PHONOLOGICAL AND MORPHOLOGICAL NATIVIZATION OF ENGLISH NOUNS

BORROWED INTO EKEGUSII: A CONSTRAINT-BASED APPROACH

BY:

ANYONA GEORGE MORARA

BA (UNIVERSITY OF NAIROBI), PGDE (KENYATTA UNIVERSITY), MA

(EGERTON UNIVERSITY)

A THESIS SUBMITTED TO THE SCHOOL OF POST-GRADUATE STUDIES IN

PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE OF

DOCTOR OF PHILOSOPHY OF THE FACULTY OF ARTS AND SOCIAL SCIENCES,

DEPARTMENT OF LINGUISTICS, LANGUAGES AND LITERATURE,

KISII UNIVERSITY.

NOVEMBER, 2017 DECLARATION DECLARATION BY THE CANDIDATE

This thesis is my original work and it has not been submitted in this or any other university known to me.

Sign…………………………………………………… Date………………………

George Morara Anyona

DAS13/60010/2014

DECLARATION BY THE SUPERVISORS

This thesis has been submitted with our approval as university supervisors.

Sign ……………………………………………………….. Date…………………………

Dr. David O. Ongarora, PhD

Department of Linguistics, Maseno University

Sign ……………………………………………………….. Date…………………………

Dr. Evans G. Mecha, PhD

Department of Lit., Lang. & Ling., Kisii University

ii PLAGIARISM DECLARATION DECLARATION BY STUDENT

i. I declare I have read and understood Kisii University Postgraduate Examination Rules and Regulations, and other documents concerning academic dishonesty. ii. I do understand that ignorance of these rules and regulations is not an excuse for a violation of the said rules. iii. If I have any questions or doubts, I realize that it is my respondibility to keep seeking an answer until I understand. iv. I understand I must do my own work. v. I also understand that if I commit any act of academic dishonesdty like plagiarism, my thesis/project can be assigned a fail grade (“F”). vi. I further understand I may be suspended or expelled from the university for academic dishonesty.

Name: George Morara Anyona Signature:…………………………..

Reg. No.: DAS13/60010/2014 Date:………………………………..

DECLARATION BY SUPERVISOR(S) i. I/we declare that this thesis has been submitted to plagiarism detection service. ii. The thesis contains less than 20% of plagiarized work. iii. I/we hereby give consent for making.

1. Name: Dr. David O. Ongarora, PhD Signature:…………………………

Affiliation:………………………………………... Date:………………………………

2. Name: Dr. Evans G. Mecha, PhD Signature:…………………………

Affiliation:………………………………………... Date:………………………………

iii DECLARATION OF NUMBER OF WORDS

Name of Candidate: GEORGE MORARA ANYONA Adm. No.: DAS13/60010/2014

Faculty: ARTS AND SOCIAL SCIENCES

Department: LINGUISTICS, LANGUAGES AND LITERATURE

Thesis Title: PHONOLOGICAL AND MORPHOLOGICAL NATIVIZATION OF ENGLISH

NOUNS BORROWED INTO EKEGUSII: A CONSTRAINT-BASED

APPROACH

I confirm that the word length of:

1) The thesis, including footnotes, is 64, 004 2) the bibliography is 3, 972

And, if applicable, 3) the appendices are 3, 219

I also declare the electronic version is identical to the final, hard bound copy of the thesis and corresponds with those on which the examiners based their recommendation fro the award of the degree.

Signed:……………………………………………… Date:……………………………….. (Candidate)

I confirm that the thesis submitted by the above-named candidate complies with the relevant word length specified in the School of Postgraduate and Commission of University Education regulations for the Masters and PhD Degrees.

Signed:………………………..Email:………………….Tel.:……………….Date:…………… (Supervisor 1)

Signed:………………………..Email:………………….Tel.:……………….Date:…………… (Supervisor 2)

iv COPYRIGHT

All rights are reserved. No aprt of this thesis or information herein may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the author of Kisii University on that behalf. © 2017, Anyona Morara George

v ABSTRACT This study examines the phonology and morphology of English nouns borrowed into EkeGusii within Optimality Theory. This theory provides that the well formedness in natural languages is constraint governed. Thus, even borrowed lexical items obey these constraints. Constraints are universal and are ranked on a language specific order. The focus of this study is to investigate the phonological and morphological adjustments that English nouns borrowed into EkeGusii undergo as they harmonize themselves with EkeGusii phonological and morphological systems as they adapt EkeGusii constraints ranking order. The study analyzes segmental, phonotactic and suprasegmental adaptations that English nouns borrowed into EkeGusii undergo in the process of being nativized. Further, the study analyzes the affixation processes that characterize these nouns. The study analyses nominal prefixation, augmentation, and classification. The study was guided by the following objectives: to describe the phonological and morphological structures of EkeGusii and English nouns, to analyze the phonological adjustments that English nouns borrowed into EkeGusii undergo during nativization and to analyze the morphological changes that the English noun borrowed into EkeGusii undergo during nativization. The study adopted the descriptive and the explanatory research designs. Data was collected from a sample of 13 interviewees using semi-structured interviews. The interviewees were purposively sampled based on the semantic domains of the nouns collected. All the 349 English nouns borrowed into EkeGusii colleccted from the field constituted the sample size of the study. secondary data was used to describe the phonological and morphological strucures of EkeGusii in response to question 1. Library study and researcher’s intuition were the sources of the secondary data. To test the validity and reliability of the research instrument, pre-testing was carried out and appropriate adjustments and corrections made on the instument. The study yielded the following results. The first objective describes the phonological and morphological structures of English and EkeGusii. It was established that: Phonologically; the acoustic nature of the vowels in the two languages differ significantly, EkeGusii syllable structure is (V)CV, while that of English is (C)V(C), and EkeGusii is characterized by a number of phonological processes not found in English. Morphologically; EkeGusii nouns, unlike those of English, are grouped into classes determined by the semantics of the root of the noun, and affixation in the two languages is different. The second objective analyzed the phonological adjustments that English nouns borrowed into EkeGusii undergo. It was established that: phonemes of English not present in EkeGusii are substituted for those present in EkeGusii, the foreign syllable structure from English is re-syllabified to EkeGusii syllable structure, and English stress is tonemized in EkeGusii.The third objective analyzed the morphological changes that English nouns borrowed into EkeGusii undergo during nativization. It was found that: English nouns enter EkeGusii nominal classes which are semantically determined, nativization of the English nouns borrowed into EkeGusii is characterized by the pre-prefix and, semantics plays a significant role in morphological nativization of English nouns in EkeGusii. This study has establisshed that Optimality Theory can successively analyse English nouns borrowed into EkeGusii and that the nouns obey EkeGusii constraint ranking. It is hoped that this study would contribute to the theoretical understanding of borrowed word phonology and morphology of EkeGusii in particular and that of Bantu languages in general.

vi DEDICATION To Moraa, my spouse and my children: Anyona, Nyantari, Nyaboke, Kerubo and Mokeira.

vii ACKNOWLEDGEMENT Many hands helped in coming up with this dissertation but given the limitations of space, I may not mention all of them, nor will I record all their individual contributions. However, the following people and institutions deserve special mentioning and recognition.

Much credit goes to my supervisors, Dr. Evans Gesura Mecha and Dr. David Ogoti Ongarora for their valuable and unending advice, guidance and direction right from the time of topic selection and refining, and indeed to the final preparation of this dissertation. You were a blessing from God. Thank you.

My sincere appreciations go to Professor Augustine Agwele of Texas State University-USA (an adjunct professor in KSU); firstly, for his scholarly advice and mentorship; secondly, for teaching me scholary patience; and thirdly and most importantly, for taking his time to read, critique and edit this work. Thanks a lot prof.

I also take this opportunity to thank Dr. Nilson Opande, former COD, Department of Ling., Lang., and Lit.; now director, Board of Undergraduate Studies; Dr. Barasa Margaret, former COD, Ling., Lang., and Lit.; now current Dean, FASS, Kisii University; for their encouragement and efforts of all manner of kinds to ensure that this work is completed within time. You are not only my senior colleagues and mentors, but also my dear friends as well.

As it is to every writer of a work of this nature, I owe a debt to all my teachers; from pre-school to this level, all my colleagues and my dear students; to you all, I gratefully acknowledge my deep indebtness.

Much credit also go to the thirteen (13) great men and women of Nyagaachi village, Echoro sub- location, of Nyamira County, led by their able assistant chief, Mr. David Orina and village elder Mr. James Aricha, for readily and willingly availing themselves for interviewing and thus providing the data that was analyzed in this study. Your eagerness and willingness to participate in this study was simply amazing. To all of you I say God bless you.

I would also like to extend my appreciations to all my course mates, Magutu, Omari, Nyoteyo and Omosa, for their unending advice and input during course work and indeed during the preparation of this thesis. To you all I say: yes we can!

viii I will not be fair if I do not mention the ladies who spent several hours typing and formatting this work during its preparation. To Regina Obonyo and Rhoda Makori, I say thank you and God bless you for your efforts, patience and time.

I will not end these aknowlegements without mentioning the following institutions: firstly, Kisii University for giving me the opportunity to be one of its pioneering PhD students in Linguistics; secondly, the department of Ling., Lang., and Lit., for providing human and material support required to undertake the programme; most notably, the language laboratory; thirdly, the Counnty government of Nyamira, for not only allowing me to conduct research in Couny, but also providing to me all the information requested for; and finally, the Government of Keny, through NACOSTI, for permitting me to carry out this work. I am sincerely grateful to all these institutions.

Finally, and by no means not least, to my dear wife Mary, my daughters; Laura, Esther and Joyline, my sons; Edwin and Robinson, my mother, Nyaboke; and all my sisters and brothers, I say a big thank you for your time, constant and continuous encouragement and support, understanding, and patience during the painful period of preparing this thesis. You gave me a peace of mind required in doing this kind of work. God bless you all.

TABLE OF CONTENTS DECLARATION...... ii ix PLAGIARISM DECLARATION...... iii

DECLARATION OF NUMBER OF WORDS...... iv

COPYRIGHT...... v

ABSTRACT...... vi

DEDICATION...... vii

ACKNOWLEDGEMENT...... viii

TABLE OF CONTENTS...... x

LIST OF ABBREVIATIONS, SYMBOLS, TABLES, TABLEAUX , FIGURES AND CHARTS ...... xvii

CHAPTER ONE

INTRODUCTION...... 1

1.0 Background to the study...... 1

1.1 Statement of the problem...... 5

1.2 Objectives of the research...... 6

1.3 Research questions...... 6

1.4 Justification and significance of the study...... 6

1.5 Scope and limitation...... 9

CHAPTER TWO

LITERATURE REVIEW AND THEORETICAL FRAMEWORK...... 12

x 2.0 Introduction...... 12

2.1 EkeGusii phonological and morphological structures...... 14

2.1.1 EkeGusii phonological structure...... 14

2.1.2 EkeGusii morphological structure...... 18

2.2 Phonological nativization...... 19

2.3 Morphological nativization...... 25

2.4 Theoretical framework...... 28

2.4.1 Tenets of Optimality Theory...... 28

2.4.2 Optimality Theory versus other Generative Approaches...... 35

CHAPTER THREE

RESEARCH METHODOLOGY...... 40

3.0 Introduction...... 40

3.1 Research design...... 40

3.2 Research site...... 41

3.3 Study populations...... 41

3.4 Sample size and sampling techniques...... 42

3.5 Data collection procedures...... 43

3.6 Data analysis...... 43

3.6.1 Procedure data analysis...... 44

3.7 Ethical considerations...... 45

CHAPTER FOUR

xi DATA PRESENTATION, ANALYSIS, RESULTS AND DISCUSSIONS...... 46

4.0 Introduction...... 46

4.1 Phonological and morphological structures of EkeGusii and English...... 46

4.1.1 EkeGusii and English vowel systems...... 47

4.1.1.1EkeGusii vowel harmony and disharmony...... 59

4.1.2 EkeGusii consonant system in comparison with English...... 63

4.1.2.1 EkeGusii consonantal processes...... 70

4.1.2.1.1 Voice dissimilation (Dahl’s Law)...... 70

4.1.2.1.2 Prenasalisation and nasal homorganisation...... 72

4.1.2.1.3 Declusterization of nasal consonant (NC) and consonant glide (CG) clusters...... 73

4.1.2.1.4 Defricativisation...... 82

4.1.2.1.5 EkeGusii Nasal re-syllabification...... 83

4.1.3 Comparison of EkeGusii and English syllable structures...... 89

4.1.3.1 EkeGusii and English phonotactics...... 102

4.1.3.1.1 English word initial phoneme sequences...... 103

4.1.3.1.2 English word final phoneme sequences...... 105

4.1.4 Comparison of EkeGusii toneme structure and English stress...... 107

4.1.4.1 EkeGusii tone structure...... 107

4.1.4.1.1 Underlying and surface tones in EkeGusii...... 109

4.1.4.1.2 Contour tones in EkeGusii...... 110

4.1.4..1.3 Tone floating in EkeGusii...... 112

xii 4.1.5 Comparison of EkeGusii and English morphological structures...... 114

4.1.5.1 Comparison of EkeGusii and English noun systems...... 115

4.1.5.2 Morphosyntactic classes and prefixes in EkeGusii noun...... 116

4.1.5.2.1 The pre-prefix or augment...... 121

4.1.5.2.2 The Prefix...... 125

4.1.5.2.3 EkeGusii noun class roots...... 127

4.1.5.2.4 Class 1 {omo-} and 2 {aba-}...... 128

4.1.5.2.5 Classes 3 {omo-} and 4 {eme-}...... 128

4.1.5.2.6 Classes 9 {e-}, 10 {chi-}, 9 (a) {en-} , 10 (a) {chin-}...... 130

4.1.5.2.7 Classes 5 {eri-/rii-}, 6 {ama-}, 7 {eke-}, 8 {ebi-}, 12 {aka-} and 14 {obo-}...... 131

4.2 Phonological nativization of English nouns in EkeGusii...... 135

4.2.1 Segmental nativization...... 136

4.2.1.1 Nativization of vowel segments...... 137

4.2.1.1.1 Nativization of English pure (monophthong) vowels...... 138

4.2.1.1.2 Nativization of English diphthongs...... 157

4.2.1.1.3 Nativization of English triphthongs...... 166

4.2.1.2 Nativization of English consonants...... 173

4.2.1.2.1 Nativization of English /f/ and /v/...... 174

4.2.1.2.2 Nativization of English /ð/ and //...... 179

4.2.1.2.3 Nativization of English /l/...... 183

4.2.1.2.4 Nativization of English /z/...... 187

xiii 4.2.1.2.5 Nativization of English /∫/ and / ʒ/...... 189

4.2.1.2.6 Nativization of English /g/ in EkeGusii...... 192

4.2.2 Phonotactic nativization...... 195

4.2.2.1 Nativization by vowel epenthesis...... 200

4.2.2.1.1 Nativization of complex syllable margins...... 201

4.2.2.1.2 Nativization of closed syllables...... 205

4.2.3 Supra-segmental nativization...... 210

4.2.4 Nativization by phonological processes...... 216

4.2.4.1 Nativization by voice dissimilation (Dahl’s Law)...... 216

4.2.4.2.1 nativization by defricativization...... 225

4.2.4.2.2 Spirantization/fricativization of the stops...... 230

4.2.4.2.3 Nativization by bilabialization of labiodentals...... 233

4.2.4.3 Nativization of nasal consonant clusters (NC) by pre-nasalization and homorganization ...... 235

4.2.4.4 Nativization by vowel height harmonization and disharmonization...... 239

4.3 Morphological nativization...... 241

4.3.1 Nativization by nominal classification...... 243

4.3.1.1 The role of semantics in morphological nativization of English nouns in English...... 252

4.3.1.2 Nativization into classes 1 and 2 Nouns...... 262

4.3.1.3 Nativization of English loans into EkeGusii in classes 1b {mo-} and 2b [ba-}...... 264

4.3.1.4 Nativization into classes 3 and 4...... 265

xiv 4.3.1.5 Nativization in classes 5 {eri-/rii-}, 6 {ama-}, 7 {eke-}, 8 {ebi-}, 12 {aka-} and 14 {obo-}...... 265

4.3.2 Prefixation and pre-prefixation...... 268

4.3.2.1 Nativization by prefixation...... 269

4.3.2.2 Nativization by pre-prefixation...... 272

CHAPTER FIVE

SUMMARY, CONCLUSIONS AND RECOMMENDATIONS...... 276

5.1 Summary...... 276

5.2 Conclusions...... 281

5.3 Recommendations...... 282

REFERENCES...... 285

APPENDICES...... 307

Appendix I: Interviewee profile form...... 307

Appendix II: Interviewee consent form...... 308

Appendix III: Research instrument: semi-structured interview guide...... 309

Appendix IV: Raw data recording/transcription form...... 312

Appendix V: Raw data...... 313

Appendix VI: Illustrations of phonological nativization...... 327

Appendix VII: Illustrations of morphological nativization...... 328

Appendix VIII: Enchoro sub-location villages...... 329

Appendix IX: Research Clearance Permit...... 330

xv Appendix X: Research Authorization Letter...... 331

Appendix XI: Memo from the Registrar, Research and Extension, Kisii University to NACOSTI ...... 332

Appendix XII: Plagiarism Report...... 333

Appendix XIII: Publication...... 336

xvi LIST OF ABBREVIATIONS, SYMBOLS, TABLES, TABLEAUX , FIGURES AND CHARTS

ABBREVIATIONS AFX affix APP applicative AUG a pre-prefix (a vowel) C consonant CON constraints (in the Optimality theory) EVAL evaluator FV final vowel GEN generator OT Optimality Theory IND indicative mood L left edge of a word N nucleus (of a syllable) NC nasal Consonant O onset consonant (of a syllable) OM object Marker P nominal Person PL plural SG singular SM subject Marker 1 3 SG class one noun in the third person singular 7 2 PL class seven noun in the second person plural R right edge of a word RT root of a word TNS tense V vowel H high for tonal tier / high tone HTS high tone spans xvii L O low for tonal tier ATR advanced tongue root RTR retracted tongue root UF underlying form of a language SF surface form of a language OCP obligatory contour principle

TBU tone bearing unit

FAITH C an Optimality Theory constraint that demands that consonants in the input

be the same as the consonants in the output FAITH V an Optimality Theory constraint that demands that vowels in the input be

the same as the vowels in the output IDENT-IO (F) the specification that the features of an input segment must be preserved in its output correspondent INDENT-IO (place) the specification that place of articulation of an input segment must be

preserved in its output correspondent *VOI voiced obstruents are not allowed

*VTV voiceless obstruents are banned in intervocalic positions

MAX- IO input segments must have output correspondents (no deletion) MAX-V input vowels must have output correspondents(no deletion) *CODA syllables are open ( must not have codas) ONSET syllables must have onsets DEP no epenthesis or addition of either a vowel or consonant

*OCP(dorstop) a markedness constraint which bans sequence of voices dorsal stop

*OCP(V) a markedness constraint which bans sequence of voiced consonants

*COPLEX(C) a constraint which bans complex consonants

*COPLEX(V) a constraint which bans a string of vowels

*BACK V back vowels are not allowed

xviii *CENTRAL V central vowel are not allowed

*LAX V lax vowels are not allowed

*HIGH V high vowels are not allowed

AGREE V vowels in a word should agree

AGREE C consonants in a word should agree

*STRIDENT (FRIC) strident fricates are not allowed

*INTERD (FRIC) interdental fricatives are not allowed

*LATERAL (C) lateral consonants are not allowed

*TRILL (C) trill consonants are not allowed

*VOICEOBS voiced obstruents are not allowed

*NONANTERCORFRIC non-anterior coronal fricatives are not allowed

*ASSIM (F) assimilation of features are not allowed in a given domain

*UNIFORMITY prohibits feature fusing

*FLOAT markedness constraint which prohibits floating of tones

SPECIFY a constraint which demands that a tone bearing unit must have a

correspondent tone

MAX IO (MORPH) morphological faithfulness constraint which demands that an input

morpheme must have an output correspondent no change

DEP IO (MORPH) a morphological faithfulness constraint which demands that there should

be no epenthesis of a morpheme

(ALIGN(AFX, R; RT, L)) a morphological alignment constraint which demands that the right edge of an affix should be aligned to the left edge of a root (a prefix) (ALIGN(AFX, L; RT,R)) a morphological alignment constraint which demands that the left edge

of an affix should be aligned to the right edge of a root (a suffix) xix STRPRES a faithfulness constraint which demands that a structure in the input should

be preserved in the output, no structure change.

xx SYMBOLS / / Slashes, enclose phonemic forms

[ ] Square brackets, enclose phonetic forms

{ } Braces, enclose morphemes

| | Pipes, enclose underlying forms

→ Arrow, a phonological form realized as

> Morphological form meaning realized as

>> Domination (constraint) in ranking

σ Syllable node

µ Syllabic molar

* Constraint violation

☞ Optimal candidate

! Fatal violation Unbroken association line indicating prelinking in Autosegmental Phonology

Broken association lines indicating linking in Autosegmental phonology

– Tone marker for level tone ` Tone marker (low tone) ´ Tone marker (high tone) ' syllable marker Ø A nonexistent segment to be substituted or inserted for another

xxi TABLES

Table 1: Some of the OT constraints from the universal set……………………… 31

Table 2: F1 and F2 formant frequencies in hertz (Hz) of EkeGusii vowels……… 51

Table 3: F1 and F2 formant frequencies in hertz (Hz) of EkeGusii vowels [i, e, ͻ, u]…...... 56 Table 4: F1 and F2 formant frequencies in hertz (Hz) of the English vowels [i, e, ͻ, u]………………………………………………………………… 56 Table 5: Typology of syllable shapes……………………………………………. 89

Table 6: EkeGusii prefixes and their stems semantic determinants…………….. 118

xxii TABLEAUX

Tableau 2.1: Modern English realization of the input /∫ip/. ……………………...34

Tableau 2.2: Comparative tableau for Yawelmani input /Ɂilk-hin/………………37

Tableau 2.3: Violation tableau for Yawelmani input /Ɂilk-hin/…………………. 38

Tableau 4.1: English output of the input / krɪsmǝs/………..……………………………. 141 Tableau 4.2: EkeGusii output of the input /kirismasi/…………………………. 142 Tableau 4.3: English input and output of /∫ɜ:t /……………………………...... 148 Tableau 4.4: EkeGusii output of the input /esati /……………………………... 148 Tableau 4.5: EkeGusii output of the input /ekerasi/…………………………… 150 Tableau 4.6: English input and output of /∫ɜ:t /………………………………….. 150 Tableau 4.7: English input and output of /sɪneɪt/………………………………. 153 Tableau 4.8: EkeGusii output of the input /eseneti/…………………………….. 154 Tableau 4.9: EKeGusii realization of the input /erokeseni/………………………160 Tableau 4.10: English output of the input /lǝʊkeɪ∫n/…………………...... 160 Tableau 4.11: EkeGusii realization of the input /waja/………………...... 171 Tableau 4.12: English input and output of /waɪǝ/………………………………… 171 Tableau 4.13: EkeGusii output of the input /eurit∫i/……………………………. 177 Tableau 4.14: English input and output of /frɪdʒ/…………………..…………….. 177 Tableau 4.15: Tonga realization of the English input /frɪdʒ/…………………..…. 179 Tableau 4.16: English output of the input /Өɜ:məs/…………………...…………. 182 Tableau 4.17: EkeGusii output of the the input /etamosi/………………... ………182 Tableau 4.18: EkeGusii output of the the input /ekerasi/………………………….186 Tableau 4.19: KiKamba realization of the English input /sku:l/………………..... 186 Tableau 4.20: EkeGusii output of the the input /esiro/…………….…………….. 188 Tableau 4.21: English output of the input /zirəʊ/……………………...………….. 189 Tableau 4.22: EkeGusii output of the input /esati /…………………………….. 191 Tableau 4.23: English output of the input /∫ɜ:t /…………… ………………….. 191 Tableau 4.24: EkeGusii output for the input /epaγi/……………………………… 194 Tableau 4.25: English output for the input /bæg/……………………………….. 194

Tableau 4.26: EkeGusii output and input of /ma.ma/……………………………... 198

Tableau 4.27: English output for the input /ǝʊ/…………..……..……………….. 199

Tableau 4.28: English output of the input / teɪk/…………………………………. 199

Tableau 4.29: English output of the input /steɪns/……………………………….. 199

Tableau 4.30: English output of the input /sku:l/…………………………………. 203

Tableau 4.31: EkeGusii output of the input /sukuru/……………………………. 203

xxiii Tableau 4.32: English output of the input /tͻ:t∫/………………………….. ………207

Tableau 4.33: EkeGusii output for the input [tͻͻt∫i]……………………………… 207

Tableau 4.34: Lanakel output of the input [t-n-ak-ol]…………………………… 208

Tableau 4.35: Lanakel output of the input [no-n-koma-i]…………..…………….. 210

Tableau 4.36: EkeGusii output of the input / oraŋgeti /……………...... 215

Tableau 4.37: EkeGusii output of the input /okokoro/…………………………… 219

Tableau 4.38: English output of the input input /bæŋk/……………………………222

Tableau 4.39: EkeGusii output of the English input / eeŋgi/…………………….222

Tableau 4.40: EkeGusii output of the input /kani/……………………..…………229

Tableau 4.41: EkeGusii output for the input /etaγisi/……………………………..232

Tableau 4.42: EkeGusii output of the input /aranda/…..……………………… 234

Tableau 4.43: EkeGusii output for the input /aranda/ (unit)…………………… 237

Tableau 4.44: EkeGusii output of the input /εntεrεa/…………….……………. 240

Tableau 4.45: EkeGusii output for the neutral input {nkooko}………………… 244

Tableau 4.46: EkeGusii output for the input {nkooko} ………………….....….. 245

Tableau 4.47: English output for the singular form input {boy}…………..…… 246

Tableau 4.48: English output of the input {boy}{-s}…………………………… 246

Tableau 4.49: English output the plural input {blanket}{-s}……………………248

Tableau 4.50: EkeGusii output of the English plural input {blanket}{-s}……….. 248

Tableau 4.51: English output of the input {tractor}{-s}………………..………… 254

Tableau 4.52: EkeGusii output of the English input {tractor}{-s}……………….. 255

Tableau 4.53: EkeGusii output for input {omo-}{te}………….………………… 257

Tableau 4.54: English output for the input {records}{-s}……………………….. 258

xxiv Tableau 4.55: English output of the input {record}………………..…………….. 259

Tableau 4.56: EkeGusii output of the English singular input {record}{-s}……... 260

Tableau 4.57: Tableau (4.57): EkeGusii output of the English input {record}{-s}…… 261

Tableau 4.58: EkeGusii output of the English input {omo-}{chiibu}…………… 263

Tableau 4.59: English output of the input {chief}…………..……………………..264

Tableau 4.60: EkeGusii output of the English input {blanket}………………….. 270

Tableau 4.61: EkeGusii output of the English input {gaβana}……………….…. 275

CHARTS Page

Chart 1: EkeGusii vowels………………………………………………………… 47 Chart 2: EkeGusii vowel diagram……………………………………………… 48 Chart 3: EkeGusii consonants…………………………………………………. 64 Chart 4: EkeGusii consonant inventory……………………………………….. 66 Chart 5: The English consonants………………………………………………. 68 Chart 6: Description of English /ʌ/ and EkeGusii /a/…………………………. 146 Chart 7: Production of the English diphthong [aɪ]……………………………. 163 Chart 8: Production of EkeGusii nativized form [ae] from the English Diphthong [aɪ]………………………………………………………… 163

xxv FIGURES Page

Figure 1: Process of candidate elimination in OT…………………………………….30

Figure 2: Process of OT realization of output………………………………………44 Figure 3: Spectrograms of EkeGusii vowels [a, e, ε, i, ͻ, o, u] collected from a Native speaker…………………………………………………………… 50 Figure 4: Plot of F1 against (F2) formant frequencies in hertz (Hz) of EkeGusii vowels……………………………………………………………………..51

Figure 5: Spectrograms of 8 British English vowels………………………………. 57

Figure 6: English diphthongs……………………………………………………….58

Fifure 7: Articulation of complex glides in EkeGusii……………………………… 76

Figure 8: EkeGusii consonant glide syllabic nodes……………………………….... 77

Figure 9: Nasal consonant syllabification in EkeGusii……………………………… 79

Figure 10: EkeGusii vowel lengthening……………………………………………… 81 Figure 11: EkeGusii compensatory lengthening…………………………………….... 81

Figure 12: EkeGusii de-linking rule……………………………………………………85

Figure 13: EkeGusii nasal re-syllabification rule…………………………………….. 86

Figure 14: Nasal delinking leading to lasal syllabification…………………………… 87

Figure 15: Nasal consonant cluster forming part of a syllable……………………….. 87

Figure 16: EkeGusii syllable nodes for /ketii/………………………………………… 100

Figure 17: EkeGusii syllable nodes for [o.mo.te] and [a.aa]………………………… 101

Figure 18: EkeGusii lexical contrasts due to tone structure…………………………. 108

xxvi Figure 19: EkeGusii lexical contrasts due to tone structure………………………….. 109

Figure 20: EkeGusii tone marking……………………………………………………. 110

Figure 21: Mende contour tones……………………………………………………… 111

Figure 22: EkeGusii LH contour tone………………………………………………… 111 Figure 23: Ekegusii level tone...... 112

Figure 24: EkeGusii floating tone ……………………………………………...... 112

Figure 25: EkeGusii prefix and pre-prefix syllable nodes ………………………….. 121

Figure 26: Substitution of the English /ɪ/ for EkeGusii [i].…………………………. 139

Figure 27: EkeGusii nativization of the English central vowels ……………………. 155 Figure 28: Substitution of the English /ǝɪǝ/ for EkeGusii [aja]……………………….. 167 Figure 29: Substitution of the English /f/ for EkeGusii [β]………………………… 176 Figure 30: Substitution of the English /Ө/ for EkeGusii [t]…………………………. 180 Figure 31: Substitution of the English /l/ for EkeGusii [r]………………………….. 183 Figure 32: Substitution of the English /z/ for EkeGusii [s]…………………………….187 Figure 33: Substitution of the English /∫/ for EkeGusii [s]……………………………. 190 Figure 34: Substitution of the English /g/ for EkeGusii [γ]…………………………….193 Figure 35: EkeGusii syllable types …………………………………………………… 197

Figure 36: English syllable types…………………………………………………… 197

Figure 37: Nativization by anaptyxis……………………………………………….. 202

Figure 38: Increase of number of Syllables in nativized word forms due to anaptyxis………204

Figure 39: Opening of the English closed syllables by paragogic epenthesis in

EkeGusii………………………………………………………………… 206

Figure 40: EkeGusii dissimilation from |γ| → [k]………………………………….. 218

Figure 41: Substitution of the English /k/ for EkeGusii [g]………………………… 220

Figure 42: Substitution of the English /d/ for EkeGusii [t]; English /k/ for EkeGusii [g]……………………………………………………………… 220 Figure 43: Realization of EkeGusii /γ/ as [g] ………………………………………. 226 Figure 44: Realization of EkeGusii // as [b] ……………………………………… 227 Figure 45: Realization of EkeGusii /r/ as [d]………………………………………. 227

xxvii Figure 46: English and EkeGusii treatment of nasal + consonant sequences……… 237

xxviii CHAPTER ONE

INTRODUCTION

1.0 Background to the study The purpose of this study is to examine the nature of the phonological and morphological processes involved in the nativization of nouns borrowed from English into EkeGusii language in order to understand the internal patterns of the phonology and morphology of these nouns. The focus of the study is on two languages in contact- EkeGusii, the borrowing language, and

English, the lending language. In this sub-section, a brief background to the study is given.

The sub-section begins with a brief description of the languages under focus, in which an outline of the languages in terms of where they are spoken, by who, and the language families to which they belong. Further, it is shown how and why the two languages came into contact among other highlights. This is then followed the background of the study.

According to Cammenga (2002:20), EkeGusii is “an Eastern Nyanza Bantu language spoken in the northern part of the region between the eastern shore of Lake Victoria, or Lake Nyanza and the eastern branch of the Great Rift Valley. This region, which includes the islands along the lakeshore, will be designated in geographical terms as East Nyanza, so that EkeGusii may be classified more specifically as an East Nyanza Bantu language. It is labeled E. 42 by Guthrie

(1971, vol. 3:45, vol. 3:11-15).” EkeGusii language is spoken exclusively in Kenya. However, in some literature, including Cammenga, it is wrongly observed that the language is more commonly known by its English name, viz. Kisii (Ongarora, 2009). The correct position is that the language does not have an English name as claimed by Cammenga. What Cammenga calls the English name of the language (Kisii) is a name commonly used to refer to the administrative

1 and business town of the region or sometimes to refer to the entire region occupied by the

AbaGusii people. Therefore, as Cammenga points out, “in the language itself, its proper name is

[ekeγusii] (EkeGusii), it is spoken by the [aβaγusii] (AbaGusii) people” (2002: 20). This is supported by the observation by Demuth (2000) and Ongarora (2009) that names of the languages in Bantu fall under classes 7/8 of the Bantu classification. {eke-}, therefore is a prefix marking language. Just like in Ongarora (2000), to be in conformity with the native speakers of the language, the name EkeGusii, which bears the class prefix to denote language, is used in this study. EkeGusii boarders the following languages: Kipsigis, a Nilotic language, to the north and partly to the East, Maasai, Cushitic, to the east and partly to the southeast, DhoLuo, which is Nilotic, is spoken to the west and southwest of EkeGusii, and finally Kuria, a Bantu language (E.43), is spoken to the south of EkeGusii and “of all the East Nyanza Bantu Languages it is probably most closely related to EkeGusii” (Cammenga, 2002:21). The language is spoken by over 2 million

AbaGusii (Kenya National Bureau of Statistics, 2009), scattered world over. According to Bosire and Machogu (2013), EkeGusii is closely related to other Bantu languages of the region described above, and share common ancestry with KiKuria and Suba, spoken in

Tanzania as well; KiNyarwanda spoken in Rwanda; and KiMeru, KiKuyu, and KiMaragoli, all spoken in Kenya. According to Crystal (1987) and Grimes (1996) English belongs to the Western Germanic branch of the Indo-European group of languages. Indo-European is the family of languages that first spread throughout Europe and many parts of southern Asia, and are now found in every part of the world because of colonialism among other factors (Crystal, 1987). This family, Grimes

(1996) observe includes several branches: Indo-Iranian, Gothic, Italic, Baltic, and Germanic,

2 among others. Of these branches, the Italic and the Germanic are of significance as far as the study of the development of English is concerned. According to Crystal (1987) English is a world language because of the political and socio- economic progress the English speaking countries have made over the past 200 years. Crystal further observes that conservative estimates put mother-tongue speakers of English as 300 million, second language users as another 300 million, and fluent users as a foreign language as

100 million people. Since this observation was made almost two decades ago, users of the language in the given areas could have increased a great deal by now. According to Ethnologue,

English is spoken by 983, 522, 920 people across the globe. Thus English is indeed a world language. Owino (2003) observes that English first came to Kenya through the Christian missionaries, traders and the British colonialists. Currently, it is estimated that there are approximately 900 million speakers of the language all over the world, many of them being second language speakers Ethnologue In Kenya, like in most other African countries, the language has few first language users. Majority of the users in the country are second language speakers who acquire the language through the education system (Owino, 2003). Several definitions of the term nativization have been advanced by linguists. For example,

Haugen (1970:432) sees nativization as the “attempted reproduction in one language, patterns previously found in another”, Hock (1991:390) defines it as “the integration of foreign words into one’s native structures”; while Mberia (2004) observes that nativization or borrowing is the adjustments that borrowed words undergo at the various linguistic levels in order to be accommodated in another language. Definitions of nativization indicate that borrowed words undergo certain linguistic processes in order to be accommodated or accepted in the target

3 language. In other words, the words undergo adaptation processes to conform to the structural constraints of the borrowing language. For borrowing of words to occur, languages must come into contact. In other words, the speakers of the borrowing language must be in contact with those of the loaning language (Haugen, 1970).

EkeGusii language has had such contact with a number of languages including, Dholuo, Maasai and Kalenjin because of sharing a common ethnic border; and English and Kiswahili due to factors such as trade, education, and colonization among others. As a result, several nouns from these languages have found their way into EkeGusii language as illustrated by (1). 1) EkeGusii borrowed nouns Ekegusii borrowed noun source noun source language gloss /eγetaβu/ Kitabu Swahili book /emesa / meza Kiswahili table /esukuru/ School English school | eretio| / eretjo/ radio English radio |riraβuani| /rirawani/ rabuon Dholuo sweet potato / ɳasae / nyasaye Dholuo God Source: Bosire & Machogu (2013) These borrowed nouns come into the target language with different linguistic structures

(phonologically and morphologically), which means that for the loanwords to be accepted into

EkeGusii grammar, they have to undergo readjustments structurally in terms of the various linguistic processes. This study deals with the phonological and morphological readjustment processes that characterize nativization of EkeGusii loans from English. Loan word nativization processes have been studied extensively in among others, Volland

(1986), Hock (1991), Yip (1993), Davidson & Noyer (1997), Jacobs & Gussenhoven (2000),

Kang (2003), Mberia (2004), Rose & Demuth (2006), Mwita (2009), Owino (2003), Meiseburg

(1993), Hall & Hamann (2003), Heinemann (2003), Dupoux & Peperkamp (2002), Lacharite &

Paradis (2005), Zivenga (2009), Raiz (2011), and Chebanne & Phili (2015). However, even if this is the case, the exact nature, especially that of the internal phonology and morphology of loan words in the target language, is not well understood. The questions being: which

4 phonological and morphological language constraints do these loans obey (target’s or source’s), given that each natural language has language particular ranking of the universal constraints

(Prince and Smolensky, 1993)? Which phonological and morphological language processes do the loans words undergo in order to be accommodated in their new environments, given that some phonological and morphological processes are language specific, affecting one language and not the other and vice versa? These are some of the questions that guided this study.

1.1 Statement of the problem

The well-formedness of natural language is constraint-governed. Even loaned lexical items obey these constraints. Constraints are universal and are ranked on a language specific basis; thus, no two languages have a similar ranking of constraints. The focus of this study is to investigate the phonological and morphological adjustments that English nouns borrowed into EkeGusii undergo to harmonize themselves with EkeGusii phonological and morphological systems.

1.2 Objectives of the research

The purpose of this study is to contribute towards the understanding of nativization of the phonological and morphological structures of English nouns borrowed into EkeGusii. The study seeks to achieve the following objectives:

i) To describe the phonological and morphological structures of English and EkeGusii.

ii) To analyse and discuss the phonological changes that English nouns borrowed into

EkeGusii undergo.

iii) To analyse and discus the morphological changes that English nouns borrowed into

EkeGusii undergo.

5 1.3 Research questions

The following questions will guide and aid the achievement of the objectives listed above:

i) How dissimilar are the phonological and morphological structures of EkeGusii and

English nouns?

ii) How does the EkeGusii phonological structure impact on the English nouns borrowed

into EkeGusii?

iii) How does the EkeGusii morphological structure impact on the English nouns borrowed

into EkeGusii?

1.4 Justification and significance of the study

Literature reviewed show that not much research has been conducted in EkeGusii, in particular in phonological and morphological nativization of loan words. Existing literature further show that research done on the language has focused on phonology, morphology and syntax

(Whiteley,1960 and 1965; Osinde, 1988; Ongarora, 1996 and 2009; Mecha, 2006 and 2013, and

Nash, 2011), among others. No known study therefore has focused on nativization of EkeGusii borrowed nouns from English or any other language. This study therefore is pioneering in the study of borrowed words into EkeGusii, and it is significant because it acts as a safeguard to the language which is exposed to an influx of loans (Zivenge, 2009). Analysis of nativization of

English nouns borrowed into EkeGusii is significant because it focuses on how EkeGusii modifies the English nounss in order to conform to its (EkeGusii) phonological and morphological systems and status in a global linguistic enclave, thus protecting the language

(EkeGusii) from extinction.

6 Available literature indicates that many studies on nativization processes have been conducted within the framework of Optimality Theory - a fairly recent constraint-based approach to phonology. However, no such research has been conducted on morphological and phonological nativization of English nouns borrowed into EkeGusii. This study sheds light on the applicability of Optimality Theory in accounting for the phenomenon of loan words nativization in EkeGusii, besides providing information to linguists who have interest of conducting constraint based analyses on not only related languages, but others as well.

Many phonological and morphological theories would have been employed in such a study.

However, Optimality Theory, a constraint based approach is used. This theory is apt in this study because its use of output (markedness) constraints such as *CODA, *COMPLEX C, which could motivate the adaptation processes, even when particular processes themselves have no precedents in the native phonology (Yip, 1993; Pardis & LaCharite, 1997; Jacobs and

Gussenhoven, 2000 and Broselow, 1999). This is because nativization involves adaptation of foreign processes, some of which, not present in both the source and target language, and therefore could not have equivalent rules if rule based approaches are used. This is one of the strongest justifications for the use of the constraint based approach over rule based approaches in this study.

There are many studies targeting language borrowing generally and loan word nativization in particular because language contact and the resultant lexical borrowing is an everyday activity

(Appel and Muysken, 1987). This study does not only shed light into the phenomenon of nativization, but also adds knowledge to the growing and continuing investigations into the same.

7 Further, this study is significant in that it is expected to shed more light on the phonological and morphological structures of EkeGusii in particular, and Bantu languages in general. This is because as (Owino, 2003) observes, the modification of the new words into EkeGusii makes it possible to observe the nature of EkeGusii linguistic constraints which characterizes the phonological and morphological structures of the language. Hyman (1970) and Owino (2003) suggest that lexical items of a given language do not provide as much opportunity in the study of its grammar as much as new or borrowed items because the new items are necessarily modified to fit into the constraints of the target language. Observations into these adjustments reveal structural features of the target language ( EkeGusii in this case). Another significance of this study is that its findings will shed light on the structure of other

Bantu languages that have similar formal content as EkeGusii; thus, enabling linguists to understand the phonology and morphology of those languages and widen their understanding of nativization and the processes involved.

1.5 Scope and limitation

This study examines English nouns borrowed into EkeGusii and their phonological and morphological adjustments in the process of their accommodation into the phonological and morphological structures of EkeGusii language. An attempt is made to define and rank the constraints determining phonemic, syllabic (phonotactic), phonological processes, and tonemic adaptation of English phonological features entering EkeGusii. Further, borrowed words also undergo morphological changes upon entering the target language. This study focuses on those morphological features that affect the noun class only, including inflections (affixations) for number and person as well as classification of the nouns into various word classes present in the recipient language.

8 Semantic changes also characterize loan words (Bloomfield, 1933). However, this particular aspect of loan word adaptation is not examined. Thus, this study does not focus on semantic adjustments on the borrowed nouns; but instead, the role of semantics in as far as it characterizes morphological nativization is examined. Only phonological and morphological adjustments are studied as they may offer interesting clues to the understanding of the phonology and morphology of borrowed words (Hall and Hamann, 2003). EkeGusii language has had contact with many languages such as Dholuo, Kalenjin, English and

Kiswahili among others. However, this study focuses only on borrowed nouns from English language because as Scotton and Okeju (1972) and Owino (2003) observe, the language of numerically and socio- economically and culturally more dominant peoples are the more likely donors in lexical borrowing while the less prominent groups are more often the borrowers. This is the case in the Kenyan situation, where EkeGusii, although surrounded by other numerically dominant groups like Kalenjin, it does not borrow from them as much as it does from English.

This is because, unlike English, these languages, are less dominant socio-economically and prestige (Owino, 2003). Although according to Owino there are a few first language speakers of

English in Kenya, the language is prestigious and occupies a culturally and socio-economically dominant position and therefore an obvious donor for most Kenyan indigenous languages including EkeGusii. This observation, coupled with the fact that English is an official language in the country, makes English the choice for this study. The study focuss on the processes of phonological and morphological nativization of English nouns in EkeGusii. There are many of such processes such as phoneme change, resyllabification, among others in phonology, and noun formation, affixation among others in morphology.

However, only those phonological and morphological processes that directly affect nativization, like affixation and resyllabification are described and analyzed given the nature of this study.

9 Focusing only on the relevant phonological and morphological processes will give ample time and space to focus on concepts that give insight into nativization processes, for purposes of detail and precision. At the phonological level, the study will focus on the constraints that determine re-syllabification, phonemic change, and phonological processes such as vowel harmony and

Dahl’s law that characterize English nouns in EkeGusii. Morphologically, several processes characterize nativization of lexical items in the borrowing language. Such processes include affixation involving tense, mood and pluralization among others, across the main parts of speech. However, this study does not focus on all these features in all the parts of speech; instead, it is confined to some affixation processes relating to the noun class only because nouns are the most loaned of all the word classes (Hockett, 1958). The processes on focus are: prefixation, augmentation and nominal classification. EkeGusii language has two dialects: EkeMaate and EkeRogoro dialects (Mecha, 2013; Ongarora,

1996; and Bosire, 1993). This study is confined to the EkeRogoro dialect of EkeGusii because it has the majority of speakers of the language. Lexical items, the units of focus in this study, vary widely in terms of domains of life: food and nutrition, house-hold appliances and utensils, clothing, transport/motoring, information technology, education, politics and governance, health, diplomacy, business and trade, legal affairs, agriculture/farming, and security and administration. The study targeted English nouns borrowed into EkeGusii from at least all the identified domains of life, without exception.

10 CHAPTER TWO

LITERATURE REVIEW AND THEORETICAL FRAMEWORK

2.0 Introduction

This study focuses on linguistic borrowing, which, has been defined variously by different language contact scholars. For example, Crystal (1987) sees borrowing as the introduction of words and other linguistic features by one language to another. To Hoffer (2005) borrowing is the process where linguistic items of one language are transferred into another language; whereas

Thomason and Kaufman (2008) view borrowing as the transferring of linguistic materials

(lexical, phonological, morphological, semantic and syntactic) from a donor language to a recipient language as a result of contact between languages. Lexical borrowing therefore deals with the transfer of words across language boundaries (Bynon, 1977). It is these lexical borrowings that are referred to as loan words (Kang, 2011). Peperkamp and Dupoux (2001) identifies two types of loanwords. Historical loanwords; that is those that have entered the borrowing language and are commonly used by monolingual speakers who never hear source forms of the loan words. The other type of loan words is the online adaptation or on-going adaptation. This, according to Peperkamp, eventually becomes historical when fully adopted into the borrowing language. This study focuses on historical loan words, in which monolingual speakers of EkeGusii were the source. Another category of the types of borrowing has been advanced in the literature. Here there are types such as: direct borrowing in which the borrowing language adopts and integrates into its system both the form and meaning of a lexical item from another language (Bynon, 1977; Asher

1994, and Owino, 2003). Bynon (1977) identifies loan translation or calques as another type.

Here the meaning of a foreign lexical item is employed as a model for a native creation instead

11 of being carried over (Owino, 2003). Other types in this category include: semantic extensions, loan shift and loan blend among others (Bynon, 1997; Asher, 1994). A number of reasons have been advanced to explain why languages borrow from one another.

For example, Hockett (1958); Weinreich (1963) and Owino (2003) give the prestige motive as one of this motivations. This is a case where the borrowers emulate the people they admire by borrowing linguistic items from them. Other reasons advanced include: the need to fill motive,

(Langacker, 1968); the need to designate new things (Weinreich, 1963; Taber, 1979; Owino,

2003); and the need to resolve a clash of homonyms (Weinreich, 1963). These reasons apply in this study. According to Danesi (1985), nativization is the entire process of the adaptation of a loan word by a speaker. Danesi observes that loan words are adjusted unconsciously and systematically to the pronunciation and grammatical patterns of the borrowing language. He further points out that loaned words are not accepted in their original shapes, but rather are restructured to conform to the articulatory and grammatical features of the borrowing language. Studies of this nature, according to Owino (2003), are important in as far as understanding of the internal structure (in terms of phonology, morphology, syntax and semantics) of loan words is concerned. Thus, borrowing processes have often been studied with the hope that they may reveal internal structural features of the borrowing language. This is because loan words usually undergo adaptation processes to conform to the structural constraints of the borrowing language phonology. This chapter reviews some of the studies conducted in the area of linguistic borrowing so far with the aim of establishing and filling in existing knowledge gaps. The review contextualizes this study and brings forth better understanding of the debates, controversies and current issues in the area of phonological and morphological nativization. The chapter reviews literature related to the study and the theoretical framework employed. It reviews literature on EkeGusii 12 phonological and morphological structures in order to shed light on the phonology and morphology of the language in order to show how they differ from those English- the loaning language. Literature on lexical borrowing, focusing on nativization; related studies on Bantu and non Bantu languages, conducted within and outside OT are reviewed with the aim of establishing existing gaps that need to be filled. The chapter is subdivided into the following subsections: 2.1; which addresses objective one of the study, 2.2; deals with objective two, 2.3; objective three, and 2.4; which deals with the theoretical framework within which the study was conducted.

2.1 EkeGusii phonological and morphological structures

In this subsection, literature related to objective one of the study is reviewed. The objective aims at describing the phonological and morphological structures of EkeGusii, which sets the basis of the analyses in subsequent subsections. The sub-section is divided into: 2.1.1, phonological structure of EkeGusii; and 2.1.2, morphological structure of EkeGusii.

2.1.1 EkeGusii phonological structure

This subsection reviews literature on the phonological structure of EkeGusii. It focuses on the phoneme structure, the syllable structure, and the toneme structure. Studies conducted in

EkeGusii phonological system so far (Whiteley, 1960; Cammenga, 2002 & Ongarora, 1996) among others, show that the language has seven vowels. This study is in agreement with these studies that EkeGusii has seven vowels, that is /a, e, ε, i, o, ͻ, u/. However, unlike the pioneering studies above, which give mere descriptions of the vowels, this study gives a spectrographic analyses of the vowels using the Praat software (Boersma &Weenink, 2010). These analyses give the exact acoustic nature of the vowels, unlike the early impressionistic analyses of the vowels. For example, the exact height and mouth positions of the vowels as recorded from native speakers are

13 determined. This indeed is a milestone in as far as the study of EkeGusii phonetics and phonology is concerned. EkeGusii is characterized by height harmony (Ongarora, 1996). Height harmony, according to

Ongarora, and in many other studies, is based on the impressionistic vowel triangle (Keith, 1997).

However, in this study, the vowels perceived to be in harmony are spectrographically analyzed and determined using spectrograms. This is based on the fact that vowel height is negatively correlated with formant 1 (F1) frequency; [ High] vowels have low F1, and [low] vowels have high F1

(Keith,1997:105). Phonotactics of a language are the constraints on the sequence or position of phonemes in words in that language. Such constraints are part of every speaker’s phonological knowledge of their language. The constraints operate on units larger than the single segment, or phoneme; that is, the syllable (Yule, 1996). Consonant phonemes generally operate at the margins of syllables either singly or in clusters (Mohamed, 2000) in languages. In EkeGusii, just like in most Bantu languages

(Mutua, 2007), all consonants function as onsets and not codas. Thus, as Wald (1989) and Mutua

(2007) observe, prenasalised series common in Bantu should be treated phonologically as an independent series rather than as a cluster of nasal plus stop. Therefore, EkeGusii [mb], [nd], [ng] and [nt] are independent (single) phonemes and not consonant clusters (Herby, 1986 & Downing,

2005). Pioneering studies in EkeGusii phonology (Whiteley, 1965 & Cammenga, 2002) among others show that EkeGusii has consonant glide sequences or consonant clusters in short. This study however, following observations by Herby (1986), Hayman and Katamba (1999), Order (1999), and, Downing (2005), suggests that EkeGusii, like other Bantu languages, lack consonant clusters.

Rather, what seems like consonant glide sequences are derived secondary consonants. Thus, where for example, there is a consonant glide [CW] sequence in Cammenga (2002), this study sees it as a

14 derived secondary articulation [Cw]. Studies in other Bantu languages support this view. For example, in LuGanda, spoken in Uganda, the first vowel in a word is deleted unless it is high (in which case it becomes a glide [w] or [j] (Katamba, 1993). The same is true of Ronga (Hargus and da Conceicao, 1999), Emai (McCarthy, 2007), and Fahiru (Otterloo, 2011). This observation is important to this study because it explains why borrowed words with consonant clusters from

English are declustered and the extent of declusterization. The observation further emphasizes the fact that EkeGusii, like other Bantu languages is a strict CV language. Thus, any form of consonant clusters is not allowed. A number of consonantal processes affecting EkeGusii language and Bantu languages generally have been described in a number of studies and scholary papers. For example, Uffmann (2013) makes a general observation that voice dissimilation (Dahl’s law) affects a number of Bantu languages. Guthrie (1967) outlines some of the languages affected, in which EkeGusii is included.

This is a process in which consonants in a given word become dissimilar in terms of voice depending on the language in question. Other consonantal processes which affect Bantu languages besides voice dissimilation include: Declusterization in which consonant clusters are broken

(Hyman and Katamba, 1999, Oden and Oder, 1999, McCarthy, 2007, and Otterloo, 2011), defricativization in which a fricative consonant becomes non fricative by losing its continuant characteristics (Cammenga, 2002), and nasal resyllabification in which nasals optionally become syllabic and bear tone (Ferguson, 1963& Hyman, 1985). These consonantal processes are important to this study because they have guided in establishing the extent to which Bantu phonologies in general and EkeGusii phonology in particular affect borrowed consonants; especially those from English. Pioneering studies in EkeGusii phonology (Whiteley, 1965; Ongarora, 1996, and Cammenga,

2002) show that EkeGusii, like most other Bantu languages such as Shona, (Kadenge, 2003);

15 Kiswahili, (Mwita, 2009); and Tonga, (Zivenge, 2009) among others , is a strict CV language. It does not allow consonant clusters. English on the other hand is not a strict CV language because it sometimes allows consonant clusters of up to three and sometimes four consonants as in the words structure /srkat∫ǝ/ and sixths /sikss respectively (Roach, 1983 and Cruttenden. 2011). This observation is important to this study because it allowed observation into how consonant clusters from English, which are not allowed in EkeGusii as in other Bantu languages, are dealt with by

EkeGusii. Tone, according to de Lacy (2007), is the use of pitch in language to distinguish lexical or grammatical meaning. While some languages are characterized by this feature, and therefore described as tone languages; others are not. Thus, in tone languages pitch differences are used to make lexical meaning differences. This is unlike in non-tone languages which are characterized by stress. This is an important observation to this study because EkeGusii, which loans words from

English, is a tonal (Cammenga, 2002; Bickmore, 2007; and Nash, 2011), while English is a stress language (O’Conner, 1967; Roach, 1983; Oden, 2005; Cruttenden, 2011, and Gussenhoven and

Jacobs, 2013). Thus, it has allowed observation into how EkeGusii loan words from English, a stressed language, are handled by EkeGusii, a tone language.

2.1.2 EkeGusii morphological structure

According to Meinhoff (1899), nouns in Bantu languages are morphologically classified into sets of classes. Bantuists have given varied numbers of such classes ranging from 22 in pro-Bantu

(Welmers, 1973), to as few as 12 in Aghem, spoken in Cameroon (Aikhenvald, 2000). The observation that nouns in Bantu are grouped into sets of classes is important to this study because it allows observation of how EkeGusii loan words from English, in which nouns are not grouped into classes, are handled in EkeGusii.

16 Morphosyntactically, Bantu nouns consist of a prefix and a stem, which are normally compulsory.

The prefix, according to Cammenga (2002) contains number and size features. The choice of these prefixes is determined by the semantics of the noun stems with which they occur (Cammenga,

2002; Kayigema, 2010). This is an important observation to this study because it shows how

EkeGusii loan words from English, in which prefixation neither marks size and number features nor determined by noun stems, are dealt with. Elwell (2005) observes that structurally, most noun prefixes in EkeGusii, like in some other Bantu languages, Kinyarwanda, (Kayigema, 2010); and Kuria, (Mwita, 2009), are divided into two parts: a pre-prefix (augment) and a prefix. This observation about the morphological structure of the

Bantu nouns generally and those of EkeGusii specifically is equally important to this study. This is because it has allowed observations into how EkeGusii loaned nouns from English, with different morphological structures are handled by EkeGusii.

2.2 Phonological nativization

This sub-section rviews literature related to objective two of the study which analyses the phonological changes that EkeGusii loan words from English undergo during nativization.

Phonological nativization involves the phonological adjustments that a loan word undergoes in the borrowing language in order to be accommodated in the phonological structure of that language. Many studies have been conducted in the area of phonological nativization. For example, Hock

(1991) discusses phonological nativization of loanwords in general terms without making reference to any specific languages. In the study, Hock discusses the phonological processes that characterize nativization generally. He observes that borrowed words have to change phonologically to suit the phonological characteristics of the receiving language. He emphasizes on the substitution of the various corresponding loaner language phonemes for those of the 17 receiving language. However, Hock’s study, unlike the present one, is not grounded on any theory. It also deals with phonological nativization only, while the present one deals with morphology as well. These are the departing points between the two studies. Hock’s study will give insight in as far as possible phoneme changes are concerned in this study.

