Codeswitching in Igbo–English Bilingualism: A Matrix Language Frame Account 9781474278140, 9781474278171, 9781474278164

Citation preview

Codeswitching in Igbo–English Bilingualism

Also Available From Bloomsbury Agreement, Pronominal Clitics and Negation in Tamazight Berber, Hamid Ouali Emerging Bilingual Speech, Anna Verschik The Languages of Urban Africa, edited by Fiona Mc Laughlin The Syntax of Mauritian Creole, Anand Syea

Codeswitching in Igbo–English Bilingualism A Matrix Language Frame Account Kelechukwu Ihemere

BLOOMSBURY ACADEMIC Bloomsbury Publishing Plc 50 Bedford Square, London, WC1B 3DP, UK BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of Bloomsbury Publishing Plc First published in Great Britain 2016 Paperback edition first published 2018 © Kelechukwu Ihemere, 2016 Kelechukwu Ihemere has asserted his right under the Copyright, Designs and Patents Act, 1988, to be identified as the author of this work. For legal purposes the Acknowledgements on p. xii constitute an extension of this copyright page. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. Bloomsbury Publishing Plc does not have any control over, or responsibility for, any third-party websites referred to or in this book. All internet addresses given in this book were correct at the time of going to press. The author and publisher regret any inconvenience caused if addresses have changed or sites have ceased to exist, but can accept no responsibility for any such changes. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data Names: Ihemere, Kelechukwu U. (Kelechukwu Uchechukwu) author. Title: Codeswitching in Igbo-English bilingualism : a Matrix Language Frame account / Kelechukwu Ihemere. Description: London ; New York : Bloomsbury Academic, 2016. | Includes bibliographical references and index. Identifiers: LCCN 2016016220 (print) | LCCN 2016023283 (ebook) | ISBN 9781474278140 (hardback) | ISBN 9781474278164 (epdf) | ISBN 9781474278157 (epub) Subjects: LCSH: Code switching (Linguistics)—Africa, West. | Languages in contact— Africa, West. | Igbo language—Grammar, Comparative—English. | English language— Grammar, Comparative—Igbo. | Igbo language—Foreign elements—English. | English language—Foreign elements—Igbo. | Igbo language—Influence on English. | English language—Influence on Igbo. | Multilingualism—Africa, West. | Africa, West—Languages. | BISAC: LANGUAGE ARTS & DISCIPLINES / Linguistics / Syntax. | LANGUAGE ARTS & DISCIPLINES / Linguistics / Phonetics & Phonology. | LANGUAGE ARTS & DISCIPLINES / Linguistics / General. Classification: LCC P115.5.A358 I43 2016 (print) | LCC P115.5.A358 (ebook) | DDC 306.44/60966—dc23 LC record available at https://lccn.loc.gov/2016016220 ISBN: HB: 978-1-4742-7814-0 PB: 978-1-3500-6655-7 ePDF: 978-1-4742-7816-4 ePub: 978-1-4742-7815-7 Typeset by RefineCatch Limited, Bungay, Suffolk To find out more about our authors and books visit www.bloomsbury.com and sign up for our newsletters.

For Ira, an inspiration in generosity and kindness

Contents List of Figures List of Tables Preface Acknowledgements List of Abbreviations and Symbols

1 2 3 4 5 6 7 8 9

Introduction Studying the Grammar of Codeswitching Theoretical Framework Comparison of Aspects of Igbo and English Grammars Methodology Embedded Language Single Words: Nouns and Adjectives Embedded Language Single Words: Verbs Embedded Language Islands Concluding Remarks and Implications

Appendix A: Sample Interview Protocol Appendix B: Summary of Findings from the Interview Protocol Notes References Index

viii viii x xii xiv 1 19 33 51 71 91 127 163 187 207 211 215 219 233

Figures 4.1 6.1 6.2 7.1 7.2 7.3 7.4 7.5 7.6

Igbo standard orthography with the corresponding IPA symbols ML according to morpheme order and late system morpheme criteria The ML: Igbo Syllable tree diagram of the word [.dɪs.kɔ.va.ra.] Syllable tree diagram of the word [.i.kloz.] Spectrogram of the word ‘discover+ra’ Spectrogram of the word ‘e+register+ghi.’ Spectrogram of the word ‘i+close’ Spectrogram of the word ‘i+destabilise’

52 118 120 141 142 145 146 147 148

Tables 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5.1 6.1 6.2 7.1 7.2

Igbo vowel correspondences and feature classes Summary of Igbo verbal morphology Summary of English verbal morphology Conjugation of zu. ‘to buy’ Six Igbo pronominal modifiers Some Igbo numerals Igbo pronouns Igbo conjunctions Some typological contrasts between Igbo and English Distribution of speakers according to occupation and sex Configuration of a basic Igbo and English NP/DP ML according to source of morpheme order and late system morpheme criteria Igbo vowel correspondences and feature classes English vowels

54 59 60 60 63 64 65 66 70 73 94 119 139 140

Figures and Tables

7.3

8.1 8.2

9.1 9.2 9.3

ML according to source of morpheme order and late system morpheme criteria: singly occurring EL verbs in Igbo–English CS ML according to source of morpheme order and late system morpheme criteria: EL islands in Igbo–English CS ML according to source of morpheme order and late system morpheme criteria: non-­internal EL islands in Igbo–English CS Summarized comparison of Igbo and English phonologies Borrowing versus codeswitching forms Codeswitching patterns in Igbo–English

ix

154 170

180 199 200 202

Preface The study reported in this book is unique in that it uses naturalistic data from fifty educated adult Igbo–English bilinguals resident in Port Harcourt, SouthEastern Nigeria to evaluate the Matrix Language Frame model of codeswitching. It establishes that the data can indeed be considered a classic case of codeswitching in that a matrix language can be clearly identified in bilingual clauses. This is established through both qualitative and quantitative analyses that make use of the typological contrasts between Igbo and English to uncover supportive evidence for the Matrix Language Frame model and its associated three principles: the Matrix Language Principle, the Asymmetry Principle and the Uniform Structure Principle. The investigation goes one step further by using spectrograms and the analysis of vowel harmony between English free morphemes and Igbo bound affixes to demonstrate that two phonologies can coexist in codeswitching and that codeswitching forms are essentially pronounced with a phonology that does not entirely resemble that of the matrix language variety. Furthermore, the study finds that the same language production mechanisms as detailed under the Matrix Language Frame model and its associated three principles underlie both single word and multi-­word codeswitching. That is, the present study, like those before it adopting the Matrix Language framework underlines the importance of the assumptions underpinning the Matrix Language Principle: (1) that language production is modular; (2) that lexical structure is both complex and abstract; and (3) that languages in contact divide responsibilities in what they may contribute toward lexical structure during the production of mixed constituents. Moreover, the study finds that Igbo–English bilinguals can always sustain ready access to their mother tongue mental lexicon during online speech production and thus Igbo–English may duly be described as a ‘classic’ case of codeswitching. However, this study is not convinced about the necessity for positing the Early System Morpheme Hypothesis (ESMH), which states that only early system morphemes may be doubled in classic codeswitching. The study finds that such morphemes can already be doubled in Igbo grammar apart from those covered by the Early System Morpheme Hypothesis. Therefore, it does not seem that the ESMH adds much value to the analysis of cases involving double

Preface

xi

morphology in Igbo–English codeswitching. Also, the present study disagrees with the notion under the Matrix Language Frame model that bare forms occur in codeswitching due to lack of ‘sufficient congruence’ between an embedded language word and its matrix language counterpart, thus leading to the activation of a compromise strategy with the result that the embedded language content morpheme is not placed in a slot projected by its matrix language counterpart. Instead this study demonstrates that such forms occur in Igbo–English because the matrix language (Igbo) already possesses a syntactic model that allows for the occurrence of bare forms in its grammar. Therefore, the position of this study is that bare forms occur in Igbo–English not due to lack of sufficient congruence between embedded language forms and their matrix language counterparts but strictly in accordance with the matrix language grammar. Moreover, they are inserted in clause structure in slots projected by the matrix language syntax. Nevertheless, the basic theoretical notion that there is a matrix language versus embedded language hierarchy is overwhelmingly supported in our data, because the two languages (Igbo and English) do not both satisfy the roles of the matrix language contained in the Morpheme Order Principle and the System Morpheme Principle of the Matrix Language Frame model.

Acknowledgements As someone who grew up in a bilingual environment, I have always been fascinated by the ease with which bilinguals can insert either lone lexical items or multi-­word sequences – or both – from the other language into discourses framed by their base language. This study is an effort at uncovering the grammatical mechanisms underlying such insertions in relation to Igbo– English intrasentential codeswitching. In this regard, I am most fortunate to be a bilingual and know many others in Port Harcourt, Nigeria. I had established many thriving contacts twelve years ago in the city when I did fieldwork for my doctoral study. Therefore, it was easy to mobilize and relocate to Port Harcourt in the summer of 2011 for the fieldwork that generated the natural Igbo–English bilingual data utilized for the analysis reported in the present study. I owe a huge debt of gratitude to the fifty men and women who allowed me to record them in various conversations with their families and friends. My special thanks go to Engineer Chima Ezeigbo and his wife for introducing me to their friends and family. Without their cooperation this study would not have been possible. My sincere thanks and gratitude go to Professor Alex Warwick, my head of department at the University of Westminster, and members of the Research Management Group for granting me a much-­needed sabbatical to complete this monograph. I have also been much inspired by the work of Professor Margaret Deuchar on Welsh–English codeswitching. Reading her various publications made me believe that the analysis of Igbo–English bilingual data might just shed some light on what we know already about the nature of intrasentential codeswitching. I would also like to thank Dr Evershed Amuzu of the University of Ghana for sending me a number of his publications, which were very helpful. I am also delighted to call Professor Carol Myers-Scotton a friend. Her incredible strength and enthusiasm about all things to do with language contact is quite simply inspiring. To Professor Karen Corrigan of Newcastle University, thank you for your continued belief in my abilities as a researcher. To my dear father Dr Ezeigbo Thomas Ihemere, thank you for pointing me in the direction of that light that burns ever brighter. You know the light, and the light is in us. Finally, to the three most important people in my

Acknowledgements 

xiii

life – Ira, Chris and Tasha – thank you so very much for your patient endurance and unconditional love. K. Ihemere Newcastle upon Tyne, October 2015

Abbreviations and Symbols A ACC AdjP ADVP AgrP AL AP ART AspP ATR AUX C CL C/COMP COND CP CS DEM DemP DO DP EC EL ENCL ESMH FEM FHC FMC FOC FP FUT GC

adjective accusative adjective phrase adverbial phrase agreement phrase abstract level asymmetry principle article aspect phrase advanced tongue root auxiliary complement clitic complementizer conditional complementizer phrase (projection of complementizer) codeswitching demonstrative demonstrative phrase direct object determiner phrase equivalence constraint embedded language enclitic Early System Morpheme Hypothesis feminine functional head constraint free morpheme constraint focus functional projection future government constraint

Abbreviations and Symbols

H HAB I IMP INCP IND INF IO IP LF MA MASC ML MLF MLP MOP NEG NegP NP NSO NUM O PART PASS PAST PERF PF PL POSM POSS POT PP PREP PRN/PRON PROG PSC Q QP

head habitual inflection imperative inceptive indicative affirmative infinitive indirect object inflectional phrase logical form minimalist approach masculine matrix language matrix language frame matrix language principle morpheme order principle negative negation phrase noun phrase new standard orthography numeral object participle passive past tense perfective phonetic form plural possessum possessive potential prepositional phrase preposition pronoun progressive pronominal subject clitic quantifier quantifier phrase

xv

xvi

S SG SMP ST SVC TP USP UW V VH VP ø 1,2,3 *

Abbreviations and Symbols

subject singular system morpheme principle step tone serial verb construction tense phrase uniform structure principle Urban Wolof Igbo vowel prefix vowel harmony verb phrase zero ‘null’ first, second, and third persons ungrammatical (unacceptable)

Phonetic symbols [ ] σ . + C N O R CVCV

indicates phonetic transcription indicates a syllable indicates syllable boundary indicates morpheme boundary coda nucleus onset rhyme/rime consonant–vowel–consonant–vowel

1

Introduction Many studies exploring language contact phenomena reveal that when speakers of different languages come into contact the interaction of the two or more languages involved can lead to several linguistic outcomes, including the use of two (or more) languages in the same utterance (codeswitching: CS); one group involved in the contact may take into its language some words from the other group’s language to refer to things and concepts that are new to the first group (such words usually go on to become established borrowings), or a situation may arise where one of the languages in contact contributes all the surface-­level forms, while some of the abstract lexical structures that underlie the surface-­ level patterns come from the other language(s) (convergence). Other outcomes of language contact can include the birth of new languages in the case of Pidgins and Creoles, lexical attrition, incomplete second language acquisition and interference, or even the death of a language. Codeswitching, which we shall define as the use of elements from more than one language within a single clause or sentence (see section 5.4 for a discussion of the unit of analysis adopted in this study), is conceivably one of the most studied and discussed outcomes of language contact. Its study has been approached from two main dimensions: (1) the linguistic and (2) the sociolinguistic. Researchers concerned with the sociolinguistic aspects of CS tend to seek to identify patterns of occurrence of CS and the impact of social-­psychological factors on this bilingual behaviour; whereas those working from a purely linguistic dimension focus on the structural aspects of CS, their aim being to uncover the syntactic and morphological characteristics of codeswitched utterances. The main focus of this volume is on characterizing the structural aspects of Igbo–English CS. Since Igbo and English are two typologically distinct languages (as we shall outline in Chapter 4 of this book), we believe that uncovering what CS mechanisms are deployed by Igbo–English bilinguals in the process of CS would shed some light on our ever evolving understanding of this pervasive bilingual behaviour from a purely linguistic perspective. In the sections that follow below,

2

Codeswitching in Igbo–English Bilingualism

we shall introduce some of the representative studies of CS from both the linguistic (in section 1.1) and sociolinguistic (in section 1.2) dimensions. Since our primary focus is linguistic rather than sociolinguistic, we shall return in chapters 2 and 3 to review the linguistic approaches introduced here in more detail. We only present an overview here of studies of CS from the sociolinguistic dimension to highlight their contributions to our understanding of the subject. This is followed in section 1.3 by an overview of the grammatical studies of CS in West Africa, the region of the world where Igbo–English is spoken. This is done with a view to highlighting the contributions that these studies have made to the furtherance of our understanding of CS. Also, the discussion in section 1.3 will reveal that not much comprehensive work has been undertaken in the linguistic study of CS involving Igbo and English; an important gap which the present book intends to fill. In section  1.4, we shall introduce the context of the research. Section 1.5 outlines the aims and organization of the rest of the book.

1.1  The linguistic dimension Starting from the 1970s researchers such as Lance (1975), Timm (1975) and Pfaff (1979) began to work on grammatical aspects of CS focusing on switching within the sentence. Using mostly Spanish–English data, these researchers sought to uncover constraints on CS. For instance, Timm (1975: 477–8) proposed that switching between a pronominal subject/object and a finite verb and between a finite verb and an infinitive complement was not allowed. For her part, Pfaff (1979) suggested that switching was prohibited between a noun and adjective. It is important to observe that, in as much as several of the earlier proposed constraints on CS have been shown in the literature to be acutely language-­specific and unmotivated by any particular theoretical approach, these initial studies pointed to the fact that CS is not haphazard, but grammatically constrained. Following this group of researchers are others1 interested in formulating constraints on points in a sentence where codeswitching can take place on the grounds of surface-­level linear differences between the languages concerned. These researchers consider restrictions on CS along the lines of dissimilarities in word order, either across clauses or on phrases within clauses. In particular, Poplack’s (1980) Equivalence Constraint (EC) is based on this premise that switching is not permitted when the syntax of two languages does not match at a potential switch point (see section 2.1 for a review of the EC).

Introduction

3

Another constraint put forward by researchers within this group is the Free Morpheme Constraint (FMC). The FMC is reviewed in section  2.2 of this volume. According to the FMC, a switch may not occur between a bound morpheme and a lexical form unless the latter has been phonologically integrated into the language of the bound morpheme (Sankoff and Poplack 1981). Over the years, various researchers such as Bhatt (1997a, 1997b, 1999, 2001a, 2001b), Couto, Deuchar and Fusser (2015), Deuchar (2006) and Legendre and Schindler (2010) have reported CS examples that appear to falsify the predictions of both the EC and the FMC.2 Yet another group of researchers3 look for explanations at a more abstract level than linear structure. These researchers assert that the grammatical organization of CS can be accounted for in terms of the principles of current syntactic theories, even though these theories were initially formulated to explain monolingual data. They do not seem to recognize any theoretical (or useful) value in acknowledging the asymmetry between any two languages participating in CS cross-­linguistically (these studies are reviewed in sections 2.3–5). On the other hand, the Matrix Language Frame (MLF) model (see Chapter 3 for a detailed discussion of this CS framework) makes the case for a distinction between the Matrix Language (ML) and the Embedded Language (EL). The ML is claimed to play a dominant role in shaping the overall morphosyntactic properties of codeswitched utterances. In other words, the model posits two hierarchies in reference to mixed constituents: the two languages do not participate equally; only one language is the source of the abstract morphosyntactic frame. In chapters 6 to 8 of this volume we will assess the veracity of the notion of asymmetry in the roles played by the languages participating in CS using data from Igbo–English CS. Furthermore, in the study of CS, a distinction is usually made between intersentential and intrasentential CS. Both types are introduced below, where we also specify that the analyses reported in this book shall focus on examples of intrasentential CS in Igbo–English.

1.1.1  Types of codeswitching Different types of CS are identified in the literature (see Clyne 2003 and Matras 2009, for instance), but researchers often make a differentiation between intersentential and intrasentential CS. Intersentential CS typically involves switching from one language to the other at sentence/clause boundaries. We illustrate this type of switching in (1) below from Igbo–English CS.4

4

Codeswitching in Igbo–English Bilingualism

(1) Speaker 5: di gi o.-­no. ya? [PAUSE] O. -ga e-­so anyi. e-­je u. ka? ‘Is your husband in? Will he go to church with us?’ Speaker 10: m̩ ba, o.-­ga-wa-­ra Enugu eci. He’ll be back next week. No, CL-go-ENCL-IND Enugu yesterday. He’ll be back next week ‘No, he travelled to Enugu yesterday.     He’ll be back next week.’ [10:5]

In (1), speaker 10’s response includes full sentences in both Igbo and English. Each of these sentences is a single clause: that is, o. gawara Enugu eci ‘he travelled to Enugu yesterday’ is a sentence in Igbo and it is a single clause; and He’ll be back next week is a sentence in English and it is also a single clause. Within each clause there is no switching of languages, but there is switching between the clauses. This type of CS is not the object of study here because the two languages are not in contact within the same clause. Consequently, all the examples in the Igbo– English data such as that in (1) are excluded from the analyses reported in this study (totalling 136 intersentential switches). Intrasentential CS, however, involves switching from one language to the other within the boundary of a single clause, as in the following examples from the Igbo–English data: (2) ha ga-­a-no. week abu. o. na Abuja They AUX-V-stay week NUM/D PREP Abuja ‘They will stay for two weeks in Abuja.’ [32:40] (3) eze anyi. ga-­e-­mehee hospital ahu. Chief our AUX-V-open hospital DEM/D ‘Our chief will open that hospital.’ [4:10] (4) i.-­ga-a-­bi.a i.-­hu. councillor ha na opening ceremony? CL-AUX-V-come INF-see councillor PRN/D PREP opening ceremony ‘Will you come to see their councillor at (the) opening ceremony?’ [30:14] (5) ha   ku. da-­ra booth dum no.-­na polling station ahu.  … They break-IND booth Q be-­at polling station DEM/D ‘They broke all (the) booths at that polling station.’ [35:26] (6) a-­si na a-­ga-­e-me wedding Ngozi ma o.-­gbake-e CL-said C CL-AUX-V-do wedding Ngozi when CL-recover-IMP ‘They said that they will hold Ngozi’s wedding when she recovers.’ [42:40] (7) anyi. ho.pu. ta-­ra government o.jo.o. a, ma anyi. ga-e-change ya … We elect-IND government bad DEM/D but we AUX-V-change it ‘We elected this bad government, but we will change it.’ [25:15]

Introduction

5

(8) mu. na di m a-­mali.te-­la    i-save ego niile o.-­na-e-­nye anyi. Me and husband my V-start-PERF INF-save money all CL-AUX-V-give us ‘My husband and I have started to save all (the) money he gives (to) us.’ [8:31] (9) o-­me-re campaign nye anyi. manifesto ya CL-do-IND campaign give us manifesto his ‘He campaigned and gave us his manifesto.’ [4:10] (10) ha maa na this government is full of corrupt people … They know C this government is full of corrupt people ‘They know that this government is full of corrupt people.’ [6:3] (11) Kachi a-­maa-ghi. ma her job di. safe na company ha Kachi V-know-NEG C her job BE safe PREP company their ‘Kachi does not know whether her job is safe at their company.’ [9:25] (12) nwunye m kwu-­ru that na i.-­ga-e-­so anyi. a-­ga u. ka Wife my say-IND C C CL-AUX-V-follow us V-go church ‘My wife said that you will go with us to church.’ [2:30]

Accordingly, this study is concerned with uncovering what constraints are in operation when Igbo and English morphemes occur together in the same clause as in the examples in 2 to 12 above. That is, do both languages participate equally in contributing content words, grammatical elements and word order? We shall expand upon and seek to address these issues in the later chapters of this volume. Also, we shall take as our unit of analysis the bilingual clause: this will be defined as a clause containing one or more morphemes from more than one language. This is discussed further in section 5.4.

1.2  The sociolinguistic dimension Studying the sociolinguistics of CS, Gumperz (1982: 75–84) put forward a number of discourse functions that CS is seen to realize such as quotations, addressee specification, interjections, reiteration, message qualification and personalization versus objectivization. An important part of Gumperz’s approach relies on the symbolic distinction between we versus they embodied in the choice of codes. In other words, the tendency is for the home (or in-­group) language to be regarded as the ‘we’ and the out-­group language as the ‘they’ code. The ‘we’ is the in-­group, informal, personalized activities, while the ‘they’ code typically signifies out-­group, more formal relations. Thus, in (13a and b below) from Igbo–English CS, Igbo serves to mark ‘village’ talk (equivalent to Gumperz’s ‘out-­

6

Codeswitching in Igbo–English Bilingualism

group talk’); both Igbo and English (CS) mark ‘talk with family members’ (the in-­group). (13)  a. anyi. ga village anyi. na-­asu. asu. su. Igbo we go village we AUX-speak language Igbo maka na u. fo.di. ndi. bi na obodo a-­na-ghi. because that some people of live PREP village V-AUX-NEG asu. asu. su. bekee … speak language English ‘When we go to the village we speak Igbo language because some of the villagers do not speak English …’ b. … mana with my family a-­na m e-use both … but with my family CL-AUX I HAB-use both ‘… but with my family I use both (i.e. Igbo and English).’ [5:1]

Gumperz’s generalization, it turns out, is not quite as clear cut as he had earlier suggested. In fact, the above examples illustrate that both languages can function as the in-­group language. This notwithstanding, stemming from Gumperz’s work, the general view is that CS acts as a contextualization cue (Auer 1995, 1998). Codeswitching contextualizes by highlighting in a certain context, against which inferences are drawn. For instance, in (13a) and (b) above, we can infer from the speaker’s utterance a sign of accommodating to the linguistic needs of the villagers who lack proficiency in English by electing to use Igbo with those monolingual speakers and CS with his family members who are bilinguals. Nevertheless, Chan (2003, 2004) observes that CS is quintessentially a textualization cue, which frames elements in a discourse that are to be interpreted in some way differently from the preceding text. The implication is that the act of switching rather than the switched code is the most essential cue. Under this analysis, contextualization is only one of the pragmatic functions of CS. In making this argument Chan proposes two taxonomies, namely the threefold motivations of CS (i.e. social, pragmatic and processing) and three levels of pragmatic meanings that CS may convey (i.e. ideational, interpersonal and textual). However, a major shortcoming of Chan’s argument, as he correctly observes, is that the taxonomies are tentative assumptions and more research is needed to substantiate these hypotheses (see also Ihemere 2007: 87–8). On the other hand, Myers-Scotton (1983, 1993a) notes that, in many of the world’s bilingual communities, fluent bilinguals sometimes engage in CS by producing discourses which, in the same conversational turn or in consecutive turns, include morphemes from two or more of the varieties in their linguistic repertoire. In various publications, Myers-Scotton develops the theme of CS as a

Introduction

7

tool for the speaker and an index for the addressee of the negotiation of interpersonal relationships, with participants cast within a ‘rational actor’ framework, weighing costs and rewards of choices made against a backdrop of awareness for all interaction types of ‘unmarked’ versus ‘marked’ choices. Her argument is based on what she terms the Markedness Model (MM). The MM is more centred on the notion that speakers make choices because of their own goals. Of course, they cannot ignore some consideration for listeners. After all, without listeners, there is no conversation (Myers-Scotton 2006: 158). Furthermore, the MM is an attempt to establish a principled procedure that both speakers and listeners use to judge any linguistic choice that they might make or hear as more or less marked, given the interaction in which it occurs. Hence, unmarked choices are those that are more or less expected, given the ingredients of the interaction (participants, topic, setting and so on.). For instance, Myers-Scotton (1993a) refers to a rights and obligations set (RO) as part of the normative expectations for each interaction type. These expectations illustrate an unmarked way to behave. Concerning language, the unmarked choice is the linguistic reflection of any specific RO set, but only in a specific interaction type (Myers-Scotton 2006: 159). Therefore, in (13a) and (b) above, for Igbo–English bilinguals the unmarked choice to use when speaking to villagers who do not speak English is Igbo. This linguistic choice is indexical of the RO set. Accordingly, in the words of Myers-Scotton (ibid.), when a speaker makes the unmarked choice, he/she is causing no social ripples because participants expect such a choice, based on experience. However, one of the most significant features of the MM is not what it has to say about unmarked choices, but what it says about marked choices. According to the model, marked choices are those that are not predicted, given the RO set that is in effect. Hence, the question: why do speakers make marked choices? The model is based on a negotiation principle, and according to Myers-Scotton (1993a: 114) it states: Choose the form of your conversation contribution such that it indexes the set of rights and obligations which you wish to be in force between speaker and addressee for the current exchange. Myers-Scotton 1993a: 114

Under this principle (and the model as a whole), making a marked choice is a negotiation for an RO set other than the one that is unmarked for the current exchange. Put simply, the speaker making a marked choice is requesting a new situation, asking for a new RO set to be in effect. Thus, a choice that is marked in interaction X would be unmarked in interaction Y, the one that the speaker

8

Codeswitching in Igbo–English Bilingualism

wishes to be in effect. Broadly speaking, a marked choice is a negotiation about the speaker’s persona (who the speaker is) and the speaker’s relation to other participants. Thus, Myers-Scotton (2006: 160) points out that making a marked choice is a negotiation about either the solidarity or the power dimension (or both). That is, as a sociolinguistic construct, the MM presupposes that, as part of their general cognitive architecture, all speakers have a markedness evaluator. This abstract component underlies the capacity to conceptualize markedness. Other researchers have also made important contributions to the sociolinguistic study of CS. For instance, Valdés-Fallis (1976) focuses on Spanish–English CS, distinguishing types such as situational, metaphorical and contextual CS. In later work, Valdés-Fallis (1981) describes CS as an interactional strategy, on the basis of a study of direct and indirect requests in which Spanish and English are switched. Kachru (1978, 1983) discusses various social motivations for CS in India between Indian languages and English, as well as classifying types of CS on the basis of such Indian–English data. Heller (1988) discusses the strategic use of CS for stylistic, conversation management and social significance effects, drawing on Canadian French–English CS data. Appel and Muysken (1987), in discussing various language contact phenomena, propose a number of social functions of CS, namely referential, directive, expressive, phatic and metalinguistic functions. Gardner-Chloros (1991) reports on French–Alsatian CS as marked and unmarked choice in Strasbourg. TreffersDaller (1992, 1994) focuses on social factors playing a role in French–Dutch CS in Brussels. Clyne (2003), focusing on Dutch and German in contact with English in Australia, and Ihemere (2007), exploring the contact between Ikwerre and Nigerian Pidgin English in Port Harcourt, consider the role of CS in the context of a range of language contact phenomena such as convergence, language choice, attitudes and shift. Recently, a number of researchers have begun exploring the manisfestaions of bilingualism under what they term ‘translanguaging’. This term is said to have been first coined by Cen Williams (1994), and Canagarajah (2011: 401) defines translanguaging as the ability of multilingual speakers to shuttle between languages, treating the diverse languages that form their repertoire as an integrated system. Furthermore, researchers working within this framework tend to argue that translanguaging differs from the notion of codeswitching in that it refers not simply to a shift or shuttle between two languages (as observed in the above definition by Canagarajah 2011: 401), but to the speakers’ construction and use of original and complex interrelated discursive practices that cannot be easily assigned to one or another traditional definition of language,

Introduction

9

but that make up the speakers’ complete language repertoire (García and Wei 2014). One may argue that translanguaging is nothing more than a reworking of existing social-­psychological views of what bilingual codeswitching is and how bilinguals deploy their multiple languages in their routine interactions. Still these researchers (for example, Celic and Seltzer 2011: 1) insist that the notion of codeswitching assumes that the two languages of bilinguals are two separate monolingual codes that could be used without reference to each other. Instead, they argue that translanguaging posits that bilinguals have one linguistic repertoire from which they select features strategically to communicate effectively. Thus, according to Celic and Seltzer (2011: 1), translanguaging takes as its starting point the language practices of bilingual people as the norm, and not the language of monolinguals, as described by traditional usage books and grammars. Much of the research employing translanguaging in the exploration of bilingualisms seem to focus more on bilingual education and learning (for example, ‘Lesson accomplishment’: Arthur and Martin (2006); ‘Translanguaging in the bilingual classroom’: Canagarajah (2011), Creese and Blackledge (2010), García (2009), Lewis, Jones and Baker (2012), and so on.). While the contributions made by these researchers and those investigating codeswitching particularly from the sociolinguistic dimension are not in doubt, the focus of the present volume is, however, purely linguistic. Therefore, in the next section we shall present an overview of some of the grammatical studies of CS in West Africa.

1.3  Grammatical studies of codeswitching in West Africa According to Amuzu and Singler (2014: 332), there is a link between the hierarchical character of language use in West Africa and the widespread occurrence of intrasentential CS. Also, the authors observe that most studies of CS in West Africa have assumed insertion, with a large number of them following the MLF model articulated by Myers-Scotton (1993b, 2002; Myers-Scotton and Jake 1995). Another point about CS studies in the region is that a majority of them appear to involve language pairs consisting of a majority language and a colonial language. Studies of CS focusing on a minority language and a majority language (such as Kabiye–Ewe: Essizewa 2007a, b), or a minority language and a colonial language (like Kabiye–French: Essizewa 2008) are rare in the region. A further point about the hierarchical character of language use in West Africa, which the authors make, is that insertion normally obtains when the language of lower status is the base language. Thus, in the case of Ewe–English bilinguals in

10

Codeswitching in Igbo–English Bilingualism

Ghana, CS tends to obtain when Ewe is the language of discourse, not when English is (Amuzu and Springer 2014: 333). Furthermore, many of the early grammatical studies of CS in West Africa were in the form of doctoral dissertations. Starting with Forson’s (1979: 160–2) Akan–English CS study, he writes that when we say a person is using language X, what we are actually saying is that he is using the grammatical system and grammatical items of that language, and not necessarily the lexical items. Thus, in Akan–English CS, the speakers are using the Akan grammatical system and items, and therefore speaking Akan. In intrasentential switching, the Akan word order is basically followed where the word order for English and Akan differs. Goke-Pariola’s Yoruba–English study, like Forson’s on Akan–English, investigates CS in a way that anticipates Myers-Scotton’s MLF model. In his Yoruba–English study, Goke-Pariola even uses the term ‘Matrix Language’, a term synonymous with Myers-Scotton’s CS model. Central to Goke-Pariola’s study is uncovering whether one language of the pair is consistently the ML. In his findings, he concludes that Yoruba, the speakers’ mother tongue, is almost invariably the ML (Goke-Pariola 1983: 44). A similar conclusion is also reached in Amuda’s (1986) dissertation on Yoruba–English CS. In his study, Amuda suggests a series of structural constraints, consistently framing them in terms of the languages involved; for example, negators in switches can only be in Yoruba with verbs in English but not vice versa (Amuda 1986: 342). In this and the other constraints that he proposes, Amuda is anticipating the MLF model. Employing the terminology of the Myers-Scotton model, the role that Amuda assigns to Yoruba in the constraints that he proposes is similar to that of ML, while the role of English is that of the EL. As Amuzu and Singler (2014: 336) note, once Myers-Scotton (1993b) had set out the basic principles of the MLF model, studies of CS in the region began to employ her model. They add that authors in the region find it insightful to posit that mixed utterances, specifically those involving intrasentential CS, draw their grammar from one language. In cases where the two languages in question are a colonial language and a local language (whether majority or minority), the local language is the source of grammar; that is, of word order and functional morphology. Included among the authors who have adopted the MLF model in their CS study are Amuzu (1998, 2005, 2010, 2014), Benítez-Torres (2009), Dreyfus and Juillard (2001), Essizewa (2007b), Haust (1995) and Obiamalu (2013b). Amuzu’s (2005) work explores the distinction that Myers-Scotton makes between ‘classic’ and ‘composite’ CS, and in his (2014: 384–407) study on mixed possessive constructions in Ewe–English CS, he argues for the analysis of the bilingual constructions as composite CS. Composite CS is switching in

Introduction

11

which the languages involved share responsibility for framing the constituents. In its emphasis on composite CS, as opposed to classic CS, Amuzu’s study departs from the usual characterization of CS in Africa as classic (in the sense of Myers-Scotton 1993b, 2002). Owens (2005) presents data from Nigerian Arabs in the city of Maiduguri. In his analysis of the data, he endorses the MLF model but argues that it should be adjusted so as to account for sociolinguistically complex language data. In the corpus that Owens examines, individual speakers use Nigerian Arabic (NA) and Hausa (H) routinely as MLs and English and Standard Arabic as ELs. His primary concern involves ML switches within the discourse between NA and H and vice versa. While these switches occur in both directions, Owens reports that the site of the switch relative to sentence boundary differs according to whether it is a switch from NA to H or vice versa. He proposes that the difference is not random and suggests that the MLF model can and should be refined to account for such differences (see Amuzu and Singler 2014: 336). Lamidi’s (2004) Yoruba–English study is based on Chomsky’s Theory of Principles and Parameters (Chomsky and Lasnik 1995) to formulate constraints on CS. Other CS studies have also emerged in the region involving alternation between two languages as proposed by Shana Poplack and her associates. The EC and the FMC (these CS constraints are reviewed in sections 2.1 and 2.2 of this volume) that Poplack (1980) posits as fundamental restrictions on possible codeswitches were challenged early on by Nartey (1982) and Madaki (1983), with Nartey drawing on data from Dangme–English CS and Madaki on that from Hausa–English. Nartey produces counterexamples to both constraints, arguing that such counterexamples are by no means rare. Although Madaki does not present direct counterexamples to the EC, he asserts that a Hausa–English utterance that sounds natural and conforms to the EC for switches becomes at best marginally acceptable when the two languages are reversed, that is, when the Hausa portions are switched to English and vice versa. He says that CS is not symmetrical (Madaki 1983: 10–11, cited in Amuzu and Springer 2014: 337). Still, Poplack (2012), Poplack and Meechan (1995, 1998) and Meechan and Poplack (1995) argue that lone Lb incorporations into otherwise La discourse do not constitute instances of CS but are instead borrowings (see Poplack and Meechan 1995: 200). The CS versus borrowing debate is discussed further in section 5.5 in connection with the Igbo–English data. However, in their study, Poplack and Meechan draw on quantitative evidence to support their argument, crucially from French adjectives and nouns in Fongbe–French and Wolof–French data. Specifically, they show that the distribution of lone incorporations is roughly

12

Codeswitching in Igbo–English Bilingualism

comparable to the distribution of analogous monolingual items. Eze (1998) likewise uses quantitative evidence in examining English nouns and verbs in Igbo–English discourse, thereby lending credence to Poplack and Meechan’s conclusion. Nevertheless, Owens (2002), using the same approach to examine the distribution of English possessed and possessor nouns in English–NA discourse in Maiduguri, finds a qualitative difference between the treatment of such nouns and their monolingual NA counterparts. Thus, Amuzu and Singler (2014) correctly observe that in Owens’ (2002) study, when differences in the data between the behaviour of lone importations and their monolingual counterparts undercut the assertion of their comparability, then the imports cannot be seen simply as borrowings in the way that Poplack intends (Amuzu and Singler 2014: 337). Also, examples from the present study appear to falsify both the EC and the FMC as we demonstrate in Chapters 6 to 8 of this volume. Crucially, the preceding overview of the grammatical studies of CS in West Africa reveals: (1) the huge popularity and successful application of the MLF model in CS studies in the region; and (2) the glaring gap in the available literature concerning a comprehensive study of Igbo–English CS. Therefore, the present study is unique in that it sets out to fill this important gap by presenting a comprehensive account of the grammar of Igbo–English CS from the perspective of the MLF model. To this end, we begin in the next section by giving a brief account of the sociolinguistic context that gave rise to Igbo–English bilingualism.

1.4  Context of the research As Dalby (1977) explains, Nigeria falls squarely within the Fragmentation Belt, a zone of extreme linguistic complexity stretching from Senegal to Ethiopia. Of the five language phyla recognized by orthodox or mainstream scholarship in African language classification, three are widely represented in Nigeria: Niger– Congo A and B, Nilo-Sahara, and Afro-Asiatic. Only the rather small group of Khoisan (or click) languages are not spoken in Nigeria (Dalby 1977; Webb and Sure 2000). Additionally, it is estimated that there are over 500 indigenous languages spoken in Nigeria today (Ihemere 2007; Lewis 2009), of which English is the official language. Other than English, Hausa spoken in the north, Yoruba in the south-­west and Igbo in the south-­east are recognized as regional major languages. These Nigerian languages are classified thus on the basis of relative sizes of native speaker population. Languages such as Arabic and Nigerian Pidgin English (NPE), in addition to the indigenous languages, are also widely

Introduction

13

used in Nigeria to fulfil various communicative purposes. However, English plays a very significant role within the polity. As the language of the former British colonial administrators, English sailed unchallenged into the exalted status of Nigeria’s official language. It is the primary medium of formal communication in education, government, international politics, the judiciary, commerce and the mass media, and the language of culture among Nigeria’s educated elite. Igboanusi (2002) adds that while it is true that, in most bilingual situations, the two languages in contact are complementary in the sense that they are equally mixed to facilitate communication based on any one of the languages, it is also true that in the Nigerian context language mixing is unidirectional in that it is only English that always interferes in mother tongue-­ oriented speech. This is as a result of the pride of place given to the English language in the Nigerian education system. In Nigeria, English is considered more prestigious than the autochthonous languages and has constitutional recognition. The constitution of the Federal Republic of Nigeria (2004) states the various stages at which indigenous languages and English language would be used for instruction in the Nigerian educational system. For the early childhood/pre-­primary education, the policy stipulates that ‘government shall ensure that the medium of instruction is principally the mother tongue or the language of the immediate community’ (2004: 12). At the primary school level in basic education, the same document goes on to state that the medium of instruction shall be the language of the environment for the first three years. During this period, English shall be taught as a subject. From the fourth year, English shall progressively be used as a medium of instruction and the language of immediate environment and French shall be taught as a school subject (ibid.: 16). In fact, English is the medium of instruction in all institutions of higher education in Nigeria. The implication is that every Nigerian educated in Nigerian schools is bilingual in at least their mother tongue and English. Knowledge of English is one of the prerequisites for any Nigerian to rise above or live in any wider context than his or her local setting. The indigenous languages, including the Igbo language, are not accorded the same status. However, Igbo and the other indigenous languages remain the main language of expression used by the monolingual speakers.

1.4.1  Igbo and its speakers In the case of Igbo, its native speakers are found to reside predominantly in five south-­eastern states of Nigeria, namely Abia, Anambra, Ebonyi, Enugu, Imo and

14

Codeswitching in Igbo–English Bilingualism

significant parts of Delta and Rivers state. There are also minority populations of Igbo speakers in parts of Akwa Ibom, Benue, Cross River, Edo and Kogi states. According to Echeruo (1998: xv), in general, there are two major dialect zones in Igbo: Owere and O. ni.ca, although quite significant variations occur within each zone. The zones are defined by a general combination of syntactical, lexical and phonological features rather than exclusively by any one of those elements. Within the O. ni.ca zone, the following sub-­regions can be identified: West Igbo, O. so.mari., Nsu. ka-Udi and Izzi-Abakaliki. The Owere zone incorporates such distinct sub-­regions as Mbaise, Afikpo-Eda, O. haafia-Bende-O. hu. hu. , IkwereAho.ada and Ngwa-Azu. mi.ni.. The speakers in this study speak the dialects of the Owere zone. It is important to emphasize that although, as with all language communities, mutual intelligibility is impaired or restricted across these dialect (and sub-­regional) zones, the basic elements of the language remain consistent throughout. With respect to socio-­economics, some of the largest commercial centres in West Africa are located in the Igbo areas of Nigeria. This has given rise to intense contact between the native speakers of Igbo and people from other ethnic groups in Nigeria and countries of the world. Additionally, most Igbo city dwellers are employed as professionals, civil servants, business owners and traders, and a good number are students at various levels of education in the local institutions of learning. All of these present good opportunities for intense language contact with people of other linguistic backgrounds. Hence, it is not surprising that CS is pervasive among the Igbo people.

1.4.2  Igbo–English bilingualism Similar to the cases in some bilingual communities around the world, Igbo attitudes towards CS have not always been positive. Some have in the past referred to Igbo–English CS derogatorily as Engligbo. Nwafor (1971: 44) describes Engligbo as ‘a new medium of communication, which is a hybrid of the English and Igbo languages’. In fact, in the early 1970s there were campaigns launched against Igbo–English CS. These campaigns appeared in newspaper articles, television and newspaper commentaries, books and journal articles. For instance, Ogbonna (1985, cited in Ahukanna 1990: 179–80) reports that the view by some in the community on the mixture of Igbo and English is that such behaviour amounts to what he describes as ‘linguistic sabotage’. Obiamalu and Mbagwu (2008: 34–7) observe that, unlike the Hausa and Yoruba, the educated Igbo man does not discuss with a fellow Igbo man in Igbo language without

Introduction

15

adding English words. In their social-­psychological study of Igbo–English CS, they report that the rather intense negative attitudes of the past are abating and that Igbo–English CS is deployed mostly consciously by educated Igbos to showcase their mastery of a prestigious language, English. Having made these points, it is important to reiterate that the present study is not concerned with the motivations for CS by Igbo–English bilinguals; rather, the aim of this study as we elucidate below is to characterize observable structural patterns of Igbo–English intrasentential CS from the perspective of the MLF model.

1.5  Aims and organization of the book The chief aim of this study is to offer a linguistic characterization of the ways in which educated Igbo–English bilinguals resident in Port Harcourt combine their two languages within the same clause. In doing this, we shall assess the universal applicability of the highly influential MLF model (Myers-Scotton 1993b, 2002, 2009) with Igbo–English data. Since Igbo and English are typologically distinct languages (see the discussion in Chapter 4 of this volume), it can be supposed that a number of novel phenomena will be uncovered in the data. Thus, this study aims to contribute to our understanding of the universal aspects of intrasentential CS and a more comprehensive knowledge of the precise nature of what happens to the grammatical structures when the two languages (Igbo and English) are in contact in the same clause. It goes without saying that many volumes and journal papers5 have appeared dealing with CS research involving different language pairs in many different locales; however, not many have focused on intrasentential CS involving Igbo and English. In fact, except for Eze’s (1997) doctoral thesis, ‘A variationist perspective on codeswitching and borrowing in Igbo–English bilingual discourse’, the current effort to my mind is the only comprehensive study of Igbo–English intrasentential CS from the perspective of the MLF model. Therefore, this study will fill an important gap and contribute directly to the body of research on the structural aspects of CS. It will do this by demonstrating through exemplifications and quantitative analysis that make use of the typological contrasts between Igbo and English what happens to the grammatical structures when the two languages are in contact in the same clause. The rationale for selecting the MLF model as the theoretical framework is predicated on the fact that it is one of the most influential CS frameworks in the literature. It has also been applied widely and successfully to the analysis of

16

Codeswitching in Igbo–English Bilingualism

intrasentential CS involving varied language pairs. Its reception, however, has not been without criticism (see for instance Backus and Boeschoten 1996; MacSwan 2000, 2005a; Sankoff and Vanniarajan 1990). This is not unexpected for a model that claims to be universally applicable. In fact, as we shall show in the next chapter, to date no linguistic model of CS has managed to escape criticisms based on counterexamples, as different CS researchers continue to adopt various frameworks in the analysis of intrasentential CS. Therefore, the present study contributes to the ongoing debate over the veracity of universal CS constraints by testing the predictions of the vastly popular MLF model with Igbo–English data. Given that the present focus in this book is on grammatical aspects of CS, other issues such as questions relating to the social-­pragmatic motivations for Igbo–English CS or the function of CS in the community will not be addressed. Also, the investigation is limited to CS between Igbo and Standard English. For this reason, NPE (Nigerian Pidgin English) sequences such as sentences, phrases and single words found in either Igbo or English conversations are not taken into consideration here (see Chapter 5 for a detailed account of the research methodology adopted in the study). The rest of the book is organized as follows. Chapter 2 discusses some of the major codeswitching models that consider most or all singly occurring EL forms as borrowings on the basis of morphosyntactic and phonological integration, such as: (1) the Free Morpheme Constraint (FMC) and the Equivalence Constraint (EC) of Poplack (1980); (2) the Government Constraint (GC) of Di Sciullo, Muysken and Singh (1986); (3) the Functional Head Constraint (FHC) of Belazi, Rubin and Toribio (1994) and (4) the Minimalist Approach to intrasentential codeswitching (MacSwan 1999, 2000, 2005a, 2005b and 2009). Chapter  3 presents a detailed discussion of the theoretical framework that underpins the analysis reported in this study, the MLF model (Myers-Scotton 1993b, 2002) and its three associated principles (the Matrix Language Principle, the Asymmetry Principle and the Uniform Structure Principle), which are tested with Igbo–English data. Chapter  4 provides a brief contrastive overview of aspects of Igbo and English grammars. This is done with a view to outlining the typological contrasts between Igbo and English. Importantly, this provides the backdrop for the analysis of Igbo–English codeswitching in the latter part of the study. Chapter  5 discusses the research methodology adopted in the present study. The chapter introduces the bilingual speakers, the sampling and data collection procedures as well as measures taken to ameliorate the observer’s paradox. It also gives information about the unit of analysis, how the Igbo– English data is classified and the criteria deployed in determining whether the

Introduction

17

bilingual data are genuine codeswitching forms rather than some kind of borrowings. Chapter 6 is the first analysis chapter focusing on singly occurring EL nouns and adjectives, which make up 53.25 per cent (1,416/2,659) of the corpus. This is followed in Chapter 7 with the analysis of singly occurring EL verbs embedded in otherwise Igbo utterances. The EL single verbs constitute 20.6 per cent (547/2,659) of the corpus. In both chapters, we shall provide a general overview of the Igbo–English codeswitching data, illustrating in more detail how the various principles of the MLF model are applied to the clauses containing the singly occurring EL nouns, adjectives and verbs. The analyses will show that the principles of the MLF model do appear to be reflected in Igbo–English, and that our data does indeed appear to represent a case of classic codeswitching. In particular, the analysis of vowel harmony between English verbs and Igbo bound affixes and the inspection of spectrograms of selected EL verbs with Igbo inflectional morphology will show: (1) that contrary to the predictions of some of the codeswitching models reviewed in Chapter 2 two phonologies can indeed co-­exist in codeswitching; and (2) that the EL single words in Igbo–English are pronounced with a phonology not wholly like that of native Igbo words or established borrowings. Nevertheless, we shall also highlight that there are problems of definition and some problematic examples, which will be discussed in the chapters. Chapter  8 focuses on the analysis of EL islands; that is, the switching of multi-­word sequences such as phrases and clauses, which are assumed by most researchers to be prototypical codeswitching forms. However, the analysis of the EL islands will reveal that in the main they too pattern just like the single word switches. Broadly, we shall demonstrate in this chapter that the same principles of asymmetry can account for both single word switches as well as EL islands. Chapter 9 presents the concluding remarks and the implications of our findings as they relate to the MLF model and the codeswitching frameworks reviewed in Chapter  2. Concerning the former, we shall compare our findings to other classic codeswitching cases and reflect on the problematic examples and their solutions. In connection with the codeswitching models reviewed in Chapter 2, we shall argue that there are strong asymmetries between any two languages participating in codeswitching cross-­ linguistically. It is this division of labour, as we demonstrate in Chapters 6 to 8 of this book, which regulates the roles played by the two languages in contact in intrasentential codeswitching.

2

Studying the Grammar of Codeswitching 2.1  Introduction It goes without saying that the study of CS attracts a great deal of interest and controversy among researchers. According to Matras (2009: 101), some authors use ‘codemixing’ to refer to language mixing within the phrase or utterance, reserving ‘CS’ for the alternation of languages in-­between utterances or phrases. Others employ ‘codemixing’ to denote the structures that are the product of language mixing and do not occur in the speech of monolinguals. Yet another use of ‘codemixing’ is as a cover term for various types of language-­mixing phenomena. In this study, however, we shall use the term CS to refer to the mixing of elements from more than one language within the same clause. Nearly every CS researcher within the broad spectrum of linguistic studies of CS would agree that CS has a grammar; however, differences arise in how each researcher characterizes the nature of this grammar. Some take the view that CS involves two monolingual grammars with constraints that determine points where switches are possible. Others assert that the grammatical organization of CS can be accounted for in terms of the principles of current syntactic theories, even though these theories were initially formulated to explain monolingual data. Yet another group (including this author) analyse as CS structures the likes of which others reject as borrowings. That is, this author and many others1 view CS as a monolingual grammar where the base language supplies the frame into which lexical items, phrases and clauses from both languages could be inserted. In the sections that follow immediately below, we shall review some of the representative models deployed in the study of intrasentential CS, starting with the EC and the FMC in sections 2.2 and 2.3; the GC and the FHC are discussed in sections 2.4 and 2.5; and the Minimalist Approach (MA) to intrasentential CS is discussed in section  2.6. Section  2.7 is the conclusion to the chapter. The reviews will be followed by a discussion of the MLF model in Chapter  3; a

Codeswitching in Igbo–English Bilingualism

20

proposal based on recognition of a basic asymmetry between the roles played by any two languages participating in CS cross-­linguistically.

2.2  The Equivalence Constraint Poplack, Sankoff and their associates2 were among the first to attempt to formulate general syntactic constraints, where they propose that Spanish– English CS can be generated by a model of grammar that is governed by two constraints, namely the EC and the FMC. The latter constraint is discussed below in section 2.3. According to the EC, CS will tend to occur at points where the juxtaposition of elements from the two languages does not violate the syntactic rule of either language (Poplack 1980: 586). In other words, CS will seem to take place only at points where the surface structures of the two languages map onto each other. This implies that a language switch can occur only at boundaries common to both languages and that switching cannot take place between any two-­sentence (or clausal) elements unless they are normally ordered in the same way (Romaine 2004: 126–7). It has to be said that this constraint has proved highly successful in the study of Spanish–English CS, but not so much when applied to highly inflectional or agglutinative languages and language pairs with sufficiently different word order, such as that between the internal linear order of Welsh and English noun phrases (NPs). In Welsh, the usual order is head followed by modifier rather than modifier followed by head as in English. Deuchar (2006: 1993) observes that the same applies to most adjectival modifiers, as is illustrated by the name of the charity Y Groes Goch, ‘the Red Cross’, where the head noun croes is followed by the modifier. A similar configuration is witnessed in (14) below from Welsh–English CS. (14) oedd gynnon ni ystafell yn Plas yn Dref, ystafell brilliant be.3S.IMP with-­us PRON.3S room in Plas yn Dref room brilliant ‘We had a room in Plas yn Dref, a brilliant room.’ (Welsh–English, Deuchar 2006: 1994)

Example (14) violates Poplack’s EC because switching has occurred between a Welsh noun ystafell and an English adjective brilliant, where the two languages have different word order. Also, the example in (15) from Croatian–English CS seems to violate the EC. (15) Ne No,

on radi taj posao I he does that job and

ja I

njega supervise him supervise

Studying the Grammar of Codeswitching

21

‘No, he does the job and I supervise him.’ (Croatian–English, Hlavac (2000: 353), cited in Myers-Scotton 2006: 256)

In English the usual word order is subject–verb–object (SVO), but in (15) the English verb supervise occurs in final position. According to Myers-Scotton (2006: 255), in Croatian, the verb typically comes last in a clause. Therefore, the switching of the English verb in (15) occurs where the grammars of both languages do not map onto each other in violation of the EC.

2.3  The Free Morpheme Constraint The FMC predicts that a switch may not take place between a bound morpheme and a lexical form unless the lexical form has been phonologically integrated into the language of the bound morpheme (Sankoff and Poplack 1981: 5). As stated, this constraint seems to suggest that it is universally applicable to all languages. However, the example in (16) from Ewe–English CS seems to violate this constraint. (16) Nyɔnu-­wó, all the time a, wó-­a-nɔ wo respect-m … Woman-PL, all the time TP 3PL-POT-be 2SG respect-PROG ‘Women, they are supposed to respect you all the time …’ (Ewe–English, Amuzu 2010: 155)

Notice immediately that the English verb respect in (16) takes the Ewe progressive suffix –m (a bound morpheme) in accordance with Ewe morphosyntax. In other words, the English verb receives the same inflection as its Ewe counterpart bu-­m ‘respect-PROG’ (see Amuzu 2010: 156). It is also important to underline that the English verb in (16) is a phonologically unintegrated lexical form. A further example can be seen in (17) from Moroccan Arabic–French CS. (17) ta jbqa j- confronter ces idées he keeps IMPERFECT-oppose these ideas ‘He keeps opposing these ideas.’ (Moroccan Arabic–French, Bentahila and Davies 1983: 315)

In (17), j-­confronter ‘opposing’ consists of a verbal inflection from Moroccan Arabic and a French infinitive verb confronter ‘to oppose’ in violation of the FMC. The main weakness in the two constraints (EC and FMC) seems to stem from the basic premise underlying their formulation. For instance, Poplack postulates that languages involved in CS are only activated alternately during

22

Codeswitching in Igbo–English Bilingualism

the production of mixed constituents rather than simultaneously as claimed in the MLF model of Myers-Scotton (1993b, 2002). The MLF model is discussed in the next chapter. Amuzu (2010), commenting on Meechan and Poplack’s (1995) analysis of the mixed do construction (this is the do ‘to be’ construction, not to be confused with the English verb ‘do’) in Fongbe–French CS also questions the fundamental premise underlying the formulation of the two constraints: that is, the view that language activation during the framing of CS constituents happens alternately (i.e. one at a time). In monolingual Fongbe do usually occurs with adverbials and locatives. However, Meechan and Poplack (1995: 186) report that the greatest proportion of do occurs in the context of French adjectival phrases. In their study, Meechan and Poplack (1995) see the French adjectivals as CS constituents because they are multi-­morphemic and analyse them as constituent insertions into the complement of do slot, as in (18) below. (18) Donc ɔ nyɛ mɔ dɔ que langue ɔ e do importante So TOP I see tell that language DEF she be important ‘So, me I see that language is important.’ (Fongbe–French, Meechan and Poplack 1995: 187.)

Given their hypothesis that language activation during the framing of CS constituents happens one at a time (see Poplack and Meechan 1995: 224), Amuzu (2010: 112) correctly observes that one can imagine (1) that during the time when Fongbe projects a CS slot for the insertion of a French adjective, French would have been deactivated, and (2) that French is only activated when its adjective and inflections are patterned out into the adjectival constituent that is inserted into the CS slot. We agree with Amuzu (2010) that this kind of analysis raises two fundamental questions:

1. How is the complement of do slot chosen as the suitable Fongbe morphosyntactic environment for the insertion of a French adjectival constituent if there is no online access to French-­origin information about the categorial status and semantic properties of the adjectival? One would assume that having direct access to French-­origin abstract lexical structure information about a given French adjective is necessary during the time when Fongbe grammar is deployed to select ascriptive do instead of nyi,3 for instance, as the copula that the adjective requires.

2. What kinds of Fongbe-­only processes guide the choice of nyi to introduce French nominal complements that are co-­referential with their subject

Studying the Grammar of Codeswitching

23

NP? Can information indicating that the French predicates are co-­ referential nominals be processed while French is not simultaneously activated with Fongbe? According to Amuzu (2010: 112), it is not enough to simply state that the French adjectivals occur as ‘constituent insertions’ after do without explaining why this is so, i.e. why the insertions are not made after nyi instead. The reasons required for the kind of insights we need in order to confront the questions above reside in principles that underpin the copula systems of the languages participating in CS; insights which lie deeper than mere surface structure CS configurations. Moreover, much earlier Di Sciullo et  al. (1986) observed that the EC is undesirable from a theoretical point of view as well as empirically inadequate. They explain that for it to be applicable to CS in natural languages there needs to be categorial equivalence. That is, if Language1 had the categories determiner and conjunction, for example, Language2 must have them also, otherwise it will be impossible to determine whether switching is possible at the point between conjunction and determiner. Nonetheless, they acknowledge that while there probably are major categories shared by all languages, there are a considerable number of categories which only occur in specific languages. And even then it is not evident that the categories in different languages will precisely correspond (1986: 3). Thus, Di Sciullo et al. suggest that their proposal (the GC is discussed in section 2.4 below) eliminates this identified problem by positing only points where switching could occur, and this with respect to a single linguistic system.

2.4  The Government Constraint Motivated by the theoretical and empirical inadequacies inherent in the EC proposal, Di Sciullo et  al. (1986) argued in their paper that the principle of government, the syntagmatic coherence principle of traditional and generative grammar provides a coherent and quite general account of allowable switching sites. Their focus, they say, is not on switching sites, but on relations between elements: when a government relation holds between elements, there can be no switching; when that relation is absent, switching is possible. No specific constraint they say needs to be stated to account for CS restrictions. These they claim fall out from general considerations of lexical integrity, constrained by the government condition, which hold for all uses of natural languages, not just for CS (1986: 4). Put more succinctly, the GC says that:

24

Codeswitching in Igbo–English Bilingualism

1. If Lq carrier has index q, then Yqmax. 2. In a maximal projection Ymax, the Lq carrier is the lexical element which asymmetrically c-­commands the other lexical elements or terminal phrase nodes dominated by Ymax. On this account, the GC states that the Lq carrier of a governed category must have the same Lq index as its governor. Thus, what the GC proposes is that switching between the governor and the governed is disallowed. Following Chomsky (1981), they posit that switching is disallowed between a complementizer and its governing head; a preposition and a following noun; a determiner and the noun; and a noun and an adjective which modifies it, and so on. Di Sciullo et  al. (1986) explain that the GC subsumes most of the cases predicted by the EC of Sankoff and Poplack (1981) and by the particularistic constraints in the earlier literature, and provides a principled explanation for the boundary-­strength mystery identified as one of the flaws in the EC. Furthermore, they stress that it is not their intention to claim that the principle of government is the only condition, capable in itself of handling all possible restrictions existing in different CS situations, but rather that it is the only universally applicable one. Additionally, they observe that in specific cases, there most certainly will be additional language-­particular constraints. An example may be that NP-internal agreement rules may block cases of switching within the NP. Of particular interest here is their claim that the GC is the only universally applicable one. Given this claim of universal applicability of the GC, all it will take to violate it is one counterexample. In fact, Nortier (1990) tested this and other constraints in her Dutch–Moroccan Arabic data and found that not just one, but most of the switches in her data violate the GC. Also, we saw earlier in example (14) above that switching is not blocked between the Welsh noun ystafell and the English adjective brilliant. Given the many counterexamples that have been put forward over the years, some researchers have sought to modify the GC to make it more universally applicable to different language pairs. For instance, in her study of Irish–English CS, Stenson (1990: 193) suggests an adaptation of the GC, whereby (1) phrase structure congruence is required for switching at the level of deep structure; and (2) the GC is one of the principles that act as a filter for further switching at the level of surface structure. In connection with her first suggested adaptation, she proposes that switching at the level of deep structure is due to lexical items from either language filling structures formed by phrase structure rules (PSRs) shared

Studying the Grammar of Codeswitching

25

by the two languages. In Stenson’s data, this will be a cliticized complementizer in Irish which may be proposed to disallow Irish–English switching between complementizer and complement (1990: 188). With regard to her second proposed adaptation, she allows for the possibility that only certain types of government affect CS by imposing the same-­language-index restriction, for example applying to verb–argument relations but not phrase-­internally. Another researcher who has sought to extend the GC proposal is Halmari (1997). She deployed the notion of government in her study of American Finnish–English CS. In her study, she notes the importance of case assignment and agreement relations in accounting for the American Finnish–English CS data, both of these notions being closely related to that of government (Halmari 1997: 99). Stemming from her investigation, she suggested that the GC of Di Sciullo et al. (1986) can be restated to show that case and agreement morphology can act as language carriers (Halmari 1997: 103). Ostensibly, she sought in her work to demonstrate that all American Finnish–English CS which adheres to the syntactic structure of American Finnish may be explained in terms of such a restatement of the GC. In spite of the spirited attempts aimed at improving the GC, Muysken’s (1991) earlier observation remains apt: the class of governors as well as the domain of government was too large, leading to the numerous counterexamples to the GC reported in the literature. We shall also show in Chapters 6 to 8 of this volume that the GC is not supported by evidence from Igbo–English CS.

