Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
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
Signhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip Datehelliphelliphelliphelliphelliphelliphelliphelliphellip
George Morara Anyona
DAS13600102014
DECLARATION BY THE SUPERVISORS
This thesis has been submitted with our approval as university supervisors
Sign helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip Datehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
Dr David O Ongarora PhD
Department of Linguistics Maseno University
Sign helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip Datehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
Dr Evans G Mecha PhD
Department of Lit Lang amp 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
thesisproject can be assigned a fail grade (ldquoFrdquo)
vi I further understand I may be suspended or expelled from the university for academic
dishonesty
Name George Morara Anyona Signaturehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
Reg No DAS13600102014 Datehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
DECLARATION BY SUPERVISOR(S)
i Iwe declare that this thesis has been submitted to plagiarism detection service
ii The thesis contains less than 20 of plagiarized work
iii Iwe hereby give consent for making
1 Name Dr David O Ongarora PhD Signaturehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
Affiliationhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip Datehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
2 Name Dr Evans G Mecha PhD Signaturehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
Affiliationhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip Datehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
iii
DECLARATION OF NUMBER OF WORDS
Name of Candidate GEORGE MORARA ANYONA Adm No DAS13600102014
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
Signedhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip Datehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip
(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
SignedhelliphelliphelliphelliphelliphelliphelliphelliphellipEmailhelliphelliphelliphelliphelliphelliphellipTelhelliphelliphelliphelliphelliphellipDatehelliphelliphelliphelliphellip
(Supervisor 1)
SignedhelliphelliphelliphelliphelliphelliphelliphelliphellipEmailhelliphelliphelliphelliphelliphelliphellipTelhelliphelliphelliphelliphelliphellipDatehelliphelliphelliphelliphellip
(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
copy 2017 Anyona Morara George
v
ABSTRACT
This study examines the phonology and morphology of English nouns borrowed into EkeGusiiwithin Optimality Theory This theory provides that the well formedness in natural languages isconstraint governed Thus even borrowed lexical items obey these constraints Constraints areuniversal and are ranked on a language specific order The focus of this study is to investigate thephonological and morphological adjustments that English nouns borrowed into EkeGusiiundergo as they harmonize themselves with EkeGusii phonological and morphological systemsas they adapt EkeGusii constraints ranking order The study analyzes segmental phonotactic andsuprasegmental adaptations that English nouns borrowed into EkeGusii undergo in the process ofbeing nativized Further the study analyzes the affixation processes that characterize thesenouns The study analyses nominal prefixation augmentation and classification The study wasguided by the following objectives to describe the phonological and morphological structures ofEkeGusii and English nouns to analyze the phonological adjustments that English nounsborrowed into EkeGusii undergo during nativization and to analyze the morphological changesthat the English noun borrowed into EkeGusii undergo during nativization The study adoptedthe descriptive and the explanatory research designs Data was collected from a sample of 13interviewees using semi-structured interviews The interviewees were purposively sampledbased on the semantic domains of the nouns collected All the 349 English nouns borrowed intoEkeGusii colleccted from the field constituted the sample size of the study secondary data wasused to describe the phonological and morphological strucures of EkeGusii in response toquestion 1 Library study and researcherrsquos intuition were the sources of the secondary data Totest the validity and reliability of the research instrument pre-testing was carried out andappropriate adjustments and corrections made on the instument The study yielded the followingresults The first objective describes the phonological and morphological structures of Englishand EkeGusii It was established that Phonologically the acoustic nature of the vowels in thetwo 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 inEnglish Morphologically EkeGusii nouns unlike those of English are grouped into classesdetermined by the semantics of the root of the noun and affixation in the two languages isdifferent The second objective analyzed the phonological adjustments that English nounsborrowed into EkeGusii undergo It was established that phonemes of English not present inEkeGusii are substituted for those present in EkeGusii the foreign syllable structure fromEnglish is re-syllabified to EkeGusii syllable structure and English stress is tonemized inEkeGusiiThe third objective analyzed the morphological changes that English nouns borrowedinto EkeGusii undergo during nativization It was found that English nouns enter EkeGusiinominal classes which are semantically determined nativization of the English nouns borrowedinto EkeGusii is characterized by the pre-prefix and semantics plays a significant role inmorphological nativization of English nouns in EkeGusii This study has establisshed thatOptimality Theory can successively analyse English nouns borrowed into EkeGusii and that thenouns obey EkeGusii constraint ranking It is hoped that this study would contribute to thetheoretical understanding of borrowed word phonology and morphology of EkeGusii inparticular 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
DECLARATIONii
ix
PLAGIARISM DECLARATIONiii
DECLARATION OF NUMBER OF WORDSiv
COPYRIGHTv
ABSTRACTvi
DEDICATIONvii
ACKNOWLEDGEMENTviii
TABLE OF CONTENTSx
LIST OF ABBREVIATIONS SYMBOLS TABLES TABLEAUX FIGURES AND CHARTS
xvii
CHAPTER ONE
INTRODUCTION1
10 Background to the study1
11 Statement of the problem5
12 Objectives of the research6
13 Research questions6
14 Justification and significance of the study6
15 Scope and limitation9
CHAPTER TWO
LITERATURE REVIEW AND THEORETICAL FRAMEWORK12
x
20 Introduction12
21 EkeGusii phonological and morphological structures14
211 EkeGusii phonological structure14
212 EkeGusii morphological structure18
22 Phonological nativization19
23 Morphological nativization25
24 Theoretical framework28
241 Tenets of Optimality Theory28
242 Optimality Theory versus other Generative Approaches35
CHAPTER THREE
RESEARCH METHODOLOGY40
30 Introduction40
31 Research design40
32 Research site41
33 Study populations41
34 Sample size and sampling techniques42
35 Data collection procedures43
36 Data analysis43
361 Procedure data analysis44
37 Ethical considerations45
CHAPTER FOUR
xi
DATA PRESENTATION ANALYSIS RESULTS AND DISCUSSIONS46
40 Introduction46
41 Phonological and morphological structures of EkeGusii and English46
411 EkeGusii and English vowel systems47
4111EkeGusii vowel harmony and disharmony59
412 EkeGusii consonant system in comparison with English63
4121 EkeGusii consonantal processes70
41211 Voice dissimilation (Dahlrsquos Law)70
41212 Prenasalisation and nasal homorganisation72
41213 Declusterization of nasal consonant (NC) and consonant glide (CG) clusters73
41214 Defricativisation82