Hall and Hamann (2003) discuss phonological loan word nativization in German in which they focus on changes that affect stressed vowels in the nativization of loanwords from Italian language into standard German. Their conclusion is that the Italian tense or long vowels such as /i: / and /ͻ: / are laxed or shortened in the process of nativization; especially when they occur between voiced consonants. This study gives a lot of insight into the current one because, firstly, it deals with phonemic change in the nativization processes, a process that affects the current study; and secondly, both studies are anchored on a theory- Optimality Theory. However, whereas Hall’s & Hamann’s study deal with vowel phonemes only, the present study deals with consonants as well. Hall & Hamann (2003) discuss phonological nativization only, while the present one deals with morphological nativization as well.

Miao (2005) studies Phonological loan word adaptation in Mandarin Chinese within the perspective of OT. He focuses on phoneme substitution patterns for consonants and re- syllabification processes in Mandarin Chinese loans from three languages- English, German, and

Italian. Miao’s study is similar to this study in that both studies deal with phoneme change and re-syllabification processes. Both studies are anchored on OT. However, the studies differ in a number of ways. For example, the target languages are different; in Miao’s it is Mandarin

Chinese, while it is EkeGusii in the current study. Another point of departure is that Miao considers only consonant phonemes, while this study considers vowels as well. Finally, while

18 this study investigates the possibility of the influence of Dahl’s Law in phonemic change during nativization, Mia’s does not.

Uffmann (1993) discusses vowel epenthesis in Shona loanwords from English. In his analyses, he shows that vowels are epenthesized on an acceptable consonant sequences (consonant clusters). According to Uffmann, Languages with strict CV syllable structure often epenthesize vowels in positions where they serve to break up consonant clusters or re-syllabifying coda consonants as shown in (2) adapted from Uffimann.

2) Vowel epenthesis in English nouns in Shona

a) Coronal + / i / [bazi] ‘bus’

b) Labial +/ u / [temu] ‘term’

In these nouns, vowels have been epenthesized ([i] in ‘buzi’ and [u] in ‘temu’) in order to avoid codas which are allowed in English but not in Shona because Shona, unlike English is a CV language. Shona like, EkeGusii (which is the focus of the current study), is a Bantu language.

Thus, the current study gets insight from the Shona study since both studies deal with phonology.

The departing points between these studies include: while Uffimann (1993) deals with vowel epenthesis only, the current study deals with other processes as well: phonemic change, re- syllabification, among others. Finally, the current study, unlike that of Uffimann which deals with phonology only, deals with morphological nativization as well.

Kim (2008) gives a phonological highlight of Huave language loans from Spanish. Among the conclusions he draws are that Spanish, the loaning language of Huave, has an open syllable structure, while Huave has a closed syllable structure. This prompts apocope (i.e. final vowel 19 deletion to Spanish loans in Huave. The exact opposite characterizes English loans in EkeGusii.

This is because English the loaning language of EkeGusii is sometimes a closed syllable system, while EkeGusii is a purely open syllable system. Besides apocope, this study describes other phonological processes. Syllable deletion, which according to Kim is a situation where an entire final syllable is deleted especially in words where Spanish stress as antepenultimate; palatalization- this takes place in the environment of tautosyllabic front vowels. Other processes include: cluster resolution, fortition (where intervocalic voiced fricatives become voiceless stops) vowel leveling (a case where one vowel overrides and changes the quality of a neighbouring vowel, which becomes a copy of the trigger vowel -vowel harmony).

Discussing morphological loan word nativization, Kim (2008) observes that generally, nouns are borrowed as bare roots from Spanish into Huave, with only phonological modifications. Kim gives more attention to verb morphological nativization.

Kim’s study gives the present study a lot of insight, even though his major focus is not nativization per-se. Phonological processes studied in both studies are the same to a large extent.

However, while Kim focuses on verb morphological adaptation, this study focuses on noun morphological nativization as well.

The major departing points between these two studies is that Kim’s study unlike the present one is not anchored on any theory. Kim merely describes the various phonological and morphological processes without any theoretical grounding. Descriptions and analyses in the present study are based on Optimality theory.

Zivenge (2009) studies phonological nativization of Tonga language. Like in the current study,

Zivenge’s study is anchored on theories unlike many others reviewed already. However, whereas

20 at the phonological level, Zivenge grounds his study on the rule based CV phonology theory, the present study is grounded on the constraint based Optimality Theory. Another difference between the two studies is based on phoneme change. The present study, unlike that of Zivenge, investigates the influence of the dissimilatory process (Dahl’s Law) in EkeGusii loan word nativization process. This is because, as Ellwel (2005) observes, EkeGusii unlike Tonga, is among the few Bantu languages characterized by this dissimilatory process. Finally, while

Zivange (2009) deals with Tonga, a Zimbabwean Bantu language, this study deals with

EkeGusii, a Bantu language spoken in Kenya.

Another study that benefited this study immensely is that of Mwihaki (1998). This study deals with Gikuyu loaned words from English and identifies three aspects of loaned word adaptation: phonemic, phonotactic and prosodic. The current study focuses on phonemic and phonotactic changes that characterize EkeGusii loaned words from English. Like Mwihaki’s, this study considers the changes that loaned words undergo at the syllable level. However, unlike

Mwihaki’s, the current study does not assign prosodic features to the loans. The major departing point between these two studies however is in their theoretical orientation. While Mwihaki’s is grounded on Auto Segmental and Metrical Phonology, this study employs the constraint based

Optimality Theory in its presentation and analysis of data. Again, while Mwihaki studies only phonological adaptation, the present study deals with morphological adaptation as well.

Mberia (2004) discusses the phonological behavior of borrowed words in Kitharaka. He focuses on Kiswahili and English loans. This study differs from the current one in the following ways: while Mberia merely discusses the phonological processes involved during nativization without any theoretical grounding the present study is anchored on a theoretical framework. Mberia’s study, however, gives a lot of insight into this study because both Kitharaka and EkeGusii are 21 Bantu languages (Guthrie, 1967-71); thus, the phonological and morphological behavior of the

English loans into both languages show similar trend. Besides, Mberia shows that Kitharaka, like

EkeGusii is affected by Dahl’s law and therefore, the study is instrumental in understanding the operation of Dahl’s law in Bantu languages. Owino (2003) deals with phonological nativization of Dholuo loanwords. This study, in as much as it gives insight into the general topic of nativization, differs from the current study in significant ways. Firstly, the two studies deal with languages from different groups: Dholuo is a

Nilotic language while Ekegusii is a Bantu one (Guthrie, 1967-71). This means that the English loans might have quite different characteristics in Dholuo as compared to EkeGusii since the phonotactics of the two target languages differ quite significantly. Secondly, the current study unlike Owino’s deal with two linguistic areas – morphology and phonology. Finally, while the proposed study will be anchored on the constraint based OT paradigm, Owino’s is anchored on

Auto segmental and CV – Phonology theories. Mutua (2007) analyzes Kikamba nativized loanwords in which he analyses Kikamba English loanwords. This study is insightful to the present study in that both studies deal with Bantu languages and are both anchored on OT. However, the two studies differ in that whereas Mutua’s study deals with only phonological nativization, the present study deals with morphological nativization as well. Secondly, the present study studies voice dissimilation in EkeGusii loanwords and the role of noun classes in the nativization process, Mutua’s does not. The present study, like those ones of Zivenge (2009), Lodhi (2000), Chimhundu (2002), Mwita

(2009) among others, deals with a Bantu language. Zivenge studies Tonga language, Lodhi and

Mwita, Kiswahili while Chimhundu researches on Shona language. This means that the characteristics of English loans in these languages are insightful in predicting the characteristics of EkeGusii loans from English. This is because the languages in the given studies, like EkeGusii

22 are Bantu and therefore have a lot in common in terms of linguistic properties (Shillington,

1995). But this does not mean that the languages are similar since they have what are referred to as unique values (Bloomfield, 1933). Bickmore (1997 and 1999) deal with EkeGusii verb tone within Optimality Theory, while

Cammenga (2002) describes EkeGusii verb tone without using any theoretical grounding. Nash

(2013) analyses EkeGusii noun and verb tonology. Like Cammenga, Nash’s analyses are not anchored on any theory. The present study, like that of Bickmore is anchored on OT. But while

Bickmore’s analyses the verb, this study analyses the noun.

2.3 Morphological nativization

This subsection reviews literature related to objective three of the study which analyses the morphological adjustments that English loan words into EkeGusii undergo in order for them to be accommodated into the morphological structure of the language. Morphological nativization involves the morphological changes that borrowed words undergo in order to be accommodated in the morphological structure of the target language. Islam (2011) studies the morphology of

Urdu loanwords from English, Arabic and Persian. He concludes that the affixation of English loans into Urdu, whether inflectional or derivational tend to be on native Urdu patterns. Further, that English loans are adaptable and that their integration takes place with native Urdu affixes, which are productive and conveniently attach to affixes. The study established that both inflectional and derivational changes are involved in the loaning process. This study is similar to the present one in a number of ways: they both deal with morphological processes affecting loaned words. Like in Islam’s study, where affixation tends to be on Urdu patterns, affixation in this study also follows EkeGusii patterns. However, the two studies differ as follows. While the current study is anchored on a theory (OT), Islam’s study is just descriptive; it is not anchored in

23 any theory. Secondly, while in Islam’s study there are both derivational and inflectional affixes in the nominal class nouns; affixations in the current study is limited to only inflectional affixes.

Thus, derivational affixes do not characterize the loaned words in the current study. This is because unlike Urdu, which allows suffixation, EkeGusii does not. Thirdly, while the prefix in

EkeGusii is characterized with a pre- prefix, the same is not a feature in Urdu, because pre- prefixing is a feature present in a few Bantu languages. Zivenge (2009) discusses the morphological features which characterize Tonga loans from

English in the perspective of Lexical Phonology and Morphology Theory. The current study on the other hand studies morphological features that characterize EkeGusii loans within the perspective of OT. In nominal nativization, Zivenge focuses on the class marking prefix which is characteristic of most Bantu Languages (Elwell 2005, Bresnan and Mchombo, 1987). This gives insight into the present study since, EkeGusii like Tonga, is Bantu and is characterized by nominal class prefixes. But the two studies differ in a significant way concerning nominal class prefixing. EkeGusii noun class prefixing is characterized by an augment or pre-prefix unlike

Tonga. Thus, EkeGusii nominal loans from English are not only prefixed, but also pre-prefixed as well to mark class, number and size. This is another departing point between the two studies. Kayigema (2010) analyzed how French and English loaned words into KiNyarwanda are allocated to what he calls key areas of influence and nominal class systems of KiNyarwanda.

The study established that bilinguals play and important role in importing of foreign words to the target language, that loaned words from languages closer to the language system of the borrowing language adopt more easily than those from more distant languages. The study confirmed that some of the key areas into which loaned words are allocated include: commerce, technology, transport, agriculture, among others. It further established the most affected parts of speech during borrowing are nouns and verbs because at least all languages of the world have

24 them; besides, they express more concrete realities as compared to the other part of speech.

These findings are beneficial to the present study because they help in the focusing of the areas of influence identified. Regarding nominal class allocation of loaned words, Kayigema (2010) observes that all the foreign nouns entering KiNyarwanda, enter into the nominal classes of KiNyarwanda, thereby adapting its morphological system. For example, nouns within the semantic field of [+ human ,

+singular] enters classes 1 and 2 as in u-mu-avoka and a-ba-voka for French avocat (lawyer) and avocats (lawyers) respectively. This observation is beneficial to the present study because,

EkeGusii, like KiNyarwanda, is a Bantu language characterized by nominal classes. Thus, The present study gets insight from Kayigema’s in as far as nominal classification of loaned words is concerned because the nominal classifications in the two languages are likely to share some characteristics. In as much as Kayigema (2010) benefits this study, the two studies are different in a number of ways. Firstly, the languages under focus (KiNyarwanda and EkeGusii) are different, though from the same language family (Bantu): Secondly, Kayigema’s, unlike the present study, focuses on the areas of loaned words allocation; that is, the areas into which the borrowed loans enter, and the nominal classes to which they are borrowed. The focus of the present study however, is in the morphological changes that the loaned words undergo besides allocations into the various nominal classes, one such change being nominal classification which does not characterize

English noun morphology. Finally, the present study, unlike Kayigema’s analyzes data using a theory; Optimality Theory. In other words, the morphological changes realized in this study are analyzed and explained theoretically, unlike in Kayigema’s.

25 2.4 Theoretical framework

This study is anchored on concepts and ideas drawn from Optimality Theory (OT) as expounded by Prince & Smolensky (1993/2004) and McCarthy (2006) in its analysis of data. McCarthy

(2006) observes that OT is a general approach to modeling human linguistic knowledge. The central argument of this theory that benefits this study is that surface or output forms of language reflect resolutions of conflict between competing constraints; that a surface form of language is optimal if it incurs the least serious violations of a set of constraints, taking into account the different hierarchical ranking of constraints by languages (Prince and Smolensky, 1993).

2.4.1 Tenets of Optimality Theory

The Optimality Theory is different from the rule based generative theories of phonology in a number of ways (Prince and Smolensky, 1993/2004). For example, OT and the Principles and

Parameters Theory (Chomsky, 1981) differ in in that while the two theories view grammatical principles as universal, they elaborate the principles differently: Parametric theory sees the principles as a set of inviolable constraints, while OT sees them as a set of hierarchically ranked and violable constraints. Thus, while language typologies are obtainable through parameter setting (switching on/off of a constraint) in Principles and Parameters Theory, the same is achievable through re-ranking of violable constraints in OT (Kager, 1999). The present study sees languages as differing in this sense; that is in the re-ranking of universal constraints.

OT dictates that an optimal output form is selected from a set of candidates based on a (re)- ranking of violable well formedness constraints; the candidate that minimally/least violates the constraints in the given ranking (which is language particular) is selected as the optimal candidate and thus appears as the surface form. These candidates are evaluated in parallel instead

26 of subject to a series of ordered rules – as in rule based theories. Additionally, the set of constraints in OT is proposed to be universal, and that the grammars of languages theoretically differ in the ranking order of the constraints. For instance, a highly ranked constraint in one language (for example *CODA in EkeGusii) may be lowly ranked in another language (such as

English). According to the theory, a violation of a highly ranked candidate is fatal, which means that such a candidate will never be optimal. The opposite is true.

According to this theory there are two main types of constraints: Constraints on the form of the output structure (the well formedness constraints) on segments, and segment organization. These being constraints grounded in universal markedness principles such as syllables must have onsets and constraints on the relationship between the input and the output aimed at the preservation of information (maintaining faithfulness of the output to the input) (Kager,1999). Kager observes that these two constraints are inherently in conflict.

This theory has three key components. Generator (GEN), the component which takes an input and generates a list of possible outputs called candidates (possible realizations of an input which are potentially infinite in number).

Constraint (CON), is another component. This provides the criteria in the form of strictly ordered violable constraints used to decide between candidates. These constraints are assumed to be universal. Universal in the sense that they affect all languages, though each language ranks them differently (which is one of the reasons behind language differences).

The third component is Evaluator (EVAL). This is the component that chooses/selects, depending on the grammar (language in question), the optimal candidate. Each candidate is evaluated by all constraints at once in parallel rather than in a serial fashion of the derivational

27 generative frameworks. The candidate (output) that violates the fewest high ranked constraints is chosen as the optimal by the grammar. Evaluation takes place by a set of hierarchically ranked constraints in the form (C1 >> C2 >> …Cn), each of which may eliminate some candidate output, until a point is reached at which only one output candidate survives. This elimination process is represented schematically in figure (1).

candidates constraints

a C >> C >> C 1 2 n b Input GEN c

c

e Output c… Figure (1): Process of candidate elimination in OT

Source: Kager (1999)

According to McCarthy (2006), OT has had significant impact on various fields of linguistics including phonology and morphology. This study employs, in its presentation and analysis of data, the premises of the theory that are most directly applicable to phonology and morphology.

Some of the constraints from the universal set, reproduced in Table (1) , were used to present and analyze phonological data generated in the study.

Table (1): Some OT Constraints from the universal set

CONSTARINT INTERPRETATION *COMPLEX (C) No complex margins/No consonant clusters 28 *COMPLEX (V) No strings of vowels/No complex vowels FAITH C The consonants in the input are the same as the consonants in the output FAITH V The vowels in the input are the same as the vowels in the output IDENT-IO (F) The specification for the features of an input segment must be preserved in its output correspondent INDENT-IO (place) The specification for place of articulation of an input segment must be

preserved in its output correspondent *VOI Voiced obstuents are not allowed

*VTV Voiceless obstruents are banned in intervocalic positions

MAX- IO Input segments must have output correspondents (no deletion) MAX- IO V Input vowels must have output correspondents(no deletion) *CODA Syllables are open ( must not have codas) ONSET Syllables must have onsets DEP No epenthesis or addition of either a vowel or consonant

*OCP(dorstop) A markedness constraint which bans sequence of voices dorsal stop

*OCP(V) A markedness constraint which bans sequence of voiced consonants

*COPLEX(C) A constraint which bans complex consonants

*COPLEX(V) A constraint which bans a string of vowels

*BACK V Back vowels are not allowed

*CENTRAL V Central vowel are not allowed

*LAX V Lax vowels are not allowed

*HIGH V High vowels are not allowed

AGREE V Vowels in a word should agree

AGREE C Consonants in a word should agree

*STRIDENT (FRIC) Strident fricates are not allowed

*INTERD (FRIC) Interdental fricatives are not allowed

*LATERAL (C) Lateral consonants are not allowed

29 *TRILL (C) Trill consonants are not allowed

*VOICEOBS Voiced obstruents are not allowed

*NONANTERCORFRIC Non-anterior coronal fricatives are not allowed

*ASSIM (F) Assimilation of features are not allowed in a given domain

*UNIFORMITY Prohibits feature fusing

*FLOAT Markedness constraint which prohibits floating of tones

SPECIFY A constraint which demands that a tone bearing unit must have a

correspondent tone

Source: Archangeli (1997); Kager (1999); Prince & Smolensky (2004); McCarthy (2007/2008)

Morphologically, Optmality Theory provides insight into various morphological phenomena, including affixation, reduplication and allomorphy (McCarthy, 2006). However, this study employs the tenets of OT that are most directly applicable to the morphology of loanword nativization: affixation alignment constraints, constraint ranking and violability, competition among candidate outputs, faithfulness and parallelism of evaluation. McCarthy (2006:38) observes, “constraint violability is pervasive in applications of OT, but there are two areas of morphology in which it assumes particular importance: affix location and Template morphology.” These will be the areas of focus in this study.

According to McCarthy and Prince (1993), affix alignment constraints demand that the edge of two constraints coincide. In particular, a constraint requiring that the left edge of an affix align with the right edge of a word (ALIGN (Affix, L; Word, L)) has the effect of declaring an affix a suffix, while a constraint requiring that the right edge of an affix align with the left edge of a root

(ALIGN (Affix, Root; L)) will have the effect of declaring this affix to be a prefix. McCarthy

30 and prince further observe that constraints on affixal alignment have also been applied to clitic

and affix order restrictions, for example, align the right edge of an affix to the left edge for a

word (ALIGN (Afx, R; Root, L)). This study analyzes affix location of EkeGusii loan words

from English, vis-à-vis the stated alignment constraints.

The input, the supposed underlying form of a grammar, plays a crucial role in this theory.

According to Prince and Smolensky (1993) the input has two main functions: to determine the

output candidates which compete for optimality, and to be referred to by faithfulness constraints

that prohibit output candidates from deviating from specifications in the input.

This study utilizes the correspondence framework of Optimality Theory by McCarthy and Prince

(1995), McCarthy (2008). This framework provides that both input and output consist entirely of

overt non-abstract phonological material. It gives a relation between the input segments and the

output segments, that is correspondence (input- output correspondence). This framework rejects

abstract outputs and strengthens the notion input – bringing on board input optimization

arguments of Prince and Smolensky (1993). Input or lexicon optimization framework provides

that an output is faithful to an input. This observation is demonstrated by the change of Old

English /sk/ to modern English /∫/ as in scip [skip] → ship [∫ip] (de Gruyler 2002). In OT, and

particularly in the input optimization approach, this change means that the input as well as the

output are the same ([∫ip]). Thus, the faithfulness constraints, such as MAX IO is obeyed at the

expense of the markedness constraint*COMPLEX C; (MAX IO >> *COMPLEX C), as analyzed

in tableau (1). input: /∫ip/ MAX IO *COMPLEX C a)  ∫ip

31 b) skip *!

Tableau (2.1): Modern English realization of the input /∫ip/.

This tableau shows that the input has been optimized; that is, it has been realized without any

change and therefore is faithful to the output. It is therefore an input as well as an output. This is

how this study treats EkeGusii loaned words from English.

2.4.2 Optimality Theory versus other Generative Approaches

Many theories could be employed in such a study, generative and non-generative. However, this

study employs the constraint based generative phonology and morphology approach, Optimality

Theory (Prince and Smolensky, 1993 and McCarthy, 2006), as compared to rule based

generative theories.

Rule based generative phonology theories, according to Clark, Yallop and Fletcher (2007)

embodies the derivational approach, whereby the output or surface form is the result of a series

of rules that operate on an underlying form of a word or morpheme generating at each stage of

the derivation a specific output which is in turn operated on by the following rules in the

derivation process. In constraint based approaches, such as Optimality Theory, on the other hand

a surface form is realized not through rule application, but by violating the least of a set of

language specific hierarchically ranked constraints which are violable.

Optimality Theory is apt in this study because its use of output (markedness) constraints such as

*CODA, *COMPLEX C, could motivate the adaptation process, even when particular processes

themselves have no precedents in the native phonology (Yip, 1993; Pardis & LaCharite, 1997;

Jacobs and Gussenhoven, 2000 and Broselow, 1999).

32 According to Smolensky (1996), the principle of the richness of the base in OT naturally allows for, and perhaps even requires, an analysis of novel input forms, which are not attested in native learning data. This makes the theory more amenable to the study of loaned word adaptation phenomenon because all the loaned words from English (which constitute inputs) into EkeGusii are novel.

The fact that OT allows for the formalization of tendencies can be seen as beneficial over rule based theories because phonologists have for long argued that tendencies (such as the historical tendency towards consonantal lenition) or for stress to fall on heavy syllables.

Again, generative phonology of the 1970s and 1980s had increasingly developed a mixed model which used both rules and constraints. OT, unlike these generative phonological theories enables phonological entity (constraints). This is an advantage according to Arbib (2006).

OT’s attempt to account for opacity such as in Sympathy Theory where failed candidates are allowed to influence the successful candidate and Stratal OT, which introduces lexical strata, has rescued much of what was proposed in the model of lexical phonology.

OT, unlike rule based theories predicts the emergency of the unmarked (TETU). Thus, a markedness constraint that is frequently violated in a language may still affect output. According to Arbib (2006) the constraint favoring CiCC over CCiC in the language of Yawelmani, for example, is not surface true (due to the fact that sequences of CCiC nature do not occur because high ranking faithfulness constraints preserve them, but when *CCC forces a vowel to be inserted, CiCC is preferred over CCiC. A major contribution of OT has been focusing attention on TETU, of which many new cases have been found.

33 Another advantage of OT over rule based generative approaches is its straightforward account of

what McCarthy (2001) calls homogeneity of target/heterogeneity of process. A rule, in rule

based approaches, specifies the structure that it applies to (target), and the operation to be

performed on the structure (process). It has long been observed, however, that rules applying

different processes to the same target tend to occur cross-linguistically and within the same

language. A rule based theory has no explanation as to why a structure should be a recurring

target. In OT, however, the explanation is straightforward; there is a markedness constraint

against the target, but whether and how the target is repaired depends on interaction with other

constraints.

McCarthy (2007) observes that the elements of a ranking argument are illustrated with a tableau

and that tableaux of two main types appear in the literature. Each type has its usefulness for

certain purposes. When the goal is to argue for ranking of constraints in a language, then the

comparative tableau format of prince (2002) is used. In this tableau, each cell (row, column)

indicates the number of violations, if any, of the constraint column incurred by candidate row as

shown in tableau (2) of the Yawelmani input /Ɂilk-hin/.

input: /Ɂilk-hin/ *Cu DEP

☞Ɂilik-hin 1

Ɂilk-hin W1 L Tableau (2.2): Comparative tableau for Yawelmani input /Ɂilk-hin/.

Source: McCarthy (2007:7)

In this kind of tableau, every cell in a loser row has symbols W and L showing whether the

constraint favors the winner or the loser, or no symbol if it favors neither. For example, in

tableau (2), the constraint *Cu, which bans syllabified consonants, favors the winner, because the

loser violates this constraint once, while the loser violates it not at all. DEP, which ban

34 epenthesis of segments in an input, favors the loser, because the winner violates this constraint

and the loser does not. The W and L annotations indicate the function of the constraint in the

system (McCarthy, 2007).

The other type of tableau in the literature as observed by McCarthy (2007) is the violation

tableau of Prince and Smolensyky (2004), which is used when the goal is to show which

members of a given set of candidates are possible winners under different rankings of a given set

of constraints. It allows for the observation of the difference in possible winners depending on

the ranking of the given set of constraints. Tableau (3) shows how a violation tableau handles the

above Yawelmani input.

input: /Ɂilk-hin/ *Cu DEP

☞Ɂilik-hin * Ɂilk-hin *! Tableau (2.3): Violation tableau for Yawelmani input /Ɂilk-hin/.

In a violation tableau, each violation of a constraint is indicated by an asterisk. When a constraint

knocks a candidate out of competition, the result is indicated by an exclamation mark. Cells are

shaded when any violations that they may contain can have no effect on the outcome because

higher ranking constraint are decisive.

This research employs the violation tableaux because the goal of the study is to show how

different candidate sets from EkeGusii and English are possible outputs under different rankings

of a given set of universal constraints. Thus, showing that languages differ in terms of how they

rank universal constraints and not in terms of rules, some of which are language specific.

OT, like all other approaches to studies of this nature is not sort of weaknesses. one challenges of

this approach is determining the range of candidates to be considered in an analysis given that 35 GEN has the potential of generating an infinite set of candidates (McCarty and Prince, 1993;

McCarthy, 2007). This is because as McCarthy observes, it is theoretically disastrous to overlook a candidate that ties or beats the intended winner since the overlooked candidate has the potential of undermining the entire analytical edifice.

Equally challenging in this theory is to determine the most appropriate set of constraints and their ranking to be used in the analysis of a given set of candidates, given that constraints are universal, meaning that they affect all languages; the difference being their different ranking in different languages.

36 CHAPTER THREE

RESEARCH METHODOLOGY

3.0 Introduction

This chapter describes the methods that were applied in carrying out this research. It is organized under the following sub sections: research design, research site, study population, sample size and sampling techniques, data collection procedures, data analysis and ethical considerations.

3.1 Research design

This study adopted both descriptive and explanatory research designs. A descriptive design attempts to show how the phenomenon under investigation is like. Mugenda & Mugenda (1999) observe that descriptive research design determines and reports the way things are. It attempts to describe such things as possible behaviour, attitudes, values and characteristics. In this study, the design allows generalized descriptions and characterization of the phonologicsl snd morphological structures of of EkeGusii and English languages. These kinds of descriptions and characterizations allude to the analyses that are eventually carried out in subsequent objectives as dictated by the theory in use (OT). This is in response to question one of the study.

Explanatory research design on the other hand identifies the extent and nature of cause and effect relationships. It assesses the impact of specific changes on existing norms and processes. Further, it analyzes situations to explain patterns of rlationships between variables (Zikmund, Babin,

Carr, & Griffin 2012). Explanatory design in this study allows for the assessment of the impacts of EkeGusii phonological and morphological structures on the phonological and morphological structures of English nouns borrowed into EkeGusii in response to questions two and three of the study.

37 3.2 Research site

The targeted research sites were Kisii and Nyamira counties which were purposively selected because this is where most native speakers of EkeGusii (the target language) reside. Through simple random sampling, Nyamira County was selected as the target research site. It is in this county that the accessible site (Nyagaachi Village) was selected for study. This was carried out as follows.

Eighty (80) out of the one hundred and twelve (112) sub-locations within the county with the desired characteristics (not within or near urban centers nor along ethnic boundaries) were purposively selected with the assistance of the County commissioner’s office, Nyamira County.

This was in order to increase the possibility of selecting a sub-location with as many native speakers of EkeGusii as possible with only first language (EkeGusii) competence. As Weinreich

(1953) observes, if a speaker is competent in more than one language, he/she may attempt to reproduce the borrowed morpheme with its original sounds, while the monolingual speaker is more likely to force the loan word to conform to the target language phonetic and phonemic pattern. From the selected sub-locations, Enchoro sub-location was randomly selected. Out of the seventeen (17) villages of the sub-location (see appendix viii), Nyagaachi Village was randomly selected, thus becoming the research site of this study.

3.3 Study populations

There were two types of population in this study; population of the participants in the study, and the population of EkeGusii loan nouns from English. The population of participants constituted all the native speakers of EkeGusii in Kisii and Nyamira Counties as its target participants population, who according to the Kenya National Bureau of statistics (2009) are 2,205,669.

38 However, the accessible population of the study was the number of EkeGusii native speakers in the selected study area in (3.2) above. According to the Kenya Bureau of Statistics (2009), a rural village in these counties has an average of 500 people distributed across ages. This was the target population of the study. It is from this population that a sample was selected to provide data. All the English nouns borrowed into EkeGusii constituted the second type of population.

Available literature indicate that the population of English nouns borrowed into EkeGusii is not known. This study therefore treated all the English nouns borrowed into EkeGusii as its target loan nouns population.

Secondary data was also used in this study. The main source of secondary data included library study in which existing literature (books, theses/dissertations, dictionaries and journals) were reviewed. It is this type of data that was used to describe the phonological and morphological structures of EkeGusii and English languages in response to question one of the study.

3.4 Sample size and sampling techniques

This sub-section addresses the characteristics and the size of the sample of the interviewees and

English the nouns borrowed into EkeGusii, and the sampling techniques that were employed to get the samples. The interviewees were adult native speakers of EkeGusii with knowledge of a given semantic domain by virtue of their training or experience. An adult in the Kenyan context is an over eighteen (18) years old person. But in this study, the preferred age is over sixty following Mecha (2013) observation that such a person has been widely exposed to the use of language in various social contexts and therefore is competent enough to provide the required data. The sample size of interviewees was selected as follows. An interviewee was purposively selected from each of the semantic domains of borrowed nouns identified. Thus, thirteen(13) 39 interviewees were selected, given that there were thirteen (13) of such domains. This was the sample size of the interviewees.

A total of 349 English nouns borrowed into EkeGusii were collected from the interviewees (see appendix (v). All these nouns constituted the sample size of the study. This is because the nouns could not be sampled any further because; first, their number was fairly small; and secondly sampling them could leave out some which could be used to explain certain phonological and morphological processes, while those which could not describe some processes could be sampled. Thus, the nouns were selected purposively to describe and explain a process when and where it occurs.

3.5 Data collection procedures

In addition to native speaker intuition, data in this study was collected from the interviewees through semi- structured interviews (see appendix ii). These interviews were based on an interview guide, a list of questions based on the various domains of life (see appendix iii).

Interviewee responseses/narratives were recorded by a voice recorder and later transcribed using a raw data recording/transcription form (see appendix iv).

3.6 Data analysis

Data analysis, according to Mugenda and Mugenda (1991: 203), is the process of bringing order, structure and meaning to the mass of information collected. It seeks to make general statements on how categories or themes of data are related. Data in this study were in form of texts and were of two types: phonological and morphological. Analyses of the data in this study were carried out within the tenets and principles of the constraint based Optimality Theoretic (OT) framework.

40 3.6.1 Procedure data analysis

English nouns borrowed into EkeGusii were analyzed against EkeGusii and English phonological and morphological constraints rankings in order to account for the various phonological and morphological changes observed phonemically, since constraint ranking between any two languages differ. This , according to the theory is carried as follows: INPUTS are subjected to the GEN component of the theory which generates an infinite set of candidates.

The candidates are then subjected to the EVAL component, which, using the CON component

(ranked on a language specific basis) assesses and selects the most harmonic candidate depending on the grammar in question. The selected candidate becomes the OUTPUT of the grammar. This is illustrated by figure(2):

EVAL

CANDIDATES CONSTRAINTS

INPUT GEN OUTPUT

Figure (2) : process of OT realization of output

Phonological and morphological forms of: English nouns borrowed from EkeGusii, English nouns, and EkeGusii nouns served as inputs to yield outputs. Constraints were ranked on language input basis. All these were aimed at establishing the constraint ranking that the borrowed nouns adapted, that of English or that of EkeGusii.

41 Acoustically, the vowels of the two languages were analyzed using the Praat computer software.

This was in order to determine the acoustic nature and differences between the vowels of the languages, thus establishing the general direction of change.

3.7 Ethical considerations

According to Kumar (1999), “in every discipline it is considered unethical to collect information without the knowledge of the participants in a research, and their expressed willingness and informed consent.” Seeking informed consent is probably the most common method in medical and social research (Bailey, 1978). It against this background the consent of the participants

(appendix ii) and government authorities were sought. This was after describing to them the aims and objectives of the study. This was to ensure that the rights of the participants were guaranteed.

Other ethical issues that were held include: maintaining confidentiality of information by participants, avoiding bias in sampling especially of the participants, use of appropriate methodology, correct reporting and appropriate use of information (Kumar, 1999). A Research Clearance Permit and a Research Authorization Letter were obtained from the

National Council for Science and Technology (appendices x and xi respectively) to ensure that the study was conducted within proper ethical confines as required by law.

42 CHAPTER FOUR

DATA PRESENTATION, ANALYSIS, RESULTS AND DISCUSSIONS

4.0 Introduction

This chapter is organized into three major sub-sections: 4.1 presents a theory-neutral generalized description of the phonological and morphological structures of EkeGusii and English, while 4.2 and 4.3 presents phonological and morphological analyses respectively of the changes involved in the nativization of English nouns borrowed into EkeGusii within the Optimality Theoretical framework. The chapter focuses on the phonological and morphological features of EkeGusii and English lanuages that are affected in the process of nativization.

4.1 Phonological and morphological structures of EkeGusii and English In this sub-section, descriptive generalizations of the phonological and morphological structures of EkeGusii in comparison with those of English are given with the aim of providing the structural differences and similarities which ultimately engender phonological and morphological nativization in EkeGusii as analyzed in sub-sections 4.2 and 4.3 respectively. The descriptions allude to the tenets of Optimality Theory which provides that a descriptive generalization is the essential intermediate step between data and analysis and that good descriptive generalizations are accurate characterizations of the systematic patterns that can be observed in the data. Therefore, according to the theory, proceeding straight from the data to the analysis, without taking time to formulate an accurate descriptive generalization , is never a good idea. The descriptive generalization mediates between the data and the analysis; it is what the analysis is an analysis of (McCarthy 2008:34). Data described in this sub-section is secondary as gathered from existing literature, including published books, dictionaries, theses and journals.

43 4.1.1 EkeGusii and English vowel systems Available literature indicate that few studies have been conducted in EkeGusii language, especially in the area of phonology. Whitley (1960), which is among the pioneering studies in the language, lists seven basic vowels, which are described by Cammenga (2002:39) as repeated in chart (1). i e ɛ a ͻ o u High + + - - - + + Mid - + + - + + - Back - - - - + + + ATR + - - + - - + Chart (1): EkeGusii vowels Other studies carried out in the language (Osinde, 1986; Ongarora, 1996; 2009 & Bosire, 1993 among others), have also confirmed that EkeGusii has seven relatively pure vowels as described in chart (1) above. This is further supported by a survey carried out by the University of

California in 1984 (Los Angles Phonological Segment Inventory Database) which found out that most Bantu languages surveyed have between five and seven vowels (Odlin, 2000). EkeGusii falls within the seven vowel system. The survey above places a seven vowel language system in the cardinal vowel diagram, designed by Jones (1956), as illustrated by chart (2).

Front Back High i u Mid high e o Mid low ε ͻ

Low a

Chart (2): EkeGusii vowel trapezium According to Johnson (1997), the short vowels in chart (2) may occur both as short and long depending on the environments they find themselves in as illustrated by (3) below. (3) Short and long EkeGusii vowels Vowel Examples of words Gloss i siba /sia/ tie 44 ii siiba /siia/ sip e embori /embori/ goat ee embeera /embeera/ grave εε orobeere /ͻrͻεεrε/ tit aa abaana /aaana/ children ͻ omoeto /ͻmͻεtͻ/ trap ͻͻ omoonia /ͻmͻͻnia/ seller u ekeguuru /ekeγuuru small pot uu ebiguuru /eiγuuru/ small pots (3) shows that every short vowel has its long counterpart, making the total number of these relatively pure EkeGusii vowels fourteen, and not seven as has previously been described. This is indeed the position taken by this study. According to Johnson (1997) and Mihalicek and Wilson (2011), in speech, the resonant frequencies of the vocal tract or the frequencies that resonate the loudest are referred to as formants. It is these formants that are seen as peaks in a spectrum. In their articulation, vowels produce several formants. However, as Mihalicek and Wilson point out, the first three of the formants, labeled F1, F2, and F3 respectively are the most informative in speech. The values of these formants differ from vowel to vowel, which leads to the distinction that is heard between vowels and other sounds. Spectrograms in figure (2) of the seven EkeGusii vowels were produced by a male adult native speaker of EkeGusii.

45 /a/ EkeGusii sound /a/ /e/sound /e/ / ε/ 5000 5000 5000

4500 4500 4500

4000 4000 4000

3500 3500 3500 ) z ) ) z z H 3000 3000 3000 H ( H ( (

y y y c c c n n n 2500 2500 2500 e e e u u u q q e e q r r 2000 e 2000 2000 F F r F 1500 1500 1500

1000 1000 1000

500 500 u 500 /i/ sounnd /i/ /o/0.594285714 /ͻ/ 5000 5000 5000 0 0 0 0 00.3045 0.071520 0.136 4500 4500 4500 Time (s) Time (s) Time (s) 4000 4000 4000

3500 3500 3500 ) ) ) z z 3000 3000 z 3000 H H H ( ( (

y y y c c c n n 2500 2500 n 2500 e e e u u u q q q e e e r r 2000 2000 r 2000 F F F

1500 1500 1500

1000 1000 1000

500 500 500

0 0 0 0 soundsounnd/u/ /o //i/ 0.07639 50005000 0 01.189 0.136 Time (s) Time (s) Time (s) 45004500

40004000

3500 3500 ) ) z z 30003000 H H ( (

y y c c n n 25002500 e e u u q q e e r r 20002000 F F

15001500

10001000

500500

00 Figure00 (3): Spectrograms of EkeGusii0.08388 0.07639vowels [a, e, ε, i, ͻ, o, u] collected from a native speaker The vowels in figureTimeTime (3)(s) (s) can be listed against their F1, and F2 as in table (2) as follows. Table (2): F1 and F2 frequencies in hertz (Hz) of EkeGusii vowels Vowel F1 F2 /i/ 540 2450 /e/ 730 2250 /ε/ 830 2100 /a/ 900 1850 /ͻ/ 750 1250 /o/ 590 1150 /u/ 520 1100

46 To come up with EkeGusii vowel trapezium as that in chart (2) above, the vowels in table (2) are

plotted by frequencies of their first two formants as in figure (4) below.

HZ F1 400

500

600

700

800

900

Figure1000 (4): Plot of F1 against (F2) formant frequencies in hertz (Hz) of EkeGusii vowels This plot shows that the first formant corresponds inversely to the height dimension:HZ F2 (high

vowels have low F1 and low vowels have a high F1), and the second formant corresponds to the 2500 2000 1500 1000 500 advancement (front/back) dimension: (with front vowels having a high F2 and back vowels

having a low F2).

In comparison to EkeGusii, English has a relatively large number of vowels which, like those of

EkeGusii, are either relatively pure or clearly gliding in nature. (Cruttenden, 2011; Roach, 1983;

O’Connor, 1967), among others, identify the following vowels as adapted in (4).

4) English vowels

a) pure vowels

Vowel Examples of words

/i/ heed, feel bead

/ɪ/ hid fill bid

/e/ head fell, bed 47 /ᴂ/ had bad mad

/ɑ:/ hard bard par

/ɒ/ hod bod

/ͻ/ hoard fall board paw

/ʊ/ hood full

/u:/ would fool booed pooh

/ʌ/ but cut hut

/ɜ:/ heard fur, bird pur

/ǝ/ accept father b) Dipthongs

/eɪ/ fail bayed pay

/aɪ/ hide file bide pie

/ͻɪ/ foil boy

/ǝʊ/ hoe foal

/aʊ/ howe’d foul bowed pow

/ɪǝ/ beard beer

/eǝ/ haired bared pair

/ʊǝ/ poor c) Triphthongs

/aɪǝ/ fire, tyre, choir society buyer,

/aʊǝ/ our, shower, flower, coward, nowadays

/eɪǝ/ player, greyer, layer, conveyor

/ǝʊǝ/ mower, slower,

48 / ͻɪǝ/ employer, enjoyable buoyant, joyous

(4) shows that English vowels, like those of EkeGusii, are grouped into categories based on their quality (Cruttenden, 2011; Gussenhove & Jacobs, 2011; Ladefoged, 2001; Roach, 1983; and

O’Connor, 1967).

Pure vowels remain constant and do not glide (that is move from one vowel to another) during their production. The vowels can either be long or short in nature as can be observed in data (4a) above. Long vowels, marked by one vowel symbol and a length marker of two dots (:), are those which take a relatively longer period to produce, for example /u:/. Short vowels on the other hand are marked by one vowel without any length marker; they take a relatively shorter period to produce (Gussenhoven & Jacobs, 2011; Cruttenden, 2011; and Ladefoged, 2001; Roach, 1983).

The English pure vowels are listed in (5).

5) English pure vowels

Vowel Description

/i:/ long, spread or non- rounded, front, high

/ɪ/ short, non-rounded, front and high vowel

/e/ short, non-rounded, front, close-mid and open-mid

/ᴂ/ short, non-rounded, front, open-mid

/ʌ/ short, non-rounded, central, open

/ɑ:/ long, non-rounded, open, central

/ɒ/ short, rounded, back, open

/ͻ:/ long, rounded, back, mid

/ʊ/ short, rounded, back, mid-close

/u:/ long, rounded, back, close 49 /ɜ:/ long, non-rounded, central, mid

/ǝ/ short, non-rounded, central, neutral

The English vowels are many in number as compared to those of EkeGusii. Thus, there are a number of vowels found in English but not in EkeGusii. Both English and EkeGusii pure vowels are characterized by length. (6) compares the English pure vowels with those of EkeGusii.

(6) Comparison between English pure vowels and EkeGusii vowels English pure vowels EkeGusii vowels

/i/ /i/

/ɪ/ /ii/

/e/ /e/

/ee/

/ᴂ/ /ε/

/εε/

/ɑ:/ /o/

/ɒ/ /oo/

/ͻ/ /ͻͻ/

/ʊ/ /u/

/u:/ /uu/

/ʌ/ -

/ɜ:/ -

/ǝ/ /a/

- /aa/

50 (6) shows that while English has twelve pure vowels, EkeGusii has fourteen. Both EkeGusii and

English vowels have both long and short vowels; EkeGusii length here is presented by doubling of the affected vowels.

The pure vowels in the two languages however are not the same, especially in quality and production. Acoustically therefore, even though these vowels share the same IPA symbols such as [i], [e], [ͻ] and [u], they are different as illustrated by tables (3) and (4) of F1 and F2 of the languages.

Table (3): F1 and F2 frequencies in hertz (Hz) of EkeGusii vowels [i, e, ͻ, u] repeated from

Table (2) above.

Vowel F1 F2

/i/ 540 2450 /e/ 730 2250 /ͻ/ 750 1250 /u/ 520 1100 Table (4): F1 and F2 frequencies in hertz (Hz) of the English vowels [i, e, ͻ, u] taken from spectrograms in figure (5).

Vowel F1 F2

/i/ 280 2250 /e/ 400 1920 /ͻ/ 590 850 /u/ 310 890

51 Figure (5): Spectrograms of 8 British English vowels (Source: Ladefoged and Keith, 2001:175) Tables (3) and (4) show that the frequencies of the first and second formants of the given vowels are different. For example, while the formants of the English vowel /i/ are 280 and 2250 for F1 and F2 respectively; the formants of the same vowel in EkeGusii are 540 and 2450 respectively.

Thus, the acoustic nature of the vowels in these languages are significantly different and therefore are heard differently. As pointed out above, there are gliding vowels in English. These according Ladefoged (2001)

Clark, Yallop and Fletcher (2007), Gussenhoven and Jacobs (2011), and Cruttenden (2011) among others are sequences of vocalic elements which form a glide within one syllable. Those made up of two such elements are called diphthongs as in (8b) while those made up of three are called triphthongs as in 8c) above.

Diphthongs have a first element (the starting point), and a second element (the point in the direction of which the glide is made). According to Roach, the RP diphthongs have as their first element sounds in the general region of [ɪ, e, a, ʊ], in which there are the diphthongs: /ɪǝ, eǝ, aɪ,

52 aʊ, ǝʊ/; and for their second element [ɪ ʊ ǝ] where there are the diphthongs: /ǝɪ, eɪ, ͻɪ/. The following figure, adapted from Roach (1983:19) gives a summary of the English diphthongs.

DIPHTHONG

centring closing

ending in ǝ ending in ɪ ending in ʊ

ɪǝ eǝ ʊǝ eɪ aɪ ͻɪ ǝʊ aʊ

Figure (6): English diphthongs

But, as pointed out above, EkeGusii diphthongal combinations unlike those of English are determined by vowel harmony and not the position of the first element as pointed out above.

English, unlike EkeGusii as pointed out above, has triphthongs. A triphthong is a glide from one vowel to another and then to a third, all produced rapidly and without interruption (Roach, 1983).

Phonologists such as Roach (1983) and Cruttenden (2011) view a triphthong as being composed of a closing diphthong with /ǝ/ added on the end. This means that a triphthongal vowel is composed of three constituent vowels. The five English triphthongs according to Roach (1983) are composed of the five closing diphthongs- /eɪ, aɪ, ͻɪ, aʊ and ǝʊ/- with an added /ǝ/. Thus, there are five triphthongs in English as shown in (4c) above.

Comparatively, EkeGusii, unlike English does not have triphthongs. Vowel gliding in EkeGusii ends at the second consonant.

53 4.1.1.1EkeGusii vowel harmony and disharmony

EkeGusii, like some other languages is characterized by vowel harmony. According to Sasa

(2009), vowel harmony is a phonological occurrence in which vowels in a certain unit (such as a word) agree with a certain other vowel (such as a vowel in the first syllable of a word or a vowel with a certain feature specification). Archangeli and Pulleyblank (2007) observe that a harmony system demands that two or more segments which are not necessarily adjacent must be similar in one way or another in terms of features. The opposite of vowel harmony is vowel disharmony. A number of types of vowel harmony have been identified and discussed. For example, Rhodes

(2010) mentions the following: tongue root harmony, height harmony, palatal harmony, rounding harmony, and labial harmony. EkeGusii language is characterized by vowel height harmony

(Ongarora, 1996). Vowel height harmony, according to Oden (1996), is a characteristic of most Bantu languages

Phonology. Oden observes that while any vowel in these languages can appear in the first root syllable of a word, affixes draw from a more restricted vowel inventory. Typically, affix vowels distinguish only three vowels: [a], and a front/back pair not of the third degree of height: [i u], [ɪ

ʊ], or [e o] but not [ɛ ͻ] depending on the language. The final vowel affix is usually drawn from

[i ~ ɪ] for negation, [ɛ ~ ɪ] for subjunctive, and [a] otherwise. This is true of EkeGusii language as discussed by Rhodes (2010). In discussing height harmony, Rhodes (2010) notes that in

EkeGusii, in addition to [a], high vowels block height harmony as demonstrated by (7). 7) EkeGusii vowel height harmony and disharmony a) tͻ-γɛɛnr-ɛ ‘let us go’ Omo-te ‘tree’ b) ͻ-rɛntir-e ‘he/she has brought’ e-ɳuͻm -ͻ ‘marriage’ ti-to-ko-ɳa-koβa-tɛβ-i ‘we will not be telling them’ EkeGusii has two mid vowel heights: High mid and low mid (Ongarora, 1996; Rhodes, 2010;

Anyona, 2011; & Cammenga, 2012). This is illustrated by chart (2) above. (7) shows that affix

54 mid vowels agree in height with root mid vowels. For example, in the word [tͻ-γɛɛnd-e], the root vowel is the mid vowel [ɛ], while the prefix vowel is the mid vowel [ͻ]. These two are in height harmony. However, as (7b) shows, if a non mid vowel intervenes between an affix mid vowel and the nearest root mid vowel, agreement is blocked. For example, the affix vowel in the word

[e-ɳuͻm-ͻ] is high mid rather than lower mid, which would match the vowel in the root. The height of the first vowel can be attributed to the presence of [u], a high vowel, between the two mid vowels. The vowels [e] and [ͻ] in this word are in vowel height disharmony. (8) below gives more examples of EkeGusii height vowel harmony and disharmony. 8) EkeGusii vowel height harmony and disharmony. i) Vowel Harmony [ͻmͻ-ɛt-ͻ] trap [ͻmͻ-ɳɛn-ɛ] owner [e-ŋgor-o] hole [omo-rem-i] farmer ii) Disharmony |eke-suunt-e| [ eγe-suunt-e] darkness [Omo-ib-i] thief [ama-is-ͻ] eyes [omo-uk-ͻ] blind person Adapted from: Bosire & Machogu (2013) In data (8i) the first syllables in the roots dictate that the prefix be in harmony (height) with it.

For example, in the word [e-ŋgor-o], the root vowel [o] is in harmony with the prefix vowel [e].

In (8ii) however, this is not the case. The first syllable vowels in the roots (which are either

[+HIGH] or [+LOW], dictate that the vowels be in disharmony with those of the prefix. For example, in the word [eγe-suunt-e], the first vowel of the root syllable [u] dictates that the vowel of the prefix be in disharmony with it instead of being in harmony; that is [+MID HIGH]. Three types of harmony have been described in the literature: total harmony, opacity and transparency

(Sasa, 2009). Sasa represents these schematically in (9). 9) a) V1 V2 V3 [ F] [ F] [ F] (total harmony)

55 b) V1 V2 V3 [ F] [F] [ F] (opacity) c) V1 V2 V3 [ F] [ F] [ F] (transparency) (Where F represents any feature of the vowels, such as [ATR], [LOW] and [ROUND], and the

Greek letters  and  the different values of the given features such as [+ ATR, -ROUND,

+HIGH]). In explaining the concept harmony, the terms trigger and target are frequently used (Sasa, 2009).

Sasa observes that the term ‘trigger’ refers to the vowel with which all other vowels agree in certain feature(s), while ‘target’ refers to the vowel(s) which agree(s) with the trigger in a given harmony domain such as a syllable or a word. In vowel harmony therefore, it is targets which harmonize with triggers. In total harmony, represented schematically by (9a), all the vowels in a domain agree with the trigger. This is present in EkeGusii language as exemplified by the word [ͻ-mͻɳɛn-ɛ], ‘owner’, given in (9) above. The trigger vowel [ɛ] in the root spreads the feature [+MID, +HIGH] to both the prefix and the suffix (the final vowel). In opacity harmony (which contains an opaque vowel), the vowel adjacent to the trigger does not agree with the trigger of the harmony. In addition, the final vowel agrees with the trigger of the harmony. This kind of harmony is equally present in EkeGusii as in the word |eke-suunt-e| >

[eγe-suunt-e] ‘darkness’. Here, the opaque vowel, [u] blocks harmony in the word except that between the root and the final vowels. The other two opaque vowels [a] and [i] trigger the same behavior in EkeGusii. Transparency harmony contains a transparent or neutral vowel. Here, the medial vowel does not agree with the trigger and the target; it does not participate in harmony. This is equally present in

56 EkeGusii language as in the word [a-ma-isͻ] ‘eyes’, where the trigger vowel [ͻ], does not agree with the medial vowel [i] or the target [a]. From the foregoing discussion on EkeGusii vowels, it has been observed that EkeGusii has seven vowels which can be classified as: /i, e, o, u/ with advanced tongue root (ATR) and /ɛ,ͻ, a/ with retracted tongue root (RTR). As can be observed from chart (2) above, only the mid vowels have advanced and retracted counterparts as shown in (10). 10) Advanced and retracted tongue root mid vowel pairing. ATR RTR /e/ /ɛ/ /o/ /ͻ/ The low retracted tongue root vowel, /a/ does not have a counterpart just like the high vowels /i, u/. Like in other languages with seven vowels such as Yoruba (Pulleyblank, 1996), only words containing mid vowels show perfect harmony as illustrated by EkeGusii examples given (11). 11) EkeGusii perfect harmony ATR RTR /esese/ esese ‘dog’ /ɛsɛsɛ/ esese ‘cough’ /omoonto/ omoonto ‘person’ /ͻmͻɛtɛ/ ‘trap’ /ookombe/ obokombe ‘hoe’ /ͻmͻxɛrɛ/ ‘Luo’ Adapted from Bosire & Machogu (2013) English is not characterized by this phonological process. According to Shapiro (2015), English lacks vowel harmony as a regular phenomenon.

4.1.2 EkeGusii consonant system in comparison with English In this section, the consonant system of EkeGusii is compared and contrasted with that of

English. Studies on EkeGusii phonology have identified a number of consonants. For example,

Cammenga (2002:53) has identified the consonants in chart (3).

Bilibial Alveolar (Alveolo-) Palatal Velar

Continuant /β/ /s/ /γ/

(Flapped liquid) /r/ 57 Obstruent [b] /t/ /k/

/g/

Affricate /c/

/dŽ/

Nasal /m/ /n/ ñ /ɳ /

Glide /w/ /y/

[Cw] [Cy]

Chart (3): EkeGusii consonants This study makes the following observations about consonant inventory in chart (3) . Firstly, it should be noted that Cammenga’s (2002) inventory of EkeGusii consonants is an improvement of Whiteley’s (1960) inventory. In Whiteley’s (1960) inventory, are the following consonants which Cammenga does not include in chart (3) above: [p], [ny], and [y/(j)]. As observed by both

Cammenga and Whiteley, the voiceless, bilabial stop [p] is only found in EkeGusii words borrowed from languages in which the sound is present, such as Kiswahili and English. It can therefore be concluded that the sound is not found in EkeGusii language, except in “one or two idiophones” as suggested by Whiteley. The idiophone suggested by Whitely would be the emphatic form /pi/ which means ‘completely’ as illustrated by (12). 12) EkeGusii ideophone with the voiceless stop /p/ i) ita pi/ita pi/ ‘kill completely’ ii) geenda pi /γɛɛnda pi/ ‘go completely’ iii) koora pi /koora pi/ ‘finish completely’ Adapted from Bosire and Machogu (2013) This data shows that /pi/ in the words emphasizes the given actions. Secondly, Cammenga (2002) replaces /ny/ with /ñ/ and names /j/ a glide instead of a semi vowel.

This study will use the IPA symbol /ɳ/ to represent the palato-alveolar nasal instead of /ny/ and rename /j/ an approximant instead of a glide. Thirdly, following observations that Bantu languages do not have consonant glide sequences, but instead that the glides (approximants in this study) are realized as secondary articulations 58 (Hayman & Katamba, 1999), what Cammenga includes as consonant glide sequences, ([Cw] and

[Cy]), will not be included in the inventory in this study. The approximant /w/ will be excluded altogether from EkeGusii consonant inventory, meaning that it will only be treated as a derived secondary consonant represented as ([C[w]]). Fourthly, the pre-nasal stops [b], [d] and [g], the voiced alveolar fricative [z] and the voiced palate-alveolar fricative [dʒ], like the secondary approximants described above will be treated as derived consonants through homorganization and defricativization. They are therefore not part of the phonological system of the language. This then means that they are equally treated as secondary derivativations. Fifthly, the affricates that Cammenga (2002) represents with the symbols /c/ and /dŽ/ are, in this study, represented as the IPA symbols /t∫/ and /dʒ/ respectively/.

EkeGusii consonants can now be represented as in chart (4).

Bilabial Alveolar (Alveo-) Palatal Velar

Continuant /ɸ/ /s/ /γ/

// [z] /x/

(tril) /r/

Obstruent [b] /t/ /k/

[d] [g]

Affricate [dʒ]

/t∫/

Nasal /m/ /n/ /ɳ / /ŋ/

Approximant [w] / j/ 59 Chart (4): EkeGusii consonant inventory Chart (4) shows that two new consonants have been added into the consonant inventory of

EkeGusii. These are:/ɸ/; voiceless, bilabial, continuant as in obuba /oɸuɸa/ ‘food’, amaraba

/amaraɸa/ ‘soil’ abasaacha /aɸasaat∫a/ ‘men’ and /x/ voiceless velar continuant as in omogesi

/omoxesi/ harvester, agaanto /axaanto/ ‘a thing’, ensagara /enzaɸara/ ‘lizard’.

Therefore, this study concludes that EkeGusii has fourteen distinctive consonants in its phonological inventory: /ɸ/, /s/, /γ/, //,/x/, /r/, /t/, /k/, /t/ /m/, /n/, /ɳ/, /ŋ/, and / j/; and six

phonetic derivatives: [z], [b], [d], [g], [dʒ] and [w].