2.5  The Functional Head Constraint Belazi et  al. (1994), working on Tunisian Arabic–French CS, began by acknowledging that there are certain syntactic regularities underlying the switching process in intrasentential CS. The question they correctly put is not whether CS obeys any structural constraints, but what the best way to characterize them is, and whether they can be made to follow from independently motivated, more general principles. Motivated by these questions, they reviewed and rejected the earlier proposed CS constraints. Following from this, Belazi et al. (1994) put forward the FHC after Abney’s (1987) proposal that there exists a unique relationship between a functional head and its complement. In contrast to lexical heads (nouns and verbs), functional heads (for example, complementizers and determiners) f-­select their complements; therefore, Belazi et al. argue that if f-­selection is a member of a set of feature-­checking processes and if we assume that one of the features being checked is language, then the

26

Codeswitching in Igbo–English Bilingualism

functional head requires the language feature of its complement to match its own. Thus, the FHC is stated as: The language feature of the complement f-­selected by a functional head, like all other relevant features, must match the corresponding feature of that functional head. Belazi et al. 1994: 228

In other words, the FHC is very similar to the GC (Di Sciullo et al. 1986) in that both seem to predict that a functional head such as a complementizer must come from the same language as its complement, thus disallowing switching at other junctures between a complementizer and its Inflectional Phrase (IP) complement. This prediction has been challenged by data from many different language pairs. For instance, Mahootian and Santorini (1996) offer numerous examples from various language pairs that violate the FHC. (19) I seen everything ’cause no cogi na C IP ‘I saw everything because I didn’t take anything.’ (Spanish–English, Sankoff and Poplack 1981: 6)

In (19), contrary to the prediction of the FHC the complementizer (C), a functional head, and its IP complement are not in the same language. In (20) to (22) below, we also observe further violations of the FHC. (20) No parce que hanno donné des cours I VP no because have given of the lectures ‘No, because they gave lectures.’ (Italian–French, Di Sciullo et al. 1986: 15) (21) E wo green dress ko He/She PAST-wear NP D ‘She wore a green dress.’ (Adaŋme–English, Nartey 1982: 187) (22) I’ll take some naemaek Q NP ‘I’ll take some salt.’ (English–Farsi, Mahootian and Santorini 1996: 466)

The above examples seem to falsify the FHC because switching takes place between inflection (I) and verb phrase (VP) in (20); between determiner (D) and noun phrase (NP) in (21); and between quantifier (Q) and NP in (22). Further examples violating this constraint from the Igbo–English data are shown in subsequent chapters of this volume. From the above examples and those from Igbo–English reported in Chapters 6 to 8, it is apparent that the FHC is not a

Studying the Grammar of Codeswitching

27

universally applicable constraint. In view of these shortcomings, MacSwan (1999, 2000, 2005a, 2005b, 2009, 2010) argues for the need to derive descriptive CS generalizations from general principles of the grammar that are independent of CS. His proposal derives from the Minimalist Program (Chomsky 1995, 2000, 2001) framework. MacSwan’s CS proposal is discussed in the section immediately below.

2.6  The Minimalist Approach The central position under this approach is the idea that nothing constrains CS apart from the requirements of the mixed grammars (MacSwan 1997: 175; 1999: 146; 2005c). MacSwan (1999, 2000, 2005a, 2005b, 2009) adopts such an approach to CS, one that assumes the minimal CS-specific apparatus. This assumption (MacSwan 1999: 146 [2000: 43]) entails, firstly, that no principle of grammar be explicitly formulated for CS, as in Poplack’s (1980) approach that suggests that CS is constrained by a sort of ‘third grammar’ (MacSwan 2000: 38); and, secondly, that the identities of particular languages are ignored for the purposes of linguistic theory. Instead, the language-­particular requirements of CS are assumed to be represented in morphology (i.e. in the lexicon). This, according to MacSwan, immediately overcomes one of the objections that Belazi et al.’s (1994) proposal faces, namely that the language feature that they propose is not independently motivated for other linguistic phenomena (see further Van Gass 2002: 98). Thus, in an MA explanation of the acceptability of CS sentences, there is an appeal to morphologically sensitive mechanisms motivated to account for grammaticality in monolingual sentences (MacSwan 1999: 147). Furthermore, the MA to CS assumes that the computational system is invariant across all languages and that parameters are part of the lexicon, which the computational system uses to build larger structures. Each lexical item introduces grammatical features, or F-features, into the derivation, which must be checked. MacSwan posits that the language faculty does not have to pay attention to the socio-­political identity of words (for example, our associations of house with ‘English’, and of u. lo. ‘house’ with ‘Igbo’). The language faculty he says is only sensitive to the fact that these lexical items have features which enter into the derivation and that these features must be checked. When features mismatch or when uninterpretable features cannot be checked, the derivation crashes, whether the set of lexical items is associated with one specific language or two (or more). Thus, MacSwan’s approach would automatically exclude all

28

Codeswitching in Igbo–English Bilingualism

single-­word insertions bearing other language inflections. Consequently, the acceptability of a linguistic expression depends on whether its features match, no matter whether it is a monolingual or a CS expression. Hence, if all syntactic variation is associated with the lexicon, as in the MA, then CS can be seen as the result of mixing two lexicons in the course of a derivation (MacSwan 2000: 45). There is no need for specific grammatical constraints to mediate the contradictory requirements of both languages. Instead an MA sees the grammar used for CS as a combination of both lexicons and the invariant computational system. Moreover, MacSwan (ibid.) notes that Chomsky (1995) and others have pointed out that the mapping to phonetic form (PF) is completely different from the syntactic component of the grammar. Syntactic operations apply at any time and in any order, while PF component operations must apply in a particular order. Therefore, if the two PF components are mixed in the same way as the two lexicons for mixing in the syntactic component, the ordering of rules will not be preserved. Rather than invoke a constraint which would allow the two systems to interface, MacSwan proposes the hypothesis that CS is impossible at PF, expressed below as the PF Disjunction Theorem (MacSwan 2000: 45):

1. The PF component consists of rules/constraints which must be (partially) ordered/ranked with respect to each other, and these orders/rankings vary cross-­linguistically. 2. CS entails the union of at least two (lexically encoded) grammars. 3. Ordering relations are not preserved under union. 4. Therefore, CS within a PF component is not possible. MacSwan (2000) explains that the PF Disjunction Theorem might be seen as an instantiation of full interpretation (FI), the requirement that every object have a sensorimotor interpretation to qualify as a legitimate representation, a kind of ‘interface condition’ (Chomsky 1995). Thus, the PF Disjunction Theorem is recast as the PF Interface Condition (PFIC) in MacSwan’s 2009 chapter ‘Generative approaches to codeswitching’ in the Cambridge Handbook of Linguistic Codeswitching, edited by Bullock and Toribio (2009: 309–35):

1. Phonological input is mapped to the output in one step with no intermediate representations. 2. Each set of internally ranked constraints is a constraint dominance hierarchy, and a language-­particular phonology is a set of constraint dominance hierarchies.

Studying the Grammar of Codeswitching

29

3. Bilinguals have a separately encapsulated phonological system for each language in their repertoire in order to avoid ranking paradoxes, which result from the availability of distinct constraint dominance hierarchies with conflicting priorities. 4. Every syntactic head must be phonologically parsed at Spell Out. Therefore, the boundary between heads (words) represents the minimal opportunity for CS. At the heart of MacSwan’s proposal is that since phonological systems cannot be mixed (according to him), CS at PF generates ‘unpronounceable’ elements which violate FI. Put another way, the PF Disjunction Theorem (recast as the PFIC) does not function as a grammatical constraint on CS; rather it is a theory about the relationship between the PF components of a bilingual’s linguistic system and is deduced from the nature of phonetic rules. It must be noted, here, as Legendre and Schindler (2010) also point out in their study of CS in Urban Wolof (UW),4 that central to MacSwan’s analysis is his lexicalist approach to the lexicon–syntax interface, as formulated within the Minimalist Program (Chomsky 1995). Under this approach, word formation of complex heads (e.g. a verb and its inflectional morphology) takes place in the lexicon pre-­syntactically; syntax manipulates the ‘opaque’ outcome of word formation under Checking Theory. At spell-­out, the complex head is sent to the phonological component for phonological interpretation. The lexicon is the repository of the language specification for each word entering the computation along with parameters relevant to the computation. For CS, this means that each word comes ‘pre-­labelled’ as belonging to language X or Y, which determines its formal feature strength and therefore its syntactic properties. Because the lexicon is the traditional repository of ‘arbitrariness’, adding a language specification to lexical items is not equivalent to imposing a CS-specific constraint or principle (Legendre and Schindler 2010: 51–2). Consequently, to the extent that it is successful, MacSwan’s proposal offers an elegant account of CS without appealing to CS-specific constraints, contra Sankoff and Poplack (1981), Di Sciullo et  al. (1986) and Belazi et  al. (1994). Also worthy of mention is MacSwan’s characterization of CS fully in the phonological component without recourse to syntactic constraints proper, leaving invariant the computational component of the grammar. This notwithstanding, we shall cite examples in Chapter 7 of this study from both the Igbo–English data and UW–French CS study by Legendre and Schindler (2010) that challenge the PFIC-based account.

30

Codeswitching in Igbo–English Bilingualism

Therefore, it will suffice here to point out that one of the shortcomings of MacSwan’s approach to the study of intrasentential CS is his non-­recognition of any theoretical (or useful) value in the existence of strong asymmetries in the roles played by any two languages participating in CS cross-­linguistically. Jake, Myers-Scotton and Gross (2002) observe that this limits both the depth and scope of any minimalist explanation of CS. They argued quite successfully that the MP was originally formulated to account for grammatical strings in monolingual data; noting that as applied to CS the MP rules out most mixed constituents on two grounds: (1) switching within a word is disallowed on phonological grounds; and (2) feature mismatches block switches, unless there is almost complete congruence of features in mixed constituents. However, evidence from the Igbo–English data seems to contravene both assumptions. For instance, 77 per cent of the singly occurring English nouns in our data occur in mixed determiner phrases (DPs) with either overt or covert post-­posed Igbo determiners and 85 per cent of the singly occurring English verbs in non-­serial verb constructions obligatorily receive Igbo inflectional morphology (see analyses in Chapters 6 and 7 of this volume). Thus, it is not always the case that when features mismatch or when uninterpretable features cannot be checked, the derivation crashes, as MacSwan claims under his PFIC-based account. As a result, I agree with Jake et al. (2002) that CS empirical data suggest that when congruence (see section 3.3 for a discussion of the role of congruence in CS) between the features of the participating languages is missing, there are two results: (1) the derivation does not crash; and (2) it is resolved in favour of one of the participating languages (see also Couto et al. 2015 in relation to Welsh– English CS). These propositions underpin the CS analyses reported in this study, as we demonstrate in later chapters.

2.7  Conclusion The CS models discussed in this chapter appear to be formulated to: (1) block CS if surface structures of two languages do not map onto each other; (2) block CS between a functional head and its complement; and (3) block CS word-­ internally between a root and an affix (see also Legendre and Schindler 2010: 49). However, we have shown even at this early stage in the study that the basic premises underpinning the CS models discussed here are violated by the cited CS examples. The evidence so far supplied seems to suggest: (1) that CS is very possible even when the surface features of the two languages participating in CS

Studying the Grammar of Codeswitching

31

do not match; (2) that CS between functional heads and their complements is not blocked; and (3) that CS between a root and an affix is permitted in CS. Consequently, we shall in the next chapter outline a CS framework that is based on recognition of a basic asymmetry between the languages participating in CS. Under this approach, only one of the participating languages is claimed to be the source of the morphosyntactic frame structuring what is termed ‘classic CS’. Classic CS includes elements from two language varieties in the same clause, but only one of these varieties is the source of morphosyntactic frame for the clause (Myers-Scotton 2002).

3

Theoretical Framework 3.1  Introduction As we noted in section 2.1 of the last chapter, the MLF model differs from the other CS models discussed in sections 2.2 to 2.6. The model posits that the language with the most critical grammatical contributions to the bilingual clause is called the Matrix Language (ML), and the other participating language, which largely supplies some content elements in the clause, is called the Embedded Language (EL). The ML construct is claimed to interact with several other assumptions about language and language production. Crucial to the notion of ML is the Uniform Structure Principle (USP), which is said to apply to all language. According to Myers-Scotton and Jake (2014: 3), the critical provision of the USP for CS is that it preferences uniform structure from the ML in bilingual constituents. We shall elaborate on these claims in the sections that follow immediately below. In sections  3.2 to 3.2.4, we shall outline in some detail the MLF model and its associated principles, which are tested in later chapters of this volume with Igbo– English CS data. The role of congruence in CS and how it will be deployed in this study is discussed in section 3.3. Section 3.4 is the conclusion to the chapter.

3.2  The Matrix Language Frame model The MLF model was first articulated by Myers-Scotton in her book Duelling Languages in 1993. Since then the model has undergone a number of refinements (1993b, 2002). Myers-Scotton (2006: 243) acknowledges that the MLF model benefited from the insights of earlier researchers who recognized the unequal participation of languages in CS. Specifically, she cites Joshi (1985), who was one of the first to refer to the frame-­building language as the ML and to the other participating language as the EL. She goes on to state the three premises of the MLF model:

34

Codeswitching in Igbo–English Bilingualism

1. The ML and EL do not participate equally in constituent structure. 2. Not all morpheme types are equal in the sense that not all types can come equally from the ML and the EL. 3. Both languages are always ‘on’ when a speaker engages in CS, although the ML is always more activated. According to Myers-Scotton (2002, 2006), the first two premises are supported by CS empirical data, while patterns in CS provide indirect support for the third premise. Jake et  al. (2002: 72) add that the key to understanding feature mismatches in CS is to recognize one of the asymmetries in language that is especially evident in CS: structural conflicts are resolved in favour of only one of the participating languages. The MLF model, they claim, captures this generalization in theoretical assumptions about the nature of linguistic competence and also about operations involved in language production. This view is conceptualized under what they term a USP and its corresponding two hierarchies that indicate how the model relates to linguistic competence. USP: A given constituent type in any language has a uniform abstract structure and the requirements of well-­formedness for this type must be observed whenever the constituent appears. In bilingual speech, the structures of the Matrix Language are always preferred, but some embedded structures are allowed if Matrix Language clause structure is observed. Myers-Scotton 2002: 8–9

The claim is that when this principle is applied to bilingual speech, it gives rise to the first hierarchy, which states that, in bilingual speech, the languages involved do not participate equally: one language uniformly sets the morphosyntactic frame and this frame is referred to as the ML. Furthermore, although the ML versus EL opposition is specific to bilingual speech, the ML construct, they claim, is not. According to Jake et al. (2002), it is derived from the USP and is the abstract grammatical frame structuring any constituent; it is simply vacuously transparent in monolingual data. In classic CS, however, only one of these languages is the source of the ML frame. The second of the two hierarchies of the USP is the distinction in the MLF model between the roles of content morphemes (similar to lexical elements) and system morphemes (similar to functional elements). This distinction is claimed to be evident in language in general; again, that such a division exists is part of general linguistic competence. The distinction is especially visible in CS because of the constraints the ML imposes on bilingual structures. Under the MLF model, the frame-­building system morphemes in mixed

Theoretical Framework

35

constituents (e.g. a mixed DP/NP or I/VP as part of a single clause makes such a clause a mixed constituent; and a clause containing just one other language lexical item is also a mixed constituent1) come from only one language, the source of the ML (see Jake et al. 2002; Myers-Scotton 2002; Myers-Scotton and Jake 2014). It has to be said that since the inception of the MLF model in 1993 its two central hypotheses have remained largely unchanged: the Morpheme Order Principle (MOP) and the System Morpheme Principle (SMP): MOP: In mixed constituents of at least one Embedded Language word and any number of Matrix Language morphemes, surface word (and morpheme) order will be that of the Matrix Language. Myers-Scotton 1993b [1997]: 83; 2002: 59; 2006: 244 SMP: In Matrix Language + Embedded Language constituents, all system morphemes which have grammatical relations external to their head constituents (i.e. which participate in the sentence’s thematic role grid) will come from the Matrix Language. Myers-Scotton 1993b [1997]: 83; 2002: 59; 2006: 244

These are testable hypotheses referring to the existence of asymmetry between the languages involved in CS. Both principles assert that only one language (the source of the frame) supplies both morpheme order (word order) and frame-­ building system morphemes to the frame.

3.2.1  Content versus system morphemes Under the MLF model, content morphemes are those that either assign or receive thematic roles. Thematic roles are also called theta roles. They are basically semantic roles, although they are often discussed by linguists as tied to syntax. Content morphemes are semantic in the sense that they refer to such relations within the sentence as whether a noun is the Agent or the Patient of the verb. Linguists say that verbs ‘subcategorize’ for different thematic roles. For example, the verb give subcategorizes for (meaning ‘can take’) three thematic roles, an Agent, a Patient (the element that is given) and a Beneficiary or Recipient (for example, John gave the man a pen). Because verbs most typically assign thematic roles and nouns typically receive them, verbs and nouns are prototypical content morphemes. Myers-Scotton adds that discourse markers (e.g. therefore, so, but) are considered content morphemes at the discourse level in this model. She explains that, even though they do not assign thematic roles within the

36

Codeswitching in Igbo–English Bilingualism

clause, they limit the interpretation of what comes after them in the clause and often become established borrowings (see section 5.5 for a discussion of the CS versus borrowing debate) when social conditions promote borrowing words from another language (2006: 245). On the other hand, system morphemes are prototypically affixes (bound morphemes) and certain function words that stand alone (e.g. determiners and clitics). Some lexical categories, such as adverbs and prepositions, straddle the division between content and system morphemes. She observes that degree adverbs, such as very, are system morphemes because they do not receive or assign thematic roles. Concerning prepositions, Myers-Scotton explains that the ones with more content do receive thematic roles when they head prepositional phrases (e.g. outside, inside, down, up), and are content morphemes; those that are satellites to verbs are system morphemes. They result in what are sometimes called phrasal verbs (e.g. up in look up the number); these do not assign thematic roles on their own. Other prepositions with little or no content, such as in or on in the clause I live on/in Buchanan Street, are system morphemes (ibid.: 245–6). This division of morphemes into content morphemes and system morphemes is claimed to apply cross-­linguistically and to be operationalized under the MOP and the SMP of the MLF model. This claim is exemplified in (23) from Ewe–English codeswitching: (23) Battery kúkú a a?        Wó  nu   má    zá-­m      fiá  é- dzidzi   ló Battery dead the Q  3PL  thing DEM use-PROG now 3S-long  ADDR ‘The dead battery? They’ve been using that thing for quite some time now.’ (Ewe–English, Amuzu 1998: 77, cited in Amuzu 2010: 223)

In (23), the singly occurring English noun battery precedes its Ewe modifiers, the adjective kúkú ‘dead’ and the definite article a ‘the’. According to Amuzu (2010: 223), this morpheme order reflects what is found in the Ewe NP. He adds that the morpheme order cannot be altered to, say, the [DET ADJ N] order in the English NP. That is, the sequences *a kúkú battery is unacceptable; even the sequence *kúkú battery a in which a retains its position in the Ewe NP is unacceptable because of the misplacement of the adjective kúkú ‘dead’. Thus, the mixed DP in (23) would support the MOP principle of the MLF model. Now consider example (24) from Swahili–English CS: (24) I – le membership kule kilabu kw-­enu ilikuwa pesa ngapi? CL 9-DEM there club LOC-yours it is money how much ‘How much is the membership (fee) at your club?’ (Swahili–English, Myers-Scotton 1993b: 95)

Theoretical Framework

37

In (24), the MOP would predict that the English noun membership must come before the Swahili demonstrative –le because, in Swahili, demonstratives typically follow the noun. However, the order of the mixed constituent i-­le membership in (24) is English. Myers-Scotton (1993b: 93–5) explains that the example does not violate the MOP because demonstratives can also precede nouns in Swahili, but as a marked order. Another seemingly problematic case for identifying morpheme order according to the MOP concerns EL elements that occur as bare forms. (25) Edwin? Ah, nilipoamka sikumuona, niliangalia dirisha z-­ao Window(s) CL 10-their ni-­ka-on –a curtain zi-­me-fung-­w-a I-CONSEC-see-INDIC CL 10-PERF-close-PASS-INDIC ‘Edwin? Ah, I didn’t see him when I woke up. I checked their window and saw (the) curtains were closed.’ (Swahili–English, Myers-Scotton 1993b: 95)

The EL noun curtain in (25) stands as a bare form without the necessary ML modifiers. For instance, the ML noun dirisha ‘window’ in the same example is followed by the possessive pronoun z-­ao ‘their’. According to Myers-Scotton (1993b: 112), bare forms like the English noun curtain in (25) are deployed by speakers to get around conflicts between the ML and the EL morpheme order. Furthermore, she explains that bare forms together with the use of the ML unmarked order and a marked morpheme order in the ML, but unmarked in the EL, are used as compromise strategies (this is discussed further in section 3.3 below) to preserve the MOP of the MLF model (see also Park 2000: 33). Also, the MOP does not apply to what Myers-Scotton (2002; 2006) terms EL islands (EL islands are the subject of discussion in Chapter 8). Other researchers have referred to these as ‘unambiguous’, ‘typical’ or ‘true’ codeswitches (see Andersson 1993; Eze 1998; Nortier 1990; Poplack, Wheeler and Westwood 1987). These EL islands are phrases within a bilingual clause; their words show structural dependency relationships – words in an island show a hierarchical structure, with some words ‘higher’ than others in the structure in which they occur (Myers-Scotton 2006: 261). (26) žib li-­ya een glas water of zo. get-­for me a glass of water or so ‘Get me a glass of water or something.’ (Moroccan Arabic–Dutch, Nortier 1990: 131, cited in Myers-Scotton 2006: 261)

38

Codeswitching in Igbo–English Bilingualism

So, in (26) the Dutch determiner phrase (DP) een glas water of zo ‘a glass of water or something’ is an EL island with een ‘a’ as its head, with the other elements as complements of the D head. Accordingly, EL islands are full constituents consisting only of EL morphemes occurring in a bilingual clause that is otherwise framed by the ML (Myers-Scotton and Jake 2014: 4). According to Myers-Scotton (2002), minimally EL islands can be composed of two content morphemes (e.g. noun and modifier) or a content morpheme and a non-­derivational system morpheme (e.g. noun and determiner). However, she points out that sequences of EL morphemes that are only juxtaposed are excluded. (27) À côté il y en a un autre gros building high-­rise at [the] side there is ART other big building high-­rise ‘Next door there is another big high-­rise building.’ (French–English, Poplack 1987: 29, cited in Myers-Scotton 2002: 139)

The problem with (27), according to Myers-Scotton, is that these morphemes (building high-­rise) follow the word order of the ML, French, not English; therefore, this is simply a sequence of two EL content morphemes in a mixed constituent under ML order (ibid.: 139). Furthermore, Myers-Scotton (2002: 149–53; 2006: 265) opines that some EL islands are more firmly integrated into the ML than others. She identifies these EL islands as ‘internal EL islands’; these EL islands are part of a larger phrase framed by a ML element, as in (28) below. (28) J’étais certain que c’était pas la real thing I be-PAST-1S certain that it be-PAST-3S NEG D-FEM real thing ‘I was certain that it was not the real thing.’ (French–English, King 2000: 100, cited in Myers-Scotton 2006: 265)

In (28), the DP is composed of the French D head la ‘the’ and the English NP complement real thing. Although there is no word order conflict in the case of the mixed DP in (28), Myers-Scotton, however, posits that while EL islands must follow the principles of well-­formedness of the EL regarding their internal structure (i.e. order within the phrase), they follow the placement rules of the ML within the clause. She argues that this is a very important similarity shared by both singly occurring EL words and islands. For instance, see example (29) below: (29) … ondan sonar   then after

lauw water’ nan lukewarm water-INSTRUMENTAL

yi kayinca I wash it

Theoretical Framework

39

‘… then I wash it with lukewarm water.’ (Turkish–Dutch, Backus 1992: 112, cited in Myers-Scotton 2002: 152)

Myers-Scotton (2002: 152, 2006: 265) explains that (29) illustrates the above claim because the instrumental phrase precedes the verb, as it would in Turkish, but not in Dutch. That is, the Dutch phrase lauw water (an internal EL island) is framed by the Turkish case marker for instrumental, giving rise to the meaning ‘with lukewarm water’. Crucially, she points out that the larger phrase is a post-­ positional phrase; with the head (the instrumental suffix -nan) of the phrase coming in final position (Turkish is a verb-­final language). Example (29) also supports the SMP (see section 3.2) because the Dutch equivalent of ‘lukewarm water’ is inflected with the Turkish instrumental case marker -nan. Another example that confirms the SMP is found in (30) below: (30) Oisitko sä voinu tämän arbetsskada-n estää?         this-ACC work-­related injury-ACC ‘Could you have prevented this work-­related injury? (Finnish–Swedish, Andersson 1993: 262, cited in Park 2000: 34)

In (30), the demonstrative and accusative (ACC) case endings (all system morphemes) in the mixed constituent tämän arbetsskada-n ‘this work-­related injury’ come from Finnish, the ML, while Swedish (the EL) contributes the content morpheme arbetsskada. However, in some examples the system morphemes can come from both languages as we observe in (31). (31) But ma-day-s a-­no      a-­ya- handisi   ku-­mu-on-­a but CL6-PL-day-PL CL6-PL-DEM CL6-DEM 1S-NEG-do INF-OBJ-see-FV ‘But these days I don’t see him much.’ (Shona–English, Crawhall 1990, cited in Myers-Scotton 2002: 301)

The mixed constituent in (31) ma-­day-s consists of the English noun day and two plural affixes, one each from Shona (ma-) and (-s) from English. This means that the noun day is doubly marked for plurality (or what is referred to in the literature as double morphology: Myers-Scotton 2002: 91). Some researchers have cited examples like (31) as counterexamples to the SMP (see Muysken 2000: 173; Park 2000: 34–5). However, Myers-Scotton (2002: 91–2) counters by stating that when EL plural affixes appear in mixed constituents, they do not violate the SMP because they (plural morphemes) are not the type of morpheme at issue under the specifications of the SMP; their status is claimed to be clarified under the Four-Morpheme (4-M) model (the 4-M is discussed in section 3.2.2 below)

40

Codeswitching in Igbo–English Bilingualism

in which they are explicitly differentiated from late outsiders and classified as early system morphemes. Nonetheless, Myers-Scotton describes double morphology as a type of ‘mistiming’ that arises when a speaker wishes to express his/her intentions by using an EL noun along with the concept of plurality. However, when the lemma2 for that noun is accessed, at the same time, its plural affix ‘slips in’, too (ibid.: 92). Another seemingly problematic case often cited as a counterexample to the SMP involves EL verbs that occur without the necessary ML inflections, as in (32). (32) Avan enne confuse-paNNiTTaan ‘He confused me.’ (Tamil–English, Annamalai 1989: 51, cited in MyersScotton and Jake 2014: 12.)

In (32), contrary to the prediction of the SMP, the EL infinitive verb stem confuse appears as a bare form without ML inflections for tense as required by Tamil. Instead the Tamil do verb construction (Myers-Scotton and Jake 2014) comes after confuse as an inflected form. Myers-Scotton (2002: 137) disagrees that bare forms, such as that in (32), violate the SMP; instead she argues that do verb construction is deployed by speakers as a compromise strategy to mitigate incongruence between the tense/aspect systems of participating languages. However, we agree with Park (2000: 35) that although bare forms do not violate the SMP as Myers-Scotton suggests, they do not directly support it, either.

3.2.2  The Four-Morpheme (4-M) model The MLF model has undergone a number of refinements since its inception in 1993, to make it a more robust and dynamic analytic framework to account for CS and other language contact phenomena. The most notable of these refinements has been in how the content–system morpheme opposition is refined under the 4-M model,3 which is outlined briefly below (for a comprehensive account of the model, see Myers-Scotton and Jake 2000a, 2000b; Myers-Scotton 2002; Myers-Scotton and Jake 2009). The model begins by distinguishing four morpheme types in terms of the difference in their role in syntactic structures broadly as well as empirical evidence about their distribution in CS. It uses the term ‘morpheme’ to include both the abstract element in language production that underlies surface forms and the actual surface-­level forms.4 Specifically, the model adds new specificity to the predictions of the MLF model and also predicts distributions across other types of linguistic phenomena rather than just systematizing what has been found already. These

Theoretical Framework

41

data distributions imply a hypothesis concerning abstract detail of language production, the Differential Access (DA) hypothesis: The different types of morpheme under the 4-M model are differently accessed in the abstract levels of the production process. Specifically, content morphemes and early system morphemes are accessed at the level of the mental lexicon, but late system morphemes do not become salient until the level of the Formulator. Myers-Scotton 2002: 17

Under this hypothesis, the definition of content morphemes remains unchanged (see section  3.2.1). However, they, along with one type of system morpheme called an early system morpheme, are specifically characterized as conceptually activated. Myers-Scotton (2002) explains that conceptually activated means that speaker prelinguistic intentions activate (or select) content morphemes and any early system morphemes that may accompany them on the surface. This activation occurs at the first level of what is termed the mental lexicon under the 4-M model. The mental lexicon is said to consist of elements called lemmas that are tagged for specific languages; the speaker’s intentions call up language-­ specific lemmas, which contain the information necessary to produce surface-­ level forms. Furthermore, lemmas in the mental lexicon that underlie content morphemes (e.g. nouns and verbs) are directly activated through the speaker’s intention. In turn, these lemmas activate the lemmas underlying early system morphemes. These early system morphemes flesh out the meaning of the lemmas of the content morphemes that call them. These system morphemes are called ‘early’ because of their early activation in the language production process. Examples of early system morphemes (Myers-Scotton 2006: 268) include plural markings, determiners (e.g. the definite article the and the indefinite articles a and an in English), and those prepositions (also called satellites) that change the meanings of phrasal verbs in certain contexts (e.g. out as in Alice looks out for her little brother; or through in the actor ran through his lines before the performance). This conceptual link between content and early system morphemes means that EL early system morphemes have more potential for appearing in bilingual clauses framed by the ML. In fact, this is the basis for the Early System Morpheme Hypothesis (ESMH), which seeks to account for the occurrence of double morphology in CS (see example 31 above). Double morphology in the Igbo–English data is discussed in section 6.2.4. ESMH:  Only early system morphemes may be doubled in classic CS. Myers-Scotton 2002: 92

42

Codeswitching in Igbo–English Bilingualism

Myers-Scotton explains that of all system morphemes only early system morphemes have the special relation to their heads that would promote their accessing when their EL heads are called in CS. Furthermore, she adds that, like their heads, early system morphemes are conceptually activated in the mental lexicon at the same time as their content morpheme heads. Thus, they are ‘available’ if any mistiming is going to occur (ibid.: 92). The other two types of system morphemes (bridge late system morphemes and outsider late system morphemes) are called ‘late’ because the model claims that they are not activated until a later production level, at a second abstract level that is called the formulator. According to the model, the formulator is viewed as an abstract mechanism that receives directions from lemmas in the mental lexicon (those underlying content morphemes); the directions from the lemmas underlying content morphemes tell the formulator how to assemble larger constituents, such as combinations of NPs/DPs and inflection (I)/verb phrases (VPs), resulting in a full clause. This is also the level where late system morphemes are activated to indicate relationships within the clause (Myers-Scotton 2006: 245, 268–9). This hypothesis is pertinent in explanations of the Igbo–English data in the coming chapters of this volume. Regarding bridge late system morphemes, they occur between phrases that make up a larger constituent, and the best example of a bridge is the associative or possessive element that occurs between a possessor noun and the element that is possessed in a number of languages. For instance, of is a bridge, as in the house of Gina. Also, the model considers the possessive -’s in English to be a bridge morpheme, as in Gina’s house. A bridge morpheme depends on the well-­ formedness conditions of a specific constituent in order for it to appear; such a constituent is not well formed without the bridge morpheme. Outsider late system morphemes like bridge late system morphemes also satisfy well-­ formedness conditions. However, they are said to differ from bridges in that the presence and form of an outsider depends on information that is outside the element with which it occurs and therefore outside its immediate constituent. That information comes from an element in another constituent or from the discourse as a whole. Myers-Scotton (2002) gives the clearest example of an outsider late system morpheme in English as the element that shows subject– verb agreement on the verb in many languages. She explains that the form of the agreement marker depends on the subject. Thus, English speakers would say the dog like-­s chewing bones, but dogs like-Ø chewing bones. The suffix -s only occurs when there is a third-­person singular content element in the present tense to call that suffix; otherwise, in English, there is no suffix (Ø = ‘zero’ marker). Following

Theoretical Framework

43

is a brief outline of another refinement to the MLF model in the form of the Abstract Level (AL) model.

3.2.3  The Abstract Level model The AL model is a model of the abstract lexical structure of entries in the mental lexicon. Under this model, the levels of abstract structure include lexical-­ conceptual structure, predicate–argument structure, and the level of morphological realization patterns. According to Myers-Scotton and Jake (2001: 125), lexical-­ conceptual structure is closest to the speaker’s intentions. They explain that, in their view, pre-­verbal intentions in the conceptualizer activate language-­specific semantic/pragmatic feature bundles at the interface between the conceptualizer and the mental lexicon; these bundles are mapped onto entries in the mental lexicon (lemmas) as lexical-­conceptual structure. Next the predicate–argument structure deals with how thematic structure is mapped onto grammatical relations. For example, in a specific language, this level may provide for the mapping of Agent to Subject, Beneficiary to Internal Object, etc. The level of morphological realization patterns refers to how grammatical relations are realized in surface configurations. This level includes morpheme order and agreement morphology (Myers-Scotton 2002: 19). Myers-Scotton (2013: 35) notes that other linguists discuss lexical structure, some with ideas similar to this model. However, she claims that what is innovative about the AL model is the notion that, in bilingual clauses, the three levels are not necessarily equally salient. Thus, in this model, the reason many EL nouns occur is that they need only convey a speaker’s intended message at the abstract lexical-­conceptual level; that is, an EL noun need not match an ML counterpart at the levels of predicate–argument structure and morphological realization patterns (Myers-Scotton and Jake 2014: 4–5). This flexibility in salience is claimed to be relevant to CS. It implies, according to Myers-Scotton (2013), that the three levels in one morpheme type can be split and combined with levels in another morpheme type. This can make a difference in CS when the issue of inserting an EL element into an ML frame arises. Therefore, it is especially relevant to explaining how EL verbs can participate in ML frames with ML verbal inflections. It is, however, important to stress that this study is not so much concerned with the psycholinguistic aspects of bilingual language production as it is with the claim under the MLF model that the frame-­building system morphemes (i.e. outsider late system morphemes) in mixed constituents come from only one language, the source of the ML. The other language, the EL, may supply content morphemes that are inserted in this frame (Myers-Scotton

44

Codeswitching in Igbo–English Bilingualism

2002: 20). This theoretical notion is formalized as three testable hypotheses claimed to be universally applicable in cases involving classic CS.

3.2.4  Principles of the MLF model and Igbo–English codeswitching 3.2.4.1  The Matrix Language Principle (MLP) According to this principle, in classic CS it is possible to identify the ML of any bilingual clause. The ML will be that of the morphosyntactic frame. Myers-Scotton (2001: 52; 2002: 25, 105) explains that users of classic CS have full access to the morphosyntactic frame of one of the participating languages that becomes the ML and anywhere from limited to full proficiency in the grammar of the other language. The assumption, then, is that because of their grammatical competence in their target ML, usually the speaker’s mother tongue (or first language: L1), classic CS users are expected to produce mixed constituents that conform to the grammar of this language. Furthermore, Myers-Scotton (2002) adds that: The ML–EL opposition refers to linguistic competence – in the sense that, psycholinguistically, the bilingual’s two or more languages do not achieve equal activation in bilingual speech. Decisions (largely unconscious) made at the prelinguistic conceptual level result in one language dominating (the ML sets the grammatical frame of such speech). The less dominant language (the EL) participates largely by supplying lexical elements that are integrated into that frame. Myers-Scotton 2002: 16

The processes determining how the ML constrains abstract grammatical structures of mixed constituents is related to the modular nature of language production and the abstract and multifaceted nature of lemmas supporting morphemes as well as the division of these morphemes into four types (see section 3.2.2 above). In this study, two specific principles of the MLF model that embody the MLP will be tested with Igbo–English data:

1. The Morpheme Order Principle (henceforth the morpheme order criterion): in mixed constituents consisting of at least one EL word and any number of ML morphemes, surface word (and morpheme) order will be that of the ML (Myers-Scotton 1993b: 83; 2002: 59; 2006: 244). 2. The System Morpheme Principle (henceforth the late system morpheme criterion): in ML+EL constituents all system morphemes which have grammatical relations external to their head constituents (i.e. which participate in the sentence’s thematic role grid) will come from the ML (Myers-Scotton 1993b: 83; 2002: 59; 2006: 244).

Theoretical Framework

45

We shall define and illustrate our two criteria fully in the later chapters of this study, where we apply them and the rest of the principles outlined below to the Igbo–English data.

3.2.4.2  The Asymmetry Principle (AP) This principle states that bilingual speech is characterized by asymmetry in terms of the participation of the languages involved in CS. In addition, it states that in classic CS only one of the participating languages is the source of the ML. This asymmetry is claimed as evidence of the universal drive in language to achieve uniformity in the structural frame of any variety, to avoid meaningless variation (Myers-Scotton 2002: 9). In this study, we shall interpret this to mean that in every bilingual clause in the Igbo–English data only one language (the ML) will be the source of inflectional morphology (which we shall define as late system morphemes) on all the verbs. This asymmetry will also be evident in mixed constituents where there is a conflict in word order between the ML and the EL. When this happens, the expectation is that the order of only one of the languages (the ML) will prevail. Both the EL and the ML will contribute content and early system morphemes (see sections  3.2.1 and 3.2.2 above). However, our expectation is that more of such morphemes will come from the ML than the EL.

3.2.4.3  The Uniform Structure Principle (USP) This principle states that a given constituent type in any language has a uniform abstract structure and the requirements of well-­formedness for this constituent type must be observed whenever the constituent appears. It also states that in bilingual speech, the structures of the ML are always preferred, but some EL structures are allowed if certain conditions are met (Myers-Scotton 2002: 8–9). To evaluate this prediction we shall focus particularly on the use of determiners in mixed DPs in Igbo–English CS. As we have already pointed out above, we shall offer detailed exemplifications and analyses relating to the three principles outlined here in subsequent chapters of this volume.

3.3  The role of congruence in codeswitching It was Weinreich (1953: 33) who much earlier argued that it stands very much to reason that the transfer of morphemes is facilitated between highly congruent

46

Codeswitching in Igbo–English Bilingualism

structures (cited in Sebba 1998: 3). Weinreich’s focus here was primarily with the paradigmatic aspect of congruence. He suggested that although bound morphemes would be more difficult to transfer than free morphemes because of their language-­specific grammatical function, transfer would be possible if there was sufficient congruence between the bound morphemes of one language and another. He used the notion of transfer to cover both borrowings and CS. However, the significance of his observation on congruence for CS cannot be overemphasized (see also Deuchar 2005). Muysken (1995: 192) uses the term equivalence rather than congruence, arguing that the guiding assumption is that equivalence between the grammars of two languages facilitates bilingual usage, be it second language learning, lexical borrowing or CS. Muysken clearly identifies both the paradigmatic and syntagmatic aspects of equivalence as follows: ‘There can be equivalence of categories (lexical elements, phonemes, phrase structure nodes, morphosyntactic features) or of relations between categories, in structuralist terms. The latter are either syntagmatic (e.g. word-­ order or agreement rules) or paradigmatic (equivalent oppositions)’ (Muysken 1995: 192f, cited in Deuchar 2005: 258.). In her Welsh–English study, Deuchar (2005: 256) uses the term ‘congruence’ to refer to a notion of equivalence between the grammatical categories or word classes of different languages, viewed from both a paradigmatic and a syntagmatic perspective. She defines paradigmatic congruence as similarity or equivalence between the grammatical categories of two languages, whereas syntagmatic congruence is defined in terms of similarity of word order (after Muysken 2000: 118). Furthermore, she explains that from a paradigmatic point of view the category of adjective, for example, may be (but is not necessarily) equivalent in two languages; whereas from a syntagmatic point of view the placement of an adjective in relation to a head noun may or may not be congruent, depending, for example, on whether or not the two languages have the same relative order for head and modifier. Deuchar explains that in English and Welsh, for example, one might argue that adjectives are paradigmatically equivalent in that they modify nouns in both languages, but that they are not syntagmatically equivalent because adjectives generally precede nouns in English but follow them in Welsh (2005: 256–7). Igbo is similar to Welsh in this respect (see the next chapter for a discussion of aspects of Igbo and English grammars that have a bearing on the CS analyses reported in this study). On their part, Myers-Scotton and Jake (1995) observe that congruence between cross-­linguistic categories is assumed to be determined mainly by the linguistic properties of the elements; that is, the nature of the three sub-­parts of

Theoretical Framework

47

lemmas supporting the elements (see the discussions in sections 3.2.2 and 3.2.3 above). The degree of congruence may be complete (if there is a match in the details of their three sub-­parts), partial (if there is a match across some of the details in the sub-­parts) or absent (if there is no match). Since complete congruence between cross-­linguistic lemmas is rare, they, for instance, state that the lemma supporting a lexical entry in one language that refers to even an easily accessible, concrete entity like ‘nose’ might not match the lemma supporting a corresponding lexical entry in another language completely ‘because of pragmatic considerations’. In addition, they also point out that lemmas supporting lexical items that code ‘more complex and less concrete’ ideas or processes may be even less regularly congruent cross-­linguistically because of ‘the semantic and pragmatic features associated with them’ (Myers-Scotton and Jake 1995: 988). Wei (2001) adopts Myers-Scotton and Jake’s notion of congruence and congruence checking in his analysis of its role in his Chinese–English CS data. He argues that the insertion of English content morphemes in Chinese is possible because they are congruent with their Chinese counterparts. On the other hand, he suggests that incongruence in semantic/pragmatic features or predicate– argument structure can lead to compromise strategies such as the production of EL islands, as in the phrase work in the lab which is inserted into an otherwise Chinese utterance (Wei 2001: 161). Also, Jake et al. (2002: 70) in their paper ‘Making a minimalist approach to codeswitching work: Adding the Matrix Language’ compare congruence checking to feature checking within the Minimalist Program (Chomsky 1995): Congruence checking at the lemma level verifies that the abstract structures projected by a content morpheme are congruent enough to be inserted. Viewed from the minimalist perspective [see section  2.6], insertion of an EL content morpheme depends on features that figure in the narrow syntax (what Chomsky calls interpretable features) and play a role at the level of logical form (LF). Deuchar 2005: 258

According to Radford (2004: 224–6, 343), a feature is (semantically) interpretable if it has semantic content: so, for example, a feature such as [Plural-Number] on the English pronoun they is interpretable, but a phonological feature like [+nasal] is uninterpretable, and so too are many grammatical/formal features such as case features. Jake et  al. (2002) argue that uninterpretable features, which are grammatical features without semantic content, like case features, do not participate in congruence checking. However, as Deuchar (2005: 258) correctly observes, Jake et al.’s conception of feature checking for congruence in fact differs

48

Codeswitching in Igbo–English Bilingualism

from feature checking in minimalism in that it involves the paradigmatic rather than the syntagmatic dimension. This, she notes, is made clear in the following, where EL and ML refer to different languages: ‘Once a speaker selects an EL content morpheme, the relevant feature matching is between the EL form and the features of an ML counterpart, not between the EL content morpheme and the ML system morpheme’ (Jake et al. 2002: 78). Furthermore, Jake et al. (2002: 69) point out that their model incorporates a basic asymmetry between the languages participating in CS, and this also affects their consideration of congruence. According to Jake and Myers-Scotton (1997: 27), when a speaker engages in intrasentential CS an EL content item is inserted into the grammatical frame prepared by the ML. Part of this process is the evaluation of its congruence with an ML counterpart. Thus, it is the EL item which is evaluated for congruence with an ML equivalent, rather than the reverse. This assumption of asymmetry in the concept of congruence under the MLF model will underpin the analysis of the Igbo–English data presented in later chapters of this volume. Moreover, it is important to point out that complete cross-­linguistic congruence across all three sub-­parts of EL and ML equivalent lemmas is not usually considered a mandatory measure of congruency. Instead, congruency is discussed under the MLF model in terms of ‘sufficient congruency’, which means anything between complete and partial congruency (see also Amuzu 2010: 30). In fact, MyersScotton (2002: 20) states that sufficient does not mean complete congruence and that an EL lexeme does not necessarily have to match an ML counterpart regarding lexical category. Hence, after Myers-Scotton (1993b, 2002) and Amuzu (2010) we will characterize the mechanism underlying the CS behaviour of our speakers (the speakers used in the study are introduced in section 5.2.1) thus: Stage 1: Once the lemma supporting an EL content morpheme is selected at the lemma (or conceptual) level to express a pre-­verbal intention. Stage 2: That lemma is checked for congruence with the lemma supporting the ML counterpart of the EL morpheme at the three levels of abstract lexical structure: lexical-­conceptual structure, predicate–argument structure and morphological realization pattern. One of two things happens:

1. The two elements are determined to be sufficiently congruent: Accordingly, the morphosyntactic frame meant for the ML content morpheme (at the functional level) is used for realizing the EL content morpheme at surface structure. That is, the EL

Theoretical Framework

49

morpheme is fully inserted into a slot projected by its ML counterpart. The morpheme order and system morpheme criteria are involved in the functional level processes as described above in sections 3.2.2 and 3.2.3. This CS pattern is discussed in sections 6.2.1.1 to 6.2.2, 6.2.4 and 7.2 to 7.3. 2. The two elements are determined not to be sufficiently congruent: A compromise strategy is activated and used with the result that the EL content morpheme is realized as either a bare form or as a part of an EL island. These CS patterns are discussed in sections 6.2.3, 7.5 to 7.5.1 and 8.2 to 8.3.6. Nevertheless, we shall argue in this study that we are not entirely convinced that incongruence between ML and EL elements is the reason behind the occurrence of bare forms in Igbo–English. Additionally, we shall show that no compromise strategies are required for their insertion either. It appears that the only considerations involved are ML well-­formedness requirements for the structures concerned.

3.4  Conclusion In this chapter, we have shown that, in contrast with the approaches to the study of intrasentential CS discussed in the previous chapter, Myers-Scotton claims that her MLF model is not a slate of descriptively based constraints, but a model with interrelated parts that offer explanations for why its constraints take the form they do. The model suggests that the nature of language production is relevant. That is, the model indicates that modelling language production is relevant to CS, but not that the framework itself is a model of language production. An important premise of the model is that while CS and other bilingual data follow most of the same syntactic principles applying to monolingual data, syntactic models devised for monolingual data do not suffice for explaining CS structures. A further premise of the model is that bilingual data, such as CS, cannot be sufficiently explained at the level of phrase structure alone. It emphasizes the abstract procedures directed by lemmas in the mental lexicon. Some of these procedures, as the discussion in the preceding sections indicates, necessarily refer to phrase structure, but also to the role of oppositions elsewhere at more abstract levels. Myers-Scotton suggests, that crucially, lemmas in the mental lexicon underlie surface-­level lexical elements; the lemmas contain

50

Codeswitching in Igbo–English Bilingualism

lexical rules and these rules contain all the necessary information to realize surface constructions. Critically, two interrelated oppositions or hierarchies are the key to the nuts and bolts of the MLF model: the members of the oppositions play unequal roles in CS structures. In the next chapter, we shall present a short overview of aspects of Igbo grammar that contrast with those of English; these will form the basis for the analysis of Igbo–English CS in the later chapters of this study.

4

Comparison of Aspects of Igbo and English Grammars 4.1  Introduction Since the concern of this study is with exploring grammatical aspects of Igbo– English CS, here we shall present brief descriptions of the structure of Igbo morphophonology and syntax that differ from those of English. The following descriptions may seem to focus more on Igbo rather than English, only because some of the units of grammar are not found in the English language. A second reason stems from the fact that Igbo is the less familiar language of the two and some readers may not have much information on its grammar, unlike that of English which is widely known. However, the descriptions do highlight the main typological differences between the two languages. In sum, the overall purpose of this chapter is to facilitate the grammatical analysis of Igbo–English CS in subsequent chapters of this volume. The present chapter is organized as follows: section  4.2 presents the Igbo orthography adopted in the study; section  4.3 outlines the main contrastive features of Igbo morphophonology which we anticipate will have a bearing on the CS analysis presented in the later chapters of this study; the focus of section  4.4 is on outlining the major structural differences between Igbo and English; and finally in section  4.5 we shall summarize the most important typological differences between Igbo and English.

4.2  Igbo orthography The Igbo orthography adopted in this study is the New Standard Orthography (NSO) reproduced in Figure 4.1 with their corresponding IPA (International Phonetic Alphabet) symbols in bold.

Codeswitching in Igbo–English Bilingualism

52











Figure 4.1  Igbo standard orthography with the corresponding IPA symbols

The NSO orthography in Figure 4.1 is an updated and modified version of the official O. nwu. Orthography (1961). Tone-­marking1 is not obligatory in Igbo, and because there has been no agreement as to whether all or only ‘low’ and ‘downstepped’ tones should be marked (Echeruo 1998: xii), the examples from Igbo–English cited in this study do not have tone markings. This is also done to further limit notational complexity (see also Zsiga 1997). Therefore, we consider that tone markings are not essential to the effective analysis of our Igbo–English CS data. Having made these clarifications, we will now begin outlining some of the salient structural differences between Igbo2 and English.

4.3  Morphophonology The morphology of Igbo is inextricably tied to its phonology. Such intertwining of morphology and phonology is discussed in the literature under the heading morphophonology. According to Hannahs (2001: 10053), in its broadest definition, morphophonology refers to the interaction of word formation (morphology) with the sound system of language (phonology). These two linguistic subsystems interact in many different ways in the languages of the world, including phonological alternation of a stem or affix, vowel harmony (VH) involving (both) stems and affixes, spread of a phonological feature as a grammatical marker, combinations of patterns of consonants and vowels in nonconcatenative morphology, and phonologically definable reduplication. It is on Igbo syllable structure and VH process that we focus in the next two sections.

Comparison of Aspects of Igbo and English Grammars

53

4.3.1  Syllable structure Igbo has a basic two-­syllable structure, as in o. ku. [.V.CV.] ‘light’, O [V] (pronoun) ‘he/she/it’ or [.C.VV.] kpo.o. ‘call’, which can expand to include forms such as VCV+CV(V), or VCV+VCV (the dots are used to mark syllable boundaries following Maximal Onset Principle3 and Igbo language-­specific Phonotactic4 constraints). Most morpheme boundaries are marked by double vowel sequences, which are subject to the process of VH (see section 4.3.2 below). Therefore, Igbo, unlike English, does not permit either coda consonant(s) or consonant clusters anywhere in its syllable structure. Nouns such as mma ‘goodness/beauty’, nnu.  nu. ‘bird’ and ngo.zi ‘blessing’ do occur in Igbo; however, in all the cases, the first sound is a syllabic nasal rather than a true consonant. Thus, these examples constitute the typical structure of disyllabic Igbo nouns: [+syllabic nasal: m̩ n̩ ŋ̩ ]-CV. Any loanwords (or borrowings5) with the disallowed structural configurations undergo vowel epenthesis throughout their syllable structure in order to be properly integrated into the language. Support for this claim can be found in the borrowed words listed in two Igbo dictionaries produced for the teaching of Igbo language in Nigerian schools: O. ko.wa Okwu: Igbo–English Dictionary (1962) by F. C. O. gbalu. and A Modern English–Igbo Dictionary (1985) by H. I. Nnaji. For instance, compare the following English-­origin words with their corresponding entries in the O. gbalu. dictionary: ‘bacteria’ [.bæk.tɪə.rɪə.], ‘bank’ [.bæŋk.], ‘latitude’ [.læ.tɪ.tjuːd.], ‘sulphur’ [.sʌl.fə.]; as they appear in the Igbo dictionary: bakiteria [.ba.ki.te.ria.], banki [.ba.ŋ̩ .ki..], latitudi [.la.ti.tu.di.], so.lu. fo. [.sɔ.lʊ.fɔ.]. It is evident from these examples that the established borrowings from English in Igbo follow strictly the latter’s syllable structure and can be conclusively said to behave like the Igbo words cited earlier in this section: o.ku. [.V.CV.] ‘fire/light’, O [V] (pronoun) ‘he/she/it’, and kpo.o. [.C.VV.] ‘call’. This information we believe will be useful in later chapters in assisting us to clarify whether the singly occurring EL words in the Igbo–English data are CS forms rather than some kind of borrowing.

4.3.2  Vowel harmony Vowel harmony (VH) is a type of vowel assimilation that is not known to be present in the phonology of English. Therefore, knowledge of how it operates in Igbo (particularly in verbs and affixes; see section 4.4.6 below) could be another useful diagnostic for disambiguating the classification of certain lone other language words as either CS forms or borrowings.

Codeswitching in Igbo–English Bilingualism

54

Table 4.1  Igbo vowel correspondences and feature classes Igbo

ɪ

i.

e

a

u

u.

o

o.

IPA [high] [round] [ATR]

i + – +

ɪ + − −

e − − +

a − − −

u + + +

ʊ + + −

o − + +

ɔ − + −

Adapted from Zsiga (1997: 232).

Vowel harmony is a phonological process by which vowels agree with other vowels in terms of a particular phonological feature (Harrison 2000: 111), such as backness, roundness, height or Advanced Tongue Root (ATR). Table  4.1 displays the eight vowels that are found in Igbo. It has been suggested by Casali (2008: 500) that VH is typically present in languages with a rich morphological system, where affixes have varying vowels based on the surrounding sounds. Igbo is one such language with a rich number of affixes (see Table  4.6) that typically harmonize with the vowel of the stem (or root) to which they are attached, as in the monolingual Igbo examples below. (33) i.-zu. ‘to buy’ (Infinitive verb) [-ATR] harmony

˅

In the literature, VH is said to operate in two main ways: (1) root control harmony, and (2) dominant-­recessive harmony. In the first type, the vowel of the affix assimilates to the root. In the second system, VH is phonologically driven (Krämer 2003: 35), and rarely found outside of ATR harmony systems. That is, one class of vowels ([±ATR]), if present in either root or affix, causes all of the other vowels within the word to assimilate to its value. In Igbo, VH falls under the first category (i.e. root control harmony). Thus, in Igbo, inflectional affixes always harmonize with their closest stem/root vowel. With the verb stem zu. ‘buy’ in (34), for instance, there is an inflectional vowel prefix a- and a negative imperative suffix -la; both affixes harmonize with their stem vowel as in: A-zu. -­la efe ahu. ‘Don’t buy that shirt/dress’. (34) a-­zu. -­la ‘don’t buy’ [-ATR]

The vowels of the affixes harmonize for the feature [-ATR]. Also, most VH processes spread the vowel values left-­to-right (Clements and Sezer 1982: 219;

Comparison of Aspects of Igbo and English Grammars

55

Finley 2008: 9), though there are a non-­negligible number of cases where right-­ to-left spreading occurs. In example (34) above, the VH process is bidirectional; spreading the stem vowel [ATR] value in both directions when an affix is attached to both ends of the stem. It is equally important to point out that VH systems do sometimes contain vowels that do not participate in the harmony process. These ‘neutral’ vowels, as they are termed, often arise from gaps in the vowel system. A neutral vowel can be of two types, opaque and transparent: the former start a new harmonic domain with their own feature specification, thus blocking the spread of the harmonic feature value (Krämer 2003: 27); while transparent vowels allow the feature to spread through the (neutral) vowel to the other side without distorting the harmony pattern (Finley 2008: 7). As observed by Kiparsky and Pajusalu (2003: 221), both transparent and opaque vowels can be neutral always, only in particular contexts, or unpredictably neutral. Crucially, in relation to Igbo, there are no neutral vowels in the language. In fact, every vowel has a harmonic counterpart (see Table  4.1 above). Therefore, concerning lone English-­origin verbs inserted into otherwise Igbo utterances, the degree to which the vowels closest to either the left or right edge of the words are mutated to become harmonic with the vowels of Igbo bound affixes will indicate: (1) that two different phonologies can mix in CS contrary to the predictions of the FMC and the PFIC-based account (both models are reviewed in sections  2.3 and 2.6 respectively); and (2) that such mixed constituents do not lead to a crash in CS.

4.4  Lexical and grammatical categories 4.4.1  Noun (N)/Noun Phrase (NP) In Igbo, the N is any word that can function as head (H) of the NP, where it is not bound. Igbo Ns are of various sizes, shapes and derivations. A few of them have just one syllable: di ‘husband’; others have two syllables: o.ku. ‘fire’; or are multisyllabic: o.tu. ru. kpo.kpo. ‘woodpecker’. There are no N classes as such in Igbo; however, Emenanjo (1978: 45) offers a number of what he terms ‘suggestive’ lexical classes of Ns. Four of these are given below:

1. Proper Ns: include names of persons, places and sacred institutions and concepts. The Ns may feature in any of the three slots in the Igbo sentence: the NP Subject (S) slot; the Indirect Object (IO) slot and the Complement (C) slot.

Codeswitching in Igbo–English Bilingualism

56

2. Common Ns: used to refer to commonplace things, places, institutions, concepts, and so on. They occupy the same syntactic slots as the proper Ns, and may be used as H of the NP. 3. Qualifactive Ns: a few Ns in this class may precede or follow the Ns they modify, such as ezigbo ‘good/goodness’, and nnukwu ‘big/bigness’, etc. These so-­called qualifactive Ns are said to be inherently semantically descriptive, and have thus been frequently called ‘adjectival nouns’ (see Emenanjo 1978; Maduka-Durunze 1990). In this study, we will refer to these Ns straightforwardly as adjectives (see section 4.3.3). 4. Adverbial Ns: frequently used as head of the NP; their most frequent place of occurrence is in the adverbial slot, and translate as adverbs (ADV) in English. From the above, we can determine that the Igbo N/NP, like that of English, can function as: (1) the S of a sentence; (2) the IO of a sentence; (3) the C of a verb in a sentence; and (4) the sentence adverbial. However, there is no grammatical gender in Igbo and its Ns are neither declined for case nor inflected for number (as is typically the situation in English). Plurality may be specified optionally by words such as the pronoun (PRN) u. mu. (literal meaning: ‘replicas’ or ‘copies of ’) and ndi. (‘people of/those of ’) (Echeruo 1998; Emenanjo 1978).

4.4.1.1  Nouns in associative constructions Similarly to English, in Igbo a N can follow another N to form an associative construction as in example (35) below. (35)

O. - na- a-­gba igwe CL (He/she/it) AUX-V-ride bicycle-N ‘He is riding Kanye’s bicycle’

Kanye Kanye-N.GEN

The situation in (35) is different from that of a language like English, where typically only the N in the genitive case is inflected. In Igbo, always, it is the genitival N which comes second. The preceding N is said to be in a pre-­genitival position. The implication of this for our study of the syntax of Igbo–English CS, then, is to see whether the same placement rules apply to lone English Ns in associative constructions in the Igbo–English data.

4.4.2  Adjectives (As) Igbo has only a limited set of items that function as true adjectives in the language: o.ma ‘good/beautiful/handsome’, o.jo.o. ‘bad/ugly’, o.cha ‘white/bright’, ojii

Comparison of Aspects of Igbo and English Grammars

57

‘black/dark’, edo ‘yellow’, uhie ‘red’ and ukwu ‘big/large’. Other non-­basic forms are o.mari.ca (emphatic form of o.ma), njo.ko.ro. (emphatic form of o.jo.o.), ajo. (a slightly stronger form of o.jo.o.), and nnukwu (emphatic form of ukwu) (see Emenanjo 1978: 70–1; Maduka-Durunze 1990: 237). A distinctive feature of true As in Igbo is that they typically occur in associative constructions, unlike the ones found in English which can also occur in predicative constructions. An associative construction is one in a non-­predicative form in which a modification is made of a head such that the modification is associated with it. In the following examples, we notice that As in Igbo typically occur in post-­nominal position in associative constructions: (36) (37) (38)

nwoke man English: u. lo. house English: efe dress English:

o.ma handsome/good ‘handsome/good man’ o.cha white ‘white house’ uhie red ‘red dress’

Although the true Igbo As occur only post-­nominally, some other adjectival forms occur both post-­nominally and pre-­nominally, as is the case with the elements Emenanjo (1978: 47–8) terms ‘qualifactive nouns’ (e.g. ogologo ‘long/ tall’, mkpu. mkpu. ‘short’, abadaba ‘broad’, and so on). (39) (40)

nwoke man English: ogologo tall English:

ogologo tall ‘this tall man’ nwoke man ‘this tall man’ or

a this a this ‘the tallness of this man’

Crucially, whereas the Igbo qualifactive Ns functioning as As can pre-/post-­ modify a N, in English, As typically occur in pre-­position to the N under DP/NP. Additionally, only the Igbo adjectival Ns (never the true adjectives), like the As of English, can occur in the predicative position following the Igbo existential copula verb di., as in (41) below: (41) Nwoke ahu. Man D ‘That man is tall’

di. BE

ogologo tall

58

Codeswitching in Igbo–English Bilingualism

The examples illustrate that in Igbo and English adjectives can function both attributively and predicatively. However, we would expect the vast majority of English-­origin As in the data to occur predicatively (as part of a clause predicate) because this seems to be the position where the structure of both languages is most compatible under the category of As. Therefore, it would be interesting to uncover how Igbo–English speakers deal with conflict sites, such as those involving post-­nominal As in CS.

4.4.3  Preposition (P) Compared to English with an array of different Ps (e.g. beside, to, in, from, with, towards, within, off, by, up, before, since and so on), Igbo has just one preposition na, with the meaning: ‘at, in, on, on top of, on the outside, within, inside of ’ and so on. Its specific meaning is derived contextually. It is found preceding Ns/NPs/DPs, as well as before verbs (Vs) and verb phrases (VPs). In both Igbo and English, Ps typically express locational relations in space or time, though we acknowledge that this is not their only function. In fact, Ps can be used to indicate the thing, place or direction from which something goes, comes or is driven or moved: from, away from, or out of. Generally, however, both in English and Igbo, Ps are said to take NP/DP complements. For example, na ahi.a afo. ‘at/in the Afo. market’; na (= H of the PP) introduces the NP (complement) ahi.a afo..

