appears that the
properties of this language will be determined fundamentally by the properties
of the languages in contact at the time of settlement. Assuming that people do not
hold meetings around a fire to plan for the creation of a new language (except
for criminal, war, or other secret activities), the only reasonable hypothesis we
can develop in this respect would be to say that the structural properties of the
emerging language will be shaped by learning hypotheses that speakers
entertained while trying to learn from one another during their interactions.
Mufwene (2001: 28) formalizes this hypothesis in terms of the Founder
Principle in order “to explain how structural features of creoles have been
predetermined to a large extent (though not exclusively) by characteristics of
the vernaculars spoken by the populations that founded the colonies in which
they developed.” According to Mufwene’s biological approach to language
evolution, the founders’ languages, which actually represent an ensemble of
idiolects, provide language learners with a pool of linguistic features on which
acquisition feeds. Going back to the case of Pitkern, the feature pool then
presumably consisted of the English vernaculars of the mutineers (which may
have led to an English koiné) as well as the indigenous language of the
Polynesians (viz., Tubaian). Adapting Mufwene’s (2001: 4) description of the
feature pool, this situation can be represented as in (1) (see also Ansaldo 2009:
100ff. for a similar discussion).
(1) English Scotts American St. Kitts Guernsey Tubaian
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
Competition and selection 115
As we can see from these two representations, the only difference between
(1), which applies to a learner who ended up speaking Pitkern, and (2), which
applies to a learner of French, is that (1) is more likely to allow a much wider
range of variation than (2) (Aboh and Ansaldo 2007; Ansaldo 2009 and
references cited therein). While the range of variation in (1) is set between
closely related varieties of English and the typologically distant Tubaian, that in
(2) lies within very closely related varieties of French only, notwithstanding
inter-idiolectal variation among speakers of the same dialect. This leads me to
conclude that the variation that is observed across speakers/learners is partly
determined by the feature pool they were exposed to. According to Mufwene
(2001: 5), the output (that is the developing idiolect) “represents variation in the
ways particular (combinations of) features were selected into the emergent
varieties.” What this means is that new varieties are not necessarily different
from their source languages with regard to their fundamental linguistic
ingredients (i.e., features) but rather with regard to which specific features
have been selected and how they have been recombined and/or modified by
learners. In this regard, Mufwene (2001: 5) concludes that “what makes the
new varieties restructured is not only the particular combinations of features
selected, often from different sources, into the new language varieties, but also
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
116 Competition and selection
the way in which the features themselves have been modified, ‘exapted’, to fit
into the new systems.”
A particularly interesting aspect of Mufwene’s approach to language contact
and language change that is evident from this very simple and brief description
is that language acquisition (whether first or second) implies contact of some
sort. This ‘contact’ happens in the mind of the speakers via the feature pool
which presents the learner with the input on which learning hypotheses are
made. Because of this, the output, that is, the developing idiolect, is partly
determined by the input in the feature pool and partly by hypotheses that the
learner entertained about that input. Clearly, this view (more than any other
theory of creolistics and language creation) provides us with an overarching
framework in which we can investigate variation within and across speakers,
learning hypotheses that speakers make, and how individual selections made
by speakers can propagate within a community as part of the communal
language. Most important of all is that this framework is universalist in that it
requires no exceptional hypothesis or device but the human language capacity
(as generally understood in modern linguistics) and how the ecology of the
speaker/learner may induce variation.
While Mufwene’s work draws a lot on the development of a language at the
population level (i.e., E-language), a major task that I undertake in this book
is to understand variation at birth. That is how learners/speakers select
linguistic features from the input they are exposed to (i.e., the feature pool
accessible to individual learners) and how in so doing they develop new
variants that in turn enter the feature pool. In order to conduct this research
in a reasonable way, we first need to know (or imagine) what linguistic
input an African learner/speaker could be exposed to on a plantation in
seventeenth- and eighteenth-century Suriname or Haiti. Based on the
findings on population factors in Chapter 2 and on discussion in previous
paragraphs, Section 4.1 proposes a scenario for the development of creoles
that is inspired by Mufwene’s approach. In Section 4.2, I further introduce the
reader to aspects of Mufwene (2002, 2005a, 2005b) that are relevant for the
theory of language contact and language acquisition that I put forward in this
book. In this regard, Section 4.3 presents the version of the competition and
selection model adopted here with regard to syntactic structures. Section 4.4
anticipates subsequent chapters in evaluating the proposed framework
on empirical data from the Suriname creoles, Saramaccan and Sranan.
Section 4.5 concludes the discussion.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 117
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
118 Competition and selection
L(koine) = Linguistic
model Cohorts of Africans
Restructuring
Simplification
Disintegration
Pidgin
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 119
It is indeed true that, after experiencing the middle passage, the Africans had
to acquire the local koiné as quickly as possible for daily survival. However, the
discussion in Chapter 2 teaches us that because of the way slavery was
structured on the Slave Coast, very many enslaved people of Aja descent
ended up on the plantations of especially Suriname and Haiti, where they
formed relatively homogeneous ethnolinguistic critical masses at different
stages of the development of these colonies. Another conclusion we reached
in Chapter 2 is that many of such enslaved Africans were probably
multilinguals, capable of speaking another language in addition to their
mother tongue. Put together, these considerations lead me to conclude that
typological linguistic similarity, related ethnicity, and cultural practices will
rather favor efforts from the Africans to learn each other’s languages rather
than switching immediately to the language of the colonist. Put more
concretely, I’m assuming that daily survival dictates that the enslaved
Africans not only learn the master’s language available to them as quickly as
possible but also to try, whenever possible, to immediately relate to and
communicate with other enslaved mates who they could identify as
belonging to the same regional ethnicity (e.g., the Aja descents from Tado to
ŋɔ́ tse, cf. Chapter 2). As I once put it during a creole meeting in Leipzig in 2005,
if an Eastern Gbe speaker meets with a Western Gbe speaker on a plantation, it
will probably take them less time to figure out how to say “à fɔ́ n gànjí
(Gungbe)?” or “ò fɔ́ n nywέn ɖé à (Ewegbe)?”/you wake.up well?/ that is
‘good morning’ in each other’s language rather than wondering about how
to say “how do you do,” or “comment allez-vous” in French (Aboh 2005a).
Note that the verb ‘wake.up’ is identical in these two Gbe languages.
Multilingualism was common in western Africa before the European
colonization and there is no good reason to assume that, after the middle
passage, the Africans suddenly lost this basic human ability. Indeed, the
closeness of the Gbe languages as discussed in Chapter 2 can only favor this
mutual learning. If we agree that some slaves (say some Eastern Gbe speakers)
may have tried to learn or accommodate with other Gbe varieties (e.g., Western
Gbe like Ewegbe), then there is no principled reason to assume that as normal
human beings endowed with language capacity they could not have tried the
same with less related languages such as Yoruba or even typologically different
languages such as Kikongo.
If we can allow ourselves to give a little bit of humanity back to these
enslaved Africans so as to entertain the situation just described (be it for a
tiny fraction of the population) then we are irremediably led to conclude that
the language contact situation on the plantations was probably more complex
than usually assumed. Actually, the commonly assumed founder scenario in
(3) appears very improbable. Various other scenarios come to mind, but the
basic hypothesis that I will be assuming in this book is described in (4) below.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
120 Competition and selection
La…….La+1……….Lb Sa……Sa+1…..Sb
In this representation, the series (La, La+1 . . . Lb) in the upper box to the left
represent the different colonial varieties or languages spoken by the colonists.
In the box to the right, the series (Sa, Sa+1 . . ..Sb) represent the African substrate
languages spoken on the plantation. The middle boxes represent koiné-like
varieties or vernaculars that must have arisen from language contact within
the colonist community and the enslaved Africans, respectively. As indicated
by the labels La vs. Lb and Sa vs. Sb, different sub-communities may have
developed different vernaculars. Assuming a process of competition and
selection, where some features are selected over others, this development
continues until stage 0 which involves levelled varieties or vernaculars in
both the colonist community and the enslaved African community.1 For
convenience sake, I refer to these varieties as plantation varieties.
As also indicated by the bold, vertical broken line labeled “contact
situations,” I assume that contacts between the colonist community and the
enslaved African community triggered mutual language learning which gave
rise to a plantation contact variety at stage 0. What this means is that
competition and selection between the lexifier and substrate varieties was
often mediated in the nascent plantation contact variety, which would evolve
into the creole as a communal vernacular. In synchronic terms therefore, the
developmental scenario sketched in (4) corresponds to a potentially universal
situation in the Caribbean where a local acrolectal variety evolved hand in hand
with the creole (typically associated with the basilect, though it is a continuum),
with the African languages becoming either extinct or relegated to very specific
symbolic functions in religion, ancestral practices, plant names, etc.
1
As extensively discussed in DeGraff (2002, 2009), children and young learners are likely to be
the crucial agents in the homogenization process of idiolects.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 121
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
122 Competition and selection
2
“Si tous les proprietaries d’esclaves, et même ceux qui n’en possèdent pas . . . connaissaient cette
langue, il n’y aurait pas toutes ces insultes, ruhinas, prejudices, vols, morts, et finalement
beaucoup de cas atroces.”
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 123
3
I also refer the reader to De Castro (2002), where one finds Fongbe equivalents of the Mina-Jeje
examples. Actually, this book is a gold mine for those interested in investigating the language
spoken by certain slaves.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
124 Competition and selection
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 125
late seventeenth and early eighteenth centuries was much more complex than
usually assumed in creolistics.4 The historical facts reviewed in this book
strongly suggest that such contact situation involves speakers of both lexifier
and substrate vernaculars learning from one another, though for different
motives. While plantation owners’ interests in learning the African
vernaculars had to do with proper management and strict control of an ever-
growing enslaved population, the Africans were probably motivated by
integration or bonding among those originating within the same parts of
Africa. This of course implies that not all Africans were native speakers of
such African vernaculars or related languages and we must be prepared to
admit that not all the slaves engaged in learning the dominant African
languages. Instead, it is likely that several members of the African population
targeted the European standard only.
Under this understanding, it appears logical to conclude that the creole
emerged as a mediation language between the plantation communities where
daily interactions are ‘negotiated’: a contact language. As indicated by DeGraff
(2001a: 251) quoting Schuchardt, “[T]he slaves spoke the creole not only with
the Whites but also among themselves while their mother tongue was still in
existence, the latter being moreover constantly revived to some extent by the
continual immigration from Africa.” From the perspective of the colonists, the
creole therefore represents the language of efficient and orderly management of
the enslaved population that guarantees a flourishing business: sugar and
tobacco production or gold extraction. For (part of) the Africans, however, it
represents the neutral language (just like official languages in modern Africa)
that guarantees successful daily interaction with everyone on the colony. Yet,
this language was probably not used for other survival purposes such as
planning an escape or a rebellion. While this could have been the status quo
in the early years, it appears that as the creole grew in functionality and
expressive power, competition between the lexifier vernaculars, the nascent
creole, and the African vernaculars reduced the latter to symbolic functions
only: religion, magic, plant names, etc.
Once we adopt this perspective, it becomes clear that the question we should
be asking ourselves is not why the African languages eventually disappeared
from the public domain on the colonies, but why the creole was so successful
as to expand in all such domains and become the language of acculturation
(also known in creolistics as ‘seasoning’) and upbringing. My answer to this
question is that the creole won the competition because of its socio-economic
and political power. Consistent with Mufwene’s competition and selection
4
Though there are good reasons to do so, I refrain from overgeneralizing these finding to all
plantation settings because I believe (like Mufwene 2001ff.) that each creole has to be considered
in the local context in which it emerged.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
126 Competition and selection
model therefore, the fate of the African languages in the plantation colonies is
strikingly similar to that of certain endangered languages in current sub-
Saharan Africa (Mufwene 2008: chapter 11).
Another advantage of the model put forth here is that there is no need for a
pidgin stage in the evolution of a creole (or a contact language). In this regard, it
is important to realize that the emergence of a ‘macaronic pidgin’ prior to the
creole (as has been postulated by some authors) would simply cause the
relevant socio-economic system to collapse, probably hindering the success
of the plantation industry itself. Recall, for instance, that one motivation put
forward by De Antonio Da Costa Peixoto to promote his description Lingoa
geral de Mina was precisely that knowing this language would prevent “insults,
vandalisms, financial losses, theft, death, and finally many more atrocious
cases.” This indicates that managing the plantation required the existence
of an elaborate and efficient medium of communication that is known to at
least most participants. However, no ‘macaronic pidgin’ can enable elaborate
successful communication, where every single agent expresses her intentions
or thoughts only embryonically and in any conceivable manner (with no
guarantee that the addressee understands the message). Absence of
communication, however, means inefficient organization of labor and low
production, which itself would translate into loss of investment and bankruptcy.
Assuming that the plantations were comparable to military regiments, a
good example that the functioning of such regiments requires efficient
communication is the so-called ‘petit-nègre’ or ‘français tirailleurs’ which,
even though it is inaccurately associated with African soldiers in the French
army, was actually a colonial instrument developed and enhanced by the
French colonial administration hoping to accomplish easy communication
with African soldiers recruited from their various colonies. Any serious
scientific work on this variety can easily show that during the First and
Second World Wars the French military officials did not give the enrolled
Africans the chance to learn the language before they were sent to the front.
Instead, what they thought was needed was an efficient medium of
communication that the African recruits could easily parse. The ‘petit-
nègre’ was designed to this purpose. (See van den Avenne 2005 for an
elaborate discussion.) In the context of colonial settlements in the New
World, I surmise that creole fulfilled a similar function and was used to that
end. The following quote from Pelleprat, cited in Aboh and DeGraff
(forthcoming) illustrates this:
We wait until they learn French before we start evangelizing them. It is French that they
try to learn as soon as they can, in order to communicate with their masters, on whom
they depend for all their needs. We adapt ourselves to their mode of speaking. They
generally use the infinitive form of the verb [instead of the inflected forms] . . . adding a
word to indicate the future or the past. . . . With this way of speaking, we make them
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 127
understand all that we teach them. This is the method we use at the beginning of
our teaching . . . Death won’t care to wait until they learn French. (Pelleprat 1655
[1965: 30–31], my translation)
Even though the first two sentences of this quote give the impression that
the instructors expect the Africans to ultimately learn French, this stands in
immediate contradiction with their evangelization mission. Indeed, immediate
and efficient conversion into Christianity forces missionaries to use the creole
as means of instruction/indoctrination. This linguistic choice is not only
assumed to be the best possible option that proves to be very efficient in this
context, but it is further motivated by the then common racist belief that it was
hopeless to try to teach proper French to people they considered to be primitive.
Because the missionaries considered Africans to be inferior humans, they also
assumed that the slaves had not evolved the necessary cognitive capacity to
master an allegedly refined and civilized language such as French. As the
author concludes, “Death won’t care to wait until they learn French.” What is
important for our discussion here, however, is that the creole being the
language of acculturation/indoctrination, must have been used fluently, if not
natively, by the White colonist community as well (more on this below).5
Putting all this together, I argue, based on the model in (4) and on the
historical facts revealed in this book, that creole languages were plantation
(or colony) creations, that is, vernaculars used by both colonists and the
enslaved Africans alike because they were efficient means of communication
in the relevant contact settings. The latter guaranteed economic growth and
political stability for the colonial power while providing the enslaved
population with some form of cohesion. This reason, I submit, explains why
the creole was retained as a vernacular even in Maroon societies, as in the
Saramaccan communities of Suriname where it appears that Gbe and Kikongo
could have vied for this ethnographic function, as they had critical masses of
speakers.
For this ecological view of the development and expansion of the creole due to
its function on the plantation to be complete, I now need to show that colonists
were indeed (native) speakers of the emerging creole. The point just made may
horrify some students of creole languages who adhere to exceptionalist
developmental scenarios that assume a pidgin-like stage in the development of
creoles. But once again, historical facts persistently point in the direction that I’m
proposing in this book. The above quote from Pelleprat already showed that
in Haiti, missionaries used the creole as the acculturation/indoctrination
5
Fattier (1996) discusses La passion de notre seigneur selon St Jean, a document written for
missionary purposes and which she shows to involve ‘hybrid’ lexical properties that are attested
in various French-based creoles in the Caribbean in the seventeenth and eighteenth centuries.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
128 Competition and selection
language.6 Another significant example illustrating this point comes from the
following official declarations in eighteenth-century Haiti in creole signed by
French officials Léger-Félicité Sonthonax, Charles Victoire Emmanuel Leclerc,
and Napoléon Bonaparte and reported in Denis (1935). The texts are quoted in
their original form, followed by my translation.7
Au Nom du Gouvernement Français: Liberté, Fraternité.
Proclamation.
Les commissaire v’lé et ordonné que tout ça yo sorti di icy soit affiché tout par tout
dans les communes, dans les habitations, dans les camps de la Colonie. Commissaire
recommendé à tous les officiers, que yo faire exécuté ça yo sorti recommende dans
proclamation ci la la.
Le Priésident de la Commission: Sonthonax
Proclamation là, li tiré dans registre consuls la République
Paris, 17 Brimer, an 10 Répibilique francé, yon et indivisible
Qui ca vous tout yé, qui couleur vous yé, qui côté papa zote vini, nous pas gardé ça:
nous savé tan seleman que zote tout libre, que zote tout égal, douvant bon Dieu et dans
zyé la Répiblique. Dans tan révolution, la France voir tout plein misère, dans la même
que tout monde te fere la guerre contre Français. Français levé les ens contre les otes.
Mes jordi là tout fini, tout fere paix, tout embrassé Français; tout Français zami; tout
hémé gouverneman, tout obéi li . . .
Signé: Bonaparte
Voyez pour zote. Zote à voir que li vélé resté libre. Li pas vélé ôté liberté à yo que yo
gagné en combattant, et que li va mainteni li de tout pouvoir à li. Li va mainteni
commerce et culture, parceque zote doit conné que sans ça, colonie-ci pas cable
prospéré . . .
Général en chef qui vini pour gouverné tout la Colonie: Leclerc
In the name of the French Government: Freedom and fraternity.
Declaration
The commissaries want and order that all that comes out of this commission be
announced and posted everywhere in the communes, plantations and camps of the
Colony. The commissary recommends that all government officials see to it that all
decisions made in this declaration be executed.
The President of the Committee: Sonthonax
The declaration is taken from the governing rules of the Republic.
Paris, 17 Brumaire, year 10 of the French Republic, one and indivisible.
Whoever you are, whatever your skin color, wherever your ancestors are from, that
does not matter to us. We only know that you are all free and all equal before God and
before the Republic. During the Revolution, France experienced a lot of suffering
because every other country fought against the French. The French were fighting each
other. But today, all of that is over. All people have made peace. All people have
6
Various manuals including a list of lexical items and sentence translations (e.g., Ducœurjoly
1803) were also published to inform potential migrants. The existence of such manuals clearly
indicates that Europeans were encouraged to learn the Creole.
7
I’m immensely grateful to Michel DeGraff for sharing all these precious documents with me and
for helping me with the translations.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 129
embraced France. All the French people are friends. They all love and obey the French
government.
Signed: Bonaparte
See for yourself. He wants the Blacks to be free. He does not want to take away the
freedom you gained through battle, a freedom that he wants to protect with all his power.
He will develop commerce and culture because you must know that without these, this
colony cannot become prosperous.
The General who has come to govern the colony: Leclerc
As is clear from the translations, these texts were signed by French officials
of the highest rank (including Napoléon Bonaparte) for all inhabitants of the
colony and not just for the newly freed Africans and their descendants. But
for this to be the case, and for Napoléon Bonaparte to have accepted to have
his declaration translated in the creole, it must have been the case that (i)
government officials were convinced that the creole was far more practical
than standard French, and (ii) that a large proportion of the colony was fluent in
the creole. In such a context, it is not unreasonable to assume that a French
version of the same declaration was also distributed on the plantations. Thus in
addition to showing that the Haitian colony was multilingual, these documents
represent very strong evidence that the emerging creole was used by most
agents in the colony and that it could be used in official contexts involving
public declarations, political statements, and religious instruction or
indoctrination. As such the creole represents a legitimate target for children
and bozal slaves, as well as European newcomers, and an alternative to French
for the inhabitants of the colony.
As the following excerpt from Long Edward (1774: 426) shows, one finds a
similar situation in Jamaica where both White Creoles and Africans spoke the
creole natively or as L2:8
The Africans speak their respective dialects, with some mixture of broken English. The
language of the Creoles [i.e., the local Whites] is bad English, larded with the Guiney
dialect, owing to their adopting the African words, in order to make themselves under-
stood by the imported slaves; which they find much easier than teaching these strangers
to learn English. The better sort are very fond of improving their language, by catching
at any hard word that the Whites happen to let fall in their hearing; and they alter and
misapply it in a strange manner . . . The Negroes seem very fond of reduplications, to
express a greater or less quantity of anything; as walky-walky, talky-talky, washy-
washy . . . and so forth. In their conversations, they confound all the moods, tenses,
cases, and conjugations, without mercy; for example; I surprise (for I am surprised); me
glad for see you (pro, I am glad to see you) . . . This sort of gibberish likewise infects
many of the white Creoles, who learn it from their nurses in infancy, and meet with much
difficulty, as they advance in years, to shake it entirely off, and express themselves with
correctness.
8
I thank Joseph Farquharson for bringing this reference to my attention.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
130 Competition and selection
This text strongly supports the scenario advocated in (4). Here we learn that
the Africans could speak their respective languages but also that the Creole
colonists tried to acquire such languages because it was easier for them to use
fragments of these languages to guarantee orderly and efficient management
rather than expecting the slaves to acquire English.9 Yet, the text also shows
that the Africans, though they could speak their own languages, were eager to
learn the target language adequately. This, I think, is a devastating blow to all
exceptionalist theories such as the pidgin-to-creole cycle, the bioprogram and
especially, the interlanguage hypothesis, which we now see can simply not hold
on historical grounds. Finally, the text confirms the conclusion already reached
in previous sections that, notwithstanding segregating rules, some colonists not
only used the emerging creole but also spoke it natively due to their intimate
contacts with the slaves who nursed them as children.
The following text from the unknown author of A short journey in the West
Indies (cf. Robert Charles Dallas 1790) is rather clear:
There was a little urchin, about seven years old [my emphasis], who constantly stunne
my ears with – “me wantee crab, me wantee crab” – to stop his noise, I took up the first
that came to my hand and gave it to him; but he immediately examined it, and finding it
had no eggs, he roared out, “Him no hab egg, him bleu maugre to hell, me no wantee
man crab, me wantee woman crab” – so he had heard his mother and the negroes
distinguish the sexes of crabs. I would have taught the little scoundrel better manners,
but his mother called him to her, kissed him a dozen times, and picked him out the best
woman crab.
Here again is a passage where we see that children pick up the emerging
creole and the plantation vernacular English from birth, a case of bilingual first
language acquisition. Additional evidence for this comes from breast feeding, a
practice that Long (1774: 276) condemns firmly when he writes
While I render all due praise to the Creole ladies for their many amiable qualities,
impartiality forbids me to suppress what is highly to their discredit; I mean, their
disdaining to suckle their own helpless offspring! They give them up to a Negroe or
Mulatto wet nurse.
Clearly there were intimate relations between the White children and the
Africans which necessarily involved linguistic exchange and language
acquisition. This is further supported by the fact that, instead of “teaching the
child better manners” both in language and behavior, as the visitor was hoping
9
A similar observation is made by Christian Oldendorp, cited in Holm (1988: 18): “In the West
Indies, the European languages tend to deviate to an extreme extent. For the most part, only those
people who learned to speak them in Europe can talk the pure European form of the language. On
the other hand, the people who were born here – the Crioles – do not speak the same kind of
language. They change it more or less; they employ words taken from elsewhere, arising from the
collision of the people of many nations.”
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.1 Competition and selection: setting the stage 131
for, the mother showed him love. Such an attitude clearly indicates that there is
a priori no repulsive attitude to the creole, just as there does not seem to be any
corrective attitude toward its speakers. This attitude toward the emerging
language clearly reflects the fact that it is the language widely spoken and
accepted on the colony.10
Yet, the colonists were very much aware of the fact that in order to maintain
the segregating rules in the colony, their children had to acquire the standard
European variety in addition to the local vernaculars (including the creole). A
measure often taken to address this problem is the creation of schools for White
children only. In this regard Long (1774: 250, my emphases), arguing for
boarding schools for the Jamaican plantation children, writes:
The utility of a boarding-school for these girls, where their number might admit of
employing the ablest teachers, where they might be weaned from the Negroe dialect,
improved by emulation, and gradually habituated to a modest and polite behaviour,
needs not, I think, any argument to prove it. Young ladies, so far accomplished as, I
think, they might be on a well-conducted plan, would insensibly acquire, on their
emerging into public life, the remaining graces and polish which are to be attained in
gentle company and conversation. They would by these means, become objects of love to
the deserving youths, whether natives or Europeans, and by the force of their pleasing
attractions soon draw them, from a loose attachment to Blacks and Mulattoes, into the
more rational and happy commerce of nuptial union.
This last excerpt further indicates that not only did White Creoles speak the
creole from birth, but it appears to have been their daily language until maturity.
That the creole was the lingua franca in the early years of plantation colonies is
therefore compatible with Mufwene’s (2001) view that these varieties became
disenfranchised later when the segregating rules of plantation societies became
the cornerstone of Creole societies based on slave economy. In the context of
this book, the following points indicate that the historical facts reviewed in
Chapter 2 and in this section have important repercussions on how to consider
the formation of the creole language.
– It is clear from this discussion that commonly used notions such as target
language need to be clarified. The facts presented here suggest that speakers
were exposed to a continuum of varieties of the substrate and lexifier
vernaculars, in addition to the emerging contact language. We therefore
reach a situation where there were potentially several competing targets
with the creole functioning simultaneously as target and playing a useful
function in the chain of information transmission (from the lexifier and
substrate languages).
10
See also Lalla and D’Costa (1991) for discussion.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
132 Competition and selection
– Because all available targets contribute to the input from which the
speakers select features into the emerging creole; and because these
speakers varied from each other, the emerging creole ‘inherited’ fea-
tures produced by both (bilingual) L1 and L2 learners of the colony
(i.e., Whites, Creoles, and Blacks alike). Contrary to what is often
assumed in the literature, this point makes it even harder to assign the
creation of the creole exclusively to a specific group of learners (e.g.,
enslaved African L2 learners or children), consistent with DeGraff
(1999: chapter 1, 2002).
– The commonly assumed restricted access to the target also becomes unten-
able. First, there were several targets (including the emerging creole).
Second, even though not every single African was exposed to what Long
(1774) referred to as ‘bad English,’ which we could interpret as ‘non-
standard colonial vernacular’ on a daily basis, there were various channels
through which most inhabitants of the colony could approximate the colonial
vernacular.
– The findings presented in this section indicate that the role of locally born
bilinguals in the development of the creole has been largely underestimated
in previous accounts. As I show in subsequent chapters, Haitian Creole
and the Suriname creoles display very many subtle structural properties of
their lexifiers that cannot be attributed to L2 learners in the ways often
assumed in creolistics (viz., fossilized L2 speech, or L2 learners with poor
access to the target). Many of such features seem to have been injected in
the language by (near) native speakers many of whom were probably child
bilingual speakers (i.e., either bilinguals in the emergent creole and the
local European koiné or in the emergent creole and in some African
language).
– Finally, the notion of substrate transfer (as understood in creolistics) appears
obsolete. As I show in Chapter 5, what one finds instead is a process of
transfer whose direction seems determined by linguistic factors such as the
properties of interfaces rather than the origin of the speaker (e.g., African
ethnicity).
I conclude from this discussion that creoles are only new with regard to
history but the bulk of their morphosyntactic features (i.e., those features which
are selected from the inputs) are to be found in the languages that contributed to
the feature pool. In the discussion that follows, I argue that the model of
competition and selection as described earlier (and in detail in Mufwene
2001, 2005a, 2008) offers an adequate conceptual framework within which
we can approach the evolution of the contact of input languages into a new
language. This discussion bears on issues of language creation and language
change on the level of the speaker.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.2 Languages as biological species 133
Language evolution
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
134 Competition and selection
topic and focus marking, Aboh 2006a, 2006b, 2007a, Chapter 3 of this book),
Mufwene is careful enough to complement this quote with the following:
(13) It is as necessary to invoke substrate influence from the Celtic languages to
account for the speciation of Latin into the Romance languages as it is to
invoke African substrate influence to account for the evolution of French and
Portuguese into various creoles . . . Both cases are clear instantiations of
Pyrrhic victory – where the prevailing language is so clearly affected by the
displaced ones (Mufwene 2005b: 12).
According to Mufwene’s work therefore, substrate or cross-linguistic
influence, which we should understand as structural interaction between two
L1’s during bilingual acquisition or between L1 and L2 during L2 acquisition is
not only undeniable, but it represents crucial ingredients in change.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.2 Languages as biological species 135
the substrate languages can only make their way in the grammar of the lexifier
koiné if the former and the latter show (partly) congruent structures. This, in
turn, means that the linguistic ecology in which the creole developed favored
the lexifier language but constrained the substrate languages. Accordingly,
substrate influence appears limited to areas of congruence.11
This view may look reasonable if one considers political and sociolinguistic
factors (e.g., the oppressing power forcing a linguistic policy on the enslaved
Africans and prohibiting African languages). However, as shown in
Section 4.1, things might not be so clear-cut. The discussion in previous
sections shows that there is ample evidence that some African languages
were spoken for some period in the colonial Caribbean and in colonial Brazil
(e.g., Long 1774; De Castro 2002; DeGraff 2002: footnote 45). This would
mean that linguistic features from the African languages could be selected into
the creole even though they might not converge with the European target
language. Also recall from that discussion that the process of competition
and selection proposed by Mufwene (2001ff.) assumes that linguistic features
of the competing languages form a feature pool on which selection operates.
Under this view, it is not clear to me how the ecology of such a language contact
situation, in which the lexifier represents the target language, constrains
selection inside the feature pool such that non-congruent substrate linguistic
features are strongly disfavored (if not excluded).
If we adopt a minimalist approach to the study of human knowledge of
language, as I do here, it appears that factors that concern speakers’ linguistic
ecology or practice are external to the computational system underlying human
language capacity. The latter, however, is subject to the competition and
selection process, which affects various linguistic modules during the
development of the idiolect in the mind of the speaker.
In order to keep a formal distinction between change at the individual level
and the population level, I adopt DeGraff’s (1999: 9) distinction between
E-creole and I-creole, as already stated in Chapter 1. An E-creole is an
abstraction of the linguistic codes of a creole community, whereas an I-creole
denotes “the development in individual speakers’ minds/brains of a grammar
that shows a certain typological distance from the grammars of the languages in
contact.”
Adopting this distinction, I submit that ecological factors affect I-creole
indirectly. Indeed, ecological factors may affect the learning setting in
various ways thus constraining the feature pool to which the learner is
exposed and which will feed into the development of her I-creole. For
11
As suggested to me by M. DeGraff, Chaudenson’s position can hold only if we have a clear
definition of congruence such that we are able to reconstruct the distinctive features that may
converge in the competing languages. See Aboh (2009b) for some discussion.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
136 Competition and selection
12
This need not mean that certain external factors (e.g., communication settings) may not favor
certain linguistic patterns, which because of their frequency or discourse prominence may in
turn favor some linguistic features and their associated parameters. To put it more concretely,
external factors such as frequency, or markedness (however defined) may favor certain syntactic
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.2 Languages as biological species 137
saying that competition and selection happen at two levels: the structuring of
the input, and the combination of variants selected from the input into a
coherent syntactic system. Distinguishing between these two levels of
analysis allows me to focus on syntactic recombination only in this book.
In addition to being perfectly compatible with Mufwene’s general
framework, the proposed distinction enables us to reformulate the question of
some similarities between creoles and their source languages (viz., the lexifier
languages and/or the native languages of the Africans) as well as across creoles
in terms of a more general question related to the emergence of new linguistic
variants: Why are certain linguistic features typically selected over others in
contact situations, and why are these features typically combined in some
particular ways?
The latter question obviously relates to even more difficult questions such as
why the competition and selection process does not affect all components of a
module (e.g., VP, IP, CP, DP in syntax) or all modules of the grammar (e.g.,
syntax, semantics, phonology, morphology, pragmatics) the same way. A
fascinating example discussed in DeGraff (2002, 2005) concerns word order.
Most Kwa languages, the major substrate languages for the Suriname creoles
and Haitian (Arends 1989; Smith 1987; Lefebvre 1998) display VO versus OV
alternation in specific aspectual contexts (e.g., progressive, Aboh 2004a,
2005c). No such aspect-dependent word order variation is found in Haitian,
Saramaccan, or Sranan. This is surprising since the Kwa speakers, being
numerically dominant in this context and having this feature in their
E-language, must have contributed this aspect-dependent word order
variation to the feature pool. Nevertheless, this putatively strong element of
the feature pool was not selected into the emerging creoles.