Phonologically, voiced EkeGusii consonants seem to occur with the mid-high vowels /e/ and /o/

(with the feature [+ATR]), while the voiceless ones occur with the mid-low vowels /ε/ and /ͻ/

(with the feature [+RTR]). The rest of the vowels occur without such restrictions. This is vowel- consonant harmony controlled y the feature [VOICE]. (13) illustrates this observation. 13) Occurrence of vowels with consonants in EkeGusii /e/ and /o/ (ATR) /ͻ/ and /ɛ/ (RTR) ebando / eβando/ ‘maize’ oboba /ͻͻa/ ‘mashroom’ egesanda /eγesanda/ ‘calabash’ etoigo/ εtͻixͻ/ ‘floods’ emondo /emondo/ ‘gizzard’ omoeto /ͻmͻεtͻ/ ‘trap’ Adapted from Bosire and Machogu (2013) Whether a vowel occurs with a voiced or a voiceless consonant in EkeGusii seems to be determined by whether the vowel is advanced or retracted tongue root. This, in fact, is what is responsible for the consonant and vowel harmonies that are observed in data (13). Words having vowels with ATR demand [+VOICE] consonants, while those with RTR demand [-VOICE] consonants. This is further exemplified by data set (11) above.

As compared to EkeGusii, there are 24 distinctive phonological units which are consonantal both in terms of their position in the syllable, that is phonologically; and also, in the majority of cases, in terms of how they are produced in vocal tract, that is phonetically (Cruttenden, 2011). These

60 consonantal phonemes are classified into two broad categories: Obstruents (those articulations in which there is a total closure or a stricture causing friction). This group is associated with a noise component which accompanies their production. They are further characterized by a distinctive opposition between voiceless and voiced types. The other category of consonants is that of sonorants. These are those consonants characterized by articulations in which there is only a partial closure or an unimpeded oral nasal escape of air. Such articulations are normally voiced, and frequently frictionless; that is, they are without the noise component of the obstruents. This class shares many phonetic characteristics with vowels.

According to Chomsky and Halle (1968), obstruents and sonorants are features that classify segments according to their noise component. Those in whose production, the constriction impeding airflow through the vocal tract is sufficient to cause noise are known as obstruents, while those in which there is no noise component are known as sonorants. The following English consonant classes belong to the obstruent class: bilabial plosives; /b, p/, alveolar plosives; /t, d/, velar plosives; /k, g/, palatal alveolar fricatives; /t∫, dʒ/, labiodental fricatives; /f, v/, dental fricatives; / θ, ð /, alveolar fricatives; /s, z/, alveolar fricatives; /∫, ʒ/, and glottal fricative; /h/.

Sonorants on the other hand, are those sounds in which there is no noisy component in their production. This group has the following classes of consonants: nasals, approximants and vowels. Vowels having been described in the previous section, the rest of these sonorants are described as follows as nasals: bilabial nasal; /m/, alveolar nasal /n/; and the velar nasal /ŋ/, approximants: the lateral approximant; /l/, post alveolar approximant; /r/, unrounded palatal aapproximant /j/; and the labial velar approximant /w/.

61 The description of the English vowels and consonants in this research has relied on Cruntenden

(2011 pp. 91-232). Details and clarification therefore can be verified. The English consonants described so far are presented in a manner and place of articulation chart as in chart (5).

Bilabial Labiodentals Dentals Alveolar Palate- Palatal Velar Glottal alveolar Plosive p b t d k g Fricative f v θ ð s z  ʒ h Affricative t dʒ Nasal m n Ŋ Lateral l Approximant w r j Chart (5): The English consonants Adapted from Jones (1972: xvii) A number of observations about the consonant inventories of English and EkeGusii described in this section can be made. Firstly, EkeGusii consonants are fewer as compared to those of

English; while EkeGusii consonants are fourteen, the English consonants are twenty-four. (14) below gives inventories of the consonants in English and EkeGusii (14) Inventories of English and EkeGusii consonants English consonants EkeGusii consonants /p/ /ɸ/ // /b/ - /f/ - /v/ - /θ/ - /ð/ - /t/ /t/ /d/ - /m/ /m/ /n/ /n/ - /ɳ/ /ŋ/ /ŋ/ /k/ /k/ /g/ - - /γ/ - /x/ /j/ /j/ /w/ - /r/ /r/ /l/ - 62 /dʒ/ - /t∫/ /t∫/ /s/ /s/ /z/ - // - /ʒ/ - /h/ - The second observation is that, while some of the consonants in the inventories are similar or are the same featurewise; others are not. This is clearly captured in (14) which shows that EkeGusii consonants: / ɸ, , ɳ, γ, x/; are absent in English, while the English consonants: /p. b, f, v, θ, ð, d, g, w, dʒ, ∫, ʒ, h/, are absent in EkeGusii. The third observation is that some consonants are shared at least in terms of phonetic features by both phonologies. These consonants are: [t, m, n, ŋ, k, j, r, t∫, s].

4.1.2.1 EkeGusii consonantal processes Like other languages EkeGusii phonology is characterized by consonantal processes. The processes described inthis section are those which affect EkeGusii phonology and therefore the

English loans in the language. they may no affect English phonology.These sub-section describes these processes.

4.1.2.1.1 Voice dissimilation (Dahl’s Law) This process, according to Uffmann (2013), is found in a number of Bantu languages. EkeGusii is one of the languages characterized by the process. Uffman defines Dahl’s law as a voicing dissimilation process in which a prefix stop which is underlyingly voiceless, is voiced if the stem or subsequent prefix starts with a voiceless segment. Guthrie (1967) observes that languages which show the effect of this law are found within his zones E20 - E50 – F20 and G60. EkeGusii language zoned E42, is within this range. It has also been argued that languages vary a great deal as to which consonants undergo the rule, which consonants trigger the rule, and how the rule affects multiple targets within the same word

(Bennett, 1967 and Davy and Nurse, 1982). This means that different languages have different 63 consonants which undergo the process, different consonants acting as triggers in different languages and different effects on targets in different languages (Bickmore, 1998). Bickmore observes that in EkeGusii, there is evidence that Dahl’s law affects the dorsal stop [k] as (15) below demonstrates. (15) The effect of Dahl’s Law on [k] in EkeGusii ’εsa/ ‘harvestﻻ-ͻkͻ- kɛsa| /ͻkͻ| ’oro/ ‘legﻻ-oko-koro| /oko| Source: Bickmore (1998)

This data shows that the voiceless velar obstruent /k/ in the prefixes {ͻkͻ-} and {oko-} respectively are substituted for by the voiced velar obstruent /γ/ in the roots {–γɛs} and {γor} respectively. Thus, the /k/ sound in the initial syllables does not assimilate the sounds in the adjacent syllables as expected in most languages, including English. Instead, it dissimilates as shown. This process is still quite productive in the synchronic phonology of EkeGusii

(Bickmore, 1998) as exemplified by the class 15 prefix {ko-} in (16). 16) Dahl’s Law in EkeGusii Word underlying form surface form gloss a) okoroota |ͻ-kͻ-rͻͻt-a| [ͻ-kͻ-rͻͻt-a] ‘dream’ okogoro |o-ko-kor-o| [o-ko-γor-o] ‘foot’ okonywa |o-ko-ηw-a| [o-ko-ηw-a] ‘drink b) ogokana |o-ko-kana| [o-γo-kana] ‘deny’ ogotuua |o-ko-tuua| [o-γo-tuua] ‘be blunt’ ogoseka |ͻkͻsεka| [ͻ-γͻsεka] ‘laugh’ Adapted from Bickmore (1998) The dissimilation process in (16a) is from the voiceless obstruent stop /k/ to a voiced obstruent fricatives such as /γ/ and the other way round in (16b).

4.1.2.1.2 Prenasalisation and nasal homorganisation Prenasalisation is the process which is responsible for the derivation of prenasalised consonants.

This process, according to Cammenga (2002), causes the nasal part of the prenasalised consonant to become homorganic with the following consonantal element. Thus, both the nasal and the consonantal elements involved in the process share place features of the consonantal element. In

64 other words, hormorganization is the process where the nasal element of the pre-nasalized consonant becomes homorganic (they both share the place features with that consonant). (17) adapted from Cammenga (2002: 87) demonstrates this observation. 17) EkeGusii nasal homorganisation Input: /ɛ-n-/ + /-γͻkͻ/ ‘hen’ Affixation ɛnγͻkͻ Prenasalisation ɛnγͻkͻ Nasal homorganisation ɛŋγͻkͻ (17) shows that the nasal /n/, which is [+alveolar] becomes [ŋ], a [+velar] consonant, a place feature of the consonant /γ/. This is nasal homorganisation. This process affects all nasal elements of all prenasalised consonants in EkeGusii (Cammenga, 2002). Thus |m+| → [mb], while |n+r| → [nd]. Thus, it can be argued that the nasal plus consonant as given here produces a secondary consonant, such as [mb], which as will be observed in section (4.1.1.3) below, is secondary realization, which is treated as a single consonant and not a cluster.

4.1.2.1.3 Declusterization of nasal consonant (NC) and consonant glide (CG) clusters This study argues that there are no consonant clusters of any nature in EkeGusii. Thus, underlying nasal consonant and consonant glide clusters are declusterized in EkeGusii surface forms. This is in agreement with Hyman and Katamba (1999) who observe that Bantu languages do not have consonant clusters. To Hyman and Katamba, the only combinations that seem to be clusters of consonants are those of the nasal consonant (NC), consonant glide (CG) and nasal consonant glide (NCG). This is the position taken in this study; that EkeGusii does not have obvious consonant clusters. What seems to be nasal consonant and nasal glide clusters are in fact secondary articulations motivated by the hormorganization process discussed in the previous sub-section. These nasal consonant and nasal glide secondary realizations are what the study refers to as declusterization.

65 Hyman and Katamba (1999) identify two kinds of consonant clusters that are of significance in the phonology of Bantu: homorganic nasal consonant sequences, also called pre-nasalized consonants, discussed above and consonant glide sequences (CG). These two at times overlap to produce a nasal consonant glide (NCG) cluster as illustrated by (18).

18) EkeGusii nasal consonant glide (NCG)

Underlying form surface form gloss

|n-βu-ate| → [mbwate] ‘hold me’

[m b w a t e]

N C G

Adapted from Katamba (1999)

(18) shows that the underlying nasal |n| is homorganized to [m], which in turn assimilates the consonant, |β|, a continuant to [b], a stop. Further, the underlying vowel |u|, which is high rounded, is assimilated to the glide [w] an equally rounded approximant, by the vowel [a], which is low. This is for ease of articulation (Katamba, 1993). (19) gives further examples of consonant glide sequences yielding hormorganization.

19) EkeGusii consonant glide hormoganization i) Input: /buata/ ‘hold’

Output: [wata]

Hormorganization process

/βu.ata/ → [wata]→ [βwu-ata]

CG

66 ii) Input: sieka ‘close’

Output: sjeka

Hormorganization process

/sieka/ → [sjeka]→ [sjeka]

CG where Cw and Cj are secondary articulations

Adapted from Hyman and Katamba (1999)

There is enough evidence in support of the consonant glide hormoganization argument advanced here as elsewhere. In LuGanda for example when two vowels are adjacent, the first vowel is deleted unless it is high (in which case it becomes a glide [w] or [j] (Katamba,1993), Similarly, in Emai, if two vowels are contained in lexical morphemes following one another, and that the vowel in the first morpheme is high, [i] or [u], the high vowel changes into homorganic glide of the appropriate place of articulation (McCarthy, 2007) as shown in (20) below repeated from

McCarthy (2007:9).

20) Emai consonant glide hormorganization

(i) /ku ame/ → [Kwame] ‘through water.’

(ii) /fi ͻpia/ → [fjͻpia] ‘throw cutlass.’

In (20i), the high vowel /u/ hormoganically changes to the labial consonant glide /w/, while in

(20ii) /i/ changes to the palatal consonant glide /j/. EkeGusii consonant glide hormoganization in

(19) above behaves the same way as the Emai hormoganization in (20). In ‘sieka’ in (19) for

67 example, the high vowel /i/, which is adjacent to the vowel /e/, changes to the homorganic consonant glide [j].

The secondary articulations as in Cw and Cj in (19) above advance the argument that instead of treating a sequence as a consonant glide (CG), it is occasionally treated as a secondary articulation on a single consonant, [wata] and [sjeka] respectively. This means that the consonant here is one (the primary one- underlined, which is accompanied with a secondary one which is a semi vowel- superscripted). Similar arguments have been advanced by Hargus and da Conceicao

(1999) who propose that Ronga language (spoken in Mozamique) has distinctively labialized consonants, for example the nasal consonant [n], in the word [nwala] ‘fingernail’, rejecting a cluster analysis on the grounds that there are no any other onset clusters in the language.

Similarly, Otterloo (2011) treats potential clusters of the type [Pj, Kw] in Pahari language

(spoken in in Northeastern Parkstan) as violating secondary articulated palatalized and labialized consonants [Pj and Kw], respectively.

Following the foregoing observations, this study argues in support of the view that EkeGusii language does not have consonant clusters. Instead, it has secondary realizations in cases of consonant glides as (21) further illustrates.

21) Ekegusii consonant glides homorganization as secondary articulations

Word underlying form surface form Gloss

a) rwana /ru-ana/ [rwana] fight

b) kwani /ku-ani/ [kwani] greet

c) chwei /t∫u-eri/ [chweri] saw

d) etia /e-ti-a/ [etja] pass it

68 e) berja /βeri-a/ [βerja] boil

f) tjana /ti-ana/ tjana swear

Example (21a) can be represented as in figure (7). rwana instead of *rwana. c vcv c cvcv

Figure (7): Articulation of complex glides in EkeGusii

Here, the realization *rwana is treated as ungrammatical because as has been argued before, it allows a cluster of consonants, which is against Bantu phonology which disallows consonant clusters.

In syllabic nodes, the syllables in (figure 7) above will be represented as in figure (8).

a) rwana σ σ σ

C V V C V

rw u a n a

b) *rwana σ σ

cc v c v

rw a n a

Figure (8): EkeGusii consonant glide syllabic nodes

69 Figure (8a) shows that the realization of the given word is grammatical in that it does not have any consonant cluster, while realization (8b) is ungrammatical because it contains a consonant cluster which is disallowed in Bantu phonology.

Herbey (1986) and Downing (2005) pose two questions about nasal consonant sequences in

Bantu phonology. They wonder if the sequence is a single segment or a cluster, and if it is a cluster how the given components are syllabified. These are the same questions that this research sought to answer.

The reason why NC sequences such as [nt] and [nd] are treated as two segments, which is rare, is that they are bi-morphemic, arising by joining of an autonomous nasal (a consonant) with another consonant. For example, in Matumbi language (spoken in Tanzania), the sequence [mb] as in the word [mbajite] “I said”, derives from “nitbajite” which is optionally realized as

[nimbajite], for ease of pronunciation (Herbey, 1986 and Downing, 2005). However, the reason why these NC clusters may not be treated as two segments, especially in Bantu languages (which favours the arguments advanced in this study), is that this would favor languages (such as

EkeGusii) with a typology of uncommon syllable structure such as onset and coda clusters which, violate the sonority sequencing principle (Sievors, 1981 & Clements, 1990).

To avoid treating and calling NC sequences consonant clusters, linguists employ the term

“prenasalised stops” (Hearth, 2003). According to Hearth, Makaa, a Bantu language spoken in

Tanzania for example, has twenty-two simple consonants and eight prenasalized stops. Equally,

Alnet (2009) lists a series of pre-nasalized consonants in Shimaore language.

70 Following the foregoing observations and arguments, this study argues that EkeGusii has pre- nasalized stops and other consonants and therefore no NC clusters in its syllable structure. (22) below gives the four pre-nasalized consonant stops in EkeGusii.

22) EkeGusii prenasalised consonant stops Prenasalised consonant Example of word Gloss a) |n+b|> /mb/ engombe [ɛŋͻmbɛ] ‘cow’ b) |n+r| > /nd/ enda [enda] ‘stomach’ c) |n+t| >/nt/ egento [eγento] ‘thing’ d) |n+k|>/ŋg/ egechanga [eγet∫aŋga]‘wire’

In (22), the NC ‘clusters’ (underlined) are treated as one consonant. In other words, there are no consonant clusters in essence. For example, (22b) can be represented syllabically as in figure (9). enda /e. nda./

V C V

σ σ

V NC V

e nd a

Figure (9): Nasal consonant syllabification in EkeGusii

The first syllable is made up of only the syllable nucleus, which is allowable in this language as in many other Bantu languages. The second syllable, it is argued, is made up of a pre-nasal consonant: a consonant proper (and not two consonants) and a vowel. Thus, it has an onset a prenasalised consonant and not an NC cluster.

71 Clements (1978) observes that there is vowel lengthening before NC clusters in most Bantu languages as illustrated by (23) adapted from Katamba (1989).

23) Ekegusii NC clusters

Word pronunciation gloss

a) omoonto /omoonto/ person

b) ebaando /eβaando/ maize

c) engombe /εŋgͻͻmbɛ/ cow

d) eyaanga /ejaaŋga/ dress

This data shows that the vowel before every nasal consonant is doubled (lengthened). For example, in omoonto in (23a) the vowel /o/ in the prefix {omo-} is doubled so that it becomes the nucleus of the initial syllable of the root {nto}. Clements (1978) observes that such lengthening regularly holds in many Bantu languages, including Yao, Hehe, Sukuma and Kuria spoken in Tanzania and Kikuyu, Luhyia, Kuria spoken in Kenya and many others.

The assumption, according to Clements (1978), is that a pre-consonantal nasal has a special prosodic status that is dominated by a vowel rather than a consonant. This normally results in syllabification of the nasal into the coda of the preceding syllable, but the fact that syllables should not be closed (Prince and Smolensky,1993) is taken to argue against positing nasals in the coda position or having closed syllables. The syllable is therefore syllabified in the onset of the following syllable, which leads to compensatory lengthening of the preceding vowel by re- association of the standard timing unit as illustrated by (24).

24) EkeGusii compensatory lengthening of vowels Input: Omoonto ‘person’ i) *moon.to nasal as Coda

72 ii) moo.nto nasal as Onset iii)*moo.n.to nasal as syllabic consonant

This data shows that it is (24ii) which necessitates compensatory lengthening. This argument depends on the assumption that the nasal in the vowel NC sequence must be in non-linear analysis (Clements, 1986). Here, the pre-nasal consonant lengthening is treated as compensatory lengthening coming from the fact that the nasal is deprived of its vowel slot because it is moved into the onset slot in the word and so a vowel must come in to fill the empty vowel space left by the nasal. This is demonstrated by Figure (10).

1 2 3

CVV CV C CVVC

[mon/to] [mo-/nto] [moo/nto

Figure (10): EkeGusii vowel lengthening

Adapted from Katamba (1989)

Figure (10) indicates that the vowel [o] moves in (in 3) to fill in the gap left behind (in 2) by the nasal [n], which is in the onset position (in 2) having moved from the coda position (a mora) (in

1) leaving behind an empty slot necessitating vowel lengthening. This is presented on a syllable node as figure (11).

*σ σ σ σ

µ µ µ µ µ µ

m o n t o m o n t o 73 Figure (11): EkeGusii compensatory lengthening

Adapted from Katamba (1989)

The phonological evidence in support of the fact that the nasal in NC combination is Onset is that in most languages, most syllables are open; that is, syllables normally end in vowels (Kager,

1999). However, the phonotactics of English as will be discussed in sub-section (4.1.1.3), do allow consonant clusters. It also allows closed syllables. It can therefore be argued that while

EkeGusii does not recognize NC and NCG sequences as consonant clusters, English does.

4.1.2.1.4 Defricativisation

Defricativization is another process that is caused by Prenasalisation. Here, according to

Cammenga (2002), if the consonantal element in the combination that is prenasalised is a voiced continuant, it loses the [+CONTINUANT] feature. In other words, it becomes [-

CONTINUANT]. This, Cammenga observes, is accounted for as rightward spreading of [-

CONTINUANT] specification of the nasal to the consonantal element. This process is described thus: Voiced continuants are turned into voiced obstruents whenever they are prenasalised. In

EkeGusii, /β/ is turned into[b], /γ/ into [g] and /r/ into [d] as illustrated by (25). 25) EkeGusii defricativisation i) /β/ →[b] input: /e-n-/ + /βori/ ‘goat’ affixation | e-n-βori| Prenasalisation / enβori/ Nasal homorganisation [embori] ii) /γ/ → [g] input: /ɛ-n-/ + /γͻri/ ‘rope’ affixation ɛnγͻri Prenasalisation ɛnγͻri Nasal homorganisation ɛŋgͻri

74 iii) /r/ → [d] input: /e-n-/ + /raγera/ affixation enraγera Prenasalisation enraγera Nasal homorganisation endaγera Adapted from Cammenga (2002) This data shows that whenever a voiced continuant obstruent is adjacent to a nasal, it loses its

[+CONTINUANT] feature and becomes [-CONTINUANT]; in other words, it is defricativised.

This confirms the fact that EkeGusii does not have the stops that are the end products of defricativization ([b], [g] and [d] respectively. Comparatively, defricativization is not a characteristic of English phonology as in EkeGusii. This can be explained by the fact that nasals plus consonants in English can be treated as consonant clusters since the language allows consonant clusters as discussed in (4.1.13) below. Like in

EkeGusii however, nasal consonant homorganization characterizes English as illustrated by (31). 26) English nasal consonant homorganization i) ink → iŋk ii) tomb → tu:m In (26i), the consonant /k/ homorganizes /n/ to [ŋ], while in (31ii), /b/ disappears as a result of being hormoganized to [m].

4.1.2.1.5 EkeGusii Nasal re-syllabification

Ferguson (1963), Hyman (1985), and Nasukawa (2004) observe that syllabic nasals, which are found in languages such as Pali, Japanese, and many Bantu and Ogoni languages, exhibit both consonantal and vocalic characteristics in terms of their tonal properties and syllabic distribution.

This is true of EkeGusii language. According to Cammenga (2002), whenever prenasals occur word initially, their nasal elements may optionally become syllabic and bear tone. This tone may or may not be distinct from that of the next tone bearer, that is, the next syllable. Such changes may occur in word initial position only. Nor does it seem to be limited to prenasals only. In explaining the nasal resyllabification process in this section, this research will, in the process,

75 explain other rules, which according to Cammenga and indeed this study are presupposed by the process. In fact, Cammenga simply refers to the various processes which finally lead to syllabification as delinking rule.

Though viewed as optional occurrence in word initial positions, nasal syllabification is a common process, especially in Bantu languages. In cases where a nasal is followed by a consonant, syllabification takes place as exemplified by (27) adapted from Cammenga

(2002:90).

27) a) nasal syllabification in word initial position

/n-to- taat∫ -ɛ/

F-1p-fetch-FV

[ntotaat∫e]

‘We will fetch (water) today’

b) /in- mo- taa ts- e/

F– 2p1-fetch-fv

[mmotaat∫e]

‘You will fetch (water) today’

c) in-a-taat∫-e/

F-3p-fetch-fv

[mbataat∫e]

‘They will fetch water today’

Syllable nasals are underlined in (27). The data indicates that the nasal element is in the word initial position. There are cases where nasals may also be syllabified before vowels as in (28).

76 (28) nasal syllabification before vowels

Input: /-e-/ ‘forget’

Suffixation -ee

Prefixation n-e-

Nasal resyllabification ne-

Pre-nasal /i/ insertion ine-

Nasal velarization -iŋe-

Output: [iŋee] ‘forget me’ Adapted from Cammenga (2002: 90)

Data sets (27) and (28) are accounted for by word initial delinking rule, which is exemplified by figure (12).



x x

[word[+nasal]] [+consonant] Figure (12): EkeGusii delinking rule Source: Cammenga (2002)

This figure shows that, the delinking rule optionally delinks, in word initial position, a pre- nasalized consonant from the syllable ([]) to which it is attached. This is what necessitates re- syllabification. This is because, the delinked word initial nasal floats, which by convention may not be relinked to the following consonant, figure (13) further exemplifies.



77 [word [+Nasal]

Figure (13): EkeGusii nasal re-syllabification rule

Source: Cammenga (2002)

This rule optionally assigns a nucleus to any floating word initial nasal. This is nasal re- syllabification. The process of nasal syllabification starts with prenalization, where word initial nasals are pre-nasalized. Pre-nasalization then triggers nasal homoganization, in which a nasal shares place feature with the consonant it precedes. Then defricativization takes place where and when applicable especially when the following consonant is a fricative(continuant). This is then followed by the nasal de-linking process, as in figure (12) above and finally, re-syllabification as in figure (13). This process is summarized in (29) as follows.

29) EkeGusii nasal syllabification process

Input /-γor-/ ‘buy’

Suffixation /-γore/

Prefixation /n-γore/

Prenasalisation /nγore/

Nasal homogenization ŋγore

Defricatirization ŋgore

Nasal delinking ŋ-gore

Nasal resyllabification ŋ.gore

Output [ŋ.go.re]

Adapted from Cammenga (2002)

78 This data show that the nasal consonant in the syllabified form forms the initial syllable of the word in which it is initial. This is after delinking itself from the syllable in which it is attached.

This means that it does not form a consonant cluster with the consonant with which it occurs.

This is illustrated by Figures (14) and (15) for the output in (29) above.

  

N C V C V

ŋ g o r e

Figure (14): Nasal delinking leading to nasal syllabification Adapted from Cammenga (2002)  

*CC V C V

*ng o r e

Figure (15): Nasal consonant cluster forming part of a syllable

Adapted from Cammenga (2002)

In figure (14), the nasal forms a syllable on its own, it has delinked itself from the syllable to which it is attached, while in figure (15) it is part of the syllable it is attached, thus forming a consonant cluster which is not allowed in EkeGusii.

Following the foregoing discussion and conclusions on nasal resyllabification which has mainly drawn from Cammenga (2002), this study supports the argument that all nasals in EkeGusii are syllabified. This observation further, supports the arguments that EkeGusii does not entertain 79 consonant clusters. This is because the delinking of the nasal from the consonant with which it occurs, makes the nasal stand on its own as a syllable. In EkeGusii, like in most languages, vowels, unlike consonants, form syllables on their own. Therefore, the nasals in this study are treated more as vowels as compared to consonants, because they occupy vocalic positions in syllables.

Syllabification of nasals by delinking as described in this section does not characterize English phonology. English, entertains consonant clusters, and as has already been observed, the nasal plus consonant combination form a cluster. Delinking a nasal from a consonant in English creates non-syllabic structures as illustrated by (30).

30) Nasal consonant delinking in English i) ink → /iŋk. / ii) ink → *[iŋ.k], ‘ink’.

(35ii) is ill formed because the nasal [ŋ] has been delinked from the consonant [k], creating two unacceptable syllables (in English), instead of one as in (35i).

4.1.3 Comparison of EkeGusii and English syllable structures The description of the syllable in this study is based on the typology of syllable inventories originally stated in Jacobson (1962) and elaborated in Clements and Keyser (1983) and Prince and Smolensky (2004). This is a typology based on syllable inventories attested across languages. It belongs to a class of substantive universals, and includes the implicational relations that hold among specific syllable shapes. De Lacy (2006) gives a typology of different languages syllable shapes repeated in table (5). Table (5): Typology of syllable shapes

Onset coda onset coda cluster Inventory Languages

80 cluster O (C)CV(C)(C) Totonak O X (C)CV(C) Dakota R O O CV(C)(C) Klamath X X CV(C) Temier R X O _ (C)VC Arabela X CV Senufo O (C)(C)V(C)(C) English O O O X (C)(C)V(C) Spanish

X O (C)V(C)(C) Finish X (C)V(C) Turkish O - (C)(C)(V) Piraha O X X - (C)V Fijian

Adapted from De Lacy (2006: 165) Key R= Required, O= Optional, X= Banned. X therefore means that:

Codas are never required Coda clusters are never required Onset clusters are never required Onsets are never banned

81 Generally, Ekegusii has a (V) CV syllable structure (Cammenga, 2002). Thus, the language is characterized with an open syllable structure and sometimes a single vowel word initially as illustrated by (31).

31) EkeGusii syllable structure

Syllable Underlying Surface EkeGusii Gloss

form

CV. CV. |ke.tii| [γetii] getii field

a) CV.CV. /βa.na/| [βana] bana predict/fore-tell

b) CV.CV.CV /tɛ.rɛɛ.ra./ [tɛ.rɛɛ.ra] tereera sing for

c) V.CV.CV /o.mo.te/ [omote] omote tree

d) V.CV.CV. CV /o.mo.γo.ri./ [o.mo.γo.ri] omogori buyer

e) V.CV.CV.CV /a.a.a.ni/ [aaani] ababani prophets

Adapted from Bosire and Machogu (2013)

98 (31a), for example, can be presented on syllable nodes as in figure (16).

Input: /ke.tii /, output: [γetii] ‘field’

σ σ

C V C V

γ e t ii

Figure (16): EkeGusii syllable nodes for /γetii/

Adapted from Cammenga (2002)

These syllable structures generally presuppose that syllables should have onsets and that the consonants in the input are the same as the consonants in the output respectively

(Smolensky & McCarthy, 1993).

There are cases of single vowel syllables in EkeGusii. This, however, is a case of onset violation, where a vowel begins in a word, especially in nouns number and class marking pre-prefixes and prefixes and in some single vowel words as illustrated by (32).

32) Single vowel syllables in EkeGusii a) i) o- mo -te ‘tree’ ii) e- me- te ‘trees’ Aug prefix root Aug prefix root 3sg 3.3 -tree (sing.) 4pl 3.4 - tree (pl.)

b) a.aa ‘pluck’ (vegetables etc.) The prefix {omo-} in (32ai) above marks the class of the noun ‘tree’, that is class 3 and number, that is singular, while the prefix {eme-} in (32aii) marks class four and plurality.

Examples (32ai and (32b) above will be represented on a syllable node as in figures (17). 99 i) o.mo.te

v.c v.cv. σ σ σ

v c v c v

o m o t e ii) a .aa

v.v σ σ

v v

a. aa Figure (17): EkeGusii syllable nodes for [o.mo.te] and [a.aa]

EkeGusii syllable structure is comparable to that of English. While EkeGusii is a (V)CV language as shown in (31), the syllable typologies given in table (3) above, show that

English is a (C) (C) V (C) (C) syllable type of language. Thus, Onsets; Codas; Onset clusters; and Coda clusters are all optional in English. They may or may not occur depending on the nature of the word as shown in (33).

33) Syllable types in English Word syllable type cat /kᴂt/ cvc boy /bͻɪ/ cv structure /strᴧ.kt∫ə/ cccv.ccv owe /əʊ/ v According to Roach (1983), if the first syllable of a word begins with a vowel (and in

English any vowel may occur though /ʊ/ is rare), the syllable is said to have a zero onset. If

100 it begins with one consonant, that consonant may be any, except /ŋ/ and /ʒ/ which are rare in this position. There are two types of two-consonant clusters in English; that which begins with /s/ as in string, sting, sway and smoke. In this case, the /s/ is pre-initial while the other consonants e.g. /t/ /w/ and /m/ initial. The other is that which begins with a consonant followed by either of the following: /l, w, j/ and /r/ as in play, tray, and quick few. The first consonant here is called the initial, while the second one post-initial. Consonant clusters are up to four. Examples of three initial consonant clusters include: split /splɪt/, stream /stri:m/, square /skwea. Equally, there are final consonant clusters which contain up to four consonants: two consonant cluster may include: bump /bᴧmp/, bent /bɛnt/, bank /bᴂŋk/, belt /bɛlt/, ask /ᴂsk/, begged /bɛgd/, and looked /lʊkt/ among others. There are two types of final three consonant clusters: final plus final plus post-final as in helps, banks, and bonds; and final plus post-final 1 Plus post-final 2 as in fifths (Roach 1983).

4.1.3.1 EkeGusii and English phonotactics Generally, all the fourteen consonants in EkeGusii occupy the onset position, while none occupies the coda position, because, the language is a CV one as has already been observed. Equally, all the vowels of the language take all the positions of a word; initial, medial and final. The same is not true of the English phonemes. For example, some consonants in English do not occupy onset positions; similarly, others do not occupy coda positions. According to Cruttenden (2011), English does not exploit, the syllable, all possible combinations of its phonemes. For instance, long vowels and diphthongs do not precede final /ŋ/; /e, ᴂ, ʌ, ɒ/ do not occur word finally; and the types of consonant cluster permitted are subject to constraints in both initial and final positions. /ŋ/ does not occur word

101 initially; no combinations are possible with /t∫, dʒ, ð, z/; /r, j, w/ can occur in clusters only as the non-initial element; such initial element sequences as /fs, mh, stl, spw/ are not allowed. Finally, only /l/ may occur before non – syllabic /m, n/; /h, r, j, w/ do not occur in word final positions; and terminal sequences such as /kf, ∫p, ɪ, ʒbd/ are not used in the language. In the following sub sections; 4.1.3.1.1 and 4.1.3.1.2, word initial, word final phoneme sequences and inflection suffix formation constraints of English are discussed respectively.

4.1.3.1.1 English word initial phoneme sequences Word initial consonant sound sequences in English vary from word to word. There are words with only a single consonant word initially, while there with four consonants.

Cruttenden (2011) observes that there are ten vowels in English, which constitute monosyllabic words as given in (34).

34) English vowels constituting monosyllabic words vowel word

/i;/ e letter

/ǝ/, /a/, /ɑ: / are

/ͻ:/, /ɜ: / err

/eɪ/ a letter

/aɪ/ i letter

/ǝʊ/ owe

/ɪǝ/ ear

/eǝ/ air

Adapted from Cruttenden (2011: 201)

102 (34) shows that one vowel makes up an English word, which is monosyllabic. Cases of vowels occurring word initially as syllables are common in English. According to

Cruttenden (2011) all vowel sounds can occur word initially in English depending on the word in question. Thus, some English words allow vowels in word initial position, while others do not.

There are cases of consonant vowel (CV) in word initial positions with an exception of the consonants /ŋ/ and /V/. All the other consonants generally occur before all vowels. In

English also are cases of consonant consonant vowel (CCV) word initially. In (35), there are two consonant cluster patterns for English word initial positions as repeated from

Cruttenden (2011).

35) Two consonant cluster patterns for English

Cluster form Examples of words

P+l, r, j ply, pray, pure t+ r, j, w try, tube, sweep k+l, r, j, w class, crush, cube b+l, r, j, blood, breed, beauty d+ r, j, w dry, dupe, dwell g+ l, r, j, w glass, grass, m+ j mew n+ j new l+ j lure, leau f+l, r, j flow, fraud, few

103 v+j view

θ+ r, j, w throw, thief s+l, r, j, w, p, t, k, m, n, f, v slow, sir, sue, spree, store, skin, smart,

∫+l, r, w, m, n shrewd, h + j hew

There are also cases of three consonant cluster patterns word initially in English as in (36), repeated from Cruttenden (2011).

36)Three consonant cluster patterns for word initial position in English Cluster form example of words s+p+l, r, j splendid, spring, spying s+t+ r, j street, skive s+k+l, r, j w screw, skew, squad

As can be observed, /s/ is the essential first element of the CCC clusters, the second element being a voiceless stop, the third element must either be /l, r, j/ or /w.

4.1.3.1.2 English word final phoneme sequences There are cases of word final vowels in English. Cruttenden (2011) observes that most of the English vowels except /e, ᴂ, ʌ/ and/ɒ/ occur word finally. Concerning cases of final vowels and consonant combinations, Cruttenden observes that /r, h, j w/ do not occur word finally. /ʒ/ occurs finally only after the vowels /i: ɑ:, u:/ and /ei/ in words of recent French origin, like liege /li:ʒ/, rouge /ru:ʒ/, beige /beiʒ/. /ŋ/ occurs only after the vowels /ɪ, ᴂ, ʌ/ and /ɒ/. There are also cases of VCC (vowel, consonant, consonant) combinations. There are a few mono-morphemic words of this kind including act, adze, axe, corpse, and lapse.

The consonants /r, h, j/ and / w/ do not combine with other consonants in word final

104 positions in English (RP). /g, ŋ/ do not occupy final position in a final CC cluster. /θ/ is of limited occurrence in this position.

Cases of English final VCCC, that is; that of a vowel followed by a consonant cluster of three, do occur in English such as: collapsed /kɒlǝpst/, text /tekst/, and prompt /prɒmpt/.

These final CCC English clusters can be divided into two groups: (i) those, which involve a combination of the two types of CC clusters, that is, /m, n, ŋ, l, s/ plus C plus /t, d, s, z, θ/.

These according to Cruttenden (2011), nearly all involve suffixes such as jumps, cults, lists, but there are monomorphic words such as mulct and calx. (ii) Those, which involve the double application of /t, d, s, z, θ /; the majority in this case involves suffixes such as fifths /fifθs/, products /prɒdʌkts/, acts /ᴂkts/; but there are two common monomorphic words, text /tekst/ and next /nekst/ (Cruttenden, 2011). Cruttenden further observes that the

CCC clusters predominantly follow short vowels. Eleven of the 49 CCC final clusters occur after only one vowel (that is, five after /ɪ/, four after /e/, one after / ʌ / and one after /

ǝ /).

Finally, there are cases of VCCCC final word syllable. The CCCC clusters occur only rarely as a result of the suffixation to CCC clusters of /t/ or /s/ morpheme as in: /-mpts/ in prompts, exempts; /-mpst/ in glimpsed; /-lkts/ in mulcts; /-lpts/ in sculpts; /-lfθs/, twelfths; /-ntθs/ thousandths. Both of these word initial and word final phoneme sequences indicate that there are cases of syllable complex margins in English.

105 4.1.4 Comparison of EkeGusii toneme structure and English stress

This section presents a tonal description of EkeGusii noun as compared with English stress.

The focus of the section is on the tonal patterns of EkeGusii noun in isolation and stress as it characterizes the English noun.

Tone has been defined differently by different phonologists. According to de-Lacy (2007), tone is the use of pitch in language to distinguish lexical or grammatical meaning.

Languages that are characterized by these feature are known as tone languages. Many language of the world are tonal (Katamba, 1993 and de Lacy, 2007). Such languages, according to Katamba and de-Lacy, have morphemes which are sometimes realized by pitch changes, that is using pitch differences to make phonemic contrasts. In tone languages therefore, pitch can be used to distinguish word meaning or convey grammatical distinctions. It is in this respect that tone languages differ from non-tone (stress) language like English where pitch does not have these functions.

4.1.4.1 EkeGusii tone structure

EkeGusii is a tone language (Bickmore, 2007; Nash, 2011; and Cammenga, 2002), in which pitch is used in the distinction of grammatical meaning more than lexical meaning.

Examples of noun lexical contrasts based on tone are given in (37).

37) EkeGusii noun tone distinction

Word Tonal realization Gloss

(i) omogaaka /ó m ò γ á à k a/ Old man 106 omogaka /ó m ò γ à k à/ aloe vera

(ii) omote /ó m ò t é/ tree

omote /ó m ò t è/ name of a person

(iii) esese /ē s ē s ē dog

(iii) esese /έsέsέ/ strain

This data shows that the distinction between the given words is as a result of contrastive vowel length, which according to Goldsmith (1990), is referred to as compensatory lengthening and tone differences. In compensatory lengthening, vowels simultaneously linked to several verb-slots are described as long and are at times phonemic, that is contrastive (Katamba, 1993). The nouns omogaaka /omoγaaka/; ‘old man’ and omogaka

/omoγaka/ ‘aloe vera’ are distinguished by the length of the first vowel of their roots as illustrated by figures (18) and (19).

C V V C V

γ a k a -γaaka

Figure (18): EkeGusii lexical contrasts due to tone structure

C V C V

γ a k a -γaka 107 Figure (19): EkeGusii lexical contrasts due to tone structure Adapted from Katamba (1993)

Figure (18) shows that compensatory lengthening takes place when a single vowel is doubly-linked with two verb slots in the underlying representation. Thus /a/ surfaces as long [a:] in omo-γaaka. This is not the case in figure (19) where, the vowel /a/ is linked to a single vowel slot, thus surfacing as a short vowel. Lexical contrast between the words in figures (18) and (19) are based on tone distinctions. This sub-section briefly describes

Ekegusii noun tone structure in which 4.1.4.1.1 describes underlying versus surfaces tones,

4.1.4.1.2 Contour tones, 4.1.4.1.3 tone preservation, and 4.1.4.1.4 tone floating.

4.1.4.1.1 Underlying and surface tones in EkeGusii

Following Bickmore (1997), Pulleyblank (1986) Nash (2011), and others on Bantu languages tone structure, this study takes the view that that the underlying tonal distinction in EkeGusii is one of high versus toneless, that is low tones are underspecified underlyingly, only introduced at a later stage to the surface through insertion (Mwita,

2012). This is in agreement with Kisseberth and Odden (2003) who observe that the surface tone of the augment and the class prefix is normally low in Bantu languages. Following these observations therefore, this study argues that Ekegusii has two basic surface tones,

High (H) and Low (L).

Surface tones are marked by accent marks, a transcription form used by Africanists (de-

Lacy, 2007: 230). High tone in this case is marked by an acute accent (´) and represented by H, while a low tone is either unmarked or marked with (-) and represented by L. This is illustrated by figure (20) and is used in this study.

EkeGusii tone realization word Gloss (i) /o  o k´i m a/ obokima ugali 108 L L H L

(ii) om o t∫ ´o k o r´o omochokoro grand child

L L H L H Figure (20): EkeGusii tone marking

Adapted from: de Lacy (2007: 26)

Figure (20) above shows that the tone bearing units (morae or vowels) in class prefixes are low (L) toned, while the first tone bearing units in the roots are high (H) toned.

4.1.4.1.2 Contour tones in EkeGusii According to Katamba (1993), Autosegmental Phonology Theory does not require a one- to-one association of elements on different tiers (tonal tier, segmental tier, and CV tier).

Elements at any one tier may be linked one-to-many with elements in another tier. The following tonal examples in figure (21) of Mende language data (Leben, 1978) repeated from Katamba (1993: 157) contain falling or rising tones. Such tones are called contour tones.

bεlε mbu mba כ k

H L H L L H ‘war’ ‘trousers’ ‘owl’ ‘rice’

Figure (21): Mende contour tones

Figure (21) shows that the Mende tonal contours are made when independent high and low tones are simultaneously linked to a single vowel (Katamba, 1993: 157). 109 EkeGusii language, like other Bantu languages, Kuria (Mwita 2012), has a rising contour

(LH) when only one of the two consecutive vowels in a long syllable is marked for tone, that is, it is high. This is illustrated by figure (22).

Word tonal gloss word tonal gloss (Singular) realization (plural) realization emoori /e m ō ó ri/ calf’ chimori c h i m ō ó r i calves

L H LH omoonto om ó ó n t o/ Person abaanto ab ā á n t o Persons

L H L H Figure (22): Ekegusii LH contour tone Adapted from Katamba (1993)

This figure shows that the long syllable which starts with a low tone and ends in a high one forms an LH contour. The figure further indicates that this is when the root of the noun starts with a vowel, which is a copy of the prefix vowel. However, this is not the case when the root starts with a consonant. In such a case, both consecutive vowels bear the same tone marking and therefore the syllable is level, that is it is pronounced with the same pitch.

This is illustrated by figure (23).

Word Tone Gloss Word Tone Gloss (Singular) realization (plural) realization obokokombe /o o – kó ó m b e/ hoe amakombe /ama – k ó ó m b e/ hoes

HH HH omogaaka /omo – γ á á k a/ old man abagaaka /aba–γ á á k a/ old men

H H HH

Figure (23): Ekegusii level tone

110 Adapted from Katamba (1993)

4.1.4..1.3 Tone floating in EkeGusii

As has already been observed, deletion of the vowels as in the examples given in data set

(45) above does not directly affect the tones which are associated with the vowels deleted, and as a result, after the deletion of the vowels, the tones simply remain on the tonal tiers with no association with the segmental tiers. This study, like others such as Oden’s (2005) is of the view that such an association creates floating tones. Figure (24) illustrates this observation. e g e n t o e k e e g e nt eke ‘this thing’

H L H L H H L HLH

Figure (24): EkeGusii floating tone

Figure (24) indicates that the high tone of the vowel /o/ in egento ‘thing’ floats at the surface. It is this floating tone that is associated with the following vowel /e/ (low toned or high toned), resulting in a falling or rising tone (a contour tone), in this case being a falling tone (HL).

4.1.4.1.4 Stress in English

111 While EkeGusii is a tone language as has been discussed above, English is a stress language. Verma and Krishnaswamy (1989) define stress as the intensity or prominence with which a syllable is produced. Thus, in the production of a word or any syntactic structure (sentence or phrase), there is always a syllable which is produced with prominence. It is this prominence that phonologists have come to refer to as stress.

According to Laver (1994), if prominence is put on syllables on isolated words, the resulting stress is referred to as word stress. Prominence given to words in sentences on the other hand is known as sentence stress. This study is interested in word or lexical stress, in particular, noun lexical stress.

Languages like English with syllables that differ in stress are stress languages. This means that these languages have more than one stress, normally, a loud or primary one which is marked by a short raised stroke [']; a medium or secondary one, marked by a short lowered stroke [‚]; and an unstressed one, which involves a non-prominent syllable containing no pitch changes and has one of these vowels /ɪ, ʊ / or /ǝ/ (Laver, 1994).

Depending on the number of syllables, class of the noun, and the nature of the word, whether compound or not, a noun will be stressed differently. Since every word has at least one or more stressed syllables (Laver, 1994), monosyllabic nouns have their only one syllable stressed. Equally, bisyllabic nouns have their primary stress on the first syllable as shown in (38).

38) English monosyllabic and bisyllabic noun stress i) monosyllabic nouns ii) Bisyllabic words / 'mæn / man / 'pǝmɪt/ permit

112 / 'strɪkt /strict / 'ekspͻ:t/ export / 'wi:k / weak / 'kɒntrʌkt/

(38) shows that in all the words, stress is placed on the first syllable of the given nouns.

The following subsection gives a brief description of stress in the English noun.

For nouns with three or more syllables (that is polysyllabic words), stress is determined by the ending of the noun in question or generally the suffix (Laver, 1994). Thus, in nouns which end in either –er or –ly, primary stress is placed on the first syllable, just like in the monosyllabics and bisyllabics above. This is illustrated by (39).

39) Primary stress on polysyllabic nouns ending in {-er} or {–ly}

/ '/ 'ɒdǝlɪ/ orderly

/ 'mænɪdʒǝ/ manager

4.1.5 Comparison of EkeGusii and English morphological structures

This sub-section focuses on the morphological processes that give insight into morphological nativization of EkeGusii loan nouns from English. It describes morphological processes that explain word building processes in EkeGusii as compared to

English. It describes the morphosyntactic classes of EkeGusii, and in doing so, the study relies heavily on Cammenga’s (2002) pioneering findings for EkeGusii. This study however, unlike Cammenga’s, which is not anchored on any theory, alludes to tenets of

Optimality Theory in its generalized descriptions.

4.1.5.1 Comparison of EkeGusii and English noun systems

According to Demuth (2000) noun classes in Bantu languages tend to be realized as grammatical morphemes rather than independent lexical items. These classes function as

113 part of a larger concordial agreement system where nominal modifiers, pronominals, and verbs are all morphologically marked with the same noun class (gender) features. Demuth further observes that the classes are presently morphologically productive in most Bantu languages; and that semantically, the classes have been reconstructed from Pro Bantu.

Thus, much of the semantics of current Bantu noun classes is no longer productive, and in some languages the number of classes has been reduced. Demuth concludes that, despite all the given observations, noun class systems especially morphologically are grammatically productive in most Bantu languages, and semantically productive to some degree. Just as

Demuth (2000) notes, EkeGusii nouns are characterized as grammatical morphemes and function as part of a larger concordial agreement system. Comparatively, most of the nouns in English, unlike those in Bantu, are realized as independent lexical items. This is cognizant of the fact that language morphological typologies exist. Haspelmath (2002), identifies three types of such languages typologies: isolating, agglutinative and fusional. He observes that some languages are close to ideal types; that is close to either completely isolating (such as Chinese and Vietnamese) or agglutinative (such as Turkish). Most languages however, are mixed types; sharing features of different given ideal types. English and EkeGusii are mixed morphological typology languages. What distinguishes them however is the degree of fusion and or agglutination (index of synthesis). For example, grammatical relations are shown mainly by means of prepositions in English, thus resembling the patterns of isolating languages. However, the derivational and inflectional morphologies of the same language are partly agglutinative and partly fusional. EkeGusii on the other hand, like most Bantu languages like Kiswahili (Haspelmath, 2002), is more agglutinating than isolating. Indeed, in an index of synthesis given by Haspelmath,

114 Kiswahili, is ranked higher than English, which therefore means that EkeGusii, is more synthetic or agglutinating than English. In the following sub-sections, the mophosyntactic classes and prefixes of EkeGusii are described in relation to English morphology.

4.1.5.2 Morphosyntactic classes and prefixes in EkeGusii noun

Nouns in Bantu are classified into sets referred to as noun classes (Meinhoff 1899).

According to Welmers (1973), there are at least 22 of these noun classes in Pro-Bantu, but individual languages have less than the Pro Bantu number. For example, Kiswahili has 16,

(Carsteins, 1991 &1993), Sesotho 15 (Demuth, 2000), Kivonjo 16 (Pinker, 1994), Aghem

12 (Aikhenvald, 2000), EkeGusii 20, (Cammenga, 2002 & Ongarora, 2009). Morphosyntactically, an Ekegusii noun consists of a prefix and a stem both of which generally compulsory. With an exception of a few classes, the prefix carries number and size features and has a (vowel) consonant vowel (V) CV syllable structure (Cammenga,

2002). (40) represents EkeGusii noun class prefixes carrying number and size features as repeated from Cammenga (2002:199).

40) Morphosyntactic noun class prefixes in EkeGusii 1 omo - 2 aβa- 1a mo- 1b Ǿ 3 omo- 4 eme- 5 eri- 6 ama- 7 eke- 8 eβi- 9 e- 10 chi- 9a e-n- 10a chi-n- 11 oro- 12 aka- 14 oβo- 15 oko- 16 a- 21 na

According to Givon (1972), Cammenga (2002) and Ongarora (2009), the choice of these prefixes is determined by the semantics of the noun stems with which they occur. In other

115 words, the prefixes carry the gender, number and size of the stems to which they are appropriately (in terms of semantics) prefixed as illustrated by (41). 41) EkeGusii noun gender prefixation a) omonyaroka ‘girl’ abanyaroka ‘girls’ omo – ɳaroka aβa- ɳaroka 1.3PSG- girl 2.3PPL- girl ‘girl’ ‘girls’ omo- gaaka aβa- gaaka 1.3PSG- ‘old man’ 2.3P PL ‘old men’ ‘old man’ ‘old men’ b) ekerandi ‘gourd’ ebirandi ‘gourds’ eke- randi eβi- randi 7.3PSG- ‘gourd’ 83PPL- ‘gourd’ ‘ gourd’ ‘gourds’ eke- moni eβi- moni 7.3PSG- cat 8.3PPL cat ‘cat’ ‘cats’ Adapted from Ongarora (2006) In (41a) above, the noun stem {nyaroka} ‘girl’ denotes ‘human’ referent, hence co-occur with singular prefix {omo-} and a plural one {aβa-}, while that in (41b) refers to an inanimate referent {randi} ‘gourd’ and accordingly co-occur with the singular prefix {eke} and the plural prefix {eβi-}. Thus, the mutual exclusivity of these prefixes stems from the gender of the nouns (Givon, 1972 & Ongarora, 2009). Table (2) shows EkeGusii prefixes, both in their singular and plural forms, and their stems semantic determinants.

Table (6): EkeGusii Prefixes and their Stems Semantic Determinants Prefix Noun stem semantics (meaning) determinants Singular Plural

1. omo- 2 aβa- personal, spiritual, animate beings, kinship terms i.e. God, angles, devils, the spirits of the ancestors and kinship terms (human referents) 1b Ø- 2aβa kinship terms (human referents) 3 omo- 4 eme- socioculturally relevant objects, events or periods; trees, parts of the body (non-human referents) 5. eri-/rii- 6. ama- various types of common nouns e.g. cultural or objects and location, tools, parts of the body, fruits

116 5. eri- 6. ama- augmentative + or pejorative- 7. eke- 8. eβi- inanimate, mostly cultural objects; some parts of the body; some animals, some shrubs or plants; language names 7. eke- 8. eβi- diminutive +or - pejorative 7. ke- no plural: adverbs, places names 9. e- 10. t∫i- many names of animals; socially or culturally relevant entities (place, objects, events); some concepts 9a. e-n 10a. t∫i-n same as 9-10 11. oro- 10a. t∫i-n social cultural and some natural objects 12. aka- 8. eβi- Diminutive 12. aka- 14. oβo- diminutive, non-pejorative 14. oβo- ama- some body parts; culturally relevant entities (objects, places, events, activities); some crop names 14. oβo- no plural: concepts 14. βo- no plural: adverbs, place names 15. 0ko- 6. ama- some body parts; abstract nouns mostly referring to activities or events conceived abstractically (usually without plural) 15. ko- infinitive marker (together with word- final suffix –a, expressing activities or events 16. a- [ase] ‘place’ only, no plural 21. ɳa- no regular plural: proper names, of persons, individual heads of cattle, and places Source: Cammenga (2002: 201)

This table shows that occurrence of prefixes with noun stems/ roots are semantically determined. Thus, the meaning of the stems to which the given prefix is attached plays a major role in its choice. In other words, occurrence of a prefix is not haphazard and without meaning. Nouns in English unlike in EkeGusii are not classified in terms of classes in the sense described above. In fact, as can be observed in table (2), the class of a given noun in

EkeGusii, like in other Bantu languages, is determined by the prefix. Prefixation in English performs different functions such as marking opposite, for example, un- in ‘unlock’. English, according to Katamba (1993) is a language that is characterized by base word morphology. Base word morphology entails the study of the lowest indivisible level of a

117 morphological construction (Kiparsky & Moahannan, 1982). McCarthy (2002) observes that an important feature of English, which differentiates it from many other languages, is that it has a high proportion of complex words, with an agglutinative morphology; and an equally large number of words with an isolated morphology. Therefore, as illustrated in

(42), English morphology is neither purely isolating nor purely synthetic.

42) English morphemes

(a) (b)

read – able leg – ible

hear – ing audi – ence

en – large magn–ify

perform – ance rend – ition

In (a) the two morphemes affixed together are different respectively- free and bound, while those in (b) are both bound. The difference as observed by McCarthy is attributable to the history of English. Most of the free morphemes in (a) belong to that part of the vocabulary of English that has been inherited directly through the Germanic branch of the Indo-

European language family to which English belongs, whereas the morphemes in (1b) have been introduced or borrowed from Latin, either directly or via French. Again, the words in

(a) are more common than those in (b) which reflects the fact that among the most widely used words, the Germanic element still predominates. This leads to the conclusion that in

English, there is a strong tendency for complex words to contain a free morpheme at their core. This is the argument this study is based on.

118 Structurally, most noun prefixes in EkeGusii, unlike in English, have a bi-morphemic form.

Thus, the prefix is divided into two elements: an initial vowel, sometimes referred to as an augment or pre-prefix, and the prefix per-se (Elwell, 2005). The pre-prefix is described in

4.1.2.2.1, the prefix in 4.1.2.2.2, and the noun roots in 4.1.2.2.3.

4.1.5.2.1 The pre-prefix or augment

The pre-prefix according to Elwell (2005), is a syllable added to the beginning of a word in certain languages. EkeGusii, unlike English, has such a syllable; especially in noun number and class marking prefixes and some monosyllabic words (in which case, the augment is just a single vowel). (43) gives the EkeGusii augment structure.

43) EkeGusii augment structure a)omote ‘tree’ i) o- mo- te ‘tree’ ii) e- me- te ‘trees’ aug 3SG tree aug 4PL tree b) eee ‘yes’ e-ee ‘yes’ Adapted from Bosire and Machogu (2013)

The prefix {omo-} in (43ai) marks: the class of the noun ‘tree’, that is class three and number, that is singular; while the prefix ‘eme-‘ in (43a ii) marks class four and plurality. The augment structures in (43) above is represented on syllable node in figure (25).

i) o.mo.t e σ

v.c v.c v σ σ σ

v c v c v

o m o t e ii) e. e e σ σ σ 119

v.v v v v v e e e Figure (25): EkeGusii prefix and pre-prefix syllable nodes Adapted from Katamba (1989)

The pre-prefixes in figure (25 i and ii) in each of the given words are made up of single vowel syllables, the vowel /o-/ in (21 i) and /e-/ in (25 ii). The output for class 1 affix is

‘omo-in figure (25 i).

This study is of the view that the vowels at the beginning of a prefix are tolerated because without them, the prefixes that result are those of classes 1b (ø-) and 2b (ø-), which carry the meaning of kinship terms or sometimes when referring to nobody in particular (that is neutrally) as illustrated by (44).

44) Ekegusii prefix of classes 1b (ø-) and 2b (ø-) i) monto / mo-nto / ‘person’ banto /βa-nto/ ‘persons’ ii) tata /tata/ ‘father’ batata /βatata/ ‘fathers’

These are described as follows.

- mo- nto -βa- nto

1bØ.3PSG person 2bØ.3PPL person

‘person’ ‘persons’

-tata -βa- tata

1bØ3PSG father 2b.3 PPL fathers

‘father’ ‘fathers’

120 The nouns in (44) differ from those data (43) in that while those (44) lack arguments, those in (43) have. In (44), where an argument lacks, the nouns prefixed refer to nobody in particular. The form ‘tata’ for example, is prefixless, it demands neither a pre-prefix, a prefix, nor both in the singular form (class1b). Of interest to note is the fact that its plural form, as can be observed, is either that of noun classes (2) or (2b).