4.4.4  Verbs and affixation Igbo Vs like those of English are found immediately preceding their complement(s) in the VP. The Vs can be classified as either active or stative. Active verbs are used for expressing action or activity, while stative verbs (e.g. the copulas: bu. , wu. di. and no.) are used for expressing qualities/states and existential notions of being. There are also a number of auxiliary (AUX) Vs which are unlike those of English. For instance, they are all bound morphemes in Igbo. That is, an Igbo AUX V can only be used in obligatory combination with a verbal derivative which makes it complete and meaningful. Emenanjo (1978: 127) gives a list of seven such Igbo Vs. However, we only list the three that are used in the Igbo dialect of our speakers (see section 1.4.1): (42) na- marks the progressive: Progressive affirmative: Nnenna na-­a-­bi.a AUX-V-come ‘Nnenna is coming’ or ‘Nnenna usually comes’

Comparison of Aspects of Igbo and English Grammars

59

Progressive negative: Nnenna a-­na-ghi. a-­bi.a V-AUX-NEG V-come ‘Nnenna is not coming’ or ‘Nnenna does not usually come’ (43) ga- marks the future: Future affirmative: Nnenna ga-­a-bi.a AUX-V-come ‘Nnenna will come’ or ‘Nnenna is going to come’ Future negative: Nnenna a-­ga-ghi. a-bi.a V-AUX-NEG V-come ‘Nnenna will not come’ or ‘Nnenna is not going to come’ (44) ga- marks the unfulfilled: Nnenna gaara-­a-bi.a AUX-V-come ‘Nnenna should have come’

In some sense, the Igbo bound AUX Vs behave like modals in English because they mark tense and aspect and assist to differentiate the different verbal constructions. Furthermore, it is important to point out that the V is the only grammatical category in Igbo that can take inflectional affixes.

Table 4.2  Summary of Igbo verbal morphology Inflectional category

Affix used to express category

Imperative:

This is formed by attaching the harmonizing open vowel suffix to the V root/stem: zu. -o. ‘buy!’ This is formed by attaching the prefix a-/e- and the suffix –le/-la to the verb: a-­zu. -­la ‘do not buy!’ This is formed by prefixing a harmonizing i-/ï- to the verb stem: i.-­zu. ‘to buy’. This is formed by attaching the –rV suffix to the verb, where V is the last vowel of the verb stem/root: zu. -­ru. ‘bought’. This is formed by attaching the harmonizing –ghi/-ghi. suffix to the verb: a-­zu. -­ghi. ‘did not buy’. This is formed by prefixing: e-/a- to the verb: a-­zu. ‘buying’. The perfective is formed by attaching the suffix –(V) la/-(V)le to the verb: zu. -o.-­la ‘has/have bought’. This is formed by attaching the inceptive extensional suffix -we/-wa to the verb: zu. -­wa-la ‘starting to buy’. This is formed by attaching the suffix –ie to the verb stem.

Negative imperative Infinitive: Indicative affirmative:

Negative indicative: The participle: The perfective: The inceptive: The conditional:

60

Codeswitching in Igbo–English Bilingualism

Table 4.3  Summary of English verbal morphology Inflectional category

Affix used to express category

Third person singular present: Past tense:

-s, -es: It pours, It washes -ed: It poured; (irregular verb stems undergo suppletion: go → went) -ed: has/have/had poured -ing: am/is/are/was/were pouring

Perfect aspect: Progressive/continuous aspect:

Note: The future tense in English is expressed by means of modal auxiliaries (will/shall), and the infinitive is marked with to: for example, to pour.

Affixation (or morphology) is obligatory for all Igbo Vs except in serial verb constructions (SVCs), where some Vs may appear without morphology. Igbo has a rich verbal morphology which it uses to express its tense and aspect system. As we have already noted in section 4.3.2, this system of affixes is inextricably linked to the Igbo VH rule. Igbo and English verbal morphology are summarized in Tables 4.2 and 4.3. It would seem from looking at the two tables that English has a rather limited verbal morphology in comparison with that of Igbo. In fact, Igbo has a host of other affixes in the form of enclitics and extensional suffixes which function as meaning modifiers. That is, they extend the meaning of the word with which they occur, such as the enclitics –cha ‘all/totally’, kwa ‘also’ and nwa ‘very self ’. We must add that the Igbo inflectional indicative affirmative rV suffix can also mark duration rather than strictly the past tense. This is why we use the term ‘indicative’ after Emenanjo (1978). In Table  4.4, we show the conjugation of the Igbo V zu. ‘to buy’, illustrating how the affixes (from Table 4.2) combine with the V to express tense, aspect and mood in the language. Table 4.4  Conjugation of zu. ‘to buy’ Present (imperative) Past Present progressive Future Present perfect Negative past Negative present progressive Negative future Negative imperative Infinitive

zu. +o. zu. +ru. na(AUX)+a+zu. ga(AUX)+a+zu. a+zu. +o.+la a+zu. +ghi. a+na+ghi. a+zu. a+ga+ghi. a+zu. a+zu. +la i.+zu.

Buy(!) Bought Is buying Will buy Has/have bought Did not buy Is not buying Will not buy Do not buy! To buy

Comparison of Aspects of Igbo and English Grammars

61

Given the intricate relationship existing between Vs in Igbo and affixes, it will be particularly interesting in later chapters to observe whether they (affixes) are able to attach to English-­origin Vs in much the same way as they do in Igbo or whether Igbo Vs can accept English affixes.

4.4.4.1  Verb serialization Another important feature of Vs in Igbo is that of verb serialization. Verb serialization is a syntactic resource which allows the speaker to express various aspects of a situation as a single cognitive package within one clause and with one predicate (Amaechi 2013: 156–7). Also, according to Aikhenvald: A SVC is a sequence of verbs which act together as a single predicate, without any overt marker of coordination, subordination or syntactic dependency of any sort. They are monoclausal; their intonation properties are the same as those of a monoverbal clause and they just have one tense, aspect, and polarity value. SVCs may also share core and other arguments. Each component of an SVC must be able to occur on its own right. Aikhenvald 2006: 1

In other words, an important feature of V serialization, then, is that the sequence of Vs shares the same S NP; they may have an intervening O between the Vs as illustrated in the example below from Igbo. (45). O  ji    aka   v o.   o. He hold  hand  weed ‘He used his hands (to) weed.’

The O of V1 is understood as the object of V2 in the clause; and both Vs share the same S in the example. The SVCs in Igbo have the following properties: ●

●

●

The two or more verbs with their complement (if any) in an SVC do not have any marker of coordination or subordination (see Amaechi 2013: 157). The VPs in the sequence are construed as occurring within the same temporal frame. Some Vs may appear with or without morphology that indicates past tense but the sentence obligatorily receives a past interpretation (see 45 above). Auxiliaries, negation, tense and aspect markers of the sequence of Vs are found with the first V of the SVC. However, extensional affixes and the open vowel suffix may be found on the other Vs in the SVC, as in (46) below. (46) O. - ga-­ra ahi.a zu. -o. efe She go-IND market buy-V dress ‘She went to (the) market and bought (a) dress.’

62

Codeswitching in Igbo–English Bilingualism

Different Igbo Vs are deployed in different types of SVCs found in the language, as we illustrate below: ●

●

●

Instrumental SVC: The Vs ji ‘hold’ and were ‘take’ are used to express instrumentality in Igbo. Both Vs are said to be syntactically similar, and occur in a complex structure [- NP VP], typical of SVCs, where they obligatorily take a complement and a VP. We see this in example (45) above, where the O of V1 aka ‘hand’ is also the instrument used to carry out the action of V2. Multi-­event SVC: In a multi-­event SVC different events which are related are formed and all the verbs share a single subject. This is illustrated in example (45) above. Dative SVC: Igbo dative SVCs indicate and distinguish the recipient of something given or transferred. They normally surface as V-V compounds, as in example (47) below. (47) O. zu. -­ta-­ra efe nye She buy-ENCL-IND dress/shirt give ‘She bought (a/the) dress/shirt (and) gave (it to) me.’

●

m 1SG.ACC

Resultative SVC: In Igbo, resultative SVCs like dative constructions also surface as V-V compounds as in (48). (48) Nze me-­re nwunye ya a-­zu. -o. efe Nze make-IND wife his V-buy-V dress ‘Nze made his wife to buy a dress.’

In the above example, V2 expresses the result of V1 and the O of V1 is regarded and understood to be the S of V2. Crucially, in relation to the analysis of the Igbo–English data, it is only in SVCs that a full Igbo V may appear bare; that is, without morphology (inflection). Therefore, if a grammar analogous to that of Igbo is in operation at the level of the VP in the bilingual data, then we will expect to see that all singly occurring EL verbs will obligatorily take Igbo verbal morphology except those in SVCs.

4.4.5  Determiners We shall define a determiner (DET/D) as a word which is typically used to determine the referential or quantificational properties of the N associated with it. In this regard, Igbo has two deictic words, six pronominal modifiers,

Comparison of Aspects of Igbo and English Grammars

63

three quantifiers and numerals which are assumed to be determiners in the language. ●

Demonstratives: One of the ways in which Igbo differs from English is with regard to how Ds are used in the language. For instance, Igbo does not have the equivalent of the English definite and indefinite articles: the, a/an. However, definiteness is realized through the use of the demonstratives a ‘this’, and ahu. ‘that’. Crucially, when these elements are used as Ds, they always follow the nouns they modify, as in the monolingual Igbo examples below: (49) (50)

U.  lo. a di. House D BE ‘This house is clean.’ U.  lo. ahu. di. House D BE ‘That/those house(s) is/are clean.’

u. cha clean u. cha clean

Furthermore, when the two Igbo demonstratives are not followed by a quantifier (Q) or a relative clause, they are always the last item in the DP. The placement rule of the two Igbo deictic words contrasts with that of English as reflected in the monolingual English translations of the two examples in (49) and (50). ●

Pronominal modifiers: there are six pronominal modifiers (which we shall label as determiners in this study, after Radford (2004: 37–40)) in Igbo (as shown in Table 4.5), and unlike their English counterparts, the Igbo ones always follow their Ns under DP.

Table 4.5  Six Igbo pronominal modifiers Person

Singular

Plural

1st 2nd 3rd

m ‘my’ gi. ‘your’ ya ‘her/his/its’

anyi. ‘our’ u. nu. ‘your’ ha ‘their’

(51) (52)

U.  lo. m House 1SG.GEN ‘My house.’ U.  lo. gi. House 2SG.GEN ‘Your house.’

Codeswitching in Igbo–English Bilingualism

64 (53) (54) (55) (56)

U.  lo. ya House 3SG.GEN ‘Her/his/its house.’ U.  lo. anyi. House 1PL.GEN ‘Our house.’ U.  lo. u. nu. House 2PL.GEN ‘Your house.’ U.  lo. ha House 3PL.GEN ‘Their house.’

In Table 4.5, we observe that, similarly to their English counterparts, the Igbo pronominal Ds can be either singular or plural and also possess the category of person (first, second and third person). ●

●

Quantifiers: only three quantifiers are recognized in Igbo: dum ‘all’, naabu. ‘two’ and niile ‘all’ (Emenanjo 1978). In the structure of the Igbo DP, the three quantifiers are used after their nominals. This contrasts with the rule in English, where they are normally pre-­posed to the N in DP. Numerals: numerals (NUMs) occur in many of the same syntactical slots as Ns (see section 4.4.1 above) in Igbo. The Igbo cardinal numbers (see Table 4.6) are found in isolation, as (S)ubject, (O)bject and second (O)bject of a verb, after the preposition, and after conjunctions.

Except for just two Igbo NUMs, otu ‘one’ and puku ‘thousand’, which precede Ns in the language, the rest follow their Ns: u. mu. ise (literal translation: ‘children

Table 4.6  Some Igbo numerals Cardinal numbers otu abu. o. ato. ano. ise isii asaa

one two three four five six seven

asato. iteghete iri iri abu. o. iri ano. nari. puku

eight nine ten twenty forty a hundred a thousand

Comparison of Aspects of Igbo and English Grammars

65

five’) ‘five children’; but otu nwata ‘one child’. The Igbo NUMs, like their English counterparts, can also stand alone in a reduced NP: for example, A-cho.-­ro. m abu. o. ‘I want two’. Thus, the typical linear order for NUMs in the Igbo DP is: N → NUM; the reverse order is usually the case in English. ●

Zero determiner: unlike English, all Igbo Ns are usually bare (i.e. without determiners) except when a definite/specific interpretation is required, in which case a demonstrative, any of the six pronominal modifiers or numerals may appear following the N. In fact, much earlier Welmers (1973: 220) observed that Igbo typically expresses generic reference by means of bare Ns, although a distinction has first to be drawn between animate and inanimate Ns. Drawing such a distinction is outside the scope of the present study. Crucially, however, there does not seem to be any direct correlation between the animacy of the N and generic reference in English. Also, in English, generic reference is not limited to bare Ns. Rather, as Quirk, Sidney and Jan (1985: 281–2) observe, the definite (the) and indefinite (a/an) articles as well as the zero determiner may all be used to express generic reference under certain conditions. Moreover, generic reading is indicated morphologically with -s on plural Ns in English.

4.4.6  Pronouns (PRNs) Unlike Ns, Igbo PRNs are marked for person and number as shown in Table 4.7. Specifically, the language has three singular, three plural and one impersonal PRNs. The three singular pronouns have each an independent and dependent form. In some contexts, the place of the PRN is occupied by what may be called Table 4.7  Igbo pronouns Independent pronouns

Dependent pronouns

1st person singular

m, mu.

I

2nd 3rd Impersonal pronoun

gi. ya

you he/she/it

1st person plural

anyi.

we

2nd

unu

you

3rd

ha

they

i/i. o/o. e/a

66

Codeswitching in Igbo–English Bilingualism

a pronominal prefix (or pronominal subject clitics), i-/i.-, o-/o.- and a-/e-, harmonizing with the verb stem vowel. In the case of the two impersonal pronouns (a-/e-), if a second PRN follows the verb, it is always either m ‘I’ or ha ‘they’, as in A-ta-ra m ya ‘I chewed it’, E-ri-­ri ha ya ‘They ate it’. Given this distinctive feature of the Igbo PRN, it will be particularly interesting to observe whether the dependent PRNs maintain the stated behaviour in the midst of EL verbs in Igbo–English CS. Generally, Igbo PRNs occur in the same syntactic positions as the Ns described in section 4.4.1 above.

4.4.7  Conjunctions Concerning conjunctions, Igbo has a small class of words that fulfil this function in the language as displayed in Table 4.8. The conjunctions can precede Ns and interrogatives or introduce clauses/ sentences.From the list in Table 4.8,Uwalaka (1997) identifies four complementizers (COMP or C) in Igbo: ka, ma, na and si. These Cs introduce various types of embedded clauses in Igbo. Uwalaka (1997: 8) observes that most of the Igbo Cs introduce more than one type of clause. Therefore, since the same C can be associated with more than one clause type, it would seem improper to say that Cs define clause types in the language. However, an interesting feature of Igbo Cs is that they can co-­occur in violation of the ‘doubly filled COMP filter’. In syntax, the ‘doubly filled COMP filter’ ensures that there is no co-­occurrence of a wh-phrase and a complementizer in a C-position. For instance, in English the avoidance of the co-­occurrence of the forms in (57) from Riemsdijk and Williams (1986): (57) I wonder who *that/*whether she saw.

Table 4.8  Igbo conjunctions ka (1) ‘like/as’ ka (2) ‘when’ ka (3) ‘(so) that’ ka (4) ‘hortative marker’ ma (1) ‘whether/if ’ ma…ma (1) ‘so…as’ ma…ma (2) ‘every…every’ ma (3) ‘but’ kwa…kwa ‘every…every’

na (1) ‘and’ na (2) ‘that’ kama/tüma ‘instead of/rather than/even if ’ tupu ‘before’ tutu ‘until’ nta (1) ‘then’ nta (2) ‘instead of ’ sï ‘that/multi-­purpose’

Comparison of Aspects of Igbo and English Grammars

67

According to the authors, examples like the one in (57) have been accounted for by assuming that C deletion must obtain so as to satisfy this filter. The introduction of the complementizer phrase (CP) necessitates an alternative formulation of this filter. Consequently, the problem in the cited example is now the obligatory deletion of C in English. However, this is not the case in some central Igbo dialects where examples like the one below come from (see also the examples in Uwalaka 1997: 4). (58) O. - ju. -­ru. si. ma unu ga-­a-bia He ask-IND  C  C   you AUX-V-come ‘He asked if/whether you will come’

Again we shall seek to uncover whether this behaviour of Igbo Cs is reflected in the bilingual Igbo–English data analysed in later chapters of this study.

4.4.8.  Sentence structure It should be apparent from the descriptions in the preceding sections that Igbo, like English, is essentially a subject–verb–object (SVO) language. Thus, the basic surface structure of a typical Igbo clause/sentence is similar to that of English as we see in (59)6 below: (59)

What the clause structure in (59) tells us is that broadly, in both languages, clauses are derived from a sequence of merger (combining) operations: (1) the verb

68

Codeswitching in Igbo–English Bilingualism

cho.ta ‘find’ is merged with the PRN ha ‘them’ to form the VP cho.ta ha ‘find them’; (2) the resulting VP is merged with the non-­finite T(tense) prefix (particle) i.- ‘to’ to form the TP i.-­cho.ta ha ‘to find them’; (3) this TP is in turn merged with the V acho. ‘trying’ to form the VP acho. i.-­cho.ta ha ‘trying to find them’; (4) the resulting VP is merged with the T AUX (prefix) na- ‘BE’ to form the T-bar na-­acho. i.-­cho.ta ha ‘is trying to find them’; (5) this T-bar is merged with its subject DP nwoke a (i.e. a merger of the nominal element and the post-­posed determiner to form the Igbo DP) ‘this man’ to form the TP nwoke a na-­acho. i.-­cho.ta ha ‘this man is trying to find them’; and (6) the resulting TP is merged with the C na ‘that’ to form the CP na nwoke a na-­acho. i.-­cho.ta ha ‘that this man is trying to find them’. Given this similarity in surface clause structure between the languages, the sentence is not the locus of our analysis per se; instead this study explores how various language elements from English are combined with elements from Igbo in mixed constituents (e.g. DP/NP, I/VP, P/PP, C/TP and so on) within bilingual clauses. We do this because, as we saw in the earlier sections, Igbo differs from English in that demonstratives, quantifiers/numerals, pronominal modifiers, and a number of the few true adjectives found in the language follow the N. Furthermore, it is important to add more information about the Igbo dependent PRNs mentioned in section 4.4.6 above (see Table 4.7). We do this because their syntactic behaviour is different from that of the other PRNs found in Igbo. According to Table 4.7, there are three dependent, short and weak PRNs in Igbo: i.-/i- ‘you (2nd sg)’, o.-/o- ‘he/she/it (3rd pl) and a-/e- ‘some person(s) (non-­ person-number specific or non-­definite)’. Each of the three weak PRNs has two forms and each form is conditioned by the [ATR] feature (see section 4.3.2 above) of an immediately following verbal element. Given the restricted subject position of occurrence of the pronominal forms and their phonological conditioning by a following verbal element, we shall analyse the dependent pronominal elements as pronominal subject clitics (CL for short). As Borer (1986) notes, clitics serve as syntactic constituents but are phonologically bound to adjacent elements of lexical categorial status. However, unlike lexical items clitics do not constitute prosodically autonomous elements and in this regard they pattern like affixes (see the descriptions of Igbo affixes in Table  4.2 above). Nevertheless, clitics unlike affixes enjoy a much larger freedom of attachment. They can attach to more than one grammatical category provided that grammatical category is located in the appropriate domain. Clitics straddle the boundaries between words and morphemes and between syntax and morphology. Their appearance can be explained in terms of syntactic dependencies, though their position follows from morphological rules. They can be attached at the beginning (proclitic) or at the

Comparison of Aspects of Igbo and English Grammars

69

end (enclitic) of their host. In addition, Zwicky (1977) explains that clitics can be either simple or special. A simple clitic belongs to the same word class as some independent word of the language that can substitute for it in a syntactic position; whereas a special clitic is not a contracted form of a self-­standing word but a form that can only occur as a bound morpheme attached to hosts in certain syntactic contexts (Katamba 1993). For example, the French singular definite article le/la is described as a simple clitic; it is realized as the proclitic l’ before hosts starting with vowels (as in l’ami ‘the friend’), but as independent words before hosts beginning with consonants (as in la parc ‘the park’). Conversely, the English genitive -’s is an example of a special clitic because it does not have a reduced form, and it never occurs without being attached to a host. The dependent, short and weak PRNs in Igbo can be viewed as falling under the category of special clitics since they appear in some special position, at the subject position before verbal elements as proclitics (see Anyanwu 2012). Some examples of the Igbo subject CLs in constructions are presented in examples (60) to (62) below: (60) (61) (62)

o.-­ta-ra 3SG.CL-chew-IND ‘He/she/it ate fish’ i.-­ta-ra 2SG.CL-chew-IND ‘You ate fish’ a-­ta-ra CL-chew-IND ‘Someone ate fish’

azu. fish azu. fish azu. fish

We shall demonstrate in later chapters that as CLs, the dependent pronominal elements can only occupy pro-argument positions in the constructions where they appear, though superficially they seem to appear at subject argument position. Also, we will take the view (after Anyanwu 2012) that the Igbo pronominal subject clitics possess the features of pro in pro-­drop languages. Since English is a language where pro licensing is not possible (see Radford 2004: 12–13), it will be interesting to uncover how Igbo–English speakers handle such parametric variation in CS.

4.5.  Conclusion The goal of this chapter is to outline some of the typological differences between Igbo and English so as to facilitate the analysis of the bilingual data from Igbo–

70

Codeswitching in Igbo–English Bilingualism

Table 4.9  Some typological contrasts between Igbo and English Categories

Igbo

English

Placement of D, Q/NUM, PRN-D in DP Placement of head N in NP Placement of A in DP/NP Ns declined for case Ns declined for number VH between V and affix (inflection) Serial verb construction Use of pronominal subject clitics Use of double complementizers

Follows the N √ √ No No Yes Yes Yes Yes

Precedes the N √ √ Yes √ No No No No

English CS in later chapters of this volume. In this regard, the most important typological differences between the two languages which we anticipate will have a bearing on the CS analyses reported in Chapters 6 to 8 are summarized in Table 4.9.7

5

Methodology 5.1  Introduction The present study progressed in five distinct stages comprising: (1) selection of the speakers; (2) data collection; (3) transcription; (4) classification of the data; and (5) data analysis. In this chapter, we shall discuss the research methodology adopted in the study; starting with a discussion of the sampling procedure and data collection strategy in sections  5.2 and 5.3 respectively. In section  5.4, we shall outline the unit of analysis. This is followed in section 5.5 by a discussion of the rationale used in the study for classifying the Igbo–English data as genuine CS forms rather than some kind of borrowings. Section 5.6 gives a summary of the bilingual data retained for the qualitative and quantitative analysis reported in Chapters 6, 7 and 8. Section 5.7 is the conclusion to the chapter.

5.2  Sampling procedure It is crucial in research of this type to make clear how the speakers were selected. Three dominant approaches inform participant selection in the field (see Nordberg 1982 and Milroy 1987). For instance, Poplack (1988) used a random sampling method in her extensive Ottawa–Hull corpus. She sampled her informants, stratified according to age and sex, from randomly selected neighbourhoods in the Ottawa–Hull region of Canada. In her study of CS in Brussels, Treffers-Daller (1994) selected her informants by means of a judgement sample. She reported that, having observed that bilinguals born in the metropolitan Brussels area switch between Dutch and French relatively frequently, she approached informants by asking people such as social workers to introduce her to ‘real Brusselers’. Another approach adopted in informant sampling is the social network method (see Backus 1996; Halmari 1997; Sankoff, Poplack and Vanniarajan 1991). Park (2000: 68) correctly observes that while

72

Codeswitching in Igbo–English Bilingualism

both the random and the judgement sample are based on isolated individuals who are considered to represent the population under investigation from a linguistic point of view, the social network method has a pre-­existing social group as its point of departure (citing Milroy 1987: 35). The goal of this study is not to offer generalizations about the language use of the entire Igbo speech community, but rather to explore a particular bilingual behaviour, that is, Igbo–English intrasentential CS, especially with regard to the structure of bilingual clauses. It would have therefore been unhelpful to adopt either the random or judgement sampling method. In view of the fact that in the present study it is essential to obtain as many instances of CS as possible, and since a number of research studies focusing on CS have already indicated that it occurs most frequently in informal conversations among in-­group members (Deuchar 2005, 2006; Eze 1998; Gumperz 1982; Park 2000; Poplack 1980), the social network method based on a pre-­existing friendship network was considered the most suitable for this study. A major advantage of this method as identified by Milroy (1987: 35) is that the researcher is able to attach him/herself to a group and, by making use of the group dynamics which influence patterns of language use, obtain very much larger amounts of spontaneous speech than is generally possible in interaction with a single individual who is isolated from his/her customary social network. Furthermore, the social network method permitted the collection of authentic natural language in interactional situations.

5.2.1  The speakers As mentioned above, the sampling was done through personal contacts among speakers who know each other well and share the same social network in Port Harcourt. Port Harcourt is the capital of Rivers State, and the metropolitan area has a population of over a million, composed of people from different parts of Nigeria (including a large Igbo population) and a significant number of foreign nationals working in the petroleum and allied industries. The city also houses Nigeria’s second largest seaport, and a busy international airport with regular links to all parts of the country and major world cities. Port Harcourt is the largest commercial and industrial centre outside Lagos. The city has three degree-­awarding universities and several higher education colleges. Through pre-­existing contacts in the city it was possible to recruit fifty (both men and women) educated adult Igbo–English bilinguals. Specifically, the speakers had to meet the following criteria:

Methodology

73

1. All speakers must have either Igbo or English as their first language. The rationale for this criterion stems from our focus in this study on characterizing the grammar of classic CS. Classic CS is defined as CS by speakers who have full proficiency in the participating language that becomes the ML, and anywhere from limited to full proficiency in the other language (Myers-Scotton 2002: 25; Myers-Scotton and Jake 2014: 2). As it turns out, all fifty informants reported Igbo to be their first language in the interviews1 (the interviews were used among other things to obtain information on the informants’ bio-­data, occupation and level of education). 2. All speakers must use Igbo and English in their routine conversations. The rationale for this criterion follows from the definition of classic CS in (1) above. 3. All speakers must have finished high school or better in Nigeria, where English is the official language of education (see section 1.4). Although Myers-Scotton (2002: 25) states that speakers need not have equal proficiency to use CS as a test for at least some aspects of grammatical structure, we wanted to ensure that the speakers were sufficiently proficient in English as to eliminate NPE and superficially bilingual speakers from our sample. Thus, all fifty speakers are educated and speak Nigeria Standard English as their second language. Thirty-­eight (or 76 per cent) out of the fifty informants are university graduates, while twelve (or 24 per cent) were university undergraduates at the time of the fieldwork for the study (Table 5.1 displays the distribution of the speakers). The conversations used in our analysis involve the fifty speakers either in group discussions (between two and four people at a time) or replies to questions

Table 5.1  Distribution of speakers according to occupation and sex Occupation

Male

Female

Total

Teachers Engineers Physicians Nurses/midwives Business owners Undergraduates Total

6 9 4 – 4 6 29

9 1 – 5 – 6 21

15 10 4 5 4 12 50

74

Codeswitching in Igbo–English Bilingualism

posed by others within the group or this researcher. Prior consent was sought and obtained before capturing all such interactions on tape. The speakers involved ranged in age from twenty to fifty-­five; according to Table  5.1, the speakers include school teachers (N = 15), engineers (N = 10), physicians (N = 4), nurses (N = 5), business owners (N = 4) and undergraduate students (12). Also, the fifty speakers were born and grew up in Nigeria and have Igbo as their first language; and were bilingual, being fluent in both Igbo and English. The second language (English) was learned at school age. Having studied in Nigeria, where English is the official language and the language of all advanced education, the informants have all been educated almost entirely in English. One of the medical doctors who took part in the study undertook advanced medical studies at the University of London in the 1970s. However, his initial medical training took place in Nigeria. He has lived and worked in the country ever since. The speakers are all middle class and some take holidays outside Nigeria at least once a year. They are either gainfully employed (including the four self-­employed business owners) or in full-­time study and reside in a multilingual city, where, as was indicated in Chapter 1 of this volume, there is intense language contact and multilingualism is the default mode for routine communication. The demographic information (see Appendix B) supplied by the informants and the researcher’s personal knowledge of these speakers confirm that they all use Igbo and English in their daily lives, and hence are Igbo–English bilinguals satisfying all the stated conditions for inclusion in the study. To ensure their anonymity, the speakers’ real names are not used in the examples. They are assigned a number (from 1 to 50) and where names appear in the examples these are pseudonyms.

5.2.2  Speakers excluded from the study Ten out of an original cohort of sixty conversationalists were not chosen as speakers and were thus excluded from the study. The utterances containing CS addressed to the ten were included in the study. However, replies from these persons were excluded entirely. Seven of the excluded conversationalists were children aged between six and thirteen years old (who were either in primary or junior secondary school at the time of the fieldwork); the remaining three were female adult non-Igbo house helps. Also, the three female house helps were NPE speakers. Consequently, these persons were excluded because they did not meet one or all of the criteria set out in section  5.2.1 above. Furthermore, the contributions from the seven children were minimal; however,

Methodology

75

their presence could not be completely avoided since all the conversations took place in the informants’ homes where these children were present during the recordings.

5.3  Data collection strategy As was mentioned in section 5.2, a number of studies on the grammatical aspects of CS still rely on data sourced from grammaticality judgements of native speakers. That is, researchers who use this approach typically ask bilingual speakers to judge the grammaticality of invented sentences containing CS. The problem with such studies is that it is possible for the judgements of two native speakers to vary with respect to a particular utterance; and as Treffers-Daller (1991) noted, such judgements tend to reflect attitudes towards language mixture rather than grammaticality or otherwise. This is mostly the case in communities where CS is stigmatized – most speakers have been known to judge ungrammatical or to deny the very utterances they have been captured on tape using. For instance, MacSwan (1999) reports that he encountered various problems when attempting to collect data through grammaticality judgements for his dissertation on Spanish–Nahuatl CS: I quickly learned from working with Jesus that he had very negative attitudes toward codeswitching. When I asked him to express judgments on particular codeswitched sentences he reacted with great discomfort. He believed that his language was losing ground among its people and the mixing of Spanish and Nahuatl was a great political disservice to the Aztec community. MacSwan 1999: 99

As a result, this Spanish–Nahuatl bilingual was excluded from the study, leading to an extremely limited sample size, according to MacSwan (1999). This method is still in use in some CS studies; some researchers employ it as a supplement to using authentic conversation data (see Couto et  al. 2015, for instance). Nevertheless, most studies on the grammar of CS continue to be based mainly on corpora of authentic everyday speech. Also, both the traditional methods of sociolinguistic interviews (Poplack 1988) and group conversations (Deuchar 2005, 2006; Stammers and Deuchar 2012; Jake et al. 2002; Myers-Scotton and Jake 2014) have been used to collect CS data. The present study is based primarily on a corpus of naturally occurring speech involving the fifty men and women described in section  5.2.1 above.

76

Codeswitching in Igbo–English Bilingualism

Specifically, the corpus consists of group conversations among members of a friendship network. From twenty minutes to one hour of tape-­recorded conversations involving each of the speakers were recorded by this researcher in the summer of 2011. It must be pointed out that this researcher did not spend every hour of each day with the speakers during the fieldwork in Port Harcourt. Since all the speakers were either employed or university students most of the recordings took place in the evenings after work and at the weekends. Also, some recordings were deleted because they contained sensitive material about the speakers and others not included in the study. As you can imagine, with the nature of naturalistic discourse one has to be careful not to infringe the rights of others. This is particularly pertinent, because at the start of the fieldwork for this study, in Port Harcourt, general elections had just been concluded in Nigeria to elect parliamentarians and the president (held on 9 and 16 April 2011 respectively). Emotions were high and most people (including the speakers) were not happy with the level of corruption among the political class, so portions of the recordings containing allegations of election rigging, financial impropriety and nepotism levelled against some well-­known Nigerian politicians had to be deleted in line with the United Kingdom Data Protection Act (1998).2 Some conversations of this type do make it into our analysis, but these have nothing in them that can identify the individual(s) being talked about. Moreover, real names have been changed to pseudonyms as we have already indicated in section 5.2.1. Therefore, the recorded conversations are composed of the speakers’ views on topics such as work, family situation, local politics and the economic situation in Nigeria, and their plans for the future. There are two main reasons for selecting group conversations for this study: (1) obtaining CS data in sufficient quantity was essential for the purpose of this study, that is, to investigate the structural characteristics of Igbo–English CS; and (2) the knowledge (from observing the group) that informal conversations among in-­group members offer the best opportunity for CS between Igbo and English. This desire to capture everyday language is the central concern of the linguist, and substantial amounts of high-­quality tape recordings of speech are needed to describe it. However, it is common knowledge that speakers will tend to shift away from their everyday normal language in situations where they are being tape-­recorded. Therefore, the very act of tape recording is likely to distort the object of investigation (see for example, Kite 1991; Labov 1972). In the next section, we will outline briefly how this problem was mitigated.

Methodology

77

5.3.1  Handling the observer’s paradox According to Milroy (1987), one of the major tasks of a fieldworker who wants to obtain reasonable quantities of good data may be seen as that of moving towards a resolution of the observer effect. In this study, all the participants are either friends or friends of friends. Also, the interview protocol (see Appendix A) was designed to elicit the speakers’ views on topics they had a common interest in, as outlined in the preceding section. In this way, the speakers paid less attention to their speech. A paradox of the communicative process is that the more relaxed the participants are in the company of the researcher, the less likely it is the transfer of personal and secret information will be inhibited. Moreover, I did not engage in any surreptitious recordings. I sought and obtained the speakers’ prior consent to capture them on tape. The conversations were recorded using a Sony M-560V micro-­cassette recorder. I was very fortunate not to encounter any objections from the speakers due to the cordial relations which I had built through extended contact with the participants both prior to and during the fieldwork phase of the study. Moreover, people in present-­day Port Harcourt are not unfamiliar with being tape-­recorded. Hence, the speakers lost self-­consciousness within seconds after I turned on the micro-­cassette recorder, which was placed on a coffee table in the middle of the room or on a side cabinet during the various conversations. The resulting corpus contains substantial examples of all the types of CS that are of interest to this study, including singly occurring EL items (see Chapters 6 and 7), as well as EL islands (see Chapter 8). Such ample occurrence of the contact phenomena under study is crucial in testing the application of the principles of the MLF model outlined earlier in section 3.2.4.

5.3.2  Transcriptions Since the focus of this study is on characterizing the structure of Igbo–English CS, only utterances containing CS were transcribed together with their immediate contexts. The transcriptions use the normal orthography of Igbo and English respectively. Also, the study limits the use of abbreviations in the glosses to only what is necessary to assist the reader’s understanding. This is illustrated in example (13a and b), repeated below for convenience. (13) a.

anyi. gaa village anyi. we go village we maka na u. fo.di. ndi.

na-­asu. asu. su. Igbo AUX-speak language Igbo bi  na obodo a-­na-ghi.

78

Codeswitching in Igbo–English Bilingualism because that some people of live PREP village V-AUX-NEG asu. asu. su. bekee … speak language English ‘When we go to the village we speak Igbo language because some of the villagers do not speak English …’ b. … mana with my family a-­na m e-use both … but with my family CL-AUX I HAB-use both ‘… but with my family I use both (Igbo and English).’ [5:1]

Moreover, as was pointed out earlier in the first chapter, in all the examples from the Igbo–English data, the switched element is highlighted in bold text, the first number within the square brackets below the examples refers to the speaker code, and the figure following the colon identifies the number of the cassette tape.

5.4  Unit of analysis As was pointed out in Chapter 3, the MLF model is a CS analytic framework that is based on the idea that there is an ML in all bilingual constituents (or what Myers-Scotton refers to as Projections of Complementizer: CP 3). The CP is the highest unit projected by lexical items; that is, the highest level in a tree of syntactic structures, and it is headed by the complementizer (C/COMP). Furthermore, Myers-Scotton (2002: 54) defines the CP as a syntactic structure expressing the predicate–argument structure of a clause, plus the additional syntactic structures needed to encode the discourse-­relevant structure and logical form of that clause. She argues that the CP is a more precise and unambiguous unit of analysis than the ‘sentence’. Thus, the MLF model is not formulated to be utilized in the analysis of monolingual CPs. Under the model, as we indicated in sections 1.2 and 3.2, an ML is the source of the abstract grammatical frame for the bilingual CP. In other words, the ML is said to include specifications at three levels of abstract grammatical structure: lexical-­conceptual structure, predicate–argument structure and morphological realization patterns (see section 3.2.3). As such, the model posits that the ML supplies specifications which form the bases on which slots are projected for permissible surface-­level morphemes in mixed (i.e. bilingual) CPs. Also, the model recognizes two types of ML, or more precisely two types of CS mechanisms. The first type, which is the concern of this study, is termed classic CS (see definition in section 3.2.4); in this type, abstract grammatical structures of mixed CPs are derived from only one of the languages participating in the

Methodology

79

CS. A convention of the MLF model adopted here is to name this ML type after the language that serves as its source. This convention is, however, not without problems. Myers-Scotton (2001: 32) has stressed that equating the ML with a language is inexact in that the ML exists only as a morphosyntactic abstraction. In contrast, languages exist as full linguistic systems when they are realized as their dialects. The other language in the bilingual CP is called the EL. As we explained earlier in section 3.2, in classic CS, the EL is claimed to be restricted by CS constraints, the SMP and MOP of the MLF model, to supplying mostly lexical material for insertion into slots in mixed CPs framed by the ML. The second type of CS mechanism is what Myers-Scotton calls ‘composite CS’. This type of CS is not the focus of the present study. However, this type of ML is a composite of abstract grammatical structure from more than one source variety in the CS contact (see Amuzu (2010, 2014) for a discussion of composite CS involving Ewe–English and other language pairs). Going back to the first paragraph of this section, we mentioned that MyersScotton (2002) uses the term ‘bilingual CP’, which she says is at the highest level in a tree of syntactic structures. She claims that other researchers have tended to use the sentence as the reference point for structural analyses with the attendant problem of possible misanalysis of sentences containing many different structural configurations (see Myers-Scotton’s account of such errors in the footnote on page 54 of her 2002 book, Contact Linguistics). Although this study uses the term intrasentential CS (in keeping with the tradition in studies exploring grammatical aspects of CS), our unit of analysis, nonetheless, is the bilingual clause. This is a clause containing one or more constituents that are mixed constituents; or a clause containing one or more EL islands that it dominates (see examples (63)–(71) below from Igbo–English). The bilingual clause as a unit of analysis is preferred over the sentence because it is the easiest to apply and allows for comparability of examples across different language pairs in CS. Moreover, the bilingual clause as a unit of analysis has also been adopted successfully in other CS studies (see for instance: Amuzu 1998, 2010, 2014: Ewe–English; Bolonyai 1999: Hungarian–English; Deuchar 2005, 2006: Welsh–English; Myers-Scotton 1993b: Swahili–English; Rahimi and Dabaghi 2013: Persian–English; and so on).

5.4.1  Classification of the Igbo–English data Under the MLF model, there are three distinguishable types of constituents within the bilingual clause:

80

Codeswitching in Igbo–English Bilingualism

1. Mixed constituent – a constituent containing morphemes from both the ML and the EL. 2. ML island – a well-­formed ML constituent that is realized as a part of a mixed constituent. 3. EL island – a well-­formed EL constituent that is realized as a part of a mixed constituent. Of these, ML islands and EL islands may constitute phrasal categories within the bilingual clause: DP/NP, VP, PP, AP and so on. It is equally important to point out that EL islands further subdivide into internal and non-­internal EL islands. According to Myers-Scotton (2006), an internal EL island is a part of a larger phrase framed by an ML element, whereas a non-­internal EL island is said to be a maximal projection entirely in the EL but dominated by an ML clause. Thus, a clause is bilingual if it is composed of at least a mixed constituent (as in Type 1 above), or an EL island (as in Type 3 above). Accordingly, this study is concerned with clause Types 1 and 3 as we illustrate in examples (3), (5), (7) and (10), repeated below as (63), (64), (65) and (66). In the examples, ‘C1’ stands for an example containing a single clause and ‘C2’ refers to an example containing two clauses; the square brackets represent clause boundaries, and the mixed constituents of interest to the analysis reported in the later chapters of this study are underlined. (63) C1[eze anyi. ga-­e-­mehee hospital ahu. ]    Chief our AUX-V-open hospital DEM/D ‘Our chief will open that hospital.’ [4:10] (64) C1[ha ku. da-­ra booth dum no.-­na  polling station ahu.  …]    They break-IND booth Q be-­at polling station  DEM/D ‘They broke all (the) booths at that polling station.’ [35:26] (65) C1[anyi. ho.pu. ta-­ra government o.jo.o. a]  C2[ma anyi. ga-­e-change  ya]    We elect-IND    government bad DEM/D  but we AUX-V-change it ‘We elected (a/this) bad government, but we will change it.’ [25:15] (66) C1[ha maa]  C2[na this government is full of corrupt people …]    They know   COMP this government is full of corrupt people ‘They know that this government is full of corrupt people.’ [6:3]

In examples (63) to (65), three English Ns, hospital, booth and government occur singly in mixed DPs/NP in which they are accompanied by Igbo early system morphemes (see sections  3.2.1 and 3.2.2): the demonstrative determiner (DEM/D) ahu. ‘that’ in (63), the quantifier (Q) dum ‘all’ in (64); the mixed NP in

Methodology

81

(65) is composed of the English N and a following Igbo A o.jo.o. ‘bad’. Also, in the same example (i.e. 65), the single English verb change (in the mixed VP: change ya ‘change it’) is inflected with the Igbo bound future auxiliary verb ga- ‘will’ and the harmonizing verbal inflectional prefix e-. These examples (63–65) fall under Type 1 as described above; this CS type is the focus of the analysis in Chapters 6 and 7. Examples (64) and (66) also from the Igbo–English data depict the two subtypes under Type 3. In (64), polling station is an English island that occurs as a part of the DP complement of the adjunct PP na polling station ahu. ‘at/in that polling station’. All the morphemes in the NP (polling station) come from English and they occur within the phrase in accordance with the grammar of English. It is, however, called an internal EL island (Myers-Scotton 2002) because it occurs as a part of a mixed DP headed by the Igbo D ahu. , and the DP is in turn a part of the PP headed by the Igbo P na; whereas the Inflectional Phrase (IP, or Tense Phrase: TP) in (66) could be described as a non-­internal EL island because the embedded clause is entirely in English. Nonetheless, the English TP is merged with the Igbo C na ‘that’ to form the CP ‘na this government is full of corrupt people’. This CS type is the subject of the analysis in Chapter  8 of this volume. The unmixed C1 in (66) can be described as an ML island. As we have already pointed out in this section, we shall not be concerned with such monolingual clauses in this study. Under the MLF model, as was pointed out in Chapter  3, the ML is the source of the frame for the CP (i.e. bilingual clause). Therefore, constituents entirely in the ML are almost ‘expected’ and do not present any problems for the model (see MyersScotton 2002: 58). However, other CS types occur in Igbo–English either as bare forms or with what has been described in the literature as ‘double morphology’ (ibid.: 91). (67) (68) (69)

ha  a-­hapu. -­la terminal ji-­ri taxi na-­a-lo.-­ta They V-leave-PERF terminal hold-IND taxi AUX-V-return-ENCL ‘They have left (the) terminal and they are returning in a taxi.’ [14:31] commissioner abu. o. ka e-me-re arrest kpo.ro. ga-­wa Abuja Commissioner two BE V-do-IND arrest take  go--ENCL Abuja ‘Two commissioners were arrested and taken to Abuja.’ [37:3] tractors ndi. ahu. e-­mebi-cha-­la Tractor-­s PL that V-damage-ENCL-PERF ‘Those tractors have been damaged completely.’ (Those tractors are beyond repair) [28:12]

82

Codeswitching in Igbo–English Bilingualism (70) m̩ ba, o-­kwu-ru that na o.-­ga-ghi. no,  CL-say-IND C  C  CL-AUX-NEG ‘No, he said that he will not go to church.’

a-­ga V-go

u. ka church [6:30]

Bare forms are defined by Myers-Scotton (2002: 113) as EL content morphemes appearing in mixed constituents framed by the ML, but they are missing the ML system morphemes that would make them well-­formed elements in such frames. Thus, according to the MLF model the lone English noun terminal in (67) and the verb arrest in (68) are bare forms because the noun occurs in the clause without a determiner from either Igbo or English, and the verb is without the necessary inflections for tense from either language. It is important to add that the model does not disallow bare forms as long as the SMP is satisfied. That is, if the clause in which the bare form appears contains late outsider system morphemes (see sections 3.2.1 and 3.2.2), then Myers-Scotton (ibid.) claims that they must come from the ML. This claim and the examples represented by (67) and (68) are explored further in sections 6.2.3 and 7.5 respectively. In (69) and (70), we notice two examples that may be described as cases of double morphology (see section 3.2.1). In (69), the plural morpheme is encoded twice, first in English tractor-­s and again in Igbo, using the Igbo optional plural morpheme ndi.. Recall that we noted earlier in section 4.4.1 that nouns in Igbo, unlike in English, are not typically declined for case or inflected for number. Therefore, we shall investigate the exact nature of the cases in Igbo–English represented by example (69) in section 6.2.4. Another example of double morphology involving the doubling of complementizers is found in (70). In this example, both languages contribute the complementizers, but the Igbo complementizer is closer to the IP/TP than the English one. Such examples are interesting for the analysis reported in this study. Therefore, we shall explore them further in section  8.3.4 of this study. Lastly, here, another interesting case warranting further exploration involves English verbs in the data that occur with the Igbo pronominal subject clitics (see section 4.4.8), as in example (71) below: (71) e-decorate-ri. hall ano. na u. lo.o.gwu. o.hu. u. ahu.  … CL-decorate-IND hall four PREP hospital new that ‘(Some person/s) They decorated four halls in that new hospital.’ [36:40]

We will discuss such examples as (71) in section 7.3. We will do so by taking the view that the Igbo pronominal subject clitics possess the features of pro in pro-­drop languages. Next, we discuss our criteria for classifying the singly occurring EL words in Igbo–English as genuine CS forms rather than some kind of borrowings.

Methodology

83

5.5  Igbo–English: Codeswitching or borrowing? An interminable debate in the CS literature among researchers interested in the grammatical study of the subject concerns how to classify other language single words embedded in the utterance of another language (see Couto 2015: 66; Poplack 2012: 644; Stammers and Deuchar 2012: 630). The problem, it seems, stems from the fact that some researchers4 analyse as CS structures the likes of which others5 reject as borrowings. Hence, it is not always apparent whether the constraints and models being proposed are generalizable to CS, borrowing, both, or neither. For instance, consider the following two examples from Igbo–English: (72) o.-­na-e-­tinye ego na account anyi. 3SG.CL-AUX-V-put money PREP account our-PRN/D ‘He puts money into our account.’ [7:30] (73) mu. na di m a-­malite-la i-save ego niile o.-­na-e-­nye anyi. Me/I and husband my V-start-PERF INF-save money all CL-AUX-V-give us ‘My husband and I have started to save all (the) money he gives (to) us.’ [8:30]

Both examples come from a married couple recounting to this researcher how their son who lives in Canada sends them money to assist with their preparations to visit him there. With respect to the analysis reported in this study, forms such as account anyi. ‘account our’ (our account) in (72) seem to violate the EC (Poplack 1980; see also section  2.2): which predicts that switching is free to occur only between sentence elements that are normally ordered in the same way by the monolingual grammars in contact. The same example also appears to violate the prediction of the FHC (Belazi et al. 1994; see also section 2.5): that the language feature of the complement f-­selected by a functional head, like all other relevant features, must match the corresponding feature of that functional head. Similarly, example (72) would appear to violate the GC (Di Sciullo et al. 1986; see section 2.4), which states that switching between the governor and the governed is disallowed. Firstly, it is obvious in (72) that the surface word order of the two languages in contact does not match, because the functional category, first person plural possessive pronominal determiner (D) anyi. ‘our’, follows the English noun (N) account instead of preceding it. Secondly, the switching is Determiner Phrase (DP)-internal; that is, the noun complement account and the determiner anyi. ‘our’ come from two different languages. Therefore, the structure of the mixed DP in (72) is unacceptable in English but resembles the typical

84

Codeswitching in Igbo–English Bilingualism

structure of an Igbo DP (which has a Complement (C)-Head (H) order; see section  4.4.5). The English well-­formedness condition for pronominal determiners as reflected in the monolingual English translation of (72) requires that the determiner should precede and not follow the nominal element. Also, i-­save ‘to save’ in (8), repeated above as (73), seems to contravene the basic assumption underlying the FMC (Poplack 1981; see also section  2.3), which states that a switch may not occur between a bound morpheme and a lexical form unless the latter has been phonologically integrated into the language of the bound morpheme. Additionally, the same example would be unacceptable as a CS form under MacSwan’s (2009, 2014) PFIC-based account (see section 2.6), which like the FMC assumes that two phonologies cannot be mixed in CS. Hence, any embedded language free morpheme bearing Matrix Language inflection(s), as in example (73), is automatically rejected as a non-CS form. However, for reasons to be explained below, we shall consider English words in Igbo to be genuine CS forms unless they are listed in an authoritative Igbo dictionary. As Deuchar (2005) and Stammers and Deuchar (2012) correctly observe, one of the reasons for drawing a distinction between switching and borrowing is that borrowed words are more likely than switches to be fully integrated into the recipient language, and some researchers wish to leave open the possibility that borrowing may be used by speakers in a different way from switches (Deuchar 2005: 256). In fact, as many researchers have reported, drawing a distinction between switches and borrowings is fraught with many problems and controversies. According to Matras (2009: 106), in the broader context of general linguistics, ‘borrowing’ usually refers to the diachronic process by which languages enhance their vocabulary, while ‘codeswitching’ is reserved for instances of spontaneous language mixing6 in the conversation of bilinguals. He states that to make such a distinction demands that specific criteria are deployed for distinguishing loanwords from single word switches. Muysken (1995: 189– 90) adds to the debate by referring to borrowing as ‘the incorporation of lexical elements from one language in the lexicon of another language’. According to him, three levels may be distinguished in the process. Firstly, a fluent bilingual spontaneously inserts lexical element l from language A into a sentence in language B. With time, the insertion of l becomes a frequent occurrence in a speech community (i.e. so-­called ‘conventionalized codeswitching’ occurs). Finally, l becomes adapted phonologically, morphologically and syntactically to the rules of language B and is fully integrated into the lexicon, being recognized as a word of language B by monolingual speakers. Under this view, the initial

Methodology

85

incorporation of such lexical elements as described by Muysken requires some level of bilingualism. Other studies have relied on frequency counts (see Jones 2005; Poplack and Sankoff 1984; Poplack, Sankoff and Miller 1988), leading to the problem of lack of comparability among them due to the absence of any uniform standard according to which a form’s frequency of occurrence could be assessed. Still others seek to solve this problem by expanding their definition of borrowings to include morphological integration, such as the addition of a source language affix (see MacSwan 2000; Poplack 2012; Poplack and Meechan 1995). By doing this, these researchers effectively ban word-­internal CS (as in example 73 above) and make their CS models unfalsifiable, if every word-­internal switch must be considered a borrowing (Legendre and Schindler 2010: 59). In fact, Stammers and Deuchar (2012: 632) point out that one controversial category repeatedly used by Poplack and her associates is that of ‘nonce borrowings’. For instance, Poplack et al. propose that lexical insertions fall into two groups: those that are structurally integrated from the onset, and those that are not. Such structural integration can come about independently of frequency of use in the speech of bilinguals. Thus, it is not necessarily the product of a prolonged, diachronic process. Poplack and her associates conclude that borrowing and CS are separate phenomena from the onset. They introduce the term nonce borrowing as a designation for on-­the-spot borrowings that are structurally integrated but have not necessarily reached a wide level of propagation within the speech community or even within a corpus. In other words, Poplack et  al. consider most single-­word switches not to be genuine switches, but to behave like borrowings (Matras 2009: 106). Furthermore, Poplack (2012) and Poplack and Meechan (1998: 129) claim that CS and borrowing differ as processes. They view CS as involving the alternation of the procedures of one language with those of another. Borrowing, they say, does not involve this alternation. That is, not only is there alternation in CS, but it results in compartmentalization of the languages involved in any resulting bilingual clause (Myers-Scotton 2002: 154). Thus, according to Poplack and Meechan (1998), since singly occurring forms like those in (72) and (73) pattern like native forms, they cannot be CS forms. We, however, disagree with this view. Instead, similar to Myers-Scotton (2002: 155), we recognize the role of abstract grammatical structure in CS. Our position is that when mixed constituents are accessed, there is necessarily interaction of the two grammars at an abstract level, even while the ML is more activated than the EL: the same abstract procedures may result in (1) monolingual Igbo discourse and (2) discourse with

Codeswitching in Igbo–English Bilingualism

86

an Igbo frame but English insertions; the two outcomes do not have the same history. The form of the bilingual outcome depends on both universal principles for bilingual clauses (for example, one language dominates in the grammatical frame) and restrictions that depend on congruence/incongruence regarding the typological characteristics of the participating languages. Moreover, this interaction differentiates CS from monolingual data. In this sense, there is, as it were, a continuum of EL elements in bilingual clauses, with single words as one end point and full phrases as the other (Myers-Scotton 2006: 254). Having stated our position in the last paragraph, we nonetheless concede that if an EL word has a dictionary entry, its status is undisputed (Myers-Scotton 2002: 41). Consequently, we agree with Deuchar (2005: 256) and Stammers and Deuchar (2012: 630) that this ‘dictionary’ criterion is a practical version of the more theoretical criterion of ‘listedness’ defined by Muysken (2000: 71) as ‘the degree to which a particular element or structure is part of a memorized list which has gained acceptance within a particular speech community’. That is, borrowings are assumed to be listed in the vocabulary of monolingual speakers of the recipient language, whereas switches are not. This criterion of knowledge by monolingual speakers can be applied to a language like Igbo, which has well over twenty million native speakers (Nigerian Census 2006) residing in SouthEastern Nigeria. Therefore, we shall assume that, for Igbo, listedness is reflected in established dictionaries of the language, in particular The Igbo Dictionary (revised and updated in 2006 by Roger Blench: Dictionary of Onicha Igbo). If an English-­origin item is found in this Igbo dictionary we shall assume that it is a borrowing, whereas if it is absent we shall assume that it is a switch. For example, a word like tebulu ‘table’ in example (74) is considered to be a borrowing since it is listed in the dictionary. (74)

o.-­di   na elu CL-BE PREP top ‘It is on top of the table…’

tebulu … table [10:5]

On the other hand, words such as account and save in (72) and (73) respectively are not considered to be borrowings because they do not appear in the dictionary. We will accept, like Deuchar (2005: 256), that this is admittedly an entirely arbitrary criterion, especially since we recognize that dictionaries reflect usage at an earlier point in time rather than the present. The main consequence of adopting this criterion for our analysis is that we may identify some single words as switches when they should be considered borrowings. However, we shall

Methodology

87

report on steps taken to counter the effects of possible errors of this kind in the relevant sections of this volume. Thus, this study examines an account of intrasentential CS in terms of the MLF model (see Chapter 3), a proposal associated with the notion of asymmetry between the languages participating in CS. We take, therefore, as our basic assumption the notion under the MLF model that the two languages (in our case Igbo and English) participating in CS do not both contribute equally to the morphosyntactic frame of a mixed constituent. That is, one language is dominant (the ML), contributing the frame-­building morphosyntactic properties; the other language (the EL) contributes certain language elements that are fitted into appropriate slots framed by the ML. In connection with this hypothesis, we propose that both multi-­word sequences, as in returning officer ahu. in (75) below, and single word insertions (as in 72 and 73 above) within the ML frame, may be accounted for by the same principles of asymmetry under the MLF model. (75) returning officer ahu. a-­bu. -­ghi. onye iberibe [PAUSE] o.- ma  ihe returning officer DEM V-BE-NEG person stupid, he knows thing ha  me-­re … they do-IND ‘That returning officer is not a stupid person, he knows what they did’ [6:3]

Accordingly, this study assesses the merit of such an account of intrasentential CS by firstly examining English-­origin single words involving nouns (Ns), adjectives (As) and verbs (Vs) (in Chapters 6 and 7 respectively); and secondly, multi-­word sequences (i.e. EL islands, in Chapter 8) attested in the Igbo–English data. The rationale for focusing on the three categories (Ns, As and Vs) for the analysis of single-­word insertions is because such EL elements have been shown in the literature to be notoriously ambiguous and are at the centre of the lack of agreement in the field. Moreover, they happen to be the most frequently occurring switched elements in this and other CS studies (see section  5.6 immediately below). Therefore, this study will among other things:

1. Show by exemplifications and quantitative analysis that make use of the typological contrasts between Igbo and English (see Chapter 4) what happens to the grammatical structures when the two languages are in contact in the same clause; 2. Reveal that there is indeed no marked distinction between multi-­word sequences and single word switches; and

88

Codeswitching in Igbo–English Bilingualism

3. Demonstrate both qualitatively and quantitatively that Igbo–English is a ‘classic case’ of CS. In doing this, we shall argue against CS frameworks: (1) that seek to prohibit switching on the grounds of surface-­level linear differences between the languages participating in CS; (2) that seek to bar CS between a functional head and its complement; and (3) that expand their definition of borrowings to include morphologically integrated single-­word insertions, thereby banning word-­internal CS between a root and an affix. Instead our study seeks to demonstrate that there are strong asymmetries between any two languages participating in CS cross-­linguistically. It is this division of labour, as we demonstrate in this study, which regulates the roles played by the two languages in contact in the same bilingual clause.

5.6  The bilingual data From the tape-­recorded conversations of the fifty Igbo–English bilingual speakers described in section 5.2.1, we identified a total of 3,026 bilingual clauses. Of this number, 77 per cent (2,330/3,026) are single-­word insertions (including Ns, As and Vs); a majority of these words occur at the syntactic boundaries where Igbo and English have different word orders, while others occur with verbal inflections from only one of the participating languages. Therefore, such structural contrasts in the data will provide ample opportunities for testing the predictions of the MLF model outlined earlier in section 3.2.4. The remaining 23 per cent (696/3,026) involve the two types of EL islands described in section 5.4.1 above. Concerning the distribution of the single-­word insertions (total = 2,330), singly occurring English Ns in otherwise Igbo utterances constitute 62 per cent (1,439/2,330), and 8 per cent (192/2,330) of the words from this lexical category are from Igbo in otherwise English utterances. Singly occurring English Vs in otherwise Igbo utterances make up 23.5 per cent (547/2,330) of the data. There is no instance in the corpus where an Igbo verb is inserted in an otherwise English utterance. This is not unusual. In fact, in Eze’s (1997: 94) Igbo–English study, he reports that the singly occurring verbs (total = 976) in his corpus are all of English origin. Singly occurring English As in otherwise Igbo utterances make up 6.5 per cent (152/2,330) of the present data sample. It is not surprising to find that singly occurring Ns constitute the most frequently switched elements in Igbo–English CS. In fact, many studies before

Methodology

89

this one have reported similar findings. For example, in Pfaff ’s (1979) study of Spanish–English CS, only 71/932 (7.6 per cent) switched elements are verbs; in contrast, 818 (87.8 per cent) switched elements are nouns. Similarly, in Poplack (1980), there are 141 nouns that are switched, as compared to thirteen verbs and thirteen verb phrases; switched nouns are 10.84 times more common than verbs in this Spanish–English corpus. In Okasha’s (1999) Arabic–English corpus, in the Generation 1 data, only twenty-­three English verbs occur, as compared to 139 singly occurring nouns; nouns are switched six times more frequently. In her Generation 2 data, verbs are even less frequent (8 verbs v. 838 nouns). TreffersDaller (1994: 3–4) points out that in her Brussels Dutch–French corpus, French nouns represent 58.4 per cent out of all lexical categories and verbs represent 8.9 per cent. Myers-Scotton (1993b) reports ninety-­one English verbs in Swahili finite clauses; 24.6 per cent of the switches; nouns represent 46.5 per cent of the switches (141 types/174 tokens). Thus, this study is not alone in finding that nouns are the most commonly switched constituent type in intrasentential CS. A chief reason often given in the CS literature for why single Ns are the most frequently switched elements in bilingual speech is that they are conveyors of semantic and pragmatic information and thus the most commonly switched items by bilinguals. According to Myers-Scotton and Jake (2014: 4), while nouns receive thematic roles, they do not supply them. Thus, when an EL noun occurs in an ML frame, the only congruence checking that is needed is that its semantic/ pragmatic features make it le mot juste from the speaker’s point of view. Also, they add that from a production point of view, EL nouns are ‘free’ in regard to grammatical checking (ibid.: 4). Therefore, it is not unexpected to observe that this fact is reflected in Igbo–English intrasentential CS. Nevertheless, to ensure a systematic comparison and avoid skewing the data, we have excluded from the analysis presented in the coming chapters names for places and people (i.e. proper nouns); as in example (76) below. (76)

o.para anyi. bi na Ottawa … o. lu. -­ru. onye Canada son our lives PREP Ottawa … he marry-IND person Canada ‘Our son lives in Ottawa. He married a Canadian.’ [8:30]

Our decision to exclude such words (as Ottawa and Canada in 76) is also based on our belief that one does not necessarily need to be a bilingual to actually use such proper nouns in their speech. Consequently, from the singly occurring English nouns we have retained 1,264 words that occur in otherwise Igbo utterances for the analysis reported in Chapter 6. As it turns out, all the singly occurring Igbo nouns (total: 192) in otherwise English utterances are proper Ns

90

Codeswitching in Igbo–English Bilingualism

(as in the examples below), and are thus excluded from the analysis reported in this study. (77) Speaker 10: … they were looking to post him to Enugu, but he decided to move to Owere … [10:5] Speaker 5: … that’s so much better. Owere is very close to Port Harcourt … [5:5] Speaker 10: … but Nneka and Obiora will see their dad now only on weekends. [10:5] Speaker 5: Don’t worry Amaka … [5:5]

No singly occurring Igbo Vs or As are found in otherwise English utterances in the Igbo–English data. Therefore, all the singly occurring English Vs (total: 547) and As (total: 152) in otherwise Igbo utterances are retained for the analysis reported in Chapters 6 and 7. This brings the combined total of all singly occurring EL words retained for the analysis to 1,963 (or 74 per cent: 1,963/2,659); of the 696 EL islands, 264 (or 10 per cent: 264/2,659) occur as internal EL islands, while the remaining 432 (or 16 per cent: 432/2,659) are non-­internal EL islands.