One could think that the OV order was simply selected against, because the
alternative VO order was congruent with that of the lexifier. But it is worth
noting that in the case of the Suriname creoles, OV was selected against even
in contexts where both the superstrate and the substrate converge (e.g., in
nominalizations: street sweeper, àlìò zà-tɔ́ street/sweep/person vs. figi-strati-
man sweep/street/person ‘street sweeper; van den Berg 2007: 175), see
Section 4.3.2. Similarly, DeGraff (2005) indicates that French has alternative
OVorder with clitics and this was also selected against in Haitian Creole. In this
language, the object must follow the verb.13 These facts underscore the point
made here that the dynamics of competition and selection in the syntactic
module must be more complex and to some extent blind to certain factors
related to the input that may have to do with say, number of speakers, prestige,
properties over others (e.g., periphrastic aspectual expressions with non-finite verbs versus
finite verbs), but these external factors do not determine how syntactic recombination operates.
13
I thank M. DeGraff and S. Mufwene for their suggestions on this issue.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
138 Competition and selection
etc. Clearly, there must be some independent reason internal to the syntactic
component why the emergent system selected against OV even when all
external factors would seem to favor it.14
Similarly Aboh (2006a, 2006b, 2007a) showed that while the Gbe
languages have influenced the expression of the complementizer and
determiner systems in the Suriname creoles to some extent, the same does
not hold for other syntactic areas such as the clausal inflectional domain (IP)
or even when it comes to modification inside the noun phrase. More study is
needed before we understand how linguistic features of competing languages
are recombined in the emerging language. This book aims to contribute to a
better understanding of this process and how it applies to language acquisition
in general.
In the context of the present discussion, however, the recombination of
linguistic features in an I-creole, the replication of such I-creoles, and the
emergence of a community of speakers of similar I-creoles leading to the
birth of the E-creole is comparable to the evolution of a population in
biology. The latter may consist of the mutation of a gene and the selection/
reproduction of individuals carrying such gene which eventually leads to a
population change. Other striking parallels between biological species and
languages (e.g., the analogy between successful communication and
biological interbreeding in Mufwene 2005b: 15) lead me to propose that
(15) Languages are species whose phenotypes – the linguistic features on which
typological classifications are based – correspond to genotypes, in this case,
syntactic structures.15
According to (15), syntactic features are expressions of structures (i.e.,
functional projections as defined in generative syntax) that are somehow
comparable to genotypes. The next section discusses this issue.
14
DeGraff (2005: 353, n.32) speculates about certain external factors that may have played a role
in favoring the rise of VO in Haitian Creole in contexts where certain French varieties use VO.
As is usually the case, the problem here too is to identify the right varieties and find out whether
they are relevant for the genesis of Haitian creole. I leave these issues for future research.
15
This claim appears in contradiction to Mufwene (2005b: 15, and footnote 13) who argues that
“Languages are species whose phenotypes – the linguistic features on which typological
classifications are based – correspond to no genotypes . . . Clearly languages and idiolects
cannot have genotypes, because they are not biological systems.” Mufwene’s theory builds on
the fundamental assumption that languages are more comparable to viral species than to either
organisms or to animal species.
I do not discuss these issues in this book, as they do not directly bear on the argumentation. It
is, however, important that the reader keeps in mind that it cannot be assumed that all phenotypic
properties of languages are rooted in the distinctive linguistic features as defined here. What this
book tries to do, instead, is to identify which linguistic features have typical realizations (i.e.,
phenotypic effects) and how such features recombine with other linguistic features in a situation
of language contact.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.2 Languages as biological species 139
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
140 Competition and selection
As we can see from this sketch, the morpheme spelled out as -na displays a
very specific semantics (i.e., habitual) in addition to a cluster of syntactic
properties: (i) it selects a VP as complement and it requires verb-movement.
The latter can be further formalized by suggesting for instance, that -na has a
formal v-feature that needs to be valued (or checked) by V (the verb). Setting
these technical details aside, what matters for the present discussion is that
when it comes to morphosyntax, this morpheme has two properties: a
selectional property that regulates the type of element or category it can
combine with, and a formal feature that governs its licensing and satisfies its
morphological requirements as an affix.
Because the boxes labeled “Phonology,” “Morphosyntax,” and “Semantics”
operate independently, though they interact during the derivation, we can
conclude that functional items are the loci of linguistic features which drive
grammar. This in turn leads me to the conclusion that functional items are the
target and carrier of change. The latter point is compatible with Borer’s (1984)
view that cross-linguistic variation is attributable to differences in abstract
features of lexical/functional items (see also Chomsky 1995, 2001; Baker
2008; and Boeckx 2010 for more elaborate discussion.) Under the traditional
view that functional items head functional projections which represent the
backbone of the clause, the representation in (18) suggests that functional
projections include licensing specifications that govern clause structure
building and determine cross-linguistic variation.
From the point of view of acquisition and change, this description further
indicates that the acquisition of grammar consists of the development of an
algorithm that recombines formal properties of diverging lexical/functional items
into a coherent system. Because acquisition and change go hand in hand, we
further reach the conclusion that in a situation of language contact, any component
(or combination) of the triplets in (16) and (18) (viz., phonology, morphosyntax,
semantics) can be affected. Consequently, we need to understand how lexical/
functional items emerge, in order to understand the emergence of new grammars.
With this discussion in mind, I assume (following Rizzi 1997; Cinque 1999;
and much related work in the cartographic approach) a description of clause
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.2 Languages as biological species 141
Under this view, while the clause structure is similar to the DNA in containing
all the syntactic features as well as their formal licensing properties (i.e., the
parameters), the projections are comparable to genes in that they encode specific
information about specific syntactic features and parameters only. There is,
however, an important evolutionary/developmental difference between the
two: DNA precedes the development of an organism, whereas the clause
structure is a consequence of learners’ experience during acquisition. This is an
important difference between acquisition of a biological trait by genetic
inheritance and the acquisition of a feature by learning. The bearer of a DNA
cannot modify it, while the carrier of a linguistic feature may modify it. These
considerations obviously break down the genotype/phenotype parallelism as
well as that between biological and cultural evolution. It is therefore important
to emphasize that the theory developed in this book builds on an analogy with
biological species and not on isomorphism.
If we grant the description in (19), however, we can further say that
manifestations of linguistic features, which Mufwene correctly treats as
phenotypes, are expressions of a combination of syntactic nodes according
to how the associated parameters have been set in the development of the
I-language, in this case the I-creole.16 In other words, the overt manifestation of
a learner’s mental grammar is nothing but the spell-out of specific
recombination of linguistic genotypes, as illustrated in (20).
16
As Michel DeGraff remarks (p.c. June 21, 2009), it is clear from this description that a certain
amount of “reverse engineering” qua linguistic analysis is necessary to uncover the genotypes that
are reflected in the phenotype (i.e., the overt manifestation of linguistic system). This exercise is
obviously not a simple one given that not all the properties of the phenotype relate to distinctive
linguistic features that are subject to competition and selection. Though further research is needed,
questions of this sort show the complexity of the issue at stake, and cast serious doubts on theories
of creoles that assume a trivial complex versus simple opposition where the presence versus
absence of morpheme types plays an important role (e.g., McWhorter 2001).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
142 Competition and selection
(20) E-language
I-Language
Linguistic experience
(e.g., LPD, language contact)
The box with broken lines in this representation is meant to show that what
is referred to as E-language (or E-creole) by DeGraff (1999) is an abstraction
over collective spell-outs of individual I-languages/-creoles. Furthermore,
representation (20) indicates that competition and selection as well as the
recombination of syntactic features happen in the middle box during the
mapping of genotype and phenotype. This actually corresponds to parameter-
setting in the generative framework (see Lardiere 2000, 2005 for some
discussion). Under this view, the languages in contact provide the speaker
with the right triggers (or context) for parameter setting. Taking this line of
thought seriously, the only difference between contact languages or creoles, on
the one hand, and other languages, on the other, is that the input for parameter-
setting is too diverse in the case of creoles. As a result, several possible options
compete for the fixation of one and the same parameter (e.g., VO vs. OV,
affixal TMA morphology vs. free morphemes, specificity vs. definiteness).
Furthermore, the recombined syntactic features that this competition gives
rise to in the I-language can also be very divergent, when compared to the
source languages. This is so because the range of variation set by the
typologically diverse competing languages in a plantation colony confronts
the speakers with a wider range of learning hypotheses than in monolingual
settings. This conclusion is compatible with the position assumed in this book,
viz., that creole languages only differ from older languages with regard to the
socio-historical contexts in which they came into being. This social context
brought together – at a certain point in time and in an isolated geographical
location – typologically different languages which otherwise would not have
met (see also Mufwene 2002, 2003 for some discussion).17
17
On a more general note, the description in (16), (18), and (19) assumes that syntactic features
have a formal component (i.e., licensing properties) and a semantic component (e.g., discourse
function, meaning). Each component can be copied separately or independently during the
recombination process that leads to the new language. According to this hypothesis, language
change boils down to the study of recombination of linguistic features. In addition, if, adopting
this view, we are able to isolate syntactic features to the extent that we can distinguish between
dominant (i.e., semantically prominent) and weak (i.e., semantically empty) syntactic features,
we may have an answer to why certain features are typically found in contact languages as well
as in bilingual acquisition. This predicts that the weak features (under certain ecologies) may
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.2 Languages as biological species 143
Table 4.1 A comparison of the left periphery of Saramaccan, English, and the
Gbe languages
Languages
become dominant under recombinations in new ecologies. I hope to come back to these issues in
future work.
18
This view also implies that the recombination of syntactic features may lead to changes across
syntactic paradigm. For instance, the specification of tense as free morpheme may have
consequences in the development of mood and aspect markers and the blocking of verb move-
ment. This of course strengthens the comparison between languages and biological species, but
I leave the issue for further research.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
144 Competition and selection
Languages
the left periphery. Yet, with the exception of the focus marker wὲ , which is
morphologically identical to that of Fongbe and Gungbe, most Saramaccan
markers derive from English (and keep the morphosyntax of English to some
extent). Maintaining the analogy between I-languages and biological
organisms, I claim that this is akin to feature recombination where the Gbe
linguistic features became dominant (Aboh 2006a, 2007a).
Though these languages show strong parallels, when it comes to the
complementizer system, other modules of the grammar may be differently
affected due, to competition and selection. Table 4.2, on the syntax of the
noun phrase, shows this: the determiner in Saramaccan and Gungbe display
semantic parallels but differ with regard to syntax (Aboh 2006a, 2007a).
Table 4.2 therefore suggests that while we can argue that the selected
semantic features are from Gbe, their morphosyntactic realization follows
English rules. By analogy again, this would mean that the features selected
from Gbe became weak under recombination. Alternatively, one could propose
that the competition and selection for the features specific versus nonspecific
implied two choices: one for the semantics of the feature and the other for the
syntax. Accordingly, both competing language types (e.g., Gbe vs. Germanic)
won part of the battle here: Gbe-type languages won on the semantic side and
English won on the syntax side. This is what Mufwene refers to as “Pyrrhic
victory” for the lexifier in (13).
Under the competition and selection process as I interpret it here, these facts
indicate that creoles or contact languages are not (approximate) replicas of
existing linguistic systems. Instead, these new languages are syntactically
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 145
19
As I explained in Chapter 1, the reader should refrain from confusing the process of hybridiza-
tion as developed in this book with the way Whinnom (1971) conceived of it to account for the
emergence of creoles. For instance, the latter spoke of stages of hybridization in ways that do not
apply to language acquisition. Also, he did not discuss feature recombination as a process that
contributes to the restructuring of a language into a new variety, the process that Mufwene
(2001, 2005b, 2008) characterizes as “speciation.”
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
146 Competition and selection
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 147
c. Kòfí ɖù àlè.
Kofi eat profit
‘Kofi made a profit.’
Other expressions include:
(23) d. ɖù àxɔ́
eat debt
‘to go bankrupt’
e. ɖù gbὲ
eat life
‘to enjoy life’
f. ɖù xwè
eat year
‘to celebrate’
g. ɖù àɖì
eat poison
‘to be angry’
h. ɖù wìnyán
eat shame
‘to be ashamed’
i ɖù nù gò
eat mouth PREP[at]
‘to boast’
j. ɖù gán
eat chief
‘be(come) chief’
k. ɖù yà
eat pain
‘to suffer’
While the examples in (23) might look diverse and unrelated at first sight, the
underlying meaning suggests that the logical subject experiences a state or
feeling. In this regard, it should be noted that these examples are not fixed or
idiomatic expressions because they are quite productive and allow further
combinations. Consider, for instance, the sequence (23e) ɖù gbὲ ‘to enjoy life’
as used in sentence (24a). In this example, the inherent object noun phrase gbὲ
‘life’ is further specified to form a possessive noun phrase literally ‘life of Kofi.’
Yet the intended meaning is not one whereby the speaker enjoys Kofi’s life but
rather refers to a state of affairs where the speaker enjoyed or liked an action
performed by Kofi. This sentence can be uttered in various contexts, such as a
situation where Kofi invites his friends to his place or to a restaurant to offer them
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
148 Competition and selection
dinner.20 Sentence (24b) on the other hand refers to a situation where the speaker
was very happy with the car and liked driving (or being transported in) it.
(24) a. Mí ɖù gbὲ Kòfí tɔ̀ n.
1PL eat life Kofi POSS
‘We enjoyed staying with Kofi.’
b. Mí ɖù gbὲ mótò lɔ́ tɔ̀ n.
1PL eat life car DET POSS
‘We enjoyed the car.’
Finally, the facts in (22), (23), and (24) indicate that the verb ɖù in Gungbe poses
no particular agentivity or animacy restriction on its external and internal
arguments, which could be both theme and experiencer. These data suggest that the
lexical entry corresponding to English ‘eat’ in the Gbe languages covers a wider
range of meanings than English ‘eat,’ whose semantics is much more restricted.
Likewise, that this verb is an ICV in Gbe indicates that it is strongly transitive,
and therefore has a different argument structure than English ‘eat,’ which is either
transitive or intransitive depending on the context. The morphosyntactic
specifications of V-eat in the two language types is represented in (25):
(25) a. Gungbe
vP
VP
V NP
[N-to-V incorporation]
b. English
vP
v VP
V[eat] …
V-to-v movement/incorporation
20
One also finds ɖù mὲ jí: eat someone on => to win over someone, ɖù gú: eat inheritance => to
inherit etc.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 149
Without getting into the intricacies of the formalism, the description in (25a)
suggests that ICV’s in Gbe are genuine light verbs that merge in little v and
select for a VP whose head is a transitive empty V. I submit that such empty V’s,
select for an NP complement, thus creating an incorporation context in which
the head N incorporates into V (Baker 1988; Hale and Keyser 1993; Aboh and
Dyakonova 2009; Aboh 2010d). We therefore reach a situation where it is the
incorporated N that lexicalizes V. Given that little v is a light verb, the
semantics of the ICV derives from the complex v + N (that has incorporated
into V). In the case of the discussed examples, I conjecture that v-ɖù has the
vague meaning of ‘to get’ (by ingestion of X, association with X, or being
affected by X).21 I further conclude that it is the selection of the appropriate
complement that further specifies the meaning of the complex v + N. In case N
is a dummy element, such as nú ‘thing’ in (22a), the verb receives the generic
meaning of ingestion on the basis of the context.
English in (25b), on the other hand, displays a lexical V-eat that is specified
as either transitive or intransitive and merges under V from where it raises to
little v in the current minimalist approach (e.g., Chomsky 1995). Contrary to
the situation in Gungbe, English eat seems to have the basic meaning of
ingestion for the purpose of feeding oneself or someone else.
Thus, let us consider the corresponding verb njan ‘eat’ in Saramaccan. With
regard to syntax, Saramaccan njan exhibits the same behavior as English eat in
that it can be transitive or intransitive. This is illustrated by the following
examples. Sentence (26) indicates that, like English eat, the verb njan may
take an internal argument.
(26) Amato njan di bakuba.
Amato eat DET banana
‘Amato ate banana.’
Example (26) is obviously compatible with the Gungbe example (22b),
in which the light verb v-ɖù selects for a complement. Accordingly, there
seems to be no clear difference between Gungbe, English, and Saramaccan
in this respect. Yet, when we contrast example (26) with those in (27),
taken from Rountree and Glock (1982: 43), we realize that Saramaccan
njan may lack a complement just as English eat (27a–b) and unlike
Gungbe ɖù (22c).
(27) a. I njan kaa no?
2SG eat already Q
‘Have you already eaten?’
21
English examples that illustrate this description include constructions such as ‘I got some food,’
‘I got money,’ or ‘I got malaria.’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
150 Competition and selection
b. Ai mi njan (kaa).
yes I eat already
‘Yes I have eaten (already).’
c. Ai mi njan soni.
yes 1SG eat something
* ‘Yes I ate.’
✓‘Yes I ate something.’
In addition, the ungrammaticality of Saramaccan (27c) under the relevant
reading, in contrast with the Gungbe example (22a), indicates that Saramaccan
njan cannot select for a dummy noun phrase. This clearly indicates that this
Saramaccan example has the same argument structure (and therefore syntax) as
the English verb eat. I therefore conclude that the syntax of njan in Saramaccan
maps onto that of English eat as represented in (25b). This would mean that like
English eat, Saramaccan nyan merges under V from where it moves to v, which
it lexicalizes.
Yet, Saramaccan njan occurs in sequences such as in (28), many of which
(e.g., to spend, to suffer, to boast, to have a headache) have literal translations
in Gbe as previously illustrated in (23).
(28) a. Hédi tá njan mí. [Rountree and Glock 1982: 39]
head PROG eat 1SG
‘I’m having a headache.’
b. Nján búka [Donice and Voorhoeve 1963: 80]
eat mouth
‘to boast’
c. Kofi ta nyan suti buka [Haabo, p.c May 22, 2007]22
Kofi PROG eat sweet mouth
‘Kofi is boasting.’
d. Njan moni
eat money
‘to spend money’
e. Njan pena
eat pain
‘to suffer’
f. Njan yai
eat year
‘to celebrate’
22
I thank Vinje Haabo for providing me with these examples.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 151
Assuming that the representation in (25b) holds for the verb njan in
Saramaccan and eat in English, the question arises of how Saramaccan
speakers developed the usages (or semantics) in (28) which are similar to
Gbe constructions. Two options seem possible here. One could propose that
such expressions in Saramaccan are calqued on Gbe and are comparable to
fixed idioms such as kick the bucket in English. But this view is unlikely given
that the sequences in (28) are highly productive and the verb can take additional
arguments as indicated in (29).23
(29) Kofi ta njan Gaamá a baka [Haabo, p.c. May 22, 2007]
Kofi PROG eat Grandman PREP back
‘Kofi is gossiping about the chief.’
Another possibility, worth considering is that the verb njan in Saramaccan
combines semantic properties of both Gungbe ɖù and English eat. More
specifically, I submit that the uses in (28) represent an extension of the
semantic specifications of eat in English, under the influence of Gbe. A look
at the entry eat in The New Oxford Dictionary of English (2001) provides us
with very interesting examples as in (30), which are reminiscent of the Gbe
usages and could have served as the transmission belt for combining English
and Gbe semantics of ‘eat’.
(30) a. Eat one’s heart out [i.e., suffer from excessive longing, especially for some-
one or something unattainable].
b. What is eating you? [i.e., What is worrying or annoying you?]
These examples are similar to the Gbe examples in (23a, 23i) and indicate
that both in Gbe and in English the lexical entry for ‘eat’ may have an
experiencer as internal argument. Under this description, Saramaccan njan
maps the semantic properties of English and Gbe ‘eat’ onto the syntax of
English. In order to see how the argument goes, let us try again to represent
the lexical items ɖù from Gungbe, eat from English, and nyan from
Saramaccan on the basis of the format in (18).
Starting with Gungbe ɖù, it appears that, unlike English ‘eat’, it does not
mean ‘to ingest something for the purpose of nutrition’ but rather has a vague
meaning that is comparable to English ‘to get’, for which emphasis is put on the
experiencer, like in Gungbe. In addition, the morphosyntax of this verb
suggests that it is a light verb. Putting these properties together, we can
describe this verb as in (31):
23
Interestingly, the usage and meaning in (29) do not seem to exist in Gbe, although I may be
wrong.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
152 Competition and selection
It is clear from these two descriptions that not only do the lexical items
used to express the meaning of ‘eating’ differ regarding their respective
semantics in Gungbe and English but also they show fundamental
differences regarding their morphosyntax. While the Gbe element is
strongly transitive and always requires a complement (even in its citation
form) the English lexical item has usages where it is transitive or
intransitive as in the contrast between the lions ate the visitor versus the
lions already ate. In addition, the Gungbe verb occurs in almost unlimited
semantic contexts with varied meanings, while the English verb has a fixed
meaning of ingesting some substance for the purpose of nutrition. Given
this background, the description of the Saramaccan verb nyan appears to
correspond to representation (33).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 153
Several observations are in order here. First, the representation in (33) clearly
indicates that the Saramaccan lexical item njan is not a replica of its counterparts
in English (32) and Gungbe (31). Instead, the newly created item appears to
share some properties of its counterparts in both source languages. Second,
rather than resulting in a mere (ad)mixture as some creolists misleadingly claim,
the process of recombination that results from the competition and selection
appears to be selective. We can indeed conclude on the basis of representation
(33) that the speakers of Saramaccan did acquire the morphosyntax of the lexical
entry corresponding to the verb eat in English concurrently with its different
usages as transitive or intransitive verb. Aside from the phonological form, the
fundamental change (or alteration) that is responsible for a difference between
English (in this case the primary lexifier) and Saramaccan, lies in the semantic
component: the semantics of the English verb has been conflated with that of
Gbe, which accounts for its wider range of usages. These usages in turn come
with their own syntax which may affect sub-domains of Saramaccan syntax.
Such an evolution suggests that language transfer is a selective process.
An apparent issue that is often raised against the feature pool idea and the
recombination of syntactic features proposed here is why one finds some
instances of recombination as opposed to others. For instance, why does
recombination here happen in the semantic component (but apparently not in
the syntactic component)? In the present case, why doesn’t the Saramaccan
verb njan display the syntax of its Gbe cognate ɖù in some well-defined cases
while displaying English syntax only in other contexts? In keeping with the
biological metaphor, it is clear that such a question does not make much sense
as it boils down to asking why mutation happens the way it happens, for
example, why is a mule not half a horse and half a donkey? The relevant
scientific question, it seems to me, is for students of evolution to understand
how recombination happens, producing the particular phenotypes that it yields.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
154 Competition and selection
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 155
However, a number of facts about these sequences suggest that tɔ́ /nɔ̀ /man are
affixes within a predicate structure. One such fact is that the Sranan and
Saramaccan element -man (derived from English man) and the Gungbe
forms -tɔ́ /-nɔ̀ , which literally mean ‘father’ and ‘mother’ respectively, are
unspecified for gender in compounds, unlike when they occur in isolation.
Therefore, while the examples in (35) can refer to both male and female person,
those in (36) have gender specifications and only refer to either a male or
female person.24
(36) a. (ò)tɔ́ cè
father 1SG.POSS
‘my father’
b. (ò)nɔ̀ cè
mother 1SG.POSS
‘my mother’
c. man-ningre / ningre uman
man black black woman
‘black man/black woman
An immediate conclusion here is that the elements -man (Sranan/
Saramaccan) and -tɔ́ /-nɔ̀ (Gungbe) do not function as lexical items when they
occur in compounds. Instead, they are somehow semantically bleached and
behave like derivational affixes (or grammaticalized items) that attach to the
element on their left. This description finds immediate support in examples
such as those under (37), where these morphemes attach to a verb phrase.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
156 Competition and selection
25
Smith and Veenstra (1998) observed similar examples in Saramaccan.
26
Contemporary Sranan displays such sequences with VO order as in the following example:
(i) Sibi-strati-man [Contemporary Sranan]
sweep street man
‘Street sweeper’
27
This issue merits further attention and I return to this discussion in Section 4.3.2.2. As
mentioned in Section 4.2.1, it is remarkable that while nominalized verb phrases display OV
order in both Gbe and English the Saramaccan equivalents maintain VO order. This correlates
with the facts discussed in DeGraff (2002), and Aboh (2006a) where it appears that even though
most Kwa languages exhibit VO versus OV alternation in the context of aspect, none of their
supposedly related creole has the alternation. These facts cannot be accounted for with uni-
versals, since the observed alternations are context-dependent and do not imply a different head
parameter (Aboh 2004a, 2005c). In addition, the compounds discussed here are similar to
English compounds, such as, gingerbread man, house-cleaning man, in which the properties
of man (as a head noun or an affix) are not clear. Assuming that there is surface congruence
between English and the Gbe languages with regard to these compounds, then the decisive
factor for its syntax in the Suriname creoles could have to do with whether or not it is treated as a
derivational affix as in Gbe.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 157
With regard to morphology, for instance, the elements. -tɔ́ /-nɔ̀ and -man
display a reduced form, which supports its analysis as an affix. Empirical
data from, on the one hand, Gungbe and, on the other, Sranan and Saramaccan
support this point. As discussed in Aboh (2005a, 2010b), and in Aboh and
Smith (2015a), Gungbe noun phrases have an initial vowel (a- or o-) as
indicated in (40):
(40) a. òhún ‘drum’
b. àgásá ‘crab’
The prefix o- can be dropped, while the prefix a- cannot.
(41) a. Kɔ̀ kú xɔ̀ òhún/hún ɖé.
Koku buy drum DET
‘Koku bought a drum.’
b. Kɔ̀ kú xɔ̀ àgásá/*gásá ɖé.
Koku buy crab DET
‘Koku bought a crab.’
Yet, o- and a- must drop in compound nouns, when the noun they attach to
occurs as a second term. The examples in (42) illustrate a noun with the initial
vowel o-:
(42) a. (ò)hún kpòtín
drum stick
‘drum stick’ (as stick/mallet to play drums)
b. sìn-(*ò)hún
water drum
‘water drum’ (played at funerals)
Those in (43) on the other hand illustrate a noun with the initial vowel a-:
(43) a. àgásá fὲn
crab foot
‘crab foot/pincers’
b. xùmὲ-(*à)gásá)
sea crab
‘sea crab’
In both cases, the initial vowel must drop when it occurs on the second
conjunct of a compound (42b, 43b). The same constraint applies to the source
nouns òtɔ́ ‘father’ and ònɔ̀ ‘mother’ which fail to retain their initial vowel in
compounds (30).
(44) a. *hún kùn òtɔ́ lɔ́
car engine person DET
‘driver’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
158 Competition and selection
4.3.2.2 X-tɔ́ /nɔ̀ /man as predicate structure A fact that one immediately
notices about the sequences X-tɔ́ /nɔ̀ /man is that they express a relation such
that X, a noun phrase, denotes an abstract entity, property, or quality that
predicates over tɔ́ /nɔ̀ /man. When X represents a verb phrase, however, it
expresses an event of which tɔ́ /nɔ̀ /man is an agent. The formed X-tɔ́ /nɔ̀ /man
sequences can be used in equative constructions such as in John is X-tɔ́ /nɔ̀ /man
which could mean ‘John is wealth-person (i.e., rich)’ or ‘John is drive-person
(i.e., driver).’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 159
spec F¢
aze/ aseh
F° NP
t /man
Under this representation, the head N tɔ́ /nɔ̀ /man incorporates into Fº where it
functions as an affix. Incorporation of N into Fº explains the morphological
change of the full form corresponding to the lexical entry man, òtɔ́ ‘father,’ ònɔ̀
‘mother’ into the reduced form -tɔ́ /nɔ̀ /ma(n), as discussed previously. On this
view, the sequences involving a (nominalized) verb phrase in (37) for Gungbe
and (39) for Sranan can be derived as in (48):
(48) FP
spec F¢
VP[Nom]
F° NP
t /man
Sranan and Gungbe differ in that the Sranan VP realizes the VO order or may
contain the verb only. In Gungbe, however, the nominalization of the VP
requires the OV order. This difference aside, what we see here is that the
Sranan forms aséh-man or sibi-strati-man map on the (Gbe) structures (33)
and (34), where man has been substituted for tɔ́ /nɔ̀ .
The case of aséh-man ‘witch’ is illuminating in this respect because it
raises an interesting issue with regard to language acquisition. Because the
speakers who created this form had spoken Gbe, I surmise that they could
have simply retained the equivalent Gbe word àzé-tɔ́ , just as they retained
aséh alongside with numerous other Gbe lexical items. Yet, this did not
happen. Instead, the form aséh-man represents convincing evidence that
these English learners did entertain learning hypotheses about English
derivational morphemes. This acquisition process, I claim, allows the
learners to relate the Gbe derivational morphemes -tɔ́ /-nɔ̀ to English -man
and -er. For the creators of the Surinamese creoles to be able to match these
English lexemes with Gbe structures, however, they have to make the right
hypothesis that tɔ́ /nɔ̀ corresponds to man in its affixal usage which paved the
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
160 Competition and selection
way for the development of the derivational morpheme -man. The latter
appears to embed properties of the derivational morpheme -er as a result of
which sequences of the type X-man express a relation such that X may be a
noun phrase or a verb phrase. When X is a noun phrase, it often represents an
abstract entity, property, or quality that predicates over man. On the other
hand, the contexts where X is a verb phrase express events of which man is
the agent.
Yet, morphosyntactic constraints on the development of -man resulted in
a change that is worth discussing. In the previous paragraphs, I mentioned
that while contexts such as (48) require the OV order in Gungbe, they are
realized as VO in Sranan and Saramaccan. Let us consider again the Early
Sranan example in (39c) repeated here as (49a), compared to its Gungbe
equivalent in (49b).
(49) a. figi-strati-man
sweep street man
‘street cleaner/sweeper’
b. àlìò-jí zà tɔ́
road-surface sweep person
‘street cleaner/sweeper’
c. street sweep-er
If indeed, the Sranan and Saramaccan affix -man were mapped on the Gbe
morphological structure as suggested in (48), one wonders why the two
languages display different word orders within the nominalized verb phrase
in its specifier. The issue appears even more complex when one considers
similar sequences in English. As example (49c) shows, the English equivalent
of (49a) also exhibits the OV order. We are therefore in a situation of strong
congruence where both source languages display OVorder in the same context
but the creole realizes the VO order. Why?
In order to answer this question, it is important to realize that, according to
the structure in (48), the morphemes -tɔ́ /-nɔ̀ and -man/-er have the
selectional requirement that the category in their specifier be a noun
phrase or a nominalized predicate (e.g., a nominalized verb phrase).
Crucially, however, these categories do not act as nominalizers (as one
could imagine from some traditional accounts of English -er).
Accordingly, the variation between, on the one hand, Sranan and
Saramaccan and, on the other, the Gbe languages results from the
structural make-up of nominalized phrases in both language types. The VO
versus OV variation that we observe in (49) is a consequence of how the two
languages nominalize verb phrases (rather than how the elements -tɔ́ /-nɔ̀ ,
and -man /-er combine with their specifier). Here again, we are hitting on an
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 161
interface issue, that is, the interface between morphology and syntax.
To make it simple, nominalization is a category changing process whereby
VPs become noun-like constituents in the context of some morphemes. This
can be informally represented in (50), where Fnr is the relevant category
(Marantz 1997):
(50) FPNR
FNR¢
FNR VP
If this is the right characterization, then the word order differences between
Gungbe, on the one hand, and Sranan and Saramaccan, on the other, can be
regarded as consequences of the different licensing properties of Fnr in both
languages. Let us therefore consider how this category is licensed in English
and Gungbe as opposed to Sranan and Saramaccan. Looking at English and
Gungbe, an observation that immediately comes to mind is that such OV
sequences are found in other contexts as well. In English, one also finds OV
order with nominalizations involving the affix -ing as in (51).
(51) a. Street sweep-ing
b. Car steal-ing
In Gungbe, on the other hand, one finds similar OV sequences in VP
nominalizations involving verb reduplication as in (52):
(52) a. àlìò-jí zìzà
road-surface sweep-sweep
‘road sweeping’
b. mótò fìn-fìn
car steal-steal
‘car stealing’
Interestingly, a common denominator for English -ing and Gungbe
reduplicated verbs is that they are also found in progressive constructions as
illustrated in (53a–b) with the verb kú ‘to die’ (Aboh and Smith 2012). Example
(53c) further indicates that when the progressive construction involves a
transitive verb in Gbe, the object must precede the verb. Finally, note that
progressive constructions in Gbe involve a clause-final particle which in
Gungbe is expressed by a floating low tone.