41215 EkeGusii Nasal re-syllabification83
413 Comparison of EkeGusii and English syllable structures89
4131 EkeGusii and English phonotactics102
41311 English word initial phoneme sequences103
41312 English word final phoneme sequences105
414 Comparison of EkeGusii toneme structure and English stress107
4141 EkeGusii tone structure107
41411 Underlying and surface tones in EkeGusii109
41412 Contour tones in EkeGusii110
41413 Tone floating in EkeGusii112
xii
415 Comparison of EkeGusii and English morphological structures114
4151 Comparison of EkeGusii and English noun systems115
4152 Morphosyntactic classes and prefixes in EkeGusii noun116
41521 The pre-prefix or augment121
41522 The Prefix125
41523 EkeGusii noun class roots127
41524 Class 1 omo- and 2 aba-128
41525 Classes 3 omo- and 4 eme-128
41526 Classes 9 e- 10 chi- 9 (a) en- 10 (a) chin-130
41527 Classes 5 eri-rii- 6 ama- 7 eke- 8 ebi- 12 aka- and 14 obo-131
42 Phonological nativization of English nouns in EkeGusii135
421 Segmental nativization136
4211 Nativization of vowel segments137
42111 Nativization of English pure (monophthong) vowels138
42112 Nativization of English diphthongs157
42113 Nativization of English triphthongs166
4212 Nativization of English consonants173
42121 Nativization of English f and v174
42122 Nativization of English eth and 179
42123 Nativization of English l183
42124 Nativization of English z187
xiii
42125 Nativization of English int and ʒ189
42126 Nativization of English g in EkeGusii192
422 Phonotactic nativization195
4221 Nativization by vowel epenthesis200
42211 Nativization of complex syllable margins201
42212 Nativization of closed syllables205
423 Supra-segmental nativization210
424 Nativization by phonological processes216
4241 Nativization by voice dissimilation (Dahlrsquos Law)216
42421 nativization by defricativization225
42422 Spirantizationfricativization of the stops230
42423 Nativization by bilabialization of labiodentals233
4243 Nativization of nasal consonant clusters (NC) by pre-nasalization and homorganization
235
4244 Nativization by vowel height harmonization and disharmonization239
43 Morphological nativization241
431 Nativization by nominal classification243
4311 The role of semantics in morphological nativization of English nouns in English252
4312 Nativization into classes 1 and 2 Nouns262
4313 Nativization of English loans into EkeGusii in classes 1b mo- and 2b [ba-264
4314 Nativization into classes 3 and 4265
xiv
4315 Nativization in classes 5 eri-rii- 6 ama- 7 eke- 8 ebi- 12 aka- and 14
obo-265
432 Prefixation and pre-prefixation268
4321 Nativization by prefixation269
4322 Nativization by pre-prefixation272
CHAPTER FIVE
SUMMARY CONCLUSIONS AND RECOMMENDATIONS276
51 Summary276
52 Conclusions281
53 Recommendations282
REFERENCES285
APPENDICES307
Appendix I Interviewee profile form307
Appendix II Interviewee consent form308
Appendix III Research instrument semi-structured interview guide309
Appendix IV Raw data recordingtranscription form312
Appendix V Raw data313
Appendix VI Illustrations of phonological nativization327
Appendix VII Illustrations of morphological nativization328
Appendix VIII Enchoro sub-location villages329
Appendix IX Research Clearance Permit330
xv
Appendix X Research Authorization Letter331
Appendix XI Memo from the Registrar Research and Extension Kisii University to NACOSTI
332
Appendix XII Plagiarism Report333
Appendix XIII Publication336
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 outputFAITH V an Optimality Theory constraint that demands that vowels in the input be
the same as the vowels in the outputIDENT-IO (F) the specification that the features of an input segment must be preserved
in its output correspondentINDENT-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 RTR)) 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
rarr Arrow a phonological form realized as
gt Morphological form meaning realized as
gtgt Domination (constraint) in ranking
σ Syllable node
micro Syllabic molar
Constraint violation
Optimal candidate
Fatal violation
Unbroken association line indicating prelinking in Autosegmental
Phonology
Broken association lines indicating linking in Autosegmental phonology
ndash Tone marker for level tone` Tone marker (low tone)acute Tone marker (high tone) syllable markerOslash A nonexistent segment to be substituted or inserted for another
xxi
TABLES
Table 1 Some of the OT constraints from the universal sethelliphelliphelliphelliphelliphelliphelliphelliphellip 31
Table 2 F1 and F2 formant frequencies in hertz (Hz) of EkeGusii vowelshelliphelliphellip 51
Table 3 F1 and F2 formant frequencies in hertz (Hz) of EkeGusii vowels[i e ͻ u]hellip 56
Table 4 F1 and F2 formant frequencies in hertz (Hz) of the English vowels [i e ͻ u]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 56
Table 5 Typology of syllable shapeshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 89
Table 6 EkeGusii prefixes and their stems semantic determinantshelliphelliphelliphelliphellip 118
xxii
TABLEAUX
Tableau 21 Modern English realization of the input intip helliphelliphelliphelliphelliphelliphelliphellip34
Tableau 22 Comparative tableau for Yawelmani input Ɂilk-hinhelliphelliphelliphelliphelliphellip37
Tableau 23 Violation tableau for Yawelmani input Ɂilk-hinhelliphelliphelliphelliphelliphelliphellip 38
Tableau 41 English output of the input krɪsmǝshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 141Tableau 42 EkeGusii output of the input kirismasihelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 142Tableau 43 English input and output of intɜt helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 148Tableau 44 EkeGusii output of the input esati helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 148Tableau 45 EkeGusii output of the input ekerasihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 150Tableau 46 English input and output of intɜt helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 150Tableau 47 English input and output of sɪneɪthelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 153Tableau 48 EkeGusii output of the input esenetihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 154Tableau 49 EKeGusii realization of the input erokesenihelliphelliphelliphelliphelliphelliphelliphelliphellip160Tableau 410 English output of the input lǝʊkeɪintnhelliphelliphelliphelliphelliphelliphellip 160Tableau 411 EkeGusii realization of the input wajahelliphelliphelliphelliphelliphellip 171Tableau 412 English input and output of waɪǝhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 171Tableau 413 EkeGusii output of the input euritintihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 177Tableau 414 English input and output of frɪdʒhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 177Tableau 415 Tonga realization of the English input frɪdʒhelliphelliphelliphelliphelliphelliphelliphellip 179Tableau 416 English output of the input Өɜməshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 182Tableau 417 EkeGusii output of the the input etamosihelliphelliphelliphelliphelliphellip helliphelliphellip182Tableau 418 EkeGusii output of the the input ekerasihelliphelliphelliphelliphelliphelliphelliphelliphelliphellip186Tableau 419 KiKamba realization of the English input skulhelliphelliphelliphelliphelliphellip 186Tableau 420 EkeGusii output of the the input esirohelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 188Tableau 421 English output of the input zirəʊhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 189Tableau 422 EkeGusii output of the input esati helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 191Tableau 423 English output of the input intɜt helliphelliphelliphelliphellip helliphelliphelliphelliphelliphelliphellip 191Tableau 424 EkeGusii output for the input epaγihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 194Tableau 425 English output for the input baeligghelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 194