5.7  Conclusion This chapter informs about the speakers, the data collection and transcription procedures, the unit of analysis and the classification of the Igbo–English data. It also makes clear the criterion used in the study to distinguish CS forms from potential borrowings. Next, we shall test the application of the three principles of the MLF model outlined in section 3.2.4 to Igbo–English data. This will involve exemplification and illustration of the principles in more detail, followed by the results of a quantitative analysis relating to each principle. Broadly, we shall demonstrate that the principles do appear to be reflected in our data, and that they do indeed appear to represent a case of ‘classic CS’. Nevertheless, we shall highlight that there are some problems of definition and a few problematic examples, which will be discussed in the relevant sections of this study.

6

Embedded Language Single Words: Nouns and Adjectives 6.1  Introduction Using the outlined typological differences inherent in the grammars of Igbo and English as our backdrop (see Chapter 4), we shall in the coming sections of this chapter evaluate the plausibility of the ML hypothesis: the notion that in classic CS the two languages participating in CS do not both contribute equally to the morphosyntactic frame of a mixed constituent. That is, one language is dominant (the ML), contributing the frame-­building morphosyntactic properties; the other language (the EL) contributes certain lexical items that are fitted into appropriate slots framed by the ML. If the hypothesis holds, it will be possible to identify an ML in all bilingual clauses in the Igbo–English data. As we indicated in section 3.2.4.1, two specific criteria will be employed to identify the ML of each bilingual clause: (1) the morpheme order criterion; and (2) the source of a particular type of system morpheme. These two criteria follow from two additional principles (the Asymmetry Principle – AP and the Uniform Structure Principle – USP) (see sections 3.2.4.2 and 3.2.4.3) to be outlined and tested in sections 6.3.2 and 6.3.3 below. In our analysis, we shall not consider the MLP to be falsified if morpheme order is compatible with both languages. When this happens, we will interpret it to mean that there is no word order conflict between the two languages concerning the mixed constituent. In which case, we will determine the ML of the bilingual clause according to the source language of the outsider late system morpheme criterion only (see section 6.2.2 for our definition of this particular system morpheme in relation to the analysis of the Igbo– English data). The expectation is that only one of the languages (either Igbo or English) will supply this particular type of system morpheme, not both. If the outsider late system morphemes come from both languages in the same bilingual clause, then the MLP could be falsified (see also Deuchar 2006 for the parameters adopted in her Welsh–English CS study).

92

Codeswitching in Igbo–English Bilingualism

We shall now test the application of the MLP on Igbo–English data involving singly occurring EL nouns and adjectives in sections 6.2 to 6.2.2. In section 6.2.3, we shall discuss the EL nouns that occur without any determiners either from Igbo or English. Section  6.2.4 discusses examples involving what might be termed double morphology (Myers-Scotton 2002: 91). This is followed in section 6.3 by the results of a quantitative analysis relating to each of the three principles (MLP, AP and USP) outlined in section  3.2.4. In the final section, section 6.4, we shall summarize the main findings from this chapter in relation to both the CS models reviewed in Chapter 2 and the principles of the MLF model tested here. Broadly, the analysis in this chapter will show that the principles of the MLF model do appear to be upheld in Igbo–English CS, and that the attested examples do indeed seem to represent a case of classic CS. However, we shall highlight problems of definition and discuss any problematic examples in the relevant sections below.

6.2  Testing the Matrix Language Principle 6.2.1  The morpheme order criterion The morpheme order criterion follows from the Morpheme Order Principle (MOP), which predicts that in ML+EL constituents consisting of singly occurring EL lexemes and any number of ML morphemes, surface morpheme order (reflecting surface syntactic relations) will be that of the ML (MyersScotton 2002: 59). To operationalize the morpheme order criterion we will interpret it to mean that it will apply wherever there is a conflict in word order between the two languages participating in CS. Although, as we saw in section 4.4.8, Igbo and English typically have the same surface word order for the most frequently occurring clause type (i.e. declarative clause), they, however, differ with reference to the linear order within the internal structure of their respective NP/DP. Specifically, Igbo and English differ in the relative order of head (H) and complement (C) within the NP/DP. The usual order in Igbo is C followed by H rather than the H – C order of English. To illustrate this difference in the configuration of the NP/DP in both languages, consider the monolingual Igbo sentences below. (78) Ha bi na [DP[NP[Nu. lo.] [Ao.hu. u. ]] They live in house new ‘They live in [DP[Dthat] [NP[Anew] [N house]]]’

[Dahu. ]] that

Language Single Words: Nouns and Adjectives

93

In (78), we observe that within the Igbo DP both the adjective (A) and determiner (D) are typically post-­posed to the nominal element (N). The reverse order is usually the case in English. Also, in Igbo, a noun (N) can follow another noun to form a genitival construction (see section 4.4.1.1), as in (35) repeated below as (79). (79) o.- na-­a-­gba [NP[Nigwe] [NKanye] He/she/it AUX-V-ride      bicycle Kanye ‘He is riding Kanye’s bicycle.’

The situation in (79) is different from that of a language like English, where usually only the N in the genitive case is inflected. Igbo Ns are neither declined for case nor inflected for number like those of English (see section  4.4.1). Therefore, in constructions like (79), it is the genitival N which comes second in the Igbo NP (see also Emenanjo 1978; Uwalaka 1997). The examples present possible conflict sites, where the grammars of both languages have conflicting rules. According to Myers-Scotton (2006: 255), in classic CS, all structural conflicts are resolved in favour of one of the participating languages identified as the ML. Therefore, our focus on mixed nominal expressions in Igbo–English CS seems appropriate for testing the veracity of the basic assumption underpinning the MLF model. Table 6.1 summarizes the typical NP/DP word order configurations in both Igbo and English. Table 6.1 and the examples in (78) and (79) clearly show that the class of items that can modify the noun in Igbo tend to follow rather than precede this lexical category; whereas the reverse is usually the case in English. It is equally important to point out that in Igbo these elements may co-­occur with some degree of flexibility in the word order that co-­relates with little or no changes in meaning (see Obiamalu 2013a: 54). The examples in (80a and b) are well-­formed structures in Igbo but not in English. (80) a. b.

n̩ nu. nu. ahu. bird DEM ‘all those birds’ n̩ nu. nu. niile bird Q ‘all those birds’

niile Q ahu. DEM

The above examples show that a demonstrative and a quantifier can exchange position without triggering any change in meaning. According to Radford (2004) and Radford et  al. (2009: 311), the most obvious manifestation of structural

*NUM

*AN

NP (posm)

+N/PRN pos-NP

A

PRN/D

NUM (ORD/CARD)

Q

DEM/D

Q

RC

English

 

Q

DET (posm-NP)

NUM (ORD/CARD)

Q

A

+N/PRN

RC

1. *NUM = only two numerals may precede the noun in Igbo (otu ‘one’ and puku ‘thousand’; see section 4.4.5). These do not occur in our corpus. 2. *AN = adjectival nouns may pre-/post-­modify the noun in Igbo (see Emenanjo 1978: 47–8, and section 4.4.2 this volume). 3. PP = preposition, always precedes its NP complement in both Igbo and English. 4. +N/PRN = as in English, either the noun or personal pronoun may function as the subject, object of a sentence or complement of a verb in a sentence. 5. A = all true adjectives in Igbo follow their nouns (see section 4.4.2); whereas the reverse order is the case in English. 6. PRN/D = pronominal determiners (see section 4.4.5) always follow the noun in Igbo (the reverse is the case in English). 7. NUM = all Igbo numerals except the two mentioned in (1) always follow the noun; the reverse obtains in English. 8. Q = quantifiers follow their nouns in Igbo, but precede them in English. 9. DEM = demonstratives always follow their nouns and are always last in the DP when not followed by an Igbo quantifier. These elements precede the noun in English. However, the functional category D may appear null in both languages. RC = the relative clause follows the noun in both Igbo and English.

Notes:

Igbo

Table 6.1  Configuration of a basic Igbo and English NP/DP

Language Single Words: Nouns and Adjectives

95

variation in syntax lies in word-­order differences. The authors add that, if we suppose that the theory of Universal Grammar (UG) incorporated into the language faculty provides human beings with a genetically transmitted template for syntactic structure (so that clauses are universally CP+TP+VP structures, and nominal expressions are universally DPs), we should expect to find that word-order differences are attributable to differences in the movement operations which apply within a given type of structure (Radford et  al. 2009: 311). Thus, by adopting the DP hypothesis (after Abney 1987) for the Igbo nominal expressions which assumes that the NP is headed by a functional element, we agree with Obiamalu (2013a) that the structures where the N/NP precedes the D (as in 78–80) seem problematic for a theory that assumes that the functional head is higher in the structure and has scope over the NP which it c-­commands. Under this view, heads are said to always precede their associated complement position, even though the surface word order in some languages may be Head – Complement (e.g. English) and in some others like Igbo Complement – Head. Consequently, in languages like Igbo, the complement (C) is said to undergo left adjunction to the Specifier (Spec) position (see Obiamalu 2013a: 56). The claim is that the universal ordering between a head and its dependents is Spec-Head-Complement, as represented in (81) below. (81)

If so, the Igbo nominal expressions which have a C–H order could be analysed as having the structure in (82) below. (82)

The structure in (82) says that the nominal expressions are headed by a functional head that takes an NP as C. The NP complement moves to the Spec position in surface syntax giving rise to the C–H order. For example, the Igbo nominal

Codeswitching in Igbo–English Bilingualism

96

expression nwoke a ‘man this’ (this man), which we saw earlier in (59), will have the structure in (83): (83)

In (83), the N nwoke is shown to move into its surface position where it appears before the DEM/D (a ‘this’); thus creating two possibilities: the N head could move to the head of the functional category in a head to head movement or the NP could move to the Spec position of the functional projection (FP). Given that there is no agreement morphology between the N and the associated functional category in Igbo, we assume the latter structure for Igbo as illustrated in (82) and (83) after Obiamalu (2013a). (84)

Also, for the purpose of the analysis reported in this chapter, we shall assume after Obiamalu (2013a: 57) that in addition to the projection of a functional category D, which in most instances has a null (or zero) realization in Igbo, more than one functional projection could be found within the Igbo nominal phrase. This is so, because the functional elements can co-­occur, as we showed in examples (80a and b) above. For instance, consider the structure in (84) above. The structure in (84) shows that there is no hierarchical order between the DemP and QP in Igbo. As we have already shown in (80a and b) above, either element can dominate the other without effecting any change in meaning. Thus, in (84),

Language Single Words: Nouns and Adjectives

97

by the movement of the NP to the Spec of the available functional projections, the noun is said to have access to the features of the functional heads in a Spec– head relationship. That is, the Q niile assigns the feature [+ plural] to the N, while the DEM ahu. assigns the features [+ definite + specific]. As Obiamalu (2013a: 58) observes, this could explain why the NP n̩ nu. nu. ‘bird’ is interpretable as both plural and definite without the need for any overt plural marker (see also section 4.4.1 in this volume). Consequently, if Igbo and not English is the ML in our data, we would expect the vast majority of NPs/DPs overtly including both Igbo and English elements to have patterns of noun modification similar to those in examples (78) to (80) above. In the sections that follow immediately below, we shall illustrate how the identified contrasts are reflected in Igbo–English.

6.2.1.1  Distribution of determiners in mixed nominal expressions As we pointed out earlier in section 4.4.5, one of the ways in which Igbo differs from English is with regard to how Ds are used in the language. For instance, Igbo does not have the equivalent of the English definite and indefinite articles: the, a/an. However, definiteness can be realized by using the two deictic words: a ‘this’, and ahu. ‘that’. Crucially, as we showed in section 6.2.1 above, when these elements are used as Ds, they always follow the nouns they modify. This is unlike English, where the articles ‘the’ and ‘a/an’ are in complementary distribution with demonstratives – ‘this’, ‘that’ – and quantifiers – ‘all’, ‘every’ and so on. Importantly, in English, all the mentioned determiners precede the nouns they modify. Given this contrast between Igbo and English, consider the examples below from Igbo– English CS: (85) ma ceremony ahu. fu-­ru n̩ nukwu ego but ceremony D cost-IND big money ‘but that ceremony cost a lot of money’ [29:13] (86) election afo. a a-­di.-ghi. mfe mao.li. election year D V-BE-NEG easy at all ‘This year’s election is not easy at all.’

[27:11]

(87)

[28:12]

i.-ma-­kwa na stew ahu. a-gba-la u. ka? CL-know-ENCL C stew D V-go-PERF off ‘Do you know that that stew has gone off?’

In the three mixed DPs, ceremony ahu. in (85), election afo. a in (86) and stew ahu. in (87), we observe that the two Igbo deictic words are post-­posed to

Codeswitching in Igbo–English Bilingualism

98

their respective singly occurring English nominal complements. Notice also that the Igbo noun afo. ‘year’ in (86) is not inflected for case as is its English counterpart in the monolingual English translation. This behaviour of the noun in the bilingual example is in line with Igbo grammar (see section 4.4.1). Generally, grammatical relations of Igbo nouns are expressed by word order (and tone). Also, consider the examples in (88) to (90) below from Igbo– English: (88)

maka father-­in-law m a-­nwu. o.-­la … because father-­in-law PRN/D V-die-PERF ‘because my father-­in-law has died …’

(89) 

u. mu. nwanyi. niile ga-­e-­yi uniform ha people of women all AUV-V-wear uniform PRN/D na olili nna di m PREP funeral father husband my ‘All the women will wear their uniform at the funeral of my father-­in-law.’ [33:16]

(90)

ma campaign ya a-­di.-ra-­ghi. very effective but campaign PRN/D V-BE-IND-NEG very effective ‘but his campaign was not very effective.’ [31:15]

[33:16]

In examples (88) to (90), all the pronominal D heads like the two deictic words earlier (in 85–87) are post-­posed to their singly occurring English noun complements. This configuration is in sharp contrast with the requirements of English grammar, where the functional elements are supposed to precede their nouns in DP. Example (90) contains the EL island very effective, which is the subject of discussion in Chapter 8. Again we observe that the pattern in (85) to (90) is reflected in (91) and (92) below from Igbo–English involving the placement of quantifiers in mixed nominal expressions. (91) ha ku. da-­ra booth dum no.-­na polling station ahu. they break-IND booth Q be-­at polling station DEM/D ‘They broke all (the) booths at that polling station.’ [35:26] (92) o. nani. ya nwete-­re contract niile i.-ru. u. zo. he only him receive-IND contract Q INF-build road si. Port Harcourt ga Owere from Port Harcourt go Owere ‘He alone received all the contracts to repair the road that goes from Port Harcourt to Owere.’ [50:45]

Language Single Words: Nouns and Adjectives

99

(93) a-­si. na e-­mechie-la university niile maka … CL-say that CL-close-PERF university all because of ‘They said that they have closed all the universities because of .…’ [11:28]

In section 4.4.5, we noted that only three quantifiers are recognized in Igbo, dum ‘all’, naabu. ‘two’ and niile ‘all’ (Emenanjo 1978). In the structure of the Igbo DP, the three quantifiers are used after their nominals (see section 6.2.1 above). This contrasts with the rule in English, where quantifiers are normally pre-­posed to the noun within DP. Of the three quantifiers, two are found with lone English nouns in the Igbo–English CS corpus, as in (5), repeated above as (91), (92) and (93). An interesting observation concerning the three singly occurring English nouns (booth, contract and university) is that they occur in the mixed nominal expressions without the usual plural inflections. It would seem that the presence of the Igbo quantifiers in the mixed DPs strips the English nouns of their plural inflections. Recall from section 4.4.1 that we pointed out that Igbo nouns are not usually inflected for number or declined for case, as is the case in English. We will discuss this further in section 6.2.4 below. It will suffice for the moment to observe that the mixed nominal expressions seem to be well formed only in one language (Igbo) and not in the other (English). This claim is further supported by the bilingual examples below involving the use of numerals as nominal modifiers in Igbo–English. (94) ha ga-­a-no. week abu. o. they AUX-V-stay week NUM/D ‘They will stay two weeks in Abuja.’

na PREP

Abuja Abuja [32:40]

(95) e-decorate-ri. hall ano. na u. lo.o.gwu. o.hu. u. ahu. CL-decorate-IND hall four PREP hospital new that ‘They decorated four halls in that new hospital.’ [36:40]

Recall from section 4.4.5 and Table 6.1 that except for two specific Igbo numerals (otu, ‘one’ and puku ‘thousand’; the two numerals do not occur in our corpus) which precede the nouns they modify under DP, the typical linear order for numerals in Igbo is: Noun → Numeral. The reverse order is usually the case in English, as we notice in the monolingual translations of (2) and (71), repeated above as examples (94) and (95) respectively. Similar to the examples in (91) to (93), when an Igbo numeral is used together with an English noun (as in 94 and 95), the noun loses its plural marker. This is against the requirements of the grammar of English, as we have already noted. Stemming from the preceding exemplifications, in a bilingual DP overtly including elements from both Igbo

100

Codeswitching in Igbo–English Bilingualism

and English where the D head follows the nominal complement, we would identify Igbo as the ML according to the morpheme order criterion.

6.2.1.2  Nouns in genitive constructions In sections 4.4.1.1 and 6.2.1, we noted that in both Igbo and English a noun can follow another noun to form a genitival relationship. However, in Igbo, always, it is the genitival noun which comes second as in (79) above. We note that English also allows an analytic type genitive (‘the N2 of N1’) alongside the synthetic type (‘N1’s N2’); still the structure of genitive constructions differs markedly in both languages as we will outline below. Our discussion is however constrained by the type of structures that occur in our Igbo–English data. Therefore, we will limit our discussion to only those cases where both the analytic (of-construction) and synthetic (N1’s N2) constructions are possible alternatives in English. Consequently, cases of ‘post-­genitive’ (96), ‘local genitive’ (97) and ‘independent genitive’ (98) (Kreyer 2003: 170) will not be considered because they do not occur in our data corpus.1 (96) a friend of hers *hers’ friend2 (97) we meet at Bill’s *we meet at of Bill (98) his car is faster than John’s *his car is faster than of John (Kreyer 2003: 170)

Examples (6), repeated below as (99), and (100) represent the types of genitive strucutres that occur in Igbo–English CS. (99)

a-­si na a-­ga-­e-­me wedding Ngozi ma CL-said C CL-AUX-V-do wedding Ngozi when o.-­gbake-e CL-recover-IMP ‘They said that they will hold Ngozi’s wedding when she recovers.’ [42:40]

(100) ma ha gba-­ra u. lo. vice-­chancellor o.ku. na the riots but they burn–IND house vice-­chancellor fire PREP the riots ‘but they burned (the) vice-­chancellor’s house in the riots’ [11:28]

Notice immediately that the situation in the bilingual examples is different from that of a language like English, where typically only the noun in the genitive case is inflected (as can be seen in the monolingual English translations). In

Language Single Words: Nouns and Adjectives

101

Igbo, the preceding noun is said to be in a pre-­genitival position, while the second noun is the possessor as in (99) and (100). As we noted earlier in this section, English also allows an analytic type genitive (e.g. the mixed NPs could become the wedding of Ngozi and the house of the vice-­chancellor respectively) alongside the synthetic type (Ngozi’s wedding and the vice-­chancellor’s house); nevertheless, Igbo and English differ in the following ways. Firstly, in the English synthetic type genitive construction, the genitive marker ‘s’ is a D which mediates (or a type of bridge system morpheme, according to Myers-Scotton 2002) between the two NPs, ‘Ngozi’ and ‘wedding’ in (99) for instance. Thus, the DP ‘Ngozi’ is the subject of the higher DP which is headed by the genitive marker ‘s’ and takes the NP ‘wedding’ as its complement. A similar explanation can also account for the analytic type of-construction of the same example. Conversely, no such overt bridge system morphemes are used in either the monolingual Igbo genitive construction in (79) or in the bilingual examples above. Secondly, whereas English allows cases where the genitive (‘N1’s N2) can be substituted by the of-construction (the N2 of N1), and vice versa (see Quirk et al. 1985: 1277), in Igbo, always, the genitival noun comes second. Thus, (101a) is ungrammatical in Igbo, but (101b as well as 99 and 100) conform to the requirements of Igbo grammar. (101) a. b.

*Ngozi n̩ risa *Ngozi wedding (feast) ‘Ngozi’s wedding’ n̩ risa Ngozi wedding Ngozi ‘Ngozi’s wedding’

Thirdly, the English of-construction requires not only the use of the bridge late system morpheme of (see section  3.2.2), but also the definite article the (an early system morpheme; see section  3.2.2), which is always pre-­posed to the nominal element. This contrasts sharply with the situation in Igbo, where genitive constructions do not include the use of overt determiners; if determiners are used at all, they are always post-­posed to the nominal elements. Therefore, to explain the acceptability of (101b) over the versions in (101a), it is important to note that the genitive is used for possessive and associative constructions in Igbo. In this language, as we have already stated in sections 4.4.1.1 and 6.2.1, both the possessive and the associative constructions appear as N+N constructions where N2 always modifies N1 as in (102) a and b from Obiamalu (2013a: 58).

102 (102) a. b.

Codeswitching in Igbo–English Bilingualism àla Īkē land Ike ‘Ike’s land’ àlà ikē land strength ‘hard surface’

In (102a), we notice a genitive construction involving possession (ownership), similar to the bilingual example in (100), where N1 u. lo. ‘house’ is the possessum and N2 vice-­chancellor is the possessor. The example in (102b), like the bilingual example in (99), represents an associative construction, where NP 1 wedding is modified by NP 2 Ngozi (proper noun). Obiamalu observes that the distinction between the two types of constructions is sometimes viewed as a distinction between alienable and inalienable possession. Furthermore, the two types of constructions have different tone patterns as shown in the monolingual Igbo examples in (102a and b) above. However, a commonality between the two types of constructions is that a Step Tone (ST) is found somewhere in the four syllables. Stemming from this, he posits that ST marks genitive construction in Igbo. He suggests a Genitive Phrase (GenP) in Igbo and the ST as its morphological spell-­out (after Cinque 2003 and Kayne 1994 among others, who assume that an abstract head mediates the relationship between the two arguments.). In Igbo, therefore, he argues that the genitive marker, in the form of an ST, is the mediator between the two arguments and establishes an R-relation between them. As a result, the ST could be construed as occupying the Gen head position, while the D head is null (Obiamalu 2013a: 58–9). Therefore, both the monolingual Igbo examples in (102) and the bilingual structures in (99) and (100), which resemble the structures in (102a and b), could be analysed as in (103) below: (103)

Language Single Words: Nouns and Adjectives

103

Following Obiamalu’s (2013a) position, we can say that in (103) the ST in the Gen head position links the possessor DP to its possessum DP complement. The movement of the DP possessum to the Spec of the higher DP spreads the effect of the ST to the other syllables producing the surface structure in (103). In fact, such an analysis provides the rationale for positing a separate D head apart from DEM and Q in (84) above.3 Crucially, in view of all the above-­identified distinctions between the structures of genitive constructions in both languages, we would identify Igbo as the ML in N/NP+N/NP constructions where the first N/NP is modified by the second N/NP.

6.2.1.3  Adjectives in Igbo–English codeswitching Recall from sections 4.4.2 and 6.2.1 that we pointed out that the elements that function as true adjectives in Igbo have two very important characteristics: (1) they occur only after the nouns they modify; and (2) they occur only in attributive contexts. The other non-­basic forms (i.e. adjectival nouns) are the only elements that can pre- or post-­modify nouns in the language. Quite dissimilar to the Igbo true adjectives, only the adjectival nouns can be preceded by the Igbo copulas di./ no. ‘BE’, verbs that can only be followed by a predicate (adjectival) nominal and PP respectively (see Emenanjo 1978; Maduka-Durunze 1990; Nwachukwu 1984; Welmers 1973). Therefore, it can be said that the Igbo adjectival nouns behave like the adjectives found in English. However, English has a wide range of true adjectives with distinctive syntactic behaviour from those found in Igbo. Crucially, English adjectives can occur as heads in adjectival phrases which function as pre-­head modifiers under NP and as head of phrases functioning as predicative complements in clause structure. Put more succinctly, Igbo true adjectives follow the nouns they modify, while English true adjectives precede their nouns under NP. Nevertheless, the predicative position is structurally compatible in both Igbo and English. In this position, adjectives follow a verbal element in the two languages. We can illustrate how the morpheme order criterion would apply to Igbo–English in the following examples: (104) o.-­zu. ta-­ra m jacket o.hu. u. mgbe o.-­ga-ra Dubai CL-buy-IND me jacket new when CL-travel-IND Dubai ‘He bought me (a/the) new jacket when he travelled (to) Dubai.’ [21:19] (105) anyi. ho.pu. ta-­ra government o.jo.o. a [PAUSE] ma anyi. we elect-IND government bad DEM/D but we ga-­e-change ya AUX-V-change it ‘We elected this bad government, but we will change it.’ [25:15]

104

Codeswitching in Igbo–English Bilingualism

(106) a-­ga m e-­yi akwa white na aki.si. red … CL-AUX I V-wear dress white CONJ headscarf red ‘I will wear (a) white dress and red headscarf …’ [11:28] (107) anyi. niile zu. -­ru. agbai we all buy-IND shoes ‘We all bought white shoes because of …’

white white

maka … because of [42:37]

(108) o.-­ga-­e-­nye ya ego transport mgbe o.-­ga-a-­pu. -­wa CL-AUX-V-give her money transport when CL-AUX-V-leave-INCP ‘He will give her transport money when she will start to leave.’ [3:7]

In (104) and (105), we notice that the two Igbo true (attributive) adjectives o.hu. u. ‘new’ and o.jo.o. ‘bad’ are post-­posed to their English-­origin nouns. This is in sharp contrast with the situation in English, as can be seen in the monolingual translations, where the order of the attributive adjectives is reversed. It is equally important to observe that in the monolingual English translations, an overt determiner is required for the DPs to be well formed in this language but not in Igbo. We will discuss indeterminate DPs further in section  6.2.3 below. In examples (106) and (107), observe that the English attributive adjectives4 are in post-­position to their Igbo nouns. We have already outlined that Igbo itself has few attributive adjectives; however, when they occur Igbo grammar uses a head-­ modifier (i.e. head N+A modifier order) structure for these adjectives (as we see in examples 104 and 105). The reverse order is seen in the monolingual English translations. Also, observe that in accordance with the requirements of Igbo grammar, the noun agbai ‘shoe/s’ in (107) is not inflected for number in the bilingual NP. We saw a similar situation earlier with the singly occurring English nouns in the mixed DPs in examples (91) to (95). The plural morpheme is described as an early system morpheme in the MLF model; and one that can be contributed by both the ML and the EL. Yet, it would seem that Igbo grammar requirements block the appearance of this morpheme type in the bilingual DPs. However, we shall show in subsequent sections that there are instances in Igbo– English when the plural marker is present in some mixed constituents. We will discuss this further in section  6.2.4 below. The example in (108) involves an Igbo noun in association with an English-­origin noun, but the order is that of Igbo as in examples (104) to (107) above. Therefore, in mixed NPs where attributive adjectives follow their head nouns, we would identify Igbo as the ML according to the morpheme order criterion. Recall that we have already stated that the Igbo adjectival nouns (as they are referred to by the Igbo grammarians Emenanjo (1978) and Maduka-Durunze

Language Single Words: Nouns and Adjectives

105

(1990)) may appear either post-­posed or pre-­posed to the nouns they modify. Therefore, it is unsurprising to find that a few of the singly occurring English nouns in the data are modified by pre-­posed Igbo adjectival nouns, as in (109) below. (109) obodo anyi. no. country our BE ‘Our country is in big trouble.’

na PREP

n̩ nukwu

trouble … big/bigness trouble [23:39]

Igbo → [NP [N n̩ nukwu (big/bigness)] [N n̩ sogbu (trouble)]] or [NP [N n̩ sogbu (trouble)] [N n̩ nukwu (big/bigness) ]]

At first glance, the pre-­posed Igbo adjectival N n̩ nukwu ‘big/bigness’ in example (109) appears to pose a problem for the morpheme order criterion; however, it is important to point out that the Igbo word n̩ nukwu is described by Emenanjo (1978: 47–8) and Maduka-Durunze (1990: 237) as a ‘qualifactive noun’. These Igbo grammarians argue that the Igbo true adjectives occur only post-­nominally, as in (104) to (108) above. Notably, while the Igbo qualifactive nouns functioning as adjectives can occur either pre-­nominally or post-­nominally (see the illustration immediately below example 109), in English, adjectives typically occur pre-­nominally within NP/DP. According to Maduka-Durunze (1990: 239), when the Igbo qualifactive nouns precede their head nouns they become emphatic in their descriptive meaning or ambiguously suggest an inherent as opposed to a descriptive meaning. Nevertheless, we can submit that when Igbo nouns are used as adjectives, as in (109), they behave like the adjectives found in English which typically occur pre-­nominally. It is also significant to observe that the NP n̩ nukwu trouble is the complement of the PP headed by the Igbo preposition na. This observation confirms our earlier claim that the Igbo copula no. can only take a PP as its complement. Since PPs take NP complements, it is then possible for the Igbo non-­basic adjectives to occur in such NPs following the copular verb. We shall comment more on the Igbo copula verbs later. Crucially, since the surface word order of the mixed NP in (109) is compatible with that of both languages, we have coded all instances in the data corpus represented by this example as ‘either’ according to the morpheme order criterion. Recall from section 6.1 that we stated that if the morpheme order is ‘either’ with respect to the switched element(s), then we would use the source language of a particular type of outsider late system morphemes to determine the ML of the clause. If the relevant system morphemes come from both languages, then the MLP could be refuted. With this in mind, we shall discuss example (109) and the others more in section 6.3 below. However, it is important to highlight

106

Codeswitching in Igbo–English Bilingualism

that, as we predicted in section 4.4.2, the vast majority of the singly occurring English adjectives occur in the predicative position following the Igbo existential copula dï ‘BE’ (see Table 6.2). The predicative position is structurally compatible in both Igbo and English. In this position, adjectives follow a verbal element in both languages. Some of the examples from Igbo–English illustrating such cases are found below in (110) to (112): (110) agwa ya di. childish i.-gwa behaviour her BE childish INF-tell ‘Her behaviour is childish, to tell you the truth.’ (111) o.-­di.-ghi. CL-BE-NEG ‘It is not nice.’

gi. you

eziokwu truth [2:30]

nice nice

(112) ma di. ya a-­di.-ghi. but husband her V-BE-NEG ‘but her husband is not happy that …’

[37:30] happy na … happy C [41:30]

The analysis here is that: (1) once the lemma supporting nice in (110), for example, is selected by the speaker; (2) it is checked for congruence (paradigmatic congruence: the existence of the category of adjective in Igbo; and syntagmatic congruence: similarity in the placement of predicative adjectives in both languages) with its Igbo counterpart m̩ ma (an adjectival noun) at the three levels of abstract lexical structure (see section 3.2.3) and is deemed to be sufficiently congruent with its Igbo counterpart; then (3) grammatical structure intended for m̩ ma is built for nice; and (4) nice is accordingly inserted into that structure following the application of the morpheme order criterion. Since Igbo and English are typically SVO languages, there is no conflict in word order between the two languages in these bilingual clauses as far as the switched elements are concerned. However, as we shall observe in later sections, we would be able to identify a ML by appealing to the source language of outsider late system morphemes of the bilingual clauses. Before moving on, we would like to explain why the Igbo attributive adjectives and their EL counterparts that occur in Igbo–English CS do not occur following the Igbo copula verbs di./no. ‘BE’. The principal reason why Igbo attributive adjectives and their EL counterparts in Igbo–English CS cannot occur following the Igbo copula verbs is because these (i.e. the Igbo di. and no.) are stative verbs with the incorporated meaning of ‘has/have the qualities of ’. Therefore, since true adjectives already fulfil this function on their own (i.e. they are already descriptive words), the Igbo

Language Single Words: Nouns and Adjectives

107

well-­formedness conditions for true adjectives will block the appearance of EL attributive adjectives as complements of the Igbo copula verbs. Put more succinctly, the non-­appearance of English attributive adjectives in the post Igbo copula verb slot can be attributed to a problem of syntagmatic incongruence between the languages. That is, while both languages share the category of adjective, they differ in how attributive adjectives are placed in NPs in relation to the head noun. English attributive adjectives are pre-­modifiers, while they are post-­modifiers in Igbo. This finding strongly confirms that Igbo grammar is in complete control of how both EL attributive and predicative adjectives are handled in Igbo–English CS.

6.2.2  The system (inflectional) morpheme criterion The second criterion for identifying the ML of a bilingual clause follows from Myers-Scotton’s System Morpheme Principle (SMP), which predicts that in ML+EL constituents, all system morphemes which have grammatical relations external to their head constituent (i.e. which participate in the sentence’s thematic role grid) will come from the ML (Myers-Scotton 1993b: 83; 2002: 59). This will involve identifying the relevant system inflectional morphemes in a bilingual clause and noting which language they come from. As we indicated in Chapter  3, Myers-Scotton makes a distinction between system and content morphemes. The distinction is on the basis that content morphemes assign or receive thematic roles, whereas system morphemes do not. Content morphemes are similar to lexical categories such as nouns, verbs and adjectives. Examples of these in Igbo would be the noun nwanyi. ‘woman’ and the verb yi ‘wear’ in (89), and the adjective o.hu. u. ‘new’ in (95). System morphemes are then further subdivided into early versus late system morphemes, according to whether or not they are conceptually activated or directly linked to the speaker’s intentions. Early system morphemes have in common with content morphemes the fact that they are conceptually activated, but differ in that they do not receive or assign thematic roles. Examples of early system morphemes are determiners such as the/a/an in English and the Igbo deictic words ahu. /a ‘that/this’, which add information about definiteness, and plural morphemes, e.g. -s in English, which add information about number. Recall from section  4.4.1 that there is no grammatical gender in Igbo and its nouns are neither declined for case nor inflected for number. However, in Igbo, plurality may be specified optionally by words such as the pronoun (PRN) u. mu. (literal meaning: ‘replicas of ’ or ‘copies of ’) and ndi. (‘people of/those of ’) (see Echeruo 1998 and Emenanjo

108

Codeswitching in Igbo–English Bilingualism

1978). According to Myers-Scotton (2002), early system morphemes are closely tied to their complements (i.e. any content morphemes that may accompany them on the surface); they depend on the specific semantic/pragmatic properties of a content morpheme and add semantic/pragmatic information to that morpheme. Late system morphemes, as we noted in sections 3.2.1 and 3.2.2, subdivide into bridge and outside late system morphemes, the latter being co-­indexed with forms outside the head of their maximal projection (Myers-Scotton 2002: 75), while the former are not. In other words, outside late morphemes are the system morphemes mentioned above which have grammatical relations external to their head. Examples are subject–verb agreement and case affixes. As currently stated, it would be difficult to apply the SMP to Igbo–English because there is no agreement morphology between subject and verb in Igbo. Additionally, as we have already noted here and elsewhere in this study, Igbo has no case affixes (see Emenanjo 1978 and Obiamalu 2013a), as is the case in English (see Radford 2004). Therefore, to operationalize this criterion we shall re-­define outside late system morphemes to mean that only the language identified as the ML will contribute such functional categories as auxiliary verb, tense, aspect, mood and sentence negation, which are associated with the verb in both languages; rather than in terms of relations outside a morpheme’s maximal projection. English inflectional categories tend to occur mostly as auxiliary verbs, modals and verbal affixes: -ed, -ing, -en (see Table  4.3). The inflectional categories are realized in Igbo mostly as verbal suffixes (see Tables 4.2 and 4.4), while others are realized with the auxiliary verbs na- and ga-. We shall select the mentioned functional categories as the main means of identifying outside late morphemes in Igbo and English. This is because they are possibly the most frequent kinds of outside late system morphemes that occur in most clauses in both languages. Crucially, Myers-Scotton and Jake (2014: 4) state that under the SMP of the MLF model, verbal inflections such as these must come from the ML. Thus, the language source of the inflectional morphology on verbs in bilingual clauses should enable us to identify the ML, and this criterion should lead to the same result for each clause as the morpheme order criterion discussed in section 6.2.1. To take an example, the ML of (100), repeated here as (113), was identified as Igbo in section 6.2.1.2 according to the morpheme order criterion: (113) ma ha gba-­ra u. lo. vice-­chancellor o.ku. na the riots but they burn–IND house vice-­chancellor fire PREP the riots ‘but they burned (the) vice-­chancellor’s house in the riots’ [11:28]

Language Single Words: Nouns and Adjectives

109

We can see that it is also Igbo according to the system morpheme criterion, since the affirmative indicative inflectional past tense suffix on the verb -ra comes from Igbo. Earlier, in section 6.2.1.3, we saw that it was sometimes not possible to apply the morpheme order criterion where the word order of a clause (or mixed constituent) was compatible with both Igbo and English, as in example (112), repeated here as (114). (114) ma di. ya a-­di.-ghi. happy but husband her V-BE-NEG happy ‘but her husband is not happy that …’

na … C [41:30]

In this case, the ML has to be identified on the basis of the source language of the inflectional morpheme alone. What happens if we apply the second criterion, the system morpheme criterion? In fact, we can see that the copula verb di., verbal vowel inflectional prefix a- and negative indicative inflectional suffix -ghi. come from Igbo. Therefore, Igbo is the ML of such examples according to the system morpheme criterion only. To summarize, the functional categories auxiliary verb, tense, aspect, mood and sentence negation are those which will assist us to identify the ML of a bilingual clause applying the second criterion. If the MLP is correct, then all verbal inflectional morphemes in the same clause will come from the same source language; additionally, this source language will be the same as that identified by the first criterion, that of morpheme order (see section 6.2.1). Applying both the morpheme order and system inflectional morpheme criteria we can characterize the operation of the MLP as thus (after Amuzu 2010) (using the insertion of the EL noun ceremony in (85), repeated here as (115), as an example): (115) ma ceremony ahu. fu-­ru n̩ nukwu but ceremony D cost-IND big ‘but that ceremony cost a lot of money’

ego money [29:13]

Stage 1: Conceptual (lemma level) – Once a speaker selects an English content morpheme, such as ceremony during Igbo–English CS: l

It is checked for congruence with its Igbo counterpart m̩ mem̩ me (an equivalent Igbo noun with the same meaning as ceremony) at the three levels of abstract lexical structure: (1) lexical-­conceptual structure: closest to the speaker’s intentions; (2) predicate–argument structure: deals with how thematic structure is mapped onto grammatical relations; and

110

Codeswitching in Igbo–English Bilingualism

(3) morphological realization patterns: this is where the morpheme order and system morpheme criteria apply. Levels (2) and (3) are not yet salient in the language production process. Information from level (1) is sent to the formulator for processing. Stage 2: Formulator – The formulator decodes the information sent from the lemma level: l

l

l

Regarding lexical-­conceptual structure; the formulator identifies that ceremony is a noun element (sufficiently congruent with its Igbo counterpart: m̩ mem̩ me); next the formulator notes that ceremony is the nominal complement of the DP under construction, and in relation to morphological realization pattern, the formulator prepares the complement slot in the DP for ceremony. Igbo dominates what happens at this stage (i.e. grammatical structure meant for m̩ mem̩ me is built for ceremony), this dominance is operationalized through the application of the morpheme order and system morpheme criteria.

Stage 3: Surface realization – Consequently, ceremony is inserted into the nominal complement slot under DP preceding the Igbo post-­posed DEM/D a ‘this’, which adds definiteness/specificity to the noun. The same process ensures that the inflectional morphology on the verb also comes from the same language as that supplying the surface word order of the mixed DP, Igbo. However, there are some other examples in Igbo–English CS that do not fall neatly into the pattern described immediately above. In the next two sections, we shall turn our attention to cases involving bare forms and double morphology.

6.2.3  Bare nouns in Igbo–English English is distinct from Igbo because it develops a system of articles to indicate different types of reference (see Disterheft 2004 for more on articles). However, when embedded in Igbo, as we have already seen in the previous sections, English nouns tend to lose their preceding determiners. This observation is further illustrated in the following three examples (example 67 is repeated here as 116). (116) ha a-­hapu. -­la terminal ji-­ri taxi na-­a-lo.-­ta they V-leave-PERF terminal hold-IND taxi AUX-V-return-ENCL ‘They have left (the) terminal and they are returning in a taxi.’ [14:31]

Language Single Words: Nouns and Adjectives

111

(117) n̩ pa m o.-­gwa-la gi na ha fe-­re exam father my CL-tell-PERF you C they pass-IND exam na Abuja? PREP Abuja ‘Has my father told you that they passed (the) exam in Abuja?’ [12:44] (118) ke mgbe service ga-e-bido? when time service FUT-V-start ‘What time will (the) (church) service start?

[13:30]

It would seem, judging from the context in the bilingual clauses, that the three NPs – terminal in (116), exam in (117) and service in (118) – express some kind of specific reference5 but without using any of the determiners encountered earlier in section 6.2.1.1. In other words, the NPs appear in contexts that require the use of overt determiners obligatorily in English but not in Igbo (see also examples 100 and 104). This claim is confirmed by the presence of a pre-­posed determiner in all the monolingual English translations accompanying the bilingual clauses in (116) to (118). According to Myers-Scotton and Jake (2001: 106), EL bare forms are content morphemes that occur in a mixed constituent frame prepared by the ML, but missing some or all of the required ML system morphemes. Therefore, a compromise strategy is activated and used with the result that the EL content morpheme is not placed in a slot projected by its ML counterpart; rather, it is realized as a bare form or as a part of an EL island. In other words, Myers-Scotton and Jake seem to suggest that EL bare forms occur in CS because the lemma supporting a lexical entry in one language might not match the lemma supporting a corresponding lexical entry in another language due to pragmatic considerations (see Myers-Scotton and Jake 1995: 988). This mismatch in lemmas (or incongruence), they claim, is what leads to the occurrence of EL forms as either bare forms or EL islands. While such an explanation might be wholly adequate in cases involving other language pairs, we, however, disagree that this is the case with the indeterminate NPs above. For instance, the EL Ns that occur as bare forms in Igbo–English CS have near direct ML equivalents, which also occupy the same syntactic positions in clause structure (either in subject or in object position). Thus, the English N terminal (as in ‘building’) with u. lo. in ‘Ha a-­hapu. -­la u. lo. ji-­ri taxi na-­a-lo.ta’, exam is the direct equivalent of the Igbo N ule in ‘Ha fe-­re ule na Abuja’, and service (as in ‘church service’) with u. ka in ‘Kee mgbe u. ka ga-e-bido’. Also, the EL bare Ns are not inserted with any noticeable compromise strategies either, as suggested by Myers-Scotton and Jake (2001: 106). Instead, they occur in exactly the same

112

Codeswitching in Igbo–English Bilingualism

syntactic position as their Igbo counterparts. Therefore, we posit that a simpler and more straightforward analysis can be given for the variation observed in the bare Ns/NPs following Igbo grammar (see also Obiamalu 2013a). Recall from section  6.2.1 that we have adopted the DP hypothesis for the Igbo nominal expressions. This hypothesis assumes that the NP is headed by a functional element (D), and accounted for the word-­order differences between the Igbo and English DP structures in terms of a head movement operation (see 81–83 above). Thus, to maximize structural symmetry between determinate and indeterminate nominals, we shall assume that the latter are DPs headed by a following null determiner in line with the ML grammar. In fact, Igbo already permits ‘null determiners’ in its grammar (see sections  4.4.5 and 6.2.1 in this volume and Obiamalu 2013a). If our supposition about the determinate DPs is correct, then the bare EL forms in (116) to (118) will have the structure in (119) below: (119)

Bare nominals are interpretable as definite, indefinite or generic, which are features associated with the functional category D (Radford 2004). Consequently, we argue that in languages like Igbo where there are bare nominals, there is a related null D head which carries the D-features. This eliminates the need for positing bare nominals as compromise strategies in Igbo–English CS. Also, it allows us to treat the insertion of bare nominals no differently to the other EL elements analysed in the earlier sections. On this basis, therefore, we would identify Igbo as the ML according to the morpheme order criterion in all the examples involving EL Ns/NPs that would obligatorily require an overt pre-­ posed determiner to be well-­formed DPs in English but not in Igbo. The examples support the system inflectional morpheme criterion because the source language of the verbal morphology expressing past tense -ri/-re in (116) and (117) and future tense, the auxiliary verb ga- in (118), is also Igbo.

6.2.4  Double morphology in Igbo–English As we pointed out in section 4.4.1, in Igbo, plurality may be optionally specified by combining nouns with the pronouns u. mu. (literal meaning: ‘replicas’ or

Language Single Words: Nouns and Adjectives

113

‘copies of ’) and ndi. (‘people of ’ or ‘those of ’) (see also Echeruo 1998: 104). As a result, we find that the EL nouns in the mixed nominal expressions in examples (91) to (95) above are not marked for plurality in either Igbo or English even though plurality is implied. In fact, we identified those examples correctly as conforming to Igbo grammar because in this language nouns are usually not declined for either case or number. However, the situation in the above mixed nominal expressions contrasts with that in examples (69) (repeated here as 120) and (121) below. (120) tractors ndi. ahu. e-­mebi-cha-­la tractor-­s PRN/PL DEM/D V-damage-ENCL-PERF ‘Those tractors have been damaged completely.’ [28:12] (121) u. fo.di. na-­a-nya buses ndi. some AUX-V-drive bus-­es PRN/PL ‘Some drive those buses …’

ahu.  … DEM/D [50:12]

In the examples, the English nominals appear to be marked for plurality in English and co-­occur with the following Igbo (optional) plural marker ndi.: tractor+-s + ndi. in (120) and bus+-es + ndi. in (121). The English plural nouns can be looked at as NP islands; for example, the plural marker –s post-­modifies its English head noun, tractor, in accordance with the requirements of English grammar, leading to the English NP island tractor+s. Crucially, in both examples, despite the presence of the English plural marker on the nouns, ndi. also occurs, in post-­position according to Igbo grammar, which controls the overall DP structure. An interesting question, however, is why do the Igbo–English speakers in the examples see fit to combine the English nouns with ndi. even though plurality has already been specified on the nouns? It is quite possible to suggest that EL plural nouns are switched as single lexical units (rather than as multi-­ morphemic elements), and treated just like singly occurring nouns by Igbo– English speakers. This might then allow for the occurance of ndi., which combines with the Igbo deictic word ahu. ‘that’ to give the meaning ‘those’ (a function word not found in Igbo). We have to add that only a minority of such examples exist in our data corpus (total = 22), and in every instance, the final modifying element (in the mixed DPs) is the Igbo deictic word ahu. . Furthermore, as we indicated earlier in Chapter 3, Myers-Scotton (2002: 92) claims that the status of plural morphemes is clarified by the refinements that the 4-M model adds to the MLF model in which they are explicitly differentiated from late system morphemes (such as the Igbo verbal inflectional morphemes)

114

Codeswitching in Igbo–English Bilingualism

and classified as early system morphemes (see section 3.2.2). Plural morphemes are so classified because they add conceptual structure to the Ns with which they occur. They do so by denoting two or more of something. Additionally, she explains that the Early System Morpheme Hypothesis (ESMH) correctly predicts that only early system morphemes may be doubled in classic CS (MyersScotton 2002: 92). According to her, from a psycholinguistic point of view, of all system morphemes, only early system morphemes have the special relation to their heads that would promote their accessing when their EL heads are called in CS. Like their heads, early system morphemes are conceptually activated (see sections 3.2.2 and 6.2.2). That is, early system morphemes are salient at the same level as their content morpheme heads in the mental lexicon. Hence, they are ‘available’ if any mistiming is going to occur. On the other hand, late system morphemes are not available yet, but only become salient at the level of the formulator. In addition, she adds that based on empirical evidence there are no examples in the literature to show that late system morphemes are doubled in classic CS (Myers-Scotton 2002: 92). Evidence from Igbo–English CS seems to support the claim by Myers-Scotton that only early system morphemes may be doubled in classic CS. In fact, besides the plural morphemes and complementizers (discussed in Chapter 8), both early system morphemes under the MLF framework, no other types of system morphemes are doubled in mixed DPs/NPs. Thus, in reference to what MyersScotton (2002: 92) describes as ‘mistiming’ at the lemma level, this results in, for instance, the doubling of plural morphemes. What this means in relation to Igbo–English CS is that due to the conceptual relationship existing between an English noun and the plural marker -s both are accessed together as soon as the noun is selected by the speaker at the conceptual (or lemma level) stage and they in tandem enter the mixed DP/NP as a semantic-­syntactic bundle; however at the formulator (or functional level) stage, where Igbo is the dominant language in charge of structuring all mixed DPs/NPs, autonomous processes are set off to project a plural slot for the Igbo optional plural marker ndi. to re-­encode plurality, thereby leading to the doubling of plural morphemes. To some degree, we agree with the preceding explanation, but it does not fully account for why the ML grammar also contributes its own version of the plural morpheme when the EL plural morpheme is already accessed with the EL noun. In her response, MyersScotton states that psycholinguistic experimentation may provide better solutions, but that the answer seems to be that the ML is the more activated of the two languages, no matter what. Therefore, for it to supply as many system morphemes as it can seems simply more efficient. Moreover, she stresses that all

Language Single Words: Nouns and Adjectives

115

system morphemes can be considered part of the morphosyntactic frame; some, she says, are simply more essential parts from the standpoint of constituent building (i.e. late system morphemes) (Myers-Scotton 2002: 92). In the absence of any psycholinguistic experimentation, we shall attempt to posit an explanation that is based on the identified distinctiveness of the ML DP structure. Recall from section  6.2.1 that in addition to the projection of a functional category D, which in some instances has a null realization in Igbo (see sections 4.4.5 and 6.2.3), more than one functional projection could be found within the Igbo nominal phrase (see Obiamalu 2013a). In effect, Igbo grammar already provides explanations for such examples as in (120) and (121) above. For instance, consider the structure of the monolingual Igbo DP in (84), repeated here as (122): (122)

The claim here is that functional elements can already co-­occur in the structure of the Igbo NP (see Obiamalu 2013a: 57). As we explained earlier in section 6.2.1, the structure in (122) shows that there is no hierarchical order between the DemP and QP in Igbo. Also, we saw from the examples in (80) a and b that either element can dominate the other without triggering any changes in meaning. Thus, we explained that by the movement of the NP to the Spec of the available functional projections, the noun is said to have access to the features of the functional heads in a Spec–head relationship. That is, the Q niile assigns the feature [+ plural] to the N, while the DEM ahu. assigns the features [+ definite + specific]. Stemming from this, we might suggest that with or without the presence of the EL plural morphemes on the EL nouns the DPs would still be interpreted as plural because our speakers are quite simply treating the EL nouns as they would Igbo nouns, which are typically uninflected for number. This could

116

Codeswitching in Igbo–English Bilingualism

explain why the Igbo noun n̩ nu. nu. ‘bird’ in (122) is interpretable as both plural and definite without the need for any overt plural marker. As applied to the bilingual examples, our view is that the EL NP islands appear to be accessed in language production as single lexical units rather than being put together on the spot every time they occur. In much the same way as explained by Backus (2003: 84). If so, then the examples do not really represent instances of double plural marking on the EL nouns; instead Igbo ndi. is a separate functional projection within the structure of the Igbo DP which combines with the deictic word ahu. to give the mixed DPs an overall plural reading in the absence of the Igbo Q niile/dum ‘all’ and NUM/D abu. o. ‘two’, ato. ‘three’, and so on. The presence of the Igbo Qs and NUMs (which possess the feature [+ plural]) in examples (91) to (95) could explain why the EL nouns in the mixed DPs are without their plural inflections (which are present in the monolingual English translations). And yet the bilingual DPs in those examples are interpretable as plural even though the nominal elements are without number markers. We can extrapolate from this that the EL nouns in (120) and (121) are only plurals in English (the EL); for our speakers they are just like Igbo nouns, which are uninflected for number. Thus, it is the grammatical instinct of our speakers which compels them to introduce ndi. into the DP structure; this functional element combines with ahu. ‘that’ to express a notion similar to the English plural demonstrative determiner ‘those’. This particular functional element is not found in Igbo. Therefore, it would seem that once a speaker selects a plurarlized singly occurring English noun, at the functional level stage, where Igbo is in control of structuring all mixed DPs/NPs, abstract grammatical processes are set off to project a D̍ slot for the Igbo functional element ndi., which combines with DemP ahu. ‘that’ to give the supra-DP both [+ plural] and [+definite + specific] reading. We are confident that ndi. is a determiner because it not only has the feature [+ plural], it also marks definiteness and specificity in Igbo; both functions are typical of determiners. Another reason for proposing that ndi. (an Igbo pronoun) is a determiner is that it can be used both post-­ nominally and pronominally (after Radford 2004: 37–40), as in the monolingual Igbo examples below: (123)

Language Single Words: Nouns and Adjectives

117

(124)

In (124), ndi. is used in place of the noun ‘rat(s)’ where it is post-­modified by another functional head, DEM a. In both examples, it combines with the DEM to express the notion ‘these’ in English. On this basis, examples (120 and 121) can be analysed as (125) below: (125)

The structure in (125) confirms our earlier claim that functional heads can co-­ occur in the Igbo NP. Crucially, (125) clearly illustrates that Igbo and not English is the ML in the examples, because it supplies the word order of the mixed DPs and the verbal inflectional morphology in the bilingual clauses.

6.3  Quantitative analysis 6.3.1  Testing the MLP Here, we will present the results of a quantitative analysis conducted in order to test the MLP. The analysis applies to the bilingual clauses containing the 1,264 singly occurring EL nouns and 152 singly occurring EL adjectives described in section  5.6 and exemplified in the immediately preceding sections. English words listed in the Igbo dictionary which appeared in otherwise Igbo clauses were considered Igbo as outlined in section 5.5. The two criteria for identifying the ML were then applied to the mixed constituents in order to test the MLP, according to which it is always possible in classic CS to identify the ML of a mixed constituent. The application of the first criterion, morpheme order, led to

118

Codeswitching in Igbo–English Bilingualism

two possible outcomes: (1) the label ‘Igbo’ if the order in a mixed constituent was compatible only with Igbo (see examples 85–95, 99, 100, 104–108, 116–118, 120 and 121); and (2) the label ‘either’ if the order was compatible with that of both languages (as in examples 109–112). The application of our second criterion, the identification of the source language of all verbal inflectional morphology (i.e. late system morphemes in this study), for identifying the ML of each bilingual clause in Igbo–English CS containing the singly occurring EL nouns and adjectives produced the results presented in Figure 6.1 and Table 6.2. The table also shows the results from the application of our first criterion (that of morpheme order). According to the summarized results in Table 6.2 and Figure 6.1, only one language, Igbo, contributes the late system morphemes in all the bilingual clauses containing the 1,416 singly occurring English nouns and adjectives. Also, the same language is the source of morpheme order in 89.4 per cent (this includes the 144 English nouns that occur with a following null determiner in contexts requiring the obligatory use of overt determiners in English but not in Igbo) of the mixed constituents; whereas the order in the remaining 10.6 per

Figure 6.1  ML according to morpheme order and late system morpheme criteria

Language Single Words: Nouns and Adjectives

119

Table 6.2  ML according to source of morpheme order and late system morpheme criteria Singly occurring EL nouns and adjectives in Igbo–English CS (Total = 1,416) Patterns of CS

Source of morpheme order

Source of late system morphemes

+ Igbo (955) 67.4% √ (144) 10%

+ Igbo (955) 67.4% √ (144) 10%

√ (128) 9% √ (40) 3% Either (37) 2.6%

√ (128) 9% √ (40) 3% √ (37) 2.6%

Either (112) 8%

√ (112) 8%

(1,416) 100%

(1,416) 100%

1.  Mixed DPs N/NP+D(post-­posed) N/NP+ø determiner(post-­posed) 2.  Mixed NPs N1(possessum)+N2(possessor) N+A(post-­posed adjectival modifier) N+N(associative construction) 3.  Predicative adjectives Copula di/no+A(predicative adjective) Totals

cent is compatible with that of both Igbo and English. There is no example in our data corpus where the word order in a mixed constituent is solely English. Additionally, Table  6.2 reveals that 67.4 per cent (total = 955/1,416) of all single English nouns in Igbo–English CS occur with post-­posed determiners (described as early system morphemes under the MLF model) from Igbo. Another 12 per cent (total = 168/1,416) occur in well-­formed NPs overtly including both Igbo and English elements with morpheme order and late system morphemes coming from Igbo only; whereas 10 per cent (N = 144/1,416) of all English nouns occur in contexts where a determiner is obligatory for a well-­ formed DP in English but not in Igbo. That is, these nouns occur with a post-­ posed ‘covert’ (Radford et al. 2009) Igbo determiner. Myers-Scotton (2002: 59) states that only if the terms of the principles, morpheme order and one type of system morpheme (i.e. late system morpheme) are satisfied by one and the same language can the ML be identified as that language. On this basis, we determine that the ML of 89.4 per cent (total = 1,267/1,416) of the bilingual clauses is Igbo unequivocally, as in Figure 6.2 below; whereas, the ML of 10.6 per cent (total = 149/1,416) of the remaining bilingual clauses is Igbo according to the late system morpheme criterion only. The results show overwhelming support for the MLP of the MLF model. Moreover, our

120

Codeswitching in Igbo–English Bilingualism

Figure 6.2  The ML: Igbo

results are not dissimilar to those reported for Ewe–English (Ewe is the ML: Amuzu 1998), Lingala–French (Lingala is the ML: Bokamba 1988), Hungarian– English (Hungarian is the ML: Bolonyai 2005) and Welsh–English (Welsh is the ML: Davies and Deuchar 2010). We shall, however, return in section 6.3.3 below to discuss some of the implications of the results reported here for Igbo–English CS. Next, we shall test the Asymmetry Principle (AP) and the Uniform Structure Principle (USP) with Igbo–English data.

6.3.2  Testing the AP As we outlined in section  3.2.4.2, this principle states that bilingual speech is characterized by asymmetry in the roles played by the two languages participating in CS. Regarding classic CS, the principle predicts that only one of the participating languages will be the source of ML, not both. We can explore this asymmetry in two specific ways: (1) asymmetry in the source of late system morphemes (defined in this study as verbal inflectional morphology); and

Language Single Words: Nouns and Adjectives

121

(2) asymmetry in the resolution of word order conflicts in mixed constituents. To test this principle, we can make use of the analysis reported in the previous sections and summarized in Table 6.2. If the AP is to be upheld, then we would expect the vast majority of bilingual clauses to have either Igbo or English as unequivocally the ML. Firstly, regarding asymmetry in the source of late system morphemes, the results from Table 6.2 revealed that Igbo contributed all (100 per cent) the late system morphemes in the clauses containing the singly occurring English nouns and adjectives. Secondly, concerning asymmetry as to whose word order prevails in conflict situations, the analysis in Table 6.2 showed that 89.4 per cent of the EL words occurred in structures where there was word order conflict between Igbo and English, and only 10.6 per cent of the EL single words occurred in structures where the word order was compatible with both Igbo and English, and hence not unequivocally identifiable. The implication of this observation is that the order of Igbo prevailed in all the 1,267 (89.4 per cent) mixed constituents exhibiting a conflict in word order between Igbo and English. Thus, the basic theoretical notion that there is an ML versus EL hierarchy is comprehensively supported in our data, because the two languages (Igbo and English) do not both satisfy the roles of the ML contained in the MOP and the SMP of the MLF model.

6.3.3  Testing the USP In section 3.2.4.3, we outlined that the USP states that in bilingual speech, the structures of the ML are always preferred, but some EL structures are allowed if certain conditions are met (Myers-Scotton 2002: 8–9). The USP stems from the 4-M model (see section 3.2.2), which makes a distinction between content and system morphemes, dividing the system morpheme into three subtypes: early system morphemes, bridge late system morphemes and outsider late system morphemes (see section  6.2.2 for our redefinition of outsider late system morphemes as they apply to the present study). We have already seen in the data under consideration in this chapter that in Igbo–English CS late system morphemes can only come from the ML of a clause (which is always Igbo). However, the USP goes beyond this to predict that other system morphemes, such as early system morphemes, will be drawn preferentially from the ML of a bilingual clause. The principle does not exclude the occurrence of EL early system morphemes, only that the occurrence of EL early system morphemes in mixed constituents is less likely than early system morphemes from the ML.

122

Codeswitching in Igbo–English Bilingualism

As it turns out, Igbo as the ML does not only supply the late system morphemes (the bound affixes on verbs used in the syntactic expression of tense–aspect– mood) of all the bilingual clauses (100 per cent; see Table  6.2), but also the overwhelming majority (98.5 per cent) of the early system morphemes (e.g. demonstrative determiners, pronominal determiners, quantifiers, numerals, pronouns, and so on) and the bridge late system morphemes (as in example 106, repeated here as 126). (126) a-­ga m e-yi akwa white na aki.si. red … CL-AUX I V-wear dress white CONJ headscarf red ‘I will wear (a) white dress and red headscarf …’ [11:28]

In (126), the Igbo coordinating conjunction na is a bridge late system morpheme; it links the two mixed NPs akwa white ‘white dress’ and aki.si. red ‘red headscarf ’. According to Myers-Scotton (2002), a bridge morpheme depends on the well-­ formedness conditions of a specific constituent in order for it to appear; such a constituent is not well formed without the bridge morpheme. Under this view, the Igbo copula verbs di. and no. (see examples 109–112) can be considered types of bridge late system morphemes because they link the subject and predicate in a sentence/clause. The verbs are required to build structures within the clause larger than just the DP/NP. Crucially, Igbo is the sole contributor of such morphemes in the mixed constituents considered here. English, the EL, contributes only 1.5 per cent of the early system morphemes in the form of the plural marker -s on EL nouns (as in examples 120 and 121 above). Recall from the discussion in section  6.2.4 that the presence of EL plural morphemes on EL nouns in mixed constituents does not constitute counterexamples to the SMP because they belong to early system morphemes, which rather than outsider late system morphemes can be doubled following the 4-M model ESMH. Thus, as we have already seen in the previous sections, the role of English (the EL) in the mixed constituents seems to be limited to supplying only content morphemes including nouns and adjectives. Furthermore, the pattern of zero determination observed in the 144 English nouns that occur in contexts where a determiner is obligatory in the former for a well-­formed DP but not in Igbo (see section 6.2.3) provides additional empirical evidence that, in our data, Igbo is the source of ML. That is, in the examples, Igbo provides the morphosyntactic frame for ML+EL constituents where Igbo system morphemes (both early and late) can appear freely, but even English early system morphemes, which are not ruled out by the USP, can hardly appear in (nominal) EL+ML

Language Single Words: Nouns and Adjectives

123

(functional element) constituents. In other words, the strict constraints on EL system morphemes by the ML well-­formedness requirements can account for the morphosyntactic features reflected in the 144 bare nouns. Due to the syntagmatic incongruence between English determiners (which are typically pre-­posed to the nominal element) and their Igbo counterparts (which typically follow the nominal element), the ML of Igbo will restrict their appearance and finally lead to their loss in mixed structures. This is underlined by the fact that in the entire corpus there is no instance where a lexical noun is in Igbo and the determiner is from English. The foregoing analysis, therefore, clearly indicates overwhelming support for the USP because, as the principle predicts, the structures of Igbo (the ML) are preferred in all the mixed constituents analysed in this chapter.