(53) a. John is dy-ing
b. Ján to kú-kû
John PROG die-die.PCL
‘John is dying.’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
162 Competition and selection
spec FNR¢
FNR IP
spec I¢
I VP
spec V¢
V DPobject
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 163
(55) AspP
spec Asp¢
Asp FPNR
to
spec FNR¢
FNR IP
¢
spec I¢
DPobject
I VP
V
tV tDPobject
With regard to intransitive verbs, there is no object DP that can raise to the
subject position [spec IP]. I argue that in such circumstances, the Gbe
languages are left with only one option: the licensing of a null expletive in
[spec IP]. This process in turn translates into a morphological requirement on
the verb which reduplicates. Under this view, verb reduplication is therefore
seen as an alternative to inflection which licenses a null subject (see, for
example, Aboh 2007b, 2009b; Aboh and Smith 2012 for a most recent
discussion). Everything else being equal, we reach the derivation in (56)
which stands for example (53b):
(56) AspP
spec Asp¢
Asp FPNR
to
spec FNR¢
FNR IP
¢
spec I¢
Expl
I VP
V-V
tV
Building on this analysis, Aboh (2005c) and Aboh and Smith (2012) further
argue that in OVV sequences such as in (52), where an object precedes a
reduplicated verb, it must be the case that the process in (56) is active in
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
164 Competition and selection
those environments too. In order for the verb to reduplicate in (52), the object
must not raise to [spec IP] as in (55) but to some other position. The obvious
immediate candidate here is [spec FPnr]. Indeed Aboh (2005c) and Aboh and
Smith (2012) demonstrate that such OVV sequences involve the structure in
(57), in which the object raises to [spec FPnr]. Consequently, an expletive is
inserted in [spec IP], which is licensed by the reduplicated (or say inflected)
verb under I. As the reader can also see from this demonstration, object raising
to [spec FPnr] blocks subsequent movement of IP to the same position (unlike
in representations 55 and 56):
(57) AspP
spec Asp¢
Asp FPNR
spec FNR¢
DPobject
FNR IP
spec I¢
Expl
I VP
V-V
tV tDPobject
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 165
(59) AspP
spec Asp¢
Asp FPNR
spec FNR¢
DP
FNR IP
street
spec I¢
Expl
I VP
sweep-ing
tsweep tDPstreet
spec Asp¢
Asp FPNR
spec FNR¢
FNR IP
Verb
spec I¢
DPobject
I VP
tVerb
tVerb tDPobject
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
166 Competition and selection
As we can see from this representation, such a language will not need a null
expletive in this context. Note also that such a language will display a VO order
in contexts comparable to Gungbe OV(V) or English OV-ing structures. Taking
these observations into account and comparing the Gungbe derivations in (55),
(56), (57), the English representation in (59), and the option presented in (60),
we now reach a typology of gerund-like constructions that will help in
understanding the Sranan and Saramaccan pattern illustrated in (49a), where
it appears that these languages display the VO order where their source
languages display an OV order. Indeed, the discussion here shows that in the
context of gerund-like constructions, languages may choose to lexicalize FPnr,
by raising the object DP in [spec FPnr]. But such languages must have a way of
licensing a null expletive in [spec IP]. In this regard, English resorts to the
suffix -ing while Gungbe employs verb reduplication. The resulting surface
order is OV-affix in English or OVV in Gungbe. On the other hand, languages
may choose not to raise the object in [spec FPnr] but in [spec IP] instead,
thus making the need of an expletive and its concomitant licensing device
unnecessary. Such languages, it appears, must lexicalize the head Fnr
otherwise, for instance, by raising the verb to Fnr. Because these different
options are made available by the human language capacity, they are not
necessarily mutually exclusive and it is perfectly conceivable that the same
language may resort to these different strategies in different contexts.
With regard to the development of the Suriname creoles, the model of
competition and selection that I’m arguing for here, which is based on the
feature pool idea, indicates that the Gbe learners of English were confronted
with these different typological options. Returning now to the discussion in
the previous paragraph, an immediate conclusion that emerges is that if a
language lacks (or loses) the relevant device for licensing a null expletive in
[spec IP], that language will not display an OV order. This is because such a
language will have to raise the DP object in [spec IP] while Fnr is lexicalized
otherwise, for instance, thanks to verb raising as shown in (60). This
conclusion sheds new light on the Suriname creoles: the loss of the -ing
affix is linked to the loss of OV pattern. Returning to the example in (49a)
repeated here as (61a), this would mean that the bracketed sequence
representing the nominalized (or gerund-like) phrase is comparable to
English expressions like that in (61b).
(61) a. [figi-strati]-man
sweep street man
‘street cleaner/sweeper’
b. [house-keep-ing] assistant
Under this characterization, the impossibility of OV sequence in the
Sranan example in (61a) as opposed to the English version (61b) is a direct
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.3 Semantic and syntactic recombination 167
consequence of the loss of affixal -ing. Assuming that the bracketed sequence
represents the structure in FPnr, this amounts to saying that (61a) can be
represented as in (62) (ignoring the AspP projection).
(62) FPNR
spec FNR¢
FNR IP
figi
spec I¢
strati
I VP
tfigi
tfigi tstrati
spec
F¢
FPNR
F° NP
spec FNR¢ -man
FNR IP
figi
spec I¢
strati
I VP
tfigi
tfigi tstrati
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
168 Competition and selection
spec V′
V FPNR
njan
spec FNR′
DP
FNR IP
spec I′
Expl
I VP
lepi-lepi
We now come full circle. During the process of competition and selection
that led to the development of the Suriname creoles, the English inflectional
morpheme -ing was lost. As a consequence the emerging languages
developed another strategy for licensing the relevant functional positions:
the verb moves to Fnr, while the DP object raises to [spec IP] as indicated in
representations (62) and (63). This accounts for the emergence of VO order in
English-like gerund constructions. Yet, because English was in competition
with the Gbe languages which exhibit object raising and non-iconic
reduplication (i.e., OVV) in such contexts, the Suriname creoles did retain
this strategy in a subset of the relevant contexts, such as in secondary
predicates illustrated in (65). The end result of this development is that
the Suriname creoles appear structurally mixed both with regard to
morphosyntax and semantics.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
4.4 Conclusion 169
4.4 Conclusion
The conclusion of this chapter is obvious: creole languages are linguistic
hybrids. They emerged from the recombination of linguistic features from
different languages. While this is the normal situation in every instance of
language acquisition, language change, and language evolution (Kroch 1989a,
1989b, 2001; Kroch and Taylor 1997; DeGraff 2001a, 2001b; Mufwene 2001,
2008; Lightfoot 2006), the case of creoles looks striking at first sight simply
because the recombination involves linguistic features from typologically
different languages (Aboh and Ansaldo 2007). This means that creoles are
noticeable only because of their (contrasting) phenotypes. Nothing in their
structures singles them out as a singular type (Muysken 1981b, 1988).
The approach developed here also shows that creole languages develop
opaque syntactic and semantic features (call it local complexity) which could
not have arisen only in the context of their source languages. If opacity is in
some sense related to complexity, we then reach the situation where the creole
languages, far from being ‘simple’ as some creolists claim, actually hide very
complex morphosyntactic and semantic structures (Aboh and Smith 2009). As
I concluded in Aboh (2009a: 340), the comparison between the Suriname
creoles and their source languages shows clearly that “the notion of
simplicity is completely irrelevant to the understanding of the structure and
the genesis of creole languages, and therefore to the study of language change
or language evolution.” To this, McWhorter (2012: 173) replies “but certainly,
so bold a conclusion cannot stand on two morphemes.” The implicit criticism
is that there is not much that we can learn about the development of new
languages and language acquisition by studying just two morphemes. I surmise
that it is probably easier to hop from one creole to the other counting and
contemplating morphs hoping to discover a holistic theory of simplicity that
would single creoles out as a type.
This book does not adhere to such an easy stance. Instead, it addresses the
development of creoles in the context of the general issue of language
acquisition and language change, by engaging in a rigorous, detailed analysis
of aspects of these languages that shed better light on what possible learning
hypotheses learners develop in the course of acquisition, and how different
linguistic features can be recombined into a coherent system representing the
learner’s I-language. Clearly, such an enterprise can only be made step by step.
As the reader must have noticed, there is indeed a lot to be learned in the study
of just two morphemes. One thing we learned in this chapter is that the
recombination of linguistic features is sensitive to interface properties, hence
the different properties of Saramaccan njan. Because the process is sensitive to
interface properties, it appears not to affect all (sub-) domains of grammar the
same way. In this regard, we learned that while njan points to syntax–semantics
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
170 Competition and selection
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:22:14, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.005
5 The role of vulnerable interfaces in language
change: the case of the D-system
171
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
172 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
The role of vulnerable interfaces in language change 173
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
174 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 175
1
In this work, I will not discuss the influence of Portuguese on Saramaccan syntax; I hope to return
to this issue in future work.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
176 The role of vulnerable interfaces in language change
Aboh (2004a, 2004c) and Aboh and DeGraff (2014). Of particular relevance
here are the properties of the determiners related to definiteness and
specificity.
By specificity, I refer to discourse-anaphoric noun phrases that link back
to referents that are pre-established in the discourse, or known or familiar to
the participants (Pesetsky 1987).2 On the other hand, definiteness refers to a
pre-identified set of referents that is not necessarily discourse-linked. These
informal definitions imply that specificity requires a smaller set of referents
than definiteness. For the sake of discussion, I tentatively submit that
specificity identifies a subset of the contextually given or assumed elements
for which a comment holds. Consistent with Kiss’s (1998) account of
exhaustive or identificational focus, this would mean that specificity
expresses exhaustive identification, unlike definiteness, which only selects a
contextually identified set.
As shown in Aboh (2004a, 2004c, 2006a) and Aboh and DeGraff (2014),
specificity and definiteness combine in some languages, such as Gungbe, in
ways that suggest the following characterization:
(i) A specific definite noun phrase is strongly D(iscourse)-linked and repre-
sents a unique referent assumed to be known to both speaker and hearer,
and which the speaker intends to refer to.
(ii) A specific indefinite noun phrase need not be D-linked. It represents an
existing referent that the hearer may not know about, but which the speaker
has in mind and intends to refer to.
More work is needed in order to reach a proper semantic formulation of the
description in (i) and (ii). For the purposes of this discussion, however, the
description of specific definites in (i) is compatible with the notion of
assertion of existence, as discussed in the literature (e.g., Bickerton
1981), while specific indefinites, described in (ii), recall Ionin’s (2006a)
notion of the speaker’s intent to refer. In the Gbe languages, these two
notions are properties of two different determiners, which exclude each other
and which I assume to be expressions of the nominal left periphery (Aboh
2004a, 2004c, forthcoming, and much related work). In these languages, noun
phrases are never ambiguous with respect to specificity because each kind has a
different marker. On the other hand, nonspecific noun phrases, whether
definite, indefinite, or generic, take no marking even though they may be
modified by an adjective, a demonstrative, a numeral, or a relative clause
(see Aboh and Essegbey 2010 for the discussion on Kwa languages in
general, and Aboh and DeGraff 2014 for a comparison of Haitian Creole and
2
This definition is compatible with Prince’s (1981) notion of assumed familiarity for topics and
suggests that topic and specificity interact (Aboh 2004b).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 177
Gungbe). I refer to such noun phrases as “bare noun phrases.”3 Building on this,
I propose that specificity and definiteness are syntactic features anchored in
distinct heads within the nominal left periphery (Aboh 2004a, 2004c,
forthcoming; Szabolcsi 1994).
The following sections describe how the creoles (Haitian, Sranan,
Saramaccan) and their lexifiers encode definiteness, specificity, and number,
as well as the types of determiners used to express these notions. In so doing,
I pay particular attention to the contexts in which the determiners can occur
and the word order variations inside the noun phrase that they may produce.
I also consider the position of the noun relatively to other nominal modifiers
such as relative clauses, adjectives, demonstratives, and genitives. I will start
with Haitian Creole, Gungbe, and French.
3
I am therefore assuming, contra Enç (1991) and much related work, that definite noun phrases are
not necessarily specific. In fact, the Gbe languages provide numerous examples of definite noun
phrases that derive their definiteness from the context or from the modifiers that they are
associated with (e.g., relative clause, demonstrative, genitive). For a similar study of the noun
phrase in a creole, Gullah, see Mufwene (1986).
4
See also Brousseau and Lumsden (1992), Bruyn (1995a, 1995b), DeGraff (1992, 1995, 2000),
Déprez (2001), Lefebvre (1998), and Aboh and DeGraff (2014) for a discussion of Haitian
Creole; Aboh (2004a, 2004b) and Lefebvre and Brousseau (2002) for studies of Gbe languages;
and Bernstein (1997, 2001a, 2001b) and Giusti (1996, 1997) for the analysis of Romance and
Germanic languages.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
178 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 179
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
180 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 181
5
What matters here is the position of the relative clause vis-à-vis the head noun and the
determiners. The interested reader is referred to Koopman (1982a, 1982b), DeGraff (1992),
Lefebvre (1998), and Aboh (2002, 2005) for the analysis of relative clauses in Haitian and
Gungbe, respectively.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
182 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 183
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
184 The role of vulnerable interfaces in language change
‘Aimé Césaire spoke with Sarkozy, but I just heard that the Aimé Césaire in
question died.’
Similarly, la cannot co-occur with personal pronouns.
(15) *Yo-menm la, yo te chape nan goudougoudou a.
3PL-EMPH DET, 3sg TNS escape in earthquake DET
‘As for them, they escaped from the earthquake.’
Examples (11–15) indicate that the Haitian marker la displays mixed
properties, some of which make it pattern like Gungbe (Gbe), while others
make it similar to French (Romance). In this regard, the discussion in previous
paragraphs suggests that French does not involve a proper specificity marker.
One might conclude from this that this language does not encode specificity
syntactically. Such a conclusion would be misguided, since French makes use
of demonstrative articles (e.g., ce, cette), in combination with the postnominal
deictic locative là, to encode specificity, as shown in (16a). The ungrammatical
example (16b) shows that the postnominal deictic element cannot occur in
postnominal position by itself, unlike in Gungbe and Haitian.
(16) a. Cette guitare-là me plaît beaucoup. [French]
that guitar there me pleases very much
‘I like that guitar (there) very much.’
b. *Guitare-là me plait beaucoup.
guitar-there me pleases very much
Based on examples similar to (16a), it has been argued that the French deictic
locative là is obviously the source of the Haitian specificity marker la, therefore the
lexifier itself contributed to reinforcing the Gbe pattern in Haitian (see also
Lefebvre 1998 and references cited therein). I will not elaborate on the possible
reinforcing role of French here, however, and the interested reader is referred to
Mufwene (2001), Siegel (2004), and Ansaldo (2005, 2009) for some ideas. Instead,
following the approach developed in Chapter 4, I propose that part of the nominal
structure in Haitian has its source in both Gbe and French with both language types
contributing to the emergence of different aspects of the morphosyntax of the noun
phrase in Haitian. Both Gbe and French provided the creators of Haitian Creole
with the relevant triggers for setting the parameters of the DP.
Anticipating the discussion in Section 5.2, the analysis proposed here
follows Szabolcsi (1987, 1994) and much related work in assuming parallels
between the noun phrase and the sentence. It is argued that the structural make-
up of the noun phrase includes three layers, starting with the core predicate
layer where the (lexical) head noun merges and introduces its arguments (e.g.,
in possessive constructions). This layer further extends to a functional layer,
which consists of distinct functional projections responsible for agreement
features and whose specifiers host noun modifiers such as a numeral and an
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 185
adjective. In Cinque’s approach (1994, 1996, 1999, 2004, 2010), this would
mean that nominal modifiers are to the noun phrase what adverbs are to the verb
phrase. Keeping the parallel between the clause and the noun phrase, I refer to
this layer as the nominal inflectional domain, whose head, the nominal
counterpart of clausal I, encodes deixis and may host definite articles
(Szabolcsi 1994).
The nominal inflectional domain is projected under a nominal left periphery
headed by D, which Szabolcsi shows to have properties of a subordinator and to
parallel the clausal left periphery C. Under Rizzi’s (1997) split-C hypothesis,
the clausal periphery involves discrete functional projections responsible for
clause-typing, topicalization, focalization, and finiteness. Adopting this view,
Aboh (2002, 2004a, 2004b, 2004c, 2005c, forthcoming) shows, on the basis of
the Gbe languages, that the nominal left periphery can be split similarly. Aboh
(2004c, forthcoming) further argues that the Gungbe specificity marker
encodes a topic projection (TopP) that projects within the nominal left
periphery. As such, the element lɔ́ links the noun phrase to the discourse. The
number marker lέ , on the other hand expresses a lower projection NumP that
links the nominal left periphery to the lower predicate in a way comparable to
finiteness in the clause. According to Aboh (2004c: 7), the noun phrase in (17a)
can therefore be represented as in (17b), where the nominal left periphery
involves the projections within DP to the left and NumP to the right (see also
Aboh forthcoming for discussion).
(17) a. Távò lɔ́ lέ́
table DET PL
‘the tables (in question)’
b. [DP [D [TopP [FP távò] [Top° lɔ́ [NumP ttávò [Num lέ [FP ttávò]]]]]]] [Aboh 2004c: 7]
I return to the details of this analysis in Section 5.2, where I show that the
variations within Haitian Creole nominal phrase and across Haitian Creole and
the Surinamese creoles can be accounted for by looking more closely into the
nominal left periphery and the properties of this periphery at the discourse–
syntax (or semantics–syntax) interface. But before getting onto this, let us take
a step back and consider the parallels we have observed between Haitian and
Gungbe. What we seem to find here is that the Haitian specificity determiner la
and the number marker yo express the nominal periphery similarly to their
Gungbe counterparts lɔ́ and lέ (Aboh 2004a, 2004c, 2006a).6 According to the
present analysis, such a parallel is regarded as an instance of pattern
transmission, in which a portion of the Gbe-like nominal left periphery has
been transferred to the creole (as explained in Section 5.2.).
6
But see Déprez, Sleeman, and Guella (2011) for a UG-based account of the emergence of
specificity encoding in language acquisition.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
186 The role of vulnerable interfaces in language change
The data further suggest that Haitian Creole shows a split in its noun phrase:
the left periphery has a structure from the Gbe languages while the nominal
inflectional structure derives from French syntax. For instance, recall from the
discussion in Chapter 3 that French and Haitian both have prenominal and
postnominal adjectives, which I assume, following Cinque (1994, 1999, 2010)
and much related work, realize the nominal inflectional domain. I submit that
such a hybrid system, involving a left periphery that has been partially influenced
by Gbe and an inflectional domain that has been influenced by French, can be
accounted for straightforwardly only if one adopts the competition and selection
approach elaborated here. In this regard, it is important to note that while the Gbe
languages typically display a limited set of adjectives denoting color, shape, and
size, most adjectival elements are derived from verbs (Aboh 2007b, Aboh and
Smith 2012). This is different from French where different types of adjectives
occupy different positions within the noun phrase (Cinque 2010). The
recombination in Haitian appears to have reproduced the contrast between
French and Gbe. Thus, the properties of the creole are largely to be found in
the inputs that the learner is exposed to; they are just recombined in new ways.
The inputs generated by these languages feed into the learners’ I-creoles, the
convergence of which leads to the E-creole. Further empirical evidence
supporting this view comes from Sranan and Saramaccan, to which I now turn.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 187
‘The banana (in question) that father brought is what he ate. His stomach is
aching because he ate that banana.’
As the sentences in (19) show, the corresponding English examples
involving bare nouns are ungrammatical.
(19) a *John went to the market to buy banana.
b. *Banana that Daddy bought yesterday.
Like Haitian and Gungbe, Sranan has a singular specificity marker that is
realized as (n)a. Note in sentences (18b) and (20a–b) that the presence of this
element to the left of the noun triggers the specific reading. In example (20a),
the first instance of bana is interpreted as specific (i.e., ‘the banana in question’)
as opposed to bana in (20b), which is understood as (in)definite.
(20) a. Kofi, teki a bana tya gi mi. [Sranan]
Kofi take DET banana carry give 1SG
‘Kofi, give me the banana in question [e.g., the one I brought yesterday].’
b. Kofi, teki bana tya gi mi.
Kofi take banana carry give 1SG
‘Kofi, give me a (or the) banana (s).’
Again, English does not make such a specific-versus-nonspecific distinction
between bare nouns and nouns that occur with the determiner.
(21) a. I like trains very much. [Generic]
b. Every day, I take the train to Amsterdam. [Definite, nonspecific]
In addition, specificity may be encoded in English by various elements
including (in)definite articles, demonstratives, and quantifiers. While Sranan
parallels Haitian and Gungbe in the way it encodes specificity, it differs from
them in lacking a distinct number marker. Hence, the plural counterpart of
example (20a) involves the plural specificity marker, as in (22).
(22) Kofi, teki den bana tya gi mi. [Sranan]
Kofi take DET banana carry give 1SG
‘Kofi, give me the bananas (in question) [e.g., the ones I brought yesterday].’
Therefore, while Gungbe and Haitian Creole lexicalize specificity and number
by means of two separate and independent morphemes, Sranan and Saramaccan
amalgamate the two features into one morpheme. These variations point to a
possible structural difference between the DP in Sranan, on the one hand, and in
Haitian and Gungbe, on the other. That this is probably the correct description is
further suggested by the fact that the specificity determiner precedes the noun in
Sranan, but follows it in Gungbe and Haitian. This pattern is replicated in relative
clauses, where the specificity determiner precedes the relativized noun, which in
turn precedes the relative clause in Sranan (18b). In Gungbe and Haitian,
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
188 The role of vulnerable interfaces in language change
however, the relative clause is sandwiched between the head noun and the
determiners, as already illustrated in (7–8). Given that the sequencing of the
noun phrase in Sranan and Saramaccan parallels that of English, I will argue in
what follows that the morphosyntax of the noun phrase in these creoles bears a
strong legacy of English.
5.1.3 Summary
It appears from the above description that Haitian and Sranan use determiners
(and number markers) in a way parallel to Gungbe and other Gbe languages.
These languages mainly distinguish between specific and nonspecific referents,
thus allowing bare nouns to occur in all possible argument positions. In
addition, such bare noun phrases show no obvious distributive differences
from noun phrases marked by the specificity or the number marker (Aboh
and DeGraff 2014). We can therefore conclude that the distribution of
determiner-like elements is regulated by discourse properties in Gbe and the
creole languages. In French and English, on the other hand, the definite and
indefinite articles are governed by purely syntactic constraints. While English
allows bare mass nouns as well as bare plural count nouns, French excludes
bare nouns except in conjoined plural noun phrases. Both Old French (Mathieu
2009) and Old English allow bare nouns, but again the distribution of such bare
nouns differs from that found in the creoles and the Gbe languages. It appears
from this description that the morphosyntax of the noun phrase in Haitian,
Saramaccan, and Sranan is different from that of the noun phrase in the source
languages. Table 5.1 recapitulates the possible feature combinations and
expressions of determiners in all five languages considered:
As the last three rows of this table show, bare nouns occur in Sranan, Haitian,
and in Gungbe whenever the noun phrase has the following feature combinations:
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.1 Comparing the creoles with their source languages 189
7
This sort of irregularity is often ignored by those who claim that creoles have simple structures. If
we consider learnability issues, the fact that the creoles assign the same form to specific indefinite
plurals and nonspecific noun phrases not only leads to ambiguity and therefore possible acquisi-
tion difficulties but also implies that the nominal module must have been completely reorganized
to allow different mechanisms of licensing and interpretation for these two types of noun phrases.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
190 The role of vulnerable interfaces in language change
Table 5.2 Function and distribution of determiners in creoles and their donor
languages (HC: Haitian Creole; SC: Saramaccan; SR: Sranan; Fr: French;
En: English; Gu: Gungbe)
Determiners HC SC SR Gu Fr En
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 191
Table 5.2 (cont.)
Determiners HC SC SR Gu Fr En
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
192 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 193
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
194 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 195
11
Note that the competition between affixal -man and -er led to the loss of the latter. We can
account for this on phonological ground: -er is phonologically weaker than -man. For instance
the former cannot be stressed while the latter can in some contexts. Cases like these strongly
support the view put forward here, viz., that the more interfaces an element activates the more
likely it is to be retained. In addition, other factors like phonological saliency and wider
distribution appear to play a role.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
196 The role of vulnerable interfaces in language change
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 197
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
198 The role of vulnerable interfaces in language change
represented here as compP. This would mean that the VO order in main clauses
derives from the underlying OV order.
(29) Matrix clauses: there is no overt COMP, thus V raises to COMP:
COMPP
The basic idea of this analysis is that in matrix clauses, comp is not
morphologically realized; consequently the verb must raise there. This
process alters the underlying OV order, giving rise to a surface subject/XP-V-
O order, where XP represents any fronted constituent. In embedded contexts
however verb movement cannot apply as indicated in (30):
(30) Embedded clauses: COMP is realized by a wh-phrase or daβ, V raising is not
possible:
COMPP
Specifier……………COMP……………..Object…………….V
Wh-phrase / dab finite/verb or auxiliary
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 199
OV- and VO-type languages simultaneously. Will she transfer properties of one
language to the other? In case transfer occurs, does it follow a particular
direction or does it target certain specific domains of the architecture of
language (e.g., interfaces)?
The acquisition facts discussed in Müller (1998) shed light on these
questions (see also Müller and Hulk 2000, 2001; Notley, van der Linden, and
Hulk 2007). Starting with monolingual L1 learners of English, French, and
Italian, no specific problem is reported with regard to the acquisition of
subordinate clauses. German children, pass through a phase where they make
some errors (e.g., using non V-final order in embedded clauses) but these errors
are few and children easily overcome this difficulty.
With regard to bilingual children, Müller (1998: 160) reports that:
When children acquire German and a second language, such as, French, Italian or
English, difficulty seems to be asymmetrical, that is, word order in French, Italian,
and English subordinate clauses does not represent a problem space for the children, in
contrast to German, where various error types are attested. This is the case indepen-
dently of whether German is the preferred language or not.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
200 The role of vulnerable interfaces in language change
12
In this regard, Notely, van Linden, and Hulk (2007: 231) argue that “the only elements that can
be retained from Hulk and Müller’s original hypothesis are the two conditions formulated in the
2000 paper, namely (i) the phenomenon must involve the interface between two modules of
grammar, and (ii) overlap between the two grammars must exist.”
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 201
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Table 5.3 The complementizer um/om in German and Dutch (adapted from Ribbert and Kuiken 2010: 43)
Obligatory Obligatory Piet nimmt die Straβenbahn, um nicht zu spat zu kommen [G]
Piet neemt de tram om niet te laat te komen [D]
Piet takes the tram comp not too late to come
‘Pieter takes the tram in order not to be on time.’
Ungrammatical Obligatory *Das ist Ein Aufsatz um zu heulen [G]
Dat is Een werkstuk om te huilen [D]
This is A paper comp to cry
‘This is a paper which could make you cry.’
Ungrammatical Optional *Michael hat versucht um eine Doktorandenstelle zu bekommen
Michael heft gepoogd om een promotieplaats te krijgen. [G]
Michael has tried comp a Ph.D.grant to get [D]
‘Michael has tried to get a Ph.D.grant.’
5.2 On the vulnerability of interfaces 203
perfectly compatible with the view I adopt in this book, viz., in a situation of
contact, comp is particularly vulnerable to transfer regardless of whether the
receiving language is the learner’s L1 or L2.
Summing up, transfer seems immediately available when learners are
confronted with competing variants. The phenomenon does not appear to
follow any particular direction: Adult or late L2 learners transfer properties
of L1 to L2 or vice versa, and bilingual L1 learners transfer properties of an
L1 to another L1. Transfer can affect all sub-domains of grammar, but the
process implies a reorganization of the receiving grammar and seems
particularly sensitive to interfaces in general (e.g., comp, representing
syntax–discourse interface, and argument structure, representing syntax–
semantics interface).
I assume, following Rizzi’s (1997) cartographic approach and much
related work, that the structural make-up of comp, is a complex system
involving elements that clause-type the sentence and peripheral
grammatical items expressing finiteness and (speech act) modality. In
between these two classes of elements appear discourse-related items such
as topic and focus. This sequencing (to which I return in Chapter 6) is
represented in (34):
Force TopicP
Discourse–Syntax
interface
Topic FocusP
Focus FinitenessP
Proposition
Finiteness
I have shown in Aboh (2003b, 2004a, 2004b, 2004c, 2006b, 2007a, 2010e,
forthcoming) and Aboh and Pfau (2011) that the Gbe languages support Rizzi’s
hypothesis in a straightforward manner. The following sentence from Gungbe
indicates that all the hypothesized positions, Force, Topic, Focus, and
Finiteness are realized by distinct markers in Gbe:
(35) Ùn kànbiɔ́ ní òsɔ́ éhè yà ògán wὲ.
1SG ask if horse this TOP chief FOC
mí ní zè è yì ná?
1PL MOOD take 3SG go give
‘I asked whether this horse, we should take it to THE CHIEF.’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
204 The role of vulnerable interfaces in language change
Abstracting away from details about the derivation, this sequencing can be
represented as follows:
(36) ForceP [clause-typing]
TopicP
Discourse–Syntax
interface
Topic FocusP
ya
Focus FinitenessP
we
Finiteness Proposition
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 205
Dem-(PL)-Adj-N-(PL) Bahamian
(PL)-Dem-(PL)-(Adj)-N-(Adj) Kriyol, Tok Pisin, Seychelles, Mauritian
Dem/PL-Adj-N-PL-Dem Berbice Dutch, Sranan
(PL)-(Adj)-N-(PL)-(Adj)-(PL)-Dem Fa d’Ambu, Nubi
(Adj)-N-(Adj)-(Dem)-(PL)-(Dem) Haitian, Papiamentu
(PL)-Adj-N-Dem Sango
(Dem)-PL-N-Adj-(Dem) Lingala
The variation among creoles with the same lexifier (Table 5.4) and across
creoles with different lexifiers (Table 5.5) suggests that such sequencing
cannot simply derive from the different types of lexifiers, but rather must
stem from a more general structural factor such as the properties of the
interfaces.
In this regard, the idea that interfaces are vulnerable to language transfer
offers a new way of looking at such variation. The variation in Tables 5.4 and
5.5 is not surprising if we assume that the clause structure and the nominal
structure include specific syntactic articulations that represent the point of
interaction between syntax and semantics. Extending this idea to the nominal
domain as proposed in Aboh (2004a, 2004c, forthcoming), we reach the
characterization already presented in Section 5.1, viz., that the determiner
system represents a corresponding articulation within the noun phrase.
DP is the highest projection that links the nominal sequence to the
discourse. It is the nominal equivalent of ForceP. NumP, on the other
hand, links the nominal left periphery to the nominal predicate. It is the
nominal equivalent of FinP.
(37) DP
D TopicP
Discourse–Syntax
interface
Topic FocusP
Focus NumberP
Number NP-predicate
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
206 The role of vulnerable interfaces in language change
5.2.3.1 Haitian and Gungbe versus French I submit that Haitian and
Gungbe manifest the structure in (38), in which the number markers lέ /yo
and the specificity markers lɔ́ /la merge in Num and Top, respectively (Aboh
2002, 2004a, 2004c, 2005d).
(38) [DP [D [TopP [Top [±specific] lɔ́ /la [NumP [Num[±plur, ±def] lέ/yo]]]]]] [Haitian, Gungbe]
An immediate consequence of this is that the D position, the nominal
equivalent of force, is never overtly realized in these languages. I refer the
reader to Aboh (forthcoming) for detailed discussion but an obvious
empirical fact compatible with this view is that there is no distributive
difference between bare nouns and nouns that are marked with a determiner
in these languages. If we assume, following Szabolcsi (1987, 1994),
Longobardi (1994) and subsequent work, that D is a subordinating
category, then the fact that bare nouns and determined nouns can be used
interchangeably as arguments in these languages suggests that this position
does not need to be overtly realized (Aboh and DeGraff 2014). This further
correlates with the fact that the so-called articles in these languages, do not
seem to play the grammatical role they play in Germanic and Romance.
Building on previous analyses of Gungbe noun phrases, I therefore argue
that Gungbe and Haitian D-elements occur postnominally because number
and specificity are checked in these languages by overt movement of the
nominal predicate to [spec NumP] and [spec TopP], as shown in (39). Note
that this fronting rule is obligatory in Gungbe and Haitian, but not in French,
where specificity and number may be checked by demonstratives, such as ce,
cet, cette, and ces.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 207
(39) DP
spec D¢
D TopP
spec Top¢
Top[±specific] NumP
spec Num¢
Num°[±plural] åP
l /yo
The same analysis carries over to the relative clause.13 According to Kayne
(1994), the relative clause involves a complementation structure similar to that
in (40), where the determiner and the relative complementizer first merge in D
and C, and the relativized noun raises to [spec CP].