In commenting on augmentation and non-augmentation, Cammenga (2002) observes that while augmentation is the basic or regular state of affairs in EkeGusii morphology, non- augmentation, which lacks an augment as in (43) above, may be characterized as the special case. He further points out that generally, both syntactic and semantic factors determine whether or not a word may take an augment. That is, the presence or absence of an augment is determined by lexical category membership and the semantics of the noun stem as has already been observed. In this respect therefore, nominal prefixes in morphosyntactic classes 1-8 and 11-15 may or may not be augmented, while the prefixes in classes 1b {ø-}, 9 (a) {e-(n)}, 10 (a) {t∫i- (n)}, 16 {a-} and 21{ɳa-}, are never augmented.

This is the view taken in this study.

The full EkeGusii prefix is generally made up of two parts: an augment (pre prefix), which is a vowel, /V/ and a prefix – proper, which is made up of a consonant and a vowel, /CV-/.

Therefore, an EkeGusii prefix takes the form V-CV, which covers nominal prefixes in classes (1-8) and (10-15) (Cammenga, 2002).

Lexically, class 5 prefix is regularly pre prefixed {eri–} in nouns of which the stem begins with a vowel, but non-pre-prefixed {rii–} in nouns of which the stem begins with a consonant. This is a case of phonologically determined allomorphy. In other words, it is the

121 sound at the beginning of the given noun which determines its pre prefixation or non pre prefixation. Otherwise the noun is one and the same thing. (Cammenga, 2002). (45), adapted from Bosire and Machogu (2013) exemplifies this observation.

45)EkeGusii class 5 prefixes (singular) a) rii-toke ‘banana’ b) eri-iso ‘eye’ rii-sosa ‘pumpkin leaves eri-ino ‘tooth’ rii-raba ‘soil’ eri-ogo ‘medicine’ rii-mama ‘dumb person’ eri-eta ‘name’

In data (45a) the nouns begin with a consonant and therefore do not allow augmentation.

(45b) on the other hand begins with a vowel and therefore allows augmentation. What qualifies them as phonologically conditioned allomorphs is the fact that they take the same prefix form in their plural, that is {ama-} as in rii-toke ama-toke and eri-so ama- iso. Classes 9 /e-/ and 16 /a-/ prefixes consist of a vowel which may not be augmented. A brief general description of the regulations of the shape or quality of the augment structure in EkeGusii language is presented as follows. (46) gives EkeGusii noun prefixes adapted from Cammenga (2002).

46) EkeGusii noun prefixes 1 o-mo- 1b Ø 2 a-βa- 3 o-mo – 4 e-me- 5 eri-/rii- 6 ama- 7 eke- 8 eβi- 9 e- 9a e-n 10 t∫i 122 10 a t∫i-n 11 oro- 12 aka- 14 oβo- 15 oko- 16 a- 21 ɳa-

A number of observations about the pre- prefix shapes in data (46) can be made. Firstly, all the prefixes with the form {CV-} allow pre prefixation except for those in classes 10 {t∫i-},

10a {t∫i- n}, 21 /{ɳa-} and allomorph {rii-} of class 5 discussed in (45) above.

Secondly, that the pre prefix is a copy of the prefix vowel except in the case of class 5

{eri-} 8 {eβi-} and 10 {t∫i-}. Thus, the augment in noun prefixes may be accounted for by a rule as in (7).

47) Noun prefix augmentation rule

The rule states that: copy the vowel of prefix {CV-} to the left of the input, such that any non-low output vowel must be [+ mid]. This, according to Cammenga (2002), includes all relevant prefixes and pre prefixes but appropriately excludes pre prefixation of all {V-} shaped prefixes. It ensures moreover that the [+high, - mid] or high, front prefix vowel of 5

/ri-/, 8 (βi-/, and 10 (a) t∫i-n/ is lowered to a [+high, + mid] or upper mid front augment vowel /e/.

4.1.5.2.2 The Prefix The structure of EkeGusii prefix has been described by a number of studies (Ongarora,

2009; Cammenga, 2002; and Whiteley, 1965). As has already been observed, there are 20 of these classes as given in (48).

48) EkeGusii noun classes Class Examples Gloss 1. omo- [omoonto] person 123 [omwaana] child 2. aβa- [aβanto] persons [aβaana] children 1bØ- [Øβaaβa)] mother [Ømaγokoro] grandmother [Øsokoro] grandfather 3. omo- [omotwe] head [omote] tree 4. eme- [emetwe] heads [emete] trees 5. rii- [riirok] foodrest eri- [eriiso)] eye 6. ama [amaγoko] footrests [amaiso] eyes 7. eke- [ekerandi] gourd 8. eβi- [eβirandi] gourds 9. e- [esese] dog [eusi] thread 10. t∫i- [t∫isese] dogs [t∫iusi)] threads 9 a e-n-|e-n-βaata| - [embaata] duck |e-n-raaγera| - [endaaγera] food 10 a) t∫i-n- |t∫i-n-βaata| [t∫imbaata] ducks |t∫i-n-raaγera| [t∫indaaγera] foods 11 oro- [oroko)] firewood 10 a) t∫i-n- [t∫iŋko] pieces of firewood 12 aka- [akaana] small honey comb 8 eβi- [eβinana] small honey combs 12 aka- [akamoonto] small person 14 oβo- [oβomoonto] small persons [oβosaat∫a] manhood abstract noun no plural 15 oko- [ͻkͻ βͻͻkͻ] arm [okoγoro] leg [oγoto] ear [okoruγa] cooking 6 ama- [amaoko] hands [amaγoro] legs [amato] ears 16 a- [ase] place [no plural] 21 ɳa- [ɳagera] blackie (cow, proper name (no plural) (ɳaγeeŋke) name of a place (proper name, no plural The prefixes are underlined. Source: Cammenga (2002)

Demuth (2002) observes that Bantu noun class systems can be characterized in two typological terms: first, noun classes normally realized as grammatical morphemes and not 124 independent lexical items, Second, the class system that morphosyntactically function as part of a large concordial agreement system, where nominal modifiers, pronominals and verbs are all morphologically marked with the same noun class (gender feature).

This study, like others in Bantu languages (Demuth, 2002; Ongarora, 2009; and Kayigema,

2010), recognizes the fact that EkeGusii noun classes tend to be realized as grammatical morphemes rather than independent lexical items. In the following sub-section, the noun classes are presented in their various grammatical morpheme forms as identified in (48) above. In particular, the descriptions in the sub-section focus on the root/base morphemes of the identified classes; prefixes and pre- prefixes having been accounted for in this and previous section.

4.1.5.2.3 EkeGusii noun class roots

It has already been observed in section (4.1.2.1) that Bantu nouns are realized as grammatical morphemes rather than independent grammatical items; and that these morphemes function as part of a large concordial agreement systems. Therefore, description of EkeGusii noun involves among other processes, the identification of the various constituent grammatical morphemes, including the root.

Katamba (1993:41) observes that, “…a root of a word is the irreducible core of that word with absolutely nothing attached to it. It is the part of a word that is always present, possibly with some modifications in the various manifestations of a lexeme.” For example,

‘talk’ in English is a root with the following word forms: talk, talk-s, talk-ing and talk-ed.

As can be seen, the form ‘talk’ cannot be reduced any further without losing its meaning.

125 This is how this study views the roots which are described in the following subsections according to their classes as identified in (48) above.

4.1.5.2.4 Class 1 {omo-} and 2 {aba-}

Nouns belonging to these classes are those within the meaning of personal, spiritual and animate beings, kinship terms, including God/gods, angles and spirits as described in table

(2) above. (49) gives examples of noun roots in classes (1) and (2).

49) EkeGusii noun roots in classes 1and 2

Noun surface underlying root gloss form form form

omoonto [omoonto] /o- mo- onto/ /-onto/ person

aug- 3psg- root abanto [aaanto] /a - a- anto] /-anto/ persons

aug-3pl- root omonyenyi [ͻmͻɳɛɳi] /ͻ- mͻ- ɳɛɳi/ /ɳeɳi/ butcher

aug- 3psg- root abanyenyi [aaɳeɳɳ] /a- a- ɳɛɳi/ / ɳɛɳi -/ butchers

aug- 3pl- root

Adapted from Cammenga (2002)

(49) shows that EkeGusii root/base form is of either {-CV} or {V-CV-} form. While the form {CV-} obeys the Onset syllabic constraint, the {V-}CV- form violates it, even if it is the realized form (Prince and Smolensky, 1993).

126 4.1.5.2.5 Classes 3 {omo-} and 4 {eme-}

Nouns in these classes are those within the meaning of events or periods, trees and parts of the body. (50) shows noun roots within these classes.

50) EkeGusii noun roots in classes 3 and 4

(i) Class 3

Noun surface Underlying Root Gloss

form form form

Omote [omote] /omo-te/ [te] tree

Omotwe [Omotwe] /o-mo-tue/ [-twe] head

Omogondo [omoγondo] /o-mo-γoondo/ [-γondo] garden ii) Class 4 emete [emete] /eme-te/ [e] trees emetwe [emetwe] /eme-twe/ [twe] heads emegondo [emeγondo] /e-me-γoondo/ [-γoondo] gardens

Adapted from Cammenga (2002)

The morphological behavior of these classes, that is 3 and 4, is just like that of classes 1 and 2, in which case they are in their singular and plural forms respectively. Cammenga

(2002) observes that the semantic motivation of classes 1 and 2 still appears to be somewhat stronger in present day EkeGusii as compared to the other classes. This indeed is a correct observation because the nouns in classes 1 and 2 almost solely deal with animate humans in singular and plural forms respectively. However, some animate human beings such as the the physically and mentally challenged, more often are taken to other classes:

127 such as 7{eke-} as in in eke-rema ‘lame person’ and 8{ebi- } in ebi-rema ‘lame persons’;

5{ri- } as in ri-tiino ‘dumb person’ and 6{ama-} as in ama-tiino ‘dumb persons’.

4.1.5.2.6 Classes 9 {e-}, 10 {chi-}, 9 (a) {en-} , 10 (a) {chin-}

Nouns in these classes are those within the meaning of animals, some events, places, and objects. (51) gives noun class roots of these classes.

51) Noun class roots for classes, 9, 10, 9 (a) and 10 (a)

Class 9 {e-}

Noun Surface underlying root gloss form form form esese [esese] /e-sese/ [-sese] dog etaaro [etaaro] /e-taaro) [-taro] journey ebuunda [eβuunda] /e-βuunda/ [-βuunda] donkey

Class 10 {chin-} chisese [t∫isese] / t∫i-sese/ [-sese] dogs chitaaro [t∫itaaro] / t∫i-taaro/ [-taaro] journeys chibuunda [t∫iβuunda] / t∫i-βuunda] [-buunda] donkeys

The roots in these classes like those in Class 9 are similar in form. However, while those in class 9 carry the singular form, those in class 10 carry the plural meaning.

Class 9a {en-} embata[embata] /e-n-βaata/ [βaata] duck endangera [endagera] /e-n-raaγera/ [raaγera] food embori [embori] /e-n-βoori/ [βoori] goat

Class 10a {chin-}

[t∫imbaata] / t∫i-n-βaata/ [βaata] ducks [endaaγera] /chi-n-raaγera/ [raaγera] foods 128 [emboori] /e-n-βoori/ [βoori] goat

Adapted from Cammenga (2002)

These data show that while the outputs (surface forms) of the roots in classes 9 and 10 are generally similar to their inputs (underlying forms), at least in structure and morphological features, those in classes 9a and 10a are not. While the outputs of these classes (9a and 10a) have voiced obstruents [b], [d] and [g], their input roots have [β], [r], and [γ] respectively, which are voiced fricatives. This is due to nasal homorganicity and voicing dissimilation explained earlier on. What this means is that EkeGusii language does not have the voiced obstruents. They only emerge at the surface as prenasals due to phonological conditioning.

4.1.5.2.7 Classes 5 {eri-/rii-}, 6 {ama-}, 7 {eke-}, 8 {ebi-}, 12 {aka-} and 14 {obo-}

These classes are marked by combination of corresponding singular and plural prefixes as in (52).

52) EkeGusii classes 5, 6, 7,8,12 and 14 prefixes

Singular plural

(a) 7 eke - 8 eβi-

12 akas - 8 eβi-

12 aka- 14 oβo-

(b) 5 eri- 6 ama-

Source: Cammenga (2002)

These singular/ plural pairing of the given prefixes is explained as follows: Firstly, a word from another class entering in any of the classes in (52a) gets the meaning of diminution besides its basic meaning, while when such a word is transferred to the classes in (52b), at least the idea of augmentation is added to its basic meaning.

129 Secondly, as has already been observed, prefixes in (52a), except that of class 12, function as regular class prefixes. This is in addition to marking diminution and augmentation just described. This, according to Cammenga (2002), underlines the fact that it is the particular combination of singular and corresponding plural prefix that constitutes some class and determines its meaning. Prefix class 12 {aka-} is the one exception since it expresses diminutive meaning only. This double function of prefixes in classes 5, 6, 7, 8, 12 and 14 entails that words belonging to a class marked by any of them cannot be transferred to the class to which it already belongs regularly in order to express diminution or augmentation.

Diminution, according to Cammenga (2002) can be achieved through class transference, with pejorative connotation, non-pejorative diminution through adjectival modification, and the expression of degrees of pejorativeness through a combination of these two means, with or without an added adverb are all exemplified in (53) as follows:

53) EkeGusii diminution by prefixation a) Diminution

/ o- mo- oNto o- mo- ke / / a- βa- anto a- ba- ke/ aug.- 1 - person aug- 1 - small aug.- 2 - person aug- 2 - small

[omoonto ͻmͻkɛ] [aβaanto aβakɛ]

‘a small person’ ‘small persons’ b) Pejorative or non-pejorative in increasing degrees.

/ a- ka- mo- onto / aug.- 12- 1 - person

[akamoonto]

‘a small person’ (pejorative or non-pejorative). 130 The plural ([oβomonto]) is always pejorative

a- ka- mo- onto a- ka- ke aug.-12- 1- person aug. 12- small

[akamoonto aγake]

‘a very small person’ c) Pejorative in increasing degrees e- ke- mo -onto e- βi- mo- nto aug.- 7 - 1 person aug. 8- 1 person

[ekemoonto] [eβimoonto]

‘small person’ ‘small persons’

/e- ke- mo- oNto e- ke- ke/ aug-7 – 1 – person aug – 7 – small

[ekemoonto eγeke]

‘a very small person’ – pejorative

Plural (eβimoonto eβike)

/e- ke – mo- onto e-ke-ke -mono/ aug-7 – 1 – person a-7 – small- very

[ekemoonto eγeke mono]

A very very small person’ (pejorative)

Plural [eβimonto eβike mono]

(54) shows examples of EkeGusii pre-prefixation or augmentation.

131 54) EkeGusii augmentation a) non – pejorative

/o- mo- onto o- mo- nene/ /a- ba- anto- a – ba- nene) aug. - 1- person aug.-1 - big aug.-2 -person – aug.-2- big

[omoonto omonene] [abaanto abanene]

‘a big person, ‘big persons’

(b) non – pejorative or more usually pejorative. /rii – mo-Nto/ /a- ma-mo-Nto/ 5 – 1 – person aug.-6-1 persons

[riimoonto] [amamoonto] rii- here means ‘big’ just like ma – c) abusive

*/rii- ke-mo-Nto/ */a -ma-ke-mo-Nto/

5- 7- 1 – person aug.- 6- 7 – 1- person

*[riikemoonto] * [amakemoonto] d) pejorative

/rii-mo-oNto rii – nene/ /a – ma- mo-oNto- a- manene/ 5 1 peson 5 – big aug- 6 – 1 person aug – big [riimoonto riinene] [amamoonto amanene]

‘very big person’ ‘very big persons’

Adapted from Cammenga (2002: 206-7)

In (54c) under augmentation, the form *[riikemoonto] is not acceptable in the view of the researcher who s a native speaker. The prefix stacking which brings in the prefix {ke-} of class 7 does not seem to add any meaning to the whole structure of the word. In fact, the

132 class 7 prefix brings in a meaning of small, so that the structure could mean *’big small person’, which, in view of this study, does not sound correct.

Classification and sub-classification of the English noun is different from that of EkeGusii.

Classification of nouns in EkeGusii is determined by the prefix which is in turn controlled by the semantics of the noun in question. This is not the case in English.

4.2 Phonological nativization of English nouns in EkeGusii

As mentioned in chapter 1 in this study, each natural language has its own structural system upon which the words are built. An arguments based on universal grammar provides that languages have certain basic properties that they tend to share. However, as Massamba

(1991) correctly observes, it is quite unlikely that any two languages share exactly the same structural forms (phonology and morphology). In other words, in addition to the universal grammar (UG) properties shared by all grammars, each grammar has some peculiar sequential constraints. This section deals with objective two of the study that analyses the phonological changes that English nouns borrowed into EkeGusii undergo during nativization. It focuses on the phonological features that EkeGusii and English grammars do not share and how the English phonological system is adjusted so that it conforms to the phonological constraints of EkeGusii grammar. Analyses in this section and indeed the next one (4.3) are carried out within the standard Optimality Theory (McCarthy & Prince, 1993;

Prince & Smolensky, 1993/2004) and the data analyzed are those which were gathered in the field (English nouns in EkeGusii). The loaned nouns are carefully and critically examined for purposes of realizing their phonological changes and how the changes can be accounted for within Optimality Theory perspectives.

133 Nativization of English nouns borrowed into EkeGusii at the phonological level is basically governed by the syllable structure of EkeGusii. This is to say that a loaned noun normally violates some constraint(s) of syllable well formedness in the target language in the process of nativization. In other words, the loaned noun avoids the syllabic structure of the source language in order for it to be accommodated in the target language. It is this avoidance that leads to conformity, because the foreign structure is avoided at the expense of the native one, hence nativization. For example, many languages avoid cluster consonants and onsets.

Other phonological features and processes besides the syllable structure also determine nativization. Phonological nativization in this study is analyzed under four broad headings:

Segmental, phonotactic, prosodic and phonological processes. Under segmental nativization (4.2.1), the focus is on the consonants and the vowels; phonotactic nativization focuses on syllable structure in (4.2.2); prosodic nativization (4.2.3) focuses on tone; while phonological processes nativization (4.2.4), focuses on a number of processes.

4.2.1 Segmental nativization

According to Sapir (1964), and Zivenge (2009), languages are loosely similar, that is, they have slightly different inventories, with some similarity. In the same way, there is some loose similarity between English and EkeGusii languages. A number of phonemes found in the English noun are not found in EkeGusii phonological system. However, this does not mean that there are no similarities at all between the two phonological systems. In other words, in as much as there are English phonemes not found in EkeGusii phonological structure, there are some phonemes found in both languages (Anyona, 2011). In order for the English phonemes to be accommodated in the new EkeGusii phonological environment, two approaches were employed by the speakers: substitution and deletion.

134 This study considered the former because it was the most common approach the speakers adopted. This is discussed under nativization of vowel phonemes in 4.2.1.1 and consonant phoneme nativization in 4.2.1.2 respectively.

4.2.1.1 Nativization of vowel segments

EkeGusii has a vowel system that is different from that of English in the same way consonants of the two languages differ. However, the vowel difference between the two languages is more pronounced as compared to that of consonants. This is probably because as Anyona (2011) points out, English has more vowels as compared to EkeGusii language.

Anyona points out that unlike EkeGusii language, which has only pure vowels or monophthongs, sometimes characterized by length; English has diphthongs and triphthongs as well, besides having more monophthong vowels comparatively. Therefore, there are many English vowels that are not found in EkeGusii phonology. Thus, most of the ‘excess’ vowels from English are collapsed into the few EkeGusii vowels. In other words, while

English has twenty- five vowels (Cruntenden, 2011; O’Connor, 2011; and Roach, 1983 among others), EkeGusii has fourteen as has already been observed in this study. Sub- section 4.2.1.1.1 analyzes nativization of English pure vowels, 4.2.1.1.2 with English diphthongs, while 4.2.1.1.3 analyzes thriphtongs.

4.2.1.1.1 Nativization of English pure (monophthong) vowels

These are those vowels which when produced, the tongue remains constant in that it does not glide. This sub-section shows how these vowels are integrated into EkeGusii phonology.

135 Nativization of English /ɪ/ into EkeGusii [i] ( /ɪ/→[i]) The English vowel / ɪ / shares almost similar features with the EkeGusii vowel [i]. They are both,

[+HIGH, -ROUND AND –BACK]. However, while the English /ɪ/ is [+LAX], EkeGusii [i] is

[–LAX], which explains why they are acoustically different as illustrated by tables (3) and (4) in section (4.1.1.) above. The English vowel /ɪ/ therefore, was realized as EkeGusii [i] as in (55).

This is in addition to other phonological changes. The substituted vowels are in bold.

(55) Nativization of English /ɪ/ to EkeGusii [i] English word Pronunciation EkeGusii nativized form pronunciation Christmas /krɪmǝs/ ekirisimasi [ekirisimasi] guitar /gɪtɑ:/ egiita [egiita] kitchen /kɪt∫ǝn/ ekicheni [ekit∫ɛni/ cabbage /kǝbɪdʒ/ ekabichi [ekait∫i] The realization in (55) is expected because the two vowels are closely related in terms of phonological features as has already been observed. They are [+FRONT, +HIGH, AND –

ROUND], differing only in [LAXNESS]; while [ɪ] is completely laxed, [i] on the other hand is slightly more tensed, though not as much as the long [i:] (O’Connor, 1967;

Cruttenden, 2011). In fact, the vowel [i] is present in both EkeGusii and English (Anyona,

2011; Cammenga, 2002), differing only in their degree of tenseness during production.

This is further supported by the acoustic differences between the vowels. To demonstrate how the English pure vowel/ ɪ/ in (55) was substituted for by the EkeGusii vowel [i], the word kirisimasi /kirisimasi/ ‘chrismas’ is presented in figure (26). Øk r ɪ s m ǝ s Ø English k Ø r Ø s Øm Ø s EkeGusii

k r i s m a s Phonemic substitution

e k i r i s i m a s i (Nativization: Borrowing complete) Figure (26): Substitution of the English /ɪ/ for EkeGusii [i]. Structural presentation adapted from Gussenhoven & Jacobs (2011) Figure (26) shows that the English front, short, high, and unrounded vowel /ɪ / is substituted for the EkeGusii front, high, tense and unrounded [i] vowel. That is, while [i] is

136 tense, /ɪ/ is lax. This phenomenon is explained by the fact that the EkeGusii phonological system does not have the lax front high vowel /ɪ/; but both vowels, that is, ([ɪ] and [i]), share many common features: they are [+High], [-Back] and [-Round] (Cruttenden, 2011;

Roach, 1983; O’Connor. 1967). This is in agreement with Kang (2011) who argues that a foreign input containing a segment absent in the target language necessitates the replacement of the foreign segment by the closest sound in the target language. Kang gives the example of the adaptation of the French high, front, rounded vowel [y] as [u] (which has the rounding and high qualities) in White Hmong as discussed by Golston and Yang

(2001). In essence, the realization of the English [ɪ] as EkeGusii [i] involves phonetic featural changes. Thus, EkeGusii prefers tense vowels to lax ones. The occurrence in figure (27) is against Optimality Theory’s markedness constraint, *TENSE (V), which prefers lax vowels to tense ones (McCarthy, 2007) . The realization therefore bans lax vowels- *LAX (V). The different realizations of the vowels in the input (English) and output (EkeGusii), imply the faithfulness constraint, IDENT IO (FEATURE), which demands that an input feature must also be in the output, no change (Kager 1999). Since change is allowed at the expense of having tense vowels, the markedness constraint dominates the faithfulness one, resulting to the ranking argument; *LAX (V) >> IDENT IO (FEATURE). OT differentiates languages on the basis of hierarchical ranking of universal constraints and not on language particular recursive rules of early generative theories. The ranking and re- ranking of constraints in this study used the Tesar and Smolensky (1993) algorithm model, which provides that given surface forms of the borrowing language (in this case EkeGusii nativized forms from English) and a set of universal constraints, it is possible to discover the correct ranking of the target language. In this model, it is assumed that an input, that is

137 the form from which the output derives, is provided (the English forms in the case of this study) and that the output is the phonologically structured representation and not a raw phonetic form (in this case the EkeGusii nativized forms from English collected from the field). Given that the initial state of the algorithm is one in which all constraints are unranked with respect to one another, that is all are undominated, the algorithm employs the principle of constraint demotion in ranking and reranking of the universal constraints in a language specific manner. Using the constraints given above, English and EkeGusii realizations of the word

‘Christmas’, /krɪmǝs/ and ekirisimasi, [ekirisimasi] respectively are analyzed in tableaux

(1) and (2) respectively. English realization. Input: /krɪsmǝs/ This realization, and indeed all the realizations in which vowels of the target language

(EkeGusii) are substituted for those of the source language (English), will rank the constraints given above as follows. IDENT IO (FEATURE/PLACE) V >> *LAX (V) which means that IDENT IO

(FEATURE) VOWEL is ranked higher and therefore dominates *LAX (V). Thus, IDENT

IO (FEATURE) VOWEL plays an important role in determining the optimal candidate in

English. In all the ranking arguments and how optimal candidates (winning candidates) in this study are established and illustrated, violation tableaux are used (McCarthy, 2007;

2008). This is because the goal of this study is to establish or select the optimal candidate in the given constraint ranking. Following this therefore, the English realization above uses violation tableau (1) to establish the ranking argument and demonstrate how the optimal candidate competitively emerges.

138 Input: / krɪsmǝs / IDENT IO (FEATURE) (V) *LAX (V) a. ☞ [krɪ.smǝs] *

b. [ki.ri.sima.si] *!

Tableau (4.1): English realization of the input / krɪsmǝs/ The winning candidate here is (a). It satisfies the faithfulness constraint, IDENT IO

(FEATURE) (V), which is highly ranked in English. Its violation of *LAX (V) is not fatal, since English allows it. Candidate (b) loses because it violate the highly ranked constraint in the language; that is, IDENT IO (FEATURE) (V). This is comparable to EkeGusii output of the same word in tableau (2) below. Input: /kirisimasi/ ‘Christmas’ This realization re-ranks the constraints as follows: *LAX (V) >> IDENT IO (FEATURE) (V). Thus, it reverses the ranking. The realization is analyzed in tableau (2).

Input: /kirisimasi/ *LAX (V) IDENT IO (FEATURE) (V) a. [krɪ.smǝs] *!

b.☞[ki.ri.si.ma.si] *

Tableau (4.2): EkeGusii realization of the Input /kirisimasi/ In this tableau, the optimal candidate is (b). This is irrespective of the fact that the candidate violates the faithfulness constraint, IDENT IO (FEATURE) (V) as illustrated by the tableau. The faithfulness constraint is dominated by the markedness one in EkeGusii unlike in English. On the contrary, candidate (a) loses because it violates a highly ranked constraint, *LAX (V), which disallows lax vowels. This, in Optimality theory terms, is a fatal violation. The realization of the English /ɪ/ as [i] in English nouns borrowed into EkeGusii is not a peculiarly EkeGusii phenomenon. English loans in White Hmong language spoken in in

Southern Chaina behave the same as illustrated by (56) below.

139 56) White Hmong nativization of English lax vowel /ɪ/ English word pronunciation Hmong realization McKinley /mǝkkɪnli/ [mekiŋli] Mitsubishi /mɪtsǝbi∫i/ [mi∫imbi∫i]. Adapted from: Goldstone and Yang (2001) (56) shows that the English lax vowel /ɪ/ (bold) is realized as White Hmong tense [i] (bold).

Golstone’s and Yang’s conclusion that short vowels not found in Hmong are borrowed into the language as the vowel that is closest to them in terms of features such as height, rounding and backness, seems to be the case in this study. Indeed, all the English lax vowels entering EkeGusii were generally tensed as shown by data set (55) above. [advise on how to handle delete or leave]

Nativization of English /æ/ into EkeGusii [a] ( /æ/→[a]) The vowels [a] and [æ] differ only in one respect; while the English /æ/ is lax, EkeGusii [a] is tensed accoustically. They are the same in all other aspects, they are: [front, non- rounded, low].

The English vowel /æ/ is realized as [a] by EkeGusii speakers as in (57). 57) Nativization of English /æ/ to EkeGusii [a] English noun pronunciation EkeGusii nativized form pronunciation taxi /tæksi/ etagisi [etaγisi] glass /glᴂs / ekerasi [ekerasi] tank /tᴂŋk/ etanki [etaŋgi] bathroom /bᴂðru:m/ ebaturumu [eaturumu] In (57), the English vowel /æ/ is realized as [a] in EkeGusii. These two vowels share phonetic and phonological features as has already been shown. This explains why the speakers substitute one for the other. The only difference which is responsible for their phonemic status is the phonetic feature [TENSE]; [a] is [+TENSE], while /æ/ is [–TENSE]. There are three possible explanations for the occurrence in (57). The first one is phonetic as explained by Yip (2002). This provides that since /æ/ and [a] both have a lowered jaw in their production, the speakers find [a] a better perceptual match for the English /æ/ since

140 the muscles of the speakers are used to this production. The second explanation, which is equally phonetic and closely related to the first one, is acoustic. EkeGusii, unlike English, does not allow lax vowels, thus English /æ/, which is lax is realized as [a], which is tensed in EkeGusii (see section4.1.1) above. The third explanation, which is visual, is that of orthographic influence. In this case, as Peperkamp (2006) observes, adaptations reflect the way native speakers are used to reading of foreign graphemes. According to Peperkamp,

French children learn to pronounce English graphemes as their native sounds. This is illustrated in (58) as adapted from Peperkamp (2006). 58) Realization of English graphemes by French children English Grapheme French Realization (Pronunciation) Example of word /œ/ but book As a result of this, Peperkamp observes that French adult speakers are likely to base their adaptations of English words on these between language grapheme to phoneme correspondence. Both the phonetic and perceptual explanations seem to influence the realization of the

English vowel /æ/ as [a] in EkeGusii besides closeness in terms of phonological features discussed in sub section 4.1.1 above. (57) above indeed shows that all the noun loans, the vowel /æ/ is realized as [a]. Optimality Theory account of this realization is the same as that discussed in section 4.2.1.1.1 above.

Nativization of English /ᴧ/ into EkeGusii [a] ( /ᴧ/→[a]) The vowels/ᴧ/ and [a] are characterized by similar feature values. They are both [-BACK] and [-HIGH]. But while the English /ᴧ/, which is absent in EkeGusii phonology, is

[+LAX], EkeGusii [a] is [–TENSE]. It is the phonetic similarity and difference that makes it possible for the realizations witnessed in (59).

59) Nativization of English /ᴧ / to EkeGusii [a] 141 English noun pronunciation EkeGusii nativized form pronunciation cut /kᴧt/ ekati [ekati] brush /brʌ∫/ eburasi [eurasi] cupboard /kʌbǝd/ ekabati [ekaati] pump /pʌmp/ epambu [epambu] In (59), the English vowel /ᴧ/ is realized as [a] by EkeGusii speakers. In fact, this is one of the vowels which did not provide much pronunciation challenge to the speakers. This is perhaps because the two sounds are produced by almost the same part of the tongue and their degree of tongue height is almost similar as illustrated by chart (6).

Front central back High

Mid ʌ Low a

Chart (6): English /ʌ/ and EkeGusii /a/ Chart (6) shows that both sounds, that is; /ʌ/ and /a/, are [+front], [+low], and [-rond].

Thus, the sounds share more phonetic features values than they differ.

Nativization of English /ɜ:/ and /ǝ/ into EkeGusii [a] ( / ɜ: /, /ǝ:/ →[a]) The vowels English /ɜ:/ and /ǝ/ are characterized by the phonetic feature values [+tense,

-round, -low], -front], while the EkeGusii vowel [a] is characterized by: [+tense, -round,

+low, front]. Both the English and EkeGusii vowels share two features ([+tense, –round]), which perhaps, together with perceptual closeness, determines the substitutions that occur as illustrated by (60) and (61).

(60) Nativization of English /ɜ:/ and to EkeGusii [a] English noun pronunciation EkeGusii nativized form pronunciation skirt /skɜ:t/ esikati [esikati] shirt /∫ɜ:t/ esati [esati] breakfast /brekfɜ:st/ burekibasiti [urekiasiti nurse /nɜ:s/ omonasi [omonasi] 142 (61) Nativization of English / ǝ:/ and to EkeGusii [a] English noun pronunciation EkeGusii nativized form pronunciation father /fɑ:ðǝ/ omobaata [omoaata] christmas /krɪmǝs/ ekirisimasi [ekirisimasi] pastor /pɑ:stǝ/ omobasita [omoasita] computer /kǝmpju:tǝ/ ekombiuta [ekompjuta] As (60) and (61) show, the English /ɜ/ and /ǝ/ are realized as EkeGusii [a] (in bold). This is a common phenomenon in loan word nativization. For example, Dholuo, a Nilotic language nativizes the two central English vowels, which are absent in its phonology, to [a] (Owino,

2003). Bantu languages like Tonga and Kalanga, spoken in Zimbabwe, and Botswana respectively, (Zivenge, 2009; Chebanne and Phili, 2015), like EkeGusii, also substitute the

English /ɜ/ and /ǝ/ for [a]. Language family does not seem to determine the substitution; rather the absence of the vowels in the borrowing languages. One feature value that the

English vowels do not share with the vowel it is substituted for in EkeGusii and the other languages, that is [a], is [+ CENTRAL]. This is a marked feature value, because many

African languages avoid it at the expense of either [FRONT]. Theoretically therefore, the realizations of /a/ instead of /ɜ:/ and /ǝ/ in (60) and (61) respectively presuppose the markedness constraint *CENTRL (V), which prohibits central vowels; but the change of the feature values violates the faithfulness constraint IDENT IO (F). Tableaux (3) and (4) ranks and re-ranks the constraints of English and EkeGusii realization of the English word shirt /∫ɜ:t/, for example. English input: /∫ɜ:t /

Input: /∫ɜ:t / IDENT IO (F) V *CENTRAL (V) a. ☞ [∫ɜ:t] *

b. esati *!

Tableau (4.3): English realization of the input /∫ɜ:t / 143 EkeGusii input: [esati]

Input: /esati / *CENTRAL (V) IDENT IO (FEATURE) (V) a. [∫ɜ:t] *!

b. ☞ [esati]

Tableau (4.4): EkeGusii realization of the input /esati/ Re-ranking of the given constraints yields different outputs. When the ranking is such that the markedness constraint *CENTRAL (V) dominates the faithfulness one, IDENT IO (F)

(V), that is *CENTRAL (V) >> IDENT IO (F) (V) as in tableau (4), EkeGusii output results. The opposite is true when the faithfulness constraint dominates the markedness constraint as in tableau (3). Thus, English tolerates the given markedness constraint as compared to EkeGusii and other African languages. The interpretation of tableau (3) for the English output is that candidate (a) is the output because it obeys the faithfulness constraint which is ranked higher in English as compared to EkeGusii. The markedness constraint on the other hand dominates the faithfulness constraint in tableau (3) to enable candidate (b) to be the output.

Nativization of English / ɑ: / into EkeGusii [a] ( / ɑ: /→[a/ aa]) These vowels share the feature values [+low, +tense, and -round]. They differ in that while the English /ɑ:/ is [-FRONT], EkeGusii [a] is [+FRONT]. The choice of [a] as a substitute therefore is expected because the two vowels share many feature values than they differ.

The substitution of /ɑ:/, which is [-FRONT], for [a] which is [+FRONT], presupposes the markedness constraint *BACK (V), which prohibits back vowels; they, especially [+LOW] ones are marked (Kager, 1999). Thus, as (62) indicates, all cases of the English /ɑ:/ coming into EkeGusii phonology were realized as either [a] or [aa], which is [+FRONT].

144 (62) Nativization of English /ɑ:/ to EkeGusii [aa] English noun pronunciation EkeGusii nativized form pronunciation car /kɑ:/ ekaa [ekaa] card /kɑ:d/ ekati [ekaati] glass /glɑ:s/ ekerasi [kerasi] garage /gærɑ:ʒ/ egarachi [γarat∫i] In (62), the open, low, back, tense English vowel /ɑ:/ is realized as EkeGusii [aa/a]. This is because EkeGusii does not have the English vowel /ɑ:/ in its phonological inventory and most importantly, /ɑ:/, a low back vowel is marked. Closer orthographic perception also plays a role. The markedness feature which is the main determinant of the substitution presupposes the markedness constraint *BACK (V), which bans back vowels. This constraint in turn means that the faithfulness constraint which demands that input and output features be the same (IDENT IO (F)), is violated. Thus, the outputs of the English input /ɑ:/ in English and EkeGusii is determined by re-ranking of these constraints as analyzed by tableaux (5) and (6) for the English word glass /glɑ: s/. EkeGusii input: /ekerasi/

Input: /ekerasi/ *BACK (V) IDENT IO (F) V a. [glɑ:s] *!

b.☞ [ekerasi] *

Tableau (4.5): EkeGusii realization of the input /ekerasi/ English input: /glɑ:s/

Input: /glɑ:s/ IDENT IO (F) V *BACK (V)

a. ☞ [glɑ:s] *

b. [ekerasi] *!

Tableau (4.6): English realization of the input /∫ɜ:t /

145 In tableau (5), candidate (b) is the output because it obeys the markedness constraint,

*BACK (V), which bans back vowels. Its violation of the faithfulness constraint is inconsequential because the constraint is lowly ranked in EkeGusii. The reranking of the constraints leads to the analysis in tableau (6). English ranks the faithfulness constraint higher than the markedness constraint, which is why it tolerates the marked feature,

*BACK, which is avoided by EkeGusii. This realization is not peculiar to English nouns borrowed into EkeGusii. Hmong, Golston and Yang (2001); Dholuo, Owino (2003); Tonga, (Zivenge, 2009); and Kalanga, Chebanne and Phili (2015), among others behave the same way. For example, in Dholuo, Owino

(2003), just like in EkeGusii, the vowel /ɑ:/ is realized as Dholuo [a] as in (63).

63) Nativization of English /ɑ:/ to Dholuo [a] English noun pronunciation Dholuo nativized form pronunciation garage /gærɑ:dʒ/ garach [garat∫] glass /glɑ:s/. gilas [gilas] card /kɑ:d/ kadi [kadi] Source: Owino (2003) This realization further confirms the fact that [+LOW] [+BACK] vowels are marked and therefore absent in most languages of the world because they are not easy to learn and produce.

Nativization of English /ɒ/ into EkeGusii [ͻ] ( / ɒ /→[ͻ]) The vowels [ɒ] and [ͻ] are characterized by the value features [+back, +round]. But while the English /ɒ/ is [+LOW], EkeGusii [ͻ]) is [-LOW]. Thus, this is the feature which determines the substitution of the [+LOW] vowel for the [-LOW] one. As observed by

Kager (1999), [+LOW, +BACK] vowels are marked, and therefore avoided by most languages. English /ɒ/ is avoided in EkeGusii as in (64).

146 (64) Nativization of English /ɒ/ to EkeGusii [ͻ]. English noun pronunciation EkeGusii nativized form pronunciation Cotton /cɒtn/ ekotini [ɛkͻtoni] Box /bɒks/ epogisi [ɛpͻγisi] Bolt / bɒlt/ eboriti [ɛͻriti] In (64), the English short back rounded English vowel /ɒ/ is realized as EkeGusii [ͻ].

Tableau analysis of this realization is the same as those of the realization in (62) above because it is the same markedness constraint involved in both cases, that is *BACK V.

Nativization of English / ɪ / into EkeGusii [e] or [ɛ ] ( / ɪ/→[e] or [ɛ]) These vowels are characterized by the following phonetic feature values: [-BACK,

-ROUND, -TENSE]. The only feature which distinguishes the two vowels is [high]; while the English /ɪ/ is [+HIGH], EkeGusii [e] is [-HIGH]. This is perhaps one of the reasons behind the realization of the English /ɪ/ as EkeGusii [e] or [ɛ] as in (65).

(65) Nativization of English /ɪ/ to EkeGusii [e] English noun pronunciation EkeGusii nativized form pronunciation college /kɒlɪdʒ/ ekorechi [ɛkͻrɛt∫i/ senate /sɪneɪt/ eseneti [seneti] elephant /elɪfǝnt/ erebanti [ɛrɛanti] In (65), the English vowel /ɪ/ is realized as EkeGusii [e] or [ɛ]. The choice between [e] and

[ɛ] is determined by vowel harmony discussed in section 4.1.1.1.1 above. This realization can be given two explanations. The first explanation is that of orthographic influence, which is perceptual or orthographic in nature, (Owino, 2003; Peperkamp, 2006), as discussed in section 4.1.1.1.2 above. The orthographic system of the vowels and not the feature values of the vowels dictate the pronunciation of the English /ɪ/ as EkeGusii [e] or

[ɛ]. For example, in college, the letter influences the realization of [ɛ], which is closely related to [e] in terms of phonetic features, and not the English /ɪ/, which is neither in the

EkeGusii orthography nor closer featurally to the [ɛ]. The second explanation is phonetic. It has already been observed in this section that the vowels are more similar phonetically than 147 they differ; they differ only in terms of [height], while [ɪ] is [+high]; [e] and [ɛ] are [-high].

Markedness has it that high vowels are more marked as compared to low vowels

(Trubetzkoy, 1969), thus, [ɪ] is more marked and therefore less natural than [ɛ]/[e].

Therefore, it is easier to produce [e]/ [ɛ] as compared to [[ɪ]. The realization in (65), like that of other vowels discussed so far, show a change of feature values between the input and output forms of the English vowel /ɪ/. Thus, in OT theoretic terms, faithfulness constraints such as IDENT IO (F) V and markedness constraint

*ASSIM (F), which prohibits assimilation of features in a given domain, are presupposed.

Thus, while English demands that the vowel in the input must be preserved in the output,

EkeGusii demands that vowels be assimilated. This results in different ranking of the constraints as demonstrated by analyses of the English and EkeGusii outputs of the word senate /sɪneɪt/ for example, in tableaux (7) and (8) below respectively. English realization Input: /sɪneɪt/ ‘senate’ Constraints ranking: IDENT IO (F) V >> *ASSIM (F)

Input: /sɪneɪt/ *ASSIM (F) IDENT IO (F) V a. ☞ [sɪneɪt]

b. [eseneti] *!

Tableau (4.7): English realization of the input /sɪneɪt/ EkeGusii realization Input: /eseneti/ This realization is presented in tableau (8) Constraint ranking: IDENT IO (F) >>*ASSIM (F)

Input: /eseneti/ IDENT IO (V) *ASSIM (F)

a. [sɪneɪt] *!

148 b. ☞ [eseneti] * **

Tableau (4.8): EkeGusii realization of the input /eseneti/ In tableau (7), the optimal candidate is (a) because it does not violate the constraint

*ASSIM (F), which is the highest ranked, while in tableau (8) candidate [b] wins because it satisfies the constraint IDENT IO (V), which is banned in EkeGusii. Other languages, for example, Dholuo, Owino (2003) also sometimes nativize the English vowel /ɪ/, to [e] as demonstrated by (66). 66) Dholuo nativization of English /ɪ/ to [e] English noun pronunciation Dholuo nativized form pronunciation mission /mɪ∫n/ misen [misen] television /telɪvɪ∫n/ telefison [telefison] elephant /elɪfǝnt/ elefant [elefant] In (66), the English vowel /ɪ/ is realized as [e] in Dholuo, just like in EkeGusii as shown in

(65). Just like in EkeGusii, this vowel is not present in Dholuo phonology(Owino, 2003). So far, under the section of English pure vowel nativization, it has been realized that in

EkeGusii loaned words from English, the central vowels /ɜ: , ǝ, ʌ /, are substituted for

EkeGusii low, front vowels [a] as illustrated in figure (27).

ɜ:

ǝ [a]

ʌ Figure (27): EkeGusii nativization of the English central vowels Source:Bright (1970: 123) These realizations, according to Owino (2003), can be attributed to phonetic factors. Owino observes that, on account of restricted physiological space associated with the lower region of the oral cavity, it can be assumed that the articulatory and auditory properties of the low vowels occurring in English are minimally differentiated. At the same time, Dholuo operates on a single low vowel, /a/. This vowel can be considered a natural rendering of the central vowels found in English. This is in fact what characterized the English central 149 vowels and indeed most other vowels upon coming into EkeGusii phonology. In other words, some of the English vowels entering EkeGusii, like those entering Dholuo, are collapsed into the EkeGusii [a], and the other few EkeGusii vowels. This of course is dependent on the phonological closeness (in terms of features) between the target and the source language and sometimes the hardness with which the incoming vowel is produced; that is its markedness status. The integration of English vowels into the vowels of EkeGusii is further illustrated by (82).

67) English vowel realization in EkeGusii English vowel EkeGusii realization /i:, ɪ/ [i] /ʌ,æ, ǝ, ɑ:/ [e, ɛ] /e, æ, ɜ:/ [e] /u, ʊ/ [u] /ͻ: ɒ/ [o, ͻ] Source: Anyona (2011) (67) shows that all the English vowels are collapsed into the seven EkeGusii vowels. This is in agreement with Bright (1970) who observes that African languages collapse English vowels into those present in their phonologies as in (68). 68) English pure vowels against their approximate African vowel phonemes English vowel Approximate African languages vowel.

/i:, ɪ / [i] /ʌ, ǝ, ɜ:, ɑ:/ [a] /e, æ / [e] /u, ʊ/ [u] /ͻ ǝʊ / [o] Adapted from Bright (1970) Cases of vowels of source languages being substituted for those of the target languages as in the case of this study are common (Golstone &Yang, 2001; Owino, 2003; Zivenge,

2009; Hussain, 2011; Kang, 2011,and Chebanne & Phili, 2015) All these studies, like the present one point to the fact that vowels and indeed sounds of the source language change

150 to or are substituted for those of the target language when they are not present in in the phonologies of the target languages. For example, as (56) above attests, the English vowel /ɪ/ is substituted for the White Hmong [i], just like in EkeGusii. The difference between White Hmong and EkeGusii studies being that while Hmong is a Miao-Yiao language, which is isolating and largely monosyllabic, spoken in Southern Chaina

(Golstone and Yang, 2001), EkeGusii on the other hand is a Bantu language, which is fairly polysyllabic and agglutinative, spoken in Kenya. The fact that the two languages share the given phonological phenomenon is illuminating.

It means that the shared feature would be regarded as a universal tendency. This observation is further supported by the fact that findings in the other studies mentioned above point to the same direction irrespective of the fact that some are accounted for by different theoretical perspectives, while others are not anchored on any theoretical perspectives. Golstone & Yang (2001) and this study are anchored on Optimality Theory, a constraint based generative theory, while Owino (2003) and Zivenge (2009) are accounted for within rule based generative theories. Hussain (2011) and Chebanne & Phili (2015) on the other hand are not anchored on any theoretical framework, yet the results of all the studies are the same. The findings of all these studies point to the universal nature of the given phonological occurrence; that is, the substitution of foreign language vowels not present in the target language for those present in the target language.

4.2.1.1.2 Nativization of English diphthongs

A diphthong, according to Roach (1983), is a vowel containing two vowels pronounced as one, gliding from one to the next in rapid succession. EkeGusii, unlike English does not have diphthongs. EkeGusii vowels which follow one another in a word, like those in other

151 Bantu languages are not realized as single units forming syllable nuclei in English. Thus, such vowels in Bantu get realized as two distinct vowels belonging to two successive syllables instead of forming the nucleus of a single syllable as is the case in English

(Chebanne and Phili, 2015). Many cases of diphthong nativization are realized as single vowels in the borrowed words as observed under the section of phonological processes nativization in this study. The common occurrence is that English diphthongs are either substituted for either by a single phoneme vowel (monophthongization) or lose its second element and lengthen the first element. This, in Optimality theory suggests the following constraints: IDENT IO (F), a faithfulness constraint, which demands that features of an input segment must be preserved in the output, no feature change, and *COMPLEX V, a markedness constraint, which bans complex vowels and MAX IO, another faithfulness constraint, which demands that input segments must have output correspondents, deletion of segments is disallowed. This subsection shows how the English diphthongs were realized in EkeGusii and how the realizations are accounted for within Optimality Theory.

Nativization of English /ǝʊ/ into EkeGusii [o] ( /ǝʊ/→[o:]) The diphthong /ǝʊ/ is made up of two pure vowels; the first of which being mid central and non-rounded, while the second is high back and rounded. The EkeGusii [o:] on the other hand is a monophthong, with the feature values [+ MID] and [+ ROUND]. It has one feature from each of the vowels of the English diphthong it substitutes: [+MID] from [ǝ], and [+round] from [ʊ]. This probably explains why the diphthong is substituted for the monophthong in EkeGusii realizations as in (69) below.

152 69) English /əʊ/ nativized to EkeGusii [o:] English noun pronunciation EkeGusii nativized form pronunciation Location /lǝʊkeɪ∫n/ erookeseni [ero: keseni] Sofa /sǝʊfǝ/ esooba [eso:a] Cocoa /kɒkǝʊ/ ekooko [eko:ko:] Pawpaw /pǝʊpǝʊ/ ripoopo [ripo:po:] Radio /reɪdiǝʊ/ ereetio: [ereetjo:] In (69), the English diphthong /ǝʊ/, is realized as the EkeGusii back mid rounded vowel

[o]. This is achieved through the process of vowel coalescence which is a common process

affecting vowel nativization (Zivenge, 2009; Owino, 2003; Chebanne & Phili, 2015).

Orthographic influence, according to Golstone & Yang (2001) and Hussain (2011), is

another reason behind this realization. In other words, orthography makes speakers focus

on the grapheme rather than the actual sound. The change noted here is that of coalescence

which in essence leads to the substitution of [o] for /ǝʊ/ in EkeGusii. This occurrence

presupposes the OT constraints given above in the given ranking: *COMPLEX V >>

{MAX OI, IDENT IO (F)}. This is because the realization of the diphthong, which leads to

monophthongization, satisfies the constraint *COMPLEX V (which is highly ranked in

EkeGusii) at the expense of violating the constraints MAX OI and IDENT IO (F) (both of

which relatively lowly ranked in the language). To account for realizations in (84), the

EkeGusii nativized word form [erokeseni] ‘location’, for example is presented in tableau

(9).

EkeGusii input: /ero: keseni/ Constraint ranking: *COMPLEX V >> {MAX OI, IDENT IO}

Input: / ero:keseni / *COMPLEX V MAX OI IDENT IO (F)

153 a. [lǝʊkeɪ∫n] *! * *

b. ☞ [ero:keseni] * *

Tableau (4.9): EKeGusii realization of the input /erokeseni/ This tableau shows that the optimal candidate is (b) even though it violates two relatively low ranked constraints in EkeGusii grammar, that is by adding of new segments and changing of the features of segments in the loanword. These violations however are not as serious as the violation of maintaining complex vowels in the loan. Therefore, EkeGusii constraint ranking prevails upon that of English. English output of the same word will be analyzed as in tableau (10). Input: /lǝʊkeɪ∫n/ location/ Output: [lǝʊkeɪ∫n] Constraint ranking: {IDENT IO(F), MAX OI} >> *COMPLEX (V)

Input: /lǝʊkeɪ∫n/ IDENT IO(F) MAX OI (F) *COMPLEX (V)

a. [lǝʊkeɪ∫n] * *

b. ☞ [ero:keseni] *!

Tableau (4.10): English realization of the input /lǝʊkeɪ∫n/ Realization of the English diphthong /ǝʊ/ as EkeGusii [o:] in not peculiar to EkeGusii loans from English only. Other languages treat the diphthong the same way. For example, the diphthong is nativized as [o:] in Urdu and Punjabi loans from English as illustrated by (70). 70) Urdu and Punjabi nativization of the English diphthong /ǝʊ/ a) Urdu realization English noun pronunciation Urdu realization pronunciation Hotel /hǝʊtǝl/ hootel [ho:tǝl] Coach /kǝʊt∫/ kooch [ko:t∫] b) Punjabi realization English noun pronunciation Urdu realization pronunciation Road /rǝʊd/ rood [ro:d] Coach kǝʊt∫ kooch [ko:t∫]

154 (70) shows that in both Urdu and Punjabi, the English diphthong /ǝʊ/ is substituted for [o], just like in EkeGusii. Hussain (2011) attributes this realization to the fact that Urdu and

Punjabi phonological systems lack the diphthong, meaning that it will be substituted for that which is in the target language that is closest in terms of features. This is indeed the situation in this study. EkeGusii phonological inventory lacks the diphthong [ǝʊ] which leads for its substitution for [oo], which as has already been observed is phonetically closer to the diphthong. Other languages with similar results include: Dholuo, Owino (2003),

KiKamba, Mutua (2007), Tonga, Zivenge (2009) and Kalanga, Chebanne & Phili (2015) among others. These studies differ with the present one in two crucial ways: some employ different theoretical approaches (Owino, 2003 & Zivenge, 2009), Chebanne & Phili (2015) like Hussain (2011) does not employ any theory; while Mutua (2007), like the present study employs Optimality Theory. These studies focused on different languages.

Nativization of English /aɪ/ into EkeGusii [ae] ( /aɪ/→[ae]) This is one of the cases where an English diphthong is substituted for an EkeGusii one. The diphthongs are characterized by the same initial element, that is [a], which is [+LOW]. The second elements; [ɪ] for English and [e] for EkeGusii, though different, share most feature values. The features are: [-LOW, -ROUND, -BACK]. This is perhaps the reason why the second elements are substitutable. [ɪ] becomes [e] because, for one, it is not present in

EkeGusii phonology. Secondly, it is the closest vowel to [e] in terms of phonetic features as has already been observed. (71] gives cases of realizations of English /aɪ/ as EkeGusii [ae].

71) Nativization of English /aɪ/ to EkeGusii [ae] English noun pronunciation EkeGusii nativized form pronunciation file /faɪl/ ebaeri [eaeri] tile /taɪl/ etaeri [etaeri] mile /maɪl emaeri [emaeri] 155 styile /stail/ esitaeri [esitaeri] In (71), the English diphthong /aɪ/ is realized as [ae] in EkeGusii nativized forms. In producing the diphthong /aɪ/ in English, the gliding begins with an open vowel which is low central [a] and moves upwards to the high front position of [ɪ]. It is one vowel, which in the given monosyllabic words serves as the peak of the syllable. Its nativized form however, does not follow this form of gliding. Instead, the tip of the tongue is lowered to the front mid close position of [e] as illustrated by charts (7) and (8).

ɪ

a Chart (7): Production of the English diphthong [aɪ] (Adapted from Roach, 1983:20)

e

a

Chart (8): Production of EkeGusii nativized form [ae] from the English diphthong [aɪ] Adapted from Roach (1983: 20) The forms realized in (71) are phonetically distant from the possible source form [ai], not only in in terms of gliding, but also syllabically. While in the source language, the combination is a diphthong, in target language the resulting combination constitute of two separate monophthongs; each in its own syllable. Thus, the diphthong is monophthongized.

For example, [etaeri] from English /taɪl/ ‘tile’, there is no diphthong in EkeGusii realization [e.ta.e.ri.]; instead, the English diphthong is split into two parts,each part forming a syllable of its own. In fact the second part of the English diphthong,[ɪ], is changed to [e]; which forms a sinle syllable in the nativized form.

156 Other languages such as Dholuo, (Owino, 2003) and Kalanga (Chebanne and Phili, 2015) also nativize the English diphthong [aɪ] to [ae]. (72) shows how Dholuo nativizes the

English [aɪ]. 72 Nativization of English /aɪ/ into Dholuo [ae] English noun pronunciation Dholuo nativized form pronunciation File faɪl fael fael Style staɪl stael stael Mile maɪl mael mael tie taɪl tael tael Source: Owino (2003) In Dholuo, unlike in EkeGusii and indeed other African languages, as (72) shows, the diphthong is treated as a single unit, like in English and not as distinct vowels in separate syllables. This, as will be discussed under phonotactic nativization, is because EkeGusii, a

Bantu language, unlike Dholuo, a Nilotic language, strictly does not allow codas.

Nativization of English /eɪ/ into EkeGusii [e] (/eɪ/ → [e:]/ [ɛ:]) The first element of the English diphthong and the EkeGusii vowel with which it is substituted are similar phonetically. The nativized form is created by dropping the final element and lengthening the first element, which is present in both phonologies as shown in

(73).