Tableau 426 EkeGusii output and input of mamahelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 198
Tableau 427 English output for the input ǝʊhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 199
Tableau 428 English output of the input teɪkhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 199
Tableau 429 English output of the input steɪnshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 199
Tableau 430 English output of the input skulhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 203
Tableau 431 EkeGusii output of the input sukuruhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 203
xxiii
Tableau 432 English output of the input tͻtinthelliphelliphelliphelliphelliphelliphelliphelliphelliphellip helliphelliphellip207
Tableau 433 EkeGusii output for the input [tͻͻtinti]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 207
Tableau 434 Lanakel output of the input [t-n-ak-ol]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 208
Tableau 435 Lanakel output of the input [no-n-koma-i]helliphelliphelliphelliphelliphelliphelliphelliphellip 210
Tableau 436 EkeGusii output of the input oraŋgeti helliphelliphelliphelliphellip 215
Tableau 437 EkeGusii output of the input okokorohelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 219
Tableau 438 English output of the input input baeligŋkhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip222
Tableau 439 EkeGusii output of the English input eeŋgihelliphelliphelliphelliphelliphelliphelliphellip222
Tableau 440 EkeGusii output of the input kanihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip229
Tableau 441 EkeGusii output for the input etaγisihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip232
Tableau 442 EkeGusii output of the input arandahelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 234
Tableau 443 EkeGusii output for the input aranda (unit)helliphelliphelliphelliphelliphelliphelliphellip 237
Tableau 444 EkeGusii output of the input εntεrεahelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 240
Tableau 445 EkeGusii output for the neutral input nkookohelliphelliphelliphelliphelliphelliphellip 244
Tableau 446 EkeGusii output for the input nkooko helliphelliphelliphelliphelliphelliphelliphellip 245
Tableau 447 English output for the singular form input boyhelliphelliphelliphelliphelliphellip 246
Tableau 448 English output of the input boy-shelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 246
Tableau 449 English output the plural input blanket-shelliphelliphelliphelliphelliphelliphelliphellip248
Tableau 450 EkeGusii output of the English plural input blanket-shelliphelliphellip 248
Tableau 451 English output of the input tractor-shelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 254
Tableau 452 EkeGusii output of the English input tractor-shelliphelliphelliphelliphelliphellip 255
Tableau 453 EkeGusii output for input omo-tehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 257
Tableau 454 English output for the input records-shelliphelliphelliphelliphelliphelliphelliphelliphellip 258
xxiv
Tableau 455 English output of the input recordhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 259
Tableau 456 EkeGusii output of the English singular input record-shelliphellip 260
Tableau 457 Tableau (457) EkeGusii output of the English input record-shelliphellip 261
Tableau 458 EkeGusii output of the English input omo-chiibuhelliphelliphelliphelliphellip 263
Tableau 459 English output of the input chiefhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip264
Tableau 460 EkeGusii output of the English input blankethelliphelliphelliphelliphelliphelliphellip 270
Tableau 461 EkeGusii output of the English input gaβanahelliphelliphelliphelliphelliphelliphellip 275
CHARTS Page
Chart 1 EkeGusii vowelshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 47Chart 2 EkeGusii vowel diagramhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 48Chart 3 EkeGusii consonantshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 64Chart 4 EkeGusii consonant inventoryhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 66Chart 5 The English consonantshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 68Chart 6 Description of English ʌ and EkeGusii ahelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 146Chart 7 Production of the English diphthong [aɪ]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 163Chart 8 Production of EkeGusii nativized form [ae] from the English
Diphthong [aɪ]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 163
xxv
FIGURES Page
Figure 1 Process of candidate elimination in OThelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip30
Figure 2 Process of OT realization of outputhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip44Figure 3 Spectrograms of EkeGusii vowels [a e ε i ͻ o u] collected from a
Native speakerhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 50Figure 4 Plot of F1 against (F2) formant frequencies in hertz (Hz) of EkeGusii
vowelshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip51
Figure 5 Spectrograms of 8 British English vowelshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 57
Figure 6 English diphthongshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip58
Fifure 7 Articulation of complex glides in EkeGusiihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 76
Figure 8 EkeGusii consonant glide syllabic nodeshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 77
Figure 9 Nasal consonant syllabification in EkeGusiihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 79
Figure 10 EkeGusii vowel lengtheninghelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 81
Figure 11 EkeGusii compensatory lengtheninghelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 81
Figure 12 EkeGusii de-linking rulehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip85
Figure 13 EkeGusii nasal re-syllabification rulehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 86
Figure 14 Nasal delinking leading to lasal syllabificationhelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 87
Figure 15 Nasal consonant cluster forming part of a syllablehelliphelliphelliphelliphelliphelliphelliphelliphellip 87
Figure 16 EkeGusii syllable nodes for ketiihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 100
Figure 17 EkeGusii syllable nodes for [omote] and [aaa]helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 101
Figure 18 EkeGusii lexical contrasts due to tone structurehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 108
xxvi
Figure 19 EkeGusii lexical contrasts due to tone structurehelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 109
Figure 20 EkeGusii tone markinghelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 110
Figure 21 Mende contour toneshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 111
Figure 22 EkeGusii LH contour tonehelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 111
Figure 23 Ekegusii level tone 112
Figure 24 EkeGusii floating tone helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 112
Figure 25 EkeGusii prefix and pre-prefix syllable nodes helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 121
Figure 26 Substitution of the English ɪ for EkeGusii [i]helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 139