(40) [DP [D the [CP mani [C° that [IP Mary will invite ti . . .]]]]]
Recall from examples (7) and (8), repeated below for convenience as (41)
and (42), that Haitian and Gungbe relative clauses differ from French relative
clauses because they display the following general properties:
1. The use of bare noun relatives.
(41) a. [Moun] ki pa travay p ap touché. [Haitian; DeGraff, p.c.]
people REL NEG work NEG FUT get-paid
‘Those who don’t work won’t get paid.’
b. [Mὲ] ɖê má wà àzɔ́ n màá yí kwέ. [Gungbe]
one REL NEG do work NEG.FUT receive money
‘Anyone who doesn’t work will not get money.’
2. The relative clause follows the noun, but precedes the specificity marker
and/or the number marker.
(42) a. [Fam ki vini wè-u] la. [Haitian; Sylvain 1936: 69]
woman REL come see-2PL DET
‘The woman (in question) who came to visit you.’
b. [Fam ki rete] yo. [Haitian; Sylvain 1936: 62]
woman REL arrest 3PL/
PL
‘The women who arrested them/or the women who were arrested.’
13
The fact that the Gungbe and Haitian definite specificity markers display the same syntactic
properties is supported by the fact that they can also function as clausal determiners, unlike
French articles (Aboh 2002, 2004a, 2004c; DeGraff 2007; Lefebvre 1998).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
208 The role of vulnerable interfaces in language change
Given that relative clauses may precede the specificity marker and a number
marker in these languages, I propose to refine the representation in (16) as in
(22), consistent with Aboh (2002, 2005d):14
(43) [DP [D° [TopP [Top° lɔ́ /la [NumP [Num° lέ/yo [CP [C° ɖĕ/ki [IP. . . ...]]]]]]]]]
Since specificity and number licensing require fronting in Gungbe and
Haitian, I conclude that the relative clause precedes the determiner in these
languages because the relative CP-clause must raise to [spec NumP] and
[spec TopP], as represented in (44a) and (44b), respectively. Representation
(44c) indicates that French does not allow such a fronting rule in its relative
clauses and the article in this language might not qualify as a nominal topic
marker (Aboh 2002, 2004c, 2005d, forthcoming; Aboh and DeGraff 2014,
forthcoming).
(44) a.
b.
c. . . . [DP [D la [NumP [Num° [CP femmei [C° que [IP Marie a invitée ti. . .]]]]]]]
[French]
Under the proposed analysis, the parallels between Haitian and Gungbe
determiner phrases can be regarded as an instance of pattern transmission
because both languages share similar properties with regard to the function
and syntax of the nominal left peripheral elements, such as the specificity
markers lɔ́ /la and the number markers lέ /yo.
At this stage of the discussion, it is worth noting that pattern transmission, as
described here, does not entail a perfect replica of the source system in the
emerging language. Recall from the discussion above that I analyze pattern
transmission as the recombination of syntactic and semantic features of a
source language into a nascent language. Such recombination can only be
partial given that all the properties of the source language cannot be
replicated intact in the emerging language. Accordingly, pattern transmission
need not mean that the source language (whether the substrate or the
superstrate) and the creole have to be isomorphic. Note, for instance, that
14
I agree with Koopman (1982a, 1982b), DeGraff (1992), and Takahashi and Gracanin Yuksek
(2008) that ki is a complementizer, but see Lefebvre (1998) for an alternative analysis.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 209
even though the Gbe determiner encoding the feature specific definite can be
equated with that of Haitian, as far as their syntax is concerned, the same does
not hold true of the Gbe determiner that expresses specific indefinite.
Indeed, all Gbe languages involve a distinct determiner that realizes specific
indefinite, follows the noun phrase, and excludes the specific definite
determiner, as illustrated in (45a–c). Examples (45d–e) indicate that the Gbe
specificity determiners can co-occur with the number marker.
(45) a. Súrù kù mótò lɔ́ wá. [Gungbe]
Suru drive car DET come
‘Suru drove the car back [i.e., he came back with the car in question].’
b. Súrù kù mótò ɖé wá.
Suru drive car DET come
‘Suru drove some car back [i.e., he came back with a certain car]’
c. *Súrù kù mótò lɔ́ ɖé wá.
Suru drive car DET DET come
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
210 The role of vulnerable interfaces in language change
such constraint, however, holds of specific indefinite noun phrases, which are not
necessarily discourse-anaphoric, but only represent a referent that the speaker
intends to refer to (Ionin 2006a, 2006b). Accordingly, the indefinite specific
determiner may be subject to ambiguity (thus allowing conflicting learning
hypotheses) unlike the definite specific determiner.
The following French examples illustrate this asymmetry. The sequence
cette guitare ‘that guitar’ in (47a) is definite specific because it represents
information that is bound by a discourse antecedent known to both the
speaker and the addressee. In example (47b), however, the sequence une
certaine guitare ‘a certain/particular guitar’ is specific indefinite. It does not
necessitate a discourse-antecedent and need not be known to the addressee,
even though it is interpreted as specific.
(47) a. J’ai acheté cette guitare qu’on a vue dans le magasin ce matin. [French]
I have buy that guitar that 3SG has seen in the shop this morning
‘I bought the guitar that we saw in the shop this morning.’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 211
(2006a, 2006b), for instance, shows that errors of article misuse in English
(i.e., overuse of the) mainly arise with specific indefinites. On the other hand,
learners seem to use the appropriately for assumed or pre-established referents
(i.e., specific definite in our terms). I interpret this asymmetry as the result of
the weakness of specific indefinite versus the strength (or prominence) of
specific definite at the syntax–discourse/semantics interface.
Recall also from our previous discussion of specific definite marking in
Haitian that the French use of the postnominal deictic locative là to encode
specificity may have reinforced the Gbe pattern. With regard to specific
indefinites, however, French does not seem to provide any congruent
syntactic paradigm that could have reinforced the Gbe pattern. This would
add to the weak discourse properties of specific indefinites in disfavoring the
emergence of a distinct specific indefinite determiner in Haitian modeled on
that in Gbe. This state of affairs further points to the potential role of
typological (di)similarities among the competing languages in favoring or
disfavoring the emergence of a pattern in the creole. Finally, the absence of a
designated indefinite specificity marker in Haitian contributed to block the
development in this creole (and in the Surinamese creoles) of a marker of
indefinite specific plural noun phrases, as I have already remarked in the
discussion related to Table 5.1. More study is needed before we achieve a
better understanding of these congruence phenomena and how they may favor
certain options while disfavoring others, but see Mufwene (2001) and Ansaldo
(2009) for some ideas.
Summing up, the facts presented here support the idea that domains of
transfer in a language contact situation involve those points (or articulations)
that interact with the interfaces. This chapter focuses on instances concerning
the syntax–discourse/semantic interface, but as I suggested in Chapter 4, it is
reasonable to think that other interfaces (e.g., syntax-phonology/morphology)
are relevant as well, as has already been demonstrated for second language
acquisition (e.g., Goad, White, and Steele 2003; Lardiere 2000).
The next section discusses the properties of the determiner phrase in Sranan
and compares them to those of the determiner phrase in English and Gungbe. It
is argued that Sranan retains only the function of the specificity feature as it
appears in the Gbe languages. The syntax of this feature is, however, modeled
on English syntax.
5.2.3.2 Sranan and English versus Gungbe I propose the internal structure
in (28) for the Sranan noun phrase. However, unlike in Haitian and in Gungbe,
in which Top and Num are realized individually by the markers lɔ́ /la and lέ /yo,
respectively, Sranan lexicalizes these two positions with a single grammatical
item. The Sranan determiner merges under Num but raises to Top to encode an
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
212 The role of vulnerable interfaces in language change
otherwise empty head. This determiner therefore binds two positions: Top
and Num.
(49) DP
spec D¢
D TopP
spec Top¢
Top[±specific] NumP
spec Num¢
Num°[±plural] åP
na/di/den
Representation (49) accounts for the fact that Sranan involves one
determiner form, na, that expresses the feature combination [+specific,
–plural], and one, den, which encodes the feature combination [+specific,
+plural]. This is clearly different from Haitian and Gungbe, where the
expressions of specificity and number are dissociated (39). In this respect,
Sranan appears similar to Germanic and Romance languages, which conflate
definiteness and number in a single morphological form (e.g., le/la, un/une
versus les, des in French, and the/a[singular] versus the/Ø[plural] in English). The
link between Sranan and the Germanic (and Romance) languages further
translates into Num-to-Top movement within the noun phrase, even though
the semantics of the determiner originates in the Gbe languages (49).
Recall that Sranan clusters with Haitian and the Gbe-type languages in
discriminating between specific and nonspecific referents only, which is why
I assume that the determiner-like elements are topic markers in these languages
as well. For instance, these languages have bare nouns that are interpreted as
(in)definite or generic depending on the context, while noun phrases that are
associated with a specificity marker are necessarily discourse-anaphoric.
According to the syntax–discourse interface hypothesis, this property (i.e.,
the split between specificity and definiteness) clearly sets these creoles and
the Gbe-type languages apart from Romance and Germanic languages, where
the two notions often derive from a single determiner overtly realized in the
noun phrase (Aboh and DeGraff 2014). We therefore face a situation where
Sranan displays a hybrid noun phrase system with the function (or semantics)
of the determiner developing from Gbe languages and the syntax resulting from
English (and possibly Portuguese, in the case of Saramaccan).
However, Sranan and Haitian are parallel (and unlike Gbe) with regard to the
expression of the specific indefinite marker. We conclude in Section 5.2.3.1 that
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 213
Haitian lacks a specific indefinite marker due to the weakness of such markers
at the syntax–discourse interface. The same holds for Sranan: the language
does not have a specific indefinite marker. With regard to plural specific
indefinites, some speakers use the combination wan-tu (‘one-two’) with the
meaning of ‘some/certain,’ as in Kofi njan wan tu bana ‘Kofi ate some
bananas.’ These Sranan facts support the analysis in terms of the
vulnerability of interfaces. Indeed, because the specific indefinite marker is
weak in discourse, it allows for ambiguous interpretation at the interfaces and
could not be selected in Haitian or in Sranan.
With this in mind, I conclude that the functional similarities between Sranan
and Gungbe with regard to the specific definite are accounted for by a substrate-
induced feature, where the Gbe languages provided the relevant trigger for the
feature [specificity] to emerge in the creole. The formal licensing of this
feature, however, deviates from the Gbe pattern (or syntax), as a consequence
of the influence of English. The latter can be measured through several
properties that are specific to English and Sranan noun phrases but absent
from the Gbe languages:
1. Modifier–noun order
Both English and Sranan display prenominal modifiers. In Gbe, however, all
modifiers must occur postnominally. Example (50) shows this contrast.
(50) a. den bigi bana [Sranan]
DET big banana
‘the/those big bananas’
b. the big banana(s)
c. àkwékwè dàxó lέ
banana big DET
‘the big bananas’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
214 The role of vulnerable interfaces in language change
b. a datra oso
DET doctor house
‘the doctor’s house’
c. den oso
3PL.POSS house
‘their houses’
In Gungbe, by contrast, possessive constructions manifest the orders
possessor–genitive–possessee and possessee–possessor–genitive.
While sequence (52a) might look similar to the English and Sranan cases in
(51b), it is worth remembering that in the Gbe languages the determiner is
postnominal. In addition, the sequence in (52b) is unavailable in English and
Sranan. Taken together, these facts could be interpreted as evidence that the two
language types (corresponding here to Gungbe versus English) do not use the
same pattern in possessive constructions; Sranan appears to have selected the
English pattern (Aboh 2002, 2004a, 2004c).
(52) a. dàwè lɔ́ sín xwé lɔ́ [Gungbe]
man DET POSS house DET
‘the man’s house [i.e., the particular house of the man in question]’
b. òxwé dàwè lɔ́ tɔ̀ n lɔ́
house man DET POSS DET
‘the man’s house [i.e., the particular house of the man in question]’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 215
(54) a. A kisi wan pikin, dan dati mofo. [Bruyn 1995a: 112, footnote 35]
3SG get DET child, then that mouth
ben langa leki turtur kayman
PST long like turtle caiman
‘She got a child, then that mouth (of her) was long like that of a turtle-caiman.’
b. den pikin disi/dyà [Bruyn 1995b: 265]
DET.PL child this/here
‘these children’
We can therefore conclude that Sranan behaves like Germanic and Romance
languages in making use of both pre- and postnominal positions for nominal
modification (as does Saramaccan). Such word order alternations inside the
noun phrase are simply unavailable in the Gbe languages, where the head noun
always precedes other nominal modifiers, in the rigid order noun-adjective-
numeral-demonstrative, and any word order alteration leads to
ungrammaticality (see Aboh 2004a, 2004c, 2010a).
4. Determiner–noun–[relative] order
Finally, I noted above that Sranan and English display relative clauses in
which the determiner precedes the head noun, which in turn precedes the
relative clause, as illustrated in (55a). The Gungbe example in (55b) indicates
that in the Gbe languages the relative clause must occur between the noun
phrase and the determiner. The relative clause appears to share the same space
as nominal modifiers in Gbe, unlike in Sranan and English (see Bruyn 1995a for
discussion of relative clauses in Sranan).
(55) a. Den uma di mi si na a wowoyo. [Sranan]
DET woman REL 1SG see LOC DET market
‘The women (in question) that I saw at the market.’
b. Náwè [ɖĕ ùn mɔ̀ n tò àxìmὲ] lɔ́ lέ. [Gungbe]
woman REL 1SG see at market DET PL
‘The women (in question) that I saw at the market.’
Even though the Sranan noun phrase is sensitive to the specific versus
nonspecific distinction (like in Gungbe and Haitian), I conclude from these
facts that the syntax of the noun phrase in Sranan differs from that of Gungbe or
Haitian noun phrases. Sranan appears to have inherited English syntax, which
is why noun phrases in both languages share a significant number of syntactic
properties.
Assuming this description, the question arises of how to account for such
asymmetry between two creole groups that are supposed to have the same
substrate Gbe languages. It seems clear to me that the answer to this question
cannot simply be the difference in lexifiers, viz., French versus English.
Indeed, even though these two languages show significant morphosyntactic
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
216 The role of vulnerable interfaces in language change
15
Even though the data discussed here support this scenario, it is still not clear to me what factors
determine the selection of some licensing conditions over others, for instance, first merge of an
operator versus movement of a goal to some designated specifier position. I hope to return to this
issue in future work.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 217
(56) a. DP b. DP
spec D¢ spec D¢
Opi
D TopP D[+specific] NumP
Sranan
5.2.3.3 Agreement in relative clauses in Gbe and the Suriname creoles The
following sentences indicate that, just like English determiners, the Sranan (and
Saramaccan) specificity markers precede the head noun, which precedes the
relative clause.
(57) a. Den uma di mi si na a wowoyo. [Sranan]
DET woman REL 1SG see LOC DET market
‘Those women that I saw at the market.’
b Di womi di mi go kai a kon. [Saramaccan]
DET man REL 1SG go call 3SG come
‘The man that I went out to call came.’
c. The man that Mary called came. [English]
According to the analysis proposed for relative clauses in (56a), Sranan and
Saramaccan are like French and English and unlike Gungbe and Haitian.
Indeed, the Surinamese creoles involve a null operator in [spec TopP] to
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
218 The role of vulnerable interfaces in language change
license specificity. The presence of this null operator in these languages as well
as in Germanic and Romance languages blocks fronting of the noun phrase
complement or relative clause to [spec NumP] and [spec TopP], unlike in
Haitian and Gungbe. Sranan, Saramaccan, and English relative clauses can
therefore be derived as in (58):
(58) [Sranan]
a. [DP [D° [TopP [Top° Den [NumP [Num° tden [CP uma [C° di [IP mi si tuma na a
wowoyo]]]]]]]]]
[Saramaccan]
b. [DP [D° [TopP [Top° Di [NumP [Num° tdi [CP womi [C° di [IP mi go kai twomi]]]]]]]]] a kon
[English]
c. [DP [D The [NumP [Num° [CP man [C° that [IP Mary called tman . . .]]]]]]] came
It is obvious from these representations that all these languages are
characterized by Germanic/Romance determiner syntax (modulo the
projection of TopP), but Saramaccan displays yet another significant
difference: it exhibits number agreement in relative clauses, a fact that is not
found in English but in Gbe instead. Recall from the Saramaccan example (37)
discussed in Chapter 3 and repeated here as (59) that this language involves two
types of relative markers (different from wh-pro forms) which agree with the
head noun (Huttar, Aboh, and Ameka 2013):
(59) a. Di fisi di mi tata kisi bigi.
DEF.PL fish REL.PL 1SG father catch big
‘The fish that my father caught is big.’
b. Dee fisi dee mi tata kisi bigi.
DEF.PL fish REL.PL 1SG father catch big
‘The fish that my father caught are big.’
In (59a), di fisi ‘the fish’ is singular and combines with a relativizer in singular
di, while in (59b) dee fisi ‘the fish/es’ combines with the plural form dee. As I
showed in Chapter 3, this sort of agreement is a counter-example to
exceptionalist views of creolization, according to which creoles typically lack
‘contextual agreement.’ On the other hand, we must ask how such agreement
patterns arose in Saramaccan and why they did not in other Surinamese creoles,
including Sranan and Ndyuka. The phenomenon seems so exceptional that it is
often not discussed in the literature. Indeed, the variation among the Surinamese
creoles is reminiscent of that between the Eastern and Western Gbe languages
discussed in Chapter 2. In Gungbe (Eastern Gbe) for instance, the relative marker
has a unique form ɖě, which does not agree in number with the head noun. This is
illustrated by the Gungbe equivalents of the Saramaccan examples.
(60) a. [[Hwèví [ɖě bàbà cè wlé] lɔ́ ]] klò.
fish REL father 1SG.POSS catch DET big
‘The fish that my father caught was big.’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.2 On the vulnerability of interfaces 219
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
220 The role of vulnerable interfaces in language change
whether Western Gbe speakers formed part of the founder population and set in
motion the restructuring process, whether at some critical point they were dominant
among the Gbe speakers, and whether they were more concentrated in some parts
of Suriname but not others. More study is needed before we can hope to answer this
question conclusively and understand the dynamics of the emergence of norms in
speech communities, and how population factors interact with the selection of
linguistic features.
Another reason, directly related to the hypothesis proposed in this chapter,
relates to the interfaces. Indeed, if we assume Kayne’s (1994) complementation
analysis of relative clauses, Saramaccan and the Western Gbe languages appear
to have the left periphery (65) in relative clauses.
(65) a. DP [Saramaccan]
spec D¢
D TopP
spec Top¢
Top[±specific] NumP
spec Num¢
Num°[±plural] CP
di/dee
spec C¢
C
di/dee IP
(65) b. DP [Gengbe]
spec D¢
D TopP
spec Top¢
Top[±specific] NumP
spec Num¢
Num°[±plural] CP
wo
spec C¢
C
ke-wo IP
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
5.3 Concluding remarks 221
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:25:06, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.006
6 The emergence of the clause left periphery
Chapter 5 shows that the nominal left periphery, the locus of interfaces between
the noun phrase and the discourse, is vulnerable in language contact and
appears to display instances of syntactic recombinations that indicate the
hybrid nature of grammars. This chapter recapitulates the findings in Aboh
(2006b, 2007a) on the emergence of the complementizer system in
Saramaccan. Like in Chapter 5, where I discuss the structural make-up of
comp, I assume here Rizzi’s (1997) cartographic approach, which postulates
structure (34) in the same Chapter 5, repeated below as (1). According to this,
the complementizer system represents the interface between the proposition
and the discourse as illustrated by the boxes to the right. ForceP, the highest
projection, expresses the clause type. Fin(iteness)P, the lowest projection, is
responsible for tense and mood specifications that match those of the
proposition realized within the INFL domain. The space between these two
functional projections hosts various discourse-related projections which
license topic and focus phrases.
(1) ForceP [clause-typing]
Force TopicP
Discourse–Syntax
interface
Topic FocusP
Focus FinitenessP
Finiteness
Proposition
In a series of studies on the left periphery in Gbe and Kwa in general, I have
shown that these languages realize the structure in (1) in a transparent way (e.g.,
Aboh 2003b, 2004a, 2004b, 2006b, 2006c, 2007a, 2009c, 2010c, 2010e, 2010f,
forthcoming; Aboh and Dyakonova 2009). An example is given again in (2a)
222
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
The emergence of the clause left periphery 223
Force TopicP
Discourse–Syntax
interface
Topic FocusP
ya
Focus FinitenessP
we
Finiteness
Proposition
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
224 The emergence of the clause left periphery
1
See Winford (1985, 1993) for the discussion on the different classes of verbs that can select fi as
well as other instances of fi in these Caribbean English Creoles.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
The emergence of the clause left periphery 225
2
This is similar to the distribution of Haitian pou derived from French pour ‘for’ as discussed in
Aboh and DeGraff (2014). I return to this in Section 6.4.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
226 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 227
as in (4) above, repeated below for convenience, as (10a). While (10a) is taken
from Jamaican Creole and can be found in many other creoles (Winford 1985,
1993), the Saramaccan equivalent example in (10b) has been under debate,
with authors disagreeing regarding the syntactic category of fu in such
examples.
(10) a. Im fi kom op ya. [Jamaican Creole; Winford 1985: 589]
3SG fi come up here
‘He ought to come up here.’
b. Amato fu bói dí ganía. [Saramaccan]
Amato fu cook DET chicken
‘Amato should cook the chicken.’
c. Amato kĕ fu Ajawa kísi dí ógífou a matu.
Amato want fu Ajawa catch DET owl LOC jungle
‘Amato wants Ajawa to catch the owl in the jungle.’
d. Amato bói dí ganía fu nján.
Amato cook DET chicken fu eat
‘Amato cooked the chicken to eat.’
In the example (10b), fu expresses deontic modality. In this example, fu
follows the subject of the clause. This is different in the embedded example
(10c) expressing irrealis mood where fu precedes the subject. The latter
example of fu seems to also introduce non-finite purpose clauses as in (10d).3
Looking at the surface order, these examples suggest that Saramaccan involves
two types of fu: fu1 precedes the subject (10c–d) contrary to fu2 which follows
the subject (10b). (See Winford 1985 and Mufwene 1989 for related
discussions.) This is the conclusion reached in Aboh (2006b, 2007a), contra
Damonte (2002) and much related work.
3
Because the Gbe languages and the creoles discussed here do not show any verbal morphology,
‘finiteness’ is determined relative to the context in which the verb occurs (see Aboh 2004a for
discussion on Gbe; DeGraff 1992, 2007 on Haitian Creole; and Veenstra 1996 on Saramaccan).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
228 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 229
Topic FocusP
Focus FinitenessP
Finiteness
fu Proposition
I discuss the topic-focus articulation in Sections 6.3 and 6.4, but the
representation in (16) seems to be compatible with the following example
reported in Veenstra (1996: 116), in which the two positions seem overtly
realized (see also Veenstra 2000).
(17) I táki táa fu a náki dí dágu.
3SG say that fu 3SG hit DET dog
‘You told him to hit the dog.’
Damonte (2002) therefore concludes that Saramaccan unique fu does not
express deontic modality but derives such semantics from the higher selecting
verb (e.g., ábi or musu). Yet, as Aboh (2006b, 2007a) shows, several facts about
Saramaccan suggest that Damonte’s (2002) analysis and related work are
misguided. The following section recapitulates a few shortcomings of this
analysis that are relevant to the discussion.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
230 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 231
c. A fu lóbi dí físi.
3SG fu love DET fish
‘He should love the fish
d. I ku en fu go. [Muysken 1987: 90]
2SG with 3SG fu go
‘You and him should go.’
As these examples show clearly, fu can be immediately preceded by different
types of subjects including weak and strong pronouns, as illustrated by the third
person singular weak form a versus the strong form hen (cf. Veenstra 1996).
Given the ungrammaticality of (19a), it is not clear to me what would motivate
total deletion of ábi only in contexts like (10a), (11a), and (20). There seems to
be no principled explanation for this deletion hypothesis given that the
contracted form á is obligatorily realized even in contexts where it is
immediately preceded by the third person weak pronoun a as in (18a). This
appears to be the most favorable context where one would expect the deletion
of àbi, which could fuse with the subject pronoun and would presumably make
the latter longer and tone-marked as in aá.
Example (18b) offers a similar, though less favorable, context where one
expects deletion to take place, but this does not happen contrary to what tenants
of a unique fu analysis commonly claim (without demonstration). Actually,
absence of ábi deletion in all these contexts conforms to a common observation
in linguistics that elements that usually delete within the verbal domain are
functional items rather than the lexical verb itself. To the best of my knowledge,
no other lexical verb of Saramaccan has been demonstrated to optionally
delete. The only option left for a unique fu analysis would be to suggest that
ábi (or musu for that matter) only deletes before fu, but this hypothesis makes
the argumentation circular and therefore unattractive.
All these facts point to what seems to me the most obvious conclusion:
Saramaccan displays a type of fu that follows the subject and encodes deontic
modality as indicated by all the examples presented thus far. This element,
labeled here as fu2 should be distinguished from another complementizer fu1
which precedes the subject and introduces irrealis and purposive subordinate
clauses (Aboh 2006b, 2007a). Indeed, while the examples involving fu2 in
(10a–b) and (20) express (weak) deontic mood, those involving fu1 in (10c–d),
(14), and (17a) convey irrealis meaning with no particular deontic force.
This observation leads me to another characteristic of fu2. Indeed, if
we assume that the combination ábi fu (and musu fu) in (11) are akin to the
English combinations have to or have got to, then we reach an interesting
characterization already mentioned in Aboh (2006b, 2007a) according to
whom this fu is comparable to English to rather than its correlate for (in
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
232 The emergence of the clause left periphery
for-to complement clauses). This leads to the conclusion that the different
usages of the morpheme fu in Saramaccan cut across those of both for and to
in English. As I show below, English does involve deontic constructions such
as John is to write a new book, where deontic modality seems to be encoded by
the preposition to.
Aboh (2006b) further shows that while Saramaccan speakers interpret
constructions like (11) as conveying strong deontic modality, those in (20)
express weak deontic modality (Palmer 1987). It seems obvious to me that
Damonte’s (2002) approach, which treats the contrast between (11) and (20) as
a mere PF phenomenon, in which the verb ábi is spelled out in one case but not
in the other, cannot account for this semantic difference. In the context of the
current discussion, however, such subtle semantic variations are compatible
with the hypothesis that the morphological form fu occurs in two distinct
syntactic positions associated with distinct semantic effects in Saramaccan:
fu1 versus fu2. This would mean that examples such as (11) are structurally
different from those in (20). The latter are monoclausal as opposed to the
former which are bi-clausal.
Finally, Damonte’s (2002) analysis, according to which Saramaccan unique
fu realizes FinP, as indicated in (16), suggests that this complementizer must
follow topicalized and focused elements but necessarily precedes the subject.
We have already seen in example (20) that this description cannot be adequate,
given that there is one type of fu which follows the subject. The following
examples complete the picture by showing that there is yet another fu which
precedes topicalized and focused elements, contrary to Damonte’s predictions.
In sentence (21a), fu precedes the focused constituent di míi ‘the child.’ In
(21b), fu precedes both the topicalized constituent and the focalized resumptive
strong pronoun hén:
(21) a. A kĕ fu [di míi]i dέ ai nján dí kúku.
3SG want fu1 DET child TOP 3SG eat DET cookie
‘As for the child, he wants him to eat the cookie.’
b. A kĕ fu [di míi]i dέ [hén]i ti nján dí kúku.
3SG want fu1 DET child TOP 3SG eat DET cookie
‘As for the child, he wants HIM to eat the cookie.’
Altogether, these facts indicate that a complementizer fu that is immediately
selected by a verb and introduces an embedded clause must precede the
topicalized and focalized constituents in addition to the subject, while a fu
that encodes weak deontic modality must follow the subject (and,
consequently, topicalized and focused constituents). Consistent with Aboh
(2006b, 2007a), therefore, I argue in the following section that these two
functional elements realize the delimiting projections of the complementizer
system, that is, ForceP and FinP.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 233
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
234 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 235
4
Although Suriname is geographically in South America, it has been treated in creolistics as
belonging in the Caribbean, along with Guyana and French Guyana. Politically both Guyana and
Suriname are included in the Caribbean Community, more commonly referred to as CARICOM.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
236 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 237
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
238 The emergence of the clause left periphery
5
See Aboh (2004a), Aboh and Pfau (2011), and references cited therein for the discussion of
questions in Gbe.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 239
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
240 The emergence of the clause left periphery
Force TopicP
Discourse−Syntax
interface
Topic FocusP
Focus FinitenessP
Finiteness
fu 2 2 Proposition
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 241
while fu2 is comparable to the preposition ‘to.’ In this regard, Winford’s (1985)
discussion on decreolization was meant to explain why certain speakers of
Caribbean English creoles substitute the complementizer fu1/fi1 for ‘for,’ while
deontic fu2/fi2 is replaced by ‘to.’ The discussion in this chapter however
suggests that we need not resort to decreolization to account for this property.
Actually, these two usages are reported in early documents suggesting that they
have always been part of the varieties of English-based creoles. Similarly, these
two usages are not typical of Standard English only, but can be found in non-
standard Englishes as well; so there is no need to resort to decreolization to
account for their emergence in Caribbean English creoles in general. From the
point of view of competition and selection, the different properties of the
morphemes fu suggest the characterization in (44) which cuts across Gbe and
English.
(44) a. fu1/fi1 [Saramaccan, Caribbean English creoles] = ní1/n-V1 [Gungbe (Gbe] =
for [English]
b. fu2/fi2 [Saramaccan, Caribbean English creoles] = ní2/n-V2 [Gungbe (Gbe)] =
to [English]
It therefore appears from this description that during the formation of the
creole, the creole forms fu1/fi1 and fu2/fi2 competed with the English
complementizers ‘for’ and ‘to,’ respectively, in addition to competing with
the Gbe n-V1/2 forms. In the context of this discussion it is important to realize
that English displays deontic expressions in which the prepositions for and to
are used in contexts similar to their creole equivalents fu1 and fu2.7
(45) a. John is to finish his book.
b. John is to come here immediately.
c. John wants for Mary to catch the owl in the jungle.
Similarly, one sometimes encounters examples such as (46) in certain non-
standard English varieties. In these constructions, the complex for-to occurs in
positions similar to those occupied by fu-type elements described by Winford
(1985) in the context of Caribbean English creoles (46b).
(46) a. Always the evenings for to get the men for to do it.
[South West English dialect; Wagner 2004: 168]
b. Mi a hop fi Jan fi go. [Jamaican Creole]
1SG PROG hope fi John fi go
‘I’m hoping for John to go.’ [Winford 1985: 602]
Though these examples are arguably formally distinct, the fact that non-
standard English involves for-to sequences may reinforce learners’ analyses of
modal ‘to’ as being associated with preposition-complementizer ‘for.’ Note
that such a surface analysis is compatible with the Gbe data where the two
7
I thank Anne Zribi-Hertz for bringing these cases to my attention.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
242 The emergence of the clause left periphery
6.1.5.1 fu2, ní2 versus to The discussion has shown that the Saramaccan
deontic marker fu2 occurs in finite sentences where it follows the subject. This
would mean that in such Subject- fu2-V constructions, the subject is marked for
nominative case. Though most creoles don’t have overt case morphology,
Saramaccan does. In this language, the third person weak pronoun exhibits
two case-sensitive forms: a for nominative and hεn/εn for accusative
(cf. Veenstra 1996).8 As shown in example (20a–b), repeated here as (47a–b),
both the nominative form a and the tonic form, expressing emphasis, can
precede deontic fu2. Saramaccan fu2 constructions therefore involve finite
clauses.
(47) a. A fu nján dí físi.
3SG fu2 eat DET fish
‘S/he should/must eat the fish.’ [Bakker, Post, and van der Voort 1994: 255]
b. Hén fu heépi mi. [Wijnen and Alleyne 1987: 43]
3SG fu help 1SG
‘It is HIM who should help me.’
This distribution is different from that of the English preposition to, which
cannot encode modality on its own (48a) but must combine with a verb or a
(modal) auxiliary that expresses tense specifications and which also licenses
nominative forms only (48b). Such constructions can be analyzed similarly to
raising constructions (e.g., John seems happy, John seems to like baseball)
where the subject of the non-finite embedded clause raises to the matrix to be
assigned nominative case.9 Finally, in non-finite embedded contexts where to is
8
Building on Veenstra (1996) one can actually propose that Saramaccan exhibits case distinction
in its entire pronominal system since weak pronouns are restricted to the subject position, while
tonic pronouns typically occur in object positions even though they can be used emphatically in
subject positions as is the case in some Romance, Germanic, and Kwa languages.