(73) Nativization of English /eɪ/ to EkeGusii [e:] or [ɛ:]

English noun Pronunciation EkeGusii Nativized Form Pronunciation cake /keɪk/ ekeeki [ekɛ:ki]‘cake’ case /keɪs/ ekeesi [eke:si]‘case’ basin /beɪsn/ ebeeseni [ɛɛ:seni]‘basin’ station /steɪ∫n/ esiteseeni [esitese:ni]‘station’

157 In (73), the English diphthong /eɪ/ is realized as [e:] or [ɛ:] in EkeGusii, depending on the vowels of the roots, which harmonises with the rest of the vowels in the word. This involves the deletion of the second element of the diphthong and lengthening the first element, as in /eɪ/ → [e:]; and changing the elements of the diphthong altogether and adding length to the new element as in /eɪ/ → [ɛ:]. These occurrences, which are both phonological and phonetic (LaCharite and Paradis, 2003), are not confined to EkeGusii phonology. Languages, for example, Dholuo (Owino, 2003), KiKamba, Mutua (2007),

Punjabi and Urdu, Haussin (2011), and Kalanga (Chabanne and Phili, 2015), are characterized by the same occurrences. In KiKamba for example, all cases of English /eɪ/ are realized as [e] as in (74). 74) KiKamba realization of the English diphthong /eɪ/ English noun pronunciation KiKamba nativized form pronunciation frame /freɪm/ bulemu [ulemu] crane /kreɪn/ keleni [keleni] crate /kreɪt/ keleti [keleti] Adapted from Mutua (2007)

4.2.1.1.3 Nativization of English triphthongs A triphthong is defined as a vowel made up of three short vowels produced as one

(O’Connor, 1967; Roach, 1983). The three vowels are treated as one because they are treated as a single unit forming the syllable nuclei in the language (Chabanne and Phili,

2015). This subsection shows how these triphthongs are realized in EkeGusii. As observed in section 4.1.1.1, there are five triphthongs in English: /eɪǝ, aɪǝ, ͻɪǝ, aʊǝ, ǝʊǝ/.

Not many loaned words with these triphthongs were realized by the speakers. While most of the diphthongs were realized in very few cases, others were not realized at all. (75) shows how these triphthongs were realized. 75)Nativization of the English triphthongs i) Realization of /ǝɪǝ/ as EkeGusii [aja] English noun pronunciation EkeGusii nativized form pronunciation 158 wire /wǝɪǝ/ egwaya [e-γwaja] wire /fǝɪǝ/ efaya [efaja] ‘ whoir /kwǝɪǝ/ ekwaya [ekwaja] ii) Realization of /eɪǝ/ as EkeGusii [aja] English noun pronunciation EkeGusii nativized form pronunciation player /pleɪǝ/ epureya [epureja] layer /leɪǝ/ ereya [ereja] iii) Realization of /ͻɪǝ/ as EkeGusii [aja] English noun pronunciation EkeGusii nativized form pronunciation employer /emplͻɪǝ/ eemburoya [eemburoja] iv) Realization of /aʊǝ/ and /ǝʊǝ/ as EkeGusii [awa] English noun pronunciation EkeGusii/ nativized form pronunciation flower /flaʊǝ/ eburawa [eurawa] shower /∫ǝʊǝ/ esawa [esawa] (75) shows that English triphthongs with the middle vowel being the front close vowel /ɪ/ are substituted for the form [aja] in EkeGusii, while those which have the middle vowel being the back, high close vowel /ʊ/ are substituted for EkeGusii [awa]. Thus, the English triphthongal realization is lost in both cases. Instead, an approximant is introduced to replace the middle element of the triphthong, thus creating an extra syllable. The introduced approximant is determined by the backness and or the roundness of the vowel.

The round vowel /ʊ/, is replaced by the labial approximant [w], while the non-rounded vowel /ɪ/ is replaced by the palatal approximant [j]. Both approximants, unlike the vowels they replace, are present in EkeGusii phonology. They are respectively closely related phonetically to the vowels they replace. This explains why the approximants are chosen during nativization. Figure (28) for the English word /wǝɪǝ/ ‘wire’ for example illustrates how triphthongs are handled by EkeGusii. ØØ w ǝ ɪ ǝ English

ØØ ØØØØ EkeGusii ØØw a j a Phonemic substitution

e γ w a j a (Nativization: Borrowing complete) Figure (28): Substitution of the English /ǝɪǝ/ for EkeGusii [aja]

159 This figure shows that the English triphthong /ǝɪǝ/ is realized as EkeGusii /aja/, in which the English short vowel /ǝ/ is replaced with EkeGusii /a/ and the English vowel /ɪ/ is replaced by the semi consonant /j/. EkeGusii phonology, like many other phonologies, does not have any triphthongal glide. Thus any triphthong that comes into it is likely to take a different form as it does in (75). Again, the phonotactics of EkeGusii does not allow any form of vowel clusters. The vowel clusters of three as in the words in (91) have to be declusterized to the acceptable phonotactic form as will be discussed under phonotactics below. Nativization of the English triphthong by vowel declusterization through a replacement of the medial vowel of the triphthong by an approximant is a common phenomenon. For example, in Dholuo (Owino, 2003), nativization of the English triphthong /ǝɪǝ/ behaves exactly the same way as in EkeGusii, even though the two languages are from totally different families (EkeGusii is Bantu, while Dholuo is Nilotic) as illustrated by (76). 76) Dholuo nativization of the English triphthong /wǝɪǝ/ English noun Pronunciation Urdu Realization Pronunciation wire wǝɪǝ gwaya [waja] choir kwǝɪǝ kwaya [kwaja] (76) shows that the triphthong is done away with by introducing a glide which takes the position of the medial vowel. This further leads to resyllabification of the word, changing from being monosyllabic to disyllabic. This is indeed what happens to the diphthong in

EkeGusii nativization. However, not all languages nativize the English triphthong by declusterization through approximant introduction. Urdu and Punjabi (Hussain, 2011) for example, nativize the

English triphthong /ǝɪǝ/ by substituting it for [ae], a diphthong as shown in (77). 77) Substitution of English /aɪǝ/ for Urdu and Punjabi [ae] i)Urdu English noun pronunciation Urdu realization Pronunciation fire /faɪǝ/ faer [faer] wires /waɪǝs/ waeles [waeles] 160 ii) Punjabi English noun pronunciation Punjabi realization pronunciation fire /faɪǝ/ faer [faer] diary /daɪǝrɪ/ daer [daer] Adapted from Hussain (2011) In (77) Urdu and Punjabi nativize the English triphthong /aɪǝ/ as [ae]. The first element in the triphthong is maintained, the medial element deleted, while the final element is changed from a mid-front vowel to a low front vowel. These changes are different from those observed in EkeGusii and Dholuo nativization of the triphthong. In these cases, the initial and the final elements of the triphthong are maintained while the medial element is replaced with an approximant. All these occur, as Owino (2003) observes, in order to create an acceptable syllable structure (to be discussed in detail under nativization by resyllabification). In the case of EkeGusii and Dholuo for example, the introduction of an approximant to replace a vowel is intended to break the complex vowel that is not acceptable in grammar of the languages, and in the process, an extra syllable is created.

Punjabi and Urdu seem to tolerate a complex vowel, a diphthong, but not a triphthong. It deletes the last element which seems to be replaced with the alveolar tap [r], which closes the syllable. Thus, Punjabi and Urdu tolerate syllable codas. In Optimality Theory perspective, the realizations witnessed in (75)– (77) are accounted for by the changes that take place. As it has already been observed above, English triphthongs are realized differently in EkeGusii loaned words, like in other language loaned words.

There is loss of the English triphthongal status due to the replacement of the middle vowel by an approximant, which creates an extra syllable. This change presupposes the following

OT constraints: IDENT IO (F), which demands that features of an input segment must be preserved in the output, no feature change, *COMPLEX V, which demands that complex vowels are not allowed and MAX IO, which demands that output segments must have input

161 correspondents, no segment addition. These constraints are ranked differently depending on the output required. EkeGusii does not allow complex vowels or clusters, therefore, it ranks the markedness constraint higher than the faithfulness constraints. Thus its ranking is:

*COMPLEX V >> {MAX IO, IDENT IO (F)}. English on the other hand tolerates complex or vowel clusters, meaning that the markedness constraint is dominated by the faithfulness constraints. Thus, the constraints are reranked as follows: {IDENT IO (F),

MAX IO} >> *COMPLEX. Given these rankings, analyses of EkeGusii and English realizations of the English triphthong /aɪǝ/ for example, are given in tableaux (11) and (12) respectively. EkeGusii realization EkeGusii input: /waja/

Input: /waja/ *COMPLEX V MAX OI IDENT IO (F) a. [waɪǝ] *!

b. ☞ [waja] * *

Tableau (4.11): Ekegusii realization of the input /waja/ In this tableau, candidate (a) loses to candidate (b) because (a) disobeys the higher ranked constraint by allowing a complex vowel, which is banned in this language. candidate (b) on the other hand wins because it obeys the determining constraint. English realization Input: /waɪǝ/ ‘wire’ Output: /waɪǝ/

Input: /waɪǝ/ IDENT IO (F) MAX OI *COMPLEX V a. ☞ [waɪǝ] *

b. [waja] *! *

Tableau (4.12): English realization of the input /waɪǝ/

162 In tableau (12), candidate (a) is the output because it obeys the highest ranked constraint as compared to candidate (b) which violates the determining constraint.

In essence, nativization of English diphthongs and tripthongs in EkeGusii loaned nouns from English generally involve monophthongization. Monophthongization does not only affect EkeGusii loans from English, but other languages too. For example, languages such as Kalanga spoken in Botswana, (Chebanne & Phili,2015); KiKamba, (Mutua, 2013);

Tonga spoken in Zimbabwe, (Zivenge, 2009); and Dholuo spoken in Kenya, (Owino,

2003) among others, behave the same way, that is, they get monophthongized. For example, in KiKamba, like EkeGusii, Kalanga and Tonga (all Bantu), there are no diphthongs or triphthongs in the strict sense of English. KiKamba therefore, like these other Bantu languages monophthongizes any diphthong and triphthong that enters into its phonology from English as in (78).

78) Monopthongization of diphthongs and triphthongs in KiKamba

Diphthong English Word Pronunciation KiKamba Realization Pronunciation

/eɪ/ frame /freɪm/ vulemu [ulemu] /aɪ/ bicycle /baɪskl/ vasikili [asikili] /iǝʊ/ radio /rediǝʊ/ letiu [letio] aʊǝ towel /taʊǝwǝl/ taulo [taulo] Adapted from Mutua (2013) (78) shows that English diphthongs and triphthongs are realized as monophthongs in

KiKamba. This is how they are treated in EkeGusii as discussed above. This is expected because KiKamba like EkeGusii is a Bantu language and therefore share common phonological features. However, analysis of Dholuo, a non Bantu language show that

English diphthongs and triphthongs are equally monophthongized (Owino, 2003). (79) demonstrates this observation.

163 79) Dholuo monophthongization and triphthongization

Diphthong/triphthong English noun pronunciation Dholuo realization pronunciation

/ǝʊ/ coat /k ǝʊt/ koti [koti] /eɪ/ grade /greɪd/ giredi [giredi] /aʊ/ scout /skaʊt/ sikaot [sikaot] /aɪ/ file /faɪl/ fael [fael] /ɪa/ gear /gɪa/ giya [gija] /ǝɪǝ/ wire /wǝɪǝ/ waya [waja] Adapted from Owino (2003) Monophthongization of English diphthongs and triphthongs in (79) mean that monophthongization is not confined to Bantu languages only, but rather that all those languages without them, irrespective of their language families.

4.2.1.2 Nativization of English consonants

Nativization consonants in English nouns borrowed from EkeGusii is done by replacing or substituting the English consonant segments not present in EkeGusii phonological system.

A number of consonant segments found in the phonological system of English do not exist in EkeGusii phonology. These include: /f/, /v/, /l/, /Ө/, /ð/, /ʒ/, /∫/, /h/. As it has already been observed in this study, the voiced plosives /d/, /g/, and /b/ only occur with nasals homorganistically in EkeGusii and are effectively regarded as pre-nasals. It has also been observed that the plosive /p/ is only found in one or two idiophones, according to Whiteley

(1960). Cammenga (2002) suggests that this plosive is “increasingly noticeable in the speech of the younger generation which has had contact with Swahili and English.” This study argues in favour of the fact that /p/ is a rare sound in EkeGusii and that if all the younger generation of the 1960s, when Whitely conducted his research on the language were using it, it was only in nativized words from the languages mentioned by Whiteley.

This section shows how the English consonants (listed above) not found in EkeGusii

164 phonological structure are realized by EkeGusii speakers. The realizations are accounted for within Optimality Theory perspectives.

As it has already been mentioned, nativization of English consonants in EkeGusii involves change or substitution of English consonants for those of EkeGusii. This in Optimality

Theory means that there is a violation of a faithfulness constraint - IDENT IO (SEG) C or

IDENT IO(F) or IDENT IO (P) and IDENT IO (VOICE), which demands that an input consonant segment or feature specification of a segment or place of articulation of a segment and voice of a segment must have an output correspondent respectively. This demand however would satisfy a number of markedness constraints such as: *VOI which prohibits voiced obstruents like [g d z], *VTV which bans voiceless obstruents in intervocalic positions as in [*ota] versus [oda] McCarthy (2007), AGREE (VOICE),

AGRREE (CONTINUANT), *SRIDENT (FRIC) which prohibits noisy fricatives

(Laparombara 2013). The markedness constraint therefore would be ranked higher than the faithfulness constraint in these realizations because, faithfulness constraints will be violated at their expense. Thus, markedness constraints will dominate faithfulness ones:

MARKEDNESS >> FAITHFULNESS. This is the ranking which prevails in the realizations of English consonants loaned into EkeGusii. Re-ranking of the constraints so that the faithfulness constraints dominate the markedness ones give opposite results. That is, English outputs are realized.

4.2.1.2.1 Nativization of English /f/ and /v/ The English labiodental fricatives share the labial feature value with EkeGusii [β], which they are substituted for. The feature value that separate the English consonants from that of

EkeGusii are: while /f, v/ are [+DENTAL], and [+ STRIDENT]; [β] is [–DENTAL] and [-

165 STRIDENT]. It is the labial feature which they share which probably make them substitutable as shown in (80).

80) Realization of English /f/ and/v/ by EkeGusii [β], [ɸ] English noun pronunciation EkeGusii nativized form pronunciation verandah /vǝrændǝ/ ebaranda [eβaranda] tv /ti:vi:/ etibii [etiβii] fashion /fæ∫n/ ebaasoni [eβaasoni] fridge /frɪdʒ/ eburichi [eurit∫i]

(80) shows that the English voiced labiodental fricatives /v/ and its voiceless counterpart /f/ are realized as [β], a voiced bilabial fricative in EkeGusii. The possible explanation to this is that EkeGusii phonology lacks these anterior strident labiodental fricatives and therefore the anterior non strident bilabial fricative substitutes them because of phonetic similarity between them. They share [+LABIALITY] and [+CONTINUANCY], differing majorly in terms of stridency. Therefore, the realization of /v/ and /f/ as [β] is as a result of phonetic similarity. Other languages nativize the English /v/ and /f/ in a similar manner. In Dholuo for example, /v/ becomes [b] (Owino, 2003). This is even when the phonetic distance between the two obstruents seem to be more than it is in EkeGusii. The consonant substitution process that takes place in data set (80) is illustrated by figure (29) which considers nativization of the English noun fridge /frɪdʒ/ to EkeGusii eburichi [eurit∫i].

Øf Ø r ɪ dʒØ English ØØØ r Ø ØØ EkeGusii

Ø Ø r i t∫Ø Phonemic substitution

e u r i t∫ i (Nativization: Borrowing complete) Figure (29): Substitution of the English /f/ for EkeGusii [β] This figure shows that the English consonant /f/ is substituted for the EkeGusii consonant [].

166 In OT, this realization would be accounted for by the fact the English segment /f/, a labiodental strident fricative changes to [],a bilabial non-strident fricative in EkeGusii segment. This suggests that a segment in the output will not be faithful to its input form, presupposing the faithfulness constraint IDENT IO (SEG) C. It also presupposes that strident fricatives are not allowed in outputs, yielding the markedness constraint

*[STRIDENT] (FRIC). Thus, the markedness constraint will therefore be ranked higher than the faithfulness one in EkeGusii outputs because faithfulness is inconsequential in determining the optimal candidate in the language. The opposite ranking of the constraints as shown in tableaux (13) and (14), yields English outputs. EkeGusii realization of English /f/ EkeGusii input: [eβurit∫i] ‘fridge’ Constraint ranking: *[STRIDENT] (FRIC) >> IDENT IO (SEG) C

Input: /eβurit∫i/ *[STRIDENT] (FRIC) IDENT IO (SEG) C

a) frɪdʒ *!

b) ☞ eβurit∫i *

Tableau (4.13): EkeGusii realization of the input /eβurit∫i/ Candidate (b) is the output because it violates a less consequential candidate in EkeGusii in a bid to satisfy the demands of the highly ranked constraint, *[STRIDENT] (FRIC), which prohibits strident fricatives in EkeGusii. Candidate (a) violates this highly ranked constraint in the language, the reason why it loses. This is compared to the English realization of the same word as follows. English input: /frɪdʒ/ ‘fridge’ Constraint ranking: IDENT IO (SEG) C >> *[STRIDENT] (FRIC)

Input: / frɪdʒ/ IDENT IO (SEG) C *[STRIDENT] (FRIC)

167 a) ☞ frɪdʒ *

b) eβurit∫i *!

Tableau (4.14): English realization of the input /frɪdʒ/ In this tableau, candidate (a) is the output because it obeys the faithfulness constraint

IDENT IO (SEG) C, which is ranked higher than the markedness constraint in English. An alternative explanation to the realizations given in (85) above can be given. In Tonga,

(Zivenge, 2009), the English voiceless labiodental fricative /f/ is realized as [v], a strident labiodental just like /f/. This is irrespective of the fact that the voiceless labiodental is present in both the phonology of English and Tonga. Zivenge attributes this occurrence to the fact that the sound is constrained in terms of occurrence to onsets of the last syllable of class 7 nouns in Tonga. This occurrence, however, can be given another interpretation. In

EkeGusii, in which as data set (96) shows, this labiodental fricative is realized as the voiced bilabial fricative [], while in KiKamba loans from English, Mutua (2013), it is realized as

[], a voiceless bilabial fricative. In White Hmong loans from English, Golston and Yang

(2001), the fricative is maintained, irrespective of the position it occupies in a word. While it is agreeable that the labiodental takes different realizations in loanwords from

English, depending on the phonology of the host languages, it can be argued that whichever segment they substitute in the target language, is normally voiced intervocalically and voiceless in voiceless environments. In Tonga (Zivenge, 2009), the English word scarf /ska: f/ is realized as sikava [sikava]. In

EkeGusii, as (96) shows, the word fridge /frɪdʒ/ is realized as eburichi [eurit∫i]. These show that the two languages realize the sound /f/ differently intervocalically; [v] in Tonga, and [] in EkeGusii. Besides these sounds sharing labiality and frication, they are voiced.

Given their environment of occurrence in the target languages ( are intervocalic) and their 168 voiced nature, the *VTV markedness constraint in Optimality Theory (McCarthy 2003), is presupposed. This constraint prohibits voiceless consonants in intervocalic positions. The realizations would also assume a number of faithfulness constraints such as IDENT IO

(SEG) C, which demands that input consonant segments must have output correspondents.

Thus, the markedness constraint dominates the faithfulness one in the realization of The

English /f/ in the above examples. The following tableau for the English input scarf /ska: f/ in Tonga realization illustrates this observation. Tonga input: [sikava] Constraint ranking in Tonga: *VTV >> IDENT IO (SEG) C

Input: /skava/ *VTV IDENT IO (SEG) C

a) skafu *!

b) ☞ sikava *

Tableau (4.15): Tonga realization of the input /skava/ Candidate (a) in the tableau is optimal because it satisfies the highest ranked constraint

(*VTV). (a) loses because it violates the constraint. Reranking the given constraints will yield English output of the given word since the languages rank constraints differently.

4.2.1.2.2 Nativization of English /ð/ and // The consonants [Ө] and [ð] are characterized by the feature values [+interdental], [+ continuant], [+coronal], and [+ anterior]. They are distinguished by the feature [voice].

While [Ө], is voiceless, [ð] is voiced. The consonant [t], with which the given English consonants substitute, share a number of feature values such as: [+anterior], and

[+coronal], which determine its choice for substitution. EkeGusii does not have interdental

169 fricatives. This is perhaps the reason behind the avoidance of the interdentals, which are marked, in EkeGusii as (81) shows.

81) Realization of the English /Ө/ and /ð/ as EkeGusii [t] English noun pronunciation EkeGusii nativized form pronunciation thermos /Өɜ:məs/ etamosi [tamosi] thief /θi:f/ etiβu [ tiβu] father /f ɜ:ðə/ omoβata [omoβata] The substitution process involved in these realizations is illustrated by figure (30) for the

English noun ‘thermos’ /Өɜ:məs/. ØӨ ɜ: m ə s Ø English ØØ Ø m Ø s Ø EkeGusii

Øt a m o s Ø Phonemic substitution

e t a m o s i (Nativization: Borrowing complete) Figure (30): Substitution of the English /Ө/ for EkeGusii [t] Adapted from Gussenhoven and Jacobs (2011)

This figure shows that the English /Ө/ is substituted for Ekegusii [t]. The voiced equivalent of /Ө/; that is, /ð/ is also substituted for [t]. This is explained as follows. Firstly, Ekegusii does not have the two interdental fricatives and therefore the nearest consonant, in terms of feature values; that is, [t] is the likely choice in this case. As has already been observed, [t] is characterized as an alveolar in EkeGusii (Whitely, 1960 and Cammenga, 2002).

However, as Cammenga (2002: 54) suggests, “though /t/ is characterized as an alveolar, it may be rendered as a voiceless interdental obstruent, possibly also as an alveolo-dental or perhaps dental-alveolar obstruent”. This study takes the position that the sound is an alveolo-dental obstruent and therefore excludes the possibility that it may be an interdental or a dental alveolar obstruent. Thus, [t], therefore is the likely choice as the substitute of the two English interdentals in English words loaned into EkeGusii.

170 Secondly, Ekegusii language lacks voiced consonants (except pre-nasal stops, nasals, and the fricative [ndʒ]) (Cammenga, 2002). This means that chances of the English voiced inter-dental /ð/ being an output in the nativized English loaned words in EkeGusii are minimal, if at all. Given that interdentals are not realized in EkeGusii and that they are not part of the phonological system of the language, a markedness constraint of place of articulation which bans interdentals is proposed; that is, *INTERD (FRIC) (McCarthy, 2003). This means that interdentals are not allowed in this language, in other words they are marked. A survey of literature seem to support this proposal: In Tonga, Zivenge (2009), English /Ө/ is realized as [s] as in theory /Өɪǝrɪ/ → [sijori] , while /ð/ is realized as [dʒ], as in leather

/leðǝ/ → [ledʒa]; in Japanese, Kay (1996), English /Ө/ is realized as [s] as in thrill /Өrɪl/ →

[siriru], among others. In all these realizations, there is avoidance of the English interdentals. Thus, the proposal that the given interdental fricatives, and the interdental position generally are marked is supported. Given that interdentals are avoided, resulting to a change of segment as shown in (97), the faithfulness constraint IDENT IO (SEG) C is violated. Tableaux (16) and (17) below analyses the effect of the ranking and re-ranking of the constraints in English and EkeGusii respectively for the English word /Өɜ:məs/

‘thermos’.

English input: /Өɜ:məs/ ‘thermos’ Constraint ranking in English: IDENT IO (SEG) C >> *[INTERD] (FRIC)

Input: /Өɜ:məs/ IDENT IO (SEG) C *[INTERD] (FRIC) a) ☞Өɜ:məs *

b) tamosi *!

171 c) ðɜ:məs *!

Tableau (4.16): English realization of the input /Өɜ:məs/ In tableau (4.16), (a) is the output because it obeys the faithfulness constraint which ranked higher than the markedness constraint. (b) and (c) on the other hand lose because they violate the constraint. This is comparable to EkeGusii ranking of the constraints as follows. EkeGusii input: /tamosi/ ‘thermos’ Constraint ranking in EkeGusii: *[INTERD] (FRIC) >> IDENT IO (SEG) C

Input: /tamosi/ *[INTERD] (FRIC) IDENT IO (SEG) C

a) Өamosi *!

b) ☞ tamosi *

c) ðɜ:məs *!

Tableau (4.17): EkeGusii realization of the input /tamosi/ This tableau shows that ranking of the markedness constraint over the faithfulness one yields EkeGusii output since EkeGusii ranks the faithfulness constraint lower than the markedness constraint.

4.2.1.2.3 Nativization of English /l/ The English consonant /l/ is realized as [r] in EkeGusii. [l] and [r] are characterized by the following feature values [+ANTERIOR], [CORONAL], among other features. They are distinguished by the feature lateral; while, [l] is [+LATERAL], [r] is [-LATERAL], meaning that it is a trill. EkeGusii phonology does not have the lateral consonant. Given that the two consonants are closely related, in terms of feature values, substituting [l] for [r] is natural as in (82). 82) Substitution of English /l/ by EkeGusii [r] (/l/ → [r]) English noun pronunciation EkeGusii nativized form pronunciation television /tεlεvi∫n/ eterebisoni [tereβisoni] class /klæs/ ekerasi [ekerasi] 172 lunch /lʌndʒ/ ranchi [rant∫i] glucose /glu:kǝʊz/ gurukosi [γurukosi] In (98), the English consonant /l/ is substituted for [r] in EkeGusii. The substitution process involved is illustrated by the English word /klæss/ ‘class’ in figure (31). Ø k  l æ s Ø English

Ø k ØØ Ø s Ø EkeGusii

Øk Ø r a s Ø Phonemic substitution

e k e r a s i (Nativization: Borrowing complete) Figure (31): Substitution of the English /l/ for EkeGusii [r] The consonant /l/ is not found in EkeGusii phonological system. This explains why it is not realized in data (82). Substitution of [r] for the English /l/ by second language speakers is a common phenomenon. For example, in Japanese, Kay (1996), like in EkeGusii, the English

/l/ is realized as [r] as in (83). 83) substitution of English /l/ for Japanese [r] English noun pronunciation Japanese nativized form pronunciation hustle hʌsl hassuru [hasuru] whistle hwɪsl hoissuru [hoisuru] last læst rasuto [rasuto] slip slɪp surippu [suripu] Adapted from Kay (1996) In Japanese, like in EkeGusii, the alveolar lateral [l] is not found in its phonological system

(Kay, 1996). That is why it is substituted for the alveolar trill as in (82 & 83) above, which is closer to it in terms of phonetic features as has already been observed. Some languages however, behave in the exact opposite of what happens in EkeGusii and

Japanese regarding these two segments. In Hawaiian, (Golston & Yang, 2001), KiKamba,

(Mutua, 2007), and Tonga (Zivenda, 2009), among others for example, English /r/ is realized as [l]. Thus, in Hawaiian; /merɪ/ merry → [mele], in Tonga; /ru:lǝ/ ruler → [lula], while in KiKamba; kri:m cream → [kelimo]. In these languages, unlike EkeGusii and

Japanese, [r] is not found in their phonological systems. What these substitutions of the English /l/ for [r] in target or borrowing languages and the other way round mean is that one of these approximants normally occur in a language and 173 not both. In other words, it is normally, only either of them which occurs in a given phonological system and not both. This however does not close out the possibility of some languages having both of these approximants, for example, in Kalanga, (Chebanne and

Phili, 2015), just like in English, both [l] and [r] are present in its phonological system.

That is why the approximants are retained in Kalanga nativized forms from English as in

(84). 84) Kalanga nativization of the English approximants /l/ and /r/ English noun pronunciation Kalanga nativized form pronunciation driver /draɪvǝ/ dirayivara [dirajivara] plastic /plæstɪk/ pulasitiki [pulasitiki] plate /pleɪt/ puleyiti [pulejiti] film /fɪlm/ filimu [filimu]

Adapted from Chebanne & Phili (2015) In this data, the English sounds /r/ and /l/ are retained in Kalanga nativized forms of

English. The explanation that can be given as to why these approximants are substitutable cross linguistically is that they share all but only one phonetic feature; they are produced at the alveolar ridge, they are approximants and voiced. But while [l] is a lateral, [r] is a trill.

The cross linguistic alternation between these segments as observed above is a common phenomenon (John, 1984). Given this kind of alternation in which some languages prefer the lateral consonant, while others the trill in their output forms, a manner of articulation markedness constraint can be presupposed. Such a constraint may read: Languages that have trills do not allow laterals (*LATERAL); languages that have laterals do not allow trills (*TRILLS) (McCarthy, 2003). The constraint *LATERAL therefore bans trills; while

*TRILL disallows laterals in outputs. These markedness constraints presuppose faithfulness constraints such as IDENT IO (SEG) C.

174 EkeGusii which lacks the lateral sound handles the English input /ekerasi/ class as analyzed in tableau (18). Constraint ranking: *LATERAL >> IDENT IO (SEG) C

Input: /ekerasi / *LATERAL IDENT IO (SEG) C

a) clæs *!

b) ☞ekerasi *

c) ekelasi *!

Tableau (4.18): EkeGusii realization of the input /ekerasi / Candidate (b) is the output even when it violates the faithfulness constraint. It’s being optimal is motivated by the constraint *LATERAL, which bans laterals in EkeGusii, and which is ranked higher than the faithfulness constraint. (a) and (c) are not optimal because they violate this constraint. Languages like KiKamba and Tonga among others, that do not have the trill segment [r], would have a different analysis as tableau (19) for KiKamba realization of the input /skulu/

‘school’ shows. Constraint ranking in KiKamba: *TRILL >> IDENT IO (SEG) C

Input: /skulu/ *TRILL IDENT IO (SEG) C

a) sukuru *!

b) ☞ sukulu *

Tableau (4.19): KiKamba realization of the input /skulu/ Candidate (b) is optimal in this tableau because it obeys the constraint that is relatively highly ranked in KiKamba, that is *TRILL, suggesting that KiKamba does not allow trills.

175 4.2.1.2.4 Nativization of English /z/ The common feature values that characterize these consonants include [+OBSTRUENT],

[+CONTINUANT], and [+ANTERIOR], among others. They are distinguished by the feature [VOICE]. While [z] is [+VOICE], [s]is [-VOICE]. EkeGusii does not have the voiced obstruent, which explains why it is not realized in the loaned words from English as in (85). 85) Substitution of English /z/ by EkeGusii [s] English noun pronunciation EkeGusii nativized form pronunciation zero /zirəʊ/ esiro [siro] zone /zͻn/ esoni [soni] maize /meɪz/ ebimeisi [meisi] gas /gᴂz/ egasi [eγasi] The substitution process involved in this realization is illustrated by the English word /zirəʊ/ ‘zero’ in figure (32). Ø z i r ǝʊ English ØØ i r Ø EkeGusii

Øs Ø r o Phonemic substitution

e s i r o (Nativization: Borrowing complete) Figure (32): Substitution of the English /z/ for EkeGusii [s] Substitution of the English /z/ for [s] in loanwords affects other languages too, especially those that lack in their phonological systems. In Kikamba, (Mutua, 2007),for example, fees /fi:z/ → viisi [iisi]. A similar occurrence is witnessed in Dholuo (Owino, 2003) as in

‘gazette’ /gǝszet/ → [gaset]. This is even when the two assimilating languages are from different language families; KiKamba being Bantu, while Dholuo being Nilotic. The substitution of [s] for [z] can be explained by the fact that the segments share all but one feature. They are: [+OBSTRUENT], [+CORONAL}, and [+ANTERIOR] among other features. They are distinguished by the feature voice. While [s] is voiceless, [z] is voiced. EkeGusii realization of [s] for English /z/ presupposes the OT markedness constraint

*VOICEOBS (McCarthy, 2003), which prohibits voiced obstruents. The segmental change assumes the faithfulness constraint IDENT IO (SEG), which is violated at the expense of 176 the markedness constraint, which must not be violated. Therefore, *VOICEOBS dominates

IDENT IO (SEG) in EkeGusii and the other way round in English. This is analyzed in tableaux (20) and (21) for the English input /esiro/ and /zirəʊ/ ‘zero’, for EkeGusii and

English respectively. Constraint ranking in EkeGusii: *VOICEOBS >> IDENT IO (SEG)

Input: /esiro/ *VOICEOBS IDENT IO (SEG) C

a) zirəʊ *!

b) ☞ esiro **

Tableau (4.20): EkeGusii realization of the input /esiro/ Candidate (b) is the output because it obeys *VOICEOBS, which is highly ranked in

EkeGusii. (a) on the other hand violates the constraint and therefore loses. English reranks the given constraints differently from EkeGusii ranking with different results. The reranking of the constraints will give the analysis of tableau (21). Constraint ranking in English: IDENT IO (SEG) >> *VOICEOBS

Input: /zirəʊ/ IDENT IO (SEG) *VOICEOBS

a) esiro *!

b) ☞ zirəʊ

Tableau (4.21): English realization of the input /zirəʊ/ This tableau shows that (b) is the output. English ranks the faithfulness constraint higher than the markedness constraint. Thus, the demand of having corresponding segments between inputs and outputs in English is more serious than allowing different outputs.

English allows many voiced obstruents in its outputs in order to be faithful to the inputs.

177 4.2.1.2.5 Nativization of English /∫/ and / ʒ/ These consonants share the feature values [+CORONAL], and [+CONTINUANT] among others. But while /∫/ and / ʒ/ are [-ANTERIOR], the consonant [s] is [+ANTERIOR]. Thus,

EkeGusii does not allow non-anterior coronal continuants, explaining why they are avoided by speakers as in (86).

86) Realization of English /∫/ as EkeGusii [s]. English noun pronunciation EkeGusii nativized form pronunciation show /∫ǝʊ/ esoo [esoo] shirt /∫ɜ:t/ esati [esati] shock /∫ͻk/ esoki [esͻki] bishop /bɪ∫ͻp/ ebisobu [eisͻu] The substitution process involved in the realizations in these data set is illustrated by figure

(33) for the English word /∫ɜ:t/ ‘esati’. Ø ∫ ɜ:t Ø English

ØØ Ø r Ø EkeGusii

Øs a t Ø Phonemic substitution

e s a t i (Nativization: Borrowing complete) Figure (33): Substitution of the English /∫/ for EkeGusii [s] (186) indicates that the English consonantal segment /∫/ is not realized by EkeGusii speakers. It is not found in EkeGusii phonological system. It is substituted for by the segment [s], which is, in terms of phonetic features, closest to it. They share the features: [+

CORONAL, + CONTINUANT, +CORONAL]. However, while [s] is [+ ANTERIOR], [∫] is [-ANTERIOR]. This points to the fact that the feature [anterior] determines the choice of

[s] and not [∫] in EkeGusii realization. Thus, EkeGusii does not allow [-anterior, +coronal

+continuant] in its outputs. These features combine to form a markedness constraint of the form: *NON-ANTERCORLFRIC (McCarthy, 2003)), which prohibits non anterior coronal fricatives in EkeGusii. Like other markedness constraints, this constraint presupposes the

178 faithfulness constraint IDENT IO (SEG), because it involves a change of an input segment in its output form. This is analyzed in tableaux (22) and (23) for the inputs /esati/ and /∫ɜ:t/

‘shirt’ for EkeGusii and English respectively. EkeGusii realization Constraint ranking: *NON-ANTERCORLFRIC >> IDENT IO (SEG)

Input: /esati / *NON-ANTERCORLFRIC IDENT IO (SEG)

a) ∫ɜ:t *!

b) ☞ esati *

Tableau (4.22): EkeGusii realization of the input /esati/ English Realization Constraint ranking: IDENT IO (SEG) >> *NON-ANTERCORLFRIC

Input: /∫ɜ:t / IDENT IO (SEG) *NON-ANTERCORLFRIC

a) ☞∫ɜ:t *

b) esati *!

Tableau (4.23): English output and input of /∫ɜ:t/ In tableau (22), the output is candidate (b) because it obeys the relatively high ranked constraint in EkeGusii. The reranking of the constraints in English yields a different output in tableau (23), which obeys the faithfulness constraint. These analyses point to the fact that non anterior coronal continuants are marked. This observation is supported by similar realization of the English /∫/ loaned words in other languages such as in Dholuo as in (87). 87) Nativization of English /∫/ in Dholuo English noun pronunciation EkeGusii nativized form pronunciation shirt /∫ɜ:t/ sati [sati] fashion /fæ∫n/ fason [fason] bishop /bɪ∫ͻp/ pisopu [pisopu] 179 Source: Owino (2003) In (87), the English consonant /∫/ is realized as [s] in Dholuo. Thus, like in EkeGusii, the consonant is equally marked in Dholuo.

4.2.1.2.6 Nativization of English /g/ in EkeGusii The consonants [g] and [γ] are closely related in terms of feature values. They are: [-

CORONAL], [+ VELAR AND [-ANTERIOR]. They are distinguished by the feature

[CONTINUANT]. While [γ] is [+CONTINUANT], [g] is [-continuant]. This explains why

EkeGusii realizes the English /g/ as [γ] as in (88). 88) Nativization of English /g/ as EkeGusii [γ] English noun pronunciation EkeGusii nativized form pronunciation bag /bᴂg/ epagi [epaγi] glue /glu:/ eguru [eγuruu] goal /gǝʊl/ egori [eγori group /gru:p/ egurubu [eγuruu] In (88), the English velar non-continuant /g/ is substituted for the voiced EkeGusii Velar fricative [γ] in order to be dissimilar in terms of features with the consonant in the adjacent syllable; that is [p]. While [γ] is [+ CONTINUANT] and [+VOICE]; [p] is [-

CONTINUANT] and [-VOICE]. Thus, Dahl’s Law of feature dissimilation described elsewhere in this study, which affects EkeGusii is in operation here. Figure (34) for the

English word /bᴂg/ ‘bag’ is an illustration of this observation. Ø bæ g Ø English ØØ ØØØ EkeGusii

Øp a γ Ø Phonemic substitution

e p a γ i (Nativization: Borrowing complete) Figure (34): Substitution of the English /g/ for EkeGusii [γ] In figure (34), the English voiced velar non continuant /g/ is substituted for by EkeGusii

[γ], a voiced velar continuant. In essence, the stop is fricativized. This, in Optimality

Theory, presupposes that in EkeGusii, assimilation of certain phonemic features such as continuancy and voice are not allowed in a word. This is feature dissimilation, which in OT

180 can be put as a markedness constraint that disallows assimilation of features in a domain such as a syllable or a word; that is *ASSIM (F) (McCarthy, 2002). This constraint presupposes that phonemes in the output forms of a word will not be the same in their input correspondents, the result being a faithfulness constraint such as FAITH (F). Using these constraints, the realization of the English word /bæg/ ‘bag’ is analyzed in tableaux (24) and

(24) for EkeGusii and English rankings respectively. EkeGusii input: /epeγi/ ‘bag’ constraint ranking: *[ASSIM] (F) >> FAITH (F)

Input: /epaγi/ *[ASSIM] (F) FAITH (F)

a) bæg *!

b) ☞ epaγi ***

Tableau (4.24): EkeGusii realization of the input /epaγi/ In this tableau, candidate (b) is optimal because it does not assimilate features across the syllables of the word. In candidate (a) which loses, there is assimilation of the features voice and stop in the word. This is not allowed in EkeGusii. English input: /bæg/ ‘bag’ Constraint Ranking: FAITH (F) >> *[ASSIM] (F)

Input: /bæg/ FAITH (F) *[ASSIM] (F)

a) ☞ bæg *

b) epaγi *!

Tableau (4.25): English realization of the input of /bæg/ Candidate (a) is the output because English ranks the faithfulness constraint higher than the markedness one, which disallows feature assimilation.

181 Besides the kind of substitution of the consonants described so far, there is also substitution of the voiceless English pre-nasal consonants /k/ and /s/ with their voiced equivalents in

EkeGusii, that is; /g/and /z/ respectively and substitution of the voiceless English stop /k/ with EkeGusii voiced fricative [γ]. This is Dahl’s law in operation (Bickmore, 1998). This process is discussed in detail under phonological processes in sub-section 4.2.4.1.

4.2.2 Phonotactic nativization In section 4.2.1 above, segmental nativization of English nouns in EkeGusii, which is one of the areas of focus in phonological nativization, has been analzed and discussed. In this section, the second area, Phonotactic nativization is analyzed and discussed. Phonotactic nativization in this study focuses on the syllable structures of the two languages in contact.

Kager (1999:91) observes that the syllable is a useful unit of phonological analysis; it is in the syllable that the rules that govern the well formedness sequences of segments in a language, consonants and vowels, are defined. Kagar (1999) observation that the syllable controls processes of insertion and deletion of segments, where either a cluster is not required or a consonant/vowel is not required word initially or finally; and that the syllable links segments to higher- level (suprasegmental) units, forms the basis of analyses in this section. The study shows syllable differences between EkeGusii and English and how the English syllable adjusts syllabically to conform to the phonotactic patterns of EkeGusii, the target language. In particular, the study shows how epenthesis breaks consonant clusters on syllable margins (onsets and codas), making the syllables open in cases where the incoming words have a closed syllable. It also shows how deletion helps to achieve similar results (that is, avoidance of closed syllables).

182 As was observed in subsection 4.1.3, EkeGusii is a (V) CV syllable structure, while English

on the other hand is a (C) C) (C) V (C) (C) (C) (C) language. This is illustrated by (89) .

89) EkeGusii and English syllable structures

a) EkeGusii syllable structure

syllable structure noun pronunciation gloss

CV mama /ma.ma/ mother V CV omote /o.mo.te/ tree

b) English syllable structure

Syllable structure word pronunciation V owe /ǝʊ / CVC take /teɪk/ CCVCC stains /steɪns/ CCCVC structure /strʌkt∫ǝ/ CVCCC facts /fᴂkts/ CCVCCCC glimpsed /glɪmpst/ These syllable types can be presented on syllable nodes as figures (35) and (36)

demonstrate. i) CV as in /ma.ma/ ‘mother’ σ σ

O N O N

C V C V

m a m a

ii) VCV as in /o.mo.te/ ‘tree’ σ σ σ

N O N O N

V C V C V

o m o t e

183 Figures (35): EkeGusii syllable types i) V as in /ǝʊ / ‘owe’

σ

N

V

ǝʊ ii) CCVCC as in /steɪns/ ‘stains’ σ

O N C

CC V CC

s t eɪ nz

Figures (36): English syllable types

The figures in (35) show that EkeGusii syllables do not allow complex margins. All margins, that is, onsets and codas are simple. English syllables as shown by figures in (36), on the other hand allow complex margins, both in the onset and coda positions.

The syllable types in (89) as illustrated in figures (35) and (36) presuppose a number of OT markedness constraints: ONSET, which demands that syllables must have onsets, *CODA, which disallows closed syllables, *COMPLEX C, which demands that syllables must not have complex margins or that syllables must not have a cluster of consonants (McCarthy,

2003). They also presuppose the faithfulness constraints: MAX IO, which provides that input segments must have output correspondents, no deletion of a segment, and DEP IO

(SEG), which provides that no segment should be added to the output; that is no epenthesis.

What differentiates the two syllable typologies in this study is the different ranking of the constraints by the languages. In (105a) for example, ONSET is ranked higher than DEP IO 184 (SEG), which is presented as ONSET >> DEP-IO. Tableau (26) for the EkeGusii input

/ma. ma/ [mama] ‘mother ‘illustrates.

Input :/ma.ma/

Constraint ranking: ONSET >> DEP-IO

Input: /ma.ma/ ONSET DEP-IO a) ☞ ma.ma b) a. ama *! *! * c) m.aa *! * Tableau (4.26): EkeGusii realization of the input /ma.ma/

In this tableau, candidate (a) is optimal because it violates neither of the constraints.

Outputs (b) and (c) lose because they violate the highly ranked constraint in the language:

ONSET. Thus, a markedness constraint dominates a faithfulness one.

The structures in (89b) on the hand allow the violation of ONSET in (i), NOCODA in (ii) and *COMPLEX in (iii). These constraints would be ranked as follows: (i) MAX IO (SEG)

>> ONSET, (ii) MAX IO (SEG) >> NO CODA, (iii) MAX- I0 (SEG) >> *COMPLEX.

This is analyzed in tableaux (27-29). i) Input: /ǝʊ/ ‘owe’

Constraint ranking in English: DEP-I0 >> ONSET

Input: /ǝʊ / MAX IO(SEG) ONSET a) ☞ ǝʊ * b) tǝʊ *! c) ǝʊǝ *! *

Tableau (4.27): English realization of the input /ǝʊ/ 185 ii) Input / teɪk/, take

Constraint ranking: MAX(seg) -IO >>NOCODA

Input: /teɪk / MAX-IO(SEG) NO CODA a) teɪ *! b) ☞ teɪk * c) teɪki *! Tableau (4.28): English realization of the input / teɪk/ iii) Input /steɪnz/ ‘steins’ constraint ranking: MAX I0 (SEG) >> COMPLEX

Input: /steɪnz / MAX-IO(seg) *COMPLEX a) seɪn *! b) siteɪnsɪ *! c) ☞ steɪns **

Tableau (4.29): English realization of the input /steɪnz/

These tableaux show that faithfulness constraints are ranked higher than markedness constraints. Thus, in English the demand that output forms be as similar as possible to their input correspondents outranks the well formedness of outputs. Therefore, faithfulness constraints dominate markedness constraints: FAITHFULNESS CONSTRAINTS >>

MARKEDNESS CONSTRAINTS. Thus, violating a faithfulness constraint in English is more serious than violating Markedness constraints. EkeGusii on the other hand as was realized in this study, unlike English, ranks markedness constraints higher than the faithfulness constraints as illustrated by tableau (25).

186 Syllable nativization or syllabification is governed by syllable well formedness in the target language (EkeGusii in this case). A loaned word in most cases violates some constraints of syllable well-formedness in the new language. This is because each language ranks constraints (from the universal set) differently (Prince and Smolensky, 1993). English loans coming into EkeGusii have a different syllable structure and therefore have to undergo syllable changes in order to be accommodated in the EkeGusii phonological system. It was discovered that EkeGusii language uses a number of strategies to change the syllable structures of the English loans coming into its phonological system. The speakers employed vowel epenthesis (anaptyxis and paragogue) strategy to deal with consonant syllable clusters and syllable codas. This is analzed in section 4.2.2.1 below.

4.2.2.1 Nativization by vowel epenthesis Epenthesis, according Katamba (1989), Eichoff (1990), Uffmann (2001 & 2004) and Rose

& Demuth (2006) among others, is where new segments appear from zero in the formerly unoccupied positions in a word or morpheme or between two previously abutting segments.

There are three types of such insertions: prothesis, which is the insertion of an initial segment, normally a vowel; anaptyxis, which is an insertion of a vowel between two consonants and paragogic insertion, which inserts a segment at the end of a word, especially a vowel (Katamba, 1989). This section considers paragogic and anaptyxistic insertions which seemed to be the preferred forms of epenthesis in EkeGusii.

Anaptyxis and paragogue were used by EkeGusii speakers to break complex syllable margins and open closed syllables of the loans from English. This is because EkeGusii neither tolerates complex syllable margins nor codas or closed syllables. Anaptyxis was found to be common in handling of complex margins, while paragogue was used to handle

187 codas in English nouns borrowed into EkeGusii. The two processes are discussed in sections 4.2.2.1.1 and 4.2.2.1.2 respectively.

4.2.2.1.1 Nativization of complex syllable margins

In section 4.1.3.1, it was observed that English is characterized by complex syllable margins of up to three consonants in onsets and four in codas (Cruttenden, 2011). EkeGusii on the other hand does not allow consonant clusters of any nature. Its phonotactics is characterized by simple syllable margins (Cammenga, 2002). Thus, any consonant clusters in EkeGusii loans from English are broken through a number of ways such as anaptyxis as shown in (90).

90) Nativization of English nouns borrowed into EkeGusii through anaptyxis. English noun pronunciation EkeGusii nativized form pronunciation store /stͻ:/ sitoo [si.too] brush /brʌ∫/ burasi [βu.ra.si] school /skʊl/ sukuru [su.ku.ru] glass /glɑ:s kerasi [ke.ra.si]

In (90), the English nouns contain a syllable with a complex margin, while their EkeGusii nativized forms do not. The nativised forms were realized through, among other processes, anaptytxis (vowel insertion); an epenthetic process. This process is illustrated by figure

(37) for the word /stͻ:/ ‘store.’

English EkeGusii

σ σ σ

O N O N O N

CC V C V C V

s t ͻ: s i t o:

Figure (37): Nativization by anaptyxis 188 Figure (37) shows that the consonant cluster /st/ in the English / stͻ:/ / is broken by insertion of the vowel [i] in EkeGusii, to create the syllables [si] and another additional one

([to:]) since the consonant /t/ is pushed into an onset position of a new syllable. Therefore, while there is only one syllable in the English word form, there are two in the EkeGusii nativized form.

In Optimality Theory, epenthesis leads to violation of a faithfulness constraint, DEP–I0 V, because the epenthetic segment has no counterpart in the input. In the case of the loaned word, the well formedness embodied in *COMPLX C among other constraints: (ONSET and NO CODA), dominate the DEP –I0 constraint (Uffiman 2004). Thus, epenthesis is opted for at the expenses of DEP- I0 V, which is opposed by the syllable markedness constraint. In other words, the cost of violating DEP –IO is less than that of the occurrence of an impossible syllable structure in the native system.

Following this argument, the nativization process in (90) above for the word /sku: l/

‘school’, for example, is analyzed by tableaux (30) and (31) for English and EkeGusii ranking of the constraints respectively.

English input: /sku:l/ ‘school’ English constraint ranking: DEP-I0 >> *COMPLEX

Input: /sku:l / DEP-IO *COMPLEX a) sukuru *! c) ☞ sku:l * Tableau (4.30): English realization of the input /sku:l/

EkeGusii input: /sukuru/ ‘school’

EkeGusii ranking: *COMPLEX >> DEP IO

189 Input: /sukuru / *COMPLEX DEP IO a) sk.uru *! c) ☞ su.ku.ru * Tableau (4.31): EkeGusii realization of the input /sukuru/

In tableau (31) candidate (b) is optimal because it does not violate the comparatively high ranked constraint in EkeGusii. The candidate avoids complex syllable margins that are banned in EkeGusii. The reverse is true in English language as tableau (29) indicates.

English tolerates complex margins or consonant clusters in onset position. English therefore ranks the markedness constraint lower than the given faithfulness constraint.

One obvious observation that can be made from this analyses is that English loan word nativization in EkeGusii through anaptyxis leads to an increase of syllables in the resulting loaned words as is illustrated in (91) repeated from (90) above.

91) Number of syllables in the English noun as compared to those in English nouns in

EkeGusii

English noun no. of syllables nativised form no. of syllables

/stͻ:/ ‘store’ 1 [si.too] 2

/brʌ∫/ ‘brush’ 1 [βu.ra.si] 3

/sku:l/ ‘school’ 1 [su.ku.ru] 3

/glɑ:s/ ‘glass’ 1 [ke.ra.si] 3

(91) shows that EkeGusii nativized forms have more syllables because, EkeGusii is a strict

CV language unlike English. This is further illustrated by figure (38) for the English word /sku:l/ ‘school’.

English EkeGusii 190 σ σ σ σ

O N C O N O N O N

CC V C C V C V C V

s k u: l s u k u r u Figure (38): Increase of number of Syllables in nativized word forms due to anaptyxis

This figure indicates that while the English word form has only one syllable, its EkeGusii nativized form has three. This is due to vowel epenthesis.

Vowel epenthesis as a strategy of loaned word adaptation in order to satisfy constraints on phonotactic and syllable structure, in the borrowing language is a common feature across languages. It is not unique to EkeGusii language, nor indeed to Bantu languages. This is demonstrated by the following examples of nativized loan words into various languages from English as adapted from Uffimann (2004) in (92).

92) Vowel epenthesis of English loan words in Yoruba, Japanese, SeTswana, Shona, Samoan, Sranan

Language example of noun pronunciation gloss Yoruba kilaasi [kilaasi] class

Japanese sutoraiko [sutoraiko] strike

SeTswana kirisimasi [kirisimasi] christmas

Shona girini [γirini] green

Samoan sikauti [sikauti] scout

Sranan buku [uku] book

In (92), there is at least an epenthetic vowel, either to break a complex margin/cluster or open a closed syllable. For example, in SeTswana, the word Christmas is nativized to

191 [kirisimasi] by epenthesizing vowel [i] three times. The first epenthesis breaks the consonant cluster [kr], the second one the cluster [sm]; while the third opens the closed vowel at the end of the loaned word. SeTswana, like EkeGusii and most other Bantu languages is a strict CV syllable structure.

4.2.2.1.2 Nativization of closed syllables While English sometimes tolerates codas or closed syllables in its syllable structure because of its nature (has diverse sources of loan words) (Cruttenden, 2011), EkeGusii has a total ban on codas, just like most other Bantu languages (Nash, 2011; Uffimann, 1994).

Thus, English syllables with codas entering EkeGusii have to break the codas. In this study, this is achievable through paragogic epenthesis, which is the insertion of a vowel at the end of a word syllable (John, 2000) as (93) shows.

93) Nativization English nouns borrowed into EkeGusii through paragogic epenthesis

English noun pronunciation EkeGusii nativized form pronunciation fine /faɪn/ baini /βa.i.ni/ torch /tͻ: t∫/ tochi /tͻ.ͻ.t∫i/ machine /mǝ∫i:n/ masini /ma.si ni/ suit /su:t/ suti /su.ti/ (93) shows that all the source forms of the words have closed word end syllables. Their nativized forms, however, are not; instead, they are all open, as illustrated by figure (39), for the English word /su:t/ ‘suit’.

English EkeGusii

σ σ σ

O V C O N O N

C N C C V C V

s u: t s u t i 192 Figure (39): Opening of the English closed syllables by paragogic epenthesis in EkeGusii

This figure shows that the closed English syllable is opened by a paragogic vowel [i]. This, like anaptyxis above has the effect of creating an additional syllable in the nativized word form. This in Optimality Theory means that, whereas EkeGusii nativized forms obey the

*CODA constraint; English sometimes does not, it instead violates it. Thus presupposing the following constraints, NO CODA and DEP IO. These two are ranked differently to analyzes the phenomena in (96). Thus, while English ranking is DEP IO >> *CODA,

Ekegusii ranking is *CODA >> DEP-IO. The English word 'torch’/ tͻ:t∫ / and its nativized form in EkeGusii ‘tochi’ [tͻͻt∫i] are analysed by tableaux (31) and (32) respectively.

English input: /tͻ:t∫ / ‘torch’

Constraint ranking: DEP IO >> *CODA

Input: / tͻ:t∫ / DEP-IO NOCODA a) ☞ tͻ:t∫ * b) tͻ:t∫i *! Tableau (4.32): English realization of the input /tͻ:t∫/

EkeGusii input: [tͻͻt∫i] constraint ranking: NO CODA >> DEP-IO

Input: / tͻͻt∫i / NO CODA DEP IO a) tͻ:t∫ *! b) ☞ tͻͻt∫i *

Tableau (4.33): EkeGusii realization for the input /tͻͻt∫i/

In tableau (32), candidate (a) is the output even though it violates the NO CODA, a markedness constraint. This is because this constraint is not of any serious consequence in

193 determining the output in English. DEP-I0 on the other hand if violated by English will mean that, the violating candidate will never be the output. The reverse is true in tableau

(33) for EkeGusii. NO CODA is of higher rank as compared to DEP – IO in EkeGusii.

The implication here is that any English syllable with coda entering EkeGusii must be changed; that is nativized to codalessness through vowel paragogic epenthesis in order to conform to the CV syllable structure of EkeGusii.

Complex syllabic margins according to Kager (1999) have a marked status. That is why languages avoid them through vowel epenthesis and consonant deletion processes. Some languages, such as Lanakel (Lynch, 1974, Blevins, 1995), avoid both types of complex margins (onsets and codas) by vowel epenthesis as (94) adapted from Kager (1999) shows.

94) Avoidance of complex margins by vowel epenthesis in Lanakel

Complex margin form epenthesized form gloss i) /t-n-ak-ol/ t i.na.gͻl you (sing.) will do it ii) /ark-ark/ ar.ga. r ikh to growl iii) /k m-n-man-n/ kam. n ima. n in for her brother

(94) shows that consonant clusters are not allowed in Lanakel, which presupposes the constraints *COPMLEX ONS and *COMPLEX CODA. In (94i), the vowel [i] breaks an onset cluster, while in (94ii), the vowel breaks a coda cluster. This is analyzed in tableau

(34) for the Lanakel input for the word /t-n-ak-ol/.

Input: /t-n-ak-ol/

Constraints: *COPMLEX ONS >> DEP IO (SEG)

Input: /t-n-ak-ol/ *COPMLEX ONS DEP IO SEG)

194 a) t-n-ak-ol *! b) ☞ t i.na.gͻl * Tableau (4.34): Lanakel realization of the input [t-n-ak-ol]

Candidate (b) is optimal because it does not allow a complex onset, which is banned in the language. Candidate (a) on the other hand loses because it violates the constraint. This phenomenon (vowel epenthesis) is common in the nativization of EkeGusii loanwords from

English as has already been discussed above.

According to Batibo (1996), vowel epenthesis involves the insertion of a vowel between two consonants or after a consonant in a syllable final position. The choice of the vowel to be epenthesized is, according to Uffimann, determined by a number of factors. According to Pulleyblank (1988) and Shinohara (1997), for example, in theie analysis of Yoruba and

Japanese respectively concluded that the maximally unmarked, underspecified or phonetically shortest vowel is inserted via default; Paradis (1996) analyzing Fula found that the stem vowel is copied into the epenthesized vowel slot; while Akinalabi (1993), working on Yoruba, Batibo (1995) on SeTswana and Smith (1997) on Sranan, concluded that epenthesis is sometimes controlled by consonant vowel assimilation, mostly labial attraction. A combination of all these contributed to the choice of the epenthetic vowel in this study. It does not, however, mean that it is only vowel sounds that are epenthesized. In some languages, such as Axininca Campa (Kager, 1999), consonant sounds are also epenthesized as shown in (95).