6.4  Summary and conclusion We set out in this chapter to test the plausibility of the MLP on the sample of singly occurring English nouns and adjectives embedded in Igbo clauses. By utilizing both qualitative and quantitative methods which make use of the contrasting elements of Igbo and English grammars, we have demonstrated that singly occurring EL words in Igbo–English CS are embedded with restrictions set by the ML (Igbo) exactly as the MLP predicts. Specifically, exploration of the relative order of head and complement within bilingual nominal expressions (see sections 6.2.1.1 to 6.2.1.3, 6.2.3, 6.2.4 and Table 6.2) indicates that the lone English nouns and adjectives overwhelmingly (89.4 per cent) pattern like Igbo nouns under DP/NP. Furthermore, this finding of DP/NP-internal switching joins a long list of studies reporting CS examples which appear to contradict Poplack’s (1980) EC proposal (see section 2.2), which predicts that a switch may occur only between sentence elements that are normally ordered in the same way by the monolingual grammars in contact. As we saw earlier in sections 6.2.1.1 to 6.2.1.3, 6.2.3 and 6.2.4, examples such as the mixed DP in (127) below occur in Igbo–English CS. (127) ma flats ndi. ahu. di.-cha but flats D DEM BE-ENCL ‘but those flats are (extremely) very expensive’

very expensive very expensive [39:25]

Example (127) was contributed in the context of a conversation about the high cost of rented accommodation in Port Harcourt. Our focus here is on the order

124

Codeswitching in Igbo–English Bilingualism

of the mixed DP flats ndi. ahu. ‘those flats’. It is evident in (127), as in the other examples cited in the earlier sections of this chapter, that Igbo and English do not share the same Phrase Structure Rule (PSR) under DP. Principally, any material accompanying an Igbo noun typically must follow the noun, while any material accompanying an English noun typically must precede the noun under DP. Therefore, since the PSRs are different from one another, the EC would block CS between Igbo and English within DP. The implication then, is that an Igbo–English bilingual is permitted (under the EC proposal) to mix only entire DP/NPs within S (sentence) or clause, and should not mix Igbo and English in the DP/NP. However, the example in (127) and those in (85–95), (99), (100), (104–108), (116–118), (120) and (121) each include an instance of CS within the DP/NP. For instance, in (127), English would require that the Igbo demonstrative ahu. , which combines with the functional element ndi. to give the meaning ‘those’ (see discussion in section 6.2.4 for how the two elements combine in Igbo DP structure) precede the English plural noun flats. In English syntax, demonstratives precede instead of following the nouns they modify. However, following the English PSR would violate Igbo syntax, which stipulates that demonstratives follow rather than precede the nouns they modify. Hence, the mixed DP flats ndi. ahu. would be erroneously blocked under the EC because it violates the English DP PSR. As we have seen throughout this chapter, all such structural conflicts are resolved in favour of the language identified as the ML in the bilingual clauses, Igbo. The examples would also falsify the predictions of both the GC (see section  2.4; Di Sciullo et  al. 1986) and the FHC (see section  2.5; Belazi et  al. 1994), which seek to prohibit switching between a functional head and its complement, between a noun and an adjective which modifies the noun, and so on. Put simply, the GC and the FHC assume like the EC that CS between the governor and the governed is blocked unless both come from the same language. Evidently, the examples analysed in this chapter seem to violate this prediction because in the vast majority of cases English nouns in our data are followed by either overt post-­posed Igbo determiners (67.4 per cent: 955/1,416) or covert determiners (10 per cent: 144/1,416). Also, 3 per cent (or 40/1,416) of the few singly occurring English attributive adjectives in Igbo–English occur in post-­ position to the Igbo nouns they modify (see section 6.2.1.3). Given that the CS frameworks are posited as universally applicable, all that was needed to falsify them was just one counterexample. However, as we have shown in this chapter, our data contains many counterexamples. These go to confirm the observations made much earlier by other CS researchers6 that the EC, GC and FHCs are not

Language Single Words: Nouns and Adjectives

125

universally applicable constraints. Moreover, the majority of the examples considered in this chapter would also falsify the MA of MacSwan (1999, 2000, 2005a, 2005b, 2009), who argues that no principle of grammar can be explicitly formulated for CS; and that the identities of particular languages involved in CS are ignored for the purposes of linguistic theory (see section 2.6). In other words, MacSwan does not seem to recognize any theoretical value in the existence of strong asymmetries in the roles played by the two languages involved in CS. In fact, evidence from Igbo–English confirms that in the type of CS structures analysed here, the structures of the language identified as the ML are preferred over those of the EL. Igbo–English is not alone in showing strong asymmetries in the roles played by the ML versus the EL (see also Ewe–English: Amuzu 2005; Hungarian–English: Bolonyai 2005; Welsh–English: Couto et al. 2015; Deuchar 2005, 2006; Finnish–English: Halmari 1997; Ewe–Kabiye: Essizewa 2007a; Persian–English: Rahimi and Dabaghi 2013; Urban Wolof–French: Legendre and Schindler 2010; Swahili–English: Myers-Scotton 1993b; these studies report findings similar to those from Igbo–English, which permits the type of CS explicitly prohibited by one or more of the putatively universal CS frameworks reviewed in Chapter 2 of this study). As the figures in Table 6.2 indicate, only a total of 152 singly occurring English adjectives are found in Igbo–English. The vast majority of these adjectives (8 per cent: 112/1,416) occur as predicative adjectives following the Igbo copula verbs di./no.; only 3 per cent (40/1,416) occur as attributive adjectives modifying a preceding Igbo (ML) noun. This finding is not unique to this study (see section  6.2.1.3). In fact, Myers-Scotton (2002: 132) observes that very few EL adjectives modifying ML nouns occur in CS corpora, possibly because of congruence problems at all levels. Although Myers-Scotton, as Deuchar (2005: 262) points out, does not specifically distinguish between congruence in relation to attributive versus predicative adjectives, this distinction is drawn by Pfaff in relation to her Spanish–English data. Spanish, like Igbo, normally has its adjectival modifiers following rather than preceding the noun. Pfaff (1979: 314, cited in Deuchar 2005: 262) finds that adjective switches are unrestricted when they take the form of predicate adjectives, but are limited within the NP. It would seem that, similarly to the cases in Pfaff ’s data, in Igbo–English CS, English adjectives can be inserted relatively freely in Igbo utterances in predicative position, where there is considerable syntagmatic as well as paradigmatic congruence; their insertion is less free and frequent in attributive position, where the syntagmatic congruence is less because of differences in the word order of the two languages.7 This would suggest that considerations of congruence may

126

Codeswitching in Igbo–English Bilingualism

be more important in our data. However, we must exercise some caution about how we interpret this finding because it is based on only a small amount of examples involving adjectives in our data. We would need to examine more qualitative and quantitative data on adjectives before we can draw firmer conclusions. Nevertheless, the test of the AP on our data of singly occurring EL nouns and adjectives (see section 6.3.2) revealed overwhelming support for the principal theoretical notion that there is an ML versus EL hierarchy, because Igbo and English do not both satisfy the roles of the ML according to our two criteria for identifying the ML of each bilingual clause. It was shown that Igbo supplies both word order and the frame-­building elements (function words and verbal inflections) into which English single words are inserted. Similarly, the testing of the predictions of the USP, in section  6.3.3, revealed that even English early system morphemes, which are not prohibited by this principle, can hardly appear in ML+EL constituents. We saw that the strict constraints on EL system morphemes by the ML well-­formedness requirements meant that the overwhelming majority of the early system morphemes (98.5 per cent) were supplied by Igbo. Stemming from the analysis of the EL single nouns and adjectives embedded in otherwise Igbo utterances, Igbo–English CS seems to represent a classic case of CS. That is, in the bilingual clauses examined in this chapter, only one of the participating languages (Igbo) is the source of the morphosyntactic frame for the clauses. The goal of the analysis in the next chapter, then, is to see whether the switching of singly occurring English verbs pattern just like the single words analysed in this chapter. We expect that when the EL single verbs occur in structures manifesting a conflict in word order between Igbo and English, such conflicts will be resolved in favour of the language identified as the ML.

7

Embedded Language Single Words: Verbs 7.1  Introduction We have already seen, in the previous chapter, that switching between Igbo and English is not blocked at those sites considered to be non-­equivalence sites for both languages; and that the functional category determiner always comes from Igbo in the mixed nominal expressions. The same language (i.e. Igbo) was also found to be in charge of word order and the overall morphosyntactic frame of the bilingual clauses containing the singly occurring EL nouns and adjectives. Here, we turn our attention to the singly occurring EL verbs in Igbo–English CS. According to Myers-Scotton and Jake (2014: 4), in contrast to nouns, verbs carry a good deal of grammatical information relevant to the phrase structures in which they occur; they assign thematic roles. They add that verbs determine how thematic roles are mapped onto predicate–argument structure. For this reason, the authors predict that switching a verb should carry a higher production cost than a noun. Perhaps this might explain why in many CS studies (this one included) single-­word nouns are switched more frequently than verbs (see section  5.6). Also, Myers-Scotton and Jake (2014: 4) note that earlier, it was assumed under the MLF model that all EL verbs must be checked for congruence with the requirements of the ML grammatical frame (see Myers-Scotton 1993b; Myers-Scotton and Jake 1995). However, the authors now posit that not all aspects of the grammatical structure projected by EL verbs are checked for congruence in CS. They say that, similarly to nouns, only the lexical-­conceptual structure of EL verbs is checked for congruence with the ML (ibid.: 4), as we discuss in the sections below. The analysis will include every singly occurring English verb in our corpus (total = 547; see also section 5.6). That is, we shall explore all singly occurring English verbs in clauses exhibiting no conflict in word order between Igbo and English (see our interpretation of the morpheme order criterion in section 6.2.1) as far as the switched elements are concerned. Example (128) illustrates switching

128

Codeswitching in Igbo–English Bilingualism

of an EL verb inflected with ML morphology, but there is no conflict in word order between Igbo and English in terms of the positioning of the switched item in clause structure. (128) anyi. believe-u. -ru. na ihe ha kwu-­ru na.… we believe-V-IND PREP thing they say-IND PREP ‘We believed in the thing they said at…’ [16:47]

In (128), the ML supplies the EL verb with the element (inflection) that marks it as a finite verb. Such EL verbs occurring in ML finite slots are discussed in section  7.2. Recall that we mentioned in section  5.4.1 that some EL verbs in Igbo–English CS receive not only the ML inflections for tense–aspect–mood, but also occur with the uniquely Igbo pronominal subject clitics (see section 4.4.8). This is illustrated in example (129) below. (129) e-develop-u-­ghi photo wedding Adao.ma na oge CL-develop-V-NEG photo wedding Adao.ma PREP time ‘They did not develop Adao.ma’s wedding photos on time.’

The focus in (129) is on the singly occurring English verb develop, which receives both the ML negative inflection marker -ghi, a late system morpheme (see our redefinition of late system morphemes in section  6.2.2), and the pronominal subject clitic e-, an early system morpheme under the MLF model (see MyersScotton 2002). The NP photo wedding ‘wedding photo’ is not an EL island because the two English nouns are only juxtaposed; that is, these morphemes follow the word order of Igbo (the ML), not English. Therefore, as Myers-Scotton (2002: 139) correctly states, this is simply a sequence of two EL content morphemes in a mixed constituent under ML order (EL islands are discussed further in the next chapter). However, the EL verbs in this type of bilingual clause, which contrasts sharply with the structure of the monolingual English translation in (129), are discussed further in section 7.3. The process of affixation in Igbo is associated with its VH (vowel harmony) rule, in that the vowel(s) of all affixes must also harmonize with the vowel(s) of the verb stem closest to the affix. We shall demonstrate in section  7.4 that the presence of VH between English (a language without VH rule) verbs and Igbo affixes provides additional empirical evidence of ML influence on all levels of grammatical structure. In section 7.4.1, we shall discuss the results of visual observation of spectrograms involving some EL verbs in Igbo–English, which tends to suggest that the ML phonological system must have simultaneous access to vowels in morphemes from two different languages (both Igbo and English). This observation, as we

Embedded Language Single Words: Verbs

129

shall show later, has implications for some of the CS models reviewed in Chapter  2 of this study. Furthermore, recall from section  4.4.4.1 that another important distinctive feature of verbs in Igbo is that of verb serialization. We noted there and in section 4.4.4 that morphology (i.e. inflection) is obligatory for all Igbo verbs except copulas (for example, bu. /di./no. ‘BE’), and some instrumental verbs, such as ji ‘hold’ and bu ‘carry/bear’, which may or may not receive morphology in the language. Therefore, the only time a full Igbo verb may appear without morphology (i.e. bare) is when such a verb occurs in an SVC (serial verb construction). As it turns out, all the bare EL verbs in Igbo–English occur in SVCs. This is illustrated in example (9), repeated here as (130). (130) o-­meV1-re campaignV2 nyeV3 anyi. CL-do/make-IND campaign give us ‘He campaigned and gave us his manifesto.’

manifesto ya manifesto his [4:10]

According to Aikhenvald (2006: 1), an SVC is a sequence of verbs which act together as a single predicate, without any overt marker of coordination, subordination or syntactic dependency of any sort. Looking at the bilingual clause in (130) we can notice immediately that it is very distinct from its monolingual English translation. For instance, in the monolingual English translation a coordinating conjunction (and) is introduced which creates two clauses instead of the single clause found in the bilingual example. Also, notice that like the Igbo verb nye ‘give’, the English verb campaign is not inflected for tense either in the bilingual clause. Instead the first verb in the series, the Igbo verb me ‘do/make’, is inflected for tense. We shall explore further the occurrence of such EL bare verbs in section 7.5. A quantitative analysis that relates to all three principles of the MLF model outlined in section 3.2.4 follows in section 7.6. In section 7.7, we shall present a summary of the discussions and conclude the chapter. To sum up, for the purpose of our investigation here, if Igbo is the ML of the bilingual clauses in Igbo–English CS, an English verb would remain bare only if such a verb is one in a series of verbs in an SVC. We expect that all verbs that are not in SVCs will receive morphology obligatorily from Igbo.

7.2  Insertion by verbal inflectional morphology In a review of the insertion of EL verbs in various language pairs, Backus (1996: 212) and Myers-Scotton and Jake (2014: 8) observe that these verbs are usually

130

Codeswitching in Igbo–English Bilingualism

non-­finite but are made finite by ‘ML means’. A similar observation is also reported by Deuchar (2005: 263–7) for the English verbs in her Welsh–English CS study. This observation by these researchers is an important one in view of our definition of the ML, for which the language of verbal inflection is an important criterion (see section  6.2.2). In this section, we will focus on the insertion in Igbo of non-­finite verbs of English. According to Myers-Scotton and Jake (2014: 8), the reason why non-­finite EL verbs are easily inserted in CS structure is due to the fact that non-­finite verbs do not carry the same costs as finite forms because their levels of predicate–argument structure and/or morphological realization patterns are not salient in structure building. They add that their only salient level of abstract structure is the level of lexical-­ conceptual structure, as is the case with nouns. Thus, such verb forms as infinitives and present participles can take ML verbal inflections without creating any congruence problems regarding the abstract levels referring to grammatical structure. That is, according to Myers-Scotton and Jake, non-­finite verbs do not send directions regarding EL late system morphemes that need to be accommodated, and, presumably, inhibited in CS, if an EL finite verb were employed. Because of this, they claim that non-­finite verbs fit the low-­cost scenario that can be observed in CS for the phrases that would contain finite verbs in monolingual data. The EL verbs are inserted into Igbo–English CS in two main ways. The EL verbs are either inserted in synthetic constructions, where the sole verb is inflected for tense, aspect and negation by the ML, or they are inserted in periphrastic constructions in which the finite auxiliary comes from the ML. These are the two main strategies adopted in Igbo for the insertion of English verbs, as we shall see shortly. Examples (131) and (132) illustrate the two types of constructions in monolingual Igbo structures. (131) Ginika bi.a-­ra Ginika come-IND ‘Ginika came yesterday.’

eci yesterday

(132) Ginika ga-­a-bi.a Ginika AUX-V-come ‘Ginika will come.’

In (131), we notice that the past tense is expressed by the presence of the inflection suffix -ra on the sole verb bi.a ‘come’. Example (132) shows that the future tense is marked by the presence of the bound auxiliary verb ga- and the vowel inflectional affix a- on the main verb bi.a. The harmonizing vowel inflectional affix (or particle) is found with nearly all Igbo verbs (see Emenanjo

Embedded Language Single Words: Verbs

131

1978: 128). As we shall see, Igbo–English bilingual speakers use the same verbal affix to integrate English verbs that are not found in the Igbo dictionary (which are all the verbs analysed in this study) into their speech. Since according to Myers-Scotton and Jake (1995) complete congruence between languages is not always possible, some English verbs in our corpus are integrated by means of the addition of the Igbo vowel particle just mentioned. This allows for the EL verbs to be properly inflected with the appropriate ML morphology which conveys the speaker’s intentions. This can be likened to the Hungarian and Welsh verbalizer suffix added to the English verbs described in Bolonyai (2005: 317) and Deuchar (2005: 264–5) respectively. According to Deuchar (2005: 264–5), the addition of a Welsh verbalizer suffix to English verbs is a way of compensating for the partial congruence between Welsh and English verbs and a way of integrating EL (i.e. English) verbs into Welsh morphologically. The following are some utterances from Igbo–English which illustrate the integration of English verbs by means of Igbo inflectional catogories (see Table 4.2). (133) ha miss-i.-ri. flight they miss-V-IND flight ‘They missed their flight…’

ha… ha [15:30]

(134) maka na ha register-ra na the wrong desk… C C they register-IND PREP the wrong desk ‘because they registered at the wrong desk…’

[17:30]

(135) ma ha book-u. -ru. na hotel but they book-V-IND PREP hotel ‘but they booked into (a) hotel’

[15:30]

Examples such as (133) to (135) make it very clear that the English verbs are not finite forms because the speaker’s intentions call for a past tense marking, but miss-­ed, register-­ed and book-­ed do not occur; the past meaning comes only from the ML inflectional morphology that the English verbs do not influence. The vowel inflectional suffixes on the EL verbs in (133) and (135) could be viewed as inserted to make the verbs satisfy Igbo (i.e. ML) phonotactics (see discussion in section 4.3.1). Similar examples are also reported in Deuchar (2005: 265) and Myers-Scotton and Jake (2014: 9). Other examples from Igbo–English are given in (136) to (139) below. (136) ha a-qualify-i.-ghi. They V-qualify-V-NEG ‘They did not qualify.’

[17:13]

132

Codeswitching in Igbo–English Bilingualism

(137) anyi. ga-­e-kick ha niile we AUX-V-kick them all ‘We will kick all of them out of office.’

out of office out of office

(138) o clean-i.-cha-­la moto she clean-V-ENCL-PERF (motor) car ‘She has finished cleaning your car.’ (139) o.-na-­a-cho. i.-start CL-AUX-want INF-start ‘She wants to start quickly.’

[19:30] gi your [29:43]

ngwa ngwa quick quick [1:3]

In the above examples, the ML only supplies all the relevant inflectional morphology on the singly occurring English verbs: the negative inflection -ghi. and the verbal paricles in (136), the auxiliary ga- and the vowel particle in (137), the perfective suffix -la, enclitic (meaning ‘completely’) and verbal particle in (138), and the infinitive prefix i.- in (139). We decided to include these examples to show that word-­internal switching, like DP/NP-internal switching, is not blocked in Igbo–English CS contrary to the predictions of the CS models reviewed in Chapter  2 of this volume. Therefore, the language production process involved in the switching of the mixed verbal expressions (similar to the process underlying the switching of mixed nominal expressions discussed in the previous chapter) can be explained as detailed below (taking the insertion of believe in (128) as an example). Stage 1: Conceptual (lemma level) – Once a speaker selects an English content morpheme, such as the verb believe in (128) during Igbo–English CS, he/she also selects Igbo as the ML of the mixed verbal expression under production. The ensuing processes are triggered to commence the building of an appropriate grammatical slot for believe in the clause: l

First, believe is checked for congruence with its Igbo counterpart kwe at the three levels of abstract lexical structure: (1) lexical-­conceptual structure: closest to the speaker’s intentions; (2) predicate–argument structure: deals with how thematic structure is mapped on to grammatical relations; and (3) morphological realization patterns: this is where the morpheme order and system morpheme criterion apply. Information from all three levels is sent to the formulator (including information about the full abstract lexical structure of the Igbo counterpart of believe = kwe).

Embedded Language Single Words: Verbs

133

Stage 2: Formulator – The formulator decodes the information sent from the lemma level: l

Regarding lexical-­conceptual structure, the formulator identifies that believe and kwe are sufficiently congruent because:

1. They encode identical concept – i.e. they have analogous lexical-­ conceptual structure; l

They have the same predicate–argument structure:

2. They are both verbs (believe is used as a transitive verb in example (128): ‘the speaker considers what is being said in the example to be true’) that assign the same thematic roles to their argument, i.e. to the subject (Agent) and to the object (Patient); l

In relation to morphological realization pattern, they both require no case marking of their arguments;

3. Igbo is the language in control of functional level processes, therefore the late system morpheme criterion (see above section 6.2.2) ensures that only Igbo supplies all the required inflectional morphology in the mixed verbal expressions. In example (128), believe receives the Igbo indicative affirmative inflection marking past tense. Also, the morpheme order criterion ensures that Igbo morphosyntactic procedures are used in framing the mixed verbal expression. Igbo, like English, is typically an SVO language as reflected in the above examples. Hence, there is no conflict in word order in the examples as far as the switched elements are concerned. Stage 3: Surface realization – Consequently, believe in (128) is inserted into the verbal slot intended for the Igbo verb kwe. The examples considered above appear to confirm the view expressed by MyersScotton and Jake (2014: 7) that the EL is active in CS at the level of lexical-­conceptual structure, when an EL verb is selected as the lemma that best satisfies the speaker’s intentions. Thus, the EL verb brings along its meaning, but it is the ML that integrates it into predicate–argument and morphological realization patterns. That is, how thematic roles are realized in the syntax is determined by the ML.

Codeswitching in Igbo–English Bilingualism

134

As we saw in the examples above, the ML (i.e. Igbo) determines how verbal inflectional morphology is realized in Igbo–English CS. There are also other English verbs inserted in Igbo with not only the late system morphemes seen in the examples above, but these verbs occur with the Igbo pronominal subject clitics (PSC) described earlier in section 4.4.8. We shall discuss such examples in the next section.

7.3  Insertion by PSC and verbal inflectional morphology As we noted in section  4.4.6, Igbo has three singular, three plural and one impersonal pronoun. Furthermore, the three singular pronouns have each an independent and dependent form. In some contexts, as we showed in section 4.4.8, the place of the pronoun can be occupied by what may be called a pronominal prefix (or pronominal subject clitic, after Anyanwu 2012), i-/i.- ‘you (singular)’, o/o.- ‘he/she/it’ and a-/e- ‘nonperson-­number specific’ (see Anyanwu 2012; Emenanjo 1978; Green and Igwe 1963), harmonizing with an immediately following verb stem vowel. Before getting to the CS examples, we shall argue after Anyanwu (2012: 376–8) that the Igbo dependent pronominal elements are PSCs. As PSCs, they can only occupy pro-argument positions in the constructions where they appear though, superficially, they seem to appear at subject argument position. Anyanwu (2012) provides supportive evidence for arguing that the Igbo PSC has the features of pro in pro-­drop languages by firstly examining some of the unique features of the Igbo pronominal elements that establish their status as clitics. The Igbo PSCs, like pronominal subject clitics in French (see Jaeggli 1981), must be adjacent to either the main or auxiliary verbs unlike the independent pronouns and lexical NP subjects, which pattern differently. (140) a. b.

O. - *naani. ga-­ra ahi.a 3SG.CL only go-IND market Eze/Unu naani. ga-­ra ahi.a Eze/2PL.PRN only go-IND market ‘Only Eze/you went to market.’ (Anyanwu 2012: 379–80)

Example (140a) is unacceptable because the Igbo PSC is split from the verb by the introduction of naani. ‘only’, which can correctly post-­modify an Igbo noun or pronoun, as in (140b). Another feature of the Igbo PSCs is that they always occur on their own without modification, while the independent pronouns can be modified by numerals, as we see in (141) below.

Embedded Language Single Words: Verbs (141) a. b.

135

O. - *ato. ka ha cho.-ro. 3SG.CL NUM FOC 3PL want-IND (self-­benefactive) Ya ato. ka ha cho.-ro. 3SG NUM FOC 3PL want-IND (self-­benefactive) ‘It is the three (of them) that they want.’ (Anyanwu 2012: 380)

Unlike both lexical NPs and independent pronouns, Igbo PSCs cannot be topicalized. This accounts for why (142a) is ungrammatical. (142) a. b.

*O. -, ka Ada nye-­re ego 3SG.CL FOC Ada give-IND money Ya/Ha/Unu/Eze, ka Ada nye-­re ego 3SG/3PL/2PL/Eze FOC Ada give-IND money ‘It is him/they/you/Eze that Ada gave money.’ (Anyanwu 2012: 380)

In Igbo, both the independent pronouns and lexical NPs can be clefted, but the PSCs cannot as illustrated in the ungrammaticality of (143a). (143) a. *O. bu. o.- ka Ada nye-­re ego It BE 3SG.CL FOC Ada give-IND money b. O. bu. ya/ha/unu/Eze ka Ada nye-­re ego it BE 3SG/3PL/2PL/Eze FOC Ada give-IND money ‘It was him/them/you/Eze that Ada gave (some) money.’ (Anyanwu 2012: 380)

Whenever there is emphasis on the subject, an independent rather than a PSC is used. This can explain the unacceptability of (144a) below. (144) a. b.

*O. - bi.a! 3SG.CL come! Ya/Unu/Ha bi.a! 3SG/2PL/3PL come ‘Let him/you/them come.’ (Anyanwu 2012: 380)

According to Anyanwu (2012: 380), an Igbo PSC cannot be conjoined with an independent pronoun or lexical NP. This is unlike the independent pronouns and lexical NPs that can be conjoined with each other, as in examples (145a–c) below. (145) a. b.

*O. - na 3SG.CL and Ya na 3SG and ‘S/he and you came.’

Ada bi.a-­ra Ada come-IND unu bi.a-­ra 2PL come-IND

136 c.

Codeswitching in Igbo–English Bilingualism Ya/Ha na Ada/gi. bi.a-­ra S/he/they and Ada/you come-IND ‘S/he/they and Ada/you came.’ (Anyanwu 2012: 380)

In addition to the above syntactic evidence, we shall show in the next section that, like the Igbo verbal affixes, the [ATR] vowel quality of pronominal clitics is conditioned by the [ATR] vowel quality of a following verbal element. Furthermore, Anyanwu (2012: 381–3) cites evidence from some pro-­drop languages such as Italian, Spanish and Degema to support his position that Igbo should be among the group of pro-­drop languages. For instance, he argues that both Italian and English pattern closely in having verb agreement markers (suffixes) while Degema and Igbo allow PSCs. Italian, English and Degema pattern in the sense of allowing the co-­occurrence of the verb agreement markers or PSCs with lexical or substantive pronouns. This option is not available in Igbo, except if the lexical NP or substantive pronoun is made a topic and adjoined to the left of a minimal clause (AGRsP), thus allowing the PSC to occur at the pre-­verbal position (as in 146). (146) Eze [o.-na-­asu. French] Eze 3SG.CL-AUX-speak French ‘Eze speaks French.’ (Anyanwu 2012: 381–2.)

Still, Italian, Degema and Igbo share the feature of having the phi-­features of the non-­substantive subject position to be recovered through verb agreement suffixes (for Italian) or PSCs (for Igbo). Thus, what is marked and recoverable through pronominal clitics in languages like Igbo is marked and recoverable through verb agreement suffixes in languages like Italian (see Anyanwu 2012: 382). Crucially, for the purpose of the CS analysis presented here, if Anyanwu’s (2012) supposition is correct, then Igbo belongs with Japanese and Chinese, for example, as languages with underived inflectional forms exhibiting neither rich system of verb agreement (like Italian and Spanish) nor rich systems of subject clitic agreement (like Degema) and yet which permit pro licensing. English, on the other hand, is characterized as a language exhibiting erratic or partial verb agreement where pro licensing is not possible (ibid.: 383; Ndimele and Kari 1997; Radford 2004). With this knowledge in mind, consider the singly occurring English verbs in examples (147)–(150) below. (147) o.-study-ri. na 3SG.CL-study-IND PREP ‘He studied in London in the 70s.’

London in the 70s London in the 70s [5:4]

Embedded Language Single Words: Verbs

137

(148) i.-si. na i.-lodge-i.-ri. na guest house ahu. 2SG.CL-say C 2SG.CL-lodge-V-IND PREP guest house DEM ‘You said that you lodged in that guest house.’  [35:7] (149) e-send–i.-ri. invitation card iri CL-send-V-IND invitation card NUM ‘(Some person/s) They sent ten invitation cards.’

[42:40]

(150) e-decorate-i.-ri. hall ano. na u. lo.o.gwu. o.hu. u. ahu. CL-decorate-V-IND hall NUM PREP hospital new that ‘(Some person/s) They decorated four halls in that new hospital.’ [36:40]

One of the interesting things about the above CS examples is that Eze (1997) claims that such single-­word switching is blocked in Igbo–English CS under the EC model (see Poplack 1981 and section  2.2 of this volume), and thus no examples are attested in his Igbo–English corpus. Contrary to Eze’s claim, Igbo– English CS clearly permits switching between verb and PSCs. Following Anyanwu (2012: 383–4), the PSCs can be analysed in two ways; they can be analysed just like the English subject pronouns, in which case the PSCs are arguments in the Spec of AGRsP as in (151). (151)

The structure in (151) represents the ‘substantive argument analysis of PSCs’ (to borrow the terms used in Anyanwu 2012: 383); it can be argued that the PSC is generated at Spec, VP from where it raises to Spec, AGRsP for nominative case marking. Also, given the morphosyntactic status of the PSCs outlined above and in Anyanwu (2012: 383), the argument analysis implies that the subject pronominal clitic having undergone a Spec-­to-Spec move operation is at some point, lowered1 to AGRs by a syntactic operation or by a purely phonological process of cliticization. On the other hand, it can be argued that the PSC cannot occupy Spec, AGRsP argument position but just a mere functional element generated at AGRs, which identifies a null pronominal in

138

Codeswitching in Igbo–English Bilingualism

subject position (cf. Rizzi 1986, cited in Anyanwu 2012: 383). This type of analysis, the pro analysis as in (152), entails that the PSC is base generated at AGRs while its Spec, AGRsP is occupied by pro. Thus, the PSC is a spellout of subject agreement features (cf. Campbell 1998, cited in Anyanwu 2012: 383). (152)

In line with the pro-argument analysis (after Anyanwu 2012), we would identify Igbo as the ML in bilingual clauses where singly occurring English verbs are immediately preceded and followed by Igbo PSCs and verbal inflectional (late system morphemes) morphemes according to both morpheme order and late system morpheme criteria (see sections  6.2.1 and 6.2.2 for our definitions of both criteria as they apply to this study). Moreover, it has often been the case that some researchers dismiss as borrowings EL language forms inflected with ML morphemes (see discussion in section  5.5). For instance, Poplack and Meechan (1998) and MacSwan (1999, 2000) argue that singly occurring EL forms such as those discussed above, which are inflected with ML bound morphemes, introduce a second phonology, and thus all the examples analysed in the immediately preceding sections are either borrowings or completely blocked from occurring in CS. We, however, argue to the contrary that CS forms can and do undergo morphophonological and syntactic adaptations where necessary so as to fit properly into the ML (see Backus 1996; Deuchar 2005, 2006; Gardner-Chloros 1991; Pfaff 1979; Stammers and Deuchar 2012; Treffers-Daller 1994; most notably Myers-Scotton 1993b, 2002 and Myers-Scotton and Jake 2014). However, an important question is how do we demonstrate that the EL single words in Igbo–English CS are not borrowings but genuine CS forms besides our claim that these words are not listed in the Igbo dictionary? In answering this question, we want to do two things: (1) to show that two phonologies can co-­exist in CS; and (2) to show that the English-­origin single words in

Embedded Language Single Words: Verbs

139

Igbo–English CS are not as phonologically adapted (integrated) as either borrowed or native words in Igbo. That is, we want to show that the EL single words are pronounced with a phonology that is not wholly like that of Igbo. To do this, we shall begin by demonstrating the co-­existence of two phonologies in Igbo–English CS in the section immediately below (i.e., section 7.4). In sections 7.4.1 and 7.4.1.2, we will utilize spectrograms to validate our claim that two phonologies can indeed co-­exist in CS. Thus, we shall determine (in addition to the fact that the EL single words are not listed in the Igbo dictionary) that the EL single words are genuine CS forms rather than some kind of borrowings.

7.4  Vowel harmony between EL verbs and ML bound morphemes An important phonological distinction between Igbo and English is that the former displays a system of vowel harmony (VH) not found in the latter. In section 4.3.2, we defined VH as a type of phonological process by which vowels agree with other vowels in terms of a particular phonological feature such as backness, roundness, height or Advanced Tongue Root (ATR). Also, in the same section, we stated that Igbo displays a robust form of ATR root control harmony. That is, Igbo affixes typically harmonize with the vowel of the stem (or root) to which they are attached. Since we have established in the preceding analysis that a grammar analogous to that of Igbo is in control of how affixes combine with English verbs in the data, then one would reasonably expect that the vowels of the Igbo affixes are harmonizing with those of the English verbs with which they occur. If this is the case, then such evidence is one more proof of ML influence in Igbo–English CS.

Table 7.1  Igbo vowel correspondences and feature classes Igbo

I

IPA [high] [round] [ATR]

i + − +

i. I + − −

Adapted from Zsiga (1997: 232).

e

a

u

e − − +

*a − − −

u + + +

u. ʊ + + −

o *o − + +

o. *ɔ − + −

Codeswitching in Igbo–English Bilingualism

140

Table 7.2  English vowels Features i:

I

u:

ʊ

ɔ

o:

ɒ

ɑ:

*ʌ

æ

e:

ε

*ә

*әʊ з:

High Low Back Front Round Tense

+ − − + − −

+ − + − + +

+ − + − + −

− − + − + −

− − + − + +

− + + − + −

− + + − − +

− + − − − −

− + − − − −

− − − + − +

− − − + − −

− − − − − −

− − − − + +

+ − − + − +

− − − − − +

Adapted from Davenport and Hannahs (2010: 113).

To establish whether the vowels of the English verbs are subject to VH, we have repeated Table  4.1 (Igbo vowel correspondences and feature classes) as Table  7.1, and added Table  7.2 (showing feature specifications for English vowels) to allow for comparative analysis. Before commenting on the relevance of both tables, we must point out that while the [ATR] feature is used particularly for the description of a number of languages (including Igbo) which show VH process, it is sometimes used in the description of English for the distinctions referred to in Table 7.2 under the feature [±Tense]. This is because advancing the root of the tongue often involves a concomitant raising of the tongue body. Thus, [+ATR] can be the equivalent of [+Tense] and [-ATR] the equivalent of [-Tense]. It is equally vital to remind the reader that we outlined earlier in section  4.4.4 (see Table  4.2) that the vowel of the Igbo indicative affirmative suffix -rV reduplicates the vowel of the preceding syllable (i.e. it matches its [±ATR] value). The same rule applies to the infinitives, which are formed by prefixing either i- or i.- to the verb stem, while the negative is formed by prefixing either a- or e- and adding the suffix -ghi or -ghi. to the verb stem. With this knowledge in mind, now consider the examples in (153) and (154), where the affixes appear to harmonize with their closest stem vowels with respect to their ATR value. (153) ndi. Shell discover-ra oil wells o.hu. u. na Bonny people of Shell discover-IND oil wells new PREP Bonny ‘People of Shell discovered new oil wells in Bonny.’ [1:27] (The Shell Oil Company discovered new oil wells in Bonny.) (154) e-­chefu-la i-close igbe a … CL-forget-NEG INF-close box DEM ‘Don’t forget to close this box …’ 

[18:43]

Embedded Language Single Words: Verbs

141

Your attention is drawn to the three English vowel sounds that have been substituted for what we will call Igbo sounds (in bold type): /ʌ/ (open-­mid back unrounded vowel) → /ɔ/ (open-­mid back rounded vowel); /ә/ (mid-­central unrounded vowel) → /a/ (open front unrounded); and /әʊ/ (fairly close/fairly back rounded diphthong) → /o/ (close-­mid back rounded vowel). Table  7.2, showing feature specifications for English vowels, reveals that the first two vowels from English (/ʌ/ and /ә/) have the feature value [-Tense], while the feature specification for the diphthong /әʊ/ is [+Tense]. Now compare the feature values for the English sounds (identified by the asterisks next to the sounds of interest) with the [ATR] feature specifications of their Igbo replacements in Table 7.1. First, discover-ra (Standard British English transcription of the stem [.dIs.kʌ.vә.]) versus our speakers’ pronunciation with the past tense suffix attached ([.dIs.kɔ.va.ra.]). Igbo, like the Geordie dialect of British English, does not have the open-­mid back unrounded vowel /ʌ/. Therefore, English words containing this sound are pronounced with the closely related open-­mid back rounded vowel /ɔ/, as shown in the transcription. Also the schwa /ә/ is not present in Igbo and replaced by the open front unrounded vowel /a/, also shown in the transcription. Figures 7.1 and 7.2 give syllable tree diagrams of our two words. Second, close (Standard British English transcription [.klәʊz.]) versus our speakers’ pronunciation with the infinitive prefix attached ([.I.klo:z.]). Again, Igbo does not have diphthongs and such sounds are realized as one long sound. In fact, a diphthong is quite simply a sequence of two vowel sounds. Therefore,

Figure 7.1  Syllable tree diagram of the word [.dIs.kɔ.va.ra.]

142

Codeswitching in Igbo–English Bilingualism

Figure 7.2  Syllable tree diagram of the word [.i.kloz.]

our speakers reflect this fact in their pronunciations by using long vowels for the diphthongs. It is rather remarkable to observe that the bilingual speakers’ vowel choices match the [±Tense/±ATR] values of the substituted English vowels. Most notably, the syllable structure of the English words remains intact. The only changes occur on either the left or the right edge of the English verbs depending on whether either a prefix or suffix is attached. On this last point, recall from the discussion in section 4.3.2 that Igbo displays a robust form of ATR root control harmony. Therefore, the Igbo affixes must look to the English verbs in order to determine their own [ATR] value, implying that a single phonological system must have simultaneous access to vowels in morphemes from two different languages; a situation which seems to violate MacSwan’s (2009) PFIC-based account (see section 2.6). This finding is similar to that reported in Legendre and Schindler (2010: 58–9) for Urban Wolof (UW). In their UW–French CS study, the researchers show, in particular, that a French root vowel can trigger ATR vowel harmony on a vowel in a Wolof suffix similar to the situation between an English verb and an Igbo affix. According to Legendre and Schindler (2010), Wolof roots harmonize according to the ATR distinction. They add that in Wolof VH is progressive – i.e. the ATR value of the initial vowel determines the ATR value of all subsequent vowels. Also, Wolof VH is not restricted to roots. When a suffix with a mid-­vowel is added to a root, the ATR value of the root determines the ATR value of the suffix vowel. For instance, in Wolof, the post-­verbal past tense marker woon has the forms [wɔ:n] and [wo:n];

Embedded Language Single Words: Verbs

143

the post-­verbal plural imperative marker leen may appear as [lε:n] or [le:n], and the demonstrative boobu has the forms [bɔ:bu] and [bo:bu]. Furthermore, they cite various examples from their data such as in example (155) below illustrating that VH was present between French and Wolof elements in UW. (155) a. pomme boobu ‘that apple’ 

b. faire-leen ‘make!’

c. peser-woon ‘weighed’ (Legendre and Schindler 2010: 57)

Using various acoustic measurements they were able to demonstrate that in example (155), the Wolof vowel takes its ATR feature from the final French vowel. The only verbs that would trigger [-ATR] harmony are [-ATR] irregular verbs such as faire. Therefore, evidence from both UW–French and Igbo–English CS appears to violate not just MacSwan’s (2009) PFIC-based account but also Sankoff and Poplack’s (1981) FMC (see section 2.3).

7.4.1  Preservation of EL syllable structure in CS forms The goal here is to demonstrate that apart from the Igbo inflectional morphology on either the left (prefix) or right (suffix) edge of the English verbs embedded in Igbo, the syllable structure of the words remains intact. In other words, the EL words retain their syllable structure once we go past the Igbo affixes. To show this, we examined the syllable structure of a sample of fifty (total = 50) singly occurring English monosyllabic and multisyllabic verbs found in bilingual clauses in the Igbo–English data. We made sure to select contributions from both the male and female speakers described in section 5.2.1. First, we used Audacity2 digital audio editor software version 2.0.3 to identify points in the larger audio files at which the fifty words occurred. From this we made fifty smaller audio files of the EL free morphemes (i.e. lone English verbs) + ML bound morphemes (i.e. the inflectional affixes identified in the sections above). We did this instead of asking the speakers to recite an a priori prepared list of fifty words. It is a well-­established fact that when speakers pay attention to their speech (as is often the case in recorded pronunciation tasks) they are more likely to shift away from their usual pattern of pronunciation. Therefore, our method avoids this and instead the analysis is done on naturally produced data. Additionally, the fifty EL verbs were selected on the basis that they contained: (1) coda consonant(s) within their syllable structure (total = 25); and (2) consonant clusters (total = 25). As we indicated in section 4.3.1, Igbo has a strict two-­syllable structure: V/CV. That is, according to the phonotactic constraints of Igbo:

144

Codeswitching in Igbo–English Bilingualism

1. It does not permit coda consonant(s) anywhere in its syllable structure. 2. It does not permit consonant clusters anywhere in its syllable structure. Therefore, if the singly occurring English verbs with Igbo inflectional morphology are some kind of borrowings, then we would expect to see vowel epenthesis at every syllable boundary in the fifty English verbs where there is either a filled coda or consonant cluster as a means by the Igbo–English bilinguals to:

1. Neutralize the coda consonant(s); and 2. Split the consonant clusters. Put simply, the bilingual speakers’ autochthonous phonological knowledge should compel them (as it were) to make the words conform to the Igbo two-­ syllable structure. If this is found not to be the case, then the only conclusion left would be to accept that the attested EL single words are genuine CS forms because they are not as phonologically integrated into Igbo as either native or borrowed words are (see the earlier examples in sections  4.3.1 and 5.5). Once the smaller audio files containing our fifty words were ready, each word was individually examined to check for vowel epenthesis in the EL words apart from the inflectional affixes. For instance, in (153), discover-­ra (Standard British English transcription of the free morpheme including syllable boundaries: [.dIs.kʌ.vә.]) versus our speakers’ pronunciation with the past tense suffix attached ([.dIs.kɔ.va.ra.]): if the word is some kind of borrowing, then we would expect to see an additional vowel inserted between the coda consonant [s] of the first syllable and the onset consonant [k] of the following syllable. Similarly, in (154), close (Standard British English transcription [.klәʊz.]) versus our speakers’ pronunciation with the infinitive prefix attached ([.I.klo:z.]): if this is a borrowed word, then we would see a vowel inserted between the consonant cluster in the onset position and another at the end of the word to neutralize the coda consonant. For the examination, we used Praat3 software version 5.3.85 to generate both wave forms and spectrograms of the fifty words. The latter was most useful for the purpose of our analysis, as we explain below. A spectrogram is a visual representation of sound. It displays the amplitude of the frequency components of the signal over time. Hence, spectrograms are very useful in the analysis of complex signals. Complex signals are those that contain more than one frequency component. Therefore, the spectrograms of the words would allow us to see a visual representation of any epenthetic vowel(s). Usually,

Embedded Language Single Words: Verbs

145

in speech, the various components that make up a complex signal do not share the same amplitude value. For our purpose amplitude refers to the loudness of the component sounds that make up each word. Differences in amplitude are shown in a spectrogram by shading. That is, the frequency components with the highest amplitude values are shown in ‘dark black’ and components with lower amplitude values are displayed in lighter shades of ‘grey’ to ‘white’ (‘white’ signifies very low amplitude or silence). In this way, a spectrogram is three-­dimensional: it shows (1) time (duration) on the horizontal axis; (2) frequency on the vertical axis; and (3) amplitude by degree of shading. The results discussed below are based on observations of the degree of shading (amplitude) in the spectrograms. It is equally important to inform the reader that vowels are usually louder on the sonority scale than consonants (see O’Grady and Archibald 2012 and Selkirk 1984). This is exactly the case for vowels in both languages. Therefore, any inserted vowel(s) (apart from the vowels of the Igbo affixes on either the left (prefix) or right (suffix) edge of the words) would be clearly visible in a spectrogram.

7.4.1.1.  Ø epenthesis at syllable boundaries with coda consonant(s) The results of the analysis indicate that the Igbo–English bilinguals do not insert a vowel sound at syllable boundaries with coda consonants. In the spectrogram of the word discover+ra [.dIs.kɔ.va.ra.], as shown in Figure 7.3, we can see that no insertion takes place within the internal structure of the English free

Figure 7.3  Spectrogram of the word ‘discover+ra’

146

Codeswitching in Igbo–English Bilingualism

morpheme. The only additional syllable is that of the Igbo rV past tense suffix; otherwise the English verb retains its syllable structure. Specifically, the spectrogram in Figure  7.3 shows that no epenthetic vowel is inserted between the coda consonant [s] of the first syllable and the following onset consonant [k] of the succeeding syllable. Even EL verbs such as e-register-ghi. in (156), which receive both the Igbo PSC and negative suffix do not change their internal syllable structure, as the spectrogram in Figure 7.4 shows. (156) e-register-ghi. aha nwa ya na oge … CL-register-NEG name child her PREP time ‘They did not register her daughter’s name on time …’

[15:19]

As in the former example, the syllable structure of the EL verb is retained as no vowel is inserted between the voiceless alveolar sibilant [s] and the following voiceless alveolar stop [t]. However, the vowels in all the EL verbs investigated in the study have been coloured by the accent of the speakers. This is not unexpected since the Igbo–English bilinguals are not native speakers of English. Nonetheless, the spectrograms indicate no vowel epenthesis within the EL free morphemes. Finally, here, Figures 7.3 and 7.4 as well as Figures 7.5 and 7.6 reveal that the Igbo–English bilinguals do not insert any extra vowels in the EL verbs with coda consonant(s). Crucially, the spectrograms confirm our earlier claim in section 7.4 that two phonologies can indeed co-­exist in CS.

Figure 7.4  Spectrogram of the word ‘e+register+ghi.’

Embedded Language Single Words: Verbs

147

Figure 7.5  Spectrogram of the word ‘i+close’

7.4.1.2.  Ø epenthesis in consonant clusters Further proof of the co-­existence of two phonologies in CS is found in the examination of the EL verbs containing consonant clusters. Again, we notice that the speakers in our study do not insert a vowel inside consonant clusters as the two spectrograms confirm. It is evident in Figure 7.5 that no vowel is inserted inside the onset consonant cluster [kl] or following the coda consonant [z] of the word i+close. The Igbo infinitive prefix i- is clearly visible to the left of the image in Figure 7.5. This observation is also corroborated by the vowel format patterns in the spectrogram. According to Fleischhacker (2005: 60), ‘st’ and ‘sr’ clusters may be repaired by inserting a vowel either ‘before the clusters or inside it’. However, our analysis contradicts this claim as the spectrogram of the word ‘i-­destabilise’ in Figure 7.6 illustrates. Figure 7.6 shows that no vowel is inserted inside the onset consonant cluster [st] or in word final position after the coda consonant [z] of the word ‘i+destabilise’. It is clearly evident from the preceding analysis that EL verbs in Igbo–English CS violate Igbo language-­specific phonotactic constraints:

1. The words retain their coda consonant(s) against the basic Igbo CV syllable structure. 2. No vowels are inserted inside consonant clusters in violation of Igbo morphophonological structure. An explanation for the lack of vowel epenthesis in the speech of these bilinguals could be because they are educated middle-­class speakers, who have had long

148

Codeswitching in Igbo–English Bilingualism

Figure 7.6  Spectrogram of the word ‘i+destabilise’

exposure to the other language through education, work and socialization. Therefore, even when they combine EL verbs with Igbo affixes they still strive to preserve the syllable structure of the EL words. Moreover, the spectrograms confirm our earlier claim that any adaptations to the words occur only on either the left or right edge of the words depending on whether a ML prefix or suffix is required. In other words, the words in our data are not wholly phonologically integrated into the ML like native or loanwords. Consequently, on the basis of this analysis (also) we conclude that the singly occurring English words inserted in otherwise Igbo utterances are genuine CS forms. Also, the examples of word-­internal CS from the Igbo–English data represent serious violations of the descriptive generalizations underlying Poplack’s (1981; 2012) FMC/NBH and MacSwan’s (2009) PFIC-based account (see sections 2.3, 2.6 and 5.5). Finally, similarly to the findings from UW–French CS (Legendre and Schindler 2010), the examples from Igbo–English CS appear to suggest that ML and EL phonology can co-­exist in CS. In the next section, we shall explore the singly occurring EL verbs in Igbo–English that occur as bare forms (i.e. without ML verbal inflections).

7.5  Bare EL verbs in Igbo–English codeswitching The analyses in the previous sections have shown that when mixed constituents are accessed by Igbo–English bilinguals, there is necessarily interaction of the

Embedded Language Single Words: Verbs

149

two grammars at an abstract level; however, the grammar of the ML (Igbo) is more activated than that of the EL (English). We have seen this in patterns of affixation on non-­finite EL verbs. The influence of the ML is shown to extend into the phonology of the EL in how the Igbo affixes harmonize with their closest vowels in the EL free morphemes (i.e. the English verbs embedded in Igbo). In sum, the analyses in all the preceding sections have shown overwhelming support for the ML versus EL construct of the MLF model. However, we have to account for the remaining examples that occur as bare forms in Igbo–English CS. It is important to underline that the MLF model does not disallow bare forms (see Myers-Scotton 2002), as long as the system morpheme criterion is satisfied (see the 144 bare EL nouns analysed in section  6.2.3). Therefore, in exploring the bare verbs we shall seek to establish the language of morpheme order and the source language of the late system morphemes. Recall that we have already stated that inflectional morphology is obligatory for all Igbo verbs except copulas (for example, bu. /di./no. ‘BE’), and some instrumental verbs, such as ji ‘hold’ and bu ‘carry/bear’, which may or may not receive inflection in the language (see sections 4.4.4, 4.4.4.1, 7.2 and 7.3). That is, all full Igbo verbs in non-SVCs receive inflection obligatorily. Verb serialization may be described, after Aikhenvald (2006: 1), Ameka (2006: 128–9) and Amuzu (2013: 19–20), as having the following features:

1. It is a monoclausal construction in which two or more VPs (including any complements and adjuncts) appear as a single predicate without any overt marker of coordination, subordination, or syntactic dependency of any sort. 2. The verbs in sequence combine to describe what is conceptualized as a single event. 3. All the verbs in it share one syntactic subject that is expressed only once, with V1. 4. Within the SVC, the individual verbs may have the same or different transitivity values. 5. Where applicable when two (or more) verbs in sequence share an object argument, the object argument appears only once with the first of the verbs. 6. The verbs can be formally marked for the same or different aspect and modality categories. Some of the above characteristics are observed in Igbo SVCs. In particular, as we outlined in section 4.4.4.1:

150

Codeswitching in Igbo–English Bilingualism

1. An SVC is the only situation when a full Igbo verb may appear bare (i.e. without morphology) as one of a succession of verbs (see Amaechi 2013). 2. The first verb in the series is usually marked for temporal reference. However, Emenanjo (1978) and Manfredi (1991) indicate that not every initial verb in a SVC bears morphology. 3. Some verbs in a SVC may take extensional affixes. Crucially, an important feature of SVCs in connection with Igbo is that one of the verbs bears morphology while the others may occur as bare forms. As it turns out, all bare EL verbs in Igbo–English CS occur in SVCs. Amaechi (2013) identifies four main types of SVCs found in Igbo, and these are also reflected in the bilingual examples below. A number of the EL verbs in Igbo–English occur in what Amaechi (2013: 159) refers to as ‘multi-­event SVCs’. In this type of SVC, different events which are related are formed; and all the verbs share a single subject as in (157) to (161). (157) o-­meV1-re campaignV2 nyeV3 anyi. manifesto ya CL-do/make-IND campaign give us manifesto his ‘He campaigned and gave us his manifesto.’

[4:10]

(158) commissioner abu. o. kaV1 e-­meV2-re arrestV3 kpo.ro.V4 gaV5-wa Abuja commissioner two BE V-do/make-IND arrest take go-ENCL Abuja ‘Two commissioners were arrested and taken to Abuja’ [37:3] (159) councillor anyi. baV1-ra government bidoV2 meV3-we embezzleV4 ego councillor our enter-IND government start  do/make-INCP embezzle  money ‘Our councillor entered government and started to embezzle money.’[42:32] (160) o.-biaV1-ra deliverV2 speech gaaV3 leeV4 that new unit … 3SG.CL-come-IND deliver speech go look that new unit ‘He came, delivered his speech and went to see that new unit …’ [43:32] (161) o-­meV1-re promiseV2 gwuoV3 3SG.CL-do/make-IND promise dig ‘He promised and dug three (water) boreholes …’

borehole ato. … borehole three [23:40]

Other EL verbs in the data occur in what is termed instrumental SVC (Amaechi 2013: 159), as in (162) to (163). (162) mu.   na nwunye m  jiV1 aka  anyi. meeV2 dismantleV3 the leaking roof Me/I and wife  my hold hand our  do/make dismantle the leaking roof ‘My wife and I dismantled the leaking roof with our hands.’ [7:32]

Embedded Language Single Words: Verbs

151

(163) e-­jiV1 ha u. gbo. m̩ miri meeV2 surveyV3 ebe a-­gaV4-a-­ru. V5 oil rig ahu. CL-hold they vehicle water do/make survey place V-FUT-V-build oil rig that ‘They used a water vehicle (ship) to survey the location where they will build that oil rig.’ [40:27]

In Igbo, the verb ji ‘hold’ is used to express instrumentality and it usually occurs in a complex structure [- NP VP], typical of SVCs, where it obligatorily takes a complement and a VP, as in (162) and (163). For instance, the object of V1 aka anyi. ‘our hands’ in (162) is also the instrumental argument of V3. Example (164) occurs in what may be termed a dative SVC. (164) nwunye m gaV1-e-­meV2 gi.  inviteV3 gwaV4-kwa gi. u. bo.chi. m wife my FUT-V-do/make you invite tell-ENCL you day I gaV5-a-­pu. V6-ta o.ru. FUT-V-leave-ENCL work ‘My wife will invite you and tell you what day I will get off work.’ [1:12]

In Igbo, dative constructions typically surface as V–V compounds; and indicate/ distinguish the recipient of something given or transferred, as in the above example (see also Amaechi 2013). The fourth type of SVC is resultative SVC (Amaechi 2013: 159), as in example (165) below. (165) o-­meV1-re m̩ ma ya sponsorV2 ya a-­gaaV3 Canada 3SG.CL-do/make-IND mother his sponsor she V-go Canada ‘He sponsored his mother and she travelled to Canada.’ [15:6]

As in monolingual Igbo resultative SVCs, in (165), we observe that V3 expresses the result of V2 and the object of V1 is regarded and understood to be the subject of V3. Additionally, in the above example, V2 is analysed as incorporating into V1 to give the complete predicate. We shall now explore the bilingual SVCs further in the section below.

7.5.1  Characteristics of bilingual SVCs in Igbo–English codeswitching Recall from section 7.5 that an important feature of verb serialization is that the sequence of verbs share the same subject NP/DP; they may have an intervening object between the verbs, as in (159), (162), (163), (164) and (165). The SVCs in our data share the same subject NP/DP: o ‘he’ in (157), commissioner abu. o. ‘two commissioners’ in (158), councillor anyi. ‘our councillor’ in (159), o. ‘he’ in (160), o ‘he’ in (161), mu. na nwunye m ‘My wife and I’ in (162), e in (163), nwunye m ‘My wife’ in (164), and o ‘He’ in (165).

152

Codeswitching in Igbo–English Bilingualism

Besides, another very important feature of verb serialization in Igbo is that the sequence of VPs in the construction act together as a single predicate, without any marker of coordination, subordination, or syntactic dependency of any kind. This is exactly the situation with all the verbs in the bilingual SVCs. For instance: the verbs me-­re ‘did’, campaign and nye ‘give’ in (157); ka ‘copular’, e-­me-­re ‘did’, arrest, kpo.ro. ‘take’ and ga-­wa ‘start to go’ in (158); ba-­ra ‘entered’, bido ‘start’, me-­we ‘start to do’ and embezzle in (159); bi.a-­ra ‘came’, deliver, gaa ‘go’, and lee ‘see/look’ in (160); me-­re ‘did’, promise, and gwuo ‘dig’ in (161); ji ‘used’, mee ‘do’ and dismantle in (162); ji ‘used’, mee ‘do’ and survey in (163); ga ‘will’, mee ‘do’, invite, and gwa ‘tell’ in (164); and me-­re ‘did’ and sponsor in (165) are all contained within a single clause (as far as Igbo is concerned), and these verbs share syntactic subject and object within their clauses. On the other hand, in the monolingual English translations of the same examples, the monoclausal bilingual SVCs are expressed as multi-­clausal constructions linked by the conjunction ‘and’. Also, other elements of the monoclausal bilingual SVCs now appear in the monolingual English translations as DP complement of a PP in (162, ‘with our hands’) and ADV clauses in (163, ‘… where they will build that oil rig’; and 164, ‘… what day I will get off work’). Lastly, in the monolingual English translations of (159) and (163), the English verbs embezzle and survey appear as ‘to + infinitive’ construction. This is absent in the bilingual SVCs where the two verbs are not inflected with the Igbo infinitive prefix i.-. Thus, the structural configuration of the bilingual SVCs resembles that of Igbo (see the monolingual Igbo SVCs in section 4.4.4.1) rather than English. We also noted in section 4.4.4.1 that in Igbo SVCs the VPs in the sequence are construed as occurring within the same temporal frame. Some verbs in the series may appear with or without morphology. Auxiliaries, negation, tense and aspect markers of the sequence of verbs are found usually with the first verb of the SVC. These requirements, it would seem, can account for why all the English verbs in the examples above occur without ML morphology even though a past or future tense reading is implied. Notice that in the monolingual English translations the EL verbs receive the appropriate finite verb inflections, which are missing in the bilingual SVCs. Moreover, we can observe in the bilingual SVCs that auxiliary, negation, tense and aspect markers are found with the first verb in the series. For example, the Igbo auxiliary verb ka ‘BE’ in (158), the -rV indicative affirmative suffix on the Igbo verb me ‘do’ in (157, 158, 161 and 165), on ba ‘enter’ in (159), and bi.a ‘come’ in (160); and the Igbo bound auxiliary gaand the verbal particle e- on me ‘do’ in (164). In (162) and (163), the Igbo instrumental verb ji ‘hold/use’ is bare but obligatorily receives a past tense

Embedded Language Single Words: Verbs

153

interpretation (see section  4.4.4.1 this volume, Amaechi 2013 and Uwalaka 1982). From this, it would seem that the English verbs are bare only because this is what is expected according to the grammar of Igbo, the ML. This finding means that Igbo not only contributes morpheme order but also the late system morphemes in the examples. Furthermore, this finding lends credence to the observation in Amuzu (2013: 20) that the few studies involving bilingual SVCs on West African languages and a European language, Amuzu (Ewe–English: 1998, 2005, 2010), Forson (Akan–English: 1979) and Quarcoo (Akan–English: 2009), have arrived at the same conclusion, that structural possibilities in bilingual SVCs closely relate to structural possibilities in SVCs in the West African languages involved. That is, each of the studies mentioned here has characterized the West African language involved as the language of grammatical structure (or ML) of the bilingual SVCs. In fact, it is not entirely unexpected to find that Igbo controls both morpheme order and the supply of late system morphemes in the bilingual SVCs. Although verb serialization occurs in all types of languages (Payne 1997), they are, however, widespread in West African, South-­east Asian, Creole, Amazonian, Oceanian and New Guinea languages (Aikhenvald 2006). Igbo happens to be one of the many West African languages in which SVCs are rampant compared to a language like English. Therefore, Amuzu (2013: 20) is correct in stating that bilingual SVCs are possible in CS only when the ML has SVCs, as is the case in Igbo, the ML in Igbo– English CS.