9
Mufwene (2005c: 233) discusses examples of headlines or caption where “to” seems to encode
irrealis.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 243
6.1.5.2 fu1, ní1 versus for In the previous sections I showed that Saramaccan
fu1 shares certain semantic and distributive properties with the English
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
244 The emergence of the clause left periphery
preposition for. This is shown again in example (14a), repeated here as (51a)
which we can compare to the English example in (45c) represented again
in (51b).
(51) a. Amato kĕ fu Ajawa kísi Dí ógífou a matu.
Amato want fu Ajawa catch DET owl LOC jungle
‘Amato wants Ajawa to catch the owl in the jungle.’
b. John would prefer for Mary to catch the owl in the jungle.
Aside from this surface similarity, however, there appear to be structural
differences between these two forms which make Saramaccan and Gungbe
pattern alike. Indeed, an important aspect of fu1 and ní1 is that both precede
the topic and focus articulation unlike English ‘for.’ An illustration was given
in (21b) repeated here as (52a) together with a grammatical Gungbe
comparable sentence (52b). In these two examples, the topicalized and
focused elements follow the subordinator. The ungrammatical English
example (52c), indicates that topicalization is excluded in the
corresponding English constructions (Rizzi 1997; Haegeman and Ürögdi
2010a, 2010b).
(52) a. A kĕ fu [dí míi dέ] [hén] nján dí kúku.
3SG want fu1 DET child TOP 3SG eat DET cookie
‘As for the child, he wants HIM to eat the cookie.’
b. Ùn kpɔ́ n ní [ví éhè yà] [àklà lɔ́ wὲ] é ɖù.
1SG look ní1 child DEM TOP cookie DET FOC 3SG DET
‘I checked whether as for this child, he ate THE COOKIE.’
c. *I would prefer for, this child, the cookie he ate.
Accordingly, while Saramaccan fu1 can precede topic and focus constituents
in embedded clauses, English apparent equivalent for cannot (51b vs. 52c).
Similarly to fu2 and ní2, which license a finite clause whose subject receives
nominative case, (unlike English to), the irrealis complementizers fu1 and ní1
also license finite clauses, contrary to English for. In these examples, the
embedded clause displays a pronoun in the nominative form, to the exclusion
of the accusative form.
(53) a. A kĕ fu a/*εn nján dí bakúba. [Saramaccan]
3SG want fu1 3SG-NOM/ACC eat DET banana
‘S/he wants her/him to eat the banana.’
b. Ùn kpɔ́ n ní à/*wè ɖù [àklà lɔ́ ].
1SG look ní1 2SG-NOM/ACC eat cookie DET
‘I checked whether you ate the cookie.’
c. It was not possible for {him/*he} to meet the deadline.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 245
Based on these facts, I conclude that while the Saramaccan morphemes fu1
and fu2 exhibit semantic properties that cut across the Gbe languages and
English, the morphosyntax appears to be mainly mapped on the Gbe
languages. Note, however, that the Gungbe ní1 and ní2 and the Saramaccan
fu1, and fu2 do not show identical distribution and semantic properties.
At this stage of the discussion, a non-trivial question that arises is why
Saramaccan (and possibly other English creoles) have the same form for the
two types of complementizers. Given the proposed analysis, the fact that the
Saramaccan complementizers fu1 and fu2 behave similarly to both Gungbe
n-V1/2 complementizers and English for-to sequences does not automatically
imply that the Saramaccan complementizers should exhibit a unique form.
After all, that Gungbe displays ní1 and ní2 and Saramaccan fu1 and fu2 could be
a simple coincidence. Certain facts about Gbe suggest a different view. Indeed,
the Gbe languages appear to display a striking correlation with regard to the
form of ‘dative’ prepositions and certain tense and mood markers.
As shown in example (54), the Gungbe ‘dative’ preposition also used to
introduce purpose clauses has the form ná in this language. Two additional
examples are given below:
(54) a. Súrù xɔ̀ kὲkέ ná ví lέ ní dó yì wéxɔ̀ mὲ.
Suru buy bicycle for child PL ni2 ride go school’
‘Suru bought a bicycle for the children to cycle to school.’
b. Súrù sán tà ná kòkló lɔ́ .
Suru cut head to chicken DET
‘Suru chop the chicken’s head off.’
Accordingly, while the preposition used to introduce purpose constructions
and new arguments is ná in Gungbe, the irrealis/conditional marker is
expressed as ní1 while the deontic marker expressing subjunctive and
injunctive is realized as ní2. In English, these notions can be expressed by
means of various auxiliaries or modals which in subordinate contexts combine
with for and to functioning as complementizers in non-finite embedded clauses.
It therefore appears that there is an asymmetry between English and Gungbe as
to the form of the complementizers and their relation to the preposition. Yet, the
broader picture from Gbe provides us with a new insight into this apparent
paradox. Here, I present data from an Eastern Gbe language, Fongbe which is
closely related to Gungbe, and two closely related Western Gbe languages,
Ewegbe and Gengbe. I start with Fongbe.
Fongbe (Eastern Gbe), has a deontic marker that occurs between the subject
and the verb and displays the form ní (hence ní2) as illustrated in (55a).
However, this example also shows that Fongbe is different from Gungbe in
that the purpose subordinator or conditional marker in (55b) has the form nú,
which is also identical to the dative preposition shown in (55c).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
246 The emergence of the clause left periphery
10
I’m grateful to James Essegbey for providing me with these examples and for sharing with me
his thoughts about aspects of the analysis presented here.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.1 Complementizer fu in Saramaccan 247
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
248 The emergence of the clause left periphery
expressions of dative, injunctive, and conditional. Only the Fongbe and Gungbe
form ní seems to be an innovation in this Gbe context.
A plausible explanation for this syncretism is that the mood markers and the
prepositions developed from the same form. It is traditionally assumed that the
dative preposition developed from the verb ‘give’ which is realized as ná
across Gbe. Two additional assumptions are needed. First I adopt without
discussion Kayne’s (2002) proposal that prepositions are merged within the
extended domain of the VP (i.e., VP-externally). Second, I follow Roberts and
Roussou’s (2003) proposal that grammaticalization is the reflex of head
movement where the element being grammaticalized moves from the lexical
domain to the functional domain and subsequently acquires the property of
being directly merged in the functional domain. (See Aboh 2009b and
Chapter 7, this book, for a discussion of serial verb constructions.) Put in the
context of the sequencing in (42), these hypotheses lead to the characterization
in (58), in which the lexical verb evolves into a preposition and then into a
subjunctive/injunctive maker and eventually into a conditional/irrealis marker.
Incidentally, it is worth mentioning that in Gungbe, the prospective aspect
marker and the future marker also have the form ná, which could have emerged
from the same source. As the structure shows, each realization corresponds to a
different syntactic position:
(58) Conditional/Irrealis (Gbe)
Tense (Gungbe)
n-V[ -a]
Prospective aspect (Gungbe)
n-V[ -a]
Dative Preposition (Gbe)
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.2 Topic, focus, and interrogation in Saramaccan 249
and form described in Table 6.1, which itself arguably results from the
grammaticalization path described in (58). This view is in support of the
process of recombination of syntactic features developed in this book.
In addition, these facts allow us to understand the role of interfaces in the
development of new linguistic patterns. Our discussion has shown that even
though the Gbe languages show a certain degree of variation with regard to
the forms selected, these languages exhibit a syncretism between dative
prepositions, complementizer-prepositions, and deontic mood markers. In all
these languages, the similar forms are used to express these distinct
grammatical functions. Because the complementizer prepositions and the
deontic mood markers relate to the complementizer domain which has been
shown to be vulnerable during acquisition in a multilingual context, this
contributes to favor cross-linguistic influence between Gbe and English such
that the functions of n-V2 in Gbe were extended to cover the usage of English
to in modal contexts.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
250 The emergence of the clause left periphery
11
According to Smith and Cardoso (2004), the reanalysis of simple nasal consonants into
prenasalized consonants in Saramaccan does not only affect words of Gbe origin but also
Germanic and Romance words. For instance, English words such as make and knife are realized
as mbei and ndefi, while words of Portuguese origin such as anel ‘ring’ and demolir ‘demolish’
are realized as andélu and dimbolí.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.2 Topic, focus, and interrogation in Saramaccan 251
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
252 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.3 Predicate fronting and doubling 253
languages in general (Aboh and Essegbey 2010), and which distinguish them
from English. The data discussed thus far lead to the characterization in (67) for
the three languages:
(67) ForceP [clause-typing]
Force TopicP
Discourse–Syntax
interface
Topic FocusP
d ; ya
Focus FinitenessP
w
Finiteness
Proposition
This representation shows that while all the languages involve relevant markers
to realize Force (English: that/for; Saramaccan: táa/fu1; Gungbe: ɖɔ̀ /ní1) and Fin
(English: to; Saramaccan: fu2; Gungbe: ní2), only Saramaccan and Gungbe
display topic markers (Saramaccan: dέ ; Gungbe yà) alongside the focus marker
(Saramaccan/Gungbe: wὲ ). It is therefore clear that the backbone of the
complementizer system in Saramaccan follows the Gbe pattern, though the
relevant markers embed semantic (and morphosyntactic) properties of certain
English complementizers (e.g., for, to). While the architecture of the
complementizer system as proposed by Rizzi (1997) is provided by UG, the
discussion here shows that the Gbe languages combined with English to trigger
the emergence of very specific properties of the Saramaccan clausal left periphery.
In the following section, I discuss predicate fronting with doubling which further
enhances the similarities between the Surinamese creoles and the Gbe languages.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
254 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.3 Predicate fronting and doubling 255
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
256 The emergence of the clause left periphery
hand, the Gbe or Kwa-type marker does not require exhaustive reading and can
be used in a wide range of contexts, including presentational focus (Aboh
2006c, forthcoming). According to Durrleman-Tame’s analysis, the different
patterns in (72) translate into two distinct structures, as shown in (73):
(73) Saramaccan/Gungbe Jamaican/Haitian
Exh(austive)P
FocP
spec Exh¢
spec Foc¢
XPFOCUS Exh FocP
Foc FinP se/a
w spec Foc¢
XPFOCUS XPFOCUS
Foc FinP
XPFOCUS
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.3 Predicate fronting and doubling 257
Among the focus-marking strategies which may combine in the derivation of verb focus
doubling constructions in Haitian, low contrastive reiteration, clefting and adverbial
focus markers are all available in French . . . Focus Raising (i.e., the raising of a focused
constituent to the clause periphery, without clefting) is not, on the other hand a natural
focus-marking strategy in French, where focus is generally phrase-final, especially for
verbs . . . This suggests that the one non-French property involved in the diachronic
development of Haitian verb focus with doubling was Focus Raising, a property
characteristically attested in Gbe.
Assuming this analysis is correct, I conclude that the contrast in (72) does
not illustrate a simple surface variation but hides a structural difference that
sets the Gbe (Kwa) languages and Saramaccan apart from other Caribbean
creoles.
A question that has been extensively discussed in the literature is how to
account for the genesis and the structure of the doubling structures in (68), (69),
and (70). I will not review the literature here but I refer the interested reader to
Glaude and Zribi-Hertz (2012) for a recent state of the art as well as a new
alternative. With regard to the genesis of such structures in Atlantic creoles,
these were considered typical of Atlantic creoles and certain Niger-Congo
languages (e.g., Kwa). As such they were seen as empirical evidence for
substrate influence. Though these constructions are highly productive in
Kwa, it is important to realize that they are also present in typologically
diverse languages such as Germanic, Romance, Semitic, Slavic, and Semitic
(Aboh and Dyakonova 2009). Accordingly, that such structures are attested in
Atlantic creoles cannot simply be claimed to illustrate substrate influence.
Instead, we need to adopt Glaude and Zribi-Hertz’s (2012) more nuanced
approach where only certain properties of these structures can be related to
the relevant African languages. Building on Aboh (2005d) and Aboh and
Dyakonova (2009), I assume that Predicate Fronting (with doubling) for the
purpose of focusing and event-relativization is such a relevant property.
Starting with the structural account for doubling, Aboh and Dyakonova
(2009) argue, on the basis of Gungbe and Russian, that sequences such as in
(68) result from parallel chains. Technical details aside, the argument goes as
follows: In Gbe languages of the Gungbe-type, the verb must be licensed for
aspect, which requires the verb root to move to the immediately relevant aspect
position. This creates a first chain Vasp-V. In the context of verb focus, the verb
root is simultaneously attracted by an aspect head within the INFL domain as
well as the focus head within the clausal left periphery, thus creating a second
chain Vfocus-V. The tail of these two chains is deleted at spell out, thus leading
to the phonetic realization of the two heads Vfocus and Vasp, as in (74)
(Chomsky 1995, 2008; Aboh 2009c).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
258 The emergence of the clause left periphery
(74)
FocP
spec Foc¢
Foc FinP
VFOCUS
TP
AspP
spec Asp¢
VASP VP
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.3 Predicate fronting and doubling 259
12
Mufwene (1987: 80) discusses examples of related predicate focus constructions in Bantu
where the focused VP is marked “as a verbal noun.” Such findings obviously reinforce the
view developed here where in a situation of competition and selection, the Gbe patterns are
reinforced by congruent Bantu patterns (see also Ansaldo 2009).
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
260 The emergence of the clause left periphery
draws on Gungbe but I assume that comparable facts can be demonstrated for
Saramaccan.
1. As noted above, event relatives involve the relative marker in Gungbe. In
addition, these constructions exclude the focus marker. Examples (79b–c)
are ungrammatical whether the focus marker precedes or follows the rela-
tivizer. Example (79d) indicates that the presence of the focus marker is
excluded even in contexts where the relative marker is not overtly realized.
Put together, these examples show that event relativization is formally
distinct from verb focusing.
(79) a. Wlé *(ɖĕ) mí wlé àgásá ɖàxó lɔ́ lέ vέ ná Kòfí.
catch REL 1PL catch crab big DET PL hurt for Kofi
‘The fact that we caught the big crabs annoyed Kofi.’
b. *Wlé wὲ ɖĕ mí wlé àgásá ɖàxó lɔ́ lέ vέ ná Kòfí
catch FOC REL 1PL catch crab big DET PL hurt for Kofi
c. *Wlé wὲ ɖĕ mí wlé àgásá ɖàxó lɔ́ lέ vέ ná Kòfí
catch FOC REL 1PL catch crab big DET PL hurt for Kofi
d. *Wlé wὲ mí wlé àgásá ɖàxó lɔ́ lέ vέ ná Kòfí
catch FOC 1PL catch crab big DET PL hurt for Kofi
2. A typical property of the Gbe languages is that they involve a clausal
determiner, that is, an article that marks nominalized clauses. In Gungbe
this article is lɔ́ which also marks noun phrases as indicated in (80)
(Lefebvre 1998; Lefebvre and Brousseau 2002; Aboh 2004a).
(80) a. Àgásá ɖàxó lɔ́ nyɔ́ n tàun.
crab big DET good very
‘The big crab is very nice.’
b. Ðě Àgàsú hɔ̀ n yì lɔ́ má nyɔ́ n kpálí.
REL Agasu flee go DET NEG good at.all
‘That Agasu fled is not nice at all.’
Because event relatives are nominal clauses, they typically involve this
determiner. This is not the case with simple verb focus constructions, which
tend to resist this determiner.
(81) a. [[Hɔ̀ n ɖĕ Àgàsú hɔ̀ n yì] lɔ́ ] vέ ná Kòfí.
flee REL Agasu flee go DET annoy for Kofi
‘That Agasu fled annoyed Kofi.’
b. *[[Hɔ̀ n Àgàsú hɔ̀ n yì] lɔ́ ]
flee Agasu flee go DET
‘Agasu FLED.’
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.3 Predicate fronting and doubling 261
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
262 The emergence of the clause left periphery
while the second part of the clause suggests that the meat of the crab was not
delicious.
Another difference between nominal relative clauses and event relatives is
that nominal relatives typically occupy the same syntactic positions as
common noun phrases (viz., as external or internal arguments, or as
adjunct), while event relatives function only as clausal subjects. The
following examples show that while nominal phrases can function as
complements, event relatives cannot. Accordingly, (85b) is ungrammatical,
unlike example (84). For (85b) to be grammatical the factive event has to
become a subject as in (85c).
(85) a. Súrù sà [àgásá ɖĕ Kòfí wlé lɔ́ ].
Suru sell crab REL Kofi catch DET
‘Suru sold the crab that we caught.’
b. *Súrù gbέ [àgásá lɔ́ ] ɖĕ Kòfí wlé]].
Suru refuse crab DET REL Kofi catch
‘Suru did not accept the fact that we caught the crab.’
c. [àgásá lɔ́ ] ɖĕ Kòfí wlé] má yì xέ Súrù kpálí.
crab DET REL Kofi catch NEG go with Suru at.all
‘The fact that we caught the crab did not please Suru at all.’
Furthermore, while nominal relative clauses involve a wide range of
constituents, event relatives involve the verb or the internal argument but
exclude the external argument. This is shown by sentences (86):
(86) a. [[Gán lɔ́ ] ɖĕ náwè lɔ́ kàn wé xlán] vέ ná mì.
chief DET REL woman DET write letter to annoy for me
‘The fact that that woman wrote a letter to THE CHIEF annoyed me.’
b. [[Wé lɔ́ ] ɖĕ náwè lɔ́ kàn xlán ògán lɔ́ ]] vέ ná mì.
letter DET REL woman DET write to chief DET annoy for Me
‘The fact that the woman wrote THAT LETTER to the chief annoyed me.’
c. ??/*[[Náwè lɔ́ ] ɖĕ kàn wé xlán ògán lɔ́ ] vέ ná mì.
woman DET REL write letter to chief DET annoy to Me
‘The fact that that woman wrote a letter to the chief annoyed me.’
Sentences (86) are very illustrative of the properties of event relatives in that
they show that while the factive reading can be obtained by fronting the internal
arguments (including double objects) or the verb expressing the Event head,
fronting of the external argument does not yield the same result.
According to Aboh’s (2005d) detailed analysis, the contrast in (86) is
indicative of the fact that event relativization does not target the verb root
(similarly to focus constructions) but rather targets the event phrase (i.e.,
minimally the verb phrase). Thus, the fronted internal DP argument and the
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.3 Predicate fronting and doubling 263
verb heading the event relate to aktionsart and can therefore be regarded as part
of a functional projection denoting the event. This would mean that the internal
argument or the verb bears an event feature that can be attracted by a head
within the clausal periphery. Consistent with this hypothesis, a factive clause
such as the bracketed sequence in (85c) can be represented as in (87), in which
CP is a shorthand for the articulated structure in (34). In this example, the verb
raises to Asp to be licensed for aspect, while the DP object is attracted within
the CP field for relativization.
(87) CP
spec C¢
DPOBJECT
C TP
AspP
spec Asp¢
VASP VP
V
DPOBJECT
On the other hand, when the event head (viz., the verb) is attracted, I
hypothesize that it is the remnant VP that is fronted (after the DP object has
raised to its licensing position). A possible argument in favor of this hypothesis
is that the fronted relativized event can be combined with a determiner as in the
Saramaccan example (76a). As discussed in Collins (1994), similar examples
are found in Fongbe and other Kwa languages as well. This would mean that a
factive clause with a fronted verb can be represented as in (88), ignoring the
intermediate movement of the object to its licensing position.
(88) CP
spec C¢
VP
C TP
AspP
spec Asp¢
VASP VP
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
264 The emergence of the clause left periphery
As in the case of focus constructions, doubling here results from the spell out
of two heads of two parallel chains that have a common tail inside VP. The
analysis proposed here indicates that factive clauses involving fronting of the
DP internal argument and the remnant VP fundamentally share the same
structure, though they are felicitous under different pragmatic conditions.
Indeed, factive clauses with verb fronting and doubling generally tend to
imply emphasis and topicality, but I hope to return to these semantic
differences in future work.
What matters for the present discussion is that if we accept to extend the
analysis just proposed to the Saramaccan example in (76a) as well as to the
French and English constructions in (77) and (78), we reach the typological
conclusion according to which Gungbe and Saramaccan involve internal
argument or remnant VP fronting (with doubling) while English and French
display insertion of an expletive factive phrase ‘the fact’ or its covert
equivalent. Following Aboh’s (2005d) analysis of such English and French
constructions, I therefore conclude that the examples in (77) and (78) can be
represented as in (89):
(89) CP
spec C¢
XPTHE FACT/Ø
C TP
that/que
AspP
VP
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.4 The HC left periphery 265
the focused verb is attracted by two higher heads simultaneously. While this
could be seen as raising structure typical of focus constructions, the discussion
has shown that a similar strategy arises in event factives as well. Here again
Saramaccan partly patterns like Gbe in allowing event factive clauses to derive
from the raising of the verb phrase to the clausal left periphery in conjunction
with a doublet that occurs internally to the proposition.
If we agree that the typological options in (90) are made available by UG,
the fact that Saramaccan, like many other Caribbean creoles, developed
the strategy in (90a) alongside Focus Raising but excludes (90c) cannot be
accidental. Instead, such a development derives from the sort of recombination
of syntactic features that I have argued for in this book. Indeed, these facts
constitute strong evidence that the relevant syntactic features for Focus Raising
and factive clauses are active in both Saramaccan and Gbe. In addition, that
these features activate the left periphery, which represents an interface between
the proposition and the discourse, further supports the view defended in
this book, viz., that interfaces are particularly vulnerable for syntactic
recombination. Before concluding this chapter, let me briefly address a
question that the alert reader might be wondering about, viz., What about
Haitian Creole? Does it show (part of) the different constructions reviewed
here that could further support the recombination analysis proposed in this
book?
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
266 The emergence of the clause left periphery
Interestingly, however, DeGraff (2007: 109) also shows that HC pou can
introduce full finite clauses as illustrated in (93):
(93) a. Kouto sa a pa fèt pou li Koupe pen. [Haitian]
knife DEM DET NEG make for 3SG Cut bread
‘This knife is not made for cutting bread.’
b. Li te ale nan fèt la pou li te ka fè
3SG TNS go to party DET for 3SG TNS capable do
yon ti danse, men lè li rive, pa te gen mizik.
DET little dance but when 3SG arrive NEG TNS have music
‘S/he went to the party to dance a bit, but when s/he arrived there was no
music.’
Thus, unlike Modern French, Haitian Creole pou can introduce both
subjectless non-finite clauses and full finite clauses (but see Aboh and
DeGraff, forthcoming for discussion). On the other hand, Haitian also shows
evidence of modal pou as illustrated in the three examples in (94):
(94) a. Se Bouki ki pou te vini. [DeGraff 2007: 109]
FOC Bouki COMP MOD TNS come
‘It’s Bouki who had to come’
b. Tout moun pou vin lan fèt La [Sterlin 1989: 143]
all people pou come to party DET
‘Everyone must/should come to the party.’
c. Jan pou danse. [Glaude 2012: 60]
Jean MOD danse
‘John is favorable to dancing.’
‘John can dance.’
The different interpretations provided by these authors suggest that pou can
convey different meanings when used as a modal. More study is needed before
we understand the semantics and morphosyntax of this element. What matters
for the current discussion, however, is that this instance of pou follows the
subject of the clause. A surface comparison therefore suggests that Haitian
Creole, just like Saramaccan and Gungbe, has two types of pou: pou1
introduces purpose or irrealis clauses and precedes the subject as indicated in
(93). pou2, on the other hand, follows the subject and encodes modality. In
addition, the focus example in (94a) suggests that pou2 follows focused
elements, which in Haitian Creole are marked by the element se (in a way
similar to many Caribbean creoles, cf. Section 6.3.1). The Haitian focus marker
is further illustrated in example (95), where it marks the focused verb. As in the
case of Kwa languages, Atlantic creoles and many other languages (Aboh and
Dyakonova 2009) such constructions lead to doubling (Harbour 2008; Glaude
and Zribi-Hertz 2012):
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
6.5 Conclusion 267
Force TopicP
Discourse–Syntax
interface
Topic FocusP
Focus FinitenessP
Finiteness
Proposition
6.5 Conclusion
The facts that I discussed in this chapter show that the left periphery of
Saramaccan and Haitian to some extent displays morphosyntactic properties
that combine specific syntactic features of the Gbe languages with those of the
lexifier European languages. Accordingly, both the Gbe languages and the
creoles display distinct discourse-related markers that occupy topic (TopP)
and focus positions (FocP). These two positions are sandwiched between
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
268 The emergence of the clause left periphery
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:27:21, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.007
7 The emergence of serial verb constructions
1
I’m very grateful to Michel DeGraff for his insightful comments and suggestions on a previous
version of this chapter as well as for sharing his judgments and intuitions on Haitian Creole with
me. I’m also thankful to Herby Glaude for providing me with additional data on SVC in Haitian.
2
In the context of creole languages, there is a debate as to whether the construction is found in
Indian Ocean Creoles or not (see Bickerton 1989, 1990b versus Seuren 1990a, 1990b, 1991).
269
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
270 The emergence of serial verb constructions
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.1 The clause in Haitian Creole, Saramaccan, and Gbe 271
own inadequacies and theoretical drawbacks which show that the scenario is
not this simple. I showed in Chapter 3 that the Relexification Hypothesis as
defined by Lefebvre (1998) is untenable, consistent with Singler (1996) and
DeGraff (2002, 2003), among others. Nonetheless, Section 7.1 presents
ample empirical evidence that lends support to the view that SVCs
developed in creole languages as a result of cross-linguistic influence.
Building on this, Section 7.2 argues that cross-linguistic influence must be
defined in a very precise way, such that it is compatible with the theory of
competition and selection adopted in this book. Indeed, a new analysis of
SVCs presented in Sections 7.3 and 7.4 indicates that verb serialization is
widely represented in languages of the world (including Indo-European)
where it takes the form of clause union, also referred to as restructuring
in the generative framework and defined as a process that unites two clauses
into one (Rizzi 1982; Wurmbrand 2001, 2004; Cinque 2004). Under this
finding, serialization appears to be a property of UG for which there does not
seem to be any obvious distinctive parameter that could set serializing
languages apart from non-serializing ones (Aboh 2003a, 2009b). This
means that, in order to demonstrate substrate influence, we can no longer
rely on a superficial one-to-one mapping of SVCs in the creoles onto
apparently corresponding SVCs in the relevant substrate languages. In
Section 7.5, I show that substrate influence can be measured by looking at
both similarities and differences between the emerging creoles and their
source languages. Under a theory of creole formation that assumes
competition and selection, with creoles being normal instances of hybrid
grammars, I account for these facts arguing that the creoles display
recombinations of linguistic features from both the superstrate and the
substrate. Section 7.6 concludes the discussion.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
272 The emergence of serial verb constructions
Table 7.1 Some contrasting properties between the creoles and their source
languages
It appears that in contexts where the creoles and the substrates share some
morphosyntactic properties, these are absent in the lexifier languages and vice
versa. As a result, Haitian and Saramaccan are more like Gbe languages vis-à-
vis their inflectional morphology and the expression and interpretation of
Aspect than they are like their respective superstrate languages.
Interestingly, this language group often displays aspect markers that evolved
from verbs. A case in point is the habitual marker in Gungbe, which evolved
from the verb nɔ̀ ‘to stay’, as illustrated in (2).
(2) Súrù nɔ̀ nɔ̀ xwégbè.
Suru HAB stay house
‘Suru used to/habitually stay/s at home.’
Similarly, the Saramaccan irrealis/future marker ó is assumed to have
developed from the English verb go, while the progressive aspect marker tá,
which can also be used to mean to wait/to stay, is analyzed as having developed
from English stand (e.g., Veenstra 1996: 13; Winford and Migge 2007: 85, 89;
see also van den Berg and Aboh 2013 on the morphosyntax of kaba/kba in
Sranan). The same could be said of the Haitian mood markers dwe and met,
which are derived from the French verbs devoir ‘must’3 and (per)mettre ‘allow’
(e.g., Glaude p.c. January 24, 2013).
Together, these facts indicate that the Gbe languages, Haitian, and
Saramaccan exhibit a cluster of properties that all appear compatible with the
emergence of verb serialization: They are isolating, they do not exhibit
agreement or inflectional morphology on the verb. They involve TMA
markers which have typically evolved from verbs (sometimes adverbs) and
therefore exhibit some form of auxiliation where both the lexical verb and its
cognate INFL-marker co-occur.
3
In the singular forms, the present tense of devoir was pronounced [dwε(t)] in the seventeenth
century.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.2 Serial verb constructions: a brief description 273
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
274 The emergence of serial verb constructions
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.2 Serial verb constructions: a brief description 275
aim of this section is to present the reader with very general aspects of SVCs
that bear on the positions defended in this book.
SVCs typically display a single overt structural subject shared by all the
verbs. In example (6), for instance, the same noun phrase is interpreted as
subject of the first and second verbs, that is, the bandits beat and killed the man.
4
Such INFL spreading is found in many Kwa languages (e.g., Akan, Baule) as well as typologi-
cally different languages (Aikhenvald 2006: 42ff.).
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
276 The emergence of serial verb constructions
5
Michel DeGraff (p.c. June 21, 2009) remarks that this constraint does not hold for Haitian, which
disallows extraction:
(i) *Ki moun Bouki pran flè bay?
wh- person Bouki take flower/s give
‘Who did Bouki gave flowers?’
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.2 Serial verb constructions: a brief description 277
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
278 The emergence of serial verb constructions
It appears from these examples that it is not possible to negate only a sub-
event of an SVC (13a–b), but nothing excludes this in a non-serializing
construction (13c–d).
6
Though this is by no means linguistic evidence, it is worth mentioning that there is a saying in
Gungbe which is compatible with the description provided here. As the reader may see, the
selected verb corresponds to “to see” not “to look.”
(i) Núkún ɖé mɔ̀ n/*kpɔ́ n nú má hù hwὲ.
eye REL see/look thing NEG kill fault
‘An eye that sees (something) is not guilty [i.e., anyone who happens to see something
by chance cannot be judged guilty].’
7
I thank Michel DeGraff for his comments on this issue.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.2 Serial verb constructions: a brief description 279
While this may look like a ‘natural’ restriction on V2, there does not seem to be
any syntactic principle that will force V2 to combine with motion verbs only when
it implies direction, transfer, or displacement. Instead, the constraints on such verb
combinations seem to be semantic in nature. Indeed, in Gungbe, for instance, it is
perfectly possible to utter sentence (16) in which the three verbs in the series (in
boldface) include yì ‘go’ following a verb of saying ɖɔ̀ xó ‘say word’, which does
not imply motion. Also notice from the translation that the whole sentence does not
imply movement (though this is possible in a metaphorical sense).
(16) Àzɔ́ nví lέ ɖɔ̀ [xó] yì jὲ gán sín àsì jí
apprentice PL say word go reach chief POSS wife on
bɔ̀ gán sè . . .
COORD chief hear
‘The apprentices talked about the chief’s wife and the chief heard them . . . ’
It therefore appears that the restrictions on the Saramaccan example in (15)
relate to the semantic specifications of the functional projection that hosts V1.
Similarly, DeGraff (1992, 2007) identifies three types of serial verbs in Haitian:
motion (17a), transfer (17b), and comparison (17c).
(17) a. Bouki voye timoun yo ale lekòl. [DeGraff 1992: 57, 58]
Bouki send kids PL go school
‘Bouki sent the kids to school.’
b. Boukinèt te pran yon flè bay Malis.
Boukinet TNS take a flower give Malis
‘Boukinèt gave a flower to Malis.’
c. Boukinèt mache (plis) pase Mari.
Boukinèt walk more pass Mari
‘Boukinèt walked more than Mary.’
With regard to SVCs encoding motion or transfer, DeGraff (1992: 58)
observes that verbs that realize the position V2 are selected from a smaller set
than verbs that encode V1: in motion-type SVCs, V2 is typically taken from (e.g.,
vini, ale, rive, soti), while it is realized as (bay, pote, mete . . .) in transfer-type
SVCs. This would mean that in Haitian serialization, V1[motion/transfer] exhibits
semantic selectional requirements on the lexical verbs V2 it co-occurs with.
Similar semantic factors could be at the source of the impossibility of example
(18a) in Haitian Creole, but not in the corresponding Gungbe example (18b).8
(18) a. *Jan kuit pen an manje.
b. Ján mὲn blέɖì lɔ́ ɖù.
John cook bread DET eat
‘John cooked bread and ate it.’