95) Consonant epenthesis in Axininca Campa Underspecified form surface form gloss /no-N-koma-i/ noŋkomati he will paddle /no-N-koma-ko-i noŋkomatakoti he will paddle for 195 /no-N-t∫hik-ako-aa-i-ro noɳt∫hikakotaatiro he will cut for it again

Adapted from Kager (1999)

(95) indicates that the underspecified word forms are without consonants in some onset positions. The realizations of these words however are with an epenthesized consonant [t].

This violates the faithfulness constraint DEP-IO (SEG), which demands that output segments must have input segments. This consonant epenthesis is out to break vowel complex margin, *COMPLX (V). Axininca Campa consonant epenthesis in (95) is analysed in tableau (35) for the input /no-N-koma-i/.

Constraint ranking: *COMPLX (V) >> DEP-IO (SEG)

Input: /no-n-koma-i/. *COMPLX (V) DEP IO (SEG) a) ☞ noŋkomati * b) no-n-koma-i *! Tableau (4.35): Lanakel realization of the input [no-n-koma-i]

Because complex vowel margins are not allowed in Axininca Campa, a consonant is epenthesized to break the complex vowel in candidate (a), the optimal candidate.

However, there was no case of consonant epenthesis observed in this study.

4.2.3 Supra-segmental nativization The third feature that characterize phonological change of EkeGusii loan nouns from

English is Supra segmental or prosodic. Supra segmental nativization in this study focused on toneme structure. As was observed in subsection 4.1.4.1, Ekegusii, is a tone language.

Katamba (1993) posits that in a tone language, pitch can be used to distinguish word meaning or to convey grammatical distinction. It is in this perspective that tone languages differ from stress (non-tone) ones such as English. In other words, pitch does not have

196 these functions in stress languages. It is against this background that it can be argued that

EkeGusii loans from English are nativized by tonemization (tone nativization). The English stressed noun loans are assigned EkeGusii toneme structure, they are tonemized as illustrated by (96).

96) Toneme nativization English nouns in EkeGusii

English noun pronunciation (stressed) EkeGusii nativized form pronunciation (toned) school /'sku:l/ esukuru [e-s u k u r u]

L H L H blanket /'blæŋkɪt/ oboranketi [oβo-r a ŋ k e t i]

L L H L H chief /'t ∫ i:f/ omochibu [o m o-t ∫ i β u]

L L H L wire /'w ǝ ɪ ǝ/ eguaya [e g w a j a]

L H L H

Adapted from Bickmore (1997)

(96) indicates that while all the English forms are stressed in the initial syllables, their

Ekegusii nativized forms are tonemized. All the prefixes in the nativized forms are toneless

(low toned), while the initial vowels of the root forms are high toned (Bickmore, 1997;

Mwita, 2012). In other words, stress in English is preserved as a variant of high tone in

EkeGusii nativized forms (Kang, 2010). Cases of such changes are not unique to EkeGusii.

197 Some other tone languages behave in a similar manner. For example, in Yoruba, spoken in

Nigeria, and Shona, spoken in Zimbabwe (Kenstowicz, 2006); and Dholuo, spoken in

Kenya (Owino, 2003), loan words from English with stressed syllables are realized with high tones in these languages as shown in (97).

97) Stress preservation as a variant of high tone in tone languages a) Yoruba English noun pronunciation (stressed) Yoruba nativized form pronunciation (toned) liberty /'lɪbǝt/ libati [liHbati] tomato /tǝ'ma:tǝʊ/ tomato [tomaHto] guarantee /gærǝn'ti:/ garanti [γarantiH] b) shona English noun pronunciation (stressed) Shona nativized form pronunciation (toned) Recipe /'resǝpi/ resipi [reHsipi]

Philosophy /fǝ'lɒsǝfi/ firosofi [firoHsofi] c) Dholuo

English noun pronunciation (stressed) Dholuo nativized form pronunciation (toned) battery /'bætri/ betiri [beHtiri] location /lǝʊ'keɪ∫n/ lokesen [lokeHsen] degree /di'gri:/ digiri [diγiHri] Adapted from Kang (2010)

In (97), the stressed syllables of English are preserved as high tones in the target languages, just like in EkeGusii. This study therefore concludes that English stress becomes high tone in EkeGusii loaned nouns from English.

The change of English stress to high tone as observed above can be accounted for within

Optimality Theory, specifically by autosegmental OT. Bickmore (1997) observes that within Optimality Theory, instead of deriving surface forms from underlying 198 representations via the serial application of a number of phonological rules, a form is grammatical if it satisfies a ranked set of constraints better than any other possible candidate does. The candidate set consists of forms created from a given input form by generator (GEN), the component that generates permutations of the input. With respect to tone, it is assumed that GEN can manipulate both tones and their associations to Tone

Bearing Units (TBUs). Thus, minimally, GEN can add and delete tones themselves, as well as manipulate (that is expand or reduce the size of) input High Tone Spans (HTS).

Following Bickmore (1997), this study analyses EkeGusii loan nouns nativization by tonemization within the Optimality Theory frameworks (McCarthy and Prince, 1995. The following constraints of these framework are used: MAX-IO (F), which demands that no feature should be deleted; DEP-IO, which prohibits feature insertion; IDENT (F), which prohibits feature changing between input and output correspondent; and UNIFORMITY, which prohibits feature fusing.

Following Goldsmith (1990) well formedness conditions for tones, that is: i) all vowels are associated with at least one tone, ii) all tones are associated with at least one vowel and iii) association lines do not cross, the following constraints ensuring that tones are linked and that tones bearing units are assigned a tone are used in this study. DEP- ET, which prohibits floating of tones (*FLOAT) (T); meaning that every tone must have a correspondent tone bearing unit (TBU) and MAX- IO (T), which specifies that each TBU must have a correspondent tone (SPECIFY (T)) (Bickmore, 1997). Bickmore (1997) ranks the constraints outlined above as (98).

98) Ekegusii toneme constraints ranking.

199 *FLOAT >> MAX IO (T) >> IDENT (H) >> DEP – IO (T) >> SPEC (T)

In the analysis of loaned words within Optimality Theory, the aim is to show that the loans obey the constraints of the target language and not those of the loaning one. This is true given what has already been indicated, EkeGusii is a tone language and therefore can be analyzed tonally while English is a stress language and cannot be analyzed in terms of tone.

The process of tonal nativization of English loans into EkeGusii starts with tonimization, which involves changing the loans from their stress form to a tone form, as illustrated in

(97) above. It is this tonemized form that is analyzed within OT in this study to account for

EkeGusii loanword nativization by tonemization as in tableau (36) for the noun /ooraŋgeti/ ‘blanket’ from the input /blæŋkɪt/.

Input: /'blæŋkɪt/ output: o β o r a ŋ g e t i

L L H L H This tone structure presupposes the following tone constraints. *FLOAT >> MAX-IO >>

IDENT-IO >> DEP-IO.

Input: oo-raŋget-i *FLOAT MAX-IO IDENT-IO DEP-IO SPEC (T) (H) (T) H

200 a) ☞ o β o- r a ŋ g e t i

H * * * b) o β o- r a ŋ g e t I

H * ! * * * *

* c) o β o- r a ŋ g e ti

H * ! * * d) o β o r a ŋ g e t i * ! * * * * * *

* *

Tableau (4.36): EkeGusii realization of the input /oo-raŋget-i / Adapted from Bickmore (1997)

The optimal candidate in tableau (35) is (a) because it obeys the highest ranking constraint

*FLOAT, which prohibits a floating high tone (H). Besides, this candidate obeys most of the constraints, violating the relatively lowly ranked SPECIFY (which demands that each tone bearing unit (TBU) must have a correspondent tone. Candidate (b) on the other hand loses because it violates *FLOAT, a fatal violation. Candidate (c) loses because it changes a low (L) feature to a H feature. This is a serious violation of tone structure in EkeGusii.

Candidate (d) deletes the feature (H) suggesting (wrongly) that the language has only L tones, which in essence are underlying.

Of significance to this study is the observation that English nouns in EkeGusii adopt the tone structure of EkeGusii in the process of being accommodated into the phonological structure of EkeGusii as shown in (97) above. 201 4.2.4 Nativization by phonological processes A phonological process involves a change of a phoneme (segment), in terms of features, which is motivated by a number of factors such as the environment of the phoneme and the phonemes with which it occurs in a word. By environment here it means the position in a word where a phoneme occurs (Katamba, 1993). For example, if a voiced English stop such as /d/ occurs after a voiceless one such as /k/ as in walked, it is devoiced as in /wͻkt/ and not */wͻkd/ and therefore the phonological process involved is that of devoicing. This section gives an analysis of various phonological processes which characterize the nativization of EkeGusii loaned words from English. These are processes which characterize the phonology of EkeGusii, the target language.

4.2.4.1 Nativization by voice dissimilation (Dahl’s Law) According to Bickmore (1998), Dahl’s law is a phonological process in which voiceless consonants in adjacent syllables become dissimilar in some Bantu languages. Guthrie

(1967) observes that languages which show the effect of Dahl’s law are found within his zones E20 - E50 – F20 and G60. It has also been argued that languages vary a great deal as to which particular consonants undergo the process, which consonants trigger it, and how the process affects multiple targets within the same word (Bennett, 1967 and Davy &

Nurse, 1982). According to Bickmore (1998), there is evidence that Dahl’s Law affects

EkeGusii as in (99). 99) Voice dissimilation (Dahl’s law) process in EkeGusii (a) /ͻ-kͻ-rͻͻt-a/ [ͻkͻrͻͻta] ‘to dream’ /o-ko-kor-o/ [okoγoro] ‘foot’ /o-ko-nyu-a/ [okoɳwa] ‘to drink (b) /o-ko-kana/ [oγokana] ‘to deny’ /o-ko-tuu.a/ [oγotuua] ‘ to be blunt’ /ͻkͻsɛka/ [ͻγͻsɛka] ‘to laugh’ Source: Bickmore (1998)

202 In (99a), in the word ‘[okoγoro]’ for example, the voiceless velar obstruent (/k/) in the prefix {oko-} demands the voiced velar obstruent (/γ/) in the root {-γor}. Thus, the [–voice] and [-continuant] features of [k] in the prefix, dissimilates to [+voice] and [+continuant] in

[γ] in the root of the word. The dissimilation process in this case is from the voiceless obstruent /k/ underlyingly to the voiced obstruent /γ/ in the surface. The reverse is true in

(99b). The [+voice] and [-continuant] underlying features in the prefixes surface as [-voice] and [+continuant] in the roots. Uffmann (2013) explains this phenomenon using the laryngeal feature [OPEN]. The consonant /k/, which is produced with open larynx, dissimilates to the next onset if the onset also contains a consonant produced in the same manner- open larynx. This means that

/k/ will lose its [open] laryngeal feature to become [g], with laryngeal [close]. However, [g] is not recognized in EkeGusii phonology, except in combination with the nasal consonant /ŋ/. Therefore, as Uffmann suggests, in order for the [g] to share its manner specification with the preceding nasal, It loses the [closed] manner feature to become [γ], a featurally mannerless continuant. The dissimilation processes in (99) can be illustrated by figures (40) and (41) for EkeGusii words [okoγoro] okogoro ‘leg’ and [oγokana] ogokana ‘to deny’. |okokoro| → [okoγoro] o k o k o r o underlying form

o k o γ o r o surface form Figure (40): EkeGusii dissimilation from |k| → [γ] |okokana| → [oγokana] o k o k a n a underlying form

o γ o k a n a surface form Figure (40): EkeGusii dissimilation from |γ| → [k] Figure (40) shows that /k/ the dorsal stop dissimilates to the dorsal continuant [γ], while it is the opposite in (41); that is, the dorsal continuant dissimilates to the dorsal stop. These dissimilation processes in Optimality Theory, suggest the Obligatory Contour Principle 203 (OCP), a phonological hypothesis that bans certain consecutive identical features in a linguistic unit such as a morpheme or word underlyingly (Prince and Smolensky 2004). In particular, dissimilation in EkeGusii seems to target the voiceless dorsal stop feature, leading to a markedness constraint, *OCP(dorstop) Uffmann (2013), which provides that there should be no sequence of voiceless dorsal stop and another voiceless stop. The dissimilation also suggests the faithfulness constraint DEP IO (F), which demands that features of an input segment must be preserved in the output, no change. Given these constraints, the dissimilation in figure (36) is analyzed in in tableau (37). Input: /okokoro/ okogoro ‘leg’ Constraints: *OCP (dorstop) >> DEP IO (Seg)

Input: /okokoro/. *OCP (dorstop) DEP IO (Seg) a) okokoro *! b) ☞ okoγoro * Tableau (4.37): EkeGusii output of the input /okokoro/

The input form of this tableau loses as candidate (a), because it violates the constraint

*OCP (dorstop), a highly ranked constraint in EkeGusii, which demands that a voiceless dorsal stop should not be in a sequence with another voiceless stop. Candidate (b) on the other hand is optimal since it obeys the constraint. Dahl’s law seems to affect English loan words into Ekegusii as (100) below shows 100) Dahl’s law in English nouns borrowed into EkeGusii (a) EkeGusii /, t, s/ features dissimilate with [g] features English noun pronunciation EkeGusii nativized form pronunciation bank /bæŋk/ ebenki [e-eŋg-i] drink /driŋk/ eturunki [e-turuŋg-i]

sink /siŋk/ esinki [e-siŋg-i]

204 (b) EkeGusii /t / features dissimilate with [r] features; /nz/ features with [t, ] features

English noun pronunciation EkeGusii nativized form pronunciation [t-iכt eturansiboti [e-turanzi:כ Transport /trəensp

Transfer /trænsfə:/ eturansiba [e-turanzi-a]

(100) shows disimillation of the feature voice in the nativized word forms, suggesting the

*OCP (voice) constraint. This is illustrated by the EkeGusii nativized forms of the words

[esiŋgi] ‘sink’ and [e-turanzi-a] ‘transfer, respectively in figures (41) and (42).

Øs ɪ ŋ k Ø English Ø s Øŋ kØ EkeGusii

Øs i ŋ g Ø Phonemic substitution

e s i ŋ g i (Nativization: Borrowing complete) Figure (41): Substitution of the English /k/ for EkeGusii [g] Ø d Ø r ɪ ŋ kØ English ØØ Ø r Øŋk Ø EkeGusii

Ø t Ø r u ŋg i Phonemic substitution

e t u r u ŋg i (Nativization: Borrowing complete) Figure (42): Substitution of the English /d/ for EkeGusii [t]; English /k/ for EkeGusii [g] In figure (41), the EkeGusii /k/, a voiceless velar stop, is substituted for [g], a voiced velar stop. This is in order for the two to be dissimilar in the feature [voice]. In figure (42), there are two substitutions of English consonants in EkeGusii. First, there is substitution of the

English voiced alveolar stop /d/ for EkeGusii voiceless stop [t], which dissimilates the features [voice] and [continuance] between the substituted [t] and the [r] in the adjacent syllable. Secondly, there is substitution of the English /k/ for EkeGusii [g] as in (42).

The dissimilation processes in the nativized words in the given data is not motivated by prefixes as in the EkeGusii non-nativized words discussed early, instead, the process seems to be triggered by the first consonants of the roots. In (100a), it is the voiceless obstruents

205 of the roots ([,t,s]) that triger the process, while in (100b) the initial consonant of the root is [t]. The [t] in the roots of the examples in (100b) is voiceless, hence the voiced pre- consonantal continuant [z]. This voiced pre-consonantal continuant triggers voice dissimilation in the adjacent syllable consonant, hence the voiceless continuant [], in the

’t-i] ‘eturansiboti’ (transport) and [e-turanzi-a] ‘eturansibaכwords [e-turanzi

(transfer).

In order to analyze the voice dissimilation shown in (100) for the English loans in EkeGusii in OT perspectives, the *OCP(feature) constraint (McCarthy 2004) is presupposed. This constraint bans consecutive identical features in segments of a word. The changes that occur in (100) also suggest the faithfulness constraint IDENT IO (F), which requires that the features of an input segment be identical in the surface. Thus, the English and EkeGusii realizations of the words given in (100) differ only in their ranking of these constraints as tableaux (38) and (39) for the English input /bæŋk/, ‘bank’, show. Input: /bæŋk/ Constraints: IDENT (F) >> *OCP (voice)

Input: /bæŋk/ IDENT (F) *OCP (voice) a) ☞ bæŋk b) eŋgi *! *

Tableau (4.38): English output of the input / bæŋk/

Output (a) is the winning candidate in the tableau above because it violates the least serious constraint, *OCP (voice) in English language. Candidate (b), which loses, on the other hand violates the serious constraint in the language; IDENT (F), which demands that the voice features in the input must be maintained in the output. This is comparable to tableau

(39) below for EkeGusii realization with reversed constraint ranking from that of English. EkeGusii input: /e.e.ŋg.i / 206 Constraint ranking: *OCP (voice) >> IDENT (F)

Input: /e.e.ŋg.i / *OCP (feature) IDENT (F) a) ☞ e.e.ŋg.i b) e.ba.ŋki *! Tableau (4.39): EkeGusii output of the input / e.e.ŋg.i /

Candidate (a) obeys the constraint OCP (voice), which is highly ranked in EkeGusii, and therefore wins, while (b) loses because it violates the constraint.

Loan word voice dissimilation (Dahl’s Law) discussed above is also influenced by the prefix structure of EkeGusii, which is different from that of the loaning language.

Basically, most nouns in EkeGusii have prefixes which have a bi-morphemic structure. The prefix itself is divisible into two elements, that is, an initial vowel, (augment or pre-prefix) and the prefix, per-se (Bickmore, 1998) as (101) demonstrates. 101) EkeGusii bi-morphemic prefix structure a) o - mo - mur - a b) a - ba - mur- a augment prefix root FV augment prefix root FV

boy boy 1.1 2.2

c) e - ki - ar - a d) e - bi - ar- a augment prefix root FV augment prefix root FV

finger 7.1 finger 8. more The augment and the prefix in (101) mark the class andthan the 1 number of the noun. For one instance, the augment and prefix in (101a) mark the noun as class 1 and that the noun is in singular, while in (101b) the augment and prefix mark the noun as class 2 and that it is in the plural form.

207 According to Katamba (1989), in many Bantu languages, there is a rule which requires that a consonant in a prefix disagree in voicing with the first consonant of the root it is attached to; a voiced stem initial segment requires a voiceless consonant in the prefix, and that a voiceless stem - initial segment requires a voiced consonant in the prefix. The Kirundi

(Burundi) examples in (102) demonstrates this observation.

102) Voice dissimilation in Kirundi Imperative 1 st person singular present Word Pronunciation Gloss Word Pronunciation Gloss a) rya /rjia/ eat turia /tu-rjia/ I eat mwa /mwa/ shave tumwa /tu-mwa/ I shave va /va/ come from tuva /tu-va/ I come from bona /ona/ seen tubona /tu-bona/ I’ve seen b) soma /soma/ read dusoma /du-soma/ I read teeka /te:ka/ cook duteka /du-te:ka/ I cook seka /seka/ laugh duseka /du-seka/ I laugh kubita /kuita/ hit dukubita /du-kuita/ I hit Source: Katamba (1989) (102a) shows that when the prefix is voiced, the stem (root) is voiceless; the opposite is true in data (101b). This is Dahl’s rule in operation which, as illustrated by (100) above, affects English loaned nouns in EkeGusii. Dahl’s Law affects English loans into other languages too, with similar results as those discussed in this research. For example, in Kitharaka, English loan words undergo the process as exemplified in (103). 103) Dahl’s Law in Kitharaka loaned words from English English word pronunciation Kitharaka nativized form pronunciation UF SF small shool /sku:l/ gasukuru |ka-sukuru| [γa-sukuru] to freeze /fri:z/ gofirisi |ko-firisi| [γo-firisi] to sheet /∫i:t/ goshiti |ko-∫iti| [γo-∫iti] Adapted from Uffmann (2013) (103) shows that in their underlying forms (UF), the prefix and the root initial consonants share the feature [- voice]. In their surface forms (SF) however, the consonants are

208 dissimilar in terms of voice. While those of the roots are voiceless, those of the prefixes are voiced. This is Dahl’s Law in operation.

4.2.4.2 Nativization by defricativization, spirantization (fricativization) and bilabialization

Defricativization as was pointed out in sub-section 4.1.2.1.4 is a process where a voiced continuant or spirant loses its continuant quality by becoming [-CONTINUANT].

Fricativization therefore can be described as the opposite of defricativization. In fricativization or spirantization, a [-CONTINUANT] sound becomes [+CONTINUANT].

Bilabialization, on the other hand is the process of changing a non-bilabial sound into a bilabial one. These processes characterize EkeGusii loan words from English and are discussed in 4.2.4.2.1, defricativization; 4.2.4.2.2, fricativization/spirantization; and

4.2.4.2.3, bilabialization.

4.2.4.2.1 nativization by defricativization As pointed out in subsection 4.1.2.1.4 above, defricativization is a process where a voiced continuant consonant, [+CONTINUANT], loses its continuant feature to become minus continuant [-CONTINUANT]. This process also affects EkeGusii loan words from English as illustrated in (104).

104) Nativization of English nounsborrowed into EkeGusii through defricativization a) Nativization of /γ/ as [g]

209 English noun pronunciation EkeGusii nativized form pronunciation UF SF tank /tæŋk/ etanki |etan-γi|→[etaŋ-gi] drink /drɪŋk/ eturunki |eturun-γi|→[eturuŋ-gi] b) Realization of // as [b] English noun pronunciation EkeGusii nativized form pronunciation

UF SF camp /cæmp/ ekambi |ekan-i|→[ekam-bi] remand /rɪmænd/ erumande |eruman-ri|→[eruman- di] c) Realization of /r/ as [d] UF SF secondary /sekǝndri/ esekenderi |ɛsɛkɛn-ri| → [ɛsɛkɛn-dɛri] [n-daγitaכ n-raγita| →[ ɛkכ conductor /kɒndʌktǝ/ ekondagita |ɛk

The substitution processes in (104) are further illustrated by figures (43), for the EkeGusii /γ/ → [g]; (44), for EkeGusii // → [b]; and (45), for EkeGusii /r/ → [d]. English word: tank /tæŋk/ Nativized from: [etaŋgi] ‘etanki’

e t a n γ i EkeGusii underlying (nativized) form

e t a ŋ g i EkeGusii surface realization Figure (43): Realization of EkeGusii /γ/ as [g]

English word: camp /kæmp/

Nativized from: [ekambi] ‘ekembi’

e k a n  i EkeGusii underlying (nativized) form

e k am b i EkeGusii surface realization Figure (44): Realization of EkeGusii // as [b]

English word: secondary /sekǝndri/

210 Nativized from: [esekenderi] ‘esekenderi’

ɛ s ɛ k ɛ n r ØØi EkeGusii underlying (nativized) form

ɛ s ɛ k ɛ n d ɛ r i EkeGusii surface realization Figure (45): Realization of EkeGusii /r/ as [d] These figures show that an underlying fricative consonant is realized as a stop, which is of the same place of articulation in the surface form of the loan word. This is phoneme defricativization. This kind of mapping is analyzed in (105) for the English words /tæŋk/

‘tank’, /sɪŋk/ ‘sink’ and /drɪŋk/ ‘drink’. 105) Output mapping English nouns in EkeGusii by defricativization (|γ|→[g]) i) |e-tanγ-i| → [etaŋgi] ii) |e-sinγ-i|→ [esiriŋgi ] iii) |e-turunγ-i/|→ [eturuŋgi]

Here, the voiced velar fricative /γ/ is realized as [g], a voiced velar stop, which is not found in EkeGusii phonology. This change is due to nasal homorganization discussed in section

(4.1.1.2.3). The nasal [n], an alveolar is homorganized to [ŋ] by [γ], a velar. It is the newly formed nasal velar [ŋ], a stop, which assimilates [γ], a velar fricative to [g], a velar stop.

Thus, EkeGusii does not have the voiced velar stop [g]; it is only realized phonetically from the voiced fricative /γ/, which is defricativized.

The realizations in (104b and c) are analyzed in (106).

106) Nativization of English nouns borrowed into EkeGusii by defricativization (||→ [b]) i) |e-kaan-i| → [ekaambi]

|ɛ-kͻͻniut-al| → [ɛkͻmbju:ta] ii) |r| → [d]

|e-rumanr-e| → [erumande]

211 |ɛsɛkͻnrari| → [ɛsɛkͻndari]

In (106i), the bilabial voiced fricative ||, underlyingly, is defricativized to [b], a voiced bilabial stop, while in (106ii) /r/, an alveolar tap is defricativized to [d] an alveolar stop.

The underlying forms in data (105) and (106) above are explained by the fact that EkeGusii phonemic inventory does not have voiced plosives [g], [d] and [b], instead it has the voiced fricatives [γ], [r] and []. Therefore, the borrowed plosive stops from English are nativized by defricativization as shown. The process of defricativization is such that the plosive is first fricativized, for example, /b/ → [], before being defricativized in the surface form as shown in (107).

107) Defricativization of the English noun in EkeGusii after fricativization camp → |kan -- i|→ [kambi] ‘camp’.

Defricativization can be explained by OT using the constraints IDENT IO (F) and

*COMPELEX C (because a homorganized nasal consonant NC, as has already been observed is treated as a singletone consonant and not a consonant cluster). (107) above is analyzed in tableau (40) for illustration.

Input: |kan-i| > /kambi/

Output: [kambi]

Constraint ranking: *COMPLEX C >> IDENT IO (F)

/ Input: /kani/ *COMPLEX C IDENT IO (F) a. ☞ kambi ** b. kani *! c. kanbi *!

212 Tableau (4.40): EkeGusii output of the input /kani/

Candidate (a) is the output in this tableau because it satisfies the highly ranked constraint in the language, *COMPLEX C. This is because, as was observed earlier, the combination, nasal consonant (NC), is treated as a pre-consonantal unit (one consonant) and not a consonant cluster. Candidates (b) and (c) violate this constraint; they have consonant clusters (violating *COMPLEX C).

Cases of loaned word nativization by defricativization has not been given a lot of attention.

This is because, as Ohala and Sole (2008) observe, defricativization is associated with nasalized fricatives and that these kind of fricatives are rare or marked and that they tend to be defricativized if voiced. Ohala and Sole further observe that defricativization is as a result of the difficulty involved in simultaneous production of nasalization and friction, both of which features are marked. Defricativization therefore aims at unmarking one of the features for ease of articulation. The target feature in this case being the voiced fricatives. This process characterizes EkeGusii and the English loaned words into EkeGusii as shown in (107) above.

Defricativization though rare, as observed by Ohala and Sole (2008), seems to be common in some Bantu languages, many of which do not treat nasal consonants as clusters of consonants as illustrated in (108) for Kiswahili spoken in East Africa.

108) Kiswahili defricativization Word UF SF Gloss kamba |kaN-a| [kamba] rope ngombe |N-γN-ɛ| [ŋͻmbe] cow kondoo |kͻ-N-rͻͻ| [kͻndͻͻ] sheep

213 pingu |piN-γu| [pingu] curffs (ad hoc)

(108) shows that nasals in Kiswahili, like in EKeGusii, are underspecified in the underlying form. They only receive feature specification on the surface (phonetically). All the fricatives, after the nasals underlying are realized as voiced stops in the surface. This is defricativization.

4.2.4.2.2 Spirantization/fricativization of the stops According to Kenstowicz (1994), spirantization is a phonological process which involves a change of stops to fricatives (spirants) through what phonologists refer to as lenition or weakening of the stops. In this study, the process can be said to be motivated by Dahl’s

Law of dissimilation discussed in (4.2.4.1) above as illustrated by (109).

109) Nativization of English nouns by fricativization of bilabial stops

English word pronunciation EkeGusii nativized form pronunciation

UF SF dip /dɪp/ tibu |tipu| tiu pastor /pʌstǝ/ basita |pasita| asita bolt /bͻlt/ boriti |pͻriti| ͻriti

In these data, for example in the English word /dɪp/ ‘dip’, which is realized as [tiu] ‘tibu in EkeGusii’, the stop /p/ is realized as [], a voiced fricative. This is because the consonant

/p/ must be dissimilar in terms of voice features with the initial root consonant, [t]. In this case, it becomes [], a [+voice and +continuant] consonant, since [t] is [-voice]. It becomes

[+ continuant] because EkeGusii does not have the stop /p/ and therefore the bilabial

214 continuant [] is picked because it is the closest consonant to /p/ in terms of features (they are both bilabials).

The English velar stop /k/ is equally fricativiced during nativization due to Dahl’s Law of voice dissimilation as exemplified in (110).

110) Nativization of English nouns in EkeGusii by fricativization of velar stops

English noun pronunciation EkeGusii nativized form pronunciation

UF SF taxi /tᴂksi/ tegisi |tekisi| [teγisi] socks /sɒks/ sogisi |sͻkisi| [sͻγisi] box / bɒks/ bogisi |ͻγisi| [oγisi] degree /dɪgri:/ tigirii |tikirii| [tiγirii]

In (110), the velar, /k/ is fricativized into the voiced velar fricative [γ] in the EkeGusii nativized forms from English. This facilitates voice dissimilation of the adjacent syllables.

For example, in the word [teγisi], ‘taxi’, the consonant [γ] of the second syllable is dissimilar to that of the first syllable, [t] at least in terms of voice. While [t] is a voiceless stop, [γ] is a voiced continuant.

This phenomenon presupposes the markedness constraint *OCP(voice) and the faithfulness constraint IDENT IO(F) as exemplified in tableau (41) for the English word /tᴂksi/ ‘taxi’

Input: /eteγisi/ Constraint ranking: *OCP (voice) >> IDENT (F)

Input: / eteγisi / *OCP (voice) IDENT (F)

a) ☞ eteγisi *

b) tᴂksi *!

Tableau (4.41): EkeGusii output of the input / eteγisi / 215 In this tableau, the optimal candidate is (a) because it obeys the constraint which bans sequences of voiceless dorsal stops and another voiceless stop. The dorsal stop /k/ does therefore not occur with the voiceless stop /t/ consecutively, hence, the voiced continuant

[γ] as the output.

Fricativization or spirantization of stops as discussed above is not a preserve of EkeGusii.

Other languages such as Kitharaka (Uffmann, 2013); and KiKuria (Oden, 1994), among others show a similar trend as as in (110).

110) Dahl’s Law in Kikuria

Word Pronunciation Gloss UF SF okogaamba |okokaamba| [okoγaamba] to say ogosooka |ͻkͻsͻͻka| [ͻγͻsͻͻka] to respect ogoteema |ͻkͻtɛɛma| [ͻγͻtɛɛma] to hit

Adapted from Uffmann (2013)

In (110) a prefix stop which is underlyingly voiceless is voiced if the stem or subsequent syllable starts with a voiceless consonant and the other way round. The prefix therefore dissimilates in voicing from the stem or a following prefix.

Dissimilation witnessed in (110) does not involve voice only, but frication as well. In cases where there are only, stops in adjacent syllables, one of stops has to be fricativiced. For

216 example, in |ͻkͻtɛɛma| (underlying), there are two stops: |k| and |t|, therefore, one has to be fricativiced to effect dissimilation. In this case it is |k| of the prefix which changes to the voiced fricative [γ]. This is what happens to EkeGusii loaned words as discussed in this sub-section.

4.2.4.2.3 Nativization by bilabialization of labiodentals This, like fricativization, is a means by which consonants of adjacent syllables get voice and place features that are dissimilar as in (111).

111) English nounsrrowed into EkeGusii nativization by bilabialization of labiodentals English noun pronunciation EkeGusii nativized form pronunciation veranda /vǝrændǝ/ baranda -ɸaranda university /ju:nɪvǝsɪtɪ/ yunibasiti -juniɸasiti fine /faɪn/ baini -aini

(111) shows that the labiodentals /v/ and /f/ are bilabialized to either the voiceless bilabial

[ɸ] or its voiced counterpart [] depending on the vowel of the syllable which they make. If the vowel is of [Retracted Tongue Root] ([RTR]), as in veranda [ɸaranda], it becomes [-

VOICE] [ɸ]. However, if it is of [Advanced Tongue Root] ([ATR]) as in tv [tibii], it becomes [+ voice] [].

Bilabialization described, like fricativization, presupposes the constraints: *OCP(voice) and IDENT IO (F) among others. The English input /vǝrᴂnǝ/ ‘veranda’ in (111) is analyzed in tableau (42).

EkeGusii input: /ɸaranda/ ‘baranda’

Constraint ranking: OCP(voice) >> IDENT IO (F)

Input: / ɸa.randa / *OCP(voice) IDENT IO (F)

217 ɸa.randa * a. ☞ vǝrᴂndǝ *! b. Tableau (4.42): EkeGusii output of the input /ɸaranda/

This tableau shows that candidate (a) is the output because it obeys the constraint

*OCP(voice), which is ranked highly in EkeGusii language. (b) loses since it disobeys the constraint in a bid to be true to the input. IDENT IO (F), which (a) disobeys is inconsequential in deciding the output EkeGusii; in any case, breaking it is motivated by the voice dissimilation process explained above.

4.2.4.3 Nativization of nasal consonant clusters (NC) by pre-nasalization and homorganization In sub- section 4.1.2.1.3, it was observed that EkeGusii language does not have obvious consonant clusters. It was argued that any nasal consonant sequence in this language is treated as a single consonant. This, it was argued, is due to the fact that EkeGusii (a Bantu language) does not allow consonant clusters or complex margins. English on the other hand allows consonant clusters (section 4.1.1.3.1). This means that loaned nouns from English into EkeGusii with complex margins will have to undergo structural and phonological changes in order to be accommodated. One such modification strategy is that of vowel epenthesis which breaks the clusters (4.2.2.1). In this sub-section, two segments, a nasal and a consonant will be treated as one; a pre-nasalized consonant made by the process of homorganization as in (112).

112) English nouns with nasal consonant sequence borrowed into EkeGusii

English noun pronunciation EkeGusii nativized form pronunciation bench /bendʒ/ ebenchi [eβendʒi]

218 drink /drɪŋk/ turunki [eturuŋgi] stamp /stæmp/ esitambu [esitambu driver /draɪvǝ/ omontereba [omontereɸa] Each of the nativized words in (112) has a nasal plus consonant sequence. Following Herby and Downing (2005), Clements (1990), and Sievors (1981), this sequence is treated as a single unit and not a complex consonant or a cluster of consonant in EkeGusii. This single unit (a prenasal consonant) is achievable through pre-nasalization and homorganization

(Cammenga 2002). In this case, the consonantal element shares place feature with the nasal element. EkeGusii loaned nouns from English, eturunki ‘drink’ and ebaranda ‘veranda’ for example, will be homorganized as in (113).

(113) EkeGusii loan onun homorganization i) drink

Input : /e-/ + /-turunγ-i/ affixation eturun-ki prenasalization [eturuŋ-gi] homorganization [eturuŋgi]

(ii) veranda

Input: /e-/+/βaranr-a/ affixation /eβaran-ra/ prenasalization /eβaran-da] homorganization [eβaranda]

Adapted from Cammenga (2002)

In (113i), prenasalization derives the prenasal consonant [ŋg], while in (113ii), the prenasal consonant derived is [nd]. These prenasals are homorganic in that the consonantal 219 elements, like the nasals, have the same place of articulation feature. For [ŋg], it is [+velar]; while for [nd] it is (+alveolar). Of significance to be noted here is that the prenasals in the loan word forms in EkeGusii are treated as a single unit (a prenasal consonant), while they are treated as a nasal consonant cluster in their English forms as illustrated in figure (47).

English EkeGusii veranda baranda

/vǝ.rᴂn.dǝ/ [βa.ra.nda]

NC NC

‘a consonant cluster’ ‘a unit’ (one consonant)

Figure (46): English and EkeGusii treatment of nasal + consonant sequences

In optimality theory, EkeGusii handling of the sequence in figure (41) presupposes the markedness constraints *COMPLEX C and the faithfulness one FAITH C as analyzed in tableau (43).

ɸa.ra.nda (unit) Input: Constraint ranking: *COMPLEX >> FAITH C

Input: / ɸa.ra.nda (unit)/ *COMPLEX C IDENT IO (labiodental) ɸa.ra.nda (unit) * a. ☞ vǝrᴂndǝ (cluster) *! b. Tableau (4.43): EkeGusii output for the input / ɸa.ra.nda (unit)/

220 The optimal candidate in this tableau is (a) because it treats the nasal plus consonant

sequence as a single unit and not a cluster. As it has already been observed, EkeGusii does

not allow consonant clusters, unlike in English as (candidate (b) in this tableau.

The argument advanced here and elsewhere in this study that there are no nasal plus

consonant clusters in EkeGusii phonology; and that the combination nasal plus consonant,

is not farfetched. In Indonesian language (spoken in Indonesia), (Clements 1983), for

example, there is nasal consonant substitution as exemplified by (114).

114) Nasal consonant substitution in Indonesian

UF SF Gloss

a) i. /mǝN-pilih/ [mǝmilih] to choose to vote

ii. /mǝN-tulis/ [mǝnulis] to write

iii. /mǝN- kasih/ [mǝŋasih] to give

b) i. /mǝN-bǝlih/ [mǝmbǝlih] to buy

ii. /mǝN-dapat/ [mǝndapat to get

iii. /mǝN-ganti [mǝŋgati] to change

Adapted from Kager (1999)

(114) shows that when an input nasal |N|, underlyingly is followed by a voiceless obstruents as in (114a), the obstruent is deleted, leaving its place of articulation to the nasal. In (114ai) the nasal is the bilabial [m], in (114aii) the alveolar [n], while in (114aiii), it is the velar [ŋ], all represented by the archiphoneme |N| underlyingly. In data (114b), however, the voiced obstruents that come after the |N| underlyingly are retained in the outputs, forming a nasal

221 plus consonant cluster in Indonesian. The observation made about (114b) is that the underlying nasal represented by the archiphoneme |N| is also realized with the place of articulation of the following obstruent just as in (114a).

Following these observations therefore, it can be generalized that since in nasal plus voiceless obstruents, the obstruent is lost, thus remaining with only the nasal which adopts the place of articulation of the obstruents, there is only one consonant in such combinations, even in cases of nasals plus voiced obstruents as in (114b). In (114b), the nasal loses its place of articulation to the obstruents, and joins the obstruents in their places of articulation.

Therefore, there are no two places of articulation. This means that the new sound created by the two, like in (114a), is one and not a cluster of consonant plus nasal.

4.2.4.4 Nativization by vowel height harmonization and disharmonization As already observed, EkeGusii language is characterized by vowel harmony. Some of the

loan words, depending on their vowel composition, undergo vowel harmony. Specifically,

EkeGusii has mid- vowel perfect harmony and advanced and retracted tongue root pairing,

which are vowel height harmonies. Similarly, EkeGusii is characterized by vowel height

disharmony in which vowels in adjacent syllables in a word do not share height features.

(115a and b) gives an illustration of vowel height harmony and disharmony respectively. (115) EkeGusii loan noun nativization by vowel height harmony and disharmony a) Nativization by vowel height harmony English noun pronunciation EkeGusii nativized form pronunciation

coat /kǝʊt/ egoti [eγoti] basin bæsɪn/ ebeseni [ɛɛsɛni] lorry /lɒri/ erori [erori]

cheque /t∫ek/ echeki [t∫ɛki]

222 b) vowel height disharmony

English noun pronunciation EkeGusii nativized form pronunciation tractor /trʌktǝ/ ekeragita [ekeraγita] feet /fɪt ebuti [euti] break /breɪk/ ebureki [eureki]

(115a) shows vowel height harmony, while (115b) vowel height disharmony. In [ɛtͻt∫i]

‘torch’ for example, the prefix vowel [ɛ] is in height harmony with the root vowel [ͻ].

They are both mid vowels. In [ekeraγita] ‘tractor’, on the other hand, the prefix vowel [e] is in disharmony with the root vowel [a], while [e] is mid high, [a] is low. In Optimality theoretic terms, vowel harmony and disharmony as in (115) presuppose the following faithfulness constraints: HARMONY (V) McCarthy (2003), and FAITH V, as demonstrated by analysis the loanword [ɛntɛrɛa] in tableau (44).

EkeGusii input: [ɛntɛrɛa]

Constraint ranking: HARMONY (V) >> FAITH V.

Input/ ɛntɛrɛa / HARMONY (V) FAITH V a. ☞ ɛntɛrɛa * b. draɪvǝ *!

Tableau (4.44): EkeGusii output for the input /ɛntɛrɛa/

In this tableau, candidate (a) is optimal while (b) loses. This is because candidate (a) does not violate the highly ranked constraint * HARMONY (V) in the EkeGusii, while on the other hand, candidate (b) loses because it violates this highly ranked constraint. Indeed,

EkeGusii unlike English requires that vowels be in harmony.

223 Other languages with vowel harmony, in particular vowel height harmony, show a similar trend as in EkeGusii. For example, in Kera, spoken in South West Chad, like in most other languages with vowel height harmony, lower vowels and high vowels cannot be in harmony (Pearce, 2003). Pearce observes that when there is a high vowel anywhere in a word, the high feature will spread as illustrated in in (116). 116) Kera (Chadic) vowel height harmony Root word gloss suffixed form gloss vowel /gi:d-/ stomach /gi:di/ her stomach /i/ /ci:(r)- head /ci:ri/ her head /i/ /gu:d-/ behind /gu:di/ her behind /u/ /ka:s-/ hand /ka:sa/ her hand /a/ Adapted from Pearce (2003) In (116), the height of the vowel in the root spreads to the suffix. In /ka:s-/ for example, the low vowel /a:/ spreads to the suffix to form /ka:sa/ in the suffix form. This indeed is in support of EkeGusii occurrences observed in (115a).

4.3 Morphological nativization Section 4.2 deals with phonological nativization (segmental, phonotactic, supra-segmental changes and phonological processes) that account for the nativization of EkeGusii loans from English. The present section (4.3) deals with objective three of the study which analyzes the morphological changes that EkeGusii loan words from English undergo during nativization. It analyzes the morphological processes that account for morphological adjustments that EkeGusii loaned words from English undergo in order to be accommodated into the morphological system of EkeGusii. The changes are accounted for within Optimality Theory (Prince and Smolensky, 1993, McCarthy, 2006).

Morphological change takes place at three levels: base word level (root), vowel level and affix level (Kaspersky 1982). Analyses of morphological change of EkeGusii loaned nouns from English in this study focus on these levels and are explained by Optimality principles 224 and guidelines. In particular, the study shows which morphological ranking is favoured by

EkeGusii outputs, given the English loan nouns into EkeGusii as inputs; whether that of

English, the loaner language or that of EkeGusii the target language. This study employs the following morphological constraints in its analysis.

Faithfulness constraints

MAX IO-(MORPH) – which demands that there should be no morpheme deletion, an input morpheme should be maintained in the output.

MAX- OI (MORPH) – this demands that there should be no addition, an output morpheme should have an input correspondent.

DEP-IO (MORPH) -no (recipient) affix epenthesis

DEP -IO (V) - no vowel epenthesis

Markedness constraints

*COMPLEX (C) – no consonant clusters are allowed in syllable margins

ONSET- syllables must have onsets.

STRPRES- a structure preservation constraint which demands that the input structure must be preserved in the output

Alignment constraints

(ALIGN (AFX, R; RT, L))- demands that the right edge of an affix should be aligned to the left edge of a root; an affix should be a prefix.

(ALIGN (AFX, L, RT, R))- demands that the left edge of an affix should be aligned to the right edge of a root; an affix should be a suffix.

Sources: Prince and Smolensky (1993/94) & McCarthy (2006)

225 4.3.1 Nativization by nominal classification As observed in 4.1.2.1 morphology of the nouns in EkeGusii and English differs in one crucial manner; while EkeGusii nouns are classified into groups known as noun classes,

English nouns do not. In other words, as Demuth (2000) observes, Bantu noun classes tend to be realized as grammatical morphemes rather than independent lexical items; that the classification is part of the larger concordial agreement systems. English nouns on the other hand are characterized as independent lexical items.

EkeGusii, unlike English, enters into a system of pairs of prefixes (morphemes) that mark the semantico-syntactic (morphosyntactic) categories of singular and plural forms as demonstrated in (117).

117) EkeGusii noun classification Noun /ɛ-nγͻͻkͻ/ [ɛŋgͻͻkͻ] ‘hen’ singular 9a /t∫i - nγͻͻkͻ/ [t∫iŋgͻͻkͻ] ‘hens’ plural 10a

Adapted from Cammenga (2002)

The noun in (134), [nγͻͻkͻ] ‘hen’, is in classes 9a in the singular form and 10a in the plural form respectively. Thus, (134) shows that all EkeGusii nouns belong to one of the noun classes identified in the language (Ongarora, 2009; Camenga, 2002; and Whitely, 1965).

The noun classes are marked by the singular and plural prefix markers. In (119) for example, the prefix {e-n-} marks singularity, while the prefix {t∫i-n-} marks plurality.

In OT theoretic terms, the morphology of the noun [n-γͻͻkͻ] in its neutral, singular and plural forms is analyzed in tableaux (45), and (46) as follows.

Neutral form analysis

Input: nkooko /nγͻͻkͻ/ ‘hen’

226 The input presupposes the constraints: DEP IO (MORPH) - no affix epenthesis, and

(ALIGN (AFX, : R; RT, L)) (a word must have prefix). The constraints are ranked as follows.

DEP IO (MORPH) >> (ALIGN (AFX, : R; RT, L))

Input nkooko DEP-IO (MORPH) (ALIGN(AFX, : R; RT, L)) a. ☞ nkooko * b. e-nkooko *!

Tableau (4.45): EkeGusii neutral form output of the input {nkooko}

Candidate (a) is optimal in the tableau because it does not violate the serious constraints

DEP-I0 (MORPH) which prohibits epenthesis of an affix to the neutral noun. This is because such affixation will change the meaning of the noun, which in its neutral form means either a ‘hen’ or ‘hens’ in constructions such as:

nkooko teiyo [ŋgͻͻkͻ teijo]- ‘hen not there’ ‘there is no hen(s)’

DEP-IO is highly ranked because violating it means that the output will have a totally different meaning. However, violating (ALIGN (AFX, : R; RT, L)) is not as serious, at and therefore violating it has little significance in determining the output, at least in this sense.

Analysis of the singular and plural form of EkeGusii word form |e-n-γͻͻkͻ| > [en-γͻͻkͻ] >

[eŋgͻͻkͻ] is as follows.

Input: nkooko /nγͻͻkͻ/

227 Constraints and their ranking: (ALIGN(AFX,: R; RT, L)) >> DEP IO (MORPH).

Input: nkooko (ALIGN(AFX, : R; RT, L)) DEP IO (MORPH). a. ☞ e-nkooko * b. nk-ooko *!

Tableau (4.46): EkeGusii singular output of the input {nkooko}

Candidate (a) is optimal because it does not violate (ALIGN (AFX, : R; RT, L)), a highly ranked constraint in singular and plural form markings in the language, (b) loses because it violates the constraint. Violating it makes the noun neutral; it does not refer to any specific hen. Tableau analysis of the plural form is similar to that of the singular form analyzed above because plural forms, like singular ones, demand prefixation.

English nouns ont the other are not categorized in the same way as the EkeGusii nouns,; that is, in groups of morphemes paired in singular, plural dichotomy determined by prefixation. Instead, they are grouped, just like nouns from all other languages, in terms of countable versus non-countable, concrete versus non-concrete, regular versus irregular nouns among others. English Plural and singular forms for the noun ‘boy’ for example, can be analyzed in OT theoretic terms as in tableau (47) and (48) as follows.

Input: [bͻɪ] – singular form

The presupposed constraints here are ranked as follows: (ALIGN (AFX, : L; RT, R)) >>

DEP IO (MORPH), ranked as; DEP IO (MORPH) >> (ALIGN(AFX, : L; RT, R))

Input: boy DEP IO (MORPH). (ALIGN(AFX, : L; RT, R))

228 a. ☞ boy * b. boy-s * *!

Tableau (4.47): English realization of the input {boy}

Candidate (a) is optimal even though it violates (ALIGN (AFX, : L; RT, R)), because the violation is not fatal. A singular regular noun in English does not require an affix (suffix).

The output satisfies DEP IO (MORPH), a constraint which is highly ranked in this case.

Violating it in this case (and in particular in this word), will be fatal.

Opposite ranking as in (47) above determines the output in the plural form of the word as shown by analysis (48). Input: boy-s- plural form. Constraint ranking: (ALIGN (AFX, : L; RT, R)) >> DEP IO (MORPH)

Input: /boy-s/ (ALIGN(AFX, : L; RT, R)) DEP IO (MORPH) a. ☞ boy-s * b. boy *! *!

Tableau (4.48): English realization of the input {boy}{-s}

The optimal candidate here is (a). It satisfies the requirement in English which requires that plural forms of regular nouns be affixed with a suffix as embodied in the constraint

(ALIGN (AFX, : L; RT, R.))

The difference between the noun morphologies of English and EkeGusii analyzed so far means that EkeGusii loaned words from English undergo a number of changes in order to be accommodated in EkeGusii morphological structure. One such change is for the English nouns to enter into EkeGusii noun classes/groups as demonstrated in (118).

(118) Nativization of English nouns borrowed into EkeGusii by nominal classification 229 English noun EkeGusii nativized EkeGusii noun classes

Singular plural form singular plural scout scouts [sikaoti] 1.omo-sikaoti 2.aβa-sikaoti record records [rɛkͻti] 9.e-rɛkͻti 10. t∫i-rɛkͻti blanket blankets [raŋgeti] 14. oβo- raŋgeti 6. ama- raŋgeti pastor pastors [βaasita] 1. omo- βasita 2. aβa-βasita

(118) shows that the English nouns in EkeGusii fall into EkeGusii noun classes in their nativized forms. The word ‘blanket’ for instance, enters into classes 14 {oβo-} and 6

{ama-} for singular and plural forms respectively. This, in OT, means that the English forms (input), adopt different structural shapes and therefore violating the faithfulness constraints: FAITH (MORPHEME) (input morphemes are the same, no change), MAX OI

(Morpheme) (output morpheme must have an input correspondent, no addition; and markedness ones: STRPRES, (ALIGN (AFX, L; RT, R)), (a suffix) and (ALIGN(AFX, R;

RT, L)), (a prefix). To illustrate this observation, the English word blankets ‘blankets’, nativized as ‘ama- ranketi’ is analyzed in tableaux (49) and (50) for English and EkeGusii outputs respectively.

English output: {blanket}{-s}

Input: {blanket}{-s}.

Constraint ranking: STRPRES >> {(ALIGN (AFX, L; RT, R)), (ALIGN (AFX, R; RT,

L))}.

230 Input: blanket-s STRPRES (ALIGN (AFX, L; RT, R)) (ALIGN(AFX, R; RT, L)). a. blanket *! b. ama- ranketi *! * c.☞ blanket-s * Tableau (4.49): English output of the input {blanket}{-s}

This tableau shows that candidate (c) is the optimal because it preserves the input structure,

a constraint which is highly ranked in English concerning plural formation, it also obeys

the relatively high ranked constraint in English which demands that plurals be affixed with

a plural marker, a suffix. Candidate (a), loses because it violates SRTPRES, which

disallows a change of structure of the input in output. Candidate (b) on the other hand,

loses because it prefixes (wrong affixation) instead of suffixing as required by English,

besides being in violation of the STRPRES. This is compared to EkeGusii output in tableau

(50) as follows.

EkeGusii input: {ama}-{ranketi}

Constraint ranking: (ALIGN(AFX, R; RT, L)) >> {STRPRES, (ALIGN (AFX, L; RT, R))}

Input: ama-ranketi (ALIGN(AFX, R; RT, L)) STRPRES (ALIGN (AFX, L; RT, R)) a. blanket *! b. ☞ ama- ranketi * * c. blanket-s *!

Tableau (4.50): EkeGusii output of the input {ama-}{ranketi}

In this tableau, the alignment constraint which demands prefixation is ranked above the

rest, while, the constraint which demands for a suffix is ranked least. This underscores the

fact that languages rank constraints differently; while EkeGusii plural demands a prefix,

English demands a suffix.

231 Loan noun nativization by nominal classification is not a preserve of EkeGusii. Some other

Bantu languages behave in a similar manner. For example, in KiNyarwanda (Kagayime,

2010), loaned words are allocated to the nominal classes by the Allocation Theory. This kind of allocation depends on either the semantics of the loan noun or its morphology.

French loan nouns into KiNyarwanda behave as in (119).

119) French nouns in Kinyarwanda nominal classes.

Loan word form French form Class Gloss u-mu-arabu arabe 1 Arab a-ba-arabu arabes 2 Arabs u-mu-note minute 3 minute i-mi-nota minutes 4 minutes i--lonji orange 5 orange a-ma-lonji oranges 6 oranges i-gi-tari hectare 7 hectare i-bi-tari hectares 8 hectares i-katoti carotte 9 carrot za-karoti carrottes 10 carrots u-rufanga franc 11 franc a-ma-fanga francs 6 francs

(119) shows that French nouns enter into Kinyarwanda nominal classes. Every French noun, depending on its semantics and morphology, joins an appropriate KiNyarwanda nominal class.

232 Nativization process in KiNyarwanda through nominal classification resembles that of

EkeGusii. In both languages, the classes into which the various nouns enter are similar and are determined by the semantics and morphology of the nouns. The only difference between the two nativization processes is that while in EkeGusii, the augment vowel is homorganic to that of the prefix vowel due to vowel harmony which characterizes

EkeGusii phonology, in KiNyarwanda on the other hand, the augment vowel is in disharmony with that of the root.

Nominal class nativization in KiNyarwanda (Kayigema, 2010), is accounted for within the allocation theory, while this research accounts it within Optimality Theory. Allocation theory accounts for the distribution of the loans into nominal classes governed by their morphology and semantics, OT on the other hand accounts choice of nominal classes by loan words as competition among constraints. Allomorphic distribution in both languages are controlled by Dahl’s Law of dissimilation.

Languages without Meinhoff’s nominal classes do not nativize the same way. In other words, nouns in these languages do not recognize nominal classification. They therefore behave differently from those with nominal classes. For example, in Urdu, (Islam 2011), plurality of noun loans is marked by suffixation, like in English, as in (120).

120) English noun plural marking in Urdu

English noun plural form Urdu singular form Urdu plural form plate plate-s plat plat-a

233 glass glass-es glas glas-a book book-s buk buk-a building building-s bilding bilding-a

Adapted from Islam (2011)

(120) shows that plural of the English noun in Urdu is achieved through the suffixation of the suffix {-a} ([æ]) on to the root. This is not the case with EkeGusii and most Bantu languages. In Bantu languages, plurality is marked by prefixation (and not suffixation) and it is a function of nominal classes, in which it is a singular- plural number pairing of the same.

As observed in section (4.1.2.2), choice of nominal class by the nativized forms is not arbitrary; rather it is determined by the semantics of the root word. In other words, as

Givon (1972) observes, noun (stem/base) semantics determines prefixes choice. For example, the noun {-mura} ‘boy’ or ‘male youth’ falls within the semantic meaning of animate, human being and in its singular form, it takes the prefix {omo-} class 1, while in its plural form it takes the prefix {aβa-} class 2. Similar nouns behave the same way. Thus, in (118) above, the English words that are borrowed into EkeGusii, enter into a specific class determined by the semantics of the noun root/base. ‘scout’ for example, means an animate human being and therefore enters class (1), singular form, and class (2), plural form, as in (118).

Because semantics plays a major role in determining the membership of noun classes and their prefixes, this study briefly interrogates its role in the process of nativization of the

English loan words into EkeGusii in the following subsection.

234 4.3.1.1 The role of semantics in morphological nativization of English nouns in English It has been observed that a noun in EkeGusii, and indeed in most Bantus languages with noun class system (Henderiks & Poulos, 1990; Givon, 1972), enters into a specific class depending on its stem gender, number and size (section 4.1.2.2). In other words, class membership of a noun depends on its meaning and that it is this meaning which determines the kind of prefix to be affixed on to it. The chosen prefix puts the noun in the class it belongs to ( see table 2 above).

Indeed, as the nouns in EkeGusii enter into their classes on the basis of their semantic content, so are the loans from English into the language as further is illustrated in (121).

121) EkeGusii nouns from English and their classes/prefixes determined by semantics

Noun class gloss root meaning omo-gabana 1 governor animate being; human aba-gabana 2 govenors animate being; human obo-ranketi 14 blanket inanimate object ama-ranketi 6 blankets inanimate objects eke-ragita 7 tractor inanimate cultural object ebi-ragita 8 tractors inanimate cultural objects e-retio 9 radio inanimate object chi-retio 10 radios inanimate objects

(121) shows that the English nouns in EkeGusii are affixed with a class marking prefix which is determined by their root meanings or semantics. For example, the noun governor

235 enters class (1) and not any other class because of its semantic features. It is: [+ANIMATE,

+HUMAN BEING]. This class demands the prefix {omo-}. The word ‘tractor’ on the other hand falls within the semantic features: [-ANIMATE, -HUMAN BEING +OBJECT], and therefore enters its appropriate class- 7{eke-}. Thus, the English nouns entering EkeGusii morphology do not enter haphazardly; but rather they are determined by their semantics.

That is, depending on the meaning of the root of the loan, an appropriate class, which preserves the meaning of the input in the output, is determined and assigned.