7.6  Quantitative analysis 7.6.1  Testing the MLP As in section  6.3.1, we shall now present the results of a quantitative analysis conducted in order to test the MLP. The analysis applies to the bilingual clauses containing the 547 singly occurring EL verbs described in section  5.6, and exemplified in the immediately preceding sections. Our two criteria (see sections 6.2.1 and 6.2.2) for identifying the ML were applied simultaneously to the mixed constituents in order to test the MLP, according to which it is always possible in classic CS to identify the ML of a mixed constituent. The application of both criteria led to the results given in Table 7.3. As we outlined in section 4.4.4, Igbo uses a variety of bound affixes to mark the grammatical categories of tense, aspect and mood. Equally, English develops

154

Codeswitching in Igbo–English Bilingualism

Table 7.3  ML according to source of morpheme order and late system morpheme criteria Singly occurring EL verbs in Igbo–English CS (Total = 547) Patterns of CS

Source of morpheme order

1. E L verbs + ML verbal inflections, as in Either (346) 63% (133) to (139)

Source of late system morphemes + Igbo (346) 63%

+Igbo (118) 21.5%

+Igbo (118) 21.5%

  ‘do/make’ type constructions

+Igbo (64) 12%

+Igbo (64) 12%

  non-‘do/make’ type constructions

+Igbo (19) 3.5%

+Igbo (19) 3.5%

Totals

(547) 100%

(547) 100%

2. M L PSCs + EL verbs + ML verbal inflections, as in (147) to (150) 3. EL verbs in SVCs

a variety of bound morphemes for verbs to mark semantic notions and grammatical categories such as number, tense and aspect. However, as the results in Table 7.3 show, there is no instance in our corpus where an English inflectional morpheme (i.e. late system morpheme) is found with any singly occurring EL verbs. Instead, all the singly occurring EL verbs in non-SVCs (464/547 or 84.5 per cent; see the examples discussed in sections 7.2 and 7.3 above) are obligatorily inflected with Igbo (the ML) verbal morphology. It is equally important to note from the results in Table 7.3 that there is no instance in the entire Igbo–English corpus where an Igbo verb is inflected with verbal morphology from English. This finding is similar to the reports from other CS studies involving a West African language and a European language (see the studies reviewed in section 1.3). For instance, Asilevi (1990: 32) reports that while English verbs in his corpus occur with Ewe tense, aspect and modal morphemes, Ewe verbs do not occur with English tense, aspect and modal morphemes. Furthermore, this finding lends credence to Myers-Scotton and Jake’s (2014) observation that in intrasentential CS it is EL non-­finite verbs that receive ML verbal inflections. Besides, recall that we outlined earlier in section  6.2.1 that these bound morphemes are the type that must come from the ML only. As the results from Igbo–English CS confirm, all necessary late system morphemes come from Igbo, the ML. The importance of the source language of verb inflection in our study cannot be overemphasized; especially since the morpheme order criterion has no purview over 63 per cent (346/547) of the examples attested in the Igbo–English

Embedded Language Single Words: Verbs

155

data. This is not entirely unexpected. As we indicated in section  4.4.8, Igbo and English are typically (SVO) languages. Consequently, 63 per cent of all singly occurring English verbs in our data occurred in bilingual clauses exhibiting no conflict in word order between Igbo and English as far as the switched elements are concerned. This situation is reflected in examples (133) to (139). Recall that such examples do not falsify the MLP as long as the late system morphemes come from the language identified as the ML. And this is exactly the case in Igbo–English, as can be seen in Table 7.3. Therefore, in terms of testing the plausibility of the MOP (see section 6.2.1), the prediction is that when there is a conflict in word order, the ML order wins out (Myers-Scotton 2006: 255); the qualitative analysis of the bilingual clauses containing the EL verbs that occur with Igbo PSCs and verbal inflectional morphology (see section  7.3 above) and the results from Table  7.3 (118/547 or 21.5 per cent) reveal that Igbo contributes both word order and late system morphemes in the examples. Specifically, the order of only one of the languages in contact in the bilingual clauses (Igbo) prevails wherever there is a conflict in word order. Call to mind that we have already identified the same language (Igbo) as the ML of the bilingual clauses containing the singly occurring English nouns (see the previous chapter). The same is true for the 83 EL verbs that occur in the uniquely Igbo bilingual SVCs (see section 7.5 above). The results in Table 7.3 confirm that the MLP is comprehensively supported in the bilingual SVCs. However, much earlier, Myers-Scotton (1993b) and Myers-Scotton and Jake (2001) suggested that bare forms occur due to a lack of sufficient congruence between EL forms and their ML counterparts, leading to the activation of a compromise strategy with the result that the EL content morpheme is not placed in a slot projected by its ML counterpart; rather, it is realized as a bare form. Congruency, as we have already noted in this study, is very important in the MLF model. With reference to the analysis presented here, Amuzu’s (2013) definition of congruence is very apt. He states that ‘singly occurring content morphemes from the EL can be inserted into constituents framed by the ML only if they show sufficient congruence with their ML content morpheme counterparts at the three levels of abstract lexical structure: lexical-­conceptual structure, predicate–argument structure, and morphological realization patterns’ (Amuzu 2013: 33). Taking the EL verb deliver in (160) as an example, we find that it is sufficiently congruent with its ML counterpart nye ‘to give’. Regarding lexical-­conceptual structure, both encode identical concept – i.e. they have analogous lexical-­conceptual structure (to deliver a speech = to give a speech). They also share predicate–argument

156

Codeswitching in Igbo–English Bilingualism

structure – i.e. they both assign a thematic role to the NP which encodes the entity ‘delivering/giving the speech’. Concerning morphological realization patterns, both are verbs. Igbo already has a syntactic model (SVC) that may be used to insert bare verbs in its grammar; therefore, regarding surface realization pattern, deliver is inserted into the verbal slot intended for the bare ML verb nye, as in the grammatical example below. (166) o.-biaV1-ra deliverV2 (nyeV2) speech gaaV3 leeV4 that new unit … 3SG.CL-come-IND deliver speech go look that new unit ‘He came, delivered his speech and went to see that new unit …’ [43:32]

The problem we have with Myers-Scotton and Jake’s (2001) suggestion for why bare forms occur in relation to Igbo–English CS is that, as we saw in the preceding analysis, the EL bare verbs, like the EL bare nouns earlier (see section  6.2.3), do not seem to be inserted as compromise strategies. That is, Igbo verbs can already occur as bare forms in SVCs without the need for positing any special strategies for the insertion of bare EL verbs. Yes, in some sense, it might be argued that those EL verbs inserted in the SVCs with what MyersScotton and Jake (2014: 11) refer to as ‘do’ constructions (64/547 or 12 per cent) can be viewed as some sort of a compromise strategy. An example of this type of strategy utilized to integrate EL verbs in Igbo–English can be found in (157) to (159) and (161) to (165) above. According to Myers-Scotton and Jake (2014: 11–12), the do construction is found in many data sets across typologically diverse languages. Furthermore, they add that the do construction consists of an ML verb form that encodes the meaning ‘do’ (see for instance the Igbo verb mee ‘do’ in the examples above), but is largely bleached of any meaning; the critical meaning in the clause (according to Myers-Scotton and Jake) comes from a non-­finite verb, often the infinitive, in the predicate called by the ‘do’ verb. All necessary ML verbal inflections occur with the ‘do’ verb and not the non-­finite EL verbs (Myers-Scotton and Jake 2014: 11–12). In connection with Igbo–English, we generally agree with Myers-Scotton and Jake (2014) that in the bilingual SVCs the Igbo do verb receives all necessary ML verbal inflections, thus leaving the EL verbs bare. Nevertheless, we are not entirely convinced that this is a particularly unique CS strategy since the same syntactic model already exists in monolingual Igbo grammar. For instance, consider the monolingual Igbo SVC below. (167) ha me-­re m̩ mem̩ me gbaa egwu Osadebe They do/make-IND celebrate dance song Osadebe ‘They celebrated and danced to Osadebe’s song.’

Embedded Language Single Words: Verbs

157

As is evident in (167), the speaker’s intentions call for a past tense marking, but the full verbs of Igbo occur as bare forms; the past tense inflection is placed on the Igbo do/make verb instead (see also the examples in Amaechi 2013). In Igbo, the verb mee is usually an active verb requiring a direct object as its complement. Like any other Igbo verb, mee occurs with the appropriate affix to mark grammatical relation, such as the infinitive i-­me, the participle e-­me, the negative e-­me-ghi, the indicative me-re, the perfective e-­me-la, the imperative me-­e and the inceptive me-­we (see Eze 1997: 92). Given that all the bilingual SVCs adhere to a strategy already in place in Igbo grammar, our position is that the same language production process implicit in the production of the EL verbs with Igbo morphology (see sections 7.2 and 7.3) is also at work in the bilingual SVCs. Like the case of the EL verbs that occur with the Igbo PSCs (see section 7.3), the morpheme order criterion ensures that the bare EL verbs are placed in slots intended for their ML counterparts; while the system morpheme criterion ensures that other ML verbs in the SVCs receive all necessary verbal inflections, thereby leaving the EL verbs bare as required by Igbo grammar. Overall, we have demonstrated, in the preceding sections, through patterns of affixation and the integration of EL bare verbs in bilingual SVCs that the MLP of the MLF is supported in Igbo–English CS.

7.6.2  Testing the AP As we outlined in sections 3.2.4.2 and 6.3.2, this principle states that bilingual speech is characterized by asymmetry in terms of the participation of the languages involved in CS. Essentially, the principle predicts that, in classic CS, only one of the participating languages is the source of ML. As in the previous chapter, we can explore this asymmetry in two specific ways: (1) asymmetry in the source of late system morphemes; and (2) asymmetry in the resolution of word order conflicts in mixed constituents. To do this, we can make use of the analysis reported in the previous sections and summarized in Table  7.3. If the AP is to be upheld, then we would expect the vast majority of bilingual clauses where there is a conflict in word order to have either Igbo or English as unequivocally the ML. According to Myers-Scotton (2006: 255), when this happens, the ML order wins out (see also section  6.2.1). Therefore, all the singly occurring EL verbs in structures where the word order is compatible with that of both languages (i.e. in structures where there is no conflict in word order between Igbo and English) are excluded from the analysis presented here.

158

Codeswitching in Igbo–English Bilingualism

Regarding asymmetry in source of late system morphemes, the results from Table  7.3 revealed that Igbo contributed all (100 per cent) the late system morphemes in the clauses ontaining the 547 singly occurring EL verbs. Concerning asymmetry as to whose word order prevailed in the 201 examples (118 EL verbs inserted with Igbo PSCs+verbal inflections and 83 EL verbs in SVCs) exhibiting a conflict in word order between Igbo and English, the analysis reported in Table 7.3 showed that without exception Igbo order rather than that of English prevailed in all such cases. As in the previous chapter, the results here provide comprehensive support for the AP. The role of the EL (English) in Igbo– English CS, as observed in the analyses reported in this and the previous chapter, seems to be limited to supplying content morphemes (nouns, adjectives and verbs), which are inserted in clause structure according to restrictions imposed by the ML, Igbo.

7.6.3  Testing the USP The USP, as we saw in sections 3.2.4.3 and 6.3.3, states that in bilingual speech the structures of the ML are always preferred, but some EL structures are allowed if certain conditions are met (Myers-Scotton 2002: 8–9). This principle, as we noted earlier, stems from the 4-M model, which makes a distinction between content and system morphemes, dividing the system morpheme into three subtypes: early system morphemes, bridge late system morphemes and outsider late system morphemes. We have already seen (in this and the last chapter) that late system morphemes can only come from the ML of a clause. However, this principle goes beyond to predict that other system morphemes, such as early system morphemes, will be drawn preferentially from the ML of a bilingual clause. While the principle does not exclude the occurrence of EL early system morphemes, it, however, states that the occurrence of EL early system morphemes in mixed constituents is less likely than early system morphemes from the ML. As it turns out, this is precisely the case in Igbo–English CS, considering that in all the bilingual clauses analysed here there is not a single instance where either an English determiner or pronoun (both early system morphemes) is found to co-­occur with a ML verb. Instead, 118 EL verbs (see Table 7.3) are found with the uniquely Igbo PSCs, which are described in MyersScotton (2002) as early system morphemes. This finding thus confirms the claim by Jake et al. (2002: 72) that in bilingual speech, the structures of the ML are always preferred. That is, in Igbo–English CS, Igbo is the source of ML, which provides the morphosyntactic frame for ML+EL constituents where Igbo

Embedded Language Single Words: Verbs

159

system morphemes (both early and late) can appear freely, but even English early system morphemes, which are not ruled out by the USP, can hardly appear in verbal EL+ML constituents. The strict constraints imposed on EL system morphemes by the ML grammar can account for the morphosyntactic features reflected in the bilingual examples analysed in this chapter.

7.7  Summary and conclusion As in the previous chapter, we have employed the MLF model to examine how singly occurring English verbs are inserted in otherwise Igbo utterances. The investigation of patterns of affixation on singly occurring English verbs embedded in Igbo (see sections  7.2 and 7.3) revealed that the 464 English verbs in non-SVCs obligatorily received Igbo verbal inflectional morphology (recall that these affixes are defined as late system morphemes in this study, and thus must be supplied by only the ML). The only singly occurring EL verbs (total = 83, see Table 7.3) that remained bare (i.e. without Igbo morphology) occured in SVCs. Only in SVCs can a full Igbo verb appear without verbal inflectional morphology (see sections 4.4.4 and 4.4.4.1). We noted in section  7.5.1 that most of the EL verbs in SVCs (64/83) co-­ occured with the Igbo verb me ‘do/make’ as the verb which carried the inflections marking grammatical relations of the bilingual clauses. ‘Do-­type’ constructions, as we explained in section 7.5.1, know no typological or geographic limits and tend to co-­occur with bare EL verbs in our data, as a way of integrating these verbs. However, we disagreed with the view expressed by Myers-Scotton and Jake (2001) – that is, in relation to Igbo–English – for why bare EL verbs are inserted in Igbo with ‘do-­type constructions’. We showed that Igbo already possessed a syntactic model for inserting bare verbs in its grammar. Igbo does this through SVCs. Therefore, we wondered whether positing ‘do construction’ for Igbo as a necessary CS-specific compromise strategy due to incongruence between the tense/aspect systems of Igbo and that of English was really necessary. Instead, our view is captured in the statement by Amuzu (2013: 20) that bilingual SVCs are possible in CS only when the ML has SVCs. This is exactly the case in Igbo, the ML in Igbo–English CS. Moreover, it would seem that switching between EL content words (verbs) and ML bound morphemes (Igbo verbal inflectional morphemes) represents serious violations of the FMC (Sankoff and Poplack 1981) reviewed in section 2.3 of this study. Recall that we cited many examples in section  2.3 from studies

160

Codeswitching in Igbo–English Bilingualism

before this one, which appear to falsify the FMC proposal. This CS proposal predicts that a switch may not occur between a bound morpheme and a lexical form unless the latter has been phonologically integrated into the language of the bound morpheme. Therefore, for Poplack and her associates all 464 singly occurring EL verbs with Igbo morphology would automatically be labelled as some kind of borrowings because they pattern like monolingual Igbo verbs. In fact, a similar fate will also befall the eighty-­three singly occurring EL verbs in the bilingual SVCs (even though not one of the EL verbs is listed in the Igbo dictionary). However, utilizing spectrograms to examine the syllable structure of a sample of the EL verbs with Igbo morphology we were able to demonstrate that these words were not as phonologically integrated as borrowed or native words of Igbo. For instance, the spectrograms (in sections 7.4.1.1 and 7.4.1.2) revealed that the speakers do not insert any vowels after syllable boundaries with coda consonants or inside consonant clusters as a way of making the EL verbs conform to the rigid two-­syllable (V/CV) structure of Igbo. We observed that any adaptations to the EL verbs occured only on either the left or right edge depending on whether a ML verbal inflectional prefix or suffix was attached. In addition, the analysis in section 7.4 revealed that the vowel of the Igbo verbal inflectional affixes must look to their closest vowel in the English verbs with which they occur to determine their own [±ATR/±Tense] value, implying that a single phonological system must have simultaneous access to vowels in morphemes from two different languages. This observation, similar to the analysis in Legendre and Schindler (2010: Urban Wolof–French CS), would appear to contravene MacSwan’s (2009) PFIC-based account; which states that two phonologies cannot mix in CS. Finally, as in the previous chapter, application of the AP to our data of singly occurring EL verbs (in section 7.6.2) revealed comprehensive support for the principal theoretical notion that there is an ML–EL hierarchy, because Igbo and English do not both satisfy the roles of the ML according to our two criteria (morpheme order and late system morpheme) for identifying the ML of each bilingual clause. As in the case of the EL nouns and adjectives, Igbo supplies the frame-­building elements into which singly occurring English verbs are inserted in Igbo–English CS. Similarly, the test of the USP (in section 7.6.3) indicated that even English early system morphemes, which are not prohibited by this principle, do not occur in the mixed constituents analysed in this chapter. Thus, the findings from this chapter seem to corroborate our earlier observation from the previous chapter that Igbo–English represents a ‘classic’ case of CS. That is, in the bilingual clauses examined across the two chapters, only one of the

Embedded Language Single Words: Verbs

161

participating languages (Igbo) is the source of the morphosyntactic frame for the clauses. The goal in the next chapter, then, is to see whether the switching of EL islands patterns just like single-­word switching in Igbo–English CS. If this is found to be the case, then it would constitute additional motivation to accept that the singly occurring EL words just analysed in this and the previous chapter are genuine CS forms rather than some kind of borrowings.

8

Embedded Language Islands 8.1  Introduction In the last two chapters, we have demonstrated that singly occurring EL words in Igbo–English are genuine CS forms rather than some kind of borrowings. Here, we shall focus on EL islands. Unlike singly occurring EL words, which divide opinion among researchers interested in explaining the grammar of intrasentential CS, EL islands are assumed to be examples of prototypical CS (see Andersson 1993; Belazi et al. 1994; Di Sciullo et al. 1986; Eze 1997, 1998; MacSwan 1999, 2000; Mahootian 1993; Poplack et  al. 1987 and so on). These researchers have argued that sequences longer than single words, such as phrases and clauses from another language, clearly constitute examples of CS, and have labelled them either as ‘unambiguous’, ‘typical’ or ‘true’ codeswitches (see Park 2000). On her part, Myers-Scotton (2002: 139) explains that EL islands are full constituents consisting only of EL morphemes occurring in a bilingual CP (i.e. a bilingual clause, as defined in this study) that is otherwise framed by the ML. Also, EL islands typically show structural dependency relations; minimally an EL island can consist of two content morphemes or a content morpheme and a non-­derivational system morpheme (as in tractor+s in example (120)). In other words, EL islands represent a break in the ML frame (Myers-Scotton 2002: 139; 2006: 261). In the coming sections, we shall present an analysis that shows that a particular type of EL island in Igbo–English CS patterns very much like the singly occurring EL words analysed in Chapters 6 and 7. That is, I agree with Myers-Scotton (2006: 264) that EL islands must follow the well-­formedness rules of the EL within the islands (i.e. order within the phrase); however, they follow the placement rules of the ML within the clause. Also, recall that a number of the examples in the previous two chapters included both singly occurring EL words and EL islands. We did not offer any analysis of the EL islands because the focus then was on EL single words. In the sections that follow, we shall repeat and analyse a number of those

164

Codeswitching in Igbo–English Bilingualism

examples in addition to some new examples from Igbo–English. However, before presenting the analysis, it is important to remind the reader that the morpheme order criterion will only apply wherever the surface structures of our two languages differ (see section  6.2.1 for our definition of the morpheme order criterion in relation to the analysis reported in this study). In all cases, we expect all verbal inflections (late system morphemes) of the matrix clauses to come from only one of the languages (i.e. from Igbo, the ML). If the late system morpheme(s) of a given matrix clause comes from both Igbo and English, then the MLP could be falsified (see section 6.2.2).

8.1.1  Types of EL islands in Igbo–English codeswitching As was indicated in section 5.6, 26 per cent (696/2,659) of the CS examples in Igbo–English consist of EL islands. Of the 696 examples, 38 per cent (264/696) occur as internal EL islands, while the majority of the examples (62 per cent or 432/696) occur as non-­internal EL islands. The analysis and discussions are divided into six sections. Section 8.2 focuses on what Myers-Scotton (2002: 149, 2006: 265) terms ‘Internal Embedded Language Islands’. Internal EL islands include an ML element to frame the EL phrase. In other words, such EL islands are part of a larger phrase framed by the ML. Section 8.3 deals with non-­internal EL islands (i.e. full constituents entirely in the EL but dominated by a ML clause). In sections  8.4 and 8.5, we shall discuss the results of a quantitative analysis that applies to the AP and the USP. This is followed by section 8.6 where we discuss some of the suggested motivations for EL islands in CS. Section 8.7 is the summary and conclusion to the chapter. All the examples analysed in this chapter qualify as EL islands because the individual elements that make up the islands display a hierarchical structure, with some words ‘higher’ than others in the structure in which they occur (Myers-Scotton 2002: 139, 2006: 261).

8.2  Internal EL islands 8.2.1  EL NPs + Igbo determiners Recall from the analysis of the singly occurring EL nouns in Igbo–English CS that Igbo determiners typically follow the nouns they modify, while the reverse is usually the case in English (see section 6.2.1.1). In light of this information, consider the following examples from Igbo–English CS.

Embedded Language Islands

165

(168) house of assembly members niile me-­re reject bill governor1 house of Assembly members all do/make-IND reject bill governor ‘All (the) house of assembly members rejected (the) governor’s bill.’ [26:13] (169) local government chairman anyi. cho.-ra i.-ru. u. lo.o.gwu. local government chairman our want-IND INF-build hospital ahu. na obodo ha that PREP village their ‘Our local government chairman wanted to build that hospital in their village.’ [6:6] (170) ha ku. wa-­si.-ri. ballot box dum they break-ENCL-IND ballot box all ‘They completely broke all the ballot boxes.’

[23:26]

As examples (168) to (170) indicate, a number of the EL islands in our data occur as sequences of two or more EL morphemes with post-­posed Igbo determiners as the final modifying elements in the DPs. All the EL elements qualify as islands because: (1) they are phrases within bilingual clauses (they are NPs); and (2) their words show structural dependency relations that make them well-­formed in the EL (English). Concerning the last point, the English words in (168) to (170) follow the order of English within the islands. Also, we notice that the English words within the islands show a hierarchical structure. For instance, members is the head of the nominal sequence in (168), where it is pre-­modified by house of assembly; the head noun chairman is pre-­modified by the noun government and the adjective local in (169) and the head noun box is pre-­ modified by the noun ballot in (170). However, in all the examples, the EL morphemes are part of full DPs dominated by Igbo determiners: the quantifiers niile and dum ‘all’ in (168 and 170), and the pronominal determiner anyi. ‘our’ in (169). Significantly, all the mixed constituents support the morpheme order criterion because with the post-­posed Igbo determiners the full DPs now have a complement–head order. This, as we have already demonstrated in Chapter 6, is the usual order in Igbo (the ML). This order contrasts with that of English (as reflected in the monolingual English translations). We also observe in example (170) that ballot box is not inflected for plurality in the bilingual DP. This is a mirror image of the situation with the singly occurring English nouns analysed in section 6.2.1.1. We observed there that it would seem that the presence of the Igbo quantifiers in the mixed DPs blocks the appearance of English plural inflections on the nouns. The same explanation applies here. Recall from the discussion in section  4.4.1 that we noted that in Igbo nouns are not usually

166

Codeswitching in Igbo–English Bilingualism

inflected for number or declined for case. Therefore, the restrictions imposed by the ML grammar can account for the loss of the EL plural morpheme on box in (170), as in the singly occurring nouns earlier. Moreover, the late system morphemes (verbal inflections) of the matrix clauses come from Igbo only in the form of the indicative affirmative suffix -rV marking past tense and the infinitive prefix i.- on ru. ‘build’ (to build) in (169). Consequently, by satisfying both criteria (morpheme order and late system morphemes), we would identify Igbo as the ML. Although Eze (1997: 162) earlier claimed that NP/DP-internal switching is prohibited in Igbo–English CS because this position is a non-­equivalence site according to the EC, the above examples clearly falsify this claim. In fact, NP/DP-internal switching is highly frequent in Igbo–English CS (see the results of the analysis summarized in Table 6.2). What is more, the internal EL islands in the examples above pattern exactly like the singly occurring English nouns in otherwise Igbo utterances analysed in Chapter  6. This is one more reason to accept the singly occurring EL lexical items as genuine examples of single-­word switching. Further examples are provided in the sections below.

8.2.2  EL NPs + Ø determiner A few of the EL islands occur in contexts where a determiner is obligatory in English for a full DP but not in Igbo, as in (171) below. (171) finance committee chairman ga-­a-wu. -zi. speaker ma finance committee chairman AUX-V-become-ENCL Speaker after ikpe ya bie case his conclude ‘(The) finance committee chairman will become (the) Speaker after his case is concluded’ [49:13]

In (171), the nominal sequence is headed by the noun chairman, which is pre-­ modified by the nouns finance and committee. We can also observe that the single English noun Speaker in the same example is without any overt determiners in the bilingual utterance. Interestingly, the EL island in (171) patterns just like the lone English noun and the 144 bare nouns analysed earlier in section 6.2.3. It would seem that this EL island is treated no differently from the singly occurring EL nouns with covert determiners by the speakers. We already know from the preceding section and the analysis in Chapter 6 that Igbo determiners are not usually pre-­posed to the noun as is the case in English. Therefore, we can

Embedded Language Islands

167

safely say that the examples support our two criteria for identifying the ML of each bilingual clause. Support for the system morpheme criterion is found in the bound future auxiliary verb ga- in (171) which comes from Igbo. This bound auxiliary verb is a late system morpheme (according to the definition adopted in this study), because it is required by the overall proposition that the clause expresses (i.e. futurity).

8.2.3  EL NPs + Igbo adjectives There are only two examples of internal EL islands in the data involving English NPs and a post-­posed Igbo adjective. Both examples involve the Igbo true adjective o.hu. u. ‘new’, as we see in examples (172) and (173) below. (172) ndi. Shell discover-ra oil wells o.hu. u. na Bonny people of Shell discover-IND  oil wells new  PREP Bonny ‘People of Shell discovered new oil wells in Bonny.’ [1:27] (The Shell Oil Company discovered new oil wells in Bonny.) (173) railway lines o.hu. u. si na Abuja ga-­a Jos na Kaduna railway lines new go PREP Abuja go-HAB Jos CONJ Kaduna ‘(The) new railway lines go from Abuja to Jos and Kaduna.’ [41:10]

Recall from the discussion in section 4.4.2 that Igbo has a very limited number of true adjectives and that these adjectives always follow the nouns they modify (see also the analysis in section 6.2.1.3). In both (172) and (173), the adjectives function attributively. That is, they post-­modify the EL NPs oil wells in (172) and railway lines in (173) respectively. As it turns out, similarly to the observation from section 6.2.1.3, the two examples support both the morpheme order and system morpheme criteria. The morpheme order resembles that of Igbo (i.e. post-­posed adjective as the final modifying element in the NP/DP) and the verbal inflections come from Igbo only. By satisfying both criteria, Igbo is determined to be the ML of the two examples. Similar to the example in (171), the demonstarative determiner in (173) is covert (i.e. zero). So, the DP in (173), as in all the other examples, is only well formed in the ML but not in the EL.

8.2.4  Igbo preposition + EL NP/DP complements Other EL islands in Igbo–English occur as NP/DP complements in prepositional phrases (PP) headed by the Igbo multi-­purpose preposition na. Recall from section 4.4.3 that we outlined that Igbo has just the lone preposition na, with the

168

Codeswitching in Igbo–English Bilingualism

meaning: ‘at, in, on, on top of, on the outside, within, inside of ’ and so on. The specific meaning of the Igbo preposition na is derived contextually, as in examples (174) to (176) below. (174) i.-ga-­a-bi.a i.-hu. councillor ha na opening ceremony? CL-AUX-V-come INF-see councillor PRN/D PREP opening ceremony ‘Will you come to see their councillor at (the) opening ceremony?’ [30:14] (175) ma ha gba-­ra u. lo. vice-­chancellor o.ku. na the riots but they burn–IND house vice-­chancellor fire PREP the riots ‘but they burned (the) vice-­chancellor’s house in the riots’ [11:28] (176) maka  na   ha register-ra na the wrong desk … C C they register-IND PREP the wrong desk ‘because they registered at the wrong desk …’ [17:30]

The PPs in the above examples express locational relations in space. Also, the order in the EL islands is compatible only with that of the EL and the words display structural dependency relations. For instance: in (174), the EL island opening ceremony is headed by the noun ‘ceremony’ and modified by the noun ‘opening’; in (175), the noun riots is pre-­modified by the definite determiner the; and in (176), the noun desk is pre-­modified by the adjective wrong and the definite determiner the, the head of the DP island. Because the structure of EL islands is often formulaic, Backus (1999) refers to such EL islands as ‘chunks’. He defines a chunk as a conventional unit, stating that reference to conventionality is crucial: a chunk must be current. Novel combinations are put together on the spot, but chunks are not (Backus 1999: 94). Backus’s notion of chunks is based on the work of Langacker (1987) in the field of Cognitive Linguistics. The idea of EL islands as chunks is very appealing because it seeks to offer some kind of generalization. However, it has been criticized for being too broad a term since it hinges on the assumption that every multi-­word/morphemic EL insertion is a ‘chunk’ (Backus 1999: 97). In fact, Myers-Scotton (2002) points out that the main problem with the notion of chunks or complex units is that there are no clear boundaries between those combinations that are conventional and those that are not. Additionally, Backus’s ideas about how certain combinations are assembled are open to dispute. For instance, he (Backus 1999: 97) suggests that if two or more EL morphemes are used in CS and they form a conventional combination in the EL, then it would be too coincidental if the speaker had produced them by two or more independent switches, compositionally building up a composite expression. From this, it is quite possible that idioms, like irregular plurals and irregular past tenses in English (and other languages), may well be contained

Embedded Language Islands

169

in single lemmas and therefore are not compositionally assembled. This notwithstanding, it is difficult to suggest that an EL island such as inside the parlour in (189) below is accessed as a unit when similar phrases such as in/ inside/outside the parlour exist in the English language. Nevertheless, since there is no conflict in word order between Igbo and English with regard to the placement of the PPs in the bilingual clauses (see section 4.4.3), we have coded all such examples as ‘Either’ according to the morpheme order criterion. However, Igbo alone supplies the late system morphemes in the form of the bound future auxiliary verb ga- and infinitive prefix i.- in (174) and the indicative affirmative past tense suffix -ra in (175 and 176) respectively. Moreover, it is evident from the preceding analysis that the same language production procedures underlying the switching of the EL single words analysed in the last two chapters are also in control of the switching of internal EL islands in Igbo–English CS. That is, we agree with Myers-Scotton (2002) that in intrasentential CS the two languages involved are both switched on (i.e. the ML and EL); the ML is more activated but the appearance of EL content and early system morphemes are not banned as is evident in single word and EL island switching. Yet, in both types of switching, it is the ML that supplies the syntactically relevant grammatical elements (i.e. late system morphemes) and projects placement of the switched elements within the clause. This, as the analysis of the internal EL islands in Igbo–English CS confirms, is exactly the case. Next, we present the results of a quantitative analysis conducted in order to test the MLP. The analysis applies to the bilingual clauses containing the 264 (out of 696) internal EL islands illustrated in the immediately preceding sections.

8.2.5  Quantitative analysis The results of the analysis (summarized in Table 8.1) appear to corroborate the findings from the previous two chapters. Generally, the results confirm that: (1) the predictions of the MLP of the MLF model are reflected in the sample of internal EL islands in the Igbo–English data; and (2) the Igbo–English data represents a classic case of CS. That is, only one language (i.e. ML = Igbo) supplies late system morphemes of the matrix clauses and wherever there is a conflict in word order, this is resolved in favour of the ML. Specifically:

Morpheme order criterion 1. + Igbo, if morpheme order is specifically Igbo as in: EL NPs + Post-­posed Igbo determiners (examples 168–170); Bare EL NPs + Post-­posed covert

170

Codeswitching in Igbo–English Bilingualism

(zero) determiners (example 171); and EL NPs + Post-­posed Igbo adjectives (examples 172–173). 2. Either, if there is no conflict in word order between Igbo and English as in: Igbo PPs + EL NP/DP complement (examples 174–176).

System morpheme criterion 3. + Igbo, if the late system morpheme(s) of the matrix clauses come from Igbo only as in (examples 168–176). Firstly, Table 8.1 confirms that the late system morphemes (all necessary verbal inflections) of the bilingual matrix clauses come from Igbo only. This result corroborates those from the last two chapters with regard to the analysis of singly occurring EL words in Igbo–English CS. Secondly, in the examples where there is a conflict in word order between Igbo and English (78 per cent or 206/264), the order of the ML (Igbo) prevails (100 per cent or 206/206) over that of English. There is no word order conflict in 22 per cent (58/264) of the examples involving EL NP/DP complements of PPs headed by the Igbo multi-­purpose preposition na. Therefore, the ML can only be determined by the source of the late system morphemes. As the results in Table 8.1 confirm, only one language (Igbo) supplies such morphemes in all the bilingual clauses. Furthermore, the results confirm our earlier finding (in section 6.2.1.3 and Table 6.2) that Igbo indeed has a limited number of attributive adjectives. That

Table 8.1  ML according to source of morpheme order and late system morpheme criteria Internal EL islands in Igbo–English CS (Total = 264) Composition of internal EL islands

Source of morpheme order

Source of late system morphemes

+ Igbo (190) 72% √ (14) 5.3%

+ Igbo (190) 72% √ (14) 5.3%

√ (2) 0.7%

√ (2) 0.7%

Either (58) 22% (264) 100%

√ (58) 22% (264) 100%

1. Mixed DPs   NP+D(post-­posed)   NP+Ø determiner(post-­posed) 2. Mixed NPs   NP+A(post-­posed adjectival modifier) 3. Mixed PPs   P+NP/DP Totals

Embedded Language Islands

171

is, the present result of only 0.7 per cent of the EL NPs post-­modified by Igbo attributive adjectives mirrors the result for singly occurring EL adjectives in Table 6.2. Recall that the Igbo attributive adjectives always occur in post-­position to their nouns (see section 4.4.2). As we observed in section 6.4, while both Igbo and English share the category of adjective, they differ, however, in how attributive adjectives are placed in NPs in relation to the head noun. English attributive adjectives are pre-­modifiers, while they are post-­modifiers in Igbo. This can account for the limited amount of switching in this position in Igbo–English. However, more exploration is needed to confirm this. Nevertheless, on the basis of the results reported in Table 8.1, the MLP of the MLF model is overwhelmingly supported by our sample data of internal EL islands. Not only this, the present analysis corroborates the finding from the last two chapters. That is, only one language (Igbo) consistently supplies the late system morphemes, and wherever there is a conflict in word order between the two languages, the ML (Igbo) order prevails. Next, we consider the examples involving what we refer to (after Myers-Scotton 2002 and 2006) as non-­internal EL islands in Igbo–English CS.

8.3  Non-­internal EL islands In this section, our focus is on the major constituents which are completely in the EL and are also well formed according to the EL morphosyntactic rules. Like the internal EL islands analysed in the previous sections, these also display structural dependency relations which qualify them as islands.

8.3.1  Determiner phrase 8.3.1.1  Direct object DPs/NPs The majority of the non-­internal EL islands occur in standard SVO clauses involving switching between an Igbo verb and an English direct object DP as in (177) to (179) below. (177) o wela-­la the camcorder … he return-PERF the camcorder ‘He has returned the camcorder …’ (178) anyi. a-­ga-ghi. the party conference … we V-go-NEG the party conference ‘We did not attend the party conference …’

[12:9]

[3:20]

172

Codeswitching in Igbo–English Bilingualism

(179) o nye-­re the invitation card he give-IND the invitation card ‘He gave the invitation card before he left.’

tutuu ya a-­la before he left [3:21]

The ML of all the cases represented by the three examples is Igbo according to the late system morpheme criterion only. This is based on the evidence that the bound affixes encoding tense and negation are contributed by Igbo, the ML. Therefore, it is the ML verbs in the examples that project the placement of the EL DP islands within the clauses. However, there is no word order conflict between the two languages in the examples. A number of the examples involve switching between an Igbo indirect object pronoun and an English direct object DP, as in (180). (180) dede m no. Lagos zu. o.-ra m new phone mgbe … elder brother my BE Lagos buy-IND me new phone when ‘My elder brother who lives in Lagos bought me (the) new phone when …’  [21:25]

As we have already stated, examples (177) to (180) are not unexpected given that both Igbo and English are typically SVO languages. However, as in all the examples considered so far, the language of verb inflection is Igbo. Therefore, the ML of the matrix clauses is Igbo according to the late system morpheme criterion only.

8.3.1.2  Subject DPs The second most frequently occurring non-­internal EL islands (see Table 8.2) involve English subject DPs embedded in Igbo, as in (181) and (182). (181) the former petroleum minister wu. onye Awu. sa the former petroleum minister BE person north ‘The former petroleum minister is from the north (i.e. northern Nigeria).’  [10:6] (182) the man e-­nye-re-­ghi anyi. aka … the man V-give-IND-NEG us hand ‘The man did not give us a hand …’ [10:6]

The EL islands in (181) and (182) occur in structures compatible with both Igbo and English. Thus, we can only determine the ML by the late system morpheme criterion. For instance, the linking verb (the copula) wu. in (181) and the bound affixes on the Igbo verb nye ‘give’ in (182) are late system morphemes because they express the speaker’s intentions. These are the type of system morphemes

Embedded Language Islands

173

that must come from only one of the languages participating in classic CS. As it turns out, the relevant system morphemes in all the examples come from Igbo only.

8.3.1.3  Adverbial DP A small number of the non-­internal EL islands occur as adverbial DPs, as in (183) and (184). (183) o.-no. na u. lo. mkpo.ro.  all these years … 3SG.CL-BE PREP house prison all these years ‘He has been in prison all these years …’ (184) nwunye ya hapu. -ru. ya three years ago … wife his leave-IND him three years ago ‘His wife left him three years ago …’

[29:6]

[29:6]

In the Igbo–English data, all the switched adverbial DPs are time adverbials as in the above examples. The source language of all verb inflections in the matrix clauses is Igbo. Besides, the adjunct DP EL islands occur in positions within the clauses where equivalent Igbo time adjunct phrases may also occur. Similar examples are cited in Myers-Scotton (1993b: 27) from Swahili–English CS.

8.3.2  Verb phrase We saw in the previous chapter that apart from the eighty-­three singly occurring EL verbs in SVCs, all 464 lone EL verbs in Igbo–English CS received Igbo verbal morphology. Therefore, it is not surprising to find that the majority of the VP EL islands (see Table 8.2) in Igbo–English are inflected with Igbo verbal inflections, as in the two examples below: (185) ha na-­a-cho. i-destabilize the whole country … they AUX-V-want INF-destabilize the whole country ‘They want to destabilize the whole country …’ [18:13] (186) a-­si. na a-­na-e-build another hospital in your area … CL-said COMP CL-AUX-V-build another hospital in your area ‘They said that they are building another hospital in your area …’ [20:4]

Before analysing the above examples, it is important to point out that in syntax the functional category ‘Inflection’ (I) is usually deployed as an umbrella terminology for the categories found in the structure of a sentence associated with the verb in most languages. These categories include auxiliary

174

Codeswitching in Igbo–English Bilingualism

verbs, tense–aspect–mood, voice, negation, agreement and so on. Under recent generative theories of syntax, such as the Minimalist Program, each of the inflectional categories is assumed to have a separate functional projection: TP (Tense Phrase), AgrP (Agreement Phrase), AspP (Aspect Phrase), NegP (Negation Phrase) and so on. As we pointed out in section  4.4.4, English inflectional categories tend to occur mostly as auxiliary verbs, modals and verbal suffixes, such as -ed, -ing, -en (see Table 4.3). On the other hand, in Igbo, most of the categories are realized as verbal affixes and bound auxiliary verbs ga- and na- (see Table 4.2, the analysis in sections 7.2 and 7.3, and Obiamalu 2013b). We saw in the last chapter that morphology is a vital aspect of Igbo grammar; therefore, some of the inflectional categories found in Igbo–English CS appear to violate the FMC and other similar models of CS that ban switching between EL free morphemes and ML bound morphemes (see sections 7.2 to 7.4.1.2). For instance, in the above examples (185) to (186), like the EL single words (verbs) analysed earlier in sections 7.2 and 7.3, we observe that it is possible to have CS structures where the inflectional categories (bound morphemes) come from Igbo and the verbal base (free morphemes) are from English. Consequently, our view is that once the Igbo–English bilingual speaker directly selects destabilize (in 185, for instance) at the lemma level (or conceptual level), for instance, the lemmas underlying early system morphemes, like the definite article the (whole country), are activated when the lemmas supporting content morphemes (such as the EL verb destabilize) point to them. These indirectly elected lemmas further realize the conceptual content of the semantic/pragmatic feature bundles. For example, in (185) the whole country adds specificity/definiteness to the VP head destabilize. In other words, the same semantic/pragmatic feature bundle activates both the whole country and destabilize; but due to lack of sufficient congruence between the tense–aspect–mood systems of both languages (i.e. Igbo and English), the late system morpheme criterion ensures that the verbal inflections of the matrix clauses come from the ML (Igbo). And the morpheme order criterion acts to place the EL DP island complements of the EL VPs immediately to the right of the EL verbs. As it turns out, this position is compatible with the order of both Igbo and English. However, for Poplack and her associates (Poplack and Meechan 1995, 1998; Eze 1997/1998), the above CS examples are borrowings because they include Igbo morphology; and they pattern like monolingual Igbo VPs (see sections 4.4.4, 7.2 and 7.3). We contend that the examples are genuine EL islands because they show structural dependency relations and, apart from the ML inflections, the VPs follow the order of the EL within the islands. Like the singly

Embedded Language Islands

175

occurring EL verbs with Igbo morphology analysed in the previous chapter, the examples above clearly violate the FMC. We have already demonstrated that the inflected EL verbs are not as phonologically integrated as either borrowed or native words of Igbo (see sections 7.4 to 7.4.1.2). Moreover, the English verbs and the following EL islands are not considered borrowings by Igbo–English bilinguals who use them in CS. Also, Matras (2009) correctly observes that such phrases are not likely to become established borrowings. In other words, the VPs, like the singly occurring EL verbs from the last chapter, do not occur in the speech of monolingual Igbo speakers. As we saw earlier in section 7.5.8, the EL VP head dismantle in example (187) is bare because the verb is one in a series of verbs in an SVC. (187) mu. na nwunye m jiV1 aka  anyi. meeV2   dismantleV3 the leaking roof Me/I and wife   my hold hand our do/make dismantle  the leaking roof ‘My wife and I dismantled the leaking roof with our hands.’ [7:32]

According to Igbo grammar, only in SVCs may a full verb appear without morphology (i.e. bare), as in the above example. It is also apparent in (187) that the first verb in the series is inflected for tense (as required by the grammar of Igbo), thus leaving the EL VP head verb bare. Another interesting observation concerning the EL VPs with Igbo inflections is that while it is possible for English-­ origin verbs to accept Igbo inflections, the reverse is not the case for Igbo verbs. As in the last chapter, there is no instance in the data where an Igbo verb/VP took English inflections. Equally fascinating, as can be observed in the above examples and in the rest of the Igbo–English data, is the fact that the functional category I (Inflection) seems to determine the type of pronoun that can fill its Spec (Specifier) position. That is, if inflection is in Igbo (which is always the case in Igbo–English CS), then the pronoun in the subject position must also be in Igbo (see also Obiamalu 2013b: 111). Thus, it would appear that the occurrence of one functional category within a given structure demands that the other functional categories must be in the same language. In Igbo–English CS, there are occurrences of EL functional categories, particularly the definite article the – determiner. However, such examples occur only as maximal projections within EL islands unaffected by the morpheme order and system morpheme criteria.

8.3.3  Prepositional phrase The third most frequent non-­internal EL islands in the sample are PPs (see Table 8.2), as in examples (188) to (189).

176

Codeswitching in Igbo–English Bilingualism

(188) anyi. ga-­ama-u. zo. pu. -o. in the morning we AUX-ENCL-early leave-V in the morning ‘We will leave very early in the morning.’ (189) ka a-­kpo.ba-­ta ha inside the parlour let V-bring-ENCL them inside the parlour ‘Let us bring them inside the parlour (living room).’

[46:9]

[48:7]

The majority of the PPs occur immediately following either a full Igbo verb (as in 188) or as part of a VP inflected with Igbo morphology (as in 186 above). Others occur following an Igbo pronoun, as in (189). The examples in (186), (188) and (189) are locative PPs in structures where word order is compatible with that of both Igbo and English. Again the EL PPs are entirely well formed according to the grammar of English. It is also interesting to note that other PPs in Igbo–English occur as set expressions (or collocations), as in (190) below. (190) John fe-­re ule ya with flying colours John pass-IND exam his with flying colours ‘John passed his exams with flying colours.’

[12:44]

Looking at the examples above, it would appear that the non-­internal EL PPs function exactly like the internal EL PPs analysed in section 8.2.4. That is, they express locational relations in space or time and often appear as single lexical units or collocations (see Myers-Scotton 2002: 140–1; Backus 1999: 94 and the discussion in section  8.2.4 above). Since there is no conflict in word order between Igbo and English in the examples, the ML of the matrix clauses is determined to be Igbo according to the system morpheme criterion only. As is evident in the examples, all late system morphemes expressing tense/aspect come from Igbo only.

8.3.4  Inflectional phrase As we pointed out in section  4.4.7, Igbo has a number of complementizers (C/COMP) used to introduce various types of embedded clauses in the language. However, before we explore their role in Igbo–English CS, it is important to point out that the Complementizer Phrase (CP) is regarded as the largest syntactic projection (see Myers-Scotton 2002: 54), which takes a sentence (IP/TP) as its complement. According to Myers-Scotton (2013: 42), complementizers qualify as bridges that join together two clauses in a hierarchical relation. Furthermore, she explains that hierarchical structure is a feature of the abstract level of predicate–argument structure. That is, complementizers index

Embedded Language Islands

177

the notion that clause 2 has a structural relationship with clause 1 (ibid.). In Igbo–English, the EL IPs are introduced by Igbo complementizers as in (191) to (193) below: (191) ha maa na this government is full of corrupt people they know C this government is full of corrupt people ‘They know that this government is full of corrupt people.’ [6:3] (192) Kachi a-­maa-ghi. ma her job di. safe na company ha Kachi V-know-NEG C her job BE safe PREP company their ‘Kachi does not know whether her job is safe at their company.’ [9:25] (193) i.-ga-­a-hu. na he shouldn’t be re-­elected 2SG.CL-AUX-V-see C he shouldn’t be re-­elected ‘you will see that he shouldn’t be re-­elected’ [41:13]

It is immediately clear from the examples above that the matrix clauses are in Igbo, while the embedded CPs are in English. Interestingly, we can notice that the complementizers and the IP complements are in different languages. Already, this is a violation of the FHC (Belazi et  al. 1994), which stipulates that the complementizer and its IP complement should come from the same language. From this, we can conclude, similarly to the studies before this one (see the review of the FHC in section  2.5), that the FHC is not supported by evidence from natural CS data. Another interesting observation in Igbo–English CS concerns the examples showing what may be termed double complementizers. This is found in a few of the examples, as in (194) and (195) below; see also (12) and (70). (194) o.-na-­a-ju. whether ma i.-ga-­e-so ha a-­ga u. ka 3SG.CL-AUX-V-ask C C 2SG.CL-AUX-V-follow them V-go church ‘She is asking whether you will follow them to church.’ [5:30] (195) i.-ma that na anyi. na-­a-gbasaa na oge 2SG.CL-know C C we AUX-V-dismiss PREP time ‘You know that we dismiss (finish) on time.’ [2:30]

We have already noted in section 4.4.7 that the Igbo complementizers can co-­ occur in violation of the ‘doubly filled COMP Filter’. Recall that we stated earlier (in section 4.4.7) that English does not permit a doubly filled COMP slot in its syntax, whereas the reverse is the case in Igbo, as examples (194) and (195) confirm. So, what is the implication of this finding? The examples indicate that, in Igbo–English CS, the ML (Igbo) complementizer has dominance over that of the EL (English). That is, the Igbo complementizer can license both Igbo and English IP complements, providing that the Igbo/English IP meets the

178

Codeswitching in Igbo–English Bilingualism

well-­formedness conditions of the Igbo complementizer (see also Obiamalu 2013b). Furthermore, as is evident in the examples, the English complementizer does not license any Igbo IP complements in our data, thus leading to the doubling of complementizers where the Igbo complementizer is closer to the Igbo IP complement. Put succinctly, in Igbo–English CS, it is the Igbo complementizer that determines the following IP complement irrespective of the IP’s language of origin. Additionally, the examples confirm Myers-Scotton’s (2013: 43) claim that complementizers often come only from the ML in some data sets. Like the four examples from Spanish–English cited in Myers-Scotton and Jake (2009: 352) showing that Spanish que ‘that’ can introduce either an English clause or a Spanish clause if it is the dominant ML in the discourse, complementizers from Igbo, the ML, are used by our speakers to introduce both Igbo and English clauses in Igbo–English CS. Not only this, but all the examples support the system morpheme criterion of the MLF model because only the ML (Igbo) supplies the late system morphemes in the matrix clauses. Also, we consider the morpheme order of the examples with double complementizers to be Igbo, because this configuration is alien to the syntax of English.

8.3.5  Adjective phrase Six English ADJPs (see Table  8.2) occur in the Igbo–English CS data, as in examples (196) and (197). (196) ma flats ndi. ahu. di.-cha very expensive but flats D DEM BE-ENCL very expensive ‘but those flats are (extremely) very expensive’ [39:25] (197) ma campaign ya a-­di.-ra-­ghi. very effective But campaign PRN/D V-BE-IND-NEG very effective ‘but his campaign was not very effective’ [31:15]

The adjective phrases, like the 112 singly occurring EL adjectives analysed in section  6.2.1.3 and reported in Table  6.2, function as predicate complements following the Igbo existential copular di.. In all the examples, Igbo is the ML of the matrix clauses according to the late system morpheme criterion. As is clear in the two examples, Igbo is the source of all verbal inflectional morphology.

8.3.6  Adverbial phrase In Igbo–English, the two locative adverbial phrases found in the data are given in (198) and (199) below:

Embedded Language Islands (198) clinic m di. near the townhall ma … clinic my BE near the townhall but… ‘My clinic is near the town hall, but …’ (199) anyi. ga-­e-kick ha niile out of office we AUX-V-kick them all out of office ‘We will kick all of them out of office.’

179

[4:3]

[19:30]

In the examples, Igbo is the ML of the matrix clauses according to the system morpheme criterion. We can tell that Igbo is the ML because the copula (a late system bridge morpheme) in (198) and verbal inflectional morphemes (the future auxiliary verb ga- and the vowel inflectional prefix e-) on the single EL verb kick in (199) are all from Igbo. In sum, the examination of the internal and non-­internal EL islands in Igbo–English CS reveals that primarily the same language production processes (see chapters 6 and 7) underpin both the switching of EL single words and EL islands. Thus, the dichotomy between single word and multi-­word switching is not as wide as some researchers would have us believe. Instead, we agree with MyersScotton (2006: 254) that there is, as it were, a continuum of EL elements in bilingual clauses, with single words as one end point and full phrases/clauses as the other.

8.3.7  Quantitative analysis It is important to reiterate here that the morpheme order criterion has no purview on the configuration of the non-­internal EL islands. Recall that we stated in the introduction to this chapter that the EL island represents a break from the ML. Therefore, when we discuss the morpheme order criterion in this section it applies only to the matrix clause.

Morpheme order criterion 1. + Igbo, if morpheme order of matrix clause is specifically Igbo as in examples 186–187 and 193–195. 2. Either, if morpheme order of matrix clause is compatible with both Igbo and English as in examples 177–185, 188–192 and 196–199.

System morpheme criterion 3. + Igbo, if the late system morpheme(s) of the matrix clauses come from Igbo only as in examples 177–199.

180

Codeswitching in Igbo–English Bilingualism

The preceding analysis and the application of our two criteria (morpheme order and late system morpheme) to the matrix clauses gave rise to the results summarized in Table 8.2. The results in Table 8.2 tend to indicate that there is no conflict in word order between Igbo and English in the vast majority (slightly over 92 per cent) of the matrix clauses. As we have already observed, this is not unexpected because both Igbo and English are typically SVO languages. That is, most of the major constituent switching in our data occurs in structures where surface word order is compatible with that of both languages. It is crucial to remind the reader that these examples are not in direct conflict with the MLP since either language could be the ML. Moreover, recall from section 6.2.1 that we stated that the morpheme order criterion will only apply wherever there is a conflict in word order between the two languages. As the analysis reported in Table 8.2 show, 7.6 per cent (33/432) of the examples in Igbo–English CS occur in structures displaying conflict in word order between Igbo and English. In the thirty-­three examples, the ML (Igbo) order prevailed over that of English in the matrix clauses (100 per cent or 33/33). Similar to the findings from Tables 6.2, 7.3 and 8.1, the results from Table  8.2 confirm that Igbo alone contributes the late system morphemes of the matrix clauses in Igbo–English CS. In other words, there is no instance in Igbo–English where the late system morphemes of the matrix clauses come from the EL (English). Table 8.2  ML according to source of morpheme order and late system morpheme criteria Non-­internal EL islands in Igbo–English CS (Total = 432) Composition of non-­internal EL islands 1. Direct Object DPs Igbo verb+EL DP in SVO clause Igbo verb+Igbo IO+EL DP in SVO clause Igbo verb+EL DP in SVC 2. Subject DPs EL DP+Igbo verb in SVO clause 3. Adverbial DP Igbo DO+EL adjunct DP 4. VP Igbo I+EL VP in SVO clause Igbo do/make construction+EL bare VP

Source of morpheme order

Source of late system morphemes

Either (125) 29%    √ (37) 9% +Igbo (16) 4%

+ Igbo (125) 29% √ (37) 9% √ (16) 4%



Either (61) 14%

√ (61) 14%

   √ (18) 4%

√ (18) 4%

   √ (68) 16%   +Igbo (6) 1%

√ (68) 16%   √ (6) 1%

Embedded Language Islands 5. PP Temporal PP Locative PP Set expressions 6. IP Matrix clause has SVO word order Double COMP 7. AP Predicative complements in SVO clause 8. ADVP Locative ADVP Totals

181

Either (27) 6%    √ (12) 3%    √ (15) 3.5%



√ (27) 6% √ (12) 3% √ (15) 3.5%

   √ (28) 6.5% +Igbo (11) 2.5%



√(28) 6.5% √ (11) 2.5%



Either (6) 1%

  √ (6) 1%

    √ (2) 0.5%     (432) 100%

  √ (2) 0.5%  (432) 100%

Consequently, the results of the analysis from the preceding sections of this chapter validate the results from the previous two chapters. That is, across the three chapters, only one language (Igbo) consistently supplies all the frame-­ building elements, and in every example exhibiting a conflict in word order the order of Igbo (the ML) wins out completely, exactly as the MLP of the MLF model predicts.

8.4  Testing the AP The AP, as we saw in section 3.2.4.2, states that bilingual speech is characterized by asymmetry in terms of the roles played by the two languages involved in CS. We have already demonstrated this asymmetry in connection with the singly occurring EL elements embedded in Igbo (see sections  6.3.2 and 7.6.2). Following the same pattern as in the previous two chapters, here we shall demonstrate that the predicted asymmetry is evident in the foregoing analysis reported in tables 8.1 and 8.2. Regarding asymmetry in the source of late system morphemes, the results from tables 8.1 and 8.2 revealed that Igbo contributed all (100 per cent) the late system morphemes of the matrix clauses. Concerning asymmetry as to whose word order prevailed in the structures exhibiting a conflict in word order between Igbo and English, again the results reported in tables 8.1 and 8.2 revealed that the EL (English) does not control the word order of the mixed constituents. In fact, whenever there was a conflict in word order between the two languages, as in the 206 examples from Table 8.1 and 33

182

Codeswitching in Igbo–English Bilingualism

examples from Table 8.2, the order of Igbo (the ML) prevailed comprehensively (100 per cent or 239/239). What is very significant about the latest finding is that these examples involve multi-­word CS (i.e. phrases), yet they pattern just like the single-­word switches analysed earlier in Chapters 6 and 7. This eliminates any objections which some might have about the results from the analysis of the singly occurring EL words in structures where the ML was identified to be Igbo. Here, in the data excluding single word CS we observe that we have comprehensive support for the AP. In sum, our data does support the AP as defined by Myers-Scotton (2002: 9). As in Deuchar’s (2006) Welsh–English study, our data also supports a quantitative notion of asymmetry, in that Igbo is the ML in all bilingual clauses where this can be unambiguously identified. Furthermore, our data does confirm that there is neither empirical nor any theoretical value in making a distinction between single versus multi-­word switching since both pattern alike in Igbo–English CS.

8.5  The USP Recall from section  3.2.4.3 that the USP states that, in bilingual speech, the structures of the ML are always preferred, but some EL structures are allowed if certain conditions are met (Myers-Scotton 2002: 8–9). That is, this principle does not prohibit the occurrence of early system morphemes (such as determiners) from the EL, only that such morphemes will be drawn mostly from the ML as well. To test this prediction on the sample of EL islands in Igbo–English CS, we decided to focus on the early system morphemes used as determiners in both Igbo and English. As we have already noted in this study, in English, the class of items normally designated as determiners include articles, demonstratives, quantifiers, numerals and pronominal modifiers. There is no direct equivalent of the English definite and indefinite articles in Igbo. However, the language uses two deictic words (a ‘this’ and ahu. ‘that’), six pronominal modifiers (m ‘my’, gi. ‘your’, ya ‘her/his/its’, anyi. ‘our’, unu ‘your’ and ha ‘their’), three quantifiers (dum ‘all’, naabu. ‘two’ and niile ‘all’) and numerals as determiners. If the USP is correct, we would expect determiners in mixed constituents to be drawn preferentially from the ML rather than from the EL. We have already established in Chapters 4 and 6 that determiners precede the nouns they modify in English, whereas in Igbo they follow their nouns. Parallel to the finding from Chapter 6 (see the results in Table 6.2) examples where the determiners are in English and the lexical nouns/NPs are in Igbo are not found in Igbo–English CS

Embedded Language Islands

183

across the three chapters (i.e. Chapters 6 to 8). This is not unexpected since the order of Igbo, the ML, is N/NP followed by determiner. That is, the strict constraints on EL system morphemes by the ML well-­formedness requirements can account for the morphosyntactic features reflected in the configuration of the internal EL NP islands post-­modified by Igbo determiners reported in Table  8.1. Put simply, the findings from the analysis in this and the last two chapters confirm that in Igbo–English CS the system morphemes (both early and late) always come from the ML, Igbo, except when the English expression within Igbo–English CS is a maximal projection of its own, in which case such an expression is an EL island (se also Obiamalu 2013b). Consequently, the prediction of the USP is reflected in Igbo–English data across the analysis presented in this and the last two chapters.

8.6  Motivations for EL islands According to Myers-Scotton (2002: 142), some researchers prefer to use the notion of ‘triggering’ to explain EL islands. For instance, much earlier Clyne (1967: 84) suggested that certain words may constitute an overlapping area between two languages which causes a speaker to lose his linguistic bearings. In his language use study of Dutch and German immigrants in Australia, he stated that a word in one language seems to trigger a switch to another language, in cases where that word (the trigger) is phonologically alike with a counterpart in the second language or otherwise ambiguous. He identifies in as a common trigger word for the group in his study. Hence, he reports that in triggers more English. The term (trigger) has also been used more broadly by researchers to state that triggering occurs if the presence of any EL morpheme is followed by another EL morpheme or phrase. However, Myers-Scotton (2002) explains that triggering is not something that the MLF model provides for, although neither does it disallow it. This said, she states that quantitative studies of the effects of supposed triggering on the segment that follows are lacking, and what evidence there is goes against triggering as a very salient factor in CS. Myers-Scotton’s view on triggering is supported by evidence from Igbo–English CS. For instance, the EL VPs analysed in section 8.3.2 occur with following EL elements. Thus, the presence of the EL verbs in examples (185), (186) and (187) could be viewed as triggering the subsequent EL islands: the DP the whole country in (185), the NP another hospital in your area in (186) and the DP the leaking roof in (187). The notion of triggering is very attractive as a way to explain

184

Codeswitching in Igbo–English Bilingualism

the nature of these VP islands. Nonetheless, a quantitative analysis of all the EL verbs attested in Igbo–English CS provides only partial support for the supposed triggering on the segment that follows. For example, of the 621 EL verbs in Igbo–English (i.e. including both singly occurring EL verbs and those in EL VP islands), 88 per cent are not followed immediately by other EL element(s). Triggering is observed in only 12 per cent of the examples. Thus, in the overwhelming majority of the examples in our data triggering is not a very salient factor motivating the switching of EL islands (see also Myers-Scotton 1993b: 203). Nevertheless, we must concede that the focus of this study is not on testing the notion of triggering. Additionally, the cited examples (the EL verbs) are not true tests of triggering, at least as Clyne intended. However, the observation is revealing about the shortcomings of the notion of triggering in relation to the linguistic behaviour of Igbo–English bilinguals. Also, as we noted earlier in section 8.2.4, Backus (1999) refers to those EL islands with structures similar to formulaic expressions as ‘chunks’. Again, the idea of EL islands as chunks is very appealing because it seeks to offer some kind of generalization. However, it has been criticized for being too broad a term since it hinges on the assumption that every multi-­morphemic EL insertion is a chunk (Backus 1999: 97). As we observed in section 8.2.4, one of the problems identified by Myers-Scotton (2002: 141) with Backus’s idea of chunks is the lack of clear boundaries between those combinations that are conventional and those that are not. Additionally, Backus’s ideas about how certain combinations are assembled are open to dispute. For instance, he (Backus 1999: 97) suggests that if two or more EL morphemes are used in CS and they form a conventional combination in the EL, then it would be too coincidental if the speaker had produced them by two or more independent switches, compositionally building up a composite expression. From this, it is quite possible that idioms, like irregular plurals and irregular past tenses in English (and other languages), may well be contained in single lemmas and therefore are not compositionally assembled. This notwithstanding, it is difficult to suggest that an EL island such as inside the parlour in (189) is accessed as a unit when similar phrases such as in/inside/outside the parlour exist in the English language. On his part, Muysken (2000: 62, cited in Myers-Scotton 2002: 142–3) argues that if an EL verb stem occurs with ML inflections and then the EL verb is followed by a complement in the EL (as in examples 185 and 186), the verb stem and the following EL material form a unit; thus, introducing an Adjacency Principle: if in a code-­mixed sentence, two adjacent elements are drawn from the same language, an analysis is preferred in which at some level of representation

Embedded Language Islands

185

(syntax, processing) these elements also form a unit (Muysken 2000: 61). Muysken’s Adjacency Principle seems to be an adaptation of the triggering argument discussed in the first paragraph of this section. As we have already noted, the notion of triggering is attractive, for the same reasons that the notion of chunking is attractive. However, the problem is that not all instances of singly occurring EL content morphemes in Igbo–English CS are followed by further EL material. Therefore, when it does happen, the evidence is not conclusive (see the results of our analysis in paragraph two of this section) that any ‘triggering’ is responsible. Or, in other words, if triggering is at work sometimes, why it is not at work all the time? We can perhaps suggest that it seems more likely that the speakers in our study switch both EL single words and islands because the switched elements convey their intentions better than the ML equivalent. That is, we agree with Myers-Scotton (2006: 265–6) that the typical switch of languages is not because speakers cannot finish a phrase in the language in which they began. There may well be words or phrases with the same ‘referring meaning’ in the ML, but speakers may switch because there is no ML equivalent that has the same pragmatic force as the EL counterpart. For instance, the example in (190) was contributed by one of the female bilingual speakers (a mother) used in the study. The utterance was captured on tape while she was chatting with her friends about how well her son had done in his law degree examination. The point here is that the EL phrase with flying colours better conveys the preferred connotation than the ML equivalent in which she would have simply stated that her son did well on the exam. Hence, our view is that the speakers in our study do not just continue in the EL due to a linguistic preference for carrying on in that language. Rather, bilinguals employ such structures for emphasis or just to draw attention to their speech. In fact, it has been suggested by Myers-Scotton (2005: 3–4) that bilinguals recognize that saying something in one style rather than another adds a new dimension to their conversational turn. That is, CS taps into the reservoirs of semantic/pragmatic fields and social and psychological associations of words and phrases in not just one language, but in two (or more). Put succinctly, Igbo–English bilinguals do not codeswitch just for the sake of it. They fulfil certain linguistic and social/pragmatic goals in the process of switching codes. This view is strengthened by the findings of sociolinguistic studies on attitudes towards Igbo–English CS. These studies report that CS is a strategy deployed mostly consciously by educated Igbo bilinguals as a way to showcase their mastery of a prestigious language: English (see Ahukanna 1990; Obiamalu and Mbagwu 2008).