8
I thank Michel DeGraff for calling my attention to these facts in Haitian.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
280 The emergence of serial verb constructions
In this case, V1[kuit: cook] cannot combine with a lexical verb V2 encoding ‘to
eat’. I conclude from these facts that verb combinations in SVCs are subject to
semantic restrictions that govern the choice of verbs that can express V2 in
combination with V1. For the time being, I have no comprehensive analysis to
offer as to what semantic principles govern the various constraints illustrated in
these examples. What matters for our discussion, though, is that SVCs do not
involve a mere syntactic combination of independent lexical verbs into one
clause. Instead, these data suggest that there is a semantic and syntactic
dependence on the verbs in a series. This latter point suggests that previous
analyses of SVCs as involving two independent verbs that discharge their
thematic-roles on shared arguments might not be right. I return to this
discussion below.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.3 Serial verb constructions: a reanalysis 281
While this view somehow holds of example (19), the facts in (18) indicate
that things are not this simple. Indeed, the structural discrepancies between the
creoles and the substrates, such as illustrated in example (18), point to the fact
that the emergence of SVCs in the creoles cannot be attributed exclusively to
substrate transfer. Lefebvre’s (1998) account, provides no principled
explanation for why very productive Gbe structures such as (18b), which
must have been part of the PLD that made speakers “deduce the availability
of verb serialization,” are not retained in Haitian Creole.
In what follows, I report the conclusions in Aboh (2009b), where it appears
that there can be no serializing parameter such that two or more heads of
predicates can share the same internal arguments. This would mean that
theories of SVCs that posit complex predicates cannot be on the right track. In
this particular regard, it also appears that the so-called verb series share similar
properties with structures such as the American English examples “he will go talk
to his advisor today” or “I requested that she come discuss this problem with me
in person,” discussed in Jaeggli and Hyams (1993).9 Accordingly, the
phenomenon of verb sequences is more common than often assumed in the
literature. In order to contextualize the discussion, let us first step back and
consider some approaches to SVCs that have been explored in the literature.
For the past twenty years or so, various parameters have been evoked to
explain serialization: lack of prepositions (Bickerton 1981), lack of inflectional
morphology and hence verb raising (Muysken 1988; Baker 1991, cited in
Lefebvre 1998), and object sharing (Baker 1989). With regard to the
supposed absence of prepositions in serializing languages, the Gungbe
sentences under (20) which involve an SVC or a prepositional phrase to
introduce an instrument clearly indicate that this view is not tenable. Indeed,
most serializing languages also have adpositions (see Chapter 3, as well as
Lefebvre and Brousseau 2002; Ameka 2003; Aboh 2005b, 2010b).
(20) a. Sέtù zé kpò xò kɔ̀ jó.
Setu take stick hit Kojo
‘Setu hit Kojo with the stick.’
b. Sέtù xò Kɔ̀ jó kpó kpò Kpó.
Setu hit Kojo PREP stick Prep
‘Setu hit Kojo with the stick.’
Similarly, the view that serialization occurs in languages which lack verbal
inflections and therefore verb raising (e.g., Muysken 1988; Baker 1991) cannot
be maintained. Aboh (2003a, 2004a, 2005c, 2009b) and Aboh and Dyakonova
(2009) present a very detailed discussion of verb movement in the Gbe
9
As I show in Aboh (2009b) and further discuss here, there appears to be no structural distinction
between these motion verb constructions and SVCs.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
282 The emergence of serial verb constructions
languages showing that the verb must raise at least to an aspect position to be
licensed in these languages. The Ewegbe example in (21) illustrates this point.
In this language, the habitual aspect marker is an affix to which the verb must
adjoin. In SVCs therefore, V1 and V2 must move to distinct aspect nodes where
they adjoin to the aspect marker.
(21) E tsɔ-na akɔɖu ɖu-na. [Ewegbe]
3SG take-HAB banana eat-HAB
‘S/he used to/habitually eat/s bananas.’
That serialization is not linked to absence of verb movement is further shown
by comparative data from Igbo (Kwa) and ╪Hoan (Khoisan), illustrated in
(22a) and (22b) respectively:
(22) a. Obi kwa-da-ra Eze. [Igbo, Kwa; Steward 1998: 183]
Obi push-fall-RV Eze
‘Obi pushed Eze down.’
b. Ma a- q║hu │’o djo ki kx’u na. [╪Hoan; Collins 2002: 1]
1SG PROG pour put.in water PART pot in
‘I am pouring water into the pot.’
Osamuyimen T. Stewart (1998) and Collins (2002) show convincingly that
these constructions share strong parallels with SVCs except for the ordering of
the verbs. While in some Kwa languages, the serializing verbs may be
separated by an intervening argument, some other Kwa (e.g., Igbo) and
Khoisan languages display verb compounding where the serialized verbs are
adjacent, with V1 systematically preceding V2. The generalization seems to be
the one in (23) where XP stands for the internal argument.
(23) a. V1 (XP) V2 [Kwa, Gungbe]
b. V1 – V2 (XP) [Kwa, Igbo; Khoisan, ╪Hoan]
Though the analyses they propose are significantly different, both Collins
(2002) and Aboh (2009b) resort to verb movement to account for the variation
in (23). Starting with (23a), for instance, these studies account for the ordering
in (23b) in terms of movement of the verb (i.e., V2) past the object.
If serialization is not a consequence of a lack of prepositions in the
serializing language or of the absence of inflectional morphology and hence
verb raising, the only option we are left with is the object sharing hypothesis
proposed by Baker (1989). According to this author, SVCs are defined by the
fact that the verbs in the series necessarily share a unique internal argument.
There have been several studies (e.g., Campbell 1989, 1992, 1996b; Da Cruz
1993, 1997, among others) indicating that object sharing, as proposed by Baker
(1989) and much related work, cannot be a defining condition on SVC.
Consider the following Gungbe sentences:
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.3 Serial verb constructions: a reanalysis 283
10
Indeed, Baker (1989: 547) analyzed examples such as in (ia–b) as covert coordination.
(i) a. Kofi sutu Amba kiri Kwaku. [Sranan]
Kofi shoot Amba kill Kwaku
‘Kofi shot Amba and killed Kwaku.’
b. Kofi naki Amba kiri en.
Kofi hit Amba kill her
‘Kofi struck Amba dead.’
Evidence for treating these examples as covert coordination comes from the fact that these
sentences display island effects typical of coordinate structures (e.g., impossibility to wh-extract
the object DPs), but see note 5.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
284 The emergence of serial verb constructions
Given the discussion in Section 7.2 where I showed that extraction for the
purpose of focus or wh-question is possible out of SVCs but impossible out of
coordinate structures, it is clear that the structures in (24) and (26) are bona fide
SVCs, though they do not involve object sharing. These facts represent a
serious argument against analyses that adopt Baker’s object sharing hypothesis.
That said, we are still left with the question of whether object sharing is ever
possible in serialization. Close scrutiny shows that it is never the case that series
of lexical verbs can share the same object (Aboh 2009b). This is because the Gbe
languages involve various instances of SVC where it cannot be claimed that all
the verbs in the sequence have lexical properties. I begin with take-series.
Sentence (27a) shows that zé ‘take,’ in Gungbe, can be used as main predicate
of the sentence, though it cannot select for abstract objects (27b).11
(27) a. Àlúkú zé gbàdó càkì àtɔ̀ n.
Aluku take corn sack three
‘Aluku took three sacks of corn.’
b. *Àlúkú zé àwá-jìjὲ/xó.
Aluku take joy/word
In SVCs, however, it is possible for this verb to be followed by an abstract
object or a DP element which the verb cannot semantically select in isolation.
(28) a. Àlúkú zé àwá-jìjὲ yì jónɔ̀ n lέ.
Aluku take joy receive guest PL
‘Aluku received the guests with joy.’
b. Àlúkú zé xó mítɔ̀ n ɖɔ̀ ná Súrù.
Aluku take word 1 PL talk PREP Suru
‘Aluku told Suru about us.’
It appears from these examples that the DP element that intervenes between V1
and V2 does not occur in that position in order to be jointly theta-marked by these
verbs. Instead, the verb zé ‘take’ in (28) does not theta-mark the following DP,
and therefore cannot be considered a full lexical verb in these series. This
conclusion is supported by the fact that, in certain Gbe languages (e.g.,
Fongbe) take-series involve an instrument DP which is introduced by a
preposition (Da Cruz 1997: 37).12
(29) a. Kɔ̀ kú sɔ́ [jìví ɔ́ ]i sέn wɔ̀ xúxú ɔ́ ná eci
Koku take knife DET cut bread DET with
‘Koku cut the bread with the knife.’
11
These facts are not new and were previously presented in Lefebvre (1991) and Da Cruz (1997)
for Fongbe. However, Gungbe and Fongbe differ in this respect. While Fongbe has two verbs
(zé, sɔ́ ) that can be translated as take, Gungbe has only one form: zé.
12
See Da Cruz (1997) for the discussion on ná/nú alternation in Fongbe.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.3 Serial verb constructions: a reanalysis 285
What is remarkable about this example is that even though the bracketed DP jìví
ɔ́ ‘the knife’ is introduced by the preposition ná (which also introduces a dative
argument as I mentioned in Chapter 6), this DP must occur immediately after V1.
Also remarkable is the fact that the form of the preposition here is ná, which, as I
have shown in Chapter 6, with examples (55e) and (55f), is allowed only when the
following DP-complement has been fronted. Aboh (2009b) concludes on the basis
of these facts that the so-called ‘shared object’ must have moved to the position
between V1 and V2 for other reasons than theta-marking and case-licensing. These
facts lead me to conclude that take-series involve a functional verb take.
A possible objection to this view would be to say that SVCs come in two
types in Gbe: (i) a combination of a functional verb and a lexical verb (27–28),
and (ii) a combination of two (or more) lexical verbs (30) (e.g., Baker 1991):
(30) Àlúkú ɖà àbɔ̀ bɔ̀ ɖù.
Aluku cook bean eat
‘Aluku made some beans to eat.’
This view does not seem to be tenable, however. In accounting for take-
series of the type discussed here, Aboh (2009b) demonstrates that V1 expressed
by the verb take merges in the functional domain of the lexical field which is
headed by V2, as schematized in (31):
(31) [CP . . . . . . [TP . . . V1TAKE . . . . . ..[VP . . . [V2 . . ...]]]]
Accordingly, V1 is a functional (or light) verb that has no (internal) θ -role
to assign, but merges within the functional domain (or the extended
projection) of V2. In this framework, the function of V1 take is to encode
the way the event expressed under V2 has been carried out (e.g., Awóyalé
1988; Lefebvre 1991).
Extending this analysis to examples like (30), which seem to involve two
genuine lexical verbs, Aboh (2009b) concluded that in an SVC, elements of the
type V1 are always functional verbs. Partial evidence which led to this
conclusion comes from the fact that in Gbe (unlike in other serializing
languages, e.g., Saramaccan) there is a general restriction on V1 such that
SVCs cannot be introduced by ditransitive verbs (e.g., ná ‘give’) in Gungbe.
(32) a. *Kòfí ná kwέi xɔ̀ xwé proi.
Kofi give money buy house
‘Kofi gave money buy a house [e.g., he gave money to buy a house].’
b. *Kòfí ná wémái Mì wà àzɔ́ n proi.
Kofi give book 1SG-ACC do work
‘Kofi gave me a book to work [e.g., he gave me a book to work with].’
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
286 The emergence of serial verb constructions
Yet, this restriction does not hold on Inherent Complement Verbs which are
fine with a ditransitive verb, as V1 (33):
(33) Kòfí ná nú ɖù vi lέ.
Kofi give thing eat child PL
‘Kofi gave food to the kids’
As one can see from the interpretation, the restriction in (32a–b) cannot be
semantic since similar combinations are grammatical in the English
translations, and indeed possible in Saramaccan (34a) and in Haitian (34b).
(34) a. Dí míi dá mi dí sópu wási hen. [Veenstra 1996: 79]
DET child give 1SG DET soap wash 3SG
‘The child gave me the soap to wash him with.’
b. Bouki ban m savon lave men m. [DeGraff, p.c. June 21, 2009]
Bouki give 1SG soap wash hand/s 1SG
‘Bouki gave me soap to wash my hands with.’
First, these types of mismatches between the creoles and the substrate
languages indicate that simple substrate influence cannot help account for the
existence of SVCs in the creoles. Second, assuming example (34a) is indeed an
SVC, we can conclude that the so-called single event reading often observed in
SVCs must be a tendency rather than a defining criterion. Indeed, the washing
need not materialize in this case since the event of giving and washing need not
be construed as the same event.
Third, the general ban on ditransitive verbs as V1 in Gungbe SVCs with
intransitive V2 (32a–b) suggests that the elements that realize this V1-position
are subject to a thematic restriction. We can therefore conclude that there is a
structural restriction on V1 in Gungbe. Now, let us recall the data in (14),
repeated here as (35) for convenience. The conclusion we reached there was
that there is a semantic restriction on SVCs such that V1 determines the type of
V2 it may combine with. Accordingly, the sequence steal-look is grammatical,
but not the sequence steal-see is not (14a–b).
(35) a. Sέtù fìn fòtó lɔ́ kpɔ́ n.
Setu steal photo DET look
‘Setu quickly looked at the photos (i.e. without being noticed by other
people).’
b. *Sέtù fìn fòtó lɔ́ mɔ̀ n.
Setu steal photo DET see
‘Setu quickly looked at the photos (i.e. without being noticed by other
people).’
Based on all these facts, we can conclude that elements of the type V1 are
subject to both semantic and structural restrictions. This property reminds us
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.3 Serial verb constructions: a reanalysis 287
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
288 The emergence of serial verb constructions
It appears from these examples that when the verb kpɔ́ n ‘to look at/for’ occurs
as V1 in an SVC, it does not convey the meaning of English to look or to see (37a).
Instead, the meaning we have here is closer to the English paraphrase watch your
words or speak properly. Taking into account the meaning of the Gungbe SVCs
rather than the morphological form of the verbs it appears that V1 expresses the
manner in which the event expressed by V2 should be performed. The right
generalization is therefore that, in V1-V2 combinations, V1 embeds the manner
component. A similar observation holds for the verb hὲn ‘to hold’, which can be
understood as to ‘cause X to spoil’ (37b), or to ‘cause X to be displaced’ (37c).
Contrary to what is often assumed in the literature, these examples show that
SVCs are not just a random combination of two lexical verbs expressing a
single event. This conclusion is further supported by the fact that even verbs
like ɖín ‘to look for’ or ‘to search for’ show semantic variation (that is tied to a
thematic restriction) when they are used as lexical or as V1 in an SVC. Observe,
for instance, that below this verb has a ‘causative’ meaning in (38a) different
from its lexical meaning in (38b):
(38) Àsíbá, ɖín nú ɖé ɖù tò àdòkɔ̀ n.
Asiba search thing INDEF eat at kitchen
a. ‘Asiba, get yourself something to eat in the kitchen.’
b. *Asiba, look for something to eat in the kitchen.’
Taking all these facts into account, I further submit that, since the meaning of
ɖín in this example is equivalent to English get (or have) in their causative
usage, there is no formal reason for us to assume that V1 and V2 ever share the
same argument in SVCs. This in turn would imply that elements of the type V1
are not lexical verbs. This observation may look like a bold statement, when
one considers example (39):
(39) Mígán ɖà àbɔ̀ bɔ̀ ɖù.
Migan cook bean eat
‘Migan cooked beans and ate.’
Traditional descriptions of similar examples conclude that the verbs in these
series combine two consecutive events (e.g., cooking and eating) into a single
event. Yet, as I demonstrate in Aboh (2009b), the analysis proposed here in
terms of V1 as functional verb extends to examples (39) as well. Observe, for
instance, that Gungbe also displays constructions like (40):
(40) Mìgàn ɖà kpònɔ̀ n lέ sɔ́ yì àhwàn.
Migan cook soldier PL take go war
‘Migan prepared the soldiers to go to war [i.e., by making some magic].’
While it is clear that an event involving cooking and eating can be assumed
in (39), such cannot be the case in (40) unless we assume cannibalism. But for
cannibals too, it makes no sense to literally cook one’s own soldiers before
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.3 Serial verb constructions: a reanalysis 289
going to war. Things being so, it must be the case that the intended meaning of
ɖà in (39) and (40) does not imply cooking in the culinary sense but rather ‘to
prepare’ whose extension could include ‘to cook.’ I therefore hypothesize that
the lexical verb ɖà in Gungbe, corresponds to the English verb to prepare (i.e.,
‘to make something ready for use,’ The New Oxford Dictionary of English).
Once we allow ourselves to make this assumption, series such as (39–40) can
be analyzed as hidden causatives that actually mean ‘to make/get/cause beans
(to be) ready to eat’, or ‘to make/get/cause the soldiers (to be) ready for war’
where only the meaning prepare is available.
Now, consider the verb ɖù, which in its lexical use in (39) means ‘to eat’ but
when used as V1 in (41) can be followed by various elements that are not
eatable in the literal sense, including for instance, the DP tò lɔ́ ‘the country’ (see
Section 4.3 for further examples). It is obvious from this example that the DP
complement cannot be said to receive the θ-role Theme from V1 (i.e., the Agent
did not literally eat the country).
(41) Yé ɖù [tò lɔ́ ] vɔ̀ .
3PL eat country DET finish
‘They ate the country finish [they ruined the country misappropriating funds].’
13
In Gungbe progressive constructions like in (37b), the sentence-final particle is a floating tone
(Aboh 2004a).
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
290 The emergence of serial verb constructions
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.4 Serialization: functional verbs and lexical verbs 291
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
292 The emergence of serial verb constructions
The proposed account for SVCs not only complements Cinque’s view on
clause union but also implies that in V1-XP-V2 combinations, only elements of
the type V1 fall in the class of functional verbs sometimes also referred to as
semi-lexical items (see also Da Cruz 1993, 1995, 1997; Hagemeijer 2001;
Cardinaletti and Giusti 2001; Cardinaletti and Shlonsky 2004).
If we extend the proposed analysis to creole languages, we can conclude that
example (47a) from Haitian is derivable as in (47b):
(47) a. Boukinèt te pran yon flè bay Malis.
Bouki TNS take DET flower give Malis
‘Bouki gave a flower to Malis.’
b. [TP Boukinet [T° te [AspP [Asp° pran [FP [AspP yon flè [Asp° bay [vP tBoukinèt [v-ext
tbay [vP tyon flè [v-appl tbay[VP2 tbay Malis]]]]]]]]]]]]
Similarly, the Saramaccan example (48a) can be partially represented as
in (48b):
(48) a. A téi dí fáka kóti dí beée. [Veenstra 1996: 4]
3SG take DET knife cut DET bread
‘He cut the bread with the knife.’
b. [TP A [T° [AspP [Asp° téi [FP [AspP di fáka [Asp° kóti [vP tA [v-ext tkóti [vP tdi fáka
[v-appl tkòti [VP2 tkòti di beée]]]]]]]]]]]]
Assuming this is the right characterization, and in light of Cinque (2004),
the question naturally arises whether we can still consider the simple presence
of SVCs in creoles as proof of substrate transfer. Let us, for instance, consider
the following examples from the superstrate languages, viz., French and
English (49):
(49) a. J’ai accompagné Jean voir ses enfants.
1SG.have accompanied John see POSS. Children
‘I accompanied John to visit his children.’
b. Let me go buy some rice.
These examples can be translated in the substrate languages, e.g., Gungbe,
as in (50):
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.4 Serialization: functional verbs and lexical verbs 293
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
294 The emergence of serial verb constructions
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.5 The proof is in the pudding 295
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
296 The emergence of serial verb constructions
Gbe and most Kwa languages freely allow restructuring of aspectual and
lexical verbs:
(55) a. Àhwàn éhè ná wá vɔ̀ gbèɖòkpó.
war DEM FUT come finish one.day
‘This war will eventually end one day (i.e., sooner or later this war will come
to an end).’
b. Míɔ́ n nyàn kànlìn lέ tɔ́ n sɔ́ n zùgbó mὲ.
fire chase animal PL exit from jungle in
‘The fire drove the animals out of the jungle.’
Given the view of creolization adopted in this book, it is not surprising that
the creoles discussed here, which emerged from the contact of largely Gbe
(Kwa) with Romance and Germanic languages, readily display both types of
restructuring found in the source languages (i.e., low and high
restructuring). Recall from the preceding discussions that Haitian mostly
displays verb series of movement and transfer as illustrated again in (56a) and
(56b), respectively:
(56) a. Bouki voye timoun yo ale lekol. [DeGraff 1992]
Bouki send child PL go school
‘Bouki sent the kids to school.’
b. Boukinèt te pran yon flè bay Malis.
Boukinet TNS take DET flower give Malice
‘Boukinèt gave Malis a flower.’
Compared to their Gungbe counterparts in (57), these examples could be
taken as strong evidence of substrate transfer of an SVC structure from the Gbe
languages into Haitian:
(57) a. Súrù zɔ́ n ví lέ yì wéxɔ̀ mὲ.
Suru order child PL go school
‘Suru sent the kids to school.’
b. Súrù zé vònvó ná Dóná.
Suru take flower give Dona
‘Suru gave Dona flower(s).’
However, some details about serialization (or low restructuring) in
Haitian cast doubt on such a one-to-one mapping. Rather, they indicate that the
Haitian structures display both Kwa and Romance properties, as can be expected
in a theory that evokes recombination of semantic and syntactic features as
argued for here. Indeed, while Gungbe (and other Gbe languages) can use the
verb take in take-series freely (58a), Haitian pran cannot always enter such
combinations. Crucially, some of my informants reject such constructions
altogether, while others reject take-series where the instrument is a full DP
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.5 The proof is in the pudding 297
(58b) (see also Lefebvre 1989).15 In the context of this discussion, it is important
to keep in mind that Gbe languages do not show such variation and allow take-
series freely. Once again, we face a situation where the creoles and their relevant
substrates are not isomorphic, though they share many similarities.
(58) a. Súrù zé kpò lɔ́ /ɖé xò Dóná.
Suru take stick DET hit Dona
‘Suru hit Dona using the/a stick.’
b. Boukinèt te pran baton (*na) frappe Malis.
Boukinèt TNS take stick DET hit Malis
‘Boukinet hit Malice using the stick.’
c. Boukinèt te pran (*yon) baton frappe Malis.
Boukinèt TNS take DET stick hit Malis
‘Boukinet hit Malice using a stick.’
In order to realize the equivalent of (58b), some Haitian speakers resort to
instrument complements introduced by ak (59a). Example (59b) shows that this
construction is parallel to French instrument constructions involving avec, the
source of the Haitian preposition.
(59) a. Boukinèt frape Malis ak yon baton. [Haitian]
b. Boukinèt a frappé Malis avec un bâton. [French]
Boukinet has hit Malis with a stick
‘Boukinet hit Malis with a stick.’
Interestingly enough, similar prepositional instrument structures are found in
Gbe along with the series in (60). In this case, Gungbe (as well as other Eastern
Gbe) display adpositions that bracket the noun phrase instrument (a property
absent from the creoles).
(60) Súrù xò Dóná kpó àfɔ̀ kpà kpó.
Suru hit Dona PREP shoe PREP
‘Suru hit Dona with a shoe.’
As already explained at the outset of this chapter, the development of the
series in a language cannot be taken as a consequence of the absence of
prepositional structures in that language (pace Bickerton 1981). In addition,
one should also note that the Haitian instrument example (59a) is akin to the
French example (59b) as opposed to the Gungbe example in (60). Together,
all these facts indicate that serialization is not a default (or unmarked) UG
strategy to compensate for the lack of prepositional phrases. Instead,
serialization, as understood here, appears to be a type of clause union that
15
I thank the Master students (M1 & M2) of the FLA at the Université de Port au Prince for
confirming these findings during my seminar, June 22 and June 29, 2014.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
298 The emergence of serial verb constructions
16
The ungrammatical example (56b) should not be confused with examples like the following
which I claim must be analyzed on a par with the English or colloquial French equivalent
restructuring structures in (ii) (Jaeggli and Hyams 1993; Aboh 2009b).
(i) Bouki voye Malis al kuit diri.
Bouki send Malis go cook rice
‘Bouki sent Malis to go cook rice.’
(ii) Bouki a envoyé Malis aller cuire du riz.
17
Example (57a) is still bad with the verb ‘prepare’ (DeGraff, p.c. June 21, 2009)
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.5 The proof is in the pudding 299
from our discussion of the Gungbe examples in (38) and (39) that the verb ɖà
does not mean ‘cook’; rather it corresponds to English ‘prepare,’ whose
extension can still include a cooking activity. This appears to explain why
Haitian grammar does not generate sentences like (62). Indeed, it appears that
Haitian, like French and English, has three such lexical verbs with different
meanings: kuit ‘to cook,’ préparé ‘to prepare,’ and paré ‘to prepare in the sense
of to get ready’ (Hall 1953). Given these distinctions in Haitian, we expect this
language to distinguish morphologically between causative ‘prepare’ and the
lexical verb ‘cook.’ This semantic opposition is not possible in Gungbe,
because the two meanings and usages are conflated in a single form ɖà.
Such lexical specifications might also be at the source of the discrepancies we
observed between Haitian and the Gbe languages throughout this chapter. In
addition, Haitian displays new lexico-semantic combinations under the influence
of French (viz., French lexical properties). This is so even though one sometimes
comes across very striking parallels of the type in (63) between the Gungbe
functional verb jὲ , which contributes to the semantic meaning ‘to fall’ (63a), and
its Haitian equivalent tonbe in (63b). Gungbe and Haitian differ because the
Gungbe verb is an inherent complement verb (i.e., it requires an object in its
citation form), while the Haitian verb is similar to French tomber in the sense that
it may or may not be used with a complement (actually a prepositional phrase). In
Haitian, the preposition is incorporated in the noun, unlike in French (63b–c).18
Therefore, French and Haitian contrast with Gungbe because the internal
argument cannot remain unrealized in this language (63a):
(63) a. Súrù jὲ *(àyí).
Suru V ground
‘Suru fell.’
b. Boukinet tonbe (atè).
Boukinet fall ground
‘Boukinet fell.’
c. Boukinet est tombé (*(par) terre).
Boukinet is fall at ground
‘Boukinet fell on the ground.’
However, in both Gungbe and Haitian, these verbal elements can be used
in inceptive constructions of the type illustrated in the pairs (64a–b) and
(64c–d).
(64) a. Súrù jὲ ògán bàí jí. [Gungbe]
Suru V chief make PCL
‘Suru started behaving as a chief.’
18
I thank Herby Glaude for bringing these data to my attention.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
300 The emergence of serial verb constructions
b. Jan tonbe joure Prezidan. [Haitian; Glaude p.c. November 10, 2008]
Jean fall insult president
‘John started insulting the president.’
c. Ùn jὲ àví ví jí. [Gungbe]
1SG V cry cry PCL
‘I started crying.’
d. M tonbe kriye, m di “Mama, li pran ti-plat mwen. [Haitian]
I fall cry 1SG say Mama 3SG take little-dish 1SG
‘I started crying: “Mama, he took my little dish.”’
[Adapted from Hall 1953: 75]19
If creoles have hybrid grammars in the sense that they recombine syntactic
and semantic features from both the substrate and superstrate languages, we
expect them to show a ‘family resemblance’ with the source languages, hence
the (dis)similarities observed here (Mufwene 2001, 2008). According to the
biological analogy adopted here, what matters is that the ecology of contact
makes possible new linguistic recombinations which, though absent in the
source languages, may appear in the creole language.
In the case at hand, this would mean that Haitian verb combinations derive
from a recombination of, on the one hand, French processes such as auxiliation,
causative construction, and high restructuring, and, on the other, Gbe low
restructuring (i.e., serialization) patterns. Put together, these combinatorial
possibilities allow Haitian to exhibit verb sequences that are absent from both
French and Gungbe. Such an unexpected recombination appears in the form of
the verb bay in its causative usage.20 I showed in the above discussions that
Haitian can use this verb in transfer series, in which it occurs as V2 (65a). This
usage is clearly parallel to the pattern displayed by the Gungbe equivalent in
(65b). Note, in addition, from the French example (65c), that similar
constructions are excluded in this language.
(65) a. Mari achete liv la bay Jan.
Mary buy book DET give John
‘Mary bought a book for John (i.e., she gave him the book).’
b. Dòsà xò wémà lɔ́ ná Dòsì.
Dosa buy book DET give Dosi
‘Dosa bought a book for Dosi (i.e., he gave her the book).’
19
I thank Michel DeGraff for helping me adapt this example.
20
These data were first brought to my attention by Herby Glaude (November 10, 2008) and then
confirmed by Michel DeGraff (June 21, 2009). I thank both of them for their suggestions on
serialization in Haitian.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.5 The proof is in the pudding 301
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
302 The emergence of serial verb constructions
A causative introduced by bay ‘give’ does not imply ‘single event reading’:
all sub-events need not be true at once (69a). A causative introduced by fè, (or
fòse), on the contrary forces single event reading. All sub-events must be true at
once and cannot be negated individually. Example (69b) is ungrammatical
because it is a contradiction. The first part of the clause implies that the baby
drank the milk, while the second part says that she didn’t. French faire
causatives also imply ‘single event reading’.
Putting these together, the fact that Haitian displays a bay causative that is
formally different from French-type fè/fòse causatives and inexistent in Gbe
would seem quite exceptional, if we were to assume substrate or superstrate
influence only. In the context of this discussion, however, what we have here
appears to be an instance of a new recombination of syntactic and semantic
features from the source languages. Indeed, bay causatives and bay series allow
the same lexical element to first merge in two functional positions which
correspond to the low and high restructuring found in Gungbe and
French, respectively. I therefore conjecture that the fact that bay can be used
in restructuring (similarly to the verb ‘give’ in Gbe series) paved the way
for its usage in high restructuring, similarly to French faire and forcer
causatives. As the reader may see, this new development in Haitian led to the
emergence of local complexity with regard to restructuring.
Quite interestingly, the discussion in the previous paragraphs reminds us of
Veenstra’s (1996) analysis of the Saramaccan verb mbéi, which has the
distributive properties illustrated in (70):
(70) a. A bì tá mbéi témbe. [Veenstra 1996: 158ff.]
3SG TNS ASP make wood
‘He was making wood-carvings.’
b. Dí kái u dí tjúba mbéi hen uwíi munjá tooná kó bè.
DET fall of DET rain make 3SG hair wet turn come red
‘The rainfall made her hair wet and turn red.’
c. Dí tjúba tá kái mbéi hen uwíi munjá tooná kó bè.
DET rain ASP fall make 3SG hair wet turn come red
‘It is raining so that her hair becomes wet and turns red.’
In analyzing the various usages of this verb, Veenstra (1996: 163) concluded
that “Saramaccan has two (distinct) lexical entries for the formative mbéi: one
for a (causative) verb meaning ‘make’, which can also participate in light verb
constructions and one for a complementizer meaning ‘so that’ expressing
causality.” According to Veenstra, the causative verb mbéi corresponds to
cases like (70b), in which the selected complement is a small clause or a
reduced IP. On the other hand, the causative complementizer mbéi
corresponds to example (70c), in which the embedded clause is a full CP.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
7.6 Conclusion 303
The two instances of mbéi differ in that the former can combine with an aspect
marker (71a), while the latter cannot (66b) (Veenstra 1996: 162):
(71) a. Dí kái u dí tjúba tá mbéi hen uwíi munjá tooná kó bè.
DET fall of DET rain ASP make 3SG hair wet turn come red
‘The rainfall is making her hair wet and turn.’
b. *Dí tjúba tá kái tá mbéi hen uwíi munjá tooná kó bè.
DET rain ASP fall ASP make 3SG hair wet turn come red
What we see here is therefore another instance in which a verb has two
usages: one lexical and the other functional. Consistent with the analysis
of SVCs as restructuring structures proposed here, the behavior of
Saramaccan mbéi is to some extent similar to that of Haitian bay,
which also appeared to target two different sites in the functional
sequence.
7.6 Conclusion
This chapter has shown that the study of SVCs in certain creoles and their
source languages can help in understanding language birth and the
recombination of syntactic and semantic features into a new idiolect.
Though it could be argued in some specific cases that SVCs reflect
substrate transfer, it is not enough to limit oneself to surface
correspondences between the creoles and their source languages. The
reason is that there appears to be no serializing parameter that could be
isolated as responsible for the emergence of serialization in creoles.