Figure 27 EkeGusii nativization of the English central vowels helliphelliphelliphelliphelliphelliphelliphellip 155Figure 28 Substitution of the English ǝɪǝ for EkeGusii [aja]helliphelliphelliphelliphelliphelliphelliphelliphellip 167Figure 29 Substitution of the English f for EkeGusii [β]helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 176Figure 30 Substitution of the English Ө for EkeGusii [t]helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 180Figure 31 Substitution of the English l for EkeGusii [r]helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 183Figure 32 Substitution of the English z for EkeGusii [s]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip187Figure 33 Substitution of the English int for EkeGusii [s]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 190Figure 34 Substitution of the English g for EkeGusii [γ]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip193Figure 35 EkeGusii syllable types helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 197
Figure 36 English syllable typeshelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 197
Figure 37 Nativization by anaptyxishelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 202
Figure 38 Increase of number of Syllables in nativized word forms due to anaptyxishelliphelliphellip204
Figure 39 Opening of the English closed syllables by paragogic epenthesis in
EkeGusiihelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 206
Figure 40 EkeGusii dissimilation from |γ| rarr [k]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 218
Figure 41 Substitution of the English k for EkeGusii [g]helliphelliphelliphelliphelliphelliphelliphelliphelliphellip 220
Figure 42 Substitution of the English d for EkeGusii [t] English k for EkeGusii [g]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 220
Figure 43 Realization of EkeGusii γ as [g] helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 226Figure 44 Realization of EkeGusii as [b] helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 227Figure 45 Realization of EkeGusii r as [d]helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip 227
xxvii
Figure 46 English and EkeGusii treatment of nasal + consonant sequenceshelliphelliphellip 237
xxviii
CHAPTER ONE
INTRODUCTION
10 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 (200220) EkeGusii is ldquoan 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 345 vol 311-15)rdquo 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 ldquoin the language itself its proper name is
[ekeγusii] (EkeGusii) it is spoken by the [aβaγusii] (AbaGusii) peoplerdquo (2002 20) This is
supported by the observation by Demuth (2000) and Ongarora (2009) that names of the
languages in Bantu fall under classes 78 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 (E43) is
spoken to the south of EkeGusii and ldquoof all the East Nyanza Bantu Languages it is probably most
closely related to EkeGusiirdquo (Cammenga 200221) 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 (1970432) sees nativization as the ldquoattempted reproduction in one language patterns
previously found in anotherrdquo Hock (1991390) defines it as ldquothe integration of foreign words
into onersquos native structuresrdquo 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 nounsEkegusii borrowed noun source noun source language glosseγetaβu Kitabu Swahili book emesa meza Kiswahili tableesukuru School English school| eretio| eretjo radio English radio|riraβuani| rirawani rabuon Dholuo sweet potato ɳasae nyasaye Dholuo God Source Bosire amp 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 amp Noyer (1997) Jacobs amp Gussenhoven (2000)
Kang (2003) Mberia (2004) Rose amp Demuth (2006) Mwita (2009) Owino (2003) Meiseburg
(1993) Hall amp Hamann (2003) Heinemann (2003) Dupoux amp Peperkamp (2002) Lacharite amp
Paradis (2005) Zivenga (2009) Raiz (2011) and Chebanne amp 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 (targetrsquos or sourcersquos)
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
11 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
12 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
13 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
14 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
(Whiteley1960 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 amp 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
15 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
Dahlrsquos 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 transportmotoring information
technology education politics and governance health diplomacy business and trade legal
affairs agriculturefarming 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
20 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 21
which addresses objective one of the study 22 deals with objective two 23 objective three
and 24 which deals with the theoretical framework within which the study was conducted
21 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 211 phonological
structure of EkeGusii and 212 morphological structure of EkeGusii
211 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 amp 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 ampWeenink 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
(Keith1997105)
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 speakerrsquos 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 amp Downing
2005)
Pioneering studies in EkeGusii phonology (Whiteley 1965 amp 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 (Dahlrsquos 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 1963amp 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 srkatintǝ and sixths sikss 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 (OrsquoConner 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
212 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
22 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 Hockrsquos 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 Hockrsquos 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 Hallrsquos amp Hamannrsquos study deal with vowel phonemes only the present study deals with
consonants as well Hall amp 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 Miaorsquos 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 Miaorsquos 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 Dahlrsquos Law in phonemic change during
nativization Miarsquos 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] lsquobusrsquo
b) Labial + u [temu] lsquotermrsquo
In these nouns vowels have been epenthesized ([i] in lsquobuzirsquo and [u] in lsquotemursquo) 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 (ie 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
Kimrsquos 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 Kimrsquos 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
Zivengersquos 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 (Dahlrsquos 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 Mwihakirsquos this study
considers the changes that loaned words undergo at the syllable level However unlike
Mwihakirsquos 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 Mwihakirsquos 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 Mberiarsquos
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 Dahlrsquos law and therefore the study is instrumental in understanding the
operation of Dahlrsquos 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 Owinorsquos deal with two linguistic areas ndash morphology and phonology Finally while the
proposed study will be anchored on the constraint based OT paradigm Owinorsquos is anchored on
Auto segmental and CV ndash 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 Mutuarsquos
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 Mutuarsquos 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 Nashrsquos analyses are not
anchored on any theory The present study like that of Bickmore is anchored on OT But while