This choice of nominal class by loan nouns as a result of their semantic features in

Optimality Theory, presupposes Faithfulness constraint which preserve the meaning of the input in the output form, that is , MAX IO (meaning). Because EkeGusii nouns must belong to a noun class, and that the noun class is marked by a prefix, an appropriate alignment constraint is also presupposed: (ALIGN (AFX, R; RT, L)), which demands that, an affix be a prefix. Therefore, the loaned word must be prefixed. This differs from affixation of plurality in English, which demands for suffix affixation (ALIGN (AFX, L;

RT,R)) (Prince and Smolensky, 2004, and McCarthy, 2006). Finally, the structure of the

English word as input changes in it nativized or output form. This means that the structure preservation constraint (STRPRES) (Golston & Yang, 2001; Aronoff, 1998; and Kiparsky,

1982) is presupposed. This constraint provides that the structure of an input form be preserved in the output (no change of structure form in the output). These constraints are ranked differently for English and EkeGusii outputs as analysis in tableaux (51) and (52) below show.

Input: {tractor}{-s}

236 Constraints and their ranking: MAX IO(meaning) >> {STRPRES, (ALIGN(AFX, L; RT,

R)), (ALIGN(AFX, R; RT, L))}.

Input: tractor-s MAX IO(meaning) STRPRES ALIGN(AFX, L; RT, R) ALIGN(AFX, R; RT, L) a. ☞ tractor-s * b. tractor *! * * c. ebi-ragita *!

Tableau (4.51): English output of the input {tractor}{-s}

Candidate (a) is the output because it only violates the relatively low ranked constraint in

English, which provides that there must be a prefix to mark class and other nominal

features, a feature not recognized by English. The rest of the other candidates lose because

they violate the highly ranked constraint MAX IO(meaning), for (b), which demands that

the meaning of the input be preserved in the output; and STRPRES in (c), which demands

that the structure of the input be preserved in the output. This is compared to EkeGusii

analysis.

EkeGusii input: ebi-ragita

Constraint ranking: MAX IO(meaning) >> (ALIGN (AFX, R; RT, L)) >> {(ALIGN (AFX,

L; RT, R)), STRPRES}.

Input: ebi-ragita MAX IO(meaning) (ALIGN(AFX,R; RT, L)) ALIGN(AFX, L; RT, R STRPRES a. tractor-s *! b. eke-ragita *! * * c.☞ebi-ragita * * Tableau (4.52): EkeGusii output of the input {ebi-}{ragita}

Candidate (c) is the optimal since it does not violate the constraint demanding that the input

meaning be preserved in outputs. This is the determining constraint. (b) loses because it

changes the meaning of the input from being in plural to singular. Candidate (a) loses 237 because it aligns the given prefix wrongly in EkeGusii; it is a suffix, yet EkeGusii demands a prefix.

The analysis of the role of semantics in morphological nativization and OT handling of the same is one of the major contributions of this research in theoretical linguistics. This is because available literature (Zivenge, 2009; Kayigema, 2010; & Islam, 2011 among others) indicate that morphological loan word nativization this far has not focused on the role that semantics plays in the process. None of these studies focuses on the role of semantics in the process of loanword nativization.

(121) above show that the English nouns are pluralized by suffixation (the suffix {-s}) in all the given cases. However, their plurals in EkeGusii nativized forms are prefixed (the prefixes differ as per the semantics of the noun root) as shown in (122).

122) Pluralization of English and EkeGusii nativized forms

English forms EkeGusii forms class semantics scouts /skaʊts/ [aβa-siikaouti] 2 animate, humam records /rkͻdz/ [t∫i-rɛkɛkͻti] 9 inanimate, object blankets /blᴂŋkɪts/ [ama-raŋgeti] 14 inanimate, obj ect pastors /pʌstǝz/ [aβa-βasita] 2 animate, human being governors /gʌvǝnǝz/ [aβa-γaβana] 2 animate, human being sacraments /sækrǝmǝmǝnt/ [ama-sakaramento] 6 inanimate, object

238 These data show that all nouns entering EkeGusii from English are affixed for class and number. This is because each word in EkeGusii belongs to a particular class and number.

Given the difference in affixation for plural marking between English words and their nativized forms in EkeGusii as indicated in (122) above, affix location constraints are presupposed, (Prince and Smolensky, 2004, and McCarthy, 2006). EkeGusii language demands the following affix location constraint: ALLIGN (AFX, R; RT, L), which states that align the right edge of an affix to the left edge of a root to mark plurality among other functions. To illustrate, EkeGusii noun ‘omote’ omo-te ‘tree’ is analyzed in tableau (53) below.

Input: o- mo- te

aug. 3.3PSG tree output: [omote]

This input presupposes the constraints: ONSET, (ALLIGN (AFX, R; RT, L)), (ALIGN

(AFX, L; RT, R)), ranked as follows: ALLIGN (AFX, R; RT, L) >> {ONSET, (ALIGN

(AFX, L; RT, R))} c. te-omo *! *

Candidate (a) is the output because it does not violate the constraint which requires that the right edge of an affix be aligned with the left edge of the root to which it is affixed. Its violation of the constraint ONSET and (ALIGN (AFX, L; RT, R)) is of little consequence in determining the output in EkeGusii. Candidate (b) loses because the affix has been affixed in the wrong part of the root, that is; to the right edge instead of the left edge as

239 demanded by the language. In essence, as McCarthy (2006) observes, the affix location alignment constraint, (ALIGN (AFX, R; RT, L)), declares that this affix be a prefix. This kind of affix location alignment affects both the singular and plural forms of EkeGusii. The plural form of the noun [omo-te] is [eme-te] ‘trees’. Its OT analysis will have similar results as in tableau (53), because the constraints and their ranking are similar. The constraint (ALIGN (AFX, R; RT, L)), which declares that this affix be a prefix means that the plural marking morpheme be a prefix.

However, constraint ranking is not the same in English language forms. In the plural form, the presupposed constraints will be: (ALIGN (AFX, L; Root, R)), which demands that the left edge of an affix be aligned to the right edge of a root, (ALIGN (AFX, R; RT, L)), and

*COMPLEX (C), which demands that there should not be a complex or cluster of consonant at the syllable margins. This is illustrated in tableau (54) which analyses the

English word records / rekͻ:ds/.

Input: /rekͻ:ds/ ‘records’

This presupposes the following constraints and their ranking: (ALIGN (AFX, L; RT, R))

>> {*COMPLEX (C), (ALIGN (AFX, R; RT, L))}

Input: /records/ (ALIGN (AFX, L; RT, R)) *COMPLEX (C) (ALIGN (AFX, R; RT, L)) b. ☞ record-s * * c. s-record *! *

Tableau (4.54): English output of the input {record}{-s}

(a) is the output in this tableau because the plural marker affix (which is a suffix in English) is correctly aligned even if it violates *COMPLEX (C), which prohibits consonant clusters at syllable margins. 240 The singular forms of the English nouns do not require an affix and therefore no affix location constraint is required. The relevant constraint in this case is MAX IO (meaning) which demands that there should be no change of meaning in the output, input meaning should be maintained. This is illustrated by the singular form ‘record’ /rekͻ:d/ as analyzed in tableau (55).

Input: {record}/rekͻ: d/

This presupposes the following constraints and their ranking: MAX-1O (meaning) >>

(ALIGN (AFX, L/R; RT, L/R)).

Input: /rekͻ:d/ MAX-1O (meaning) (ALIGN (AFX, L/R, RT L/R)) a. rekͻ:d-s *! b. s-rekͻ:d *! c. ☞ rekͻ:d *

Tableau (4.55): English output of the input {record}

Candidates (a) and (b) lose in the tableau because they are affixed; affixation and prefixation respectively. These affixations are banned in singular forms of English which ranks them highly in the language and which demands that a singular form should not be affixed with any morpheme in English. (ALIGN (AFX, L/R; RT, L/R)) is satisfied in (a) and (b) because the candidates are affixed as required by the constraint: (a) aligns the left

241 edge of an affix to the right edge of a root - a suffix); while (b) aligns the right edge of an affix to the right edge of a root- a prefix). However, this satisfaction is inconsequential because the constraint is relatively low ranked in the language regarding singular forms in the grammar of English.

The data indicate that all the English nouns borrowed by EkeGusii have to be nativized; that is, they have to enter into a given a noun class. These classes, as has been observed in this subsection, are marked by prefixation. The English noun loans into EkeGusii are therefore prefixed in order to be admitted into the various EkeGusii noun classes. The constraint which demands for this prefixation as has been observed is (ALIGN (AFX, R;

RT, L)) that is; align the right edge of an affix to the left edge of a root. To illustrate,

EkeGusii loaned word [ɛrɛkͻti] ‘record’ is analyzed in tableu (56).

Input: {e}-{rekoti} /rekͻ:d/ ‘record’

{e-}{rekoti} [ ɛ- rɛkͻti ]

9.3.3PSG- record

This means that the word has been nativized into class 9 and that it is in the third person singular. This presupposes the following constraints and their ranking: (ALIGN (AFX, R;

RT, L)) >> DEP IO (MORPH).

Input: e-rekoti (ALIGN (AFX, L/R; ROOT L/R)) DEP IO (MORPH). a. ☞ e-rekoti * b. record *! c. record-s *!

Tableau (4.56): EkeGusii output of the singular input {e-}{rekoti} 242 Candidate (a) is optimal because it does not violate the alignment constraint which is highly ranked in EkeGusii. Violating it is fatal because the given word will not be prefixed for class and therefore will not be classified. The loaned word in tableau (54) above has been effectively prefixed and nativized into class 9 marked by the prefix {e-} or {ɛ-}.

The right edge of the prefix {ɛ-} is correctly aligned to the left edge of the root {-rɛkͻti} as demanded by the constraint. The constraint DEP IO (MORPH) is of no consequence here, though it is of great significance in determining English outputs, where it is relatively highly ranked.

EkeGusii plural form of the word {ɛrɛkͻti} ‘record’, behaves in a similar manner in terms of affixation only that changing it to plural will change it in nominal class and number as illustrated in tableau (55) below.

Input: records / rekͻ:ds /

Output: chi-rekoti [t∫i- rɛkͻti] ‘records’

10.3PPL record

Here, the noun is in class 10 and in plural. The constraints pre-supposed are the same as those used in the analysis of the singular form in tableau (56) above: (ALIGN (AFX, R;

RT, L)) >> DEP IO (MORPH).

Input: chi-rekoti (ALIGN (AFX, L/R, RT L/R)) DEP IO (MORPH). a. ☞ chi-rekoti * b. record *! c. record-chi *! Tableau (4.57): EkeGusii output of the input {chi-}{rekoti}

243 Candidate (a) is optimal since it violates the less serious constraint in the tableau, DEP IO

(MORPH). The rest of the candidates violate the serious constraint and therefore are fatal violations.

Most of the borrowed nouns into EkeGusii from English, it is observed, seem to favour certain classes over others. In fact, majority of the borrowed nouns enter classes: 9 (e-) ,10 chi-, 9 (a) e-n- and 10 (a) t∫i-n-; a few enter classes 1 {omo-}, 2 {aβa-} and 6 {ama-}; and in rare cases into other classes such as 14 {oβo-} as in {obo-}{ranketi} ‘blanket’ The rest of the classes do not seem to be favoured at all. This is because most of the borrowed nouns name newly invented things, objects and names of places (institutions), and that these nouns belong to the mentioned classes Kayigema (2010).

The following subsections give a description and analysis of how EkeGusii loan words from English are nativized into EkeGusii nominal classes. The nominal classes on focus are: 1,2, 3,4, 14, 6, 7, 8 and 12.

4.3.1.2 Nativization into classes 1 and 2 Nouns Classes 1 and 2 nouns refer to human beings, with class 1 denoting the singular form of the noun, while class 2 denotes the plural. In EkeGusii, the augment in the singular form is

{o-}, while in the plural it is {a-} as in o-mo-nto ‘person’ and a-ba-nto ‘persons’ respectively. The nominal prefix on the other hand is {mo-} in the singular form and {ba-} in the plural form. Loaned words from English entered these classes in EkeGusii as (123) shows.

123) Nativization of English nouns borrowed into EkeGusii in noun classes 1 and 2

English noun EkeGusii form nativized form (class 1) nativized (form class 2)

244 chief chiibu o-mo-chiibu [o-m-t∫i-iβu] a-ba-chiibu [aβa-t∫iiβu] governor gaabana o-mo-gabaana [omo-γa-aɸana] a-ba-gabana[aβa-γaaɸana] pastor baasita o-mo-baasita [omo-βa-asita] a-ba-baasita [aβa-βaasita] councilor kansara o-mo-kansara [omo-ka-anzara] a-ba-kansara [aβa-kaanzara ]

All the English nouns borrowed into EkeGusii in (123) have common semantic features; they refer to animate beings (specifically human beings). Upon entering EkeGusii language, the English words are morphologically nativized as has already been observed.

In their singular forms, they are prefixed with the singular and person marker prefix

{omo-} of nominal class 1, while in their plural form, the prefix changes to {aβa-} of nominal class 2, which marks plurality and person. This is unlike in their English forms where in the singular form, it is not affixed at all; while in the plural it is suffixed. As has been observed in this research, these affixations in EkeGusii and the non-affixation in the

English singular form presuppose the OT markedness constraint (Align (Afx, R; root, L)), which demands that the affixes have to be prefixed, which is in violation of faithfulness constraints such as ONSET and DEP IO (MORPH), which prohibit onsetless syllables and epenthesis of an affix (morpheme) respectively. These arguments are captured in tableaux

(58) and (59) for EkeGusii and English outputs for the English inputs {chief} and {chief}{- s}.

Input: {o-}{mo-}{chiibu} ‘omochiibu’

Constraint ranking: (Align (Afx, R; root, L)) >> {DEP IO (MORPH), ONSET}

Input: 0-mo-chiibu (ALIGN (AFX, L/R, RT L/R)) DEP IO (MORPH) ONSET a. ☞ o-mo- chiibu * * b. chief *!

245 c. chief-s *! Tableau (4.58): EkeGusii output of the English input {omo-}{chiibu}

In this tableau, the optimal candidate is (a) because it obeys the alignment constraint which is highly ranked in EkeGusii. (b) is not optimal because it disobeys the alignment constraint which leads to a change of meaning of the input from being singular to being neutral. (c) loses because it does not only affix a nonexistent affix in EkeGusii, but also a wrong alignment; a suffix instead of a prefix. The plural form ‘a-ba-chiibu’, chiefs will have a similar analysis. This is compared to the English realization in tableau (59).

Input: {chief}

Constraint ranking: {DEP IO (MORPH), ONSET} >> (Align (Afx, R; root, L))

Input: {chief} DEP IO (MORPH) ONSET (ALIGN (AFX, L/R, RT L/R)) a. o-mo- chiibu *! b. ☞ chief * c. chief-s *! Tableau (4.59): English output of the input {chief}

Candidate (b) is the output in this tableau because it is faithful to the input as required by the constraint DEP IO (MORPH), which is the most highly ranked of the given set of constraints. It requires that the singular forms of the input be maintained in shape in their outputs. The plural form of English, unlike that of EkeGusii requires an alignment constraint which demands for a suffix (English plural is marked by suffixation) and not a prefix.

246 4.3.1.3 Nativization of English loans into EkeGusii in classes 1b {mo-} and 2b [ba-}

These classes have the characteristics of classes 1 and 2 only that they lack the augment element as shown in the loan words (124).

124) Nativization of English nouns borrowed into EkeGusii in noun classes 1Ø and 2Ø

Source noun nativized form 1bØ- 2bØ- chief |t∫i-ɸu| [mot∫iβu] [ɸat∫iiβu] pastor |ɸasita| [moɸasita] [ɸaɸaasita] councilor |kansara| [mokanzara] [ɸakaanzara]

father |ɸaata| [moɸaata] [ɸaɸaata]

In all the cases in (124), both the singular and the plural forms of the nativized nouns are marked by {Ø-} in both classes. As Cammenga (2002) observes, these are non-augmented forms which are acceptable in the language under certain circumstances as in, [t∫iiβu taijͻ]

‘chief(s) is not there’, in a case where somebody was checking if there is a chief(s) present.

Here, the root may carry the meaning of plural or singular. Therefore, form classes considered here are instances of lexically determined allomorphy. Cases of zero ([Ø-]) prefixation as Cammenga observes are rare.

4.3.1.4 Nativization into classes 3 and 4

These classes according to Kayigema (2010) denote to things like trees, ditches, rivers, natural phenomena, and some parts of the body. Class 3 denote singular forms while 4 denote plurals.

247 Only one word was collected into these classes as (125) shows nativization of loans into classes 3 and 4.

125) Nativization of English nouns borrowed into EkeGusii noun classes 3 and 4

Word nativized form class 3 class 4 motor car /tͻkaa/ [tͻkaa] ͻmͻ-tͻkaa ɛmɛ-tͻkaa

Nativization in these classes were found to be like nativization in classes 1 and 2 above.

4.3.1.5 Nativization in classes 5 {eri-/rii-}, 6 {ama-}, 7 {eke-}, 8 {ebi-}, 12 {aka-} and 14

{obo-}

As observed in section 4.1.2.2.7, these classes are marked by a combination of corresponding singular and plural prefixes as in (143) below, repeated from section

4.1.2.2.7

126) EkeGusii noun classes 5, 6, 7,8,12 and 14 prefixes

Singular plural

(a) 7 eke - 8 eβi-

12 aka - 8 eβi-

12 aka - 14 oβo-

(b) 5 eri- 6 ama-

As observed already, whenever a word belonging to some other class is transferred to any of the classes in (126a) at least the idea of diminution is necessary added to its basic meaning. Whenever some such word is transferred to the class in (126b), at least the idea of augmentation is added to its basic meaning. Loan words too behave the same way. Words

248 from other classes transferred into the classes in (126a) above get the idea of diminution and those entering (126b) get the idea of augmentation as demonstrated in (127).

127) Nativization of English nouns in EkeGusii by diminution and augmentation i) Source word nativized form classes grease /gri:z -ris-i 7/eke-risi/ [ekerisi] 8 [eɸi-risi] ‘neutral’ school /skʊl/ -sukuur-u 7/eke-sukuuru/[eγe-sukuuru]8[eβisukuuru] ‘small’schools

12/ aka-sukuru/ [aγasukuru] 8[oβosukuru] ‘small schools’ skirt /skɜ:t/ -sikaat-i 7/eke-sikaati/ [eγesikaati] 8 [eβisikaati] ‘smmal skirt’

12/aka-sikaati/ [aγasikaati] 8 [eβi-sikaati] room /ru:m/ -rum-u 12/aka-ruumu/ [akaruumu] 14 [oβoruumu] ‘small rooms’

7/eke-ruumu/ [ekeruumu] 8 [eβiruumu]

ii) Source word nativized form classes torch /tͻ:t∫/ /tͻt∫i/ 5 /rii-tͻͻti∫/ [riitͻͻt∫i] 6[amatͻͻt∫i] ‘big torches’ governor /gʌvǝnǝ/ /kaɸana/ 5/rii-kaɸaana/ [riiγaβaana] 6[amaγabaana]‘big governors’

(127) shows that loaned words belonging to a given class when transferred to any of the classes identified in (127i) will be deminutivized. For example, the word ‘skirt’ is borrowed into classes 9 {e-} in singular and 10 {t∫i-} in plural, [e-sikaati] and [t∫i-sikaati] respectively. However, as data (127i) show, the word can be transferred into classes 7 and

8 and get the meaning of diminution: 7 [aγa-sikaati] ‘a small skirt’, 8 [eβi-sikaati] ‘small skirts’.

249 Words borrowed into other classes and then transferred to classes 5 (rii-) and 6 (ama-), get augmented as in (127ii). For example, the word ‘governor’ is nativized into classes 1 [omo-

γaβaana] for singular, and 2 [aβa- γaβaana] for plural respectively. When transferred to classes 5 (rii-) and 6 (ama-), it gets the meaning of augmentation (big in stature); ‘a big governor’ (which may be pejorative or non-pejorative) in class 5, and big governors in class 6.

Other Bantu languages such as KiNyarwanda, (Kayigema, 2010) and Tong, (Zivenge,

2009), treat loaned words in a similar manner. In other words, the loan words into these languages are nativized through nominal classification. For example, in KiNyarwanda

(spoken in Rwanda), French words into it are nativized as in (128).

128) French noun nativization in Kinyarwanda.

French noun Kinyarwanda morphological form nominal class gloss chauffeur u-mu-shoferi 1 driver chauffeurs a-ba-shoferi 2 drivers médaille u-mu-dari 3 medal médailles i-mi-dari 4 medals coat i-koti/ri-koti 5 coat coat a-ma-koti 6 coats quinine i-kinini 7 tablet quinine ibi-kinini 7 tablets 250 (128) shows that French loans into KiNyarwanda, like those of EkeGusii are allocated particular noun classes dependent on the semantic features of the noun. The noun

‘chauffeur’ (driver) for example, enters class 1 for singular and 2 for plural, because these are classes reserved for the semantic features [+animate, +human]. The noun ‘médaille’

(medal) on the other hand is allocated classes 3 and 4 because it is characterized by the features; [+inanimate, -human, +object].

4.3.2 Prefixation and pre-prefixation

The previous section has analyzed how EkeG English nouns in EkeGusii are nativized in the various nominal classes present in EkeGusii. One of the main determinants of these classes as was observed is affixation. The prefix, it was observed, determines whether a noun belongs to class 1, 2 or 3 among others. The prefixes on the other hand are determined by the semantics of the roots of the loan nouns. In this section, the nature of the prefix determining noun classes of the loan words and how the loan words from English are prefixed in order to be accommodated into the morphological structure of EkeGusii are analyzed.

It has been observed that EkeGusii has two types of prefixes; that, is the prefix per-se and the pre-prefix (augment) (Cammenga, 2002; Ongarora, 2008; and Whiteley, 1965). Section

(4.3.2.1) below deals with the prefix, while (4.3.2.2) analyzes the pre-prefix.

4.3.2.1 Nativization by prefixation Cammenga (2002) observes that EkeGusii roots are regularly prefixed by at least one of the morpho-syntactic class prefixes. The kind of prefix affixed on to a root depends on the

251 semantic content of the noun root (Gívon, 1972; Henderikse and Poulos, 1990). This is illustrated in (129).

(129) Nativization of English nouns borrowed into EkeGusii by prefixation

English noun EkeGusii form prefix class prefixed form carrot /kᴂrǝt/ -karati 9 {e-} eka-rati

blanket / blᴂŋkɪt/ -raŋgti 14 {bo -} bo-raŋgeti ticket /tɪkɪt/ -tikɛti 9 {e-} e-tikɛti cabbage /kᴂbɪdʒ/ -kaβit∫i 9 {e-} eka-βit∫i

(129) shows that whenever an English noun enters EkeGusii morphology, it undergoes class prefixation in order to be accommodated. The prefix chosen by a noun is not arbitrary; it is determined by the semantics of the noun. Katamba (1993) observes that nouns in Bantu are grouped into classes often on a minimally semantic basis which is dependent on what the nouns refer to, whether human/animate or on the basis of other important properties denoted by the noun. For example, the English noun ‘carrot’ falls into class prefixes 9{e-} for singular and 10{t∫i-} for plural. Classes 9 and 10 prefixes accommodate nouns within the semantic content of animals, people, body parts, tools, instruments, household effects, natural phenomena among others. A large number of nouns are accommodated within these semantic content classes. This explains why most of the borrowed words fall into the classes.

The prefix has CV syllable structures except that of class (9) which has a syllable structure of V as in (146) above. In OT theoretic terms, this prefix structure presupposes dominance of the markedness constraint ONSET over (Align (AFX, R; RT, L)), ranked as ONSET >>

252 (ALIGN (AFX, R; RT, L)). The nativized form of the word ‘blanket’, (146) can be analyzed in tableau (60).

Input: {bo-}{ranketi}

Input: bo-ranketi (ALIGN (Afx, R; Root, L)) DEP IO (MORPH) a. ☞ bo- ranketi * b. ranketi *! c. blanket *!

Tableau (4.60): EkeGusii output of the English input {bo-}{ranketi}

Candidate (a) is the output because it does not violate the serious constraints. Violating

DEP IO (MORPH) is not as fatal as violating the alignment constraint. (ALIGN (AFX, R;

RT, L)) demands that loans be affixed with prefixes.

The shape of the allomorphs of EkeGusii, like many other Bantu languages Kikuyu,

(Mwihaki, 1998), Kitharaka (Mberia, 2004) KiKamba (Mutua, 2007); and KiNyarwanda,

(Kayigema, 2010) among others) prefixes are determined by Dah’l Law of voice dissimilation discussed at length in section 4.2.4.1 above.

Some other Bantu languages also nativize loaned nouns by prefixation. For example,

Tonga regularly prefixes loaned noun to mark class just like in EkeGusii as illustrated by

(130).

130) Tonga English noun prefixation

English Tonga morphological form prefix function apostle mu-positoli mu- class 1 marker apostles va-positoli va- class 2 marker machine mu-china mu- class 3 marker 253 machines mi-china mi- class 4 marker girl ri-gelu ri- class 5 marker girls ma-gelu ma- class 6 marker school chi-kolo chi- class 7 marker schools zvi-kolo zvi- class 8 marker

Source: Zivenge (2009)

(130) shows that, Tonga, like EkeGusii and most other Bantu languages, nativize noun loans by prefixation to allocate them appropriate nominal classes determined by the semantics of the given noun. For example, the noun ‘apostle’, positoli in Tonga, is prefixed with {mu-}, which carries the semantic features [+animate, +human] of class one. Its plural form {va-}, marks class 2. The difference between Tonga and EkeGusii as data (130) shows is that, while the Tonga prefix is a strict{CV-} syllable form, EkeGusii prefix allows an augment and therefore is a {(V)CV-} form. (130) also shows that unlike EkeGusii prefix, (and quite uncharacteristically of Bantu phonology), the Tonga nominal prefix

({zvi-}), has a cluster of consonants or a complex margin.

4.3.2.2 Nativization by pre-prefixation An augment (pre-prefix) is the vowel that is affixed to the prefix in Bantu lexical items

(Kayigema, 2010). According to Kayigema, common nouns of all kinds allow an augment, while proper nouns, and other nouns denoting kinship terms, places among others, do not take the augment. Different languages use different vowels as augments depending on a number of phonological factors such as whether a language is characterized by vowel harmony or not. KinNyarwanda for example, utilizes four vowels or augments {u-, o-, i-, a-}, while EkeGusii has only three as illustrated by (131).

254 131) EkeGusii augments

Augment nominal classes prefixed

{a-} 2,6,12 as in a-ba-nto ‘people’, a-ma-riso, ‘eyes’, a-ka-gaaka,’small old man’

{e-} 4,5,7,8,9 as in e-mete(tree), e-riso(eye), e-geita(gate), e-bi-ita(gates),esese (dog)

{o-} 1,3,14,15 as in o-mo-onto (person), o-bo-koombe (hoe) o-ko-gooro (leg)

The rest of the remaining nominal prefix classes (10, 11, 16, and 21) do not take augments.

Cammenga (2002) observes that in most instances, the full EkeGusii prefix properly consists of an augmented prefix, that is, the classical Bantu combination of an augment vowel, {v-}, also called pre-prefix or initial vowel, with a prefix proper, usually consisting of a consonant plus a vowel, {cv-} as discussed in section 4.2.2.2.1. This means that underlyingly, the representation of a full EkeGusii prefix has the form {v-cv-}. According to Cammenga this form covers nominal prefixes in at least classes 1-8 and 10-15. Indeed, as Cammenga (2002) and Kayigema (2010) observe, the presence or absence of the augment is determined lexically by lexical category membership or lexically determined allomorphy. Common nouns of all types are normally pre-prefixed across Bantu languages.

Nouns denoting proper names, kinship terms and places among others, on the other hand are not augmented as illustrated by Kinyarwanda examples in (132).

132) ( i) KiNyarwanda nominal augmentation Noun morphological form nominal class gloss umuntu u-mu-ntu 1 person abantu a-ba-ntu 2 persons umuserebenya u-mu-serebenya 3 lizard imiserebenya i-mi-serebenya 4 lizards

255 (ii) KiNyarwanda non-augmentation Noun nominal class gloss

-data 1a Ø- my father

-nyogukuru 1a Ø- my grand mother

-Kigali 1aØ- name of a place

(Kigali)

- Kivu 1aØ- name of a lake (Kivu)

Adapted from Kayigema (2010)

In (132i), the nominal category of common nouns which allow augmentation. (132ii) on the other hand gives a category of nouns that denote kinship and place, and therefore do not allow augmentation. This is in support of EkeGusii morphological behavior regarding augmentation.

With respect to EkeGusii nouns (the focus of this study), lexical category membership determines that nominal prefixes in morphosyntactic classes 1-8 and 11-15 may or may not be augmented, whereas the other prefixes, 1b {Ø-}, 9 (a) {e- (n)}, 10 (a) {t∫i-(n)}, 16{a-} and 21{ɳ-} are not augmented (Cammenga, 2002; Kayigema, 2010). It therefore means that borrowed words from English into EkeGusii that fall within the morphosyntactic classes 1-8 and 11-15 may or may not be augmented as illustrated by (133).

133) prefixation with augmentation of English nouns borrowed into EkeGusii

Word morphological form class and number gloss i) omogabana [omoγaβana] o-mo-gaban-a 1 SG governor ii) abagabana [aβaγaβana] a-ba-gaban-a 2 PL governors iii) risiti [risiti] ri-sit-i 5SG receipt 256 iv) amarisiti [amariisiti] a-ma-risit-i 6PL receipts v) ekereti |ekereti| [egereti] e-ge-ret-i 7SG crate vi) ebireti [eβireti] e-bi-ret-i 8PL crates vii) agasukuru |akasukuru| [aγasukuru] a-ga-sukur-u 12SG micro school viii)ebisukuru [eβisukuru] e-bi-sukur-u 8PL micro schools ix) oboranketi [oβoraŋgeti] o-bo-ranket-i 14SG blanket x) amaranketi [amaraŋketi] a-ma-ranket-i 6PL blankets

The following observations are made about (133). Firstly, with the exception of classes 1 and 2, very few loans are admitted into the rest of the classes. In fact, some of the classes

(3, 4, and 15), did not admit any; while classes (5) and (14) admitted one loan each.

Secondly, the prefixes in each class have two elements: initial vowel (the augment) and the prefix per se. In all the cases however, the augment is not compulsory. It may or may not be there, though its absence leads to a difference in meaning as was observed in section

4.1.2.1.2. For example, (133i and ii) above can do away with the augment as in (134).

134) prefixation without augment i) mogabana [moγaβana] mo- gabana 1SG governor ii) bagabana [βaγaβana] ba- gabana 2PL governor

(134) shows that a prefix can do without an augment and still carry the gender and number features of the noun it is attached to. Pre-prefixation in (134) above presupposes the constraints: ONSET (syllables must have onsets) and DEP V (which prohibits vowel epenthesis, either prothesis or anaptyxis). ONSET in this particular case is ranked higher than DEP V; that is DEPV is dominated by ONSET. Thus: ONSET >> DEP V. This is analyzed in tableau (61). 257 Input: {gabana}

Input: gaβana DEP IO (V) ONSET a) ☞ o-mo-gaβana * *

b) mo-gaβana *!

c) gaβana *!

Tableau (4.61): EkeGusii output of the input {gaβana}

This tableau shows that candidate (a) is the winner. The candidate wins because it obeys the constraint DEP V, which is ranked higher than ONSET. EkeGusii prohibits onsets especially in nouns that refer to particular number and gender.

CHAPTER FIVE

SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

5.1 Summary

This study investigates the nature of phonological and morphological features and processes that characterize nativization of English Nouns borrowed into EkeGusii in order to understand the phonology and morphology of EkeGusii. The study examines how

English nouns borrowed into EkeGusii are adjusted phonologically and morphologically in

EkeGusii phonological and morphological environments in order to be accommodated. The study targeted the phonological and morphological processes and features that account for the differences between English and EkeGusii phonology and morphology. Thus, the selected phonological and morphological processes and features were those that enabled the observation and accounting for the phonological and morphological changes that affect

English nouns entering EkeGusii linguistic environment. This was achieved through a step-by-step procedural exploration of the objectives of the study in Chapter four. This

258 chapter gives a summary of the findings, conclusions, recommendations and suggestions for further research.

The first objective of the study describes the phonological and morphological structures of the two languages under investigation- EkeGusii and English. Phonologically, findings indicate that the vowel systems of the two languages differ. While EkeGusii has a total of fourteen pure or monophthong vowels, English has twenty-five: twelve monophthongs, eight diphthongs, and five triphthongs. It was also established that the acoustic nature of

EkeGusii vowels as produced by native speakers of the language differs significantly from that of English.

Another finding is that the two languages under investigation have some consonants found in both, while other consonants are found in only one of the languages and not the other.

This is one of the main contributions of this study.

Some phonological processes were found to affect EkeGusii noun phonology and not the

English noun phonology. These include: feature dissimilation, prenasalization, homorganization, declusterization of nasal consonants and consonant glides, defricativization and nasal re-syllabifiation.

Phonotactically, the study established that the syllable structures of the two languages are different in that, while EkeGusii is a strict (V)CV language, English is a (C) V (C), language in which case the consonants can be in cluster forms depending on the word in question. It was further observed that English, unlike EkeGusii allows clusters of consonants of up to three in the onset and four in the coda positions of the syllable.

259 Prosodically, findings of this study established that while EkeGusii is a tone language,

English is a stress language.

Morphologically, the findings established that EkeGusii nouns like those in most other

Bantu languages, are grouped into morphosyntactic class systems in which the classes tend to be realized as grammatical morphemes rather than independent lexical items. This is not the case with the English noun, which is realized as an independent lexical item in most cases. EkeGusii noun morphology, like the morphology of other Bantu languages, it was further established, is characterized by a pre-prefix (augment). This is not the case in

English.

The second objective of the study analyzed the phonological changes that English nouns borrowed into EkeGusii undergo during nativization. Analyses were carried out within

Optimality Theory.

Segmentally, the findings of the study show that English sounds not present in EkeGusii are substituted for those present in EkeGusii phonology. The substitution involves phonemic and feature change which is consistent with OT which provides that languages rank constraints differently depending on their grammar and that it is this ranking which is responsible for language differences.

Findings further established that English diphthongs and triphthongs are monophthongized in EkeGusii. This process is also consistent with OT’s argument that puts markedness constraints in conflict with faithfulness constraints. Thus, the markedness constraint

*COMPLEX (V) is in conflict with the faithfulness constraint DEP IO (FEATURE) in this

260 case. Analyses indicated that EkeGusii ranks *COMPLEX (V) higher than DEP IO

(FEATURE), while the opposite is true in English.

Phonotactically, findings of the study show that EkeGusii, unlike English is a strict (C)V language. All the foreign syllable structures of English were re-syllabified in EkeGusii to conform to its syllable structure. English syllables with complex margins were changed by epenthesis, which broke the complex margins and opened the syllables in EkeGusii. OT’s explanation of this observation is that, while English allows consonant clusters and complex onsets and codas, EkeGusii disallows them. Thus, EkeGusii ranks *COMPLEX

(C) constraint higher as compared to English.

Analysis further show that closed syllables from English into EkeGusii were opened through epenthesis (paragogic). In OT, this is explained by a number of constraints such as, *CODA, MAX IO (SEG), and IDENT IO. Findings established that*CODA is ranked higher in EkeGusii as compared to English; while MAX OI(SEG) ranks high in English as compared to EkeGusii.

Suprasegmentally, the findings established that English nouns with stressed forms entering

EkeGusii are tonemized. Thus, the change of the feature stress in English to the feature tone in EkeGusii is as explained in OT by the constraints IDENT IO (FEATURE) and

SPEC (T), which demands that each tone bearing unit (TBU) must have a corresponding tone. It was established that while English prefers stress by ranking IDENT (FEATURE)-

STRESS highly as opposed to SPEC (T), EkeGusii does the opposite.

Findings further show that phonologically, a number of processes characterize nativization.

Such processes include voice dissimilation, phoneme fricativization or spirantization,

261 phoneme defricativization, phoneme bilabialization, and vowel harmony and disharmony.

Analyses indicate that these processes are governed by EkeGusii constraint ranking. For example, consonants in the English nouns in EkeGusii undergo voice dissimilation. This process is determined by Optimality Theory markedness constraint *OCP (VOICE), which is ranked over the faithfulness constraints such as IDENT IO (FEATURE). This finding is one of the major contributions of this study since no known study has ever targeted voice dissimilation in loan words.

Another finding that is of significance to this study is that nasal consonant clusters from

English lose their cluster status through prenasalization and hormorganization. Treating these combinations as single units in EkeGusii is supported by OT markedness constraint which bans complex vowels, *COMPLEX (C).

The third objective of the study analyzes the morphological changes that English nouns borrowed into EkeGusii undergo before they are accommodated in the morphological system of the language. Analyses focused on nominal classification, prefixation and pre- prefixation.

Findings show that English nouns enter the nominal classes (a characteristic of EkeGusii morphology), and that nominal classification is determined by the semantic features of the borrowed noun. This characteristic, in Optimality Theory, is explained by principles which preserve input meaning in the outputs, such as MAX IO (meaning); principles dealing with affix alignment such as (ALIGN (AFX, R;RT, L)) and (ALIGN(AFX, L;RT, R)); and principles which preserve structure such as (STRPRES). The finding that semantics plays a

262 major role in morphological nativization is another major contribution of this study to theoretical linguistics because it shed light on the role of semantics in nativization.

Analysis indicate that affixation processes in the languages under study differ. Optimality

Theory handles affixation using Affix Alignment Principles among other constraints.

Findings show that during morphological nativization, the English plural marking suffix {‘- s} is dropped and prefixes used in its place in the nativized forms. The prefixes used to mark EkeGusii plurals are determined by the class to which the noun in question belongs.

Therefore, English nouns in EkeGusii are prefixed for plural variously. This feature, besides being governed by alignment constraints of OT, is also explained by featural markedness constraints such as *OCP (VOICE) and *VTV, among others. This occurrence makes contribution to theoretical linguistics because it sheds light on the role of affixation in nativization.

5.2 Conclusions

Based on the findings of this study, the following conclusions are drawn. Firstly, the phonological and morphological systems of EkeGusii and English are significantly different. Phonologically, the phonemic, phonotactic, prosodic and phonological processes between the languages are different; while morphologically, noun classification systems and affixation processes differ quite significantly between the two languages.

Of significance to note are the phonological findings that: EkeGusii and English vowel segments differ acoustically as spectrographic analyses show; English stress is tonemized in EkeGusii; and EkeGusii phonological processes not present in English such as voiced stops fricativization and defricativization, vowel harmonization and disharmonization,

263 feature dissimilation, and declusterization of nasal plus consonant clusters characterize

English nouns in EkeGusii.

It is also worth noting that morphologically: the semantics of the stems of English nouns borrowed into EkeGusii determine the nominal class into which the nouns enter, the bi- morphemic structure of EkeGusii prefix characterize English nouns borrowed into

EkeGusii; and affixations in the English nouns obey that of EkeGusii in which plurality and singularity are prefixed and class paired.

Another conclusion is that the phenomenon of noun nativization in EkeGusii can be accounted for within Optimality Theory, a constraint- based approach. Through this theory, an explanation to the phonological and morphological adjustments of English nouns borrowed into EkeGusii is possible. Phonologically, the main strategies employed in the nativization of English nouns in EkeGusii are motivated by markedness constraints such as

*OCP (V), *CODA, and *COMPLEX, which dominate the faithfulness constraints such as

IDENT IO and MAX IO. The opposite is true in the analysis of English noun. This observation lead to the conclusion that English allows marked constraints as compared to

EkeGusii. Morphologically, alignment constraints such as (ALIGN (AFX, R;RT, L)), which outrank faithfulness constraints such as STRPRES, motivate nativization of English nouns in EkeGusii. Thus, phonological and morphological nativization of English nouns in

EkeGusii is motivated by EkeGusii ranking of the universal linguistic constraints proposed in OT. Therefore, ranking of constraints in EkeGusii is responsible for the outputs of

EkeGusii English nouns in EkeGusii. This rules out any possibility that the target language has influence in the phonology and morphology of the target language besides the lexical item itself (Owino, 2003). 264 5.3 Recommendations

In the description of EkeGusii vowels, a spectrographic acoustic analysis of the vowels was attempted. Many areas of EkeGusii phonetics and phonology such as, consonant segments, pitch, tone, and intensity, among others, require such an, especially in these areas.

Therefore, this study recommends spectrographic (computer software) analyses of all the aspects of EkeGusii phonetics and phonology for better understanding and documentation of the language.

This study describes the phonological and morphological structures of EkeGusii as a basis for analyzing English nouns in EkeGusii. In the analysis of the morphology of the English nouns, it was established that semantics plays a major role in determining the noun classes into which the English noun enters, a major observation, yet this is not given the attention it deserves in this study or anywhere else. It is therefore recommended that a study focusing on the same be conducted in an effort to shed more light into EkeGusii loan words nativization. A number of phonological processes were found to characterize nativization of English nouns in EkeGusii. These processes were not given the attention they deserve given the broad nature of this study. It is therefore recommended that a study specifically focusing on all the phonological processes characterizing English nouns in EkeGusii, including the ones identified in this study be carried out so as to shed more light into the phenomenon of loan word nativization. EkeGusii has borrowed heavily from a number of languages, most notably: Kiswahili,

English and Dholuo. Yet this study has focused only on English. It is therefore recommended that other studies be directed to these languages too, if a comprehensive inventory of all the EkeGusii loan words is to be made. This is because there is a likelihood

265 that EkeGusii indigenous words are facing extinction due to this borrowing and nativization of these foreign words. Analyses in this study focused on the noun class. But other classes, especially the verb, which is rich in morphology, are equally borrowed and nativized. Therefore, a study on these other classes is recommended as it would shed more light on EkeGusii loaned words nativization, especially morphologically. As much as this study has provided important data and advanced illuminating discussions, there were a number of interesting areas that remain unexplored. This study, it is hoped, will stimulate further inquiry into the areas of EkeGusii orthography, syntax and nativization, so as to deepen the phonological and morphological understanding of linguistic integration. Since phonological and morphological systems of a language are important for the development of the orthographies of a language, it is also hoped that other such studies will stem from the current one, since EkeGusii is a language without a comprehensive orthography. Developing orthography for a language ensures the language’s continued existence and its assertiveness as an independent language that can handle loans fully.

The theoretical framework designed for this research is a constraint based generative one

(Optimality Theory) and findings of the study are best explained by such a paradigm. Other researches may also emerge, testing the same language phenomena but taking other linguistic theoretical paradigms, to enhance understanding of English loans in the EkeGusii linguistic environment, from a number of theoretical approaches.

The study is also hoped to be used as a basis to further constructive studies relating to

Bantu languages, other than EkeGusii. Since EkeGusii is a Bantu language, the findings

266 from this study can be an ‘eye opener’ and insightful to the understanding of similar languages in a diglossic situation with English.

REFERENCES Abdul, M., (2004). The effects of language contact: the case of Arab and Igbo.

(Unpublished PhD dissertation, Khartoum University of Africa). Aikhenvald, A., Y., (2000). Classifiers, a typology of noun categorization devices.

Oxford: OUP. Alderete, N., (1999). Morphologically governed accenti in Optimality Theory.

(Unpublished PhD thesis, University of Massachusetts Amherst). Anyona, M., (2011). Phonological influence of EkeGusii in the pronunciation of English. (Unpublished MA thesis, Egerton University). Appel, C., Muysken, P., (1981) Halfway between Quenchua and Spanish: the case of

reflexivization. In Highfield, A., and Valdman, A.,(eds.). Historicity and variation

in Creole studies. Michigan: Ann Arbor. Arbib, M., (Ed.). (2006). Handbook of brain theory and neural networks (2nd edn.). Mas:

MIT press. Archangeli, D., and Pulleyblank, D., (2007). Harmony. In The Cambridge Handbook of

phonology, Paul de Lacy(ed.), 353-378. Cambridge: Cambridge University press. Archangeli, D., and Pulleyblank, D., (1992). Grounded phonology. Cambridge: MIT

press. Archangeli, D., (1997). Optimality Theory: an introduction to linguistics in the 1990s.

Optimality Theory: an overview, (ed.). By Diana Archangeli and D. Terence

Langendeon,1-32. Oxford: Blackwel. Asher, R., and Simpson, J., (1994) (eds.). The encyclopedia of language and linguistics.

Oxford, Pergamon press Ltd. R.W. Bailey, K., (1978). Methods of social research (3rd ed.).New York: Free Press. Barker, M., (1969). The phonological adaptation of French loanwords into Vietnamese.

Mon- khmer studies Journal 3. 138-147.

267 Batibo, H., (1996). Loanword clusters nativization rules in Tswana and Swahili: a

comparative study. South African Journal of African languages, Vol. 16, 2. 33-41. Belvins, J., (1974). Lanakel Phonology. (Phd dissertation, University of Hawaii). Bennett, P., (1967). Dahl’s Law and Thagisu. African language studies.8:127-159. Bickmore, L., (1998). Metathesis and Dahl’s law in EkeGusii. Studies in linguistic

sciences. Vol.28. Bicmore, L., (1998). Opacity effects of Optimal Domains Theory:evidence from EkeGusii.

In Maddieson, I., and Hinnebusch, T., (eds.). Trends in African linguistics 2:

Language and History. Bickmore, L., (1991). Compensatory lengthening in Kinyambo. Katamba, I., (ed.):

Lacustrine Bantu Phonology: Afrikanistiche Arbeitspapiere, 25:75-103. Bickmore, L., (1997). Problems in constructing high tone spared in EkeGusii. Lingua.

102:4, 265-290. Block, B., and Tragger, G., (1972). Outlines of linguistic analysis. Linguistic society of

America. Bloomfield, L., (1933). Language. New York: Holt, Reihart &Winston. Boersma, P., and Weenink, D., (1992). Doing phonetics by computer. Version 6.0.2.1 SIL

EncoreTM: Summer Institute of Linguistics. Boersman, P., and Hamann, V., (2009). Loanword adaptation as first language

phonological perception. In Calabrese and Wetzels, p. 11-58. Booij, G., and Rabach, J., (1984). Morphology and prosodic domains in lexical phonology.

In. Veira, S., (ed.). Yearbook. Vrije: Vrije University. Vol.1, Number1. 88-102. Bosha, I., (1993). The Influrnce of Arabic language on Kiswahili with a trilingual

dictionary (Swahili- Arabic- English). Der es Salaam: Der es Salaam University

Press. Brown, F., (2004). Adaptation of English words in Scandinavian languages. In Colonel, P.,

(ed.). Linguistic fieldwork. Oslo: University of Oslo. 22-40. Bosire, B., (1993). Ekegusii: a comparative study of Rogoro and Maate dialects (MA thesis, University of Nairobi). Bosire, K., and Machogu, G., (2013). Authoritative EkeGusii dictionary:endabaro

endabasiay’EkeGusii. Nairobi; EkeGusii encyclopedia project.

268 Bresnan, J., and Mchombo, S., (1987). Topic, pronoun, and agreement in Chichewa:

language 63(4) 741-782. Bright, J. (1970). Teaching English as a second language. London: Longman group Ltd. Broselow, E. (2006). Loanword phonology. In Keith Brown (ed.), Encyclopedia of

language and linguistics, vol. 7. 2nd edn., 286–290. Oxford: Elsevier. Broselow, E., (2009). Stress adaptation in loanword phonology: Perception and

learnability. In Paul Boersma & Silke Hamann (eds.), Phonology in perception,

191–234. Berlin & New York: Mouton de Gruyter. Brown, C., (2000). The interrelation between speech perception and phonological

acquisition from infant to adult. In John Archibald (ed.), Second language

acquisition and linguistic theory, 4–65. Oxford: Blackwell. Burenhult, N., (2001). Loanword phonology in Jahai. Lund University Department of

Linguistics Working Papers 48. 5–14. Broselow, E., (1999). Stress, epenthesis, and segment transformation in Selayarese loans. Proceedings of the Annual Meeting, Berkeley Linguistics Society 25. 311–325. Bynon, T., (1977). Historical linguistics. Cambridge: CUP. Calabrese, A., (2009). Perception, production and acoustic inputs in loanword phonology. In Calabrese and Wetzels (2009), 59–114. Calabrese, A., and Wetzels, L., (eds.). 2009. Loan phonology. Amsterdam and Philadelphia: John Benjamins. Cammenge, J., (2002). Phonology and Morphology of Ekeusii. Postfach: Rudiga Koppe

Verlag. Campbell, G., (1991). Compendium Of the word’s Languages. Vol.2. London.Routledge. Chang, C., (2009). English loanword adaptation in Burmese. Journal of the Southeast Asian Linguistics Society 1. 77–94. Carsteins, V., (1991). The morphology and syntax of determiners phrase in Kiswahili.

(Unpublished PhD dissertation, UCLA). Carsteins, V., (1993). On the morphology and DP structure. In Sam, A., Mchombo (ed.).

Theoretical aspects of Bantu grammar. Stanford: CSLI publications. Change, C., (2009). English loanword adaptation into Burmese. Journal of the South east

Asian linguistics society 1.77-94. Chebanne, A., and Phili, C., (2015). The phonologization of English loanwords in Kalanga.

Marang: Journal of language and literature, vol.26.

269 Chimhundu, H., (1979). Some problems relating to incorporation of loanwords in the

lexicon. (PhD dissertation, University of Zimbabwe,) Dissertation Abstracts

International Vol. 11, Number 1. 75-91.

Chomsky, N., and Halle, M., (1968, 1991). The sound patterns of English. New York:

Harper and Row.

Clark, J., Yallop, C., and Fletcher, J., (2007). An introduction to phonetics and phonology.

Malden: Blakwell Publishing.

Clements, G., (2001). Representational economy in constraint based phonology. In Alan

Hall(ed.). Distincive feature theory, 71-146. Berlin & New York. Mounton de

Gruyter.

Clements, G., (1986). Compensatory lengthening and consonant germination in Luganda.

Leo Wetzels and Engin Sezer (ed.). Studies in compensatory lengthening.

Dordrecht: Foris: Laboratory Phonology I.

Clements, G., (1976). Vowel harmony in non linear generative phonology. Havard: MS.

Clements, G., and Keyser, S., (1983). CV Phonology: a generative phonology theory of the

syllable. Cambridge: Mass MIT Press.

Collins, B., and Mees, I., (2013). Practical phonetics and phonology. Oxon: Routledge.

Crysal, D., (1987). The Cambridge encyclopedia of Language. Cambridge: CUP.

David, O., (2005). Introducing phonology. New York.: Cambridge University press.

270 Davy, J., and Nurse, D., (1982). Synchronic versions of Dahl’s Law: the multiple

application of a phonological dissimilation rule. Journal of African languages and

linguistics 4:157-195.

Davidson, L., (2007). The relationship between the perception of non-native phonotactics

and loanword adaptation. Phonology 24. 261–286.

Davidson, L., and Noyer, R., (1997). Loanword phonology in Huave Nativization and the

ranking of faithfulness constraints. Proceedings of the west coast conference on

formal linguistics 15. 65-79.

De Lacy, P., (ed.). (2007). The Cambridge handbook of phonology. Cambridge: CUP.

Demuth, K., (2000). Bantu noun class systems Loanward and acquisition evidence of

sematic proclivity. In G senft (ed.) classification systems. Cambridge University

press PP. 270-292.

Denesi, M., (1985). Canadian Italian: a case in point of how language adapts to

environment. Italians in Ontario. Multicultural history society of Ontario.

Doupoux, E. & peperkamp, S., (2002). The phonetic filter hypothesis:how phonology

impacts speech perception (and vice versa). Paper presented at the second

International Conference on contrast in pholnlogy. University of Toronto, May 3-5,

2002. Retrieved April, 10 2015, from http//www chss.

Utoronto.ca/contrast/dupouxpepakamp. ppt1.

Eichhoff, J., (1990). Aspects of German borrowing into American English. In Nelde, P.

(Ed.). Language in contact and conflict. Weshbaden: Franz stainer Verlag aMBH.

271 Elwell, R., (2005). A morphosyntactic analysis of the EkeGusii verb. (Unpublished

Honours’ thesis, University of Albany).

Ferguson, C. A., (1963). Assumptions about nasals: a sample study in phonological

universals. In J. Greenberg (ed.) Universals of language, 53-60, Cambridge, MA:

MIT press.

Gay, L., (1981). Educational research: compentence for analysis and application. Charcles

E. Mairil publishing company: London.

Givon, T., (1972). Studies in Chibemba and Bantu grammar. In. Studies in African

linguistics. Supplement 3.

Gleason, H., A., (1961). An introduction to descriptive linguistics. New Delhi: Oxford

GOK (2009). Kenya Bureau of statistics. Government printers.

Goldsmith, J., (1990). Autosegmental and metrical phonology. Oxford: Basil Blackwel.

Good, J., (2005). Reconstructing morpheme order in Bantu: the case of causativization and

applicativization. Dia chronica 22:3-57.

Golston, C., and Yang, P., (2001). White Hmong loanword phonology. In Caroline Fery,

Antony Dubach Green & Ruben Van de Vijver(eds.) Proceedings of HILP5, 40-57.

Grimes, B., (1996). Ethnologue. Languages of the World. Dallas: Texas summer Institute

of Linguistics.

Grosjean, F., (1982). Life with Two Languages. An Interoduction to Bilingualism.

Cambridge, MA: Harvard University Press.

272 Gussenhoven, C., and Jacobson, H., (2013). Understanding phonology. New York:

Routledge.

Guthrie, G., (1967-71). Comparative Bantu: Farnborough. Gregg. Vols 1-4.

Hadebe, S., (2002). The standardization of Ndebele languagethrough dictionary masking.

Oslo: University of Oslo.

Hafez, O., (1996). Phonological and morphological integration of loanwords into Egytian

Arabic. Egypty/Monde Arabic, premiere serie 27-28. 383-410.

Hall, T., and Hamann, S., (2003). Loanword Nativization in German. Journal of Zeitschrift

fur sprachwissenchaft 22.1 (2003), 56-85.

Hancin-Bhatt, B., (1994). Segment transfer: a consequence of a dynamicsystem, Second

language research 10.241-261.

Harjulla, L., (2005). The Ha noun class system revisted. MS

Haspelmath, M., and Sims, A., (2002).Understanding morphology. London. Hodder

Education.

Haugen, E. (1950). The analysis of linguistic borrowing. Language 26. 210–231.

Herd. J., (2005). Loanword adaptation and the evaluation of similarity. Toronto working

papers in linguistics 24.65-116.

Haugen, E., (1950). The analysis of linguistic borrowing. Language, 26: 210-231.

Hock, H., (1991). Principles of historical linguistics. New York: Mounton de Gruyter.

Hocket, C., (1970). A course in modern linguistics. New York: MacMillan.

273 Hoffer, B., (2005). Language borrowing and diffusion: an overview. In International

communication studies, XL:200. Retrieved from http// www. uri. Edu /iaics/ content

/2002vlln4/01%20Bates%20L. %20 Hoffer. Pdf.

Holden, K., (1976). Assimilation rates of borrowings and phonological productivity.

Language 52. 131–147.

Hooper, J., (1976). An introduction to Natural Generative Phonology. New York:

Academic Press.

Hsieh, F., &and Kenstowicz, M., (2008). Phonetic knowledge in tonal adaptation:

Mandarin and English loanwords in Lhasa Tibetan. Journal of East Asian

Linguistics 17. 279–297.

Hussain, Q., (2011). Vowel substitution: a comparative study of English loans in Punjabi

and Urdu. International Journal of linguistics. 31: E31.

Hyman, L., (1985). A theory of phonological weight. Dordrecht: Foris.

Hyman, L., (1979). Aghem grammatical structure. SCOPL No.7 Califonia: UCLA.

Hyman, L., (1970). The role borrowing in the justification of phonological grammars,

Studies in African linguistics1. 1-48.

Hyman, L., and Byarushengo, E., (1984). A model of Haya tonology. In Clements, G., and

Goldsmith J., (eds.). Autosegmental studies in Bantu tone. Dordrecht: Foris.

Islam, R., A., (2011). The morphology of loanwords in Urdu: the Persian, Arabic and

English strands.( Unpublished PhD thesis, Newcastle university).

274 Ito, C., and Kenstowicz, M., (2009). Mandarin loanwords in Yanbian Korean I: Laryngeal

features. Onin Kenkyuu [Phonological Studies] 12. 61–72.

Ito, C., Kang, Y., and Kenstowicz, M., (2006). The adaptation of Japanese loanwords into

Korean. MIT Working Papers in Linguistics 52. 65–104.

Itô, J., and Mester, A., (1995). Japanese phonology. In John Goldsmith (ed.). The

handbook of phonological theory:817-838. Cambridge, MA and Oxford: Blackwel.

Itô, J., and Mester, A., (1999). The phonological lexicon. In Natsuko Tsujimura (ed.), The

handbook of Japanese linguistics, 62-100. Malden, MA and Oxford: Blackwell.

Itô, J., and Mester, A., (2001). Covert generalizations in Optimality Theory: the role of

stratalfaithfulness constraints. Studies in phonetics and morphology 7. 273-299.

Jacobs, H.. and Gussenhoven, C., (2000). Loan phonology: perception, salience, the

lexicon and OT. In Joost Dekkers, Frank van der leeuw, and Jeron vande Weijer

(eds.), Optimalitytheory: phonology, syntx and acquisition, pp.193-210.