Similarly, it cannot be claimed that the phenomenon is linked to the lack of
V-to-T movement in these languages. Indeed, the Gbe languages display
short V-to-Asp movement as well as serialization. Given that SVCs are
regarded as instances of restructuring constructions, their distributions
across languages appears more general and such constructions are found in
both the lexifiers and the substrate languages, although the congruence is not
total. In such a context, great analytic precision is needed to isolate the
linguistic features (syntactic and semantic) that are recombined into the
creole. In light of the above discussions, it appears that the debate
regarding whether or not SVCs are found in Indian Ocean Creoles (e.g.,
Bickerton 1989, 1990b versus Seuren 1990a, 1990b, 1991) is misguided. The
question should be reformulated as whether Indian Ocean Creoles have
clause union phenomena of the types analyzed here as restructuring,
and, if they do, whether they display low or high restructuring.
Downloaded from https:/www.cambridge.org/core. University of Exeter, on 11 Mar 2017 at 01:52:58, subject to the Cambridge Core
terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.008
8 Conclusions: some final remarks on hybrid
grammars, the creole prototype, and language
acquisition and change
The discussion in this book shows that during the seventeenth and eighteenth
centuries, the Kingdom of Allada on the Slave Coast played an important role in
the Triangular Trade. During this period, socio-political and economic factors
in the Bight of Benin contributed to making the locals, namely people of Aja
descent, victims of a globalized world economy in which enslaved Africans
were treated as livestock and part of the goods exchanged. Our conclusions
drawn on the basis of both geopolitics and socio-economic factors corroborate
work on population factors in the Caribbean and in Suriname according to
which speakers of Kwa languages represented a significant part of slave cohorts
exported to the Caribbean plantation colonies. While these Kwa people,
including significant numbers of Gbe speakers, probably did not always
constitute the most numerous ethnic group in the Caribbean and in Suriname
during the slavery period, they appear to have been instrumental in shaping the
culture of the emerging colonies. Two related factors underlie this view:
(i) more and more Kwa/Gbe people were enslaved during the transition of the
colonies from the homestead to the plantation phase, when there was demand
for larger numbers of slaves; and (ii) the Gbe/Kwa people originated from a
homogeneous cultural background, a factor that probably allowed them to form
tighter sub-communities (compared to other West African or Central African
slaves). The structure of such communities could have helped them exert more
convergent and thus stronger pressures on the emerging linguistic patterns
during the early stages of the nascent language. Accordingly, the ecology of
the plantation communities in Haiti and Suriname favored speakers of Gbe/
Kwa languages and enabled them to contribute a lasting founder effect.
While this perspective is compatible with the competition and selection
model argued for in this book, and in previous studies such as Mufwene
(2001, 2008) and Ansaldo (2009), it also suggests that monolithic theories
that account for creole genesis exclusively by invoking inheritance from the
superstrate, substrate influence, language universals, or fossilization of some
early interlanguage stage are untenable. Instead, the discussion in this book
shows that creole languages emerged from the recombination of linguistic
304
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
8.2 Evaluating complexity and simplicity 305
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
306 Some final remarks on hybrid grammars
studies which postulate creole simplicity and prototype thus making creole
languages an exceptional typological class. These critiques show that ideas
about creole simplicity and prototype usually ignore the socio-historical facts
discussed in Chapter 2 of this book and further rely on mere morphological
counts measuring morphological size and shape (i.e., what DeGraff refers to as
“bit complexity”) as well as trivial surface syntax. For instance, it is not
uncommon to see complexity metrics defined in terms of the size of the class
of marked members of a phonemic system, the number of tonemes, or the
number of distinct grammatical or syntactic rules in a language.
It is obvious that such informal metrics fail to take into account a
fundamental aspect of language: the relation between form and meaning. As
already mentioned in Aboh and Smith (2009) and Aboh (2009a), linguistic
complexity cannot be evaluated without factoring in the relation between form
and meaning, and how such relations compare cross-linguistically. For
instance, it could be argued that a complex form is one that results in
excessive processing cost in evaluating its meaning. Accordingly, in order to
show that an aspect of a language is more complex than that of another, we
would have to show that a given form in one language triggers excessive
processing cost compared to its equivalent in the other language (cf. Hawkins
2004). Thus, by implication, a form can be assumed to be simple if it yields
little (or almost no) processing cost. An evaluation method along these lines
predicts that a simple aspect of a language will typically result in fast
computation of the meaning, while a complex aspect on the contrary will
result in slow computation.
Such an evaluation of complexity and simplicity can be carried out thanks to
new experimental methods in psycholinguistics as well as new computational
models of language. Scholarly exchanges between experts of language
acquisition and evolution, psycholinguists, and computational cognitive
scientists will undoubtedly help us better understand human cognition, and
shed light on how issues of linguistic structural complexity are related to
typological variation. Note, however, that the conclusion I just made about
slow versus fast computation as potentially related to structural complexity
relies on the crucial observation, often ignored in theories of creole simplicity,
that a meaningful notion of complexity can only be one that is relative to a
given module of a grammar, rather than to a grammar as a whole. It seems to
me unlikely that a language will be either overall complex and therefore
allows costly processing and slow computation in all aspects of its grammar,
or alternatively simple and permits easy processing and fast computation in
all components of its grammar. Both excessive complexity and excessive
simplicity would prevent a language from meeting the requirement of
learnability and easy processability, and that of satisfying the often
complex communicative needs of its users, respectively.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
8.2 Evaluating complexity and simplicity 307
Aside from the fact that no serious study as that proposed here, viz., one in
which theory informs experimental work to investigate the relation between
structure, meaning, and complexity, has been conducted on creole languages,
this conclusion further indicates that claims about simplicity as an across-
the-board property of creole languages cannot be correct. Indeed for such
claims to be taken seriously, they would have to show on empirical and
experimental grounds that parsing in creole languages requires almost no
processing and allows fast computation in all modules of these grammars. I
leave it to proponents of creole simplicity and prototype to show that this is
indeed the case, but the task is certainly not a trivial one, as the following
paragraphs show.
The discussion here presents several cases of local complexity that
indicate that claims about creole simplicity and prototype do not typically
provide us with the full picture and therefore fail to answer a fundamental
question in linguistic study: How does structural complexity or simplicity
relate to language acquisition and linguistic variation? By way of
illustration, let us consider a commonly used metric of syntactic
complexity that basically counts the number of distinct syntactic rules in a
module of two languages. In this regard, a common point of comparison
between creoles and their Indo-European source languages (at the exclusion
of their substrate languages) is that of questions. For instance, English
displays yes–no and wh-phrase questions involving subject-auxiliary
inversion. Saramaccan and English-based creoles, in general, lack such a
rule. In comparing the English wh-question (1) to the Saramaccan example
in (2), one could conclude that Saramaccan, which combines properties of
English syntax and Gbe syntax, as shown in Chapter 6, appears to display a
simpler syntax.
(1) What did you cook?
(2) Andí wε i bói?
what FOC 2SG cook
‘What did you cook?’
However, as is often the case, such an uninformed comparison would miss
the fact that English resorts to auxiliary-subject inversion in addition to wh-
fronting, while Saramaccan involves a focus marker (absent in English) that
attaches to the fronted wh-phrase. Let us see how these two examples could be
represented structurally. Following the discussion in Chapter 6 (see also Rizzi
1997; Aboh 2004a), I’m assuming that fronted wh-phrases move to a focus
position cross-linguistically. Accordingly, both English and Saramaccan
display wh-movement to [spec FocP]. The arrows illustrate movement
operations in these two languages.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
308 Some final remarks on hybrid grammars
Counting these arrows, as one may, we realize that where English displays
two instances of movement (i.e., wh-movement and auxiliary movement to
Foc), Saramaccan exhibits one arrow only. Instead of raising a tense specifying
pleonastic element (i.e., do) to Foc, Saramaccan developed a focus marker (i.e.,
wε) that fills in this position. We could stop here and conclude that English
(with its two movement operations) has a more complex syntax than
Saramaccan which displays only one movement operation. Actually, most
exceptionalist views on creole languages do just that (e.g., McWhorter 1998,
2001, 2011).
However, for such a conclusion to be firmly established, we must also show
that auxiliary movement of do to Foc (an instance of internal merge in
minimalist syntax) is more complex than base generation of wε in Foc (a case
of external merge in Minimalism). No current theory of creole simplicity takes
this further step. Indeed, even though there are some ideas in the literature
suggesting that internal merge (i.e., movement) is costlier and therefore more
complex than external merge (i.e., base generation) (e.g., Chomsky 1995) no
experimental study has proven this to be generally correct in all the relevant
cases.
Even if we were to grant the view that the syntactic operation underlying
constituent or head movement is generally costlier than base-generation of a
lexical item, the fact that Saramaccan does display wh-movement in content
questions would lead us to a paradox that cannot be handled by theories of
creole simplicity grounded in imperfect second language acquisition. Indeed,
this language would appear to have developed a complex option for part of the
syntax of forming content questions, while the other part would involve a simpler
syntax. How come the creators of the creoles acquired wh-fronting rules even
though they failed to acquire subject–auxiliary inversion? For the picture to be
complete, one would also need to take into account the fact that English subject–
auxiliary inversion derives from the presence of auxiliaries in this language as
opposed to Saramaccan, where verbal conjugations and their related auxiliary
paradigm were replaced by an intricate system of TMA markers. Likewise, the
complementizer system responsible for subject–auxiliary inversion in English
displays a different morphosyntax in Saramaccan, where it involves a series of
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
8.2 Evaluating complexity and simplicity 309
discourse markers with their own syntax and semantics. Recall, for instance, from
the discussion in Chapter 6 that one property of such discourse markers is that, in
addition to attracting focused or topicalized constituents, they can also take scope
over the proposition which is therefore fronted. Consequently, some discourse
markers that realize the C-domain in Saramaccan can also occur in sentence-final
position (as shown in Chapter 6).
It is not clear to me how these facts would be accommodated in a holistic
theory of creole simplicity that is only concerned with counting morpheme
types and shapes or movement types without ever considering the relations
between these forms or operations and their semantics. Indeed, all English-
based creoles that I know of developed wh-fronting rules in content questions
even though such rules are not a priori required by human cognition and cannot
be assumed to be default options. Many languages of the world (e.g., Mandarin
Chinese, Japanese, Hindi, and Iraqi Arabic, to name only a few) display wh-in-
situ content questions, where the wh-phrase remains in its base position.
Likewise, the Romance and Germanic languages also allow some content
questions in which it is not immediately clear that the construction involves
movement. A case in point is subject question involving lexical verbs (e.g.,
Who came to the party? Who did what? Qui est venu à la fête ? Qui a fait
quoi ?). Clearly, the input that the creators of creole languages were exposed to
could also have favored the choice of in-situ wh-questions, but this did not
happen. If indeed, movement is costly and complex (and presumably difficult
to acquire), why are English-based creoles not similar to Mandarin Chinese?
What drives the kind of local complexity just described?
One can raise similar questions with regard to relative clauses in creoles.
Most Atlantic creoles involve relative clauses similar to those of their source
languages (Romance, Germanic, and Niger-Congo). In addition, Saramaccan
displays number agreement in relative clauses: the Saramaccan relative marker
agrees in number with the head noun, as discussed in Chapters 3 and 5 (see also
Huttar, Aboh, and Ameka 2013). Given the combination of wh-movement and
number agreement in Saramaccan relative clauses, this language appears to be
more complex in this domain than English is. One is therefore tempted to ask
why Saramaccan and more generally other Atlantic creoles did not resort to an
apparently much simpler or less complex option such as coordination. What
aspects of the input that the slaves were exposed to led them to develop a
structure like example (37b) in Chapter 3 (viz., dee fisi dee mi tata kisi bigi ‘the
fish that my father caught are big’), instead of relying on coordination or
juxtaposition (e.g., ‘my father caught fish, the fish are big’)?
Note that, according to the pidgin-to-creole cycle, one expects creoles to
exhibit this sort of paratactic structures rather than embedding constructions
such as relative clauses. It is not clear to me how these theories would explain
the link between the pidgin phase and the creole phase and how that relates to
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
310 Some final remarks on hybrid grammars
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
8.3 Hybridism drives opacity (and complexity) 311
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
312 Some final remarks on hybrid grammars
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
8.4 Beyond creoles 313
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
314 Some final remarks on hybrid grammars
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
8.4 Beyond creoles 315
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
316 Some final remarks on hybrid grammars
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:31:19, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.009
References
Abney, Paul S. 1987. The English noun phrase in its sentential aspect. PhD dissertation,
MIT, Cambridge, MA.
Aboh, Enoch O. 2001. Clitiques phonétiques ou clitiques tout court. In Claude Muller,
Paulo de Carvalho, Laurence Labrune, Frédéric Lambert, and Katja Ploog (eds.),
Clitiques et cliticisation. Paris: Honoré Champion, 461–481.
2002. La morphosyntaxe de la péripherie gauche nominale. Recherches linguistiques
de Vincennes 31: 9–26.
2003a. Les constructions à objet préposé et les series verbales dans les langues kwa.
In Patrick Sauzet and Anne Zribi-Hertz (eds.), Typologie des langues d’Afrique
et universaux de la grammaire. Vol. II: ‘Benue-Kwa, Soninke, Wolof’. Paris:
L’Harmattan, 15–40.
2003b. Focus constructions across Kwa. In Cege Githiora, Heather Littlefield, and
Victor Manfredi (eds.), Kinyira njira! Step firmly on the pathway! Trenton, NJ:
Africa World Press, 7–22.
2004a. The morphosyntax of complement-head sequences: Clause structure and word
order patterns in Kwa. New York: Oxford University Press.
2004b. Left or right? A view from the Kwa peripheral positions. In David Adger,
Cécile De Cat, and George Tsoulas (eds.), Peripheries: Syntactic edges and their
effects. Dordrecht: Kluwer Academic Publishers, 165–189.
2004c. Topic and focus within D. Linguistics in The Netherlands 21: 1–12.
2005a. Pattern and feature competition: Toward a syntactic account to contact-
induced transfer. Conference on Creole language structure between substrates
and superstrates. MPI Leipzig (Germany) June 3–5, 2005.
2005b. The category P: The Kwa paradox. Linguistic Analysis 32: 615–646.
2005c. Object shift, verb movement and verb reduplication. In Guglielmo Cinque and
Richard Kayne (eds.), The Oxford handbook of comparative syntax. New York:
Oxford University Press, 138–177.
2005d. Deriving relative and factive constructions in Kwa. In Laura Brugè,
Giuliana Giusti, Nicola Munaro, Walter Schweikert, and Giuseppina Turano (eds.),
Contributions to the thirtieth incontro di grammatica generativa. Venice:
Cafoscarina, 265–285.
2006a. The role of the syntax-semantics interface in language transfer. In
Claire Lefebvre, Lydia White, and Christine Jourdan (eds.), L2 acquisition and
creole genesis: Dialogues. Amsterdam: Benjamins, 221–252.
317
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
318 References
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 319
Aboh Enoch O and Roland Pfau. 2011. What’s a wh-word got to do with it? In
Paula Benincà and Nicola Munaro (eds.), Mapping the left periphery: The
cartography of syntactic structures. New York: Oxford University Press, 91–124.
Aboh, Enoch O. and Norval Smith (eds.) 2009. Complex processes in new languages.
Amsterdam: Benjamins.
2009. Simplicity, simplification, complexity and complexification: Where have the
interfaces gone? In Enoch O. Aboh and Norval Smith (eds.), Complex processes in
new languages. Amsterdam: Benjamins, 1–25.
2012. The morphosyntax of non-iconic reduplications: A case study in Eastern Gbe
and the Surinam creoles. In Enoch O. Aboh, Norval Smith, and Anne Zribi-Hertz
(eds.), The grammar of reiteration in creole and non-creole languages.
Amsterdam: Benjamins, 27–75.
2015a. Migrations, ethno-dynamics and geolinguistics in the Eastern Aja-Tado
cultural area. In Pieter Muysken and Norval Smith (eds.), Surviving the Middle
Passage: The West Africa-Surinam Sprachbund. Berlin: Mouton, 43–66.
2015b. Non-iconic reduplications in Eastern Gbe and Surinam. In Pieter Muysken
and Norval Smith (eds.), Surviving the Middle Passage: The West Africa-Surinam
Sprachbund. Berlin: Mouton, 241–260.
Aboh, Enoch O., Norval Smith, and Tonjes Veenstra. 2013. Saramaccan. In Susanne
Maria Michaelis, Philippe Maurer, Martin Haspelmath, and Magnus Huber (eds.),
The survey of pidgin and creole languages. Vol. I: English-based and Dutch-based
languages. Oxford University Press, 27–38.
Aboh, Enoch O., Roland Pfau, and Ulrike Zeshan. 2005. When a Wh-word is not a Wh-
word: The case of Indian Sign Language. In Tanmoy Bhattacharya (ed.), The
yearbook of South Asian languages and linguistics. Berlin: Mouton de Gruyter,
11–43.
Adande, Alexis B. A. 1984. Tɔgudo-Awute, capitale de l’ancien royaume d’Allada.
Unpublished PhD dissertation, Université de Paris I.
Aikhenvald, Alexandra, and R. M. W. Dixon. (eds.) 2006. Serial verb constructions:
A Cross-linguistic typology. New York: Oxford University Press.
Alleyne, Mervyn C. 1971. Acculturation and the cultural matrix of creolization.
In Dell Hymes (ed.), Pidginization and creolization of languages. Cambridge
University Press, 169–186.
1979. On the genesis of languages. In Kenneth C. Hill (ed.) The genesis of language.
Ann Arbor, MI: Karoma, 89–107.
1980. Comparative Afro-American: An historical-comparative study of English-
Based Afro-American dialects. Ann Arbor, MI: Karoma.
Ameka, Felix. 1992. Focus constructions in Ewe and Akan: A comparative perspective.
MIT Working Papers in Linguistics 17: 1–25.
2003. Prepositions and postpositions in Ewe (Gbe): Empirical and theoretical
considerations. In Parick Sauzet and Anne Zribi-Hertz (eds.), Typologie des
langues d’Afrique et universaux de la grammaire. Paris: L’Harmattan, 41–67.
2010. Information packaging constructions in Kwa: Micro-variation and typology. In
Enoch O. Aboh and James Essegbey (eds.), Topics in Kwa syntax. Studies in
Natural Language and Linguistic Theory 78. Dordrecht: Springer, 141–176.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
320 References
Anglade, Pierre. 1998. Inventaire étymologique des termes creoles des caraïbes
d’origine africaine. Paris: L’Harmattan.
Anonymous. 1983. Éléments de recherche sur la langue fon. Cotonou: République du
Bénin.
Ansaldo, Umberto. 2005. Typological convergence and admixture in Sri Lanka
Malay: The case of Kirinda Java. Unpublished manuscript, University of
Amsterdam.
Ansaldo, Umberto. 2009. Contact languages: Ecology and evolution in Asia.
Cambridge University Press.
Arends, Jacques. 1989. Syntactic developments in Sranan. Unpublished PhD dissertation,
University of Nijmegen.
1995a. The socio-historical background of creoles. In Jacques Arends, Pieter Muysken,
and Norval Smith (eds.), Pidgins and creoles: An introduction. Amsterdam:
Benjamins, 15–24.
1995b. Demographic factors in the formation of Sranan. In Jacques Arends (ed.), The
early stages of creolization. Amsterdam: Benjamins, 233–285.
1998. The development of complementation in Saramaccan. In Bernard Caron (ed.),
Proceedings of the eleventh International Congress of Linguistics. Oxford:
Pergamon (CD-ROM).
2009. Syntactic developments in Sranan: Creolization as a gradual process.
www.dbnl.org/tekst/aren012synt01_01/colofon.htm
Aronoff Mark, Irit Meier, and Wendy Sandler. 2005. The paradox of sign language
morphology. Language 81: 301–344.
Avolonto, Aimé. 1995. Pour une approche minimaliste des verbes à objets inhérents en
fongbe. PhD Dissertation, University of Quebec at Montreal.
Awóyalé, Yiwọla. 1988. Complex predicates and verb serialization. MIT Lexicon
Working Papers 28. MIT Center for Cognitive Science. MI, 1–35.
Bailey, Beryl L. 1966. Jamaican Creole syntax: A transformational approach.
Cambridge University Press.
Baker, Mark C. 1985. The mirror principle and morphosyntactic explanation. Linguistic
Inquiry 16: 373–415.
1988. Incorporation: A theory of grammatical function changing. The University of
Chicago Press.
1989. Object sharing and projection in serial verb constructions. Linguistic Inquiry
20: 513–553.
1991. On the relation of serialization to verb extensions. In Claire Lefebvre (ed.),
Serial verbs: Grammatical, comparative and cognitive approaches. Amsterdam:
Benjamins, 79–102.
2008. The macroparameter in a microparametric world. In Theresa Biberauer (ed.),
The limits of variation. Amsterdam: Benjamins, 351–374.
Bakker, Peter. 2014. Creolistics: Back to square one? Journal of Pidgin and Creole
Languages 29: 177–194.
Bakker, Peter, Daval-Markussena Aymeric, Parkvall Mikael, and Plag Ingo. 2011.
Creoles are typologically distinct from non-creoles. Journal of Pidgin and
Creole Languages 26: 5–42.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 321
Bakker, Peter, Marike Post, and Hein van der Voort. 1994. TMA particles and
auxiliaries. In Jacquues Arends, Pieter Muysken, and Norval Smith (eds.),
Pidgins and Creoles: An introduction. Amsterdam: Benjamins, 247–258.
Baptista, Marlyse. 2009. Economy, innovation and degrees of complexity in creole
formation. In Enoch O Aboh and Norval Smith (eds.), Complex processes in new
languages. Amsterdam: Benjamins, 293–315.
Baptista, Marlyse and Jacqueline Guéron (eds.) 2007. Noun phrases in creole
Languages: A multi-faceted approach. Amsterdam: Benjamins.
Bernstein, Judy. 1997. Demonstratives and reinforcers in Romance and Germanic
languages. Lingua 102: 87–113.
2001a. Focusing the “right” way in Romance determiner phrases. Probus 13: 1–29.
2001b. The DP hypothesis: Identifying clausal properties in the nominal domain. In
Mark Baltin and Chris Collins (eds.), The handbook of contemporary syntactic
theory. Oxford: Blackwell, 536–561.
Bickerton, Derek. 1981. Roots of language. Ann Arbor, MI: Karoma.
1984. The language bioprogram hypothesis. Behavioral and Brain Sciences 7: 173–221.
1988a. Creole languages and the bioprogram. In Frederic Newmeyer (ed.),
Linguistics: The Cambridge survey. Vol. II: Linguistics theory: Extensions and
implications. Cambridge University Press, 268–284.
1988b. A dialogue concerning the linguistic status of creole languages. In
Newmeyer, Frederic (ed.), Linguistics: The Cambridge survey. Vol. II: Linguistics
theory: Extensions and implications. Cambridge University Press, 302–306.
1989. Seselwa serialization and its significance. Journal of Pidgin and Creole
Languages 4: 155–183.
1990a. Language and Species. The University of Chicago Press.
1990b. If it quacks like a duck . . . A reply to Seuren. Journal of Piding and Creole
Languages 5: 293–303.
1999. How to acquire language without positive evidence: What acquisitionists can
learn from creoles. In M. DeGraff (ed.), Language creation and language change,
Cambridge, MA: MIT Press, 49–74.
2008. Bastard tongues: A trail-blazing linguist finds clues to our common humanity in
the world’s lowliest languages. New York: Hill and Wang.
Bobyleva, Ekaterina. 2013. The development of the nominal domain in creole
languages: A comparative-typological approach. Utrecht: LOT.
Boeckx, Cedric. 2010. What principles and parameters got wrong. Unpublished
manuscript, Catalan Institute for Advanced Studies.
Booij, Geert. 2005. The grammar of words. New York: Oxford University Press.
Borer, Hagit. 1984. Parametric syntax: Case studies in Semitic and Romance
languages. Dordrecht: Foris.
2005. Structuring sense. Vol. I: In name only. New York: Oxford University Press.
Brousseau, Anne-Marie and John S. Lumsden. 1992. Nominal structure in Fongbe. The
Journal of West African Languages 22: 5–25.
Bruyn, Adrienne. 1995a. Grammaticalization in creoles: The development of determiners
and relative clauses in Sranan. Studies in Language and Language Use 21.
Amsterdam: IFOTT.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
322 References
1995b. Noun phrases. In Jacques Arends, Pieter Muysken, and Norval Smith
(eds.), Pidgins and Creoles: An introduction. Amsterdam: Benjamins,
259–270.
Burzio, Luigi. 1986. Italian syntax: A government and binding approach. Dordrecht:
Reidel.
Bybee, Joan L. 1985. Morphology: A study of the relation between meaning and form.
Amsterdam: Benjamins.
Byrne, Francis. 1987. Grammatical relations in a radical creole. Amsterdam:
Benjamins.
Cable, Seth. 2004a. Predicate clefts and base-generation: Evidence from Yiddish and
Brazilian Portuguese. Unpublished manuscript, MIT, Cambridge, MA.
2004b. Restructuring in English. Unpublished manuscript, MIT, Cambridge, MA.
Campbell, Lyle. 1996. Typological and areal issues in reconstruction. In Jacek Fisiak
(ed.), Linguistic reconstruction. Berlin: Mouton, 49–72.
Campbell, Richard. 1989. The grammatical structure of verbal predicates. PhD
dissertation, University of California, Los Angeles.
Campbell, Richard. 1992. Serial verbs and unaccusativity. MIT Working Papers 17:
41–51.
1996a. Specificity operators in SpecDP. Studia Linguistica 2: 161–188.
1996b. Serial verbs and shared arguments. The Linguistic Review 13: 83–118.
Capo, Hounkpati B. C. 1991. A comparative phonology of Gbe. Berlin: Foris.
Cardinaletti, Anna and Giuliana Giusti. 2001. Semi-lexical motion verbs in Romance
and Germanic. In Norbert Corver and Henk van Riemsdijk (eds.), Semi-lexical
categories: On the function of content words and content of function words. Berlin:
Mouton, 371–414.
Cardinaletti, Anna and Ur Shlonsky. 2004. Clitic positions and restructuring in Italian.
Linguistic Inquiry 35: 519–557.
Cardinaletti, Anna and Michal Starke. 1999. The typology of structural deficiency:
A case study of the three classes of pronouns. In Henk Van Riemsdijk (ed.), Clitics
in the languages of Europe. Berlin: Mouton de Gruyter, 145–233.
Chaudenson, Robert. 1992. Des îles, des hommes, des langues: Éssai sur la créolisation
linguistique et culturelle. Paris: L’Harmattan.
2001. Creolization of language and culture. London: Routledge.
2003. La créolisation: Théorie, applications, implications. Paris: L’Harmattan.
Cheng, Lisa L. and Rint Sybesma. 1999. Bare and not-so-bare nouns and the structure of
NP. Linguistic Inquiry 30: 509–542.
Chomsky, Noam. 1977. On wh-movement. In Peter W. Culicover (ed.), Formal syntax.
New York/San Francisco: Academic Press, 71–132.
1986. Knowledge of language: Its nature, origin and use, Westport, CT: Praeger.
1995. The minimalist program. Cambridge, MA: MIT Press.
2001. Derivation by Phase. In Micheal Kenstowixcz (ed.), Ken Hale: A life in
language. Cambridge, MA: MIT Press, 1–52.
2008. On Phases. In Robert Freidin, Carlos P. Otero, and Maria Luisa Zubizarreta
(eds.), Foundational issues in linguistic theory: Essays in honor of Jean-Roger
Vergnaud. Cambridge, MA: MIT Press, 133–166.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 323
Cinque, Guglielmo. 1994. On the evidence for partial N-movement in the Romance
DP. In Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi, and
Raffaella Zanuttini (eds.), Paths towards Universal Grammar. Washington, DC:
Georgetown University Press, 85–110.
1996. The “antisymmetric” program: Theoretical and typological implications.
Journal of Linguistics 32: 447–464.
1999. Adverbs and functional heads: A cross-linguistic perspective. New York:
Oxford University Press.
2004. Restructuring and functional heads: The cartography of syntactic structures.
New York: Oxford University Press.
2010. The syntax of adjectives: A comparative study. Cambridge, MA: MIT Press.
Clements, Clancy J. 1996. The genesis of a language: The formation and development of
Korlai Portuguese. Amsterdam: Benjamins.
Collins, Chris. 1994. The factive construction in Kwa. Unpublished manuscript, Cornell
University.
1997. Argument sharing in serial verb constructions. Linguistic Inquiry 28: 461–497.
2002. Multiple verb movement in Hoan. Linguistic Inquiry 33:1–29.
Comrie, Bernard. 1989. Language universals and linguistic typology. Chicago
University Press.
Croft, William. 2000. Explaining language change: An evolutionary approach. London:
Longman.
Curtin, Philip D. 1969. The Atlantic slave trade: A census. Madison, WI: University of
Wisconsin Press.
Da Cruz, Maxime. 1992. ECP et l’alternance nu/na en fongbe. Unpublished manuscript,
University of Quebec at Montreal.
1993. Les constructions sérielles du fongbe: Approches sémantique et syntaxique.
PhD dissertation, University of Quebec at Montreal.
1995. Aspectual verbs fó, v ‘finish’ in Fongbe. The Linguistic Review 12: 361–380.
1997. Serial verb constructions and null arguments in Fon. In Rose-Marie Déchaine
and Victor Manfredi (eds.), Object positions in Benue-Kwa. The Hague: The
Holland Academic Graphics, 31–45.
Damonte, Federico. 2002. The complementizer layer in Saramaccan. In Manuel Leonetti,
Olga Fernàndez Soriano, and Victoria Escandell Vidal (eds.), Current issues in
generative grammar. Tenth Colloquium on Generative Grammar: Selected papers.
Madrid: Universidad Alcalà de Henares, Servicio de Publicaciones, 31–50.
De Castro, Yeda Pessoa. 2002. A língua Mina-Jeje no Brasil. Um falar Africano em
Ouro Preto do Século XVIII. Apoio cultural. Fundação de Amparo à Pesquisa do
Estado de Minas Gerais. Papemig.
De Cauna, Jacques (ed.) 2004. Toussaint Louverture et l’indépendance d’Haïti. Paris:
Karthala-SFHOM.
Déchaine, Rose-Marie and Victor Manfredi (eds.) 1997. Object positions in Benue-
Kwa. The Hague: Institute of Generative Linguistics/Holland Academic Graphics.
Debien, Gabriel. 1956. Études antillaises (XVIIIe siècle). Paris: Association Marc Bloch.
1974. Les esclaves aux Antilles françaises (XVIIe–XVIIIe siècles). Basse-Terre:
Société d’histoire de la Gouadeloupe.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
324 References
DeGraff, Michel. 1992. Creole grammars and acquisition of syntax: The case of Haitian
Creole. PhD dissertation, University of Pennsylvania.
1995. On certain differences between Haitian and French predicative constructions.
In J. Amastae, G. Goodall, M. Montalbetti, and M. Phinney (eds.), Contemporary
research in romance linguistics. Amsterdam: Benjamins, 237–256.
1997. Verb syntax in, and beyond creolization. In Liliane Haegeman (ed.), The new
comparative syntax. London: Longman, 64–94.
(ed.) 1999. Language creation and language change. Cambridge, MA: MIT Press.
2001a. On the origin of creoles: A cartesian critique of neo-Darwinian linguistics.
Linguistic Typology 5: 213–310.
2001b. Morphology in creole genesis. In Michael Kenstowicz (ed.), Ken Hale: A life
in language. Cambridge MA: MIT Press, 53–121.
2002. Relexification: A reevaluation. Anthropological Linguistics 44: 321–414.
2003. Against creole exceptionalism. Language 79: 391–410.
2005. Morphology and word order in “creolization” and beyond. In
Guglielmo Cinque and Richard Kayne (eds.), The Oxford handbook of
comparative syntax. New York: Oxford University Press, 293–372.
2007. Kreyòl Ayisyen, or Haitian Creole. In John Holm and Peter Patrick (eds.),
Comparative creole syntax: Parallel outlines of 18 creole grammars. London:
Battlebridge Publications, 101–126.
2009. Language acquisition in creolization (and language change): Some cartesian-
uniformitarian boundary conditions. Language and Linguistic Compass 3/4: 888–971.
D’Elbée, François. 1671. Journal du voyage du Sieur Delbée. In Jean de Clodoré (ed.),
Relation de ce qui s’est passé dans les Isles et Terre-ferme de l’Amerique pendant
la derniere guerre avec l’Angleterre, et depuis en execution du Traitté de Breda ii.
Paris.
Den Besten, Hans. 1977. On the interaction of root transformations and lexical deletive
rules. In Studies in West Germanic syntax. PhD dissertation, University of Tilburg.
Den Dikken, Marcel. 1998. Predicate inversion in DP. In Artemis Alexiadou and
Chris Wilder (eds.), Possessors, predicates and movement in the Determiner
Phrase. Amsterdam: Benjamins, 177–214.