Bickmorersquos analyses the verb this study analyses the noun
23 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 Islamrsquos 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) Islamrsquos study is just descriptive it is not anchored in
23
any theory Secondly while in Islamrsquos 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 Kayigemarsquos 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 Kayigemarsquos 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 Kayigemarsquos analyzes data using a
theory Optimality Theory In other words the morphological changes realized in this study are
analyzed and explained theoretically unlike in Kayigemarsquos
25
24 Theoretical framework
This study is anchored on concepts and ideas drawn from Optimality Theory (OT) as expounded
by Prince amp Smolensky (19932004) 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)
241 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 19932004) 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 onoff 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 minimallyleast 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 ndash 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) (Kager1999) 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 choosesselects
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 gtgt C2 gtgt hellipCn) 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
Input GEN
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 INTERPRETATIONCOMPLEX (C) No complex marginsNo consonant clusters
28
Cn
C2C
1a
b
c
c
e
chellip
gtgt gtgt
Output
COMPLEX (V) No strings of vowelsNo complex vowelsFAITH C The consonants in the input are the same as the consonants in the outputFAITH V The vowels in the input are the same as the vowels in the outputIDENT-IO (F) The specification for the features of an input segment must be preserved
in its output correspondentINDENT-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 amp Smolensky (2004) McCarthy (20072008)
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 (200638)
observes ldquoconstraint violability is pervasive in applications of OT but there are two areas of
morphology in which it assumes particular importance affix location and Template
morphologyrdquo 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-agrave-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 ndash 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 int as in scip [skip] rarr ship [intip] (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 ([intip]) Thus the faithfulness constraints such as MAX IO is obeyed at the
expense of the markedness constraintCOMPLEX C (MAX IO gtgt COMPLEX C) as analyzed
in tableau (1)
input intip MAX IO COMPLEX C
a) intip
31
b) skip
Tableau (21) Modern English realization of the input intip
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
242 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 amp 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)
OTrsquos 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 targetheterogeneity 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 LTableau (22) Comparative tableau for Yawelmani input Ɂilk-hin
Source McCarthy (20077)
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 (23) 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
30 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
31 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 amp 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 amp 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
32 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 commissionerrsquos 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 heshe 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
33 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 2205669
38
However the accessible population of the study was the number of EkeGusii native speakers in
the selected study area in (32) 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 thesesdissertations 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
34 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
35 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 responsesesnarratives were recorded by a voice recorder and later transcribed using
a raw data recordingtranscription form (see appendix iv)
36 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
361 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
37 Ethical considerations
According to Kumar (1999) ldquoin every discipline it is considered unethical to collect information
without the knowledge of the participants in a research and their expressed willingness and
informed consentrdquo 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
40 Introduction
This chapter is organized into three major sub-sections 41 presents a theory-neutral generalized
description of the phonological and morphological structures of EkeGusii and English while 42
and 43 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
41 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 42 and 43 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 200834) Data described in this sub-section is secondary as
gathered from existing literature including published books dictionaries theses and journals
43
411 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 (200239) as repeated
in chart (1) i e ɛ a ͻ o uHigh + + - - - + +Mid - + + - + + -Back - - - - + + +ATR + - - + - - +Chart (1) EkeGusii vowels Other studies carried out in the language (Osinde 1986 Ongarora 1996 2009 amp 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 vowelsVowel Examples of words Gloss i siba sia tie
44
ii siiba siia sipe embori embori goatee embeera embeera graveεε orobeere ͻrͻεεrε titaa abaana aaana childrenͻ omoeto ͻmͻεtͻ trapͻͻ omoonia ͻmͻͻnia selleru ekeguuru ekeγuuru small potuu ebiguuru eiγuuru 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
Time (s)0 0136
0
5000
Fre
quency (
Hz)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000 ε
Time (s)0 007152
0
5000
Fre
qu
en
cy (
Hz)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000sound ee
Time (s)0 03045
0
5000F
req
uen
cy (
Hz)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000EkeGusii sound a
Figure (3) Spectrograms of EkeGusii vowels [a e ε i ͻ o u] collected from a native speaker The vowels in figure (3) can be listed against their F1 and F2 as in table (2) as followsTable (2) F1 and F2 frequencies in hertz (Hz) of EkeGusii vowels Vowel F1 F2i 540 2450e 730 2250ε 830 2100a 900 1850ͻ 750 1250o 590 1150u 520 1100
46
Time (s)0 008388
0
5000
Fre
quency (
Hz)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000sound ou
Time (s)0 007639
0
5000
Fre
qu
en
cy (
Hz)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000sounnd i
a
Time (s)0 1189
0
5000
Fre
qu
en
cy (
Hz)
0594285714u
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000o
Time (s)0 007639
0
5000
Fre
qu
en
cy
(H
z)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000sounnd ii
Time (s)0 0136
0
5000
Fre
qu
ency (
Hz)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000ͻ
400
500
600
700
800
900
1000
2500 2000 1500 1000 500
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
Figure (4) Plot of F1 against (F2) formant frequencies in hertz (Hz) of EkeGusii vowelsThis plot shows that the first formant corresponds inversely to the height dimension (high
vowels have low F1 and low vowels have a high F1) and the second formant corresponds to the
advancement (frontback) 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
OrsquoConnor 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
HZ F2
HZ F1
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 howersquod 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 amp Jacobs 2011 Ladefoged 2001 Roach 1983 and