186

Codeswitching in Igbo–English Bilingualism

8.7  Summary and conclusion As in the last two chapters, we have in this one considered the application of the MLP, the AP and the USP associated with the MLF model on the sample of EL islands attested in Igbo–English CS and have observed that all three principles are upheld in both qualitative and quantitative analyses. Furthermore, the results confirmed our earlier finding that Igbo–English CS does seem to be a classic case of CS. We observed that in Igbo–English CS internal EL islands are subject to greater ML control and less EL autonomy. That is, Igbo, and not English, is in charge of the configuration of the mixed constituents. Generally, although EL islands must follow the principle of well-­formedness of the EL within the islands, they follow the placement rules of the ML within the clause. In this regard, they pattern just like the singly occurring EL forms analysed in the previous two chapters. Finally, when Igbo–English bilinguals continue in the Embedded Language it is not necessarily because of a linguistic preference for carrying on in that language. Rather they seem to be highlighting their mastery of a prestigious language, and in so doing they satisfy various self-­selected linguistic and social/pragmatic goals.

9

Concluding Remarks and Implications This study is unique in that it uses new data from Igbo–English to test the Matrix Language Principle, Asymmetry Principle and Uniform Structure Principle of the highly influential Matrix Language Frame model of Myers-Scotton (1993b, 2002). The study also incorporates the analysis of language-­specific phonotactic constraints, the presence of vowel harmony between English verbs (free morphemes) and Igbo bound affixes and visual observation of spectrograms to conclusively establish that the Embedded Language single words in Igbo– English intrasentential codeswitching are genuine codeswitching forms rather than some kind of borrowings. In doing this, the study underscores the fact that the same language production mechanisms underlie both the switching of Embedded Language single words and multi-­word (Embedded Language islands) sequences in Igbo–English intrasentential codeswitching. Furthermore, the findings from this study provide both empirical and quantitative support for the ML versus EL hierarchy in codeswitching as predicted by the Matrix Language Frame model. That is, both Igbo and English do not simultaneously satisfy the roles of the Matrix Language as defined by the Morpheme Order Principle and the System Morpheme Principle of the Matrix Language Frame model. We shall now outline the specific findings of this study and their implications for the Matrix Language Frame model and codeswitching research generally.

9.1  The goal The central theme of this study is the evaluation of an account of intrasentential codeswitching in terms of the Matrix Language Frame model (Myers-Scotton 1993b, 2002), a proposal associated with the notion of asymmetry in the roles played by the two languages participating in what is termed ‘classic’ codeswitching. This study explores this asymmetry by testing the three associated principles of

188

Codeswitching in Igbo–English Bilingualism

the Matrix Language Frame model on the Igbo–English intrasentential codeswitching data.

9.1.1  The Matrix Language Principle According to the Matrix Language Principle, in classic codeswitching it is possible to identify the Matrix Language of any bilingual clause. The Matrix Language will be that of the morphosyntactic frame (Myers-Scotton 2001: 52; 2002: 25, 105). This study uses two specific criteria to identify the Matrix Language of each mixed constituent in the Igbo–English codeswitching data:

9.1.1.1  The morpheme order criterion This criterion follows from the Morpheme Order Principle, which states that in mixed constituents consisting of at least one EL word and any number of ML morphemes, surface word (and morpheme) order will be that of the ML (MyersScotton 1993b: 83; 2002: 59; 2006: 244). Our interpretation of the morpheme order criterion in relation to our data is that it will apply wherever there is a conflict in word order between Igbo and English. Using both qualitative and quantitative methods which make use of the contrasting elements of Igbo and English (see Chapter 4), we demonstrate that in every instance where there is a conflict in word order between Igbo and English, the order of only the Matrix Language (Igbo) prevails over that of the Embedded Language (English) exactly as the Morpheme Order Principle predicts (see the analyses reported in sections 6.2.1.1 to 6.2.1.3, 6.2.3 and 6.3.3, 7.3, 7.5 to 7.6.3, 8.2 to 8.2.3, 8.2.5 and 8.3.7 to 8.5).

9.1.1.2  The late system (inflectional) morpheme criterion Our late system morpheme criterion follows from the System Morpheme Principle, which states that in Matrix Language + Embedded Language constituents all system morphemes which have grammatical relations external to their head constituents (i.e. which participate in the sentence’s thematic role grid) will come from the ML (Myers-Scotton 1993b: 83; 2002: 59; 2006: 244). In this study, we define this to mean that only one of the languages participating in Igbo–English codeswitching will contribute verbal inflections (bound morphemes and auxiliary verbs) used in the syntactic expression of tense–aspect–mood– negation of the bilingual clauses. Without exception, the results reported in Tables 6.2, 7.3, 8.1 and 8.2 reveal that Igbo, the Matrix Language, contributes the

Concluding Remarks and Implications

189

late system morphemes of every bilingual clause in Igbo–English codeswitching exactly as the System Morpheme Principle of the Matrix Language Frame model predicts.

9.1.2  The Asymmetry Principle The Asymmetry Principle states that bilingual speech is characterized by asymmetry in terms of the participation of the languages involved in codeswitching. In addition, it predicts that in classic codeswitching only one of the participating languages is the source of the Matrix Language. This asymmetry is claimed as evidence of the universal drive in language to achieve uniformity in the structural frame of any variety, to avoid meaningless variation (MyersScotton 2002: 9). This study evaluates this prediction in two specific ways: (1) asymmetry in the source of frame-­building elements (i.e. all verb inflections) of all bilingual clauses; and (2) asymmetry in the resolution of conflict in word order of mixed constituents. Using results from the analyses reported in Tables 6.2 (singly occurring EL nouns and adjectives), 7.3 (singly occurring EL verbs), 8.1 (internal EL islands) and 8.2 (non-­internal EL islands), the study shows that:

1. Verb inflections (i.e. late system morphemes) of all bilingual clauses analysed in this study (100 per cent or 2,659/2,659) come only from Igbo, the Matrix Language. 2. Sixty-­four per cent (1,707/2,659) of the Embedded Language elements (single words and Embedded Language islands) occur in structures where there is a conflict in word order between Igbo and English. The remaining 36 per cent (952/2,659) occur in structures where there is no conflict in word order between the two languages as far as the switched elements are concerned. The results of the test of the Asymmetry Principle reveal that the order of Igbo (the Matrix Language) prevails in all the 1,707 (100 per cent or 1,707/1,707) mixed constituents exhibiting a conflict in word order between Igbo and English. A morpheme count of the entire sample reveals that Igbo (the ML) on average contributes the vast majority of the content morphemes in the bilingual clauses (over 88 per cent). The Embedded Language (English) supplies only about 12 per cent of such morphemes. Thus, the basic theoretical notion that there is a Matrix Language versus Embedded Language hierarchy is overwhelmingly reflected in Igbo–English codeswitching because the two languages do not both satisfy the roles of the ML

190

Codeswitching in Igbo–English Bilingualism

contained in the Morpheme Order Principle and the System Morpheme Principle. And the dominance of the Matrix Language (Igbo) extends even to which of the two languages contributes most of the content morphemes in the bilingual clauses.

9.1.3  The Uniform Structure Principle The Uniform Structure Principle states that in bilingual speech, the structures of the ML are always preferred, but some EL structures are allowed if certain conditions are met (Myers-Scotton 2002: 8–9). The Uniform Structure Principle stems from the 4-M model, which makes a distinction between content and system morphemes, dividing the system morpheme into three subtypes: early system morphemes, bridge late system morphemes and outsider late system morphemes (see Chapter 3). This principle does not prohibit the occurrence of Embedded Language early system morphemes, only that the occurrence of Embedded Language early system morphemes in mixed constituents is less likely than early system morphemes from the Matrix Language. The assessment of this principle reveals that the Matrix Language supplies the overwhelming majority (98.4 per cent) of the early system morphemes in the bilingual clauses: demonstratives, pronominal modifiers, quantifiers, numerals, pronominal prefixes (i.e. subject clitics), pronouns and so on. Also, there is no instance in Igbo–English codeswitching where the determiners are in English and the lexical nouns/noun phrases are in Igbo. Instead, English nouns/noun phrases took Igbo determiners. That is, the strict constraints on Embedded Language system morphemes by the Matrix Language well-­formedness requirements mean that Igbo system morphemes (both early and late) could appear freely, but even English early system morphemes, which are not ruled out by the Uniform Structure Principle, could hardly appear in nominal Embedded Language + Matrix Language constituents, except when the English expression within Igbo– English codeswitching is a maximal projection of its own, in which case such an expression is an Embedded Language island (see Chapter 8). Furthermore, the strict constraints on Embedded Language system morphemes by the Matrix Language well-­formedness requirements can also account for the morphosyntactic features reflected in the 144 EL nouns and 14 EL noun phrases that occur without any determiners in contexts where a determiner is obligatory for a well-­formed DP in English but not in Igbo (see sections 6.2.3 and 8.2.2). Since English determiners (early system morphemes) are typically pre-­posed to their nominal complements but post-­posed in Igbo, the Matrix Language of Igbo

Concluding Remarks and Implications

191

would restrict their appearance and finally lead to their loss in mixed determiner phrases. A similar explanation can account for the English-­origin bare verbs in serial verb constructions in Igbo–English codeswitching. We notice in the analysis, in sections 7.5 and 7.5.1, that these verbs tend to co-­occur with other Igbo verbs that carry all the inflections marking tense–aspect–mood–negation in the bilingual clauses. Thus, allowing the English-­origin verbs to remain bare in accordance with Igbo grammar requirements for verbs in serial verb constructions. It is equally fascinating to observe that while it is possible for English-­origin verbs to accept Igbo verbal inflections, the reverse is not the case for Igbo verbs (see Chapters 6, 7 and 8). Moreover, the analysis of the Igbo–English codeswitching data reveals that the functional category Inflection seems to determine the type of pronoun that can fill its Specifier position. That is, if inflection is in Igbo (which is always the case in Igbo–English intrasentential codeswitching), then the pronoun in the subject position must also be in Igbo. Consequently, it would appear that the occurrence of one functional category within a given structure demands that the other functional categories must be in the same language. The dominance of the Matrix Language is also noticeable in the analysis presented in section  8.3.4, which shows that in Igbo–English codeswitching the Igbo complementizers can introduce both Igbo and English Inflectional Phrase complements. English complementizers do not license Igbo Inflectional Phrase complements in the Igbo–English data. Lastly, the above clearly indicate that the structures of Igbo (the Matrix Language) are preferred over those of the Embedded Language (English) in the mixed constituents analysed in this study exactly as the Uniform Structure Principle predicts. In general, the results of the evaluation of the Matrix Language Frame model reveal that Igbo–English codeswitching does seem to be a ‘classic’ case of codeswitching. That is, only one of the participating languages (Igbo) is the source of the morphosyntactic frame for the mixed constituents in bilingual clauses. This finding from Igbo–English codeswitching provides strong empirical support for the assertions by Myers-Scotton (2002: 112) that in classic codeswitching:

1. One language is the sole source of the frame of bilingual clauses. In the case of Igbo–English codeswitching, Igbo fulfils this role. 2. Although the Matrix Language Frame model allows for the possibility that the Matrix Language may change as the conversation, topic or participants change, this is not the case in our data corpus, thus confirming the rarity of such occurrences in classic codeswitching.

192

Codeswitching in Igbo–English Bilingualism

3. Even the Matrix Language of bilingual clauses containing bare Embedded Language forms does not change: Igbo remains the single source of the frame-­building elements in all such cases in Igbo–English codeswitching. The overall implication is not to say that the formulation of the Matrix Language Frame model embodies an a priori assumption about which language will become the Matrix Language in classic codeswitching, as both Igbo and English have the same opportunity to become this language. Instead, the Matrix Language Frame model is formulated as a dynamic analytic framework which makes allowances for the possibility that the Matrix Language may change as the conversation, topic and participants change. However, empirical codeswitching data suggests that such occurrence is rare to non-­existent in most corpora, as is the case in Igbo– English codeswitching. Consequently, the model’s predictive power lies in its recognition that there will be asymmetry between the Matrix Language and the Embedded Language in their roles in setting the morphosyntactic frame of the bilingual clause. And the consistency with which Igbo supplies both the frame-­ building elements and sets morpheme order wherever there is a conflict in word order in Igbo–English codeswitching bears this out. Next, we shall compare our conclusions to those on other language pairs identified as classic cases of codeswitching.

9.2  Igbo–English codeswitching and other classic cases Myers-Scotton (1993b) developed the Matrix Language Frame model in connection with her Swahili–English codeswitching data. Example (200) from Myers-Scotton (2002: 57) clearly illustrates the type of codeswitching that the model was designed to account for: (200) U-na-­wez-a ku-­m-pat-­a a-­me-va-­a 2S-NONPST-able-FV INF-OBJ-find-FV 3S-PERF-wear nguo ny-­ingine bright kama color y-­a red clothes CL9-other bright as   color CL9-ASSOC red ‘You can find her [that] she is wearing other bright clothes [such] as red [ones].’

The morpheme order of example (200) is that of Swahili. For instance, the head noun nguo ‘clothes’ is followed by its modifiers instead of being pre-­modified by them as is the situation in English. Moreover, the verb inflections (bound morphemes) marking case and subject–verb agreement (outsider late system

Concluding Remarks and Implications

193

morphemes in the Matrix Language Frame model) are contributed by Swahili only. Therefore, the Matrix Language of this bilingual clause is Swahili according to both the Morpheme Order and System Morpheme Principle of the Matrix Language Frame model. According to Myers-Scotton (1993b: 236), in forty conversations, she reports that in one example the modifier–head order unexpectedly reflects English in an otherwise Swahili utterance where the outsider late system morphemes seemingly come from Swahili. With reference to the lone example, she argues that there are pragmatic reasons for this violation. As in Igbo–English codeswitching, then, the Matrix Language Principle is comprehensively upheld in the bilingual clauses analysed. In their study of codeswitching among African township residents in South Africa, Finlayson, Calteaux and Myers-Scotton (1998) report, similar to our finding, in reference to grammatical structure that within any bilingual clause only one language (the Matrix Language) provides the grammatical frame in the data studied. Similarly, Amuzu (1998, 2005) also reports that in Ewe–English codeswitching, Ewe, like Igbo, functions as the Matrix Language. Therefore, English content morphemes may only occur as fully integrated codeswitching forms in slots projected by their Ewe counterparts, with which they must, as a prerequisite, be sufficiently congruent. If an English content morpheme is not sufficiently congruent with its Ewe counterpart, the English morpheme is realized as a codeswitching form through a compromise strategy in a bare form construction or Embedded Language island. Also, Boussofara-Omar (1999, 2003) successfully applied the Matrix Language Frame model to account for codeswitching between a variety of standard Arabic and local Tunisian Arabic. She reports that Tunisian Arabic is the Matrix Language into which elements of the standard Arabic are inserted (Boussofara-Omar 2003: 33). In Deuchar’s (2006) study, she tested the principles of the Matrix Language Frame model on a sample of Welsh–English codeswitching data. She reports, similar to our findings, that her data was overwhelmingly compatible with the Matrix Language Principle, the Asymmetry Principle and the Uniform Structure Principle, thus concluding that Welsh–English codeswitching is a ‘classic’ case of codeswitching (Deuchar 2006: 2005). Furthermore, in their study of Persian– English codeswitching, Rahimi and Dabaghi (2013) report that the application of the three principles of the Matrix Language Frame model to their data reveals strong support for the framework. They observe that in the bilingual clauses in their corpus only one of the participating languages (Persian) is the source of the morphosyntactic frame for the clauses (Rahimi and Dabaghi 2013: 14). Among other studies reported by Myers-Scotton (2002) to have adopted

194

Codeswitching in Igbo–English Bilingualism

the Matrix Language Frame model successfully in their analysis of codeswitching data are those of Hlavac (2000: Croatian–English codeswitching) and Türker (2000: Turkish–Norwegian codeswitching). Although Backus’s (1996) study on Turkish– Dutch codeswitching does not explicitly test the Matrix Language Frame model, it makes extensive use of the basic assumptions of the Myers-Scotton model. Finally, here, we can be confident in our conclusion that Igbo–English codeswitching is indeed a classic case from Myers-Scotton’s (2002: 110–12) discussion of classic codeswitching in terms of how it is associated with proficiency and stability. Myers-Scotton suggests that classic codeswitching is found in situations of stable bilingualism, which, as we outlined in section 1.4, is true of Nigeria (a former colony of the British Empire from 1865 to 1960). Secondly, she states that classic codeswitching typically occurs between speakers who have sufficient proficiency in one of the participating languages to use it as the sole source of the morphosyntactic frame of bilingual clauses (2002: 110). This is evidently the case for the Igbo–English bilingual speakers in our study. They acquired Igbo as their first language at home, were then exposed to English at school age and are fluent in both (see section 5.2.1). In the following section, we shall highlight some problematic examples arising from the analysis of Igbo– English intrasentential codeswitching and their suggested solutions.

9.3  Igbo–English: some problematic examples and suggested solutions As is clearly evident from the analysis presented in this study, the Matrix Language Frame model is a very powerful analytic framework and offers a straightforward explanation of the grammatical properties of Igbo–English codeswitching. Even in the seemingly problematic examples involving double morphology, doubling of complementizers and bare forms (i.e. EL nouns with zero determiners and EL verbs without Igbo morphology), the framework’s recently (post 2001) formulated 4-Morpheme model and the Uniform Structure Principle appear to make the correct predictions on all counts. Firstly, double morphology, as in example (120) repeated below as (201), has been suggested by some analysts (e.g. Muysken 2000 and Park 2000) as a counterexample to the System Morpheme Principle. (201) tractors ndi. ahu. e-­mebi-cha-­la tractor-­s PRN/PL DEM/D V-damage-ENCL-PERF ‘Those tractors have been damaged completely.’ [28:12]

Concluding Remarks and Implications

195

In example (201) we notice what might be viewed as double plural marking; the English noun tractor-s co-­occurs with the Igbo optional plural morpheme ndi. ‘replicas of/those of ’ (Echeruo 1998: 104). However, the problem of double morphology is dealt with by the Early System Morpheme Hypothesis of the Four-Morpheme model. According to the Early System Morpheme Hypothesis, only early system morphemes may be doubled in classic codeswitching (MyersScotton 2002: 92). Myers-Scotton explains that plural morphemes are not the type of morpheme predicted by the System Morpheme Principle that must come from the Matrix Language only. In fact, plural affixes are early system morphemes and thus can be contributed by both the Matrix Language and the Embedded Language. Therefore, the Early System Morpheme Hypothesis makes the correct prediction with regard to the Igbo–English codeswitching data. As we explained earlier in section 6.2.4, Myers-Scotton (ibid.) adds that the motivation for the hypothesis is that of all system morphemes only early system morphemes have the special relation to their lexical heads that would promote their accessing when their Embedded Language heads are called in codeswitching. Like their heads, early system morphemes are conceptually activated. That is, such morphemes are salient at the same level as their heads (content morphemes) in the mental lexicon. Therefore, they are available if any ‘mistiming’ is going to occur. On the other hand, late system morphemes, like the bound elements used to express tense–aspect–mood–negation, are not doubled in our data. Thus, not only do the examples analysed in this study support both the Morpheme Order Principle and the System Morpheme Principle, they also confirm Myers-Scotton’s (2002: 88) claim that late system morphemes are accessed much later than early system morphemes at the level of the formulator (see section  3.2.2). That is, outsider late system morphemes become available when larger constituents than just the DP are being built. Moreover, we also argued in section 6.2.4 following Igbo grammar (see also Obiamalu 2013a) that it is the grammatical instinct of our speakers which compels them to introduce ndi. into the DP structure. This functional element combines with ahu. ‘that’ to express a notion similar to the English plural demonstrative determiner ‘those’. The plural demonstrative determiner ‘those’ is not found in Igbo. Therefore, it would seem that once a speaker selects a pluralized singly occurring English noun (e.g. tractor+s), at the functional level stage (or the formulator, see sections  3.2.2 and 6.2.2), where Igbo (the ML) is in control of structuring all mixed DPs, abstract grammatical processes are set off to project a Dˈ slot for the Igbo functional element ndi., which combines with DemP ahu. that to give the supra-DP both [+ plural] and [+definite + specific] reading. Recall also from section  6.2.4

196

Codeswitching in Igbo–English Bilingualism

that following Obiamalu (2013a: 57) we demonstrated that in addition to the projection of a functional category D, which in some instances has a null realization in Igbo (see sections  4.4.5 and 6.2.3), more than one functional projection could be found within the Igbo nominal phrase. In effect, Igbo grammar already possesses a syntactic model to account for double morphology apart from the Early System Morpheme Hypothesis of the Four-Morpheme model. Secondly, doubling of complementizers, as in (194) repeated below as (202): (202) o.-na-­a-ju. whether ma i.-ga-­e-so ha 3SG.CL-AUX-V-ask  COMP  COMP 2SG.CL-AUX-V-follow them a-­ga  u. ka V-go church ‘She is asking whether you will follow them to church.’ [5:30]

Recall that we stated earlier (in sections 4.4.7 and 8.3.4) that English does not permit a doubly filled COMP slot in its syntax, whereas the reverse is the case in Igbo as the above example illustrates. According to Myers-Scotton (2013: 42), complementizers qualify as bridges that join together two clauses in a hierarchical relation. And hierarchical structure is a feature of the abstract level of predicate-­ argument structure (see sections 3.2.2 and 3.2.3). Thus, complementizers index the notion that clause 2 has a structural relationship with clause 1. Furthermore, Myers-Scotton (ibid: 43) notes that complementizers often come only from the Matrix Language in some data sets. This is exactly the case in Igbo–English codeswitching (see section 8.3.4). So, what is the implication of the doubling of complementizers in the Igbo–English data? As we outlined earlier in section 8.3.4, the examples of double complementizers in Igbo–English codeswitching indicate that the Matrix Language (Igbo) complementizer has dominance over that of the Embedded Language (English). That is, the Igbo complementizer could license both Igbo and English Inflectional Phrase complements, providing that the Igbo–English Inflectional Phrase meets the well-­formedness conditions of the Igbo complementizer. Additionally, as is evident in example (202) and the other examples cited in section 8.3.4, English complementizers do not license any Igbo Inflectional Phrase complements in Igbo–English codeswitching (see also Obiamalu 2013b), thus leading to the doubling of complementizers where the Igbo complementizer is closer to the Igbo Inflectional Phrase complement in (202). Put simply, in Igbo–English codeswitching, it is the Igbo complementizer that determines the following Inflectional Phrase complement irrespective of the Inflectional Phrase’s language of origin. This finding is similar to the case of

Concluding Remarks and Implications

197

Spanish que ‘that’ reported in Myers-Scotton and Jake (2009: 352), which can introduce either an English or a Spanish clause if Spanish seems to be the dominant Matrix Language in the discourse. Thirdly, the Uniform Structure Principle can better account for the morphosyntactic features seen in the EL nouns with zero determiner in Igbo– English, as in example (116), repeated below as (203). (203) ha a-­hapu. -la terminal ji-­ri taxi na-­a-lo.-ta They V-leave-PERF terminal hold-IND taxi AUX-V-return-ENCL ‘They have left (the) terminal and they are returning in a taxi.’ [14:31]

In relation to Igbo–English codeswitching, we disagree with the view by MyersScotton and Jake (2001: 106) that bare forms occur in codeswitching as a compromise strategy due to lack of sufficient congruence between the EL content morpheme and its ML counterpart (see section 6.2.3). Our view is that the compromise strategy proposal seems unnecessary for Igbo–English since the Matrix Language already has a syntactic model that accommodates nominals with zero determiners. For instance, we demonstrated in section 6.2.3 that the EL bare nouns in Igbo–English CS have near direct ML equivalents, which also occupy the same syntactic positions in clause structure (either in subject or object position). So the English noun terminal (as in ‘building’) is the equivalent of the Igbo noun u. lo. in (204). (204) ha a-­hapu. -la   terminal (u. lo.) ji-­ri taxi na-­a-lo.-ta They V-leave-PERF terminal hold-IND taxi AUX-V-return-ENCL ‘They have left (the) terminal and they are returning in a taxi.’ [14:31]

Also, it would seem that the EL noun is not inserted with any noticeable compromise strategy since Igbo already permits the occurrence of such forms in its grammar. In fact, much earlier Welmers (1973: 220) observed that Igbo typically expresses generic reference by means of bare nouns, although a distinction has to be first drawn between animate and inanimate nouns. As we stated earlier in section 4.4.5, drawing such a distinction falls outside the scope of the present study. However, an important point to make about the bare nouns (i.e. EL nouns with zero determiners) is that there does not seem to be any direct correlation between the animacy of the noun and generic reference in English. In English, generic reference is not limited to bare nouns. Rather, as Quirk et al. (1985: 281–2) observe, the definite (the) and indefinite (a/an) articles as well as the zero determiner may all be used to express generic reference under certain conditions. Thus, generic reading is also indicated morphologically with -s on

198

Codeswitching in Igbo–English Bilingualism

plural nouns in English. Crucially, as the analysis in section 6.2.3 shows, all the bare nouns occur in contexts where a determiner is obligatory in English for a well-­formed DP but not in Igbo. This notwithstanding, we also showed that the empty D slot in the examples was occupied by a post-­posed covert determiner in accordance with Igbo grammar. Again no compromise strategy is needed to account for their insertion apart from the requirements of Igbo grammar. In fact, the Uniform Structure Principle, which states that in bilingual speech the structures of the Matrix Language are always preferred, appears to offer an account that recognizes the dominant influence of the ML grammar rather than a view based on insertion by a compromise strategy. The same principle can also account for the Embedded Language verbs in Igbo–English codeswitching which occur without Matrix Language inflections. Such Embedded Language verbs, as in example (165) repeated below as (205), tend to co-­occur with other Igbo verbs in serial verb constructions (see sections 7.5 and 7.5.1). (205) o-­meV1-re m̩ ma ya sponsorV2 ya a-­gaaV3 Canada 3SG.CL-do/make-IND mother his sponsor she V-go   Canada ‘He sponsored his mother and she travelled to Canada.’ [15:6]

Verb serialization is a structure alien to Modern English but common in Igbo. It is also the only type of construction in which a full Igbo verb may appear bare (i.e. without verbal inflectional morphology) as one of a succession of verbs. Additionally, we noticed that most of the Embedded Language verbs in (as in 205) serial verb constructions in our data seem to co-­occur with the Igbo verb mee ‘do/make’. This verb carried the inflections for tense–aspect–mood–negation, thereby leaving the EL verbs bare in accordance with Igbo grammar. Like the case of the bare EL nouns, we reject the idea that the bare EL verbs in Igbo– English occur as a result of a compromise strategy. Instead, as we demonstrated in section 7.5.1, Igbo already has a syntactic model where native Igbo verbs can occur as bare forms without positing any compromise strategies. Our conclusion, following Amuzu (2013: 20), is that bilingual serial verb constructions are possible in codeswitching only when the Matrix Language has serial verb constructions, as is the case in Igbo. In other words, bare forms in Igbo–English codeswitching are not the result of the activation of any compromise strategies but rather Embedded Language forms that occur in clause structure in accordance with pre-­existing Matrix Language syntactic models. Next, we discuss the implications of our findings for the codeswitching models reviewed in Chapter 2.

Concluding Remarks and Implications

199

9.4  The codeswitching versus borrowing debate As we began to outline in Chapter  1 and section  5.5, a pervasive issue in the study of bilingual speech concerns the necessity of distinguishing true CS, by definition restricted to bilinguals, from borrowings which are not (Swigart 1992; see also Couto et al. 2015: 66; Legendre and Schindler 2010: 57; Poplack 2012: 644; Stammers and Deuchar 2012: 630). According to Legendre and Schindler (2010: 57), borrowings are sometimes defined as fully phonologically and morphologically integrated into the source (dominant) language (e.g. MacSwan 2009; Poplack 1980). Furthermore, we agree with Legendre and Schindler (2010) that expanding the definition of borrowings to include morphological integration, such as the addition of a Matrix Language affix, effectively makes these researchers’ bans on word-­internal CS unfalsifiable, if every word-­internal switch must be considered a borrowing. Taking the position that borrowings are characteristic of monolingual speech (Legendre and Schindler 2010; Pfaff 1979), and that monolinguals would be unable to pronounce words in a foreign language, this study defines borrowings in terms of listedness (after Muysken and Deuchar; see section 5.6). That is, we assume borrowings to be listed in the vocabulary of monolingual speakers of the recipient language, whereas switches are not. Thus, for a language like Igbo with over twenty million native speakers (Nigerian Census 2006) in South-Eastern Nigeria listedness is reflected in established dictionaries of the language. Also, we assume that borrowings would comply fully with the syllable structure and phonotactics of the recipient language, Igbo. Therefore, if a word is not listed in the Igbo dictionary and does not comply fully with Igbo syllable structure and phonotactic constraints it is not a borrowing, temporary or otherwise. Such words are treated as genuine CS forms in this study. The phonological differences between Igbo and English, displayed in Table 9.1, assist in distinguishing between borrowings and true switches in Igbo–English codeswitching. Given the copious contact between Britain and Nigeria, through

Table 9.1  Summarized comparison of Igbo and English phonologies Trait

Igbo

English

Vowel harmony Consonant clusters Coda consonant(s)

[ATR] Root controlled X X

X √ √

200

Codeswitching in Igbo–English Bilingualism

Table 9.2  Borrowings versus codeswitching forms Borrowed English word

As listed in the Igbo dictionary

Igbo–English

bacteria [.bæk.tɪə.rɪə.]

bakiteria [.ba.ki.te.ria.]

bank [.bæŋk.]

banki [.ba.ŋ̩ .ki..]

latitude [.læ.tɪ.tjuːd.]

latitudi [.la.ti.tu.di.]

sulphur [.sʌl.fə.]

so.lu.fo. [.sɔ.lʊ.fɔ.]

table [.teɪ.bəl.]

tebulu [te.bʊ. lʊ.]

ceremony ahu. (as in 85) ‘that ceremony’ u. lo. vice-­chancellor (as in 100) ‘(the) vice-­chancellor’s house’ e-register-ghi. (as in 156) ‘did not register’ i-destabilize (as in 185) ‘to destabilize’ o me-­re campaing (as in 157) ‘he campaigned’

colonization and membership of the Commonwealth organization of former British colonies, English has loaned many words to Igbo and the other Nigerian indigenous languages. These borrowings have been systematically ‘Igbonized’. That is, the borrowings are phonologically indistinguishable from Igbo words. In fact, the uneducated Igbo monolingual speakers do not identify them as English words, and they are listed in dictionaries as Igbo words. Table 9.2 shows a list of the phonological characteristics of English borrowings into Igbo versus codeswitching forms. In Table 9.2, borrowings show evidence of vowel epenthesis, strict adherence to the Igbo syllable structure and other segmental processes indicative of full phonological integration into Igbo. On the other hand, the codeswitching examples from Igbo–English do not share the same features as the borrowed words. Therefore, we consider the examples from Igbo–English analysed in this study to be genuine codeswitching forms rather than instances of borrowings. Also, since the Igbo–English examples do not show similar levels of phonological integration into Igbo – or none at all – they represent valid examples of head-­ internal switches, violating the Phonetic Form Interface Condition just like the Wolof–French examples studied by Legendre and Schindler (2010). As it turns out, the application of Poplack (2012) and Poplack and Meechan’s (1998) Nonce Borrowing Hypothesis (see section 5.5) to the Igbo–English data would have either erroneously blocked all the 1,963 lone English nouns,

Concluding Remarks and Implications

201

adjectives and verbs from occurring in Igbo–English codeswitching or labelled them as some kind of borrowings, thereby leaving us with nothing to analyse in the form of single-­word switches. Nevertheless, like Myers-Scotton (2002: 155), we agree with Poplack and Meechan (1998) that there are bilingual clauses framed by a Matrix Language and that lone English words can appear in these frames with Matrix Language morphosyntactic integration. We also agree that the codeswitching/nonce borrowing and attested loanwords are not distinct linguistically. However, our position differs from that of Poplack and Meechan (1998), because, similarly to Myers-Scotton (2002: 155), we recognize the role of abstract grammatical structure in codeswitching. As we noted in section 5.5, our position is that when mixed constituents are accessed, there is necessarily interaction of the two grammars at an abstract level, even while the Matrix Language is more activated than the Embedded Language. This interaction differentiates codeswitching from monolingual data; while the same abstract procedures may result in (1) monolingual Igbo discourse and (2) discourse with an Igbo frame but English insertions, the two outcomes do not have the same history. The form of the bilingual outcome depends on both universal principles for bilingual clauses (for example, one language dominates in the grammatical frame) and restrictions that depend on congruence/incongruence regarding the typological characteristics of the participating languages. In other words, the roles of the two languages involved in codeswitching are never the same. One language supplies the frame-­building properties into which content morphemes from the other are inserted. Igbo is the language that supplies the morphosyntactic frame into which English content morphemes are inserted in Igbo–English codeswitching. Other analysts such as Muysken (2000) do not distinguish codeswitching and borrowing based on surface distributions like Poplack and her associates, but instead on formal, surface characteristics. For instance, Muysken (2000: 69) states that code-­mixing (i.e. codeswitching) can be viewed as involving words with different language indexes inserted into a phrase structure for a clause C, while lexical borrowing may be seen as involving formatives inserted into an alien word structure. Furthermore, Muysken (2000: 72, 75) asserts that code-­ mixing has the ordinary, supra-­lexical, productive properties of syntax; and insertion is constituent-­internal, alternation is phrase- or clause-­peripheral. We agree with Myers-Scotton (2002: 157) that Muysken (2000) is correct in identifying these distinctions; however the fact still remains that both Embedded Language single words and islands occur in the same Matrix Language Frame in Igbo–English codeswitching (see analysis in Chapters 6 to 8 of this volume).

202

Codeswitching in Igbo–English Bilingualism

Accordingly, as we have already noted in this study, the Matrix Language Frame model is primarily based on the thinking that in the syntax of intrasentential codeswitching one language (the Matrix Language) is dominant and determines the overall morphosyntactic properties of codeswitched utterances, while the other language (the Embedded Language) supplies largely certain lexical categories, and the role of its grammar is limited by restrictions imposed by the Matrix Language. So, such Embedded Language single words as in Chapters 6 and 7 and the Embedded Language islands in Chapter  8 are analysed as codeswitching forms in this study. Interestingly, Table 9.3 shows that Igbo allows switching between functional heads (D, I, COMP, Neg) and their complement, NP, VP, IP, and VP, respectively. As in Urban Wolof codeswitching (Legendre and Schindler 2010), the highly frequent occurrence of head-­internal CS in Igbo–English codeswitching would appear to falsify the predictions of the codeswitching models reviewed in Chapter 2 of this study. Recall that the codeswitching constraints are stated as universally applicable and designed to ban exactly the type of codeswitching identified in Table 9.3 (i.e. the Equivalence Constraint, see section 2.2, Poplack 1980; the Free Morpheme Constraint, see section  2.3; the Government Constraint, see section  2.4, Di Sciullo et al. 1986; and the Functional Head Constraint, see section 2.5, Belazi et al. 1994). Also, the identified codeswitching patterns would seem to violate MacSwan’s (2009) Phonetic Form Interface Condition whereby the boundary between heads (words) represents the minimal opportunity for CS. As we have already indicated above, Igbo–English

Table 9.3  Codeswitching patterns in Igbo–English Patterns of codeswitching

Igbo–English

Example

D#NP

√

I#VP

√

NEG#VP

√

COMP#IP

√

…polling station ahu. … (as in 5) ‘…that polling station …’ …anyi. ga-­e-change ya … (as in 7) ‘…we will change it …’ e-register-ghi. aha nwa ya na oge … (as in 156) ‘They did not register her daughter’s name on time …’ … i.-ga-­a-hu. na he shouldn’t be re-­elected (as in 193) ‘… you will see that he shouldn’t be re-­elected.’

Concluding Remarks and Implications

203

codeswitching is not alone in permitting head-­internal codeswitching. Other language pairs allowing similar codeswitching patterns include Ewe–English (Amuzu 1998, 2005, 2010), Moroccan Arabic–French (Bentahila and Davies 1983), Lingala–French (Bokamba 1988), Hungarian–English (Bolonyai 2005), Welsh–English (Couto et al. 2015; Deuchar 2005, 2006; Stammers and Deuchar 2012), Kabiye–Ewe (Essizewa 2014), Finnish–English (Halmari 1997), and Wolof–French (Legendre and Schindler 2010). The highly frequent occurrence of such morphosyntactically well-­integrated forms in codeswitching cannot be due to chance. Instead, we agree with MyersScotton (2002, 2014) that such forms testify to the fact of the existence of strong asymmetries between any two languages participating in classic codeswitching. Nevertheless, it is important to point out that we do not deny that established borrowings are phonologically integrated into the recipient language to varying degrees; however, the position taken in this study is to say that a feature distinguishing codeswitching from established borrowings is that, in codeswitching, Embedded Language content morphemes inflected with Matrix Language morphemes generally retain their Embedded Language phonology. At the very least, we state that speakers employ a phonology for these forms that is not wholly identical to that of the Matrix Language variety. For instance, see sections 7.4 to 7.4.1.2 and in particular the spectrograms, in Figures 7.3 to 7.6, which show that when Igbo–English bilinguals combine Embedded Language verbs with Igbo affixes they still strive to preserve the syllable structure of the Embedded Language words. Moreover, the spectrograms confirm that any adaptations to the words occur only on either the left or right edge depending on whether a Matrix Language inflectional prefix or suffix is required. In other words, the Embedded Language single words in our data are not wholly phonologically integrated into the Matrix Language like native or loanwords found in Igbo. Consequently, on the basis of this analysis (also) we conclude that the English single words embedded in otherwise Igbo utterances are genuine codeswitching forms and that two phonologies can indeed co-­exist in codeswitching. Another reason why we do not consider the English single words in our data corpus to be borrowings is that there are equivalent words/phrases in Igbo which the speakers could have easily used instead of the English words. The same reasons are also applicable to the Embedded Language islands in Igbo– English codeswitching. That is, the speakers in our study do not just continue in the Embedded Language due to a linguistic preference for carrying on in that language. Rather, bilinguals employ such structures for emphasis or just to draw attention to their speech. It is important to note that under the Matrix Language

204

Codeswitching in Igbo–English Bilingualism

Frame model the Igbo equivalents do not necessarily need to be actual or real surface-­level Igbo morphemes. They may be unspecified Igbo content morpheme lemmas that are stored in the mental lexicon of the speaker as Generalized Lexical Knowledge of Igbo (see Amuzu 2010; Myers-Scotton 2002). Furthermore, the analysis of Embedded Language islands in Chapter  8 provides additional empirical support for our claim (after Myers-Scotton 2006: 260–1) that these full phrases from the Embedded Language pattern much like singly occurring Embedded Language words. This is further motivation to accept single words as codeswitched elements, as part of a continuum of Embedded Language material in bilingual constituents (see Myers-Scotton 2002 and Jake et al. 2002). This claim is illustrated by the internal Embedded Language island in (170), repeated below as (206). (206) ha ku. wa-­si.-ri. ballot box they break-ENCL-IND ballot box ‘They completely broke all the ballot boxes.’

dum all [23:26]

This example shows that internal Embedded Language islands are subject to more Matrix Language control and less Embedded Language independence. For example, the NP ‘ballot boxes’ in (206) represents such an island occurring in a DP framed by the Igbo quantifier dum ‘all’. Thus, the entire DP is under the control of the Matrix Language. It is evident in this example that the two English nouns have their own word order; however, they are congruent enough with their Matrix Language (Igbo) counterparts at the level of lexical-­conceptual structure to be able to appear in this DP. The full DP is assigned by the Igbo verbal assembly (the transitive verb ku. wa-­si.-ri. ‘(completely) broke’). That is, in this example, therefore, Igbo, not English, projects placement of the overall structure (in the object position). Hence, example (206) conforms to Igbo well-­formedness conditions and is further confirmation that when Embedded Language islands are projected, the Matrix Language is not entirely inhibited. In other words, although Embedded Language islands must follow the principle of well-­formedness of the Embedded Language within the islands, they follow the placement rules of the Matrix Language within the clause. In this regard, it is the position of this study that they are just like singly occurring Embedded Language forms, and thus both can be explained straightforwardly by the Matrix Language Frame model and its associated three principles: the Matrix Language Principle, the Asymmetry Principle and the Uniform Structure Principle.

Concluding Remarks and Implications

205

9.5  Conclusion This study has evaluated the Matrix Language Frame model of codeswitching with Igbo–English data and concluded that the data can indeed be considered a classic case of codeswitching, in that a Matrix Language can be clearly identified in bilingual clauses. We have established this by a qualitative and quantitative analysis, uncovering supportive evidence for the Matrix Language Principle, the Asymmetry Principle and the Uniform Structure Principle. The investigation went one step further by using spectrograms and the analysis of vowel harmony between English free morphemes and Igbo bound affixes to demonstrate that two phonologies can co-­exist in codeswitching, and that codeswitching forms are essentially pronounced with a phonology that does not entirely resemble that of the Matrix Language variety. The study also found that the same language production mechanisms as detailed under the Matrix Language Frame model and its associated three principles underlie both single-­word and multi-­word codeswitching. That is, the present study, like those before it adopting the Matrix Language framework (see Amuzu 2010: 277), underlines the importance of the assumptions underpinning the Matrix Language Principle: (1) that language production is modular; (2) that lexical structure is both complex and abstract; and (3) that languages in contact divide responsibilities in what they may contribute towards lexical structure during the production of mixed constituents. However, unlike the educated Ewe speakers described in Amuzu (2010), who engage in what he characterized as ‘composite codeswitching’ due to their frequent inability to sustain ready access to their mother tongue mental lexicon during online speech production, Igbo– English bilinguals can always sustain ready access to their mother tongue mental lexicon during online speech production. Therefore, they engage in classic rather than ‘composite’ codeswitching. Myers-Scotton (2002: 105) explains that composite codeswitching occurs in such phenomena as language attrition and shift. It occurs when speakers because of psycholinguistic or socio-­political factors do not have full access to the morphosyntactic frame of the participating language that is the desired source of the Matrix Language. Or, possibly the notion of a target Matrix Language is not clear to the speakers themselves. The result is that a composite Matrix Language frames the bilingual Complementizer Phrase (or the bilingual clause). Thus, in effect, composite codeswitching necessarily entails convergence.

206

Codeswitching in Igbo–English Bilingualism

This, as the descriptions of the speakers in section 5.2.1 and the analyses reported in Chapters 6 to 8 confirm, does not apply to our speakers, and thus we can be confident that Igbo–English is a classic case of codeswitching. We must acknowledge that this study represents a modest effort at characterizing all the possible codeswitching patterns in Igbo–English. Perhaps a much larger corpus might reveal more problematic examples than we have identified in the present study. Also, as is evident in the preceding chapters, we have focused only on discussing particular grammatical aspects of Igbo–English codeswitching to the exclusion of others such as the distribution of discourse particles and conjunctions. We assume that findings in these areas will no doubt provide additional insights into the operation of Igbo–English codeswitching. The desire to focus on what we considered to be more salient aspects of the grammar of Igbo–English codeswitching prevented us from exploring all things grammatical when it comes to the codeswitching behaviour of our speakers. Another aspect of Igbo–English intrasentential codeswitching that this study has not explored in any detail concerns uncovering what motivates the speakers to codeswitch in the first place. A number of reasons have been adduced in the literature as motivating factors for codeswitching. These range from social-­pragmatic to grammatical considerations. In fact, our interview schedule, reproduced as Appendix A, which was completed by all the fifty Igbo–English bilingual speakers used in this study, yielded ample sociolinguistic data including information about language choice and use. Not much of this information has been discussed or analysed in the present study. Again we were limited by the aim of the present study, which was grammatical rather than social. There were also the added constraints of space and the time needed to incorporate such detail in a study with a purely linguistic focus. Therefore, the exploration of the data from the sociolinguistic interviews will form the basis of any future study of Igbo–English codeswitching undertaken by this researcher.

Appendix A

Sample Interview Protocol The speakers were asked the questions below in the context of informal discussions. I made observational notes and obtained the speakers’ consent to capture their responses on tape. Following Ihemere (2007: 330) the speakers were allowed to give lengthy and detailed responses if they so desired. This led to a rich and diverse set of sociolinguistic data on our speakers regarding language choice and use. Although the speakers were at liberty to respond in either Igbo or English, we noted that most of the responses involved intrasentential codeswitching (as reflected in the codeswitching examples analysed in this study).

Section 1:  General 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Speaker’s name: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sex: □Male ǀǀ □Female Age: Level of Education: □Senior secondary school ǀǀ □University graduate ǀǀ □Undergraduate ǀǀ □Other Your occupation: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Where do you live: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Do you live with your spouse and / or your children? □Yes ǀǀ □No Do you live with your parents, brother(s) and sister(s)? □Yes ǀǀ □No What is your mother tongue? . . . . . . . . . . . . . . . . . . . . . What is your spouse’s mother tongue? . . . . . . . . . . . . . . . . . . . . . . How would you describe the formal educational level of each of the following people? Spouse: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Father: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mother: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12. Is your mother tongue the dominant language spoken in your home? . . . . . . . . . . . . . . . . . . . .

208

Appendix A

Section 2:  Language choice/use (close relatives) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

What language do you use when you talk to your father? What language do you use when you talk to your mother? What language do you use when you talk to your grandfather? What language do you use when you talk to your grandmother? What language do you use when you talk to your brother(s)? What language do you use when you talk to your sister(s)? What language do you use when you talk to your male children? What language do you use when you talk to your female children? What language do you use when you talk to your wife? What language do you use when you talk to your husband? What language do you use when you talk to your male grandchildren? What language do you use when you talk to your female grandchildren? What language do you use when you talk to your daughter-­in-law? What language do you use when you talk to your son-­in-law? What language do you use when you talk to your father-­in-law? What language do you use when you talk to your mother-­in-law?

Language choice/use (with non-­close relatives) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

What language do you use when you talk to people of your father’s age? What language do you use when you talk to people of your mother’s age? What language do you use when you talk to people of your grandfather’s age? What language do you use when you talk to people of your grandmother’s age? What language do you use when you talk to your brothers’ age mates? What language do you use when you talk to your sisters’ age mates? What language do you use when you talk to your male children’s age mates? What language do you use when you talk to your female children’s age mates? What language do you use when you talk to your wife’s age mates? What language do you use when you talk to your husband’s age mates? What language do you use when you talk to your male grandchildren’s age mates? What language do you use when you talk to your female grandchildren’s age mates? What language do you use when you talk to your daughter-­in-law’s age mates? What language do you use when you talk to your son-­in-law’s age mates? What language do you use when you talk to your father-­in-law’s age mates? What language do you use when you talk to your mother-­in-law’s age mates? What language do you use when you talk to people same age as you?

Sample Interview Protocol

209

Section 3:  Social network 1. Do you have any close friends? 2. Who are they? (The goal here is to elicit as many names as the respondent can 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

recall.) How old are they? How many of your friends are male and how in many female? What language do you use when you meet your male friends? What language do you use when you meet your female friends? Who are your present neighbours? (Elicit more names.) Do you ever talk to your neighbours? When usually? What language do you use when you talk to your male and/or female neighbours? Are you a member of your church group? If yes, how often and where do you meet? Are your friends and/or neighbours members of your church group? What way do you talk when you attend the group meetings?

Section 4:  Language attitudes 1. How often do you use English words and expressions when you talk in your

2.

3. 4.

5.

6.

mother tongue about the following types of topics? Work: □Very often ǀǀ □Often ǀǀ □Rarely ǀǀ □Never ǀǀ Family situation: □Very often ǀǀ □Often ǀǀ □Rarely ǀǀ □Never ǀǀ Politics: □Very often ǀǀ □Often ǀǀ □Rarely ǀǀ □Never ǀǀ Economic situation in Nigeria: □Very often ǀǀ □Often ǀǀ □Rarely ǀǀ □Never ǀǀ Future plans: □Very often ǀǀ □Often ǀǀ □Rarely ǀǀ □Never How often do you use English words and expressions when you are conversing with relatives and friends who speak little or no English? □Very often ǀǀ □Often ǀǀ □Rarely ǀǀ □Never What is the main reason why you mix expressions from English and your mother tongue?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . In your opinion, which group of speakers of your mother tongue most regularly mix expressions from your mother tongue and English? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Should we stop mixing expressions from your mother tongue and English when we converse? □Yes ǀǀ □No ǀǀ □No opinion ǀǀ Any additional comments: . . . . . . . . . . . . . . . . . . . . . Can we stop mixing expressions from your mother tongue and English when we converse? □Yes ǀǀ □No ǀǀ □No opinion ǀǀ Any additional comments: . . . . . . . . . . . . . . . . . . . .

210

Appendix A

7. How would you describe your feeling or attitude toward the mixing of expressions



from your mother tongue and English? □Very Positive ǀǀ □Positive ǀǀ □Negative ǀǀ □Very Negative ǀǀ □Both positive and negative ǀǀ □Neither positive nor negative ǀǀ Any additional comments: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Appendix B

Summary of Findings from the Interview Protocol 1  The speakers The interview schedule revealed that the fifty speakers used in this study ranged in age from twenty to fifty-­five years. The sample was composed of school teachers (30 per cent or 15/50), engineers (20 per cent or 10/50), physicians (8 per cent or 4/50), nurses (10 per cent or 5/50), business owners (8 per cent or 4/50), and undergraduate students (24 per cent or 12/50). Also, the fifty speakers were born and grew up in Nigeria and have Igbo as their mother tongue. They are bilingual, being fluent in both Igbo and English. The second language (English) was learned at school age. Having studied in Nigeria, where English is the official language and the language of all advanced education, the informants have all been educated almost entirely in English. Therefore, they can be described as balanced bilinguals.

2  Language choice/use As it turned out, the speakers all reported that they mix both Igbo and English when conversing with their immediate family members and friends, who are themselves Igbo– English bilinguals. In fact, the comment by speaker 5 in (13), repeated below as (207), sums up the language choice/use patterns of our speakers nicely. (207)  a. anyi. ga village anyi. na-­asu. asu. su. Igbo We go village we AUX-speak language Igbo maka na u.fo.di. ndi. bi na obodo a-­na-ghi. because that some people of live PREP village V-AUX-NEG asu. asu. su. bekee … speak language English ‘When we go to the village we speak Igbo language because some of the villagers do not speak English …’

212

Appendix B

b. … mana with my family a-­na m e-use both … but with my family CL-AUX I HAB-use both ‘… but with my family I use both (i.e. Igbo and English).’ [5:1] That is, they use codeswitching to speakers who are bilingual in Igbo and English but with the monolingual Igbo speakers they use Igbo only.

3  Social network The speakers used in this study were members of a friendship network based around membership of a local pentecostal church bible study cell. This also meant that they lived in the same area of Port Harcourt at the time of the fieldwork in the summer of 2011. The group met every Wednesday for bible study in the cell leader’s house. They also visited the homes of members in the evenings and weekends for chats and dinners. I did not record the bible study meetings due to their religious nature. However, the conversations recorded for the analysis reported in this study come from the non-­religious visits by the friends on most evenings and weekends. A major advantage of employing the social network method for the data collection was that it allowed me to make use of the group dynamics to obtain more substantial amounts of spontaneous speech than would have been possible in interaction with a single individual who is isolated from his/her customary social network. Furthermore, as we pointed out in section  5.2, the social network method permitted the collection of authentic natural language in interactional situations. Crucially, it allowed me to observe at close quarters the language choice/use patterns of the fifty speakers. Both the conversational and observational data confirmed the self-­reported data from the interview schedule conducted well before capturing the speakers on tape.

4  Language attitudes The self-­reported data confirmed that our speakers indeed have very positive attitudes towards codeswitching and regularly use English words and expressions when conversing in their mother tongue on such topics as work, family situation, local politics, economic matters and future plans. Again the codeswitching examples discussed in this study reflect this view. They did not see the need for why codeswitching should be curtailed or banned. Instead they all agreed that it is a marker of one’s level of education and proficiency in English. This confirms Myers-Scotton’s (2005: 3–4) claim that bilinguals recognize that saying something in one style rather than another adds a new dimension to their conversational turn. She adds that codeswitching taps into the reservoirs of

Summary of Findings from the Interview Protocol

213

semantic/pragmatic fields and social and psychological associations of words and phrases in not just one language, but in two (or more). In other words, as we observed earlier in section  8.6, Igbo–English bilinguals do not codeswitch just for the sake of it. They fulfil certain linguistic and social/pragmatic goals in the process of switching codes. In fact, we agree with Ahukanna (1990) and Obiamalu and Mbagwu (2008) that codeswitching is a strategy deployed mostly consciously by educated Igbo bilinguals as a way to showcase their mastery of a prestigious language (English).

Notes Chapter 1:  Introduction 1 Poplack (1980); Poplack, Wheeler and Westwood (1987); Poplack and Meechan (1995); Sankoff, Poplack and Vanniarajan (1991). 2 See subsequent chapters of this volume for counterexamples to the EC and the FMC from Igbo–English CS. 3 See Belazi, Rubin, and Toribio (1994): Functional Head Constraint (FHC); Di Sciullo, Muysken and Singh (1986): Government Constraint (GC); MacSwan (1997, 1999, 2000, 2005a, 2005b, 2005c, 2009): Minimalist Approach (MA). 4 The Igbo–English CS data were collected by me (the researcher) in 2011 from informal conversations by educated adult Igbo–English bilinguals in Port Harcourt, Nigeria (see sections 5.2 and 5.3 for information about the speakers and the data collection strategy). Henceforth, in all the examples from the Igbo–English data, we will highlight a switched element in bold; the first number within the square brackets refers to the speaker code and the following number identifies the number of the cassette tape. The abbreviations and symbols used in the gloss are found in the list of abbreviations and symbols on pages xiv–xvi. 5 For example: Backus (1992): Turkish–Dutch; Berk-Seligson (1986): Spanish– Hebrew; Bolonyai (2005): Hungarian–English; Couto et al. (2015): Welsh–English; Halmari (1997): Finnish–English; Hlavac (2000): Croatian–English; Jake et al. (2002): Spanish–English; Myers-Scotton (1993b): Swahili–English; Nortier (1990): Arabic–Dutch; Park (2000): Korean–Swedish; and so on.

Chapter 2:  Studying the Grammar of Codeswitching 1 For example: Amuzu (2005; 2010; 2014); Couto et al. (2015); Backus (1992); Berk-Seligson (1986); Deuchar (2005, 2006); Halmari (1997); Hlavac (2000); Jake et al. (2002); Myers-Scotton (1993b); Nortier (1990); Park (2000); Rahimi and Dabaghi (2013); Stammers and Deuchar (2012); Türker (2000) and so on. 2 See Poplack (1980; 1981; 1990; 2012); Poplack and Meechan (1998); Poplack, Wheeler and Westwood (1987); Poplack, Sankoff and Miller (1988); Sankoff and Poplack (1981); Sankoff and Mainville (1986); Sankoff, M’Barek and Montpetit (1987); Sankoff, Poplack and Vanniarajan (1991).

216

Notes

3 This Fongbe (nyi = nyɛ) copula takes co-­referential nominal predicates as complements and functions either as an equative copula or as an ascriptive copula depending on whether the complement nominal is definite or generic. 4 Wolof is one of the languages spoken in Senegal. It is also spoken in Gambia and Mauritania, both countries are in West Africa.

Chapter 3:  Theoretical Framework 1 Not in the original text, but this author’s understanding of what a mixed constituent means. 2 Lemmas are defined as the morphological and syntactic properties which a word is said to inherently possess, which determine its co-­occurrence and selectional restrictions – after Levelt’s (1989) Speech Production model. 3 It is important to make clear that in positing the 4-M model, Myers-Scotton (2002: 75) states that she and her associate Janice Jake do not accept the determiner as head of NP analysis (Abney 1987). Therefore, she and her associate hold the view (as in traditional grammar) that the noun heads an NP, a verb heads a VP, an adjective heads an AdjP, and so on. 4 To be consistent this study adopts the same definition.

Chapter 4:  Comparison of Aspects of Igbo and English Grammars 1 See Anyanwu (1998), Clark (1990) and Liberman and Schultz (1993) for further discussion of the tonal system of Igbo. 2 All the examples cited in this chapter are for illustration only. They do not form part of the Igbo–English CS data analysed in subsequent chapters of this volume. 3 Maximal Onset Principle: onsets must be fully formed, in accordance with the principles of the language, before coda formation is carried out with the remnant. 4 Phonotactic constraints concern restrictions on the patterning of sounds in the words of any given language (see Roca and Johnson 1999: 696–7). 5 See Chapter 5 for a discussion on the CS versus borrowing debate. 6 The phrase structure tree diagram does not show any movement operations in the generation of the Igbo clause. We shall return to such matters in subsequent sections of this study. 7 Only potential conflict sites in Igbo–English speech are included in Table 4.9.

Notes

217

Chapter 5:  Methodology 1 See Appendix A for a sample of the interview questions. 2 As a researcher working in a United Kingdom institution of higher education, I am bound by law to respect the privacy of others and by the guidelines set by my institution’s research ethics committee. Information about the UK Data Protection Act and the University of Westminster research ethics guidelines can be found at the following links: UK DPA http://www.legislation.gov.uk/ukpga/1998/29/contents; UoW research ethics information https://www.westminster.ac.uk/research/research-­ framework/research-­ethics (both accessed 4 April 2016). 3 This constitutes an approximate equivalent to a clause (see Carter et al. 2011: 156). 4 For example: Backus (1992), Berk-Seligson (1986), Halmari (1997), Hlavac (2000), Jake et al. (2002), Myers-Scotton (1993b), Nortier (1990), Park (2000), Türker (2000) and so on. 5 For example: Adalar and Tagliamonte (1998), Budzhak-Jones (1998), Eze (1997/1998), MacSwan (1999, 2005a, 2005b); Poplack (1980, 1981), and so on. 6 Language mixing is used interchangeably with codeswitching in Matras (2009: 101).

Chapter 6:  Embedded Language Single Words: Nouns and Adjectives 1 For a detailed discussion of genitive construction, see Altenberg (1982) and Quirk et al. (1985). 2 The asterisk is used to mark examples as ungrammatical. 3 See Obiamalu (2013a: 59) for a parallel structure in the verbal genitive and evidence for the analysis of the possessum DP as having moved from the complement position to the Spec position. 4 Few English attributive adjectives are acceptable in the N+A slot in the Igbo NP. The few that do occur are mostly colour adjectives (as in examples 106 and 107). 5 We use the term ‘specific reference’ here to suggest that the interlocutors have a shared knowledge about the people or thing(s) referred to in the utterance. 6 Amuzu (2010); Backus (1992); Bentahila and Davies (1983); Eliasson (1989); Halmari (1997); Myers-Scotton (1993b; 2002); Park (2000). 7 A similar explanation is also reported in Deuchar (2005) for the insertion of English adjectives in Welsh–English CS. In Welsh, as in Igbo and Spanish, attributive adjectives post-­modify their nouns; whereas predicative adjectives occur in the same position as their English counterparts.

218

Notes

Chapter 7:  Embedded Language Single Words: Verbs 1 Anyanwu (2012: 383) notes that in the Principles and Parameters approach adopted in his study feature lowering is not a permissible operation (see also Radford 1997). 2 The Audacity software is a free open-­source digital audio editor and recording computer software application. It was first developed in May 2000 by Dominic Mazzoni and Roger Dannenberg at Carnegie Mellon University. 3 Praat is a free scientific computer software package for the analysis of speech in phonetics. It was first designed in 1995, and continues to be updated, by Paul Boersma and David Weenink of the University of Amsterdam.

Chapter 8:  Embedded Language Islands 1 …bill governor in (168) is not an EL island, because the N+N associative construction represents a sequence of two EL content morphemes in a mixed constituent under ML order. We analysed such cases together with the examples cited in section 6.2.1.2.