Denis, Serge. 1935. Notre créole. Orleans: G. Luzeray.
Déprez, Viviane. 2001. Determiner architecture and phrasal movement in French
lexifier creoles. In Josep Quer, Jan Schroten, Mauro Scorretti, Petra Sleeman,
and Else Verheugd (eds.), Romance languages and linguistic theory 2001.
Amsterdam: Benjamins, 49–74.
2004. Morphological number, semantic number and bare nouns. Lingua 115: 857–883.
2007. Nominal constituents in French lexifier creoles: Probing the structuring role of
grammaticalization. Journal of Pidgin and Creole Languages 22: 263–308.
Déprez Viviane, Petra Sleeman, and Hakima Guella. 2011. Specificity effects in L2
determiner acquisition: UG or pragmatic egocentrism? In Mihaela Pirvulescu,
María Cristina Cuervo, Ana T. Pérez-Leroux, Jeffrey Steele, and Nelleke Strik
(eds.), Selected proceedings of the fourth conference on Generative Approaches to
Language Acquisition – North America (GALANA 2010). Somerville, MA:
Cascadilla Proceedings Project, 27–36.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 325
Diouf, Sylviane A. 2007. Dreams of Africa in Alabama: The slave ship Clotilda and the
story of the last Africans brought to America. New York: Oxford University Press.
Donati Caterina and Chiara Branchini. 2013. Challenging linearization: Simultaneous
mixing in the production of bimodal bilinguals. In Theresa Biberauer and
Ian Roberts and (eds.), Challenges to linearization. Berlin: Mouton, 93–128.
Donicie, Anton C. R. and Jan Voorhoeve. 1963. De Saramakaanse Woordenschat.
Bureau voor Tallonderzoek in Suriname van de Universiteit van Amsterdam.
Ducœurjoly, S. J. 1803. Manuel des habitants de Saint-Domingue. Paris: Arthus
Bertrand.
Durrleman-Tame, Stephanie. 2008. The syntax of Jamaican Creole: A cartographic
perspective. Amsterdam: Benjamins.
Eltis, David. 2011. Africa, slavery, and the slave trade, mid-seventeenth to mid-
eighteenth centuries. In Nicholas Canny and Philip Morgan (eds.), The Oxford
handbook of the Atlantic world: 1450–1850. New York: Oxford University Press,
271–286.
Emmer, Peter C. 2005. Les Pays-Bas et la traite des noirs. Paris: Karthala.
Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22: 1–26.
Essegbey, James. 1999. Inherent complement verbs revisited: Towards an understanding
of argument structure in Ewe. Wageningen: MPI series, Ponsen & Looijen bv.
Fattier, Dominique. 1996. La passion de notre Seigneur selon St. Jean en langage nègre:
Une scripta créole? Études créoles 19: 9–30.
Foley, William A. and Robert D. Van Valin.1984. Functional syntax and universal
grammar. Cambridge University Press.
Fouchard, Jean. 1972. Les marrons de la liberté. Paris: Éditions de l’École.
Gibson, Kean. 1986. The ordering of auxiliary notions in Guyanese Creole. Language
62: 571–586.
Giusti, Giuliana. 1996. Is there a FocusP and a TopicP in the noun phrase structure?
University of Venice Working Papers in Linguistics 6: 105–128.
1997. The categorial status of determiners. In Liliane Haegeman (ed.), The new
comparative syntax. London: Longman Linguistics Library, 95–123.
Glaude, Herby. 2012. Aspects de la syntaxe du l’haïtien. Paris: Anibwe.
Glaude, Herby and Anne Zribi-Hertz. 2012. Verb focus in Haitian: From lexical
reiteration to predicate cleft. In Enoch O. Aboh, Norval Smith, and Anne Zribi-
Hertz (eds.), The morphosyntax of reiteration in creole and non-creole languages.
Amsterdam: Benjamins, 77–134.
Glock, Naomi. 1986. The use of reported speech in Saramaccan. In George Huttar and
Kenneth Gregerson (eds.), Pragmatics in non-Western perspective. Publications in
Linguistics. Dallas: Summer Institute of Linguistics, 35–61.
Goad, Heather and Lydia White. 2004. Ultimate attainment of L2 inflection: Effects of
L1 prosodic structure. In Susan Foster-Cohen, Michael Sharwood Smith,
Antonella Sorace, and Mits Ota (eds.), EUROSLA yearbook, Vol.IV. Amsterdam:
Benjamins, 119–145.
Goad, Heather, Lydia White, and Jefferey Steele. 2003. Missing inflection in L2
acquisition: Defective syntax or L1-constrained prosodic representations?
Canadian Journal of Linguistics 48: 243–263.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
326 References
Granfeldt, Jonas. 2000. The acquisition of the Determiner Phrase in bilingual and
second language French. Bilingualism: Language and Cognition 3: 263–280.
Guillemin, Diana. 2011. The syntax and semantics of a determiner system: A case study
of Mauritian Creole. Amsterdam: Benjamins.
Haddican, Bill. 2001. Gengbe typing morphemes and determiners. Unpublished
manuscript, New York University.
Haeberli, Eric. 1999. Features, categories and the syntax of A-positions: Synchronic and
diachronic variation in the Germanic languages. PhD dissertation, University of
Geneva.
Haegeman, Liliane. 1995. The syntax of negation. Cambridge University Press.
Haegeman, Liliane and Barbara Ürögdi. 2010a. Referential CPs and DPs: An operator
movement account. Theoretical Linguistics 36: 111–152.
2010b. Operator movement, referentiality and intervention. Theoretical Linguistics
36: 233–246.
Hagemeijer, Tjerk. 2001. Underspecificatoin in serial verb constructions. In
Norbert Corver and Henk van Riemsdijk (eds.), Semi-lexical categories: On the
function of content words and content of function words. Berlin: Mouton, 414–451.
Hale, Ken and Samuel J. Keyser. 1993. On argument structure and the lexical expression
of syntactic relations. In Ken Hale and Samuel. J. Keyser (eds.), The view from
building 20. Cambridge, MA: MIT Press, 53–109.
Hall, Robert A. 1953. Haitian Creole: Grammar, texts, vocabulary. Volume 43.
Philadelphia: American Folklore Society.
1962. The life cycle of pidgin languages. Lingua 11: 151–156.
Harbour, Daniel. 2008. Klivaj predika, or predicate clefts in Haitian. Lingua 118:
853–871
Hartmann, Katharina and Malte Zimmermann. 2007. Exhaustivity marking in Hausa: A
reanalysis of the particle nee/cee. In Enoch O. Aboh, Katharina Hartmann, and
Malte Zimmermann (eds.), Focus strategies in Niger Congo and Afro-Asiatic: On
the interaction between focus and grammar in some African languages. Berlin:
Mouton, 249–272.
Hawkins, J. A. 2004. Efficiency and complexity in grammar. New York: Oxford
University Press.
Haznedar, Belma and Bonnie D. Schwartz. 1997. Are there optional infinitives in child
L2 acquisition? In Elizabeth Hughes, Mary Hughes, and Annabel Greenhill (eds.),
Proceedings of the 21st annual Boston University conference on language
development. Somerville, MA: Cascadilla Press, 257–268.
Heim, Irene. 1991. Articles and definiteness. Published in German as “Artikel und
definitheit,” in Armin von Stechow and Dieter Wunderlich (eds.), Semantics: An
International handbook of contemporary research. Berlin: Mouton, 487–535.
Hengeveld, Kees. 1989. Layers and operators in Functional Grammar. Journal of
Linguistics 25: 127–157.
Henri, Fabiola. 2012. Attenuative verbal reduplication in Mauritian: A morpho-
semantic approach. In Enoch O. Aboh, Norval Smith, and Anne Zribi-Hertz
(eds.), The morphosyntax of retiteration in creole and non-creole languages.
Amsterdam: Benjamins, 203–233.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 327
Holm, John. 1988. Pidgins and creoles. Vol. I: Theory and structure. Cambridge
University Press.
Huttar, George, Enoch O. Aboh, and Felix Ameka. 2013. Relative clauses in Suriname
creoles and Gbe languages. Lingua 129: 96–123.
Ionin, Tania. 2006a. This is definitely specific: Specificity and definiteness in article
systems. Natural Language Semantics 14: 175–234.
2006b. A comparison of article semantics in L2-acquisition and creole languages.
In Claire Lefebvre, Lydia White, and Christine Jourdan (eds.), L2 acquisition and
creole genesis: Dialogues. Amsterdam: Benjamins, 253–273
Jaeggli, Osvaldo, and Nina M. Hyams. 1993. On the independence and interdependence
of syntactic and morphological properties: English aspectual come and go. Natural
Language and Linguistic Theory 11: 313–346.
Kangni, Atah-Ekoué. 1989. La syntaxe du ge. Franckfurt AM: Peter Lang.
Kayne, Richard S. 1975. French syntax. Cambridge, MA: MIT Press.
1994. The antisymmetry of syntax. Cambridge, MA.: MIT Press.
2002. On some prepositions that look DP-internal: English of and French de. Catalan
Journal of Linguistics 1: 71–115.
Kiss, Katalin É. 1998. Identificational focus versus information focus. Language 74:
245–273.
Klein, Wolfgang and Clive Perdue. 1997. The Basic Variety (or: Couldn’t natural
languages be much simpler?). Second Language Research 13: 301–347.
Kluge, Angela. 2000. The Gbe languages varieties of West Africa: A quantitative
analysis of lexical grammar features. MA dissertation, University of Wales,
College of Cardiff.
2005. A synchronic lexical study of Gbe language varieties: The effects of different
similarity judgment criteria. Linguistic Discovery 3: 22–53.
Koopman, Hilda. 1982a. Les constructions relatives. In Claire Lefebvre, Hélène Magloire-
Holly, and Nanie Piou (eds.), Syntaxe de l’Haïtien. Ann Arbor, MI: Karoma, 167–203.
1982b. Les questions. In Claire Lefebvre, Hélène Magloire-Holly, and Nanie Piou
(eds.), Syntaxe de l’Haïtien. Ann Arbor, MI: Karoma, 204–241.
1984. The syntax of verbs: From verb movement rules in the Kru languages to
Universal Grammar. Dordrecht: Foris.
Kouwenberg, Silvia. 2009. The invisible hand in creole genesis: Reanalysis in the
formation of Berbice Dutch. In Enoch O. Aboh and Norval Smith (eds.),
Complex processes in new languages. Amsterdam: Benjamins, 115–158.
2010. Creole studies and linguistic typology: Part 2. Journal of Pidgin and Creole
Languages 25: 359–380.
Kroch, Anthony. 1989a. Reflexes of grammar in patterns of language change. Language
Variation and Change 1, 2: 199–244.
1989b. Function and grammar in the history of English: Periphrastic do. In Ralph W.
Fasold and Deborah Schiffrin (eds.), Language Change and Variation. Current
Issues in Linguistic Theory, vol. 52. Amsterdam: Benjamins, 133–172.
1994. Morphosyntactic variation. In Katharine Beals (ed.), Papers from the 30th
regional meeting of the Chicago Linguistics Society: Parasession on variation and
linguistic theory. Chicago Linguistic Society, 180–201.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
328 References
2001. Syntactic change. In Mark Baltin and Chris Collins (eds.), The handbook of
contemporary syntactic theory. Malden, MA: Blackwell, 699–729.
Kroch, Anthony and Ann Taylor. 1997 Verb movement in Old and Middle
English: Dialect variation and language contact. In Ans van Kemenade and
Nigel Vincent (eds.), Parameters of morphosyntactic change. Cambridge
University Press, 297–325.
Anthony Kroch, Ann Taylor, and Donald Ringe. 2000. The Middle English verb-second
constraint: A case study in language contact and language change. In Susan
C. Herring, Pieter van Reenen, and Lene Schøsler (eds.), Textual parameters in
older languages. Amsterdam: Benjamins, 353–392.
Labouret, Henri and Paul Rivet. 1929. Le royaume d’Arda et son évangélisation au
17ème siècle. Paris: Institut d’Ethnologie.
Lardiere, Donna. 2000. Mapping features to forms in second language acquisition.
In John Archibald (ed.), Second language acquisition and linguistic theory.
Malden MA: Blackwell, 102–129.
2005. On morphological competence. Unpublished manuscript, Georgetown University.
Lalla, Barbara and Jean D’Costa. 1991. Language in exile: Three hundred years of
Jamaican Creole. Tuscaloosa, AL: The University of Alabama Press.
Languages of Suriname: Saramaccan-English dictionary. www.sil.org/americas/suri
name/Saramaccan/English/SaramEngDictIndex.html
Law, Robin. 1991. The Slave Coast of West Africa, 1550–1750. New York: Oxford
University Press.
1994. The slave trade in seventeenth-century Allada: A revision. African Economic
Histrory 22: 59–92.
1997. The Kingdom of Allada. Leiden: CNWS Publications, Leiden University.
2011. Africa in the Atlantic world, c.1760–c.1840. In Nicholas Canny and
Philip Morgan (eds.), The Oxford handbook of the Atlantic world: 1450–1850.
New York: Oxford University Press, 585–601.
Lefebvre, Claire. 1989. Instrumental take serial constructions in Haitian and in Fon.
Revue canadienne de linguistique 34: 319–339.
1991. Take serial verb constructions in Fon. In Claire Lefebvre (ed.), Serial verbs:
Grammatical, comparative and cognitive approaches. Amsterdam: Benjamins, 37–79.
1998. Creole genesis and the acquisition of grammar: The case of Haitian creole.
Cambridge University Press.
Lefebvre, Claire and Anne Marie Brousseau. 2002. A grammar of Fongbe. Berlin/New
York: Mouton.
Leimgruber, Jakob R. E. 2009. Modelling variation in Singapore English. PhD
dissertation, Pembroke College, University of Oxford.
Ler, Soon Lay Vivien. 2005. An in-depth study of discourse particles in Singapore
English. PhD dissertation, National University of Singapore.
Lim, Lisa. 2011. Tone in Singlish: Substrate features from Sinitic and Malay. In
Claire Lefebvre (ed.), Creoles, their substrates and language typology.
Amsterdam: Benjamins, 271–287.
Lightfoot, David. 2006. How new languages emerge. Cambridge University Press.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 329
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
330 References
1987. An issue on predicate clefting: Evidence from Atlantic creoles and African
languages. In Philippe Maurer and Thomas Stolz (eds.), Varia Creolica. Bochum:
Brockmeyer, 71–89.
1989. Equivocal structure in some Gullah complex sentences. American Speech 64:
304–326.
1991. Language genesis and human evolution. (Review article on Bickerton 1990.)
Diachronica 8: 239–254.
1994. Genèse de population et genèse de langue. Plurilinguismes 8: 95–113.
1996. The founder principle in creole genesis. Diachronica 13: 83–134.
1999. On the language bioprogram hypothesis: Hints from Tazie. In Michel DeGraff
(ed.), Language creation and language change: Creolization, diachrony, and
development. Cambridge, MA: MIT Press, 95–127.
2001. The ecology of language evolution. Cambridge University Press.
2002 Competition and selection in language evolution. Selection 3: 45–56.
2003. Créoles, écologie sociale, évolution linguistique. Paris: L’Harmattan.
2005a. Créoles, écologie sociale, évolution linguistique: Cours donnés au Collège de
France durant l’automne 2003. Paris: L’Harmattan.
2005b. Language evolution: The population genetics way. In Guenther Hauska (ed.),
Gene, Sprachen, und ihre Evolution. Universitaetsverlag Regensburg, 30–52.
2005c. How many bes are there in English? In Salikoko S. Mufwene, Elaine J.
Francis, and Rebecca S. Wheeler (eds.), Polymorphous linguistics: Jim
McCawley’s legacy. Cambridge, MA: MIT Press, 225–246.
2008. Language evolution: Contact, competition, and change. New York: Continuum
International Publishing Company Group.
2009. Restructuring, hybridization, and complexity in language evolution. In
Enoch O. Aboh and Norval Smith (eds.), Complex processes in new languages.
Amsterdam: Benjamins, 367–400.
2010. SLA and the emergence of creoles. Studies in Second Language Acquisition 32:
359–400.
2014. Globalisation économique mondiale des XVIIe-XVIIIe siècles, émergence
des créoles, et vitalité langagière. In Arnaud Carpooran and Yannick Bosquet-
Ballah (eds.), Langues créoles, mondialisation et éducation, Proceedings of the
13th colloquium of the Comité International des Études Créoles, Mauritius
2012. Louis, Mauritius: Creole Speaking Unit, 23–79.
Müller, Natascha. 1998. Transfer in bilingual first language acquisition. Bilingualism:
Language and Cognition 1: 151–171
Müller, Natascha and Aafke Hulk. 2000. Bilingual first language acquisition at the
interface between syntax and pragmatics. Bilingualism: Language and Cognition
3: 227–244.
2001. Cross-linguistic influence in bilingual first language acquisition: Italian
and French as recipient languages. Bilingualism: Language and Cognition 4:
1–21.
Muysken, Pieter. 1981a. Half-way between Quechua and Spanish: The case for
relexification. In Arnold Highfield and Albert Valdaman (eds.), Historicity and
variation in creole studies. Ann Arbor, MI: Karoma, 52–78.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 331
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
332 References
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 333
Roberts, Ian and Ana Roussou. 2003. Syntactic change: A minimalist approach to
grammaticalization. Cambridge University Press.
Roeper, Thomas. 1999. Universal bilingualism. Bilingualism: Language and Cognition
2: 169–186.
Roodenburg, Jasper. 2004. Pour une approche scalaire de la déficience nominale: La
position du français dans une théorie des “noms nus.” LOT PhD series 99.
Utrecht: LOT.
Ross, J. Robert. 1967. Constraints on variables in syntax. PhD dissertation, MIT,
Cambridge, MA.
Rountree, S. Catherine. 1992. Saramaccan grammar sketch. Languages of the
Guianas 8. Paramaribo: Summer Institute of Linguistics.
Rountree, S. Catherine and Naomi Glock. 1982. Saramaccan for beginners: A
pedagogical grammar of the Saramaccan language. Paramaribo: Institut voor
Taalwetenschap and Summer Institute of Linguistics.
Saah, Kofi. 2010. Relative clauses in Akan. In Enoch O. Aboh and James Essegbey
(eds.), Topics in Kwa syntax. Dordrecht: Springer, 91–107.
Schumann, Christian. 1783. Neger-Englishes Wörterbuch. Manuscript published in
André Kramp (1983) Early creole lexicography: A study of C. L. Schumann’s
manuscript dictionary of Sranan, pp. 44–305. Dissertation, University of
Leiden.
Schumann, John H. 1978. The pidginization process: A model for second language
acquisition. Rowley: Newbury House.
Sebba, Mark. 1987. The syntax of serial verbs: An investigation into serialization in
Sranan and other languages. Amsterdam: Benjamins.
1997. Contact languages: Pidgins and creoles. Houndmills: PALGRAVE.
Seuren, Pieter. 1990a. Still no serials in Seselwa: A reply to “Seselwa serialization and
its significance.” Journal of Pidgin and Creole Languages 5: 271–292.
1990b. Verb syncopation and predicate raising in Mauritian Creole. Linguistics 28:
809–844.
1991. The definition of serial verbs. In Francis Byrne and Thom Huebner (eds.),
Development and structures of Creole. Amsterdam: Benjamins, 193–206.
Shappeck, Marco. 2011. Quichua-Spanish language contact in Salcedo, Ecuador:
Revisiting Media Lengua Syncretic Language Practices. PhD dissertation,
University of Illinois, Urbana-Champaign.
Schwartz, Bonnie and Rex Sprouse. 1996. L2 cognitive states and the full transfer/full
access model. Second Language Research 12: 40–77.
Siegel, Jeff. 2004. Morphological elaboration. Journal of Pidgin and Creole Languages
19: 333–362.
2008. The emergence of pidgin and creole languages. Oxford University Press.
Silveira, Luis. 1945. Œuvre nouvelle de langue generale du mina. Manuscript
de la bibliothèque publique de Evora et de la bibliothèque nationale de
Lisbone. Département des publications et bibliothèque Agence générale des
colonies.
Singler, John Victor. 1988. The homogeneity of the substrate as a factor pidgin/creole
genesis. Language 64: 27–51.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
334 References
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
References 335
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
336 References
Clive Upton (eds.), A handbook of varieties of English. Vol. II: Morphology and
syntax. Berlin: Mouton, 154–174.
Wakefield, John C. 2012. A floating tone discourse morpheme: The English equivalent
of Cantonese lo1. Lingua 122: 1739–1762.
Weerman, Fred. 2011. Diachronic change: Early versus late acquisition. Bilingualism:
Language and Cognition 14: 149–151.
Whinom, Keith. 1971. Linguistic hybridization and the “Special Case” of Pidgins and
Creoles. In Dell H. Hymes (ed.), Pidginization and creolization of languages.
Cambridge University Press, 91–115.
Wijnen, Beppie and Mervyn Alleyne. 1987. A note on fu in Saramaccan. In Mervin
C. Alleyne (ed.), Studies in Saramaccan language structure. Caribbean Culture
Studies 2, University of Amsterdam, 41–51.
Winford, Donald. 1985. The syntax of Fi complements in Caribbean English Creole.
Language 61: 588–624.
1993. Predication in Caribbean English Creoles. Amsterdam: Benjamins.
Winford, Donald and Bettina Migge. 2007. Substrate influence on the emergence of the
TMA systems of the Surinamese creoles. Journal of Pidgin and Creole Languages
22: 73–99.
Wurmbrand, Suzanne. 2001. Infinities. Restructuring and clause structure. Berlin:
Mouton.
2004. Two types of restructuring: Lexical versus functional. Lingua 114: 991–1014.
Yuka, Lendzemo Constantine and Esohe Mercy Omoregbe. 2010. The internal structure
of the Edo verb. California Linguistic Notes 35: 1–19.
Zwart, Jan-Wouter. 1997a. Morphosyntax of verb Movement: A minimalist approach to
the syntax of Dutch. Dordrecht: Kluwer.
1997b. The Germanic SOV languages and the universal base hypothesis. In
Liliane Haegeman (ed.), The new comparative syntax. London: Longman
Linguistics Library, 246–267.
2006. Complementizer agreement and dependency marking typology. In Marjo
van Koppen, Frank Landsbergen, Mika Poss, and Jenneke van der Wal (eds.),
Special issue of Leiden Working Papers in Linguistics 3: 53–72.
Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 07 Jul 2017 at 03:33:50, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167.010
Author index
Aboh, i, ii, iii, iv, ix, x, xi, xii, xiii, xiv, xvi, 3, 8, Bailey, 214
9, 10, 11, 12, 13, 14, 17, 41, 42, 43, 44, 45, Baker, 140, 149, 193, 281, 282, 283, 284, 285,
53, 61, 69, 87, 88, 92, 98, 101, 102, 107, 113, 298, 334
119, 123, 134, 135, 137, 138, 139, 141, 143, Bakker, 6, 145, 168, 230, 242
144, 145, 146, 149, 154, 155, 156, 157, 159, Bernstein, 177, 214
162, 163, 171, 173, 176, 180, 181, 185, 186, Bickerton, x, 6, 7, 61, 62, 83, 84, 85, 86, 87,
192, 193, 196, 198, 203, 205, 206, 207, 208, 88, 89, 90, 91, 92, 93, 109, 171, 175, 176,
214, 215, 217, 221, 222, 226, 227, 228, 229, 194, 209, 269, 270, 281, 297, 303, 330
230, 231, 232, 233, 234, 236, 238, 239, 246, Bobyleva, xvi, 64, 87, 92, 101, 204, 214
248, 249, 250, 251, 252, 253, 254, 256, 257, Boeckx, 140
258, 259, 260, 261, 262, 264, 266, 270, 271, Booij, 101, 102, 103, 104
272, 276, 281, 282, 284, 285, 288, 289, 290, Borer, 101, 140
291, 298, 307, 318, 330, 334 Brousseau and Lumsden, 177
Aboh and Ansaldo, 13, 115, 143, 145, 169, Bruyn, xv, 16, 156, 177, 213, 215,
194, 195 Burzio, 287
Aboh and DeGraff, 17, 43, 61, 77, 87, 88, 126, Bybee, 89, 193
173, 176, 177, 179, 180, 182, 188, 194, 206, Byrne, 258, 270, 333
208, 212, 225, 265, 266, 310
Aboh and Dyakonova, 14, 47, 139, 146, 149, Cable, 65, 295
222, 254, 257, 266, 281 Campbell, 216, 282
Aboh and Essegbey, 41, 42, 44, 45, 46, 47, 176, Capo, 20, 41, 42
250, 253 Cardinaletti and Giusti, 292
Aboh and Nauze, 236 Cardinaletti and Shlonsky, 292
Aboh and Pfau, 203, 238 Cardinaletti and Starke, 69
Aboh and Smith, 61, 155, 161, 162, 163, 164, Chaudenson, ii, x, xi, 7, 18, 33, 59, 62, 63, 64,
167, 168, 169, 170, 186, 289, 306 65, 67, 68, 69, 70, 72, 73, 74, 83, 93, 99, 109,
Aboh, Hartmann, and Zimmermann, 47 117, 118, 134, 135, 175, 179
Aboh, Pfau, and Zeshan, 88 Cheng and Sybesma, 87
Aboh, Smith, and Veenstra, 280 Chomsky, 2, 3, 9, 13, 105, 140, 149, 257, 294,
Adande, 20, 22, 23, 24, 32 308, 322
Aikhenvald, 269, 275 Cinque, 12, 14, 89, 102, 140, 185, 186, 192,
Alleyne, 13, 91, 97, 331 193, 271, 287, 291, 292, 293, 294, 295, 317,
Ameka, xv, 44, 45, 46, 281 318, 324
Anglade, 17, 58 Clements, ii, 101
Anonymous, 246 Collins, 263, 276, 282, 321
Ansaldo, ii, xvi, 8, 13, 98, 114, 184, 211, Comrie, 89
259, 304 Croft, 3, 139
Ansaldo, Matthews, and Lim, 61 Curtin, 29, 323
Arends, 10, 16, 33, 35, 39, 53, 97, 137, 255,
322, 329 d’Elbée, 23, 24, 29, 30, 31, 36, 37
Aronoff, Meir, and Sandler, 312 Da Cruz, 246, 282, 284, 292
Avolonto, 145 Damonte, 227, 228, 229, 232, 233
337
Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 06 Jan 2020 at 22:09:37, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
338 Author index
Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 06 Jan 2020 at 22:09:37, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Author index 339
Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 06 Jan 2020 at 22:09:37, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Language index
340
Downloaded from https://www.cambridge.org/core. UVA Universiteitsbibliotheek, on 14 Mar 2020 at 09:13:15, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Language index 341
Downloaded from https://www.cambridge.org/core. UVA Universiteitsbibliotheek, on 14 Mar 2020 at 09:13:15, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
342 Language index
230, 231, 232, 233, 234, 235, 236, 161, 166, 167, 170, 171, 175, 177, 186,
237, 238, 239, 240, 241, 242, 243, 244, 187, 188, 189, 190, 191, 192, 195, 205,
245, 248, 249, 250, 251, 252, 253, 254, 206, 211, 212, 213, 214, 215, 216, 217,
255, 257, 258, 259, 263, 264, 265, 266, 218, 221, 254, 270, 272, 283, 320, 321,
267, 271, 272, 273, 278, 279, 280, 285, 333, 335, 336
286, 292, 293, 294, 298, 302, 303, 307, Sri Lanka Malay, 195, 320
308, 309, 310, 311, 318, 319, 320, 322, Standard English, 118, 241
323, 325, 328, 331, 332, 333, 334, Swedish, 107
335, 336
Seychelles, 93, 204, 205 Tayo, 93, 205
Singapore English, ii, 196, 200, 328 Tok Pisin, 93, 205
Sinitic, 100, 102, 107, 310, 328 Tubaian, 111, 113, 114, 115
South Hollandic Dutch, 106
Spanish, ii, 24, 29, 31, 48, 49, 65, 74, 78, 79, Western Gbe, 41, 106, 119, 218, 219, 220, 221,
107, 109, 214, 270, 330, 333, 335 245, 246, 247
Sranan, ix, 9, 10, 11, 47, 59, 91, 92, 93, 116,
136, 137, 154, 155, 156, 158, 159, 160, Yoruba, 20, 34, 38, 40, 58, 119, 255
Downloaded from https://www.cambridge.org/core. UVA Universiteitsbibliotheek, on 14 Mar 2020 at 09:13:15, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Subject index
acquisition, i, ii, v, xiii, xiv, xv, 3, 5, 6, 7, 13, 15, colonial language, 18, 60, 79, 121
51, 60, 61, 62, 72, 74, 84, 85, 90, 91, 93, 94, Compagnie des Indes Occidentales, 23
95, 96, 97, 98, 99, 107, 108, 110, 111, 114, Company of Royal Adventurers, 29
116, 130, 134, 136, 138, 140, 141, 142, 145, competition and selection, xiii, 9, 13, 14, 17,
159, 169, 173, 174, 185, 192, 194, 195, 196, 59, 61, 98, 116, 117, 120, 125, 132, 133, 134,
199, 200, 210, 211, 216, 249, 294, 304, 305, 135, 136, 137, 141, 142, 143, 144, 145, 153,
306, 307, 308, 313, 314, 316, 317, 324, 325, 154, 166, 168, 170, 186, 192, 210, 241, 254,
326, 327, 328, 329, 330, 331, 332, 333, 335, 259, 264, 271, 294, 304, 312, 315,
336 complementizer system, 11, 14, 144, 170, 196,
Adia/Adja. See Aja 197, 222, 226, 229, 232, 238, 239, 240, 249,
Agni, 58 252, 253, 308
Aja, 47 complexity, xii, 141, 169, 305, 306, 307, 311,
Aja descent, 36, 38, 39, 40, 41, 47, 52, 58, 59, 312, 315, 319, 326, 329, 330
60, 114, 117, 119, 123, 304 compound, 57, 108, 154, 155, 156, 157
Aja Nengre, 52 comp-system, 11, 14, 233, 269
Aja-Tado, 20, 34, 36, 39, 331 Congo, vi, 6, 8, 55, 56, 57, 58, 257, 309,
Allada, vii, x, 16, 17, 19, 20, 21, 22, 23, 24, 26, 318, 326
27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, Congos, 55, 56
39, 48, 51, 53, 54, 58, 304, 319, 328 congruence, xiii, 51, 63, 66, 135, 156, 160,
Arada. See Allada 211, 303
Ashanti, 58 contextual inflection, 101, 102, 104, 109
contrastive focus, 172, 173, 250, 251
Badagri, 16, 22 Costa Peixoto, 122, 123, 124, 126
Bantu, 39, 54 creoles, xi, 55, 56
Baoule, 58
bare nouns, 2, 10, 11, 43, 77, 87, 88, 171, 178, Dahomey, 22, 35
179, 180, 181, 187, 188, 189, 206, 212, 310, Danish African Company, 29
322, 324 definite determiner, 66, 68, 71, 77, 81, 85, 181,
Basic Variety, 107, 327 182, 209, 210
Beni, 20, 24, 26, 27, 32 definiteness, 9, 10, 12, 14, 63, 67, 68, 71, 80,
Bight of Benin, 19, 33, 38, 47, 58, 304 87, 88, 142, 172, 176, 177, 190, 212, 216,
bilingualism, ii, 5, 173, 333 310, 311, 326, 327, 331
bozals, 18 deontic modality, 14, 224, 225, 226, 227, 229,
Brazil, 28, 29, 122, 124, 135 231, 232, 234, 235, 236, 237, 240, 247, 268
break in transmission, 7, 83, 174 derivational morphology, 12, 85, 194, 195
Bulfinch Lamb, 33, 35 determiners, xiv, 2, 14, 47, 71, 72, 77, 79, 108,
176, 177, 179, 180, 181, 182, 186, 188, 189,
Central African languages, 17 190, 191, 192, 207, 209, 216, 217, 261, 311,
Chevalier Marchais, 31 321, 325, 326, 329
clausal left periphery, 14, 110, 185, 226, 249, DNA, 139, 141
251, 252, 253, 257, 265, 268 Dutch, 21, 23, 24, 27, 28, 30, 33, 34, 37
clause union. See restructuring Dutch West India Company, 28, 34
343
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 15 Feb 2020 at 07:19:01, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
344 Subject index
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 15 Feb 2020 at 07:19:01, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
Subject index 345
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 15 Feb 2020 at 07:19:01, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167
346 Subject index
Downloaded from https://www.cambridge.org/core. University of Texas Libraries, on 15 Feb 2020 at 07:19:01, subject to the Cambridge Core
terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139024167