OrsquoConnor 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 amp 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 vowelsEnglish 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 2450e 730 2250ͻ 750 1250u 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 2250e 400 1920ͻ 590 850u 310 890
51
Figure (5) Spectrograms of 8 British English vowels(Source Ladefoged and Keith 2001175)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 (198319) 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
4111EkeGusii 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 frontback 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 lsquolet us gorsquo Omo-te lsquotreersquo b) ͻ-rɛntir-e lsquoheshe has broughtrsquo e-ɳuͻm -ͻ lsquomarriagersquo ti-to-ko-ɳa-koβa-tɛβ-i lsquowe will not be telling themrsquo
EkeGusii has two mid vowel heights High mid and low mid (Ongarora 1996 Rhodes 2010
Anyona 2011 amp 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 disharmony8) EkeGusii vowel height harmony and disharmonyi) 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 personAdapted from Bosire amp 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 lsquotriggerrsquo refers to the vowel with which all other vowels agree in
certain feature(s) while lsquotargetrsquo 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] lsquoownerrsquo
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| gt
[eγe-suunt-e] lsquodarknessrsquo 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] lsquoeyesrsquo 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 ɛͻ 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 pairingATR RTRe ɛo ͻThe 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 RTResese esese lsquodogrsquo ɛsɛsɛ esese lsquocoughrsquoomoonto omoonto lsquopersonrsquo ͻmͻɛtɛ lsquotraprsquoookombe obokombe lsquohoersquo ͻmͻxɛrɛ lsquoLuorsquoAdapted from Bosire amp Machogu (2013)English is not characterized by this phonological process According to Shapiro (2015) English
lacks vowel harmony as a regular phenomenon
412 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 (200253) 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 ntilde ɳ
Glide w y
[Cw] [Cy]
Chart (3) EkeGusii consonantsThis study makes the following observations about consonant inventory in chart (3) Firstly it
should be noted that Cammengarsquos (2002) inventory of EkeGusii consonants is an improvement
of Whiteleyrsquos (1960) inventory In Whiteleyrsquos (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 ldquoone or two
idiophonesrdquo as suggested by Whiteley The idiophone suggested by Whitely would be the
emphatic form pi which means lsquocompletelyrsquo as illustrated by (12)
12) EkeGusii ideophone with the voiceless stop pi) ita pi ita pi lsquokill completelyrsquoii) geenda pi γɛɛnda pi lsquogo completelyrsquoiii) koora pi koora pi lsquofinish completelyrsquoAdapted from Bosire and Machogu (2013)This data shows that pi in the words emphasizes the given actionsSecondly Cammenga (2002) replaces ny with ntilde 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 articulations58
(Hayman amp 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 tint 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]
tint
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 areɸ voiceless bilabial continuant as in obuba oɸuɸa lsquofoodrsquo amaraba
amaraɸa lsquosoilrsquo abasaacha aɸasaatinta lsquomenrsquo and x voiceless velar continuant as in omogesi
omoxesi harvester agaanto axaanto lsquoa thingrsquo ensagara enzaɸara lsquolizardrsquo
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 lsquomaizersquo oboba ͻͻa lsquomashroomrsquoegesanda eγesanda lsquocalabashrsquo etoigo εtͻixͻ lsquofloodsrsquoemondo emondo lsquogizzardrsquo omoeto ͻmͻεtͻ lsquotraprsquoAdapted 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 tint dʒ labiodental fricatives f v dental
fricatives θ eth alveolar fricatives s z alveolar fricatives int 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-alveolar
Palatal Velar Glottal
Plosive p b t d k gFricative f v θ eth s z ʒ hAffricative t dʒNasal m n ŊLateral lApproximant w r j
Chart (5) The English consonantsAdapted 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 consonantsEnglish consonants EkeGusii consonants p
ɸ
b -f -v -θ -eth -t t d -m mn n- ɳ ŋ ŋk kg -- γ - xj jw - r rl -
62
dʒ -tint tints sz - - 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 eth d
g w dʒ int 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 tint s]
4121 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 phonologyThese sub-section describes
these processes
41211 Voice dissimilation (Dahlrsquos 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 Dahlrsquos 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 ndash 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 different63
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 Dahlrsquos law affects the dorsal stop [k] as (15)
below demonstrates
(15) The effect of Dahlrsquos Law on [k] in EkeGusii |ͻkͻ- kɛsa| ͻkͻ-لاεsa lsquoharvestrsquo |oko-koro| oko-لاoro lsquolegrsquo 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 ndashγɛs 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) Dahlrsquos Law in EkeGusii Word underlying form surface form gloss a) okoroota |ͻ-kͻ-rͻͻt-a| [ͻ-kͻ-rͻͻt-a] lsquodreamrsquo
okogoro |o-ko-kor-o| [o-ko-γor-o] lsquofootrsquookonywa |o-ko-ηw-a| [o-ko-ηw-a] lsquodrink
b) ogokana |o-ko-kana| [o-γo-kana] lsquodenyrsquoogotuua |o-ko-tuua| [o-γo-tuua] lsquobe bluntrsquoogoseka |ͻkͻsεka| [ͻ-γͻsεka] lsquolaughrsquo
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)
41212 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 homorganisationInput ɛ-n- + -γͻkͻ lsquohenrsquoAffixation ɛnγͻkͻ Prenasalisation ɛnγͻkͻNasal 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+| rarr [mb]
while |n+r| rarr [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 (4113) below is
secondary realization which is treated as a single consonant and not a cluster
41213 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| rarr [mbwate] lsquohold mersquo
[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 lsquoholdrsquo
Output [wata]
Hormorganization process
βuata rarr [wata]rarr [βwu-ata]
CG
66
ii) Input sieka lsquoclosersquo
Output sjeka
Hormorganization process
sieka rarr [sjeka]rarr [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] (Katamba1993) 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 (20079)
20) Emai consonant glide hormorganization
(i) ku ame rarr [Kwame] lsquothrough waterrsquo
(ii) fi ͻpia rarr [fjͻpia] lsquothrow cutlassrsquo
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 lsquosiekarsquo 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] lsquofingernailrsquo 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 tintu-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] ldquoI saidrdquo derives from ldquonitbajiterdquo 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 amp Clements 1990)
To avoid treating and calling NC sequences consonant clusters linguists employ the term
ldquoprenasalised stopsrdquo (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 stopsPrenasalised consonant Example of word Gloss
a) |n+b|gt mb engombe [ɛŋͻmbɛ] lsquocowrsquo
b) |n+r| gt nd enda [enda] lsquostomachrsquo
c) |n+t| gtnt egento [eγento] lsquothingrsquo
d) |n+k|gtŋg egechanga [eγetintaŋga]lsquowirersquo
In (22) the NC lsquoclustersrsquo (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 Smolensky1993) 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 lsquopersonrsquo
i) moonto nasal as Coda
72
ii) moonto nasal as Onset
iii)moonto 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
[monto] [mo-nto] [moonto