References Abney, S. (1987), ‘The English Noun Phrase in its Sentential Aspect’, PhD diss., Massachusetts Institute of Technology, Cambridge, Massachusetts. Adalar, N. and S. Tagliamonte 1998, ‘Borrowed Nouns; Bilingual People: The Case of the “Londrali” in Northern Cyprus’, The International Journal of Bilingualism, 2 (2): 139–59. Ahukanna, J. G. W. (1990), ‘Bilingualism and Code-­mixing in Language Use in Nigeria: The Case of Igbo–English Bilinguals’, in N. Emenanjo (ed.), Igbo Language and Culture, 175–291, Onitsha: Central Books Ltd. Aikhenvald, A.Y. (2006), ‘Serial Verb Constructions in Typological Perspective’, in A. Y. Aikhenvald and R. M. W. Dixon (eds), Serial Verb Constructions: A Cross-­ linguistic Typology, 1–68, Oxford: Oxford University Press. Altenberg, B. (1982), The Genitive v. The of-­construction: A Study of Syntactic Variation in 17th Century English, Lund: LiberFörlag. Amaechi, M. (2013), ‘Case Checking in Igbo Serial Verb Constructions’, Research on Humanities and Social Sciences, 3 (9): 155–64. Ameka, F. K. (2006), ‘Ewe Serial Verb Constructions in their Grammatical Context’, in A. Y. Aikhenvald and R. M. W. Dixon (eds), Serial Verb Constructions: A Cross-­linguistic Typology, 124–43, Oxford: Oxford University Press. Amuda, A. A. (1986), ‘Yoruba-English Code-­switching in Nigeria: Aspects of its Functions and Form’, PhD diss., University of Reading, United Kingdom. Amuzu, E. K. (1998), ‘Aspects of Grammatical Structure in Ewe-English Codeswitching’, M. Phil diss., University of Oslo, Norway. Amuzu, E. K. (2005), ‘Ewe-English Codeswitching: A Case of Composite rather than Classic Codeswitching’, PhD diss., Australian National University, Canberra. Amuzu, E. K. (2010), Composite Codeswitching in West Africa: The Case of Ewe-English Codeswitching, Saarbrucken: Lambert Academic Publishing. Amuzu, E. K. (2013), ‘Bilingual Serial Verb Constructions: A Comparative Study of Ewe-English and Ewe-French Codeswitching’, Lingua, 137 (2013): 19–37. Amuzu, E. K. (2014), ‘Producing Composite Codeswitching: The Role of the Modularity of Language Production’, International Journal of Bilingualism, 18 (4): 384–407. Amuzu, E. K. and J. V. Singler (2014), ‘Codeswitching in West Africa’, International Journal of Bilingualism, 18 (4): 329–45. Andersson, P. (1993), ‘Finns and Americans in Sweden: Patterns of Linguistic Incorporation from Swedish’, in G. Extra and L. Verhoeven (eds), Immigrant Languages in Europe, 249–69, Clevedon: Multilingual Matters.

220

References

Annamalai, E. (1989), ‘The Language Factor in Code-­mixing’, International Journal of the Sociology of Language, 74: 47–54. Anyanwu, R. J. (1998), Aspects of Igbo Grammar: Ponetics, Phonology, Morphology, and the Tonology of Nouns, Hamburg: Lit Verlag. Anyanwu, O. (2012), ‘Pronominal Subject Clitics in Igbo’, Theory and Practice in Language Studies, 2 (2): 377–88. Appel, R. and P. Muysken (1987), Language Contact and Bilingualism, London: Edward Arnold. Arthur, J. and P. Martin (2006), ‘Accomplishing lessons in postcolonial classrooms: Comparative perspectives from Botswana and Brunei Darussalam’, Comparative Education, 42 (2): 177–202. Asilevi, F. (1990), ‘English-Ewe Code-­mixing in Conversational Discourse: A Case of English as a Second Language in Ghana’, MA diss., University of Ghana, Legon. Auer, P. (1995), ‘The Pragmatics of Code-­switching: A Sequential Approach’, in L. Milroy and P. Muysken (eds), One Speaker, Two Languages: Cross-­disciplinary Perspectives on Code-­switching, 115–35, Cambridge: Cambridge University Press. Auer, P. (1998), ‘Introduction’, in P. Auer (ed.), Code-­switching in Conversation, 1–24, London: Routledge. Backus, A. (1992), Patterns of Language Mixing: A Study of Turkish-Dutch Bilingualsim, Wiesbaden: Harrassowitz. Backus, A. (1996), Two in One: Bilingual Speech of Turkish Immigrants in the Netherlands, Tilburg, Netherlands: Tilburg University Press. Backus, A. (1999), ‘Evidence for Lexical Chunks in Insertional Codeswitching’, in E. Lanza and R. Else (eds), Language Encounters Across Time and Space, 93–109, Oslo: Novus Forlag. Backus, A. (2003), ‘Units in Codeswitching: Evidence for Multimorphemic Elements in the Lexicon’, Linguistics, 14: 83–132. Backus, A. and H. Boeschoten (1996), ‘Review of Social Motivations for Codeswitching: Evidence from Africa and Duelling Languages: Grammatical Structure in Codeswitching’, International Journal of Applied Linguistics, 6 (1): 129–53. Belazi, H. M., E. J. Rubin and A. J. Toribio (1994), ‘Code-­switching and X-bar Theory: The Functional Head Constraint’, Linguistic Inquiry, 25 (2): 221–37. Benítez-Torres, C. M. (2009), ‘Inflectional vs. Derivational Morphology in Tagdal: A Mixed Language’, in M. Matondo, F. McLaughlin and E. Potsdam (eds), Selected Proceedings of the 38th Annual Conference on African Linguistics: Linguistic Theory and African Language Documentation, 69–83, Sommervile: Cascadilla Proceedings Project. Bentahila, A. and E. E. Davies (1983), ‘The Syntax of Arabic-French Codeswitching’, Lingua, 59 (4): 301–30. Berk-Seligson, S. (1986), ‘Linguistic Constraints on Intrasentential Code-­switching: A Study of Spanish–Hebrew bilingualism’, Language in Society, 15 (3): 313–48.

References

221

Bhatt, R. M. (1997a), ‘Constraints, Code-­Switching and Optimal Grammars’, Lingua, 4 (102): 123–51. Bhatt, R. M. (1997b), ‘Code-Switching and the Functional Head Constraint’, World Englishes, 1 (16): 171–6. Bhatt, R. M. (1999), Verb Movement and the Syntax of Kashmiri, London: Kluwer Academic Press. Bhatt, R. M. (2001a), ‘Constraint Demotion and Null-­subjects in Spanish L2 Acquisition’, in J. Camps and C. Wiltshire (eds), Romance Syntax, Semantics and their L2 Acquisition, 121–35, Philadelphia: John Benjamins. Bhatt, R. M. (2001b), ‘Code-­switching: Structural Models’, in R. Mesthrie (ed), Concise Encyclopedia of Sociolinguistics, 456–61, Amsterdam: Elsevier Science. Blench, R. (2006), Dictionary of Ònìchà Igbo, Benin: Ethiope Press. Bokamba, E. G. (1988), ‘Code-­mixing, Language Variation, and Linguistic Theory: Evidence from Bantu Languages’, Lingua, 76 (1): 21–62. Bolonyai, A. (1999), ‘The Hidden Dimensions of Language Contact: The Case of Hungarian-English Bilingual Children’, PhD diss., University of South Carolina. Bolonyai, A. (2005), ‘English Verbs in Hungarian-English Codewitching’, in K. T. Cohen, K. R. McAlister and J. MacSwan (eds), Proceedings of the Fourth International Symposium on Bilingualism, 317–27, Somerville, MA: Cascadilla Press. Borer, H. ed. (1986), The Syntax of Pronominal Clitics, Syntax and Semantics, vol. 19, New York: Academic Press. Boussofara-Omar, N. (1999), ‘Arabic Diglossic Switching: An Application of MyersScotton’s MLF model’, PhD diss., University of Texas, Austin. Boussofara-Omar, N. (2003), ‘Revisiting Arabic Diglossic Switching in light of the MLF model and its Sub-­models: The 4-M model and the Abstract Level model’, Bilingualism, Language and Cognition, 6 (1): 33–46. Budzhak-Jones, S. (1998), ‘Against Word-­internal Code-­switching: Evidence from Ukrainian-English Bilingualism’, International Journal of Bilingualism, 2 (2): 161–82. Bullock, B. E. and A. J. Toribio (2009), ‘How to Hit a Moving Target: On the Sociophonetics of Codeswitching’, in L. Isurin, D. Winford and K. de Bot (eds), Multidisciplinary Approaches to Codeswitching, 189–206, Amsterdam: John Benjamins. Campbell, L. (1998), Historical Linguistics: An Introduction, 2nd edn, Cambridge, MA: MIT Press. Canagarajah, S. (2011), ‘Translanguaging in the Classroom: Emerging Issues for Research and Pedagogy’, in L. Wei (ed.), Applied Linguistics Review, 2: 1–27, Berlin, Germany: De Gruyter Mouton. Carter, R., M. McCarthy, G. Mark and A. O’Keeffe (2011), English Grammar Today, Cambridge: Cambridge University Press. Casali, R. F. (2008), ‘ATR Harmony in African Languages’, Language and Linguistics Compass, 2 (3): 496–549.

222

References

Celic, C. and K. Seltzer (2011), Translanguaging: A CUNY-NYSIEB guide for educators. New York: The City University of New York. Chan, B. (2003), Aspects of the Syntax, the Pragmatics and the Production of Codeswitching: Cantonese and English, New York: Peter Long. Chan, B. (2004), ‘Beyond Contextualisation: Codeswitching as a Textualisation Cue’, Journal of Language and Social Psychology, 23 (1): 7–27. Chomsky, N. (1981), Lectures on Government and Binding, Dordrecht: Foris. Chomsky, N. (1995), The Minimalist Program, Cambridge, MA: MIT Press. Chomsky, N. (2000), ‘Minimalist Inquiries: The Framework’, in R. Martin and D. Michaels (eds), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, 89–155, Cambridge, MA: MIT Press. Chomsky, N. (2001), ‘Derivation by Phase’, in M. Kenstowicz (ed.), Ken Hale: A Life in Language, 1–51, Cambridge, MA: MIT Press. Chomsky, N. and H. Lasnik (1995), ‘The Theory of Principles and Parameters’, in N. Chomsky (ed.), The Minimalist Program, 13–127, Cambridge, MA: MIT Press. Cinque, G. (2003), ‘A Note on Verb/Object Order and Head/Relative Clause Order’, PhD diss., University of Venice. Clark, M. M. (1990), The Tonal System of Igbo, Providence: Foris Publications. Clements, G. and E. Sezer (1982), ‘Vowel and Consonant Disharmony in Turkish’, in H. van der Hulst and N. Smith (eds), The Structure of Phonological Representations II, 213–55, Dordrecht: Foris Publications. Clyne, M. (1967), Transference and Triggering, The Hague: Nijhoff. Clyne, M. (2003), Dynamics of Language Contact: English and Immigrant Languages, Cambridge: Cambridge University Press. Couto, M. C. P., M. Deuchar and M. Fusser (2015), ‘How Do Welsh-English Bilinguals Deal with Conflict? Adjective-Noun Order Resolution’, in G. Stell and K. Yakpo (eds), Code-­switching Between Structural and Sociolinguistic Perspectives, 65–84, Berlin: De Gruyter. Crawhall, N. (1990), ‘Unpublished Shona–English data’, cited in C. Myers-Scotton (2002), Contact Linguistics: Bilingual Encounters and Grammatical Outcomes, Oxford: Oxford University Press. Creese, A and A. Blackledge (2010), ‘Translanguaging in the Bilingual Classroom: A Pedagogy for Learning and Teaching?’, The Modern Language Journal, 94 (1): 103–15. Dalby, D. (1977), Language Map of Africa and the Adjacent Islands, London: International African Institute. Davies, P. and M. Deuchar (2010), ‘Using the Matrix Language Frame model to measure the extent of Word Order Convergence in Welsh-English Bilingual Speech’, in A. Breitbarth, C. Lucas, S. Watts and D. Willis (eds), Continuity and Change in Grammar, 77–96, Philadelphia: John Benjamins. Davenport, M. and S. J. Hannahs (2010), Introducing Phonetics and Phonology, London: Routledge.

References

223

Deuchar, M. (2005), ‘Congruence and Welsh-English Code-­switching’, Bilingualism, Language and Cognition, 8 (3): 255–69. Deuchar, M. (2006), ‘Welsh-English Code-­switching and the Matrix Language Frame model’, Lingua, 116 (2006): 1986–2011. Di Sciullo, A. M., P. Muysken and R. Singh (1986), ‘Code-­mixing and Government’, Journal of Linguistics, 22 (1): 1–24. Disterheft, D. (2004), Advanced Grammar: A Manual for Students, Upper Saddle River, NJ: Pearson/Prentice Hall. Dreyfus M. and C. Juillard (2001), ‘Language-­switching in the Everyday Lives of School Children in Dakar and Ziguinchor (Senegal): Variation on the Use of French and Wolof ’, Cahiers d’Études africaines, 41 (163/164): 667–96. Echeruo, M. J. C. (1998), Igbo–English Dictionary, New Haven: Yale University Press. Eliasson, S. (1989), ‘English-Maori Language Contact: Codeswitching and the Free Morpheme Constraint’, Reports from Uppsala University Department of Linguistics, 18: 1–28. Emenanjo, N. (1978), Elements of Modern Igbo Grammar. Ibadan: Oxford University Press. Essizewa, K. E. (2007a), ‘Language Contact Phenomena in Togo: A Case Study of Kabiye-Ewe Codeswitching’, in D. L. Payne and J. Peña (eds), Selected Proceedings of the 37th Annual Conference on African Linguistics, 30–42, Somerville, MA: Cascadilla Proceedings Project. Essizewa, K. E. (2007b), ‘Sociolinguistic Aspects of Kabiye-Ewe Bilingualism in Togo’, PhD diss., New York University, New York. Essizewa, K. E. (2008), ‘Sociolinguistic Aspects of Language Change Among KabiyeFrench Bilinguals’, Journal of West African Languages, 35, 23–31. Essizewa, K. E. (2014), ‘Lexical Insertions in Kabiye-Ewe Codeswitching in Lome’, International Journal of Bilingualism, 18 (4): 408–27. Eze, E. (1997), ‘Aspects of Language Contact: A Variationist Perspective on Codeswitching and Borrowing in Igbo–English Bilingual Discourse’, PhD diss., University of Ottawa, Ottawa. Eze, E. (1998), ‘Lending Credence to a Borrowing Analysis: Lone English-­origin Incorporations in Igbo Discourse’, International Journal of Bilingualism, 2 (2): 183–201. Federal Republic of Nigeria. (2004), National Policy on Education, Abuja: Federal Ministry of Education, Nigeria. Finlayson, R., K. Calteaux and C. Myers-Scotton (1998), ‘Orderly Mixing and Accommodation in South African Codeswitching’, Journal of the Sociology of Language, 2 (3): 395–420. Finley, S. (2008), ‘Formal and Cognitive Restrictions on Vowel Harmony’, PhD diss., Johns Hopkins University, Baltimore. Fleischhacker, H. (2005), ‘Similarity in Phonology: Evidence from Reduplication and Loan Adaptation’, PhD diss., University of California, Los Angeles.

224

References

Forson, B. (1979), ‘Code-­switching in Akan-English Bilingualism’, PhD diss., University of California, Los Angeles. García, O. (2009), Bilingual Education in the 21st Century: A Global Perspective, Oxford: Wiley-Blackwell. García, O. and L. Wei (2014), Translanguaging: Language, Bilingualism and Education, Basingstoke: Palgrave Macmillan. Gardner-Chloros, P. (1991), Language Selection and Switching in Strasbourg, Oxford: Clarendon Press. Goke-Pariola, A. (1983), ‘Code-­mixing Among Yoruba-English Bilinguals’, Anthropological Linguistics, 25 (1): 39–46. Green, M. M. and G. E. Igwe (1963), A Descriptive Grammar of Igbo, London: Oxford University Press. Gumperz, J. J. (1982), Discourse Strategies, Cambridge: Cambridge University Press. Halmari, H. (1997), Government and Codeswitching: Explaining American Finnish. Amsterdam: John Benjamins. Hannahs, S. J. (2001), ‘Morphophonology’, in N. J. Smelser and P. B. Baites (eds), International Encyclopedia of the Social and Behavioral Sciences, 15 (1): 10053–8, New York: Elsevier. Harrison, P. (2000), ‘Making Sense: Embodiment and the Sensibilities of the Everyday’, Society and Space, 18: 497–517. Haust, D. (1995), Codeswitching in Gambia. Eine soziolinguistische Untersuchung von Mandinka, Wolof und English in Kontakt, Köln: Rüdiger Köppe. Heller, M. (1988), ‘Strategic Ambiguity: Code-­switching in the Management of Conflict’, in M. Heller (ed.), Codeswitching, Anthropological and Sociolinguistic Perspectives, 77–96, Berlin: Mouton de Gruyter. Hlavac, J. (2000), ‘Croatian in Melbourne: Lexicon, Switching and Morphosyntactic Features in the Speech of Second-Generation Bilinguals’, PhD diss., Monash University. Igboanusi, H. (2002), Igbo–English in the Nigerian Novel. Ibadan: Enicrownfit. Ihemere, K. U. (2007), A Tri-Generational Study of Language Choice and Shift in Port Harcourt, Boca Raton, FL: Universal Publishers. Jaeggli, O. (1981), Topics in Romance Syntax. Dordrecht: Foris Publications. Jake, J. and C. Myers-Scotton (1997), ‘Codeswitching and compromise strategies: Implications for lexical structure’, The International Journal of Bilingualism, 1 (1): 25–39. Jake, J., C. Myers-Scotton and S. Gross (2002), ‘Making a Minimalist Approach to Codeswitching Work: Adding the Matrix Language’, Bilingualism: Language and Cognition, 5 (1): 69–91. Jones, M. C. (2005), ‘Some Structural and Social Correlates of Single Word Intrasentential Code-­switching in Jersey Norman French’, French Language Studies, 15: 1–23. Joshi, A. (1985), ‘Processing of Sentences with Intrasentential Switching’, in D. R. Dowty, L. Kartunnen and A. Zwicky (eds), Natural Language Parsing, 190–205, Cambridge: Cambridge University Press.

References

225

Kachru, B. (1978), ‘Codeswitching as a Communicative Strategy in India’, in J. E. Atlas (ed.), International Dimensions of Bilingual Education, 107–24, Washington DC: Georgetown University Press. Kachru, B. (1983), The Indianization of English: The English Language in India. Delhi: Oxford University Press. Katamba, F. (1993), Morphology, London: Macmillan Press. Kayne, R. (1994), The Antisymmetry of Syntax, Cambridge, MA: MIT Press. King, R. (2000), The Lexical Basis of Grammatical Borrowing: A Prince Edward Island French Case Study, Amsterdam: John Benjmains. Kiparsky, P. and K. Pajusalu (2003), ‘Towards a Typology of Disharmony’, The Linguistic Review, 20: 217–41. Kite, M. E. (1991), ‘Observer Biases in the Classroom’, Teaching of Psychology, 18: 161–4. Krämer, M. (2003), Vowel Harmony and Correspondence Theory. Berlin: Mouton de Gruyter. Kreyer, R. (2003), ‘Genitive and of-­construction in Modern Written English. Processability and Human Involvement’, International Journal of Corpus Linguistics, 8 (2): 169–207. Labov, W. (1972), ‘Some Principles of Linguistic Methodology’, Language in Society, 1: 97–154. Lamidi, M. T. (2004), ‘The Scopal Authority of Heads in Yoruba-English Code-­ switching’, Nordic Journal of African Studies, 13 (1): 76–94. Lance, D. M. (1975), ‘Spanish-English Code-­switching’, in E. Hernandez-Chavez and A. Beltramo (eds), El lenguaje de los chicanos, 138–53, Arlington, VA: Center for Applied Linguistics. Langacker, R. W. (1987), Foundations of Cognitive Grammar: Theoretical Prerequisites. Stanford, CA: Stanford University Press. Legendre, G. and M. Schindler (2010), ‘Codeswitching in Urban Wolof: A Case for Violable Constraints in Syntax’, ReVEL, 4: 47–75. Levelt, W. J. M. (1989), Speaking: From Intention to Articulation, Cambridge, MA: MIT Press. Lewis, M. P. (ed.) (2009), Ethnologue: Languages of the world, 16th edn, Dallas: SIL International. Lewis, G., B. Jones and C. Baker (2012), ‘Translanguaging: Origins and Development from School to Street and Beyond’, Educational Research and Evaluation: An International Journal on Theory and Practice, 18 (7): 641–54. Liberman, M. and J. M. Schultz (1993), ‘The Phonetic Interpretation of Tone in Igbo’, Phonetica, 50 (3): 147–60. MacSwan, J. (1997), ‘A Minimalist Approach to Intrasentential Codeswitching: Spanish-Nahuatl Bilingualism in Central Mexico’, PhD diss., University of California, Los Angeles. MacSwan, J. (1999), A Minimalist Approach to Intrasentential Codeswitching, New York: Garland Publishing.

226

References

MacSwan, J. (2000), ‘The Architecture of the Bilingual Language Faculty: Evidence from Intrasentential Codeswitching’, Bilingualism: Language and Cognition, 3 (1): 37–54. MacSwan, J. (2005a), ‘Codeswitching and Generative Grammar: A Critique of the MLF model and some Remarks on “Modified Minimalism” ’, Bilingualism: Language and Cognition, 8 (1): 1–22. MacSwan, J. (2005b), ‘Remarks on Jake, Myers-Scotton and Gross’s Response: There is no “matrix language” ’, Bilingualism: Language and Cognition, 8 (2): 277–84. MacSwan, J. (2005c), ‘Précis of a Minimalist Approach to Intrasentential Codeswitching’, Italian Journal of Linguistics, 17 (1): 55–92. MacSwan, J. (2009), ‘Generative Approaches to Codeswitching’, in B. E. Bullock and A. J. Toribio (eds), Cambridge Handbook of Linguistic Codeswitching, 309–35, Cambridge: Cambridge University Press. MacSwan, J. (2010), ‘Understanding Codeswitching Theories’, Proceedings from the Annual Meeting of the Chicago Linguistic Society, 44: 1–32, Chicago: University of Chicago Press. MacSwan, J. (2014), ‘Code-­switching and Linguistic Theory’, in T. K. Bhatia and W. Ritchie (eds), Handbook of Bilingualism and Multilingualism, 223–350, Oxford: Blackwell. Madaki, R. (1983), ‘A Linguistic and Pragmatic Analysis of Hausa-English Codeswitching’, PhD diss., University of Michigan. Maduka-Durunze, O. N. (1990), ‘Igbo Adjectives as Morphologized Relatives’, Studies in African Linguistics, 21 (2): 237–50. Mahootian, S. (1993), ‘A Null Theory of Codeswitching’, PhD diss., Northwestern University. Mahootian, S. and B. Santorini (1996), ‘Codeswitching and the Complement/Adjunct Distinction’, Linguistic Inquiry, 27 (3): 464–79. Manfredi, V. (1991), ‘Agbo and Elugbo: Igbo Linguistic Consciousness, its Origin and Limitations’, PhD diss., Harvard University. Matras, Y. (2009), Language Contact, Cambridge: Cambridge University Press. Meechan, M. and S. Poplack (1995), ‘Orphan Categories and Adjectival Expression: A Comparative Study of Wolof-French and Fongbe-French Bilingual Discourse’, Language Variotion and Change, 7 (1995): 169–94. Milroy, L. (1987), Observing and Analysing Natural Language, Oxford: Oxford University Press. Muysken, P., ed. (1991), ‘Needed: A Comparative Approach’, in European Science Foundation, 253–72, Strasbourg: European Science Foundation. Muysken, P. (1995), ‘Code-­switching and Grammatical Theory’, in P. Muysken and L. Milroy (eds), One Speaker Two Languages: Cross-­disciplinary Perspectives on Code-­switching, 177–98, Cambridge: Cambridge University Press. Muysken, P. (2000), Bilingual Speech: A Typology of Code-­mixing, Cambridge: Cambridge University Press.

References

227

Myers-Scotton, C. (1983), ‘The Negotiation of Identities in Conversation: A Theory of Markedness and Code Choice’, International Journal of the Sociology of Language, 44: 115–36. Myers-Scotton, C. (1993a), Social Motivations for Codeswitching: Evidence from Africa, Oxford: Clarendon Press. Myers-Scotton, C. (1993b), Duelling Languages: Grammatical Structure in Codeswitching. Oxford: Clarendon Press. Myers-Scotton, C. (2001), ‘The Matrix Language Frame Model: Developments and Responses’, in R. Jacobson (ed.), Codeswitcing Worldwide II, 23–58, Berlin: Mouton de Gruyter. Myers-Scotton, C. (2002), Contact Linguistics: Bilingual Encounters and Grammatical Outcomes. Oxford: Oxford University Press. Myers-Scotton, C. (2005), ‘Supporting a Differential Access Hypothesis: Codeswitching and other Contact Data’, in J. L. Kroll and A. De Groot (eds), Handbook of Bilingualism, Psycholinguistic Approaches, 326–48, New York: Oxford University Press. Myers-Scotton, C. (2006), Multiple Voices: An Introduction to Bilingualism, Malden, MA: Blackwell Publishers. Myers-Scotton, C. (2013), ‘Cost Makes a Difference: Codeswitching Frequencies in Bilingual Language Production’, in K. U. Ihemere (ed.), Language Contact: A Multidimensional Perspective, 29–51, Newcastle upon Tyne: Cambridge Scholars Publishing. Myers-Scotton, C. and J. Jake (1995), ‘Matching Lemmas in a Bilingual Language Competence and Production Model: Evidence from Intrasentential Codeswitching’, Linguistics, 33: 981–1024. Myers-Scotton, C. and J. Jake (2000a), ‘Testing the 4-M model: An Introduction’, The International Journal of Bilingualism, 4 (1): 1–8. Myers-Scotton, C. and J. Jake (2000b), ‘Four Types of Morpheme: Evidence from Aphasia, Codeswitching, and Second Language Acquisition’, Linguistics, 38 (6): 1053–100. Myers-Scotton, C. and J. Jake (2001), ‘Explaining Aspects of Codeswitching and Their Implications’, in J. Nicol (ed), One Mind, Two Languages: Bilingual Language Processing, 84–116, Oxford: Blackwell. Myers-Scotton, C. and J. Jake (2009), ‘A Universal Model of Codeswitching and Bilingual Processing and Production’, in B. E. Bullock and A. J. Toribio (eds), Cambridge Handbook of Linguistic Codeswitching, 336–57, Cambridge: Cambridge University Press. Myers-Scotton, C. and J. Jake (2014), ‘Nonfinite Verbs and Negotiating Bilingualism in Codeswitching: Implications for a Language Production Model’, Bilingualism: Language and Cognition, 17 (3): 511–25. Nartey, J. (1982), ‘Codeswitching, Interference or Faddism? Language Use Among Educated Ghanaians’, Anthropological Linguistics, 24 (2): 183–92.

228

References

Ndimele, O. M. and E. Kari (1997), ‘A Minimalist Account of Pro-­drop in Degema’, Nigerian Language Studies, 5: 44–55. Nnaji, H. I. (1985), A Modern English-Igbo Dictionary, Onitsha: Gonaj Books. Nordberg, B. (1982), ‘Sociolingvistisk Datainsamling’, in M. Thelander (ed), Tal-­språksforkning I Norden, 92–119, Lund: Mål-­material-metoder. Nortier, J. (1990), Dutch-Moroccan Arabic Codeswitching among Moroccans in the Netherlands, Dordrecht: Foris Publications. Nwachukwu, P.A. (1984), ‘Stative Verbs in Igbo Syntax’, Journal of the Linguistic Association of Nigeria, 3: 61–74. Nwafor B. U. (1971), ‘ “Engligbo” or the Spoken Language of the Educated Igbo’, Ibadan, 29: 41–4. Obiamalu, G. O. (2013a), ‘The Functional Category D in a Language without Determiners: The Case of Igbo’, Journal of Theoretical Linguistics, 10 (2): 50–67. Obiamalu, G. O. (2013b), ‘On the Role of Functional Categories in Code-­switching: The Igbo-English Examples’, Unizik Journal of Arts and Humanities, 14: 90–119. Obiamalu, G. O. and D. U. Mbagwu (2008), ‘Motivations for Codeswitching among Igbo-English Bilinguals: A Linguistic and Socio-­psychological Survey’, Journal of African Studies, 5: 27–39. Ogbalu, F. C. (1962), Okowa Okwu: Igbo–English/English-Igbo Dictionary, Onitsha: University Publishing Company. Ogbonna, C. C. (1985), ‘Immortalizing Igbo Language’, in Nigerian Statesman, 8: 5, Owerri: Imo Newspapers Ltd. O’Grady, W. D. and J. Archibald, J. (2012), Contemporary Linguistic Analysis: An Introduction, Toronto: Pearson Longman. Okasha, M. (1999), ‘Structural Constraints on Arabic-English Codeswitching: Two Generations’, PhD diss., University of South Carolina, Columbia, South Carolina. O. nwu. Committee (1961), The Official Igbo Orthography as Recommended by the O.nwu. Committee in 1961: Notes on Script and Spelling for Teachers, Enugu: Government of Eastern Nigeria. Owens, J. (2002), ‘Processing the world piece by piece: Iconocity, lexical insertion, and possessives in Nigerian Arabic codeswitching’, Language Variation and Change, 14 (2): 173–209. Owens, J. (2005), ‘Hierarchicalized Matrices: Codeswitching among Urban Nigerian Arabs’, Linguistics, 43 (45): 957–93. Park, H. S. (2000), ‘Korean-Swedish Code-­switching: Theoretical Models and Linguistic Reality’, PhD diss., University of Uppsala, Sweden. Payne, T. E. (1997), Describing Morphosyntax, Cambridge: Cambridge University Press. Pfaff, C. W. (1979), ‘Constraints on Language Mixing: Intrasentential Code-­switching and Borrowing in Spanish/English’, Language, 55: 291–318. Poplack, S. (1980), ‘“Sometimes I’ll Start a Sentence in Spanish y Termino en Español”: Toward a Typology of Code-­switching’, Linguistics, 18: 581–618.

References

229

Poplack, S. (1981), ‘Syntactic Structure and Social Function of Code-­switching’, in R. Duran (ed.), Latino Discourse and Communicative Behaviour, 169–84, New Jersey: Ablex. Poplack, S. (1987), ‘Contrasting Patterns of Code-­switching in two Communities’, in E. Wande, J. Anward, B. Nordberg, L. Steensland and M. Thelander (eds), Aspects of Multilingualism, 51–77, Helsinki: Helsinki University Press. Poplack, S. (1988), ‘Language Status and Language Accommodation along a Linguistic Border’, in P. Lowenberg (ed.), Language Spread and Language Policy: Issues, Implications, and Case Studies, 90–118, Washington DC: Georgetown University Press. Poplack, S. (1990), ‘Variation Theory and Language Contact’, in European Science Foundation, 33–65, Strasbourg: European Science Foundation. Poplack, S. (2012), ‘What does the Nonce Borrowing Hypothesis Hypothesize?’, Bilingualism: Language and Cognition, 15 (3): 644–8. Poplack, S. and S. Sankoff (1984), ‘Variable Concord and Sentence Plural Marking in Puerto Rican Spanish’, The Hispanic Review, 52 (2): 205–22. Poplack, S., S. Wheeler, and A. Westwood (1987), ‘Distinguishing Language Contact Phenomena: Evidence from Finnish-English Bilingualism’, in P. Lilius and M. Saari (eds), The Nordic Languages and Modern Linguistics, 6: 33–56, Helsinki: University of Helsinki Press. Poplack, S., D. Sankoff, and C. Miller (1988), ‘The Social Correlates and Linguistic Processes of Lexical Borrowing and Assimilation’, Linguistics, 26 (1): 47–104. Poplack, S. and M. Meechan (1995), ‘Patterns of Language Mixture: Nominal Structure in Wolof-French and Fongbe-French Bilingual Discourse’, in L. Milroy and P. Muysken (eds), One Speaker, Two Languages, 199–232, Cambridge: Cambridge University Press. Poplack, S. and M. Meechan (1998), ‘How Languages Fit Together in Codeswitching’, International Journal of Bilingualism, 2: 127–38. Quarcoo, M. (2009), ‘Codeswitching in Academic Discussions: A Discourse Strategy by Students in the University of Education, Winneba’, PhD diss., University of Ghana, Legon. Quirk, R., G. Sidney and S. Jan (1985), A Comprehensive Grammar of the English Language, New York: Longman. Radford, A. (1997), Syntax: A Minimalist Introduction, Cambridge: Cambridge University Press. Radford, A. (2004), Minimalist Syntax: Exploring the Structure of English, Cambridge: Cambridge University of Press. Radford, A., M. Atkinson, D. Britain, H. Clashen and A. Spencer (2009), Linguistics: An Introduction, Cambridge: Cambridge University Press. Rahimi, M. and A. Dabaghi (2013), ‘Persian-English Codeswitching: A Test of the Matrix Language Frame Model’, System, 41 (2): 322–51. Riemsdijk, V. H. and E. Williams (1986), Introduction to the Theory of Grammar. Cambridge, MA: MIT Press.

230

References

Rizzi, L. (1986), ‘Null Objects in Italian and the Theory of Pro’, Linguistic Inquiry, 17: 501–57. Roca, I. and W. Johnson (1999), A Course in Phonology, Oxford: Blackwell Publishing. Romaine, S. (2004), ‘The Bilingual and Multilingual Community’, in T. K. Bhatia and W. C. Ritchie (eds), The Handbook of Bilingualism, 385–406, Oxford: Blackwell Publishing. Sankoff, D. and Poplack, S. (1981), ‘A Formal Grammar for Code-­switching’, Papers in Linguistics, 14 (1): 3–45. Sankoff, D. and M. Mainville (1986), ‘Codeswitching of Context-­free Grammars’, Theoretical Linguistics, 13: 75–90. Sankoff, D., M. Nait M’Barek and C. Montpetit (1987), ‘VSO/SVO bilingual Syntax’, New Ways of Analysing Variation, NWA VE XVI, Austin, Texas. Sankoff, D., S. Poplack and S. Vanniarajan (1990), ‘The Case of the Nonce Loans in Tamil’, Language Variation, 2 (1): 71–101. Sankoff, D., S. Poplack and S. Vanniarajan (1991), ‘The Empirical Study of Code-­ switching’, European Science Foundation, 181–206, Strasbourg: European Science Foundation. Sebba, M. (1998), ‘A Congruence Approach to the Syntax of Codeswitching’, International Journal of Bilingualism, 2 (1): 1–21. Selkirk, E. (1984), ‘On the Major Class Features and Syllable Theory’, in M. Aronoff and R. T. Oehrle (eds), Language Sound Structure: Studies in Phonology Presented to Morris Halle by his Teacher and Students, 107–36, Cambridge, MA: MIT Press. Stammers, J. R. and M. Deuchar (2012), ‘Testing the Nonce Borrowing Hypothesis: Counter-­evidence from English-­origin Verbs in Welsh’, Bilingualism: Language and Cognition, 15 (3): 630–43. Stenson, N. (1990), ‘Phrase Structure Congruence, Government, and Irish-English Code-­switching’, in R. Hendricks (ed.), The Syntax of the Modern Celtic Languages, 167–197, San Diego: Academic Press. Swigart, L. (1992), ‘Two Codes or One? The Insiders’ View and the Description of Codeswitching in Dakar’, Journal of Multilingual and Multicultural Development, 13 (1 and 2): 83–102. Timm, L. A. (1975), ‘Spanish-English Code-­switching: El Porque and How-­not-to’, Romance Philology, 28: 473–82. Treffers-Daller, J. (1991), ‘Towards a Uniform Approach to Codeswitching and Borrowing’, European Science Foundation, 259–79, Strasbourg: European Science Foundation. Treffers-Daller, J. (1992), ‘French-Dutch Codeswitching in Brussels: Social Factors Explaining its Disappearance’, Journal of Multilingual and Multicultural Development, 13: 143–56. Treffers-Daller, J. (1994), Mixing Two Languages: French–Dutch Contact in Comparative Perspective. Berlin: Mouton de Gruyter.

References

231

Türker, E. (2000), ‘Turkish-Norwegian Codeswitching: Evidence from Intermediate and Second Generation Turkish Immigrants in Norway’, PhD diss., University of Oslo. Uwalaka, M. A. (1982), ‘Igbo consecutization revisited’, Journal of Linguistic Association of Nigeria, 1: 63–72. Uwalaka, M. A. (1997), Igbo Grammar, Ibadan: The Pen Services. Valdés-Fallis, G. (1976), ‘Social Interaction and Codeswitching Patterns: A Case Study of Spanish-English Alternation’, in G. D. Keller, R. V. Teschner and S. Viera (eds), Bilingualism in the Bicentennial and Beyond, 86–96, New York: Bilingual Press. Valdés-Fallis, G. (1981), ‘Codeswitching as a Deliberate Verbal Strategy: A Microanalysis of Direct and Indirect Requests among Chicano Speakers’, in R. Duran (ed.), Latino Language and Communicative Behavior, 95–107, Norwood, NJ: Ablex Publishing. Van Gass, K. (2002), ‘Grammatical Constraints on Intrasentential Codeswitching: Evidence from English-Afrikaans Codeswitching’, Stellenbosch Papers in Linguistics, 31: 91–113. Webb, V. and K. Sure (eds) (2000), African Voices: An Introduction to the Languages and Linguistics of Africa, Cape Town: Oxford University Press. Wei, L. (2001), ‘Lemma Congruence Checking between Languages as an Organizing Principle in Intrasentential Codeswitching’, International Journal of Bilingualism, 5 (2): 153–73. Weinreich, U. (1953), Languages in Contact: Findings and Problems. New York: Linguistic Circle of New York. Welmers, W. E. (1973), African Language Structures. Berkeley: University of California Press. Williams, C. (1994), ‘Arfarniad o ddulliau dysgu ac addysgu yng nghyd-­destun addysg uwchradd ddwyieithog’ [An Evaluation of Teaching and Learning Methods in the Context of Bilingual Secondary Education], PhD diss., University of Wales, Bangor. Williamson, K. (1972 [2006]), The Igbo Dictionary, 2nd edn, Benin: Ethiope Press. Zsiga, E. C. (1997), ‘Features, Gestures, and Igbo Vowels: An Approach to the Phonology-­phonetics Interface’, Language, 73 (2): 227–74. Zwicky, A. M. (1977), On Clitics, Bloomington: Indiana University Press.

Index 4-M. See Four Morpheme model abstract level, 5, 49, 95, 162, 190, 218, 224 activated, 25, 26, 27, 39, 47, 49, 57, 95, 117, 121, 124, 162, 183, 188, 216, 224 addressee specification. See discourse functions Adjacency Principle, 198, 199 adjectival nouns, 65, 104, 113, 114, 115 adjective, 4, 24, 26, 28, 42, 54, 103, 116, 117, 129, 136, 137, 179, 181, 182, 185, 192 Advanced Tongue Root, 62, 152 adverbs, 42, 65 affix, 35, 36, 46, 60, 62, 64, 79, 95, 98, 140, 142, 143, 155, 171, 222 Agent, 40, 50, 145 agreement morphology, 29, 50, 106, 118 alternation, 14, 23, 60, 95, 224 animacy, 74, 219 associative construction, 65, 66, 112, 129 asymmetry, 5, 21, 24, 36, 40, 53, 56, 97, 131, 132, 171, 172, 195, 196, 201, 204, 210 Asymmetry Principle, 19, 53, 101, 131, 201, 204, 212, 229, 230 ATR. See Advanced Tongue Root attitudes, 10, 17, 18, 84, 199 attributive, 113, 114, 116, 117, 136, 137, 184, 185 auxiliary verb, 90, 118, 119, 122, 142, 165, 181, 183 bare forms, 43, 46, 57, 90, 91, 120, 121, 161, 162, 163, 169, 170, 171, 214, 219, 220 base language, 12, 23 bilingual, 2, 7, 8, 12, 13, 16, 17, 18, 19, 21, 34, 38, 39, 43, 44, 47, 50, 51, 53, 54, 57, 71, 76, 77, 78, 80, 81, 82, 83, 84, 87, 88, 89, 90, 94, 95, 96, 98, 99, 101, 108, 109, 110, 111, 112, 114, 116, 117, 118, 119, 121, 126, 127, 128, 129, 131, 132, 133,

135, 136, 138, 139, 140, 141, 143, 151, 155, 156, 157, 163, 164, 165, 166, 169, 170, 171, 172, 173, 174, 177, 179, 180, 181, 183, 184, 188, 193, 195, 196, 199, 202, 203, 204, 206, 208, 210, 212, 214, 220, 222, 224, 229, 230, 231 bilingualism, 10, 15, 17, 95, 214 borrowings, 14, 15, 19, 21, 23, 42, 54, 61, 80, 92, 93, 94, 95, 96, 98, 100, 151, 152, 157, 174, 175, 177, 188, 189, 201, 222, 223, 224, 228, See language contact phenomena bound morpheme, 5, 25, 78, 94, 174 bridge morpheme. See morphemes case, 2, 5, 12, 16, 21, 29, 35, 43, 44, 45, 46, 55, 65, 66, 74, 75, 76, 79, 84, 91, 97, 98, 100, 101, 102, 103, 104, 108, 109, 110, 117, 118, 119, 121, 123, 138, 142, 145, 150, 151, 152, 156, 157, 158, 166, 169, 171, 172, 173, 174, 175, 178, 180, 183, 189, 191, 197, 200, 206, 208, 210, 211, 212, 214, 218, 220, 230, 231 categories, 27, 42, 54, 64, 89, 97, 99, 117, 118, 119, 162, 166, 167, 187, 188, 189, 208, 226 chunks, 182, 198 classic codeswitching, 14, 36, 39, 47, 51, 53, 82, 87, 88, 100, 101, 102, 103, 124, 127, 131, 166, 171 clause, 2, 5, 6, 7, 18, 23, 25, 36, 38, 39, 40, 42, 43, 44, 49, 51, 53, 67, 70, 72, 75, 76, 77, 87, 88, 89, 90, 91, 95, 97, 98, 101, 102, 104, 113, 115, 117, 118, 119, 121, 128, 132, 133, 136, 138, 140, 141, 144, 149, 165, 170, 172, 174, 177, 178, 181, 183, 191,192, 193, 194, 195, 200, 202, 204, 210, 212, 218, 219, 220, 224, 229, 230 clitics, 42, 75, 77, 78, 79, 91, 140, 146, 149, 206 coda consonant, 61, 156, 157, 158, 159, 160

234

Index

codemixing, 23 Codeswitching, 2, 4, 6, 8, 10, 13, 15, 17, 19, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 34, 35, 36, 39, 42, 44, 46, 49, 51, 54, 56, 58, 60, 62, 65, 67, 69, 71, 73, 75, 77, 79, 81, 83, 85, 87, 89, 91, 93, 94, 96, 98, 100, 102, 106, 108, 110, 112, 114, 116, 118, 120, 122,124, 126, 128, 131, 133, 136, 138, 140, 142, 144, 146, 149, 151, 153, 155, 157, 159, 161, 163, 165, 167, 170, 172, 174, 178, 180, 182, 184, 186, 188, 190, 192, 194, 196, 198, 200, 202, 206, 210, 214, 218, 220, 223, 226, 229, 231 colonial language, 12, 13 complementizer, 28, 29, 30, 75, 76, 87, 91, 191, 192, 218 complementizers. See complementizer complements, 26, 29, 36, 44, 67, 108, 113, 115, 117, 118, 162, 181, 184, 188, 191, 192, 195, 207, 208, 218 composite codeswitching, 13 compromise strategy, 46, 57, 121, 169, 170, 173, 212, 219, 220 computational system, 31 conditional, 68 congruence, 28, 35, 38, 53, 54, 55, 56, 96, 99, 116, 119, 137, 139, 142, 143, 144, 169, 188, 219, 224 congruence checking. See congruence conjunction, 27, 133, 141, 165 conjunctions, 73, 75, 231 consonant clusters, 61, 156, 157, 160, 174 constituents, 14, 26, 35, 38, 40, 44, 49, 52, 77, 87, 88, 90, 102, 117, 133, 135, 138, 169, 171, 172, 173, 177, 178, 185, 202, 206, 208, 216, 229 constraints, 4, 7, 13, 14, 19, 23, 24, 25, 26, 28, 29, 32, 34, 39, 57, 61, 88, 93, 135, 137, 138, 156, 160, 173, 197, 201, 206, 222, 226, 231 constructions, 13, 35, 58, 65, 66, 68, 69, 71, 78, 103, 110, 111, 112, 113, 142, 146, 164, 165, 167, 170, 173, 208, 220 content morphemes, 40, 42, 44, 47, 49, 117, 118, 133, 140, 172, 177, 188, 216, 224 contextualization cue, 8 convergence. See language contact phenomena conversation, 9, 10, 84, 94, 135, 208, 210

conversations, 19, 81, 82, 84, 85, 86, 98, 212 copula, 26, 27, 66, 115, 116, 117, 119, 133, 137, 186, 193 corpus, 14, 21, 80, 84, 85, 86, 95, 98, 99, 104, 109, 110, 115, 123, 129, 135, 139, 143, 150, 167, 208, 212, 228, 231 Creoles. See language contact phenomena CS. See Codeswitching demonstrative, 43, 45, 74, 89, 103, 126, 133, 136, 156, 216 derivation, 31, 32, 35 determiner, 27, 28, 30, 35, 44, 71, 74, 77, 89, 91, 93, 94, 103, 114, 121, 122, 126, 128, 129, 133, 135, 139, 172, 179, 180, 181, 182, 184, 189, 197, 206, 208, 216, 219, 220 dialect zones, 17 dictionaries. See dictionary dictionary, 61, 94, 96, 127, 143, 151, 152, 174, 222, 223 Differential Access hypothesis, 47 directive. See social functions of codeswitching discourse, 7, 8, 13, 14, 15, 18, 40, 49, 85, 87, 95, 192, 219, 224, 231 discourse functions, 7 distribution, 14, 15, 46, 82, 98, 107, 231 do construction, 26, 170, 173 dominant-­recessive. See vowel harmony double morphology, 45, 46, 47, 90, 91, 102, 120, 214, 216, 218 DP hypothesis, 105, 122 Early System Morpheme Hypothesis, 47, 124, 216, 218 EC. See Equivalence Constraint EL. See Embedded Language Embedded Language islands, 21, 43, 44, 55, 86, 88, 89, 97, 98, 100, 121, 140, 175, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 193, 194, 196, 197, 198, 200, 204 elements, 2, 7, 8, 17, 23, 24, 27, 28, 34, 36, 38, 39, 43, 44, 47, 51, 54, 55, 56, 57, 66, 72, 77, 78, 91, 93, 94, 95, 96, 97, 98, 99, 103, 104, 106, 107, 108, 109, 111, 113, 116, 122, 123, 125, 126, 129, 135, 136, 138, 139, 145, 146, 156, 165, 169, 174,

Index 178, 179,183, 193, 195, 197, 198, 199, 202, 204, 210, 212, 216, 229 Embedded Language, 5, 38, 40, 101, 139, 141, 143, 145, 148, 150, 152, 154, 156, 158, 160, 162, 164, 166, 169, 171, 173, 175, 177, 178, 179, 181, 183, 185, 187, 189, 191, 193, 195, 197, 199, 200, 201, 202, 204, 205, 206, 210, 212, 216, 218, 220, 224, 226, 228, 229 epenthesis, 61, 157, 158, 159, 160, 223 Equivalence Constraint, 4, 19, 226 ESMH. See Early System Morpheme Hypothesis expressive. See social functions of codeswitching feature checking, 55, 56 feature mismatches, 35, 39 FHC. See Functional Head Constraint finite verb, 4, 140, 142, 165, 170 FMC. See Free Morpheme Constraint formulator, 49, 120, 124, 144, 145, 216 Four-Morpheme model, 45, 46 frame-­building, 38, 39, 50, 97, 101, 138, 174, 204, 210, 224 Free Morpheme Constraint, 5, 19, 25, 226 Functional Head Constraint, 19, 29, 226 functional projection, 106, 125, 126, 218 GC. See Government Constraint generative grammar, 27 genitive, 65, 78, 103, 110, 111, 112, 113 Government Constraint, 19, 27, 226 grammar, 13, 15, 23, 24, 26, 31, 32, 34, 51, 58, 59, 71, 82, 84, 90, 108, 109, 111, 114, 117, 122, 123, 124, 125, 137, 152, 162, 166, 170, 171, 173, 177, 180, 188, 189, 190, 216, 218, 219, 220, 226, 231 grammatical, 4, 5, 7, 12, 13, 15, 18, 19, 23, 31, 32, 34, 35, 38, 39, 40, 50, 51, 52, 54, 55, 56, 59, 60, 64, 65, 68, 77, 82, 84, 87, 88, 93, 95, 96, 97, 99, 108, 116, 117, 118, 119, 120, 126, 139, 140, 142, 144, 166, 167, 170, 171, 173, 183, 202, 212, 214, 216, 224,231 head movement, 106, 122 heads, 29, 34, 36, 49, 105, 107, 108, 113, 124, 125, 127, 216, 226

235

hierarchical structure. See Embedded Language Islands Igbo, 2, 4, 5, 6, 7, 8, 9, 10, 13, 14, 15, 16, 17, 18, 19, 21, 24, 26, 28, 29, 30, 31, 32, 34, 35, 38, 39, 42, 44, 46, 47, 49, 51, 53, 54, 56, 57, 58, 59, 60, 61, 62, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 85, 86, 87, 88, 89, 90, 91, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 131, 132, 133, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 204, 205, 206, 208, 210, 212, 214, 216, 218, 219, 220, 222, 223, 224, 226, 228, 229, 230, 231 Imperative, 68 inceptive, 68, 171 incongruence, 46, 55, 57, 96, 117, 121, 135, 173, 224 indeterminate, 114, 121, 122 Indicative affirmative, 68 Infinitive, 62, 68, 69 infinitive complement, 4 inflection, 25, 30, 49, 71, 79, 94, 140, 141, 142, 144, 145, 162, 167, 171, 186, 189, 208 inflectional morphology, 21, 34, 35, 53, 118, 120, 127, 128, 141, 143, 144, 145, 146, 156, 157, 162, 169, 173, 220 Inflectional Phrase, 30, 90, 208, 218 inflections, 26, 32, 46, 50, 91, 98, 109, 118, 126, 138, 140, 142, 161, 165, 167, 170, 171, 172, 173, 178, 179, 180, 181, 184, 187, 188, 189, 198, 204, 208, 211, 220 in-­group, 7, 8, 81, 85 insertion, 12, 26, 55, 57, 88, 94, 119, 122, 137, 141, 142, 144, 158, 170, 182, 198, 220, 224 interference. See language contact phenomena

236 interjections. See discourse functions internal EL island. See Embedded Language Islands Internal EL islands. See Embedded Language Islands interpersonal relationships, 9 intersentential, 5, 6 intrasentential, 5, 12, 13, 18, 19, 21, 23, 29, 35, 56, 57, 81, 88, 97, 99, 167, 177, 183, 201, 202, 208, 214, 226, 231 IP. See Inflectional Phrase judgement sample, 80, 81 language contact phenomena, 2, 10, 46 language mixing, 16, 23 language production, 38, 39, 46, 47, 50, 51, 57, 120, 126, 144, 171, 183, 193, 201, 230 lemma, 55, 56, 116, 119, 120, 121, 124, 144, 145, 188 lexical attrition. See language contact phenomena lexical- conceptual structure, 50, 142, 145 lexical form, 5, 25, 174 lexicon, 31, 32, 34, 47, 49, 50, 57, 94, 124, 216, 229, 230 linguistic, 2, 4, 8, 9, 12, 15, 17, 18, 19, 23, 27, 31, 32, 34, 39, 46, 51, 54, 55, 56, 60, 81, 88, 137, 197, 198, 199, 200, 228, 231 listedness, 96, 222 loanwords, 61, 94, 161, 224, 228 majority language, 12 marked choices, 9 markedness evaluator. See Markedness Model Markedness Model, 9 Matrix Language Frame, 5, 229, 230 Matrix Language, 5, 13, 19, 38, 39, 40, 51, 55, 94, 102, 201, 202, 204, 206, 208, 210, 212, 214, 216, 218, 219, 220, 222, 224, 226, 228, 229, 230 mental lexicon, 47, 50, 57, 230 message qualification. See discourse functions metalinguistic functions. See social functions of codeswitching metaphorical, 10

Index Minimalist Approach, 19, 23 minority language, 12 mixed constituent, 40, 43, 44, 45, 89, 97, 101, 119, 121, 127, 128, 129, 140, 166, 202 mixed constituents, 5, 26, 35, 40, 45, 50, 51, 53, 64, 77, 88, 89, 91, 95, 114, 127, 128, 132, 133, 135, 161, 166, 172, 174, 179, 195, 196, 200, 202, 204, 206, 208, 224, 230 mixed nominal expressions, 103, 107, 109, 123, 139, 144 ML. See Matrix Language, See Matrix Language MM. See Markedness Model modals, 68, 118, 188 monolingual, 5, 8, 12, 15, 16, 23, 26, 31, 32, 35, 39, 57, 62, 72, 87, 90, 93, 94, 95, 96, 102, 108, 109, 110, 111, 112, 114, 121, 125, 126, 135, 140, 141, 142, 164, 165, 170, 174, 179, 188, 189, 222, 223, 224 MOP. See Morpheme Order Principle morpheme order criterion, 51, 101, 102, 113, 115, 116, 118, 119, 122, 139, 145, 167, 171, 178, 179, 183, 188, 193, 194, 202 Morpheme Order Principle, 40, 51, 102, 201, 202, 206, 216 morphemes, 7, 8, 39, 40, 42, 44, 45, 46, 47, 49, 50, 51, 52, 53, 54, 55, 67, 77, 87, 89, 90, 91, 101, 102, 111, 115, 116, 117, 118, 119, 121, 123, 124, 125, 128, 129, 131, 132, 133, 135, 138, 140, 142, 146, 151, 152, 155, 156, 159, 162, 166, 167, 169, 171, 172, 173, 174, 177, 178, 179, 180, 182, 183, 184, 185, 186, 187, 188, 190, 192, 193, 194, 195, 196, 197, 198, 199, 201, 202, 204, 206, 211, 212, 216, 224, 228, 229, 230 morphological, 2, 50, 56, 62, 77, 87, 95, 112, 120, 142, 144, 145, 169, 170, 222 morphophonology, 59, 60 morphosyntactic, 5, 19, 26, 36, 39, 51, 54, 56, 88, 97, 101, 125, 133, 135, 138, 139, 145, 150, 172, 173, 175, 185, 197, 202, 206, 208, 210, 212, 214, 219, 224, 226, 230 morphosyntactic frame, 5, 36, 39, 51, 56, 97, 101, 125, 133, 138, 139, 175, 202, 208, 210, 212, 214, 224, 230

Index mother tongue, 13, 16, 51, 230 multilingual, 10, 83 multi-­morphemic, 26, 198 native speakers, 16, 17, 84, 96, 159, 222 Negative imperative, 68, 69 Negative indicative, 68 neutral vowel. See vowel harmony Niger– Congo, 15 Nigerian Pidgin English, 10, 19 nominal, 26, 66, 67, 77, 94, 103, 105, 106, 107, 108, 109, 110, 111, 113, 120, 122, 123, 125, 126, 134, 135, 139, 144, 179, 180, 206, 218 Nonce Borrowing Hypothesis, 223 noun, 4, 24, 28, 30, 40, 42, 43, 44, 45, 46, 49, 50, 54, 91, 93, 99, 103, 104, 107, 108, 109, 110, 111, 112, 114, 115, 116, 117, 119, 120, 123, 124, 125, 126, 127, 135, 136, 137, 139, 146, 179, 180, 182, 185, 206, 210, 216, 219 numerals, 72, 73, 74, 77, 104, 109, 133, 147, 196, 206 object, 4, 6, 25, 32, 70, 76, 85, 104, 121, 145, 162, 164, 165, 171, 185, 186, 219, 229 objectivization. See discourse functions observer’s paradox, 19, 86 opaque. See vowel harmony orthography, 59, 60, 86 out-­group, 7 paradigmatic, 54, 56, 116, 137 participle, 68, 171 Patient, 40, 145 perfective, 68, 144, 171 personalization. See discourse functions PF Disjunction Theorem, 32, 34 PF Interface Condition, 32 PFIC. See Phonetic Form Interface Condition phatic. See social functions of codeswitching phonetic form, 32 phonologically integrated, 5, 25, 94, 157, 161, 174, 189, 228 phonologies, 21, 64, 94, 151, 152, 159, 160, 174, 222, 228, 230

237

phonology, 21, 32, 60, 61, 151, 152, 161, 162, 228, 230 phonotactics, 143, 222 phrases, 4, 19, 21, 23, 24, 26, 35, 42, 43, 49, 67, 96, 99, 113, 142, 177, 179, 181, 183, 187, 189, 192, 193, 196, 198, 199, 206, 208, 228, 229 Pidgins. See language contact phenomena plural, 45, 46, 47, 73, 74, 91, 93, 107, 109, 114, 117, 123, 124, 125, 126, 133, 136, 146, 156, 179, 180, 216, 220 Port Harcourt, 10, 18, 81, 85, 86, 100, 108, 135 possessor, 15, 49, 111, 112, 113, 129 possessum, 112, 113, 129 post-­nominally, 66, 115 pragmatic functions, 8 predicate–argument structure, 50, 56, 87, 119, 139, 142, 144, 145, 169, 190 predicative, 66, 113, 116, 117, 129, 137 pre-­nominally, 66, 115 prepositions, 42, 47 pro licensing, 78, 149 pro-­drop, 78, 91, 146, 149 proficiency, 8, 51, 82, 214 pronominal, 4, 71, 72, 73, 74, 75, 77, 78, 79, 91, 93, 94, 104, 108, 133, 140, 146, 149, 150, 179, 196, 206 pronominal subject, 4, 75, 77, 78, 79, 91, 140, 146 pronouns, 74, 75, 122, 133, 146, 147, 148, 149, 150, 206 Qualifactive. See nouns qualifactive nouns, 66, 115 qualitative, 15, 80, 135, 138, 169, 200, 202, 230 quantifier, 30, 72, 89, 103, 104, 229 quantitative, 14, 15, 18, 80, 97, 100, 102, 127, 135, 138, 141, 166, 178, 183, 196, 197, 198, 200, 201, 202, 230 quotations. See discourse functions referential. See social functions of codeswitching region, 4, 12, 13, 14, 15, 80 reiteration. See discourse functions repertoire, 8, 10, 12, 34 rights and obligations set, 9

238

Index

RO. See rights and obligations set root, 35, 36, 62, 68, 98, 152, 153, 155 root control harmony. See vowel harmony second language acquisition. See language contact phenomena sentence, 2, 4, 5, 6, 14, 24, 40, 52, 64, 65, 70, 76, 77, 87, 88, 93, 94, 104, 117, 118, 119, 133, 135, 136, 187, 190, 198, 202 serial verb, 35, 69, 208, 220 singly occurring English nouns, 35, 99, 109, 115, 128, 132, 135, 169, 179, 180 singly occurring English verbs, 35, 138, 139, 144, 149, 151, 157, 169, 173, 174 situational, 10 SMP. See System Morpheme Principle social functions of CS, 10 social network, 80, 81 sociolinguistic, 2, 4, 7, 10, 12, 15, 84, 199, 231 sounds, 14, 62, 154, 158 spectrogram, 157, 158, 159, 160 spectrograms. See spectrogram speech community, 81, 94, 95, 96 structural conflicts, 39, 103, 136 structural dependency relationships. See Embedded Language Islands subject argument position, 78, 146 suffixes, 69, 118, 143, 149, 188 surface features, 35 surface-­level linear differences, 4, 98 switching, 4, 5, 6, 8, 13, 21, 24, 25, 27, 28, 29, 30, 35, 93, 94, 98, 135, 136, 138, 139, 144, 150, 173, 175, 180, 183, 185, 186, 188, 193, 194, 196, 198, 199, 201, 226 switching sites, 27 syllable, 60, 61, 64, 153, 154, 155, 156, 157, 158, 159, 160, 161, 174, 222, 223, 228 syntactic, 2, 5, 23, 24, 29, 32, 34, 46, 57, 65, 70, 75, 77, 78, 87, 88, 98, 102, 105, 113, 121, 122, 124, 133, 141, 149, 150, 151, 162, 165, 170, 173, 190, 203, 218, 219, 220 syntagmatic, 27, 54, 56, 116, 117, 135, 137 syntax, 4, 34, 40, 55, 59, 65, 75, 77, 105, 136, 145, 187, 188, 191, 192, 199, 218, 224, 226 system inflectional morphemes. See System Morpheme Principle System Morpheme Principle, 40, 51, 117, 201, 202, 204, 206, 212, 214, 216

system morphemes, 39, 42, 45, 47, 49, 50, 53, 91, 115, 117, 118, 121, 123, 124, 125, 129, 132, 133, 138, 166, 167, 169, 172, 173, 180, 183, 184, 188, 194, 195, 196, 197, 206, 212, 216 textualization cue, 8 thematic roles, 41 tone, 60, 108, 112 translanguaging, 10, 12 transparent. See vowel harmony triggering, 103, 125, 197, 198, 199 typological differences, 59, 78, 79, 101 Uniform Structure Principle, 19, 53, 101, 131, 201, 206, 208, 212, 214, 219, 220, 229, 230 Universal Grammar, 105 unmarked choice, 9, 10 Urban Wolof, 34, 155, 174, 226 utterances, 2, 5, 13, 21, 23, 64, 83, 84, 86, 98, 99, 100, 137, 138, 143, 161, 173, 180, 226, 228 UW. See Urban Wolof verb phrase, 30 verb serialization, 70, 141, 164, 165, 166 verbs, 13, 15, 21, 29, 40, 42, 46, 47, 50, 53, 61, 64, 67, 70, 71, 75, 91, 97, 98, 99, 113, 115, 116, 117, 118, 133, 137, 138, 139, 140, 141, 142, 143, 145, 146, 152, 153, 155, 156, 157, 159, 160, 161, 162, 163, 164, 165, 166, 167, 169, 170, 171, 172, 173, 174,186, 187, 188, 189, 197, 198, 201, 203, 204, 208, 214, 220, 224, 228 VH. See vowel harmony vowel, 21, 60, 61, 62, 63, 64, 68, 70, 75, 119, 140, 142, 143, 144, 146, 149, 152, 153, 154, 155, 156, 157, 158, 159, 160, 174, 193, 201, 223, 230 vowel harmony, 21, 60, 140, 152, 155, 201, 230 West Africa, 4, 12, 13, 15, 17 word order, 4, 7, 13, 24, 25, 40, 44, 53, 54, 93, 101, 102, 103, 105, 115, 116, 119, 120, 127, 129, 132, 137, 138, 139, 140, 145, 169, 171, 172, 183, 184, 185, 186, 190, 194, 195, 202, 204, 210, 229