Oxford:OUP.

Jakobson, R., (1962). Selected writings I: phonologicalstudies. Gravenhage: Mouton.

John, A., (2000). The Handbook of phonology theory. Amazon: Blackwell publishers.

Jones, D., (1956). The pronunciationof English. Cambridge: CUP.

John, H., (1994). Borrowing. In John H. (ed.) Yearbook. Vol. 2, 1 Harmondsworth:

University of Warsaw 122-140.

Johnson, K., (1977). Acoustic and auditory phonetics. Cambridge: CUP.

275 Kadenge, M., and Mudzingwa, B., (2012). Comparing Chishona loanwords of monolingual

and bilingual speaker: an Optimality Theory analysis. South African Journal of

African languages, 2012, 32 (20: 141-151).

Kadenge, M., (2003). The influence of the Shona phonology on the English: speech of First

Language (L1) Shona Speakers of English-With Specific Reference to Vowels

(Honours Dissertation, University of Zimbabwe).

Kager, R. (1999). Optimality Theory. Cambridge: Cambridge University Press.

Kang, H., (1996). English loanwords in Korean. Studies in Phonetics, Phonology, and

Morphology 2. 21–47.

Kang, Y., (2010). Tutorial overview: suprasegmental adaptation in loanwords. Lingua. 120,

2295-2310.

Kang, Y., (2011). Loanword phonology. In Mark van Oostendorp, Colin Ewen, Elizabeth

Hume, & Karen Rice (eds.). Companion to phonology. Wiley: Blackwell.

Katamba, F., (1989). An introduction to phonology. London: Longman Academic Press.

Katamba, F., (1993). Morphology. Palgrave: Hampshire.

Katayama, M., (1998). Optimality Theory and Japanese loanword phonology (PhD.

dissertation, University of Califonia, Santa Cruz).

Kay, G., (1996). English loanwords in Japanese. World Englishes. Vol. 14 . 1, 67-76

Kayigema, L., (2010). Loanword allocation in Kinyarwanda. (UnpuplishedMA thesis,

University of South Africa, South Africa).

276 Kenstowicz, M. (2005). The phonetics and phonology of loanword adaptation. In S.- J.

Rhee (ed.), Proceedings of ECKL: Proceedings of 1st European conference of

Korean linguistics 17-32:Hankook publishing co.

Kenstowicz, M., and Shuchato, A., (2006). Issues in loanword adaptation: a case study

from Thai. Lingua 116. 921-949.

Kenstowicz, M., and Sohn, H., (2001). Accentual adaptation in North kyungsang Korean.

In Michael Kenstowicz (ed.), Ken Hale. A life in language, 239-270. Campridge,

MA: MIT Press.

Kihm, A., (2005). What’s is in a noun? Noun classes, gender and nounness. MS.

Kim, H., (2008). Loanword adaptation between Japanese and Korean: Evidence for L1

feature-driven perception. Journal of East Asian Linguistics 17. 331–346.

Kim, Y., (2008). Topics in the phonology and of San Francisco del mar Huave

(Unpublished PhD dissertation, University of Califonia, Berkeley).

Kiparsky, P., and Mohanan, S., (1982). From Cyclic Phonology to Lexical Phonology. In

van de Hulst, H., and Smith, N., (eds.). The structure of phonological

representations. New Delly: Indiana University Press. Vol. 1 1, Number 1. 130-

153.

Kombo, D., and Tromp, L., (2006). Proposal and Thesis writing: an introduction. Nairobi:

Paulines publiations Africa.

Kumar, R., (1999). Research Methodology a step-by-step guide for beginners. London:

SAGE Publications Ltd.

277 LaCharite, D., and Paradis, C., (2005). Category preservation and proximity versus

phonetic approximation in loanword adaptation. Linguistic inquiry 36. 223-258.

Ladefoged, P., and Keith, J., (2001). A course in Phonetics. Boston. Cernage learning, Inc.

Langacker, R., (1968). Language and its structure: some fundamental linguistics concepts.

New York: Holt, Reihart & Winston.

Laparombara, H., (2013). An Optimality Theory account of phonological variation in

Kimbrlek Krioi. (Unpublished senior thesis, Yalle University).

Libert, B., (1971). Linguistics and the new language teacher. New York: MacMillan.

Lodhi, A., (2000). Oriental influences in Swahili: a study in languages and culture

contacts.

Lombard, L., (ed.). (2001). Segmental phonologyin Optimality Theory: constraints and

representation. Cambridge & New York: CUP.

Massamba, D., (1991). Phonological theory: History and development. Dar es salaam: Dar

es salaam University press.

Massaro, D., and Michael, C., (1983). Phonological context in speech perception.

Perception and psychophysics 34. 338-348.

Mathangwana, J., (1999). Ikalanga phonetics and phonology: a synchronic and diachronic

study. Califonia: C.S.L.I Publishers.

278 Mberia, K., (2004). The phonology of borrowed words in Kitharaka. In Buregeya, A.,

Schroeder, H., and Eshun, F., (eds.). Occasional papers in language and linguistics

(PP.45-57). Department of Linguistics and African languages: University of Nairobi

press.

McCarthy, A., C., (2002). An introduction to English morphology. Edinburgh: Edniburgh

University Press.

McCarthy, J., and Prince, A., (1993). Prosodic morphology: constraint interaction and

satisfaction. London: Routledge.

McCarthy, J., (2001). A thematic guide to Optimality Theory. Cambridge: Cambridge

University press.

McCarthy, J., (ed.). (2003). Optimality Theory in phonology: a reader. Malden, MA and

Oxford: Blacwel.

McCarthy, J., (2003). Phonological processes: assimilation. In Frawley, J. (ed.).

International encyclopedia of linguistics (2nd edn.), 320-323: Oxford University

press.

McCarthy, J., (2006). Morphology: optimality Theory. Amherst, Elsevier Ltd.

McCarthy, J., and Prince, A., (1994). The emergence of the unmarked: optimality in

prosodic morphology. London: Equinox.

Mchombo, S., (1967). Bantu grammatical reconstructions. Tervuern: Tervuren

Mchombo, S., (2004). Th.e syntax of Chichewa. Cambridge: CUP.

279 Mecha, G., (2006). The phonology and morphology of EkeGusii reduplication : an

optimality theoretic approach. (Unpublished masters thesis, Kenyatta university).

Mecha, G., (2013). Information structure in EkeGusii. (Unpublished PhD thesis, University

of Nairobi, 2013).

Meeussen, A., (1993). Comparative Bantu: test cases for method African language studies

14, 6-18.

Miao, R., (2005). Loanword adaptation in Mandarin Chinese: perceptual, phonological

and sociological factors (unpublished PhD dissertation, Stony brook University,

2005).

Mihalicek, V., Wilson, C., (Eds.). (2011). Language files. (11nth edn.). The Ohio

University Press: Columbus.

Mugenda, O., and Mugenda, A., (1999). Research methods; quantitative and qualitative

approaches. Nairobi: ACTS Press.

Mutua, B., (2007). A constraint based analysis of Kikamba nativized loanwords

(Unpublished masters thesis, Kenyatta University).

Mwihaki, A., (2001). Consonant–vowel harmony: Evidence from the phonotactics of

loanword adaptation. Poznan Studies in Contemporary Linguistics 37. 139–145.

Mwihaki, A., (1998). Loanword nativization: a generative view of phonological adaptation

of Gikuyu loanwords ( Unpublished PhD thesis, Kenyatta University).

Mwita, L., (2012). Noun tonology in Kuria. International Journal of Humanities and social

sciences 2(8): 159-168.

280 Mwita, L., (2009). The adaptation of Swahili loanwords from Arabic: a constraint based

analysis. The journal of pan African studies, vol.2, no.8.

Nash, C., (2011). Tone in EkeGusii: a constrint based description of nominal and verbal

tonology. (Unpublished PhD thesis, University of Califonia, Santa Barbar).

Nasukawa, K., (2004). Word final consonants: arguments against a coda analysis.

Proceedings of the 58th conference of the Tohoku English literacy society, 47-53.

Nurse, D., and Philemon, G., (2006). Commontense tense aspect markers in Bantu. in

journal of African languages and linguistics.

Nurse, D., and Philemon, G., (2006). The Bantu languages. London: Routledge.

O’ Connor, J. (1967). Better English pronunciation. Cambridge: CUP.

Oden, D., (2005). Introducing Phonology. Cambridge: Cambridge University press.

Oden, D., (1996). The phonology and morphology of Kimatuumbi. Oxford: Clarendon

press.

Ogechi, N., (2011) (ed.). Themes in language, education & development in Kenya. Nairobi:

Nsemia Inc. Publishers.

Ohala, J., and Sole, M. (2008). Turbulence and phnology. UC Berkeley phonology lab

annual report.

Ohso, M., (1971). A phonological study of some English loanwords from in Japanese.

(Unpublished MA thesis, the Ohio State University).

Ongarora, D., (2009). Bantu morphosyntax: a study of EkeGusii (Unpublished PhD thesis,

Jawaharlal Nehru University, India).

281 Ongarora, D., (1996). Vowel harmony in the Rogoro dialect of Ekegusii ( Unpublished MA

thesis, Egerton University).

Osinde, K., (1988). Ekegusii phonology: an analysis of the major consonants (MA thesis,

University of Nairobi).

Owino, D., (2003). Phonological nativization of Dholuo loanwords (Doctoral

dissertation, University of Pretoria).

Paradis, C., and Lacharite, D., (2009). English loanword in Old Quebec French: Fewewer

bilinguals does not mean a great increase in native phonetic approximation.

Langues et linguistique 32. 82-117.

Paradis, C., (1996). The inadequacy of filters and faithfulness in loanword adaptation. In

Jacques Durand, and Benard Laks (eds.), Current trends in phonology: modes and

methods, pp509-534. Salford: University of Saloford Publication.

Paradis, C., and Lacharite, D., (1997). Preservation and minimality in loanword adaptation.

Journal of linguistics. 33, pp. 379-430.

Pearce, M., (2003). Vowel harmony in Kera (Chadic). (MA dissertation, University College

London).

Peperkamp, S., and Dupoux, E., (2003). Reinterpreting loanword adaptations: the role of

perception. In Sole, M., Recasens,D., and Romero, J., (eds.), procedings of the 15th

international congress of phonetic sciences, 367-370. Barcelona: Causal

productions.

282 Peperkamp, S., and Dupoux, E., (2001). Loanword adaptations: laboratoire de sciences

cognitives et psycolinguitique, Universite, de Paris 8 http://www. Ehess.

Fr/centres/LSCP/persons/ dupoux/loanwords. pdf.

Plag, I., (2002). Morphological productivity, structural constraints in English Derivation.

Berlin: Routledge.

Plag, I., (2003). Word formation in English. Cambridge: CUP.

Prince, A., and Smolensky, P., (1993/2004). Prosodic morphology: constraint ineraction

and satisfaction in generative grammar. Oxford: Blacwell.

Pulleyblank, D., (1988). Vocalic underspecification in Yoruba. Linguistic Inquiry 19. 233-

270.

Quirk, R., and Greenbaum, J., (1990). A university grammar of English. England:

Longman.

Raiz, A., (2011). The morphology of loanwords in Urdu: the Persian, Arabic and English

strands (PhD dissertation, Newcastle University).

Roach, P., (1983). English phonetics and phonology.Cabridge :CUP.

Robins, R., (1964). General linguistics. An introductory survey. London: Longman.

Rodgers, J., (1995) Using diachronic linguistics in the classroom language. In. Modern

Language Journal. Vol. 111, Number 2. 212-222.

Rose, Y., and Demuth, K., (2006). Vowel epenthesis inloanword adaptation:

representational ad phonetic considerations: Lingua 116. 1112-1139.

Russel, K., (1995). Morphemes and candidates in Optimality Theory. MS. Rutgers

Optimality Archive.

283 Salim, J., (2013). A contrastive study of English-Arabic noun morphology. International

Journal of English linguistics 3 (3) , 122-132.

Sasa, T., (2009). Treatment of vowel harmony in Optimality Theory. (Unpublished Phd

dissertation, Iowa University).

Schutz, J., (2004). English loanwords in Fijian. In Geraghty and Tent (2004b), 253-294.

Scotton, C., and Okeju, J., (1972). Loanword integration in Ateso. Anthropological

linguistics, 14(9) 368-382.

Selinker L., (1972). Rediscovering inter-language: New York, Longman Group Ltd.

Shabiro, M., (2015). A unique case of vowel harmony in English (Iambaste). Retrieved

from Michael Shapiro’s wepsite, “semantic neostructuralism in peircean

linguistics”.

Shademan, S., (2003). Epenthetic vowel harmony in Farsi (MA thesis, University of

Califonia, Los Angles).

Shilington, K., (1995). History of Africa. Oxford: Macmillan publishers limited.

Shinohara, S., (2006). Perceptual effects In final cluster reduction patterns. Lingua116.

1046-1078.

Shinohara, S., (1997). Analyse phonolgique de l’ adaptation Japonaisedes mots

entrangers. (PhD dissertation, Paris III).

Sing, R., (1981). On some redundant compounds in modern Hindi. Lingua 56, p. 345-351.

Smolensky, P., (1996). The initial state and “Richness of the Base” in Optimality Theory.

(Unpublished Ms., Johns Hopkins University, Baltimore (ROA-154.)).

284 Singabapha, C., (1998). The phonologization of English loanwords in Ikalanga and the

implications of this process for the English spelling and pronunciation by Ikalanga

speakers. (BA dissertation, University of Botswana).

Silverman, D., (1992). Multiple scansions in loanword phonology: evidence fro

Cantonese. Phonology 9.289-328.

Spencer, A., and Zwicky, A., (eds.) (1998). The handbook of morphology. Oxford:

Blakwell publishers.

Steel, S., (1995). Towards a theory of morphological information. Language71: 260-309.

Taber, C. R., (1979). French loanwords in Sango: the motivation of lexical borrowing. In

Hancock, I., (ed.). Readings in creole studies. Ghent: scientific studies.

Takagi. N., and Mann, V., (1994), A perceptual basis for the systematic phonological

correspondences between Japanese loanwords. Journal of phonetics, 22, 243-356.

Thomason, S., and Kaufman, T., (2008). Linguistic contact, creolization, and genetic

linguistics. Berkeley: University of California Press.

Trubetzkoy, N., (1969). Principles of Phonology. Berkeley and Los Angeles: Califonian

University press.

Uffmann, C., (2004). Vowel epenthesis in loanword phonology. (PhD dissertation, Philipps

University, Morbug, Germany).

Uffmann, C., (2013). Set the controls for the heart of the alternation: Dahl’s Law in

Kitharaka. Nordlyd. 40.1:323-337.

Uffmann, C., (2001). Patterns of vowel epenthesis (not only) in Shona loanwords, In:

Anthony Dubach Green, Caroline Féry & Ruben van de Vijver (eds.), Proceedings

of HILP 5. Universität Potsdam.

285 Uffimann, C., (1993). Distinctive features: an introduction. (Lecture notes, Stanford:

Philipps University, Marburg).

Vendelin, I., and Peperkamp, S., (2006). The influence of orthography on loanword

adaptations. Lingua 116. 996–1007.

Verma, K., and Krishnaswamy, N., (1989). Modern linguistics: an introduction. Delhi:

OUP.

Weinreich, H., (1953). Language in Contact: findings and problems. The Hague: Mouton

and Co. Ltd.

Werker, J., and Toes, C., (1984). Phonemic and phonetic factors in adult cross language

speech perception. Journal of Acoustic Society of America. 75. 1866-1878.

Whiteley, W., (1967). Loanwords in linguistic description: a case study from Tanzania,

East Africa. Ranch, I., and Scott, C., (eds.). Approaches in linguistic methodology:

Mounton deGruyer.

Whiteley, W., (1965). A practical introduction to Gusii. Nairobi: EALB.

Witeley, W., (1960). Tense system of Gusii. Kampala: East African Institute of Research.

Winter, W., (1992). Borrowing and non-borrowing in Walapi. In E.H. Jahr, (Ed.).

Language contact: theoretical and empirical studies. Berlin: Mounton deGruyer.

Wu, H., (2006). Stress to tone: a study of tone loans in Mandrine Chinese. MIT working

Papers in linguistics 52. 227-253.

Wunderlich, D., (1996). Minimalist morphology: the role of paradigms. Yeaybook of

Morphology 1995: 93-114.

Yip, M., (2006). The symbiosis between perception and grammar in loanword phonology.

Lingua116 950-975. 286 Yip, M., (2002). Necessary but not sufficient: Perceptual influences in loanword

phonology. Journal of phonetic society of Japan 6. 4-21.

Yip, M., (1993). Cantonese loanword phonology and Optimality Theory. Journal of East

Asian Linguistics 2. 261-291.

Zikmund, W., Babin, J., Carr, J., & Griffin, M. (2012). Business research methods: with

qualtrics printed access card. Cengage Learning.

Zivenga, W., (2009). Phonological and Morphological Nativization of English Loans in

Tonga (Unpublished PhD dissertation, UNISA).

Zwicky, A., (1985). How to describe inflection. In proceedings of BLSII, ed. by Mary

Niepokuj, Mary vanClay, Vassiliki Nikifondou and Deborah Feder, 372-386.

University of Califonia, Berkeley.

287 APPENDICES

Appendix I: Interviewee profile form

1. Name……………………………………………………………………………………

2. Age……………………………………………………………………………………..

3. Gender…………………………………………………………………………………..

4. County…………………………………………………………………………………..

5. Sub- County……………………………………………………………………………..

6. Area of current residence (village)……………………………………………………..

7. Area of former residence (if any)………………………………………………………..

8. First language…………………………………………………………………………….

9. Knowlede of:

(i) Kiswahilili language………………………………………………………………..

(ii) English language………………………………………………………………….

(iii) Any other language(s):

10. Occupation………………………………………………………………………………..

11. Level of education………………………………………………………………………..

12. School(s) attended………………………………………………………………………..

288 Appendix II: Interviewee consent form I, Mr/Mrs/Ms……………………………………………………………………….., agree to participate in the research exercise being conducted by Mr. George Morara Anyona of Kisii

University. I wish to state that he has made me aware of what he requires and have voluntarily and willingly accepted to volunteer information pertaining to EkeGusii language for purposes of the research.

ID . No…………………………………………………………………………..

Sub-location …………………………………………………………………….

County…………………………………………………………………………..

Date……………………………………………………………………………..

Sign………………………………………………………………………………

289 Appendix III: Research instrument: semi-structured interview guide Interview questions in this study were based on the following thirteen (13) domains of life of the nouns.

Domain of the Questions nouns

1. FOOD AND 1. Ninki orantebie igoro ye’chindagera ne’binyugwa mokoroisia na NUTRITION koria aaiga, amo na chinkaki mogochiooni?

(What can you tell me about the food and drinks you prepare and sell here and the time you sell them?) 2. HOUSE-HOLD 1. Ntebo anko igoro ye’gasi yao ya’kera rituko. APPLIANCES AND UTENSILS (Narrate your daily work) 2. Ntebie ebinto bionsi bire nyomba aiiga gwansera ase mokorara, mokorugera, mogwesiberia, mogoikaransa. ( mention all the household things found in this house, in: the bedroom, batheroom, kitchen and sitting room) 3. AGRICULTURE 1. Aye no’yomo bwa’baremi aaiga, koranche narrateigoro yo’boremi bwao. (You are one of the farmers in this area. Please, tell me more about your farming activities).

2. Ntebi igoro ye chinchera chioboremi chigokorekana aaiga. (Tell me more about the types of farming activities carried out in this area).

3. Ntebi ebimeria biria bigosimekwa aaiga. Etugo rende? (Tell me the type of crops that are grown in this area? What about animals?) 4.TRANSPORT 1. Ntebi’ango igoro ye’gasi yao? Ndi gwaansete korosia chigari? AND MOTORING (when did you become a mechanic?)

2. Mechando ki okonyora? (Have you ever faced any challenged?)

(What challenges do you face?)

3. Ntebi’anko emechando emenene yechigari okorosia.

(Talk about the major mechanical problems you deal with )

4. Ntebianko buna ebioma aoao biegari bigokora emeremo.

290 (Tell me how a vehicle parts work)

5. Ntebi aina ye chigari okorosi

(Tell me about the kinds of vehicles that come for repair) 5. HEALTH 1. Enkaki engana ngaki gwkorete agasi buna omonyagitari? (For how long did you work as a nurse).

2. Ntebi buna enyagitari egokora egasi. (Tell me about how a hospital generally works).

3. Ntebi igoro ye’gasi yaobuna omorwaria. (Narrate to me about your work as nurse.)

4. Ntebi igoro ya’ marwaire acoria abanto aaiga. (Tell me about the most prefernt diseases in this area.)

5. Ntebi’ank buna abarwaia aoao na’bakorie’gasi be’nyagitari bagokora emeremo. (Tell me about how the various hospital personel work). 6. RELIGION 1. Ntebi’anko ekanisa yao. (Tell me the denomination you belong to?)

2. Intebie igoro ye’ kanisa eyio ne’chinde omanyete). (Tell me more about this denomination and any other that you know of).

3. Teba ebinto biria bigokorekana ekninisa rituko roi’gosasima. (Narrate the activities that during your worshiping day). 7. EDUCATION 1. Ntebi igoro yo’bogima bwao bwo’gosoma. (Tell me about your education life).

2. Ntebi igoro ye’binto birie bigokorekana esukuru rituko rie’sukuru. (Narrate about the activities that take place in school in a normal day).

3. Teba igoro ya’ baria bonsi bakobwaterana igoro ya’mangana ye’sukuru. (Talk about all the stakeholders in a school set up).

8. POLITICS 1. Kwabeire ime ya siasa amatuko amange. Ntebi’nko ebirogo biria abanasiasa bakorwanerera. (You have been in politics for so long, Tell me about the various political positions that politicians

291 vie for).

2. Nonyare kongeresa igoro ya siasa ye’nse? (Can you tell me something about national politics)? 9. LEGAL AFFAIRS 1. Gwakorire egasi ye’koti ase enkaki enyinge. Ntebi igoro ye’ chikoti chia’kenya buna chibangire na gokora egasi.(You have worked in courts for long. Tell me about the court system in Kenya). 2. Ntebi igoro yabakorie’gasi be’koti. (Tell me about the personnel of the courts).

10.ECURITY AND ADMINISTRATION 1. Ntebi igoro ye’gasi yao. (Tell me more aout your work). 2. Ntebi buna obogambi be’nse bobankire. (Describe the structure of the national administration). 11.INFORMATION, 1. Egasi yao no’korosia ebito ebi. Nkorosio’re binto binde otatiga COMMUNICATION ebio ndoche abuo? Koranche ntebi igoro ye’binto ebio.( You earn a & TECHNOLOGY leaving by repairing thins things . Do you repair anything else besides what I see on the shelves? Please tell me more about them). 2. Tell me about the changes you have witnessed over time concerning your work 12.BUSINESS/ 1. Koranche ntebi igoro ye’biasara biao. (Please, tell more about TRADE your business). Ninki ogokora kera rituko as egasi eyio? (What do on a daily basis?) 2. Ntebi igoro ye’binto biria okogora na koonia aaiga. (Tell me about the goods and services you buy and sell here).

13. CLOTHING 1. Koraanche ntebi egasi yao. (Please tell me what you do). Ntebi gochi’me mono igoro ye’gasi eyio yao). Tell me more about your work 2. Iyaanka’ki egetaamba bune’ke keraroisie? (what type of clothes can such piece of clothe make?) Naende gento kende?(Anything else?)

292 Appendix IV: Raw data recording/transcription form

Semantic domain…………1 (food and nutrition)

LOAN NOUN PHONOLOGICAL FORM MORPHOLOGICAL FORM SOURCE WORD FORM

293 Appendix V: Raw data Semantic domain…………1 (food, drinks and nutrition) - By a Hotelier

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL FORM SOURCE WORD FORM burekibasiti /βurekiβasiti/ burek-i-basit-I breakfast /brekfa:st/ ranchi /ranchi/ ranch-i lunch /lʌndʒ/ saba /saβa/ sab-a supper /sʌpǝ/ ekaroti /ekarͻti/ e-karot-i carrot /kǝrǝt chikaroti /t∫ikarͻti/ chi-karot-i carrots /kǝrǝts/ ekabichi /ekaβit∫i/ e-kabich-i cabbage /kǝbɪdʒ/ chikabichi t∫ikaβit∫i chi-kabich-i cabbages /kǝbɪdʒiz/ ekeki /ɛkɛki/ e-kek-i cake /keɪk/ chikeki /t∫ikɛki/ chi-kek-i cakes /keɪks/ ekerimu /ekerimu/ eke-rim-u cream /kri:m/ ebirimu /eβirimu/ ebi-rim-u - - esota /ɛsͻta/ e-sot-a soda /sɒdǝ/ chisota /t∫isͻta/ chi-sot-a - - gurukosi /γurukosi/ gurukos-i glucose /glu:kǝʊz/ eturunki /eturuŋki/ e-turunk-i drink /drɪŋk/ ekoko /ekoko/ e-kok-o cocoa /kɒkǝʊ ripaipai /ripaipai/ ri-paip-ai pawpaw /pǝʊpǝʊ amapaipai /amapaipai/ ama-paip-ai - esigara /esiγara/ e-sigar-a cigeratte /sɪgǝret/

Semantic domain…………2 (Household appliances and utensils) – By a house wife 294

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM etochi /ɛtͻt∫i/ e-toch-I torch /tͻ:t∫/ chitochi /t∫itͻt∫i/ chi-toch-i torches /tͻ:t∫ɪz/ ekerasi /ekerasi/ eke-ras-i glass /glᴂs ebirasi /eβirasi/ ebi-ras-i glasses /glᴂsiz/ etamosi /etamosi/ e-tamos-i thermos /θǝmɒs chitamosi /t∫itamosi/ chi-tamos-i - - ebirichi /ebirit∫i/ e-birich-i fridge /frɪdʒ/ chibirichi /t∫iβirit∫i/ chi-birich-i fridges /frɪdʒɪz oboranketi /oβoraŋketi/ obo-ranket-i blanket /blᴂŋkɪt/ amaranketi /amaraŋketi/ ama-ranket-i blankets /blᴂŋkɪts/ etauro /etauro/ e-taur-o towel /tǝwɛl/ chitauro /t∫itauro/ chi-tau-ro towels /tǝwɛlz/ erong’I /ɛrͻnŋi/ e-rong-i long trouser /lɒŋtrǝʊsǝ/ chirong’i /t∫irͻŋi/ chi-rong-’i long trousers /lɒŋtrǝʊsǝs/ esati /esati/ e-sat-i shirt /∫ɜ:t/ chisati /t∫isati/ chi-sat-i shirts /∫ɜ:ts/ ekebesiti /ekeβesiti/ eke-besit-i vest /vɛst/ ebibesiti /eβiβesiti/ ebi-besit-i vests /vests/ egoti /eγoti/ e-got-i coat /kǝʊt/ chigoti /t∫iγoti/ chi-got-i coats /kǝʊts/ echaketi /et∫aketi/ e-chaket-i jacket /dʒʌkɛt/ chichaketi /t∫it∫aketi/ chi-chaket-i jackets /dʒʌkɛt esikati /esikati/ e-sikat-i skirt /skɜ:t/ chisikati /t∫isikati/ chi-sikat-i skirts /skɜ:ts/ esogisi /ɛsͻγisi/ e-sogis-i socks /sɒks/ chisogisi /t∫isͻγisi/ chi-sogis-i - - eburaosi /eβuraosi/ e-buraos-i blouse /blǝʊz/ chiburaosi /t∫iβuraosi/ chi-buraos-i blouses /blǝʊzɪz/ etanki /etaŋki/ e-tank-i tank /tᴂŋk/ chitanki /t∫itaŋki/ chi-tank-i tanks /tᴂŋks/ ebetiruumu /ɛβɛtiruumu/ e-betiruum-u bedroom /bɛdru:m/ chibetiruumu /t∫iβɛtiruumu/ chi-betiruum-u bedrooms /bɛdru:ms/ ekicheni /ekit∫eni/ e-kichen-i kitchen /kɪt∫ǝn chikicheni /t∫ikit∫eni/ chi-kichen-i kitchens /kɪt∫ǝns ebaturuumu /eβaturuumu/ e-baturuum-u bathroom /bᴂðru:m/ chibaturuumu /t∫iβaturuumu/ chi-baturuum-u bathroom /bᴂðru:mz/ esinki /esiŋki/ e-sink-i sink /sɪŋk/ chisinki /t∫isiŋki/ chi-sink-i sinks /sɪŋks/ eburasi /eβurasi/ e-buras-i brush /brʌ∫/ chiburasi /t∫iβurasi/ chi-buras-i brushes /brʌ∫ɪz/ egeita /eγeita/ egeita gate /geɪt 295 ebiita /eβiita/ e-biit-a gates /geɪts/ eswenta /eswenta/ eswenta sweater /swetǝ/ chiswenta /t∫iswenta/ chi-swent-a sweaters /swetǝz/ esiting’iruumu /esiting’iruumu/ e-siting’iruum-u sitting room /sɪtɪŋru:m/ chisiting’Iruumu /t∫isitiŋiruumu/ chi-siting’iruum-u sitting rooms /sɪtɪŋru:ms/ ekabati /ekaβati/ e-kabat-i cupboard /kʌbǝd/ chikabati /t∫ikaβati/ chi-kabat-i cupboards /kǝbǝdz/ ebiicha /eβiicha/ e-biich-a picture /pɪkt∫ǝ/ chibiicha /t∫iβiit∫a/ chi-biich-a pictures /pɪkt∫ǝz/ egasi /eγasi/ e-gas-i gas /gᴂz/ chigasi /t∫iγasi/ chi-gas-i gases /gᴂzɪz/ esoba /esoβa/ e-sob-a sofa /sǝʊfǝ/ chisoba /t∫isoβa/ chi-sob-a sofas /sǝʊfǝz/ esito /esito/ e-sito-o store /stͻ:/ chisito /t∫isito/ chi-sito-o stores /stͻ:s/ ebeseni /ɛβɛsɛni/ e-besen-i basin /beɪsn chibeseni /t∫iβɛsɛni/ chi-besen-i basins /beɪsns/ ebatira /eatira/ e-atir-a bottle /bɒtl/ epogisi /epͻγisi/ e-pogis-i box /bɒks/ ekotoni /ɛkͻtoni/ e-koton-i cotton /kɒtn/

Semantic domain…………3 (Agriculture/ farming)- By an agricultural officer

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

296 ekeragita /ekeraγita/ eke-ragit-a tractor /trʌktǝ/ ebiragita /eβiraγita/ ebi-ragit-a tractors /trʌktǝz/ eraini /eraini/ e-rain-i line /lain/ chiraini / t∫iraini/ chi-rain-i lines /lains/ ebambu /eβambu/ e-bamb-u pump /pʌmp/ chibambu / t∫ibambu/ chi-bamb-u pumps / pʌmps/ omokirigacha /omokiriγat∫a/ omo-kirigach-a agriculture officer abakirigacha /aβakirigat∫a/ aba-kirigach-a agriculture officers egurubaro /egurubaro/ e-gurubar-o wheel barrow /wi:lbǝrǝʊs/ chigurubaro /t∫iγuruβaro/ chi-guruba-ro wheelbarrows /wi:lbǝrǝʊ/ eeka /eeka/ e-ek-a acre /eɪkǝ/ chieka / t∫ieka/ chi-ek-a acres /ekǝs/ egirati /eγireti/ e-giret-i grade /greɪd/ chiegireti /t∫iegireti/ chi-e-giret-i grade ones esirasi /esirasi/ e-siras-i slush /slʌ∫/ chisirasi / t∫isirasi/ chi-siras-i slushes /slʌ∫iz/ egwaya |eγuaya| /eγwaya/ e-gway-a wire /wǝɪǝ/ chiwaya |t∫iγuaya| / t∫iwaya/ chi-gway-a wires /wǝɪǝs/ etiibu /etiiβu/ e-tiib-u dip /dɪp/ chitiibu /t∫itiiβu/ chi-tiib-u dips /dɪps/ ebuti /eβuti/ e-but-i feet /fi:t/ chibuti /t∫iβuti/ chi-but-i feets /fi:ts/

Semantic domain…………4 (transport and motoring) – By a motor mechanic

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL FORM SOURCE WORD FORM

297 erori /erori/ e-ror-I lorry /lɒri/ chirori /t∫irori/ chi-ror-i lorries /lɒriz/ esitarinki /esitarinki/ e-sitarink-i steering /stɪǝrŋ/ chisitarinki /t∫isitarinki/ chi-sitarink-i - ebureki /eβureki/ e-burek-i break /breɪk/ chibureki / t∫iβureki/ chi-burek-i breaks /breɪks/ ekerachi /ekera t∫i/ eke-rach-i clutch /klʌt∫/ ebirachi /eβirat∫i/ ebi-rach-i clutches / klʌt∫ɪz/ ekaa /ekaa/ e-ka-a car /kɑ:/ chikaa /t∫ikaa/ chi-ka-a cars /kɑ:/ ebetirori /eβetirori/ e-betiror-i petrol /petrǝl/ omontereba /ͻmͻntɛrɛβa/ omo-ntereb-a driver /draɪvǝ/ abantereba /aβantɛrɛβa/ aba-ntereb-a drivers /draɪvǝz omomakanika /omomakanika/ omo-makanik-a mechanic /mǝkænɪk/ abamakanika /aβamakanika/ aba-makanik-a mechanics /mǝkænɪks/ eboriti /ɛβͻriti/ e-borit-i bolt /bɒlt/ chiboriti /t∫iβͻriti/ chi-borit-i bolts /bɒlts/ ebetiri /ɛβɛtiri/ e-betir-i battery /bætrɪ/ chibetiri /t∫iβɛtiri/ chi-betir-i batteries/bætrɪz/ echeki /et∫ɛki/ e-chek-i jerk /dʒɜ:k/ chicheki /t∫it∫eki/ chi-chek-i jerks /dʒɜ:ks/ etagisi /etaγisi/ e-tagis-i taxi /tæks/ chitagisi /t∫itaγisi/ chi-tagis-i taxis /tæksɪz/ omokondagita /ͻmͻkͻndaγita/ omo-kondagit-a conductor /kǝndʌktǝ/ abakondagita /aβakͻndaγita/ aβa-kondagit-a conductors /kǝndʌktǝz/ etiseri /etiseri/ e-tiser-i diesel /di:zl/ emasini /emasini/ e-masin-i machine /mǝ∫i:n/ chimasini /t∫imasini/ chi-masin-i machines /mǝ∫i:ns/ egerechi /egarat∫i/ e-garech-i garage /gærɑ:ʒ/ chigerechi /t∫igarat∫i/ chi-garech-i garages gærɑ:ʒɪz/ omotoka /ͻmͻtͻka/ omo-tok-a motor car /mǝʊtǝ kɑ:/ emetoka /ɛmɛtͻka/ eme-tok-a motor cars //mǝʊtǝ kɑ:z/ etaeri /etaeri/ e-taer-i tile /taɪl/ chitaeri /t∫itaeri/ chi-taer-i tiles /taɪls/

Semantic domain…………5 (Health) – By Health practitioner- nurse

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL FORM SOURCE WORD FORM

298 mareri /marɛri/ marer-I malaria /mǝleǝriǝ/ tibii /tiβii/ tib-ii tp /ti:pi:/ ekeriniki /ekeriniki/ eke-rinik-i clinic /klɪnɪk/ ebiriniki /eβiriniki/ ebi-rinik-i clinics /klɪnɪks/ mateneti /mateneti/ e-matenet-i maternity /mǝtɜ:nǝti/ chimateneti /t∫imatɛnɛti/ chi-matenet-i maternities / mǝtɜ:nǝtiz/ omonasi /omonasi/ omo-nas-i nurse /nɜ:s/ abanasi /aβanasi/ aba-nas-i nurses /nɜ:sɪz/ emocheri /emͻt∫eri/ emo-cher-i mortuary /mͻ:t∫ǝri/ chimocheri /t∫imͻt∫ɛri/ | chi-mocher-i mortuaries /mͻ:t∫ǝriz/ eambiurensi eambiurenzi|/eambjurenzi/ e-ambiurens-i ambulance/æmbjǝlǝns/ t∫iambiurensi |t∫iambiurenzi|/t∫iambjurenzi chi-ambirens-i ambulances /æmbjǝlǝnsɪz *eteresa */ɛtɛrɛsa/ *e-teres-a dresser *chiteresa *t∫itɛrɛsa/ *chi-teres-a dressers omoteresa /ͻmͻtɛrɛsa/ omo-teresa dresser /dresǝ/ abateresa /aβatɛrɛsa/ aba-teres-a dressers /dresǝz/ ebandechi /eβandet∫i/ e-bandech-i bandage /bændɪdʒ/

Semantic domain…………6 (religion) - By Church elder

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

299 biechii /βiet∫ii/ biech-ii PAG /pi:eɪdʒi:/ katoriki /katoriki/ katorik-i catholic /kæθɒlɪk/ esitiee /ɛsitiee/ esit-iee SDA /esdi:eɪ/ omobata /omoβata/ omo-bat-a father /fǝðǝ/ ababata /aβaβata/ aba-bat-a fathers /fǝðǝz/ omobasita /omoβasita/ omo-basit-a pastor /pʌstǝ/ ababasita /aβaβasita/ aba-basit-a pastors /pʌstǝz/ ebuku /eβuku/ e-buk-u book /bʊk/ chibuku /t∫ibuku/ chi-buk-u books /bʊks/ ekiboti /ekiiβͻti/ e-kibot-i key board /ki:bɒdz/ chikiboti /t∫ikiiβͻti// chi-kibot-i key board /ki:bɒdz/ egiita /eγiita/ e-giit-a guitar /gɪtɑ:/ ebiita /eβiita/ e-biit-a guitars /gɪtɑ:z/ ekerisimasi /ekerisimasi/ eke-risimas-i Christmas /krɪmǝs/ ebirisimasi /eβirisimasi/ ebi-risimas-i - risakaramenti /risakaramenti/ ri-sakarament-i sacrament /sækrǝmǝnt/ amasakaramenti /amasakaramenti/ ama-sakarament-i sacraments /sækrǝmǝnts/ omokatoriki /omokatoriki/ omo-katorik-i a catholic /ækæθlɪk/ abakatoriki /aβakatoriki/ aba-katorik-i catholics /ækæθlɪks/ kirisito /kirisito/ kirisit-o Christ /kraɪst/ ekorasi /ɛkͻrasi// e-koras-i chorous /kɒrǝz/ chikorasi /t∫ikͻrasi chi-koras-i chorouses / kɒrǝzɪz/ ekwaya |ekuaya|| /ekwaja/ e-kway-a choir /kwaɪǝ/ chikwaya |t∫ikuaja| /t∫ikwaja/ chi-kway-a choirs /kwaɪǝz/ omokiristo /omokiristo/ omo-kirist-o a Christian /æ krɪst∫ǝn/ abakirisito /aβakirisito/ ab-akirisit-o Christians /krɪst∫ǝns/ omoesitiee /ͻmͻɛsitiee/ omo-esit-iee an SDA /æn abaesitiee /aβaesitiee/ aba-esit-iee SDAs /es di:eɪz/ omobisopu /omoβisͻpu/ omo-bisop-u bishop /bɪ∫ǝp/ ababisobu /aaisͻpu/ aba- bisop-u bishops /bɪ∫ǝps/

Semantic domain…………7 (education) - By an educationst

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM esukuru /esukuru/ e-sukur-u school /sku:l/ 300 chisukuru /t∫isukuru/ chi-sukur-u schools /sku:lz/ ekerasi /ekerasi/ eke-ras-i class /klæs/ ebirasi /eβirasi/ ebi-ras-i classes /klæsɪz/ eburakiboti /eβurakiβͻti/ e-burakibot-i blackboard /blækbͻ:d/ chiburakiboti / t∫iβurakiβͻti/ chi-burakibot-i blackboards /blækbͻ:dz/ chooka /ɛt∫ͻͻka/ e-chook-a chalk /t∫ͻ:k/ chichooka / t∫it∫ͻͻka/ chi-chook-a pieces of chalk /t∫ͻ:k/ eyunibomu /eyunibomu/ e-yunibom-u uniform /ju:nfͻ:m/ chiyunibomu / t∫iyuniβͻmu/ chi-yunibom-u uniforms /ju:nfͻ:mz/ enasari /enasari e-nasar-i nursery /nɜ:s/ chinasari / t∫inasari/ ch-inasar-i nurseries/ nɜ:sɪz/ eburemari /eβurɛmari e-buremar-i primary /praɪmǝrɪ/ chiburemari / t∫iβurɛmari/ chi-buremar-i primaries /praɪmǝrɪz/ esekondari /ɛsɛkͻndari e-sekondar-i secondary /sekǝndri/ chisekondari / t∫isɛkͻndari/ chi-sekondar-i - eyunibasity /eyuniβasiti e-yunibasit-i university /ju:nɪvǝsɪtɪ/ chiyunibasiti / t∫iyuniβasiti/ chi-yunibasit-i universities /ju:nɪvǝsɪtɪs/ ekorechi ekoret∫i e-korech-i college /kɒlɪdʒ/ chikorechi / t∫ikoret∫i/ chi-korech-i colleges /kɒlɪdʒɪz/ ebaeri /eβaeri/ e-baer-i file /faɪl/ chibaeri / t∫iβaeri/ chi-baer-i files /faɪls etasita /etasita/ e-tasit-a duster /dʌstǝ/ chitasita / t∫itasita chi-tasit-a dusters /dʌstǝz/ ebenchi /eβendƷi/ e-bench-i bench /bendʒ/ chibenchi /t∫iβendƷi chi-bench-i benches /benʒɪz/ etamu /etamu/ e-tam-u term /tǝm/ chitamu /t∫itamu/ chi-ta-u terms tǝmz/ etigiri /etiγiri/ e—tigir-i degree /dɪgri:/ chitigiri /t∫itiγiri/ chi-tigir-i degrees /dɪgri:z/ ekosi /ekosi/ e-kos-i course /cɒs/ chikosi / t∫ikosi/ chi-kos-i courses /cɒsiz/ ewiki |eγuiki|/eγwiki/ e-wik-i week /wi:k/ chiwiki |t∫iγuiki| /t∫iwiki/ chi-wik-i weeks /wi:ks/

Semantic domain…………8 (politics, governance and security) – By an assistant chief

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

301 *eburesitenti /eβuresitenti/ e-buresitent-I president /prezɪdǝnt/ *chiburesitenti / t∫iβuresitenti/ chi-buresitent-i presidents /prezɪdǝnts/ omoburesitent /omoβurɛsitɛnti/ omo-buresitent-i president/prezɪdǝnt/ ababuresitenti /ababurɛsitɛnti/ aba-buresiten-ti presidents /prezɪdǝnts/ *egabana *chigabana omogabana /omoγaβana/ omo-gaban-a governor /gʌvǝnǝ/ abagabana /aβaγaβana/ aba-gaban-a governors /gʌvǝnǝz/ *eseneta *chiseneta omoseneta /omoseeneta/ omo-senet-a senator /sɪnætǝ/ abaseneta /aβaseeneta/ aba-senet-a senators /sɪnætǝz/ *ekansara *chikansara omokansara /omokanzara/ omo-kansar-a counsillor /kaʊnsǝlǝ/ abakansara /aβakanzara/ aba-kansar-a counsillors /kaʊnsǝlǝz/ etigiteta /etiγiteta/ e-tigitet-a dictator /dɪkteɪtǝ/ chitigitete / t∫itiγiteta/ chi-tigitet-a dictators /dɪkteɪtǝz/ ekambi /ekambi/ e-kamb-i cump /kʌmp/ chikambi / t∫ikambi/ chi-kamb-i cumps /kʌmps/ esiteseni /esiteseni/ e-sitesen-i station /steɪ∫n/ chisiteseni / t∫isiteseni/ chi-sitesen-i stations/steɪ∫ns/ etibisoni /etiβisoni/ e-tibison-i division /dɪvɪ∫n/ chitibisoni / t∫itiβisoni/ chi-tibison-i divisions /dɪvɪ∫ns/ erokeseni /erokeseni/ e-rokesen-i location /lɒkeɪ∫n/ chirokeseni / t∫irokeseni/ chi-rokesen-i locations /lɒkeɪ∫ns/ etisiturigiti /etisituriγiti/ e-tisiturigit-i district /dɪstrɪk/ chitisiturigiti / t∫itisituriγiti/ chi-tisiturigit-i districts /dɪstrɪks/ omoepi /omoepi/ omo-eep-i an AP /eɪpi:/ abaepi /aβaepi/ aba-eep-i Aps /eɪpi:s/ omoporisi /omoporisi/ omo-poris-i a police(man/woman) abaporisi /abapͻrisi/ aba-poris-i police(man/woman) ekomitii /ɛkͻmitii/ e-komit-ii committee /kɒmiti:/ chikomitii / t∫ikͻmitii/ chi-komit-ii committees/kɒmiti:s/ erekoti /ɛrɛkͻti/ e-rekot-i record /rekͻ:d/ chirekoti /t∫irɛkͻti/ chi-rekot-i records /rekͻ:dz/ omotisii /omotisii/ omo-tis-i a DC /di:zi:/ abatisii /aβatisii/ aba-tis-i DCs /d:zi:z/ omotioo /omotioo/ omo-tio-o a DO /di:ǝʊ/ abatioo /aβatioo/ aba-tio-o Dos /di:ǝʊz omochibu /omo t∫iβu/ omo-chib-u a chief /t∫i:f/ abachibu /aβac t∫iβu/ aba-chib-u chiefs /t∫i:fs/ omosabuchibu /omosaβuchiβu/ omo-sab-u-chib-u sub chief /sʌbt∫i:f/ abasabuchibu /aβasaβut∫iβu/ aba-sab-u-chib-u sub chiefs/ sʌbt∫i:fs/ omokiraki /omokiraki/ omo-kirak-i clerk /klǝk/ abakiraki /aβakiraki/ aba-kirak-i clerks /klǝks/ 302 erumande /erumande/ e-rumand-e remand /rɪmǝnd/ chirumande / t∫irumande/ chi-rumand-e remands /rɪmǝnds/ sekiuriti /sekiuriti/ sekiuriti security sɪkjʊǝrǝti/ omotureti /omotureti/ omo-turet-i traitor /treɪtǝ/ abatureti /aβatureti/ aba-turet-i traitors treɪtǝz/ risabu /risau/ ri-sabu reserve /rɪsɜ:v/ amasbu aamasau/ a-masabu reserves /rɪsɜ:vz/

Semantic domain……………….9 (legal affairs) –By court officer

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

303 *echachi /et∫at∫i/ e-chachi judge /dʒʌdʒ/ *chichachi /t∫it∫at∫i/ chi-chachi judge /dʒʌdʒɪz/ omochachi /omot∫at∫i/ omo-chach-I judge /dʒʌdʒ/ abachachi /aβachat∫i/ aba-chach-i judges /dʒʌdʒɪz/ *emachisitureti /emat∫isitureti/ e-machisituret-i margistrate /mædʒɪstreɪt/ *chimachisitureti /t∫imat∫isitureti/ chi-machisituret-i margistrates /mædʒɪstreɪts/ omomachisitureti /omomat∫isitureti/ omo-machisituret-i margistrate /mædʒɪstreɪt/ abamachisitureti /aβamat∫isitureti/ aba-machisituret-i margistrates /mædʒɪstreɪts/ ekoti /ekoti/ e-kot-i court /kͻ:t/ chikoti / t∫ikoti/ chi-kot-i courts /kͻ:ts/ omoroya /omoroja/ omo-roy-a lawyer /lͻ:jǝ/ abaroya /aβaroja/ aba-roy-a lawyers /lͻ:jǝz/ omopurosekiuta /omopurosekiuta/ omo-purosekiut-a prosecutor /prɒsɪkju:tǝ/ abapurosekiuta /aβapurosekiuta/ aba-purosekiut-a prosecutors /prɒsɪkju:tǝz/ ebaini /eβaini/ e-bain-i fine /faɪn/ chibaini / t∫iβaini/ chi-bain- fines /faɪnz/ ekesi /ekesi/ e-kes-i case /keɪs/ chikesi / t∫ikesi/ chi-kes-i cases /keɪsɪz

Semantic domain.…10 (information, communication & technology) – By an IT expert

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

304 emesechi /emese t∫i/ e-mesech-I message /meseɪdʒ/ chimesechi / t∫imeset∫i/ ch-imesech-i messages /meseɪdʒiz/ eretio |eretio|/eretjo/ e-ret-io radio /reɪdɪǝʊ/ chiretio |eretio|/t∫iretjo/ chi-ret-io radios /reɪdɪǝʊs/ etibii /etiβii/ e-tib-ii TV /ti:vi:/ chitibii /t∫itiβii/ chi-tib-ii TVs /ti:vi:z/ emobaeri /ɛmͻβaɛri/ e-mobaer-i mobile /mǝbaɪl/ chimobaeri /t∫imͻβaɛri/ ch-imobaer-i mobiles /mǝbaɪlz/ eirioo /eirioo/ e-irio-o aerial /earɪǝl/ chiirioo /t∫iirioo/ chi-irio-o aerials /earɪǝls/ ekombiuta |ɛkͻmbiuta| /ɛkͻmbjuta/ e-kombiut-a computer /kɒmju:tǝ/ chikombiuta |t∫ikͻmbiuta| /t∫ikͻmbjuta/ chi-kombiut-a computers /kɒmju:tǝz/ esenema /ɛsɛnɛma/ e-senem-a cinema /sɪnǝmǝ/ chisenema / t∫isɛnɛma/ chi-senem-a cinemas /sɪnǝmǝz/

Semantic domain…………11 (business/trade) By a Business man

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

305 ebucheeri /ebut∫eeri/ e-bucheer-I butchery /bʊt∫ǝrɪ/ chibucheeri /t∫ibut∫eeri/ chi-bucheer-i butcheries /bʊt∫ǝrɪz/ ekiro /ekiro/ e-kir-o kilo /kɪlǝʊ/ chikiro / t∫ikiro/ chi-kir-o kilos /kɪlǝʊz/ risiti /risiti/ ri-sit-i receipt /rɪsi:t/ amarisiti /amarisiti/ ama-risit-i receits/rɪsi:ts/ eoteri /eoteri/ e-oter-i hotel /hǝʊtel/ chioteri | t∫ioteri| /t∫joteri/ chi-oter-i hotels /hǝʊtels/ epaa /epaa/ e-pa-a bar /bɑ:/ chipaa / t∫ipaa/ chi-pa-a bars /bɑ:z/ epia |epia| /epja/ e-p-ia beer /bɪǝ/ chipia / t∫ipia/ chi-pi-a - etaoni /etaoni/ e-taon-i town /taʊn/ chitaoni / t∫itaoni/ ch-itaon-i towns /taʊnz/ esubamaketi /esuβamaketi/ e-subamaket-i supermarket /su:pǝmǝkeɪt/ chisubamaketi / t∫isuβamaketi/ chi-subamaket-i supermarkets /su:pǝmǝkeɪts/ eturei /eturei/ e-ture-i tray /teɪ/ chiturei / t∫iturei/ chi-ture-i trays /teɪz/ eresesi /eresesi/ e-reses-i license /lɪasǝns/ chiresesi / t∫iresesi/ chi-reses-i licenses /lɪasǝnsɪz/ ekireti /ekereti/ e-kiret-i crate /kreɪt/ ebireti /eβireti// e-biret-i crates /kreɪts/ etasani /etasani/ e-tasan-i dozen /dɒzn/ chitasani / t∫itasani/ chi-tasan-i dozens /dɒznz/ esimiti /esimiti/ e-simit-i cement /sɪmǝnt/ chisimiti / t∫isimiti/ chi-simit-i cements/sɪmǝnts/ echenchi /et∫endƷi/ e-chench-i change t∫eɪndʒ/ chichenchi / t∫ichendƷi/ chi-chench-i - rinoti /rinoti/ ri-not-i a note /nǝʊt/ amanoti /amanoti/ ama-not-i notes /nǝʊts/ esirinki /esirinki/ e-sirink-i shilling /∫ɪlɪŋ/ chisirinki / t∫isirinki/ chi-sirink-i shillings /∫ɪlɪŋz/ emarigiti /emariγiti/ e-marigit-i market /m ɑ:kɪt/ chimarigiti / t∫imariγiti/ chi-marigit-i markets/ m ɑ:kɪts/ ebanki /eβaaŋki/ e-bank-i bank /bæŋk/ chibanki / t∫iβaaŋki/ chi-bank-i banks bank /bæŋks/ echeki /ɛt∫ɛki/ e-chek-i cheque /t∫ek/ t∫it∫ɛki/ chi-chek-i cheques /t∫eks/

Semantic domain…………13 (sports) – By a youths officer

LOAN WORD PHONOLOGICAL FORM MORPHOLOGICAL SOURCE WORD FORM FORM

306 gemusi /γemusi/ gemus-I games /geɪmz/ etiriri /etiriri/ e-tirir-i drill /drɪl/ chitiriri / t∫itiriri/ chi-tirir-i drills/drɪls/ egori /egori/ e-gor-i goal /gǝʊl/ chigori / t∫igori/ chi-gor-i goals /gǝʊlz/ eneti /ɛnɛti/ e-net-i net /nǝt/ chineti / t∫ineti/ chi-net-i nets /nǝts/ esiboti /esβboti/ e-sibot-i sport /spͻ:t/ chisiboti / t∫isiβoti/ chi-sibot-i sports /spͻ:ts/ eribarii /eriβarii/ e-ribar-ii file /faɪl/ chiribarii / t∫iriβarii/ chi-ribar-ii files /faɪls/ emeratoni /emeratoni/ eme-raton-i marathon /mærǝӨǝn/ chimaratoni /t∫imaratoni/ chi-maraton-i marathons/mærǝӨǝnz/ gwoking’i resi |γuͻkiŋiresi| /gwͻkiŋresi/ gwoking-i- res-i walking race /wͻ:kiŋreɪs/ *ekabuteni */ekaβuteni/ *e-kabuten-i captain /kæpteɪn/ *chikabuteni */t∫ikaβuteni/ *chi-kabuten-i captain/kæpteɪns/ omokabuteni /omokaβuteni/ omo-kabuten-i/ captain /kæpteɪn/ abakabuteni /aβakaβuteni/ aba-kabuten-i/ captain/kæpteɪns/

Appendix VI: Illustrations of phonological nativization

Source word form Nativized form phonological proces

christmas /krɪsmǝs/ [ekirisimasi] segment change

307 taxi /tæksɪ/ [etagisi] segment change sofa /sǝʊfǝ/ [esoba] monophthongization wire /wǝɪǝ/ [ewaya] monophthongizaiton finen /faɪn/ [eaini] bilabialization vest /vest/ [eesiti] bilabialization store /stͻ:/ [esitoo] expenthesis of [i] fine /faɪn/ [ebaini] epenthesis of [i] coat /kǝʊt/ [eγoti] vowel harmonization basin /bæsn [εεsεni] vowel harmonization

Pastor /pʌstǝ/ [asita] stop fricativization bolt /bͻlt/ [ͻriti] stop fricativization drink /drɪŋk/ |eturun-γi|→[eturuŋgi] fricative defricativization camp /kæmp/ |ekan-i|→[ekembi] fricative defricativization bank /bæŋk/ [eeŋgi] voice dissimilation location /lǝʊ'keɪ∫n/ [rokeHseni] stress tonemization degree /dɪgri:'/ [tigiHrii] stress tonemization school /sku:l./ [su.ku.ru.] syllable change

Appendix VII: Illustrations of morphological nativization

Source word form Nativized form Morphological process scout o-mo-sikaoti nominal classification (1)

308 scout-s a-ba-sikaoti nominal classification (2) motor-car o-mo-tokaa nominal classification (3) motor-car-s e-me-tokaa nominal classification (4) torch (very big) ri-toochi nominal classification (5) torch-es ama-toochi nominal classification (6) school (very small) e-ke-sukuru nominal classification (7)

(small school (deminution)) school-s e-bi-sukuru nominal classification (8)

(small schools (deminition)) record ε-rεkͻti nominal classification (9) record-s chi-rεkͻti nominal classification (10) room (very small) aka-ruumu nominal classification (12)

deminution room-s (very small) obo-ruumu nominal classification (14) ticket e-tiketi prefixication ticket-s chi-tiketi prefixication governor o-mo-gabana pre-prefixication governor-s a-ba-gabana pre-prefixication

Appendix VIII: Enchoro sub-location villages 1. Sarama

2. Nyagenke

3. Ikaraancha

4. Esuguta 309 5. Nyakoria

6. Nyankarankania

7. Nyantaro

8. Nyagaachi

9. Motagara

10. Mariba A

11. Mariba B

12. Chumura

13. Mosobeti

14. Getukora

15. Enchoro

16. Ikarancha

17. Nyando

310 Appendix IX: Research Clearance Permit

311 Appendix X: Research Authorization Letter

312 Appendix XI: Memo from the Registrar, Research and Extension, Kisii University to NACOSTI

313 314 07-Nov-2017 10: 35AM (UTC + 0300)

Appendix XII: Plagiarism Report

07-Nov-2017 10: 35AM (UTC + 0300)

315 316 317 Appendix XIII: Publication

318 319 320 321 322 323