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
micro micro micro micro micro micro
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 (4113) 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
41214 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 defricativisationi) β rarr[b]
input e-n- + βori lsquogoatrsquoaffixation | e-n-βori|Prenasalisation enβoriNasal homorganisation [embori]
ii) γ rarr [g]input ɛ-n- + γͻri lsquoropersquoaffixation ɛnγͻriPrenasalisation ɛnγͻriNasal homorganisation ɛŋgͻri
74
iii) r rarr [d]input e-n- + raγeraaffixation enraγeraPrenasalisation enraγeraNasal 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 (4113) below Like in
EkeGusii however nasal consonant homorganization characterizes English as illustrated by (31)26) English nasal consonant homorganizationi) ink rarr iŋkii) tomb rarr tumIn (26i) the consonant k homorganizes n to [ŋ] while in (31ii) b disappears as a result of
being hormoganized to [m]
41215 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
(200290)
27) a) nasal syllabification in word initial position
n-to- taatint -ɛ
F-1p-fetch-FV
[ntotaatinte]
lsquoWe will fetch (water) todayrsquo
b) in- mo- taa ts- e
Fndash 2p1-fetch-fv
[mmotaatinte]
lsquoYou will fetch (water) todayrsquo
c) in-a-taatint-e
F-3p-fetch-fv
[mbataatinte]
lsquoThey will fetch water todayrsquo
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- lsquoforgetrsquo
Suffixation -ee
Prefixation n-e-
Nasal resyllabification ne-
Pre-nasal i insertion ine-
Nasal velarization -iŋe-
Output [iŋee] ltingebegt lsquoforget mersquoAdapted 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- lsquobuyrsquo
Suffixation -γore
Prefixation n-γore
Prenasalisation nγore
Nasal homogenization ŋγore
Defricatirization ŋgore
Nasal delinking ŋ-gore
Nasal resyllabification ŋgore
Output [ŋgore]
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 syllabificationAdapted 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 rarr iŋk
ii) ink rarr [iŋk] lsquoinkrsquo
(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)
413 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
R OO
O (C)CV(C)(C) Totonak
X (C)CV(C) Dakota
XO CV(C)(C) Klamath
X CV(C) TemierR X
O _ (C)VC Arabela
X CV Senufo
O O OO (C)(C)V(C)(C) EnglishX (C)(C)V(C) Spanish
X O (C)V(C)(C) Finish
X (C)V(C) Turkish
O XO - (C)(C)(V) PirahaX - (C)V Fijian
Adapted from De Lacy (2006 165)Key
R= Required O= Optional X= BannedX therefore means that
Codas are never requiredOnset clusters are never required
Coda clusters are never requiredOnsets 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 |ketii| [γetii] getii field
a) CVCV βana| [βana] bana predictfore-tell
b) CVCVCV tɛrɛɛra [tɛrɛɛra] tereera sing for
c) VCVCV omote [omote] omote tree
d) VCVCV CV omoγori [omoγori] omogori buyer
e) VCVCVCV aaani [aaani] ababani prophets
Adapted from Bosire and Machogu (2013)
98
(31a) for example can be presented on syllable nodes as in figure (16)
Input ketii output [γetii] lsquofieldrsquo
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 amp 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 lsquotreersquo ii) e- me- te lsquotreesrsquo Aug prefix root Aug prefix root 3sg 33 -tree (sing) 4pl 34 - tree (pl)
b) aaa lsquopluckrsquo (vegetables etc)The prefix omo- in (32ai) above marks the class of the noun lsquotreersquo 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) omote
vc vcv σ σ σ
v c v c v
o m o t e ii) a aa
vv σ σ
v v
a aa Figure (17) EkeGusii syllable nodes for [omote] and [aaa]
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 EnglishWord syllable typecat kᴂt cvc boy bͻɪ cvstructure strᴧktintə cccvccv owe əʊ vAccording 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
eg 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 strim
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)
4131 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 tint dʒ eth 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 ndash syllabic m n h r j w do not occur in
word final positions and terminal sequences such as kf intp ɪ ʒbd are not used in the
language In the following sub sections 41311 and 41312 word initial word final
phoneme sequences and inflection suffix formation constraints of English are discussed
respectively
41311 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 ltegt
ǝ a ɑ are
ͻ ɜ err
eɪ a letter ltagt
aɪ i letter ltigt
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
int+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
41312 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 andɒ 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 liʒ rouge ruʒ 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
414 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
4141 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 oacute m ograve γ aacute agrave k a Old man106
omogaka oacute m ograve γ agrave k agrave aloe vera
(ii) omote oacute m ograve t eacute tree
omote oacute m ograve t egrave 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 lsquoold manrsquo and omogaka
omoγaka lsquoaloe verarsquo 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 41411 describes underlying versus surfaces tones
41412 Contour tones 41413 tone preservation and 41414 tone floating
41411 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 (acute) 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 kacutei m a obokima ugali
108
L L H L
(ii) om o tint acuteo k o racuteo 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
41412 Contour tones in EkeGusiiAccording 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
k כ bεlε mbu mba
H L H L L Hlsquowarrsquo lsquotrousersrsquo lsquoowlrsquo lsquoricersquo
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) realizationemoori e m ō oacute ri calfrsquo chimori c h i m ō oacute r i calves
L H LHomoonto om oacute oacute n t o Person abaanto ab ā aacute n t o Persons
L H L HFigure (22) Ekegusii LH contour toneAdapted 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 ndash koacute oacute m b e hoe amakombe ama ndash k oacute oacute m b e hoes
HH HH
omogaaka omo ndash γ aacute aacute k a old man abagaaka abandashγ aacute aacute k a old men
H H HH
Figure (23) Ekegusii level tone
110
Adapted from Katamba (1993)
41413 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 Odenrsquos (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 lsquothis thingrsquo
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 lsquothingrsquo 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)
41414 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 [sbquo] 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
maelign man pǝmɪt permit
112
strɪkt strict ekspͻt export
wik 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 ndasher or ndashly 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 ndashly
ɒdǝlɪ orderly
maelignɪdʒǝ manager
415 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 Cammengarsquos (2002) pioneering findings for EkeGusii This study
however unlike Cammengarsquos which is not anchored on any theory alludes to tenets of
Optimality Theory in its generalized descriptions
4151 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
4152 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 amp1993) Sesotho 15 (Demuth 2000) Kivonjo 16 (Pinker 1994) Aghem
12 (Aikhenvald 2000) EkeGusii 20 (Cammenga 2002 amp 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 (2002199)
40) Morphosyntactic noun class prefixes in EkeGusii1 omo - 2 aβa-1a mo-1b Ǿ3 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 prefixationa) omonyaroka lsquogirlrsquo abanyaroka lsquogirlsrsquoomo ndash ɳaroka aβa- ɳaroka13PSG- girl 23PPL- girl lsquogirlrsquo lsquogirlsrsquo omo- gaaka aβa- gaaka 13PSG- lsquoold manrsquo 23P PL lsquoold menrsquo lsquoold manrsquo lsquoold menrsquo
b) ekerandi lsquogourdrsquo ebirandi lsquogourdsrsquo eke- randi eβi- randi 73PSG- lsquogourdrsquo 83PPL- lsquogourdrsquo lsquo gourdrsquo lsquogourdsrsquo eke- moni eβi- moni 73PSG- cat 83PPL cat lsquocatrsquo lsquocatsrsquo
Adapted from Ongarora (2006)In (41a) above the noun stem nyaroka lsquogirlrsquo denotes lsquohumanrsquo referent hence co-occur
with singular prefix omo- and a plural one aβa- while that in (41b) refers to an
inanimate referent randi lsquogourdrsquo 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 amp 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 ie God angles devils the spirits of the ancestors and kinship terms (human referents)
1b Oslash- 2aβa kinship terms (human referents)
3 omo- 4 eme- socioculturally relevant objects events or periods trees parts ofthe body (non-human referents)
5 eri-rii- 6 ama- various types of common nouns eg 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 tinti- many names of animals socially or culturally relevant entities (place objects events) some concepts
9a e-n 10a tinti-n same as 9-10
11 oro- 10a tinti-n social cultural and some natural objects12 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 names15 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 ndasha expressing
activities or events16 a- [ase] lsquoplacersquo only no plural21 ɳa- no regular plural proper names of persons individual heads of
cattle and placesSource 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 lsquounlockrsquo
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 amp 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 ndash able leg ndash ible
hear ndash ing audi ndash ence
en ndash large magnndashify
perform ndash ance rend ndash 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
41221 the prefix in 41222 and the noun roots in 41223
41521 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 lsquotreersquo
i) o- mo- te lsquotreersquo
ii) e- me- te lsquotreesrsquo
aug 3SG tree aug 4PL tree
b) eee lsquoyesrsquo
e-ee lsquoyesrsquo Adapted from Bosire and Machogu (2013)
The prefix omo- in (43ai) marks the class of the noun lsquotreersquo that is class three and
number that is singular while the prefix lsquoeme-lsquo in (43a ii) marks class four and plurality
The augment structures in (43) above is represented on syllable node in figure (25)
i) omot e σ
vc vc v σ σ σ
v c v c v
o m o t e
ii) e e e σ σ σ119
vv v v v v e e e Figure (25) EkeGusii prefix and pre-prefix syllable nodesAdapted 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
lsquoomo-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 (oslash-) and 2b (oslash-) 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 (oslash-) and 2b (oslash-)
i) monto mo-nto lsquopersonrsquo banto βa-nto lsquopersonsrsquo
ii) tata tata lsquofatherrsquo batata βatata lsquofathersrsquo
These are described as follows
- mo- nto -βa- nto
1bOslash3PSG person 2bOslash3PPL person
lsquopersonrsquo lsquopersonsrsquo
-tata -βa- tata
1bOslash3PSG father 2b3 PPL fathers
lsquofatherrsquo lsquofathersrsquo
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 lsquotatarsquo 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 oslash- 9 (a) e-(n) 10 (a) tinti- (n) 16 a- and 21ɳa- 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 ndash 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 erindash in nouns of which the stem begins
with a vowel but non-pre-prefixed riindash 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 lsquobananarsquo b) eri-iso lsquoeyersquorii-sosa lsquopumpkin leaves eri-ino lsquotoothrsquorii-raba lsquosoilrsquo eri-ogo lsquomedicinersquorii-mama lsquodumb personrsquo eri-eta lsquonamersquo
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 prefixes1 o-mo-1b Oslash2 a-βa-3 o-mo ndash 4 e-me-5 eri-rii-6 ama-7 eke-8 eβi-9 e-9a e-n10 tinti
122
10 a tinti-n11 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 tinti-
10a tinti- 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 tinti- 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) tinti-n is lowered to a [+high + mid] or upper mid front augment
vowel e
41522 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 classesClass Examples Gloss 1 omo- [omoonto] person
123
[omwaana] child 2 aβa- [aβanto] persons
[aβaana] children 1bOslash- [Oslashβaaβa)] mother
[Oslashmaγokoro] grandmother [Oslashsokoro] 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] gourd8 eβi- [eβirandi] gourds 9 e- [esese] dog
[eusi] thread 10 tinti- [tintisese] dogs
[tintiusi)] threads 9 a e-n-|e-n-βaata| - [embaata] duck
|e-n-raaγera| - [endaaγera] food10 a) tinti-n- |tinti-n-βaata| [tintimbaata] ducks
|tinti-n-raaγera| [tintindaaγera] foods 11 oro- [oroko)] firewood 10 a) tinti-n- [tintiŋko] 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
[oβosaatinta] manhood abstract nounno plural
15 oko- [ͻkͻ βͻͻkͻ] arm [okoγoro] leg
[oγoto] ear [okoruγa] cooking
6 ama- [amaoko] 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 underlinedSource 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 rootbase morphemes
of the identified classes prefixes and pre- prefixes having been accounted for in this and
previous section
41523 EkeGusii noun class roots
It has already been observed in section (4121) 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 (199341) observes that ldquohellipa 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 lexemerdquo For example
lsquotalkrsquo in English is a root with the following word forms talk talk-s talk-ing and talk-ed
As can be seen the form lsquotalkrsquo 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
41524 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 Godgods 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 [aaanto] a - a- anto] -anto persons
aug-3pl- root
omonyenyi [ͻmͻɳɛɳi] ͻ- mͻ- ɳɛɳi ɳeɳi butcher
aug- 3psg- root
abanyenyi [aaɳeɳɳ] a- a- ɳɛɳi ɳɛɳi - butchers
aug- 3pl- root
Adapted from Cammenga (2002)
(49) shows that EkeGusii rootbase 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
41525 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 7eke- as in in eke-rema lsquolame personrsquo and 8ebi- in ebi-rema lsquolame personsrsquo
5ri- as in ri-tiino lsquodumb personrsquo and 6ama- as in ama-tiino lsquodumb personsrsquo
41526 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 glossform form form
esese [esese] e-sese [-sese] dog
etaaro [etaaro] e-taaro) [-taro] journey
ebuunda [eβuunda] e-βuunda [-βuunda] donkey
Class 10 chin-
chisese [tintisese] tinti-sese [-sese] dogs
chitaaro [tintitaaro] tinti-taaro [-taaro] journeys
chibuunda [tintiβuunda] tinti-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-
[tintimbaata] tinti-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
41527 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 7812 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]
lsquoa small personrsquo lsquosmall personsrsquo
b) Pejorative or non-pejorative in increasing degrees
a- ka- mo- onto
aug- 12- 1 - person
[akamoonto]
lsquoa small personrsquo (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]
lsquoa very small personrsquo
c) Pejorative in increasing degrees
e- ke- mo -onto e- βi- mo- nto
aug- 7 - 1 person aug 8- 1 person
[ekemoonto] [eβimoonto]
lsquosmall personrsquo lsquosmall personsrsquo
e- ke- mo- oNto e- ke- ke
aug-7 ndash 1 ndash person aug ndash 7 ndash small
[ekemoonto eγeke]
lsquoa very small personrsquo ndash pejorative
Plural (eβimoonto eβike)
e- ke ndash mo- onto e-ke-ke -mono
aug-7 ndash 1 ndash person a-7 ndash small- very
[ekemoonto eγeke mono]
A very very small personrsquo (pejorative)
Plural [eβimonto eβike mono]
(54) shows examples of EkeGusii pre-prefixation or augmentation
131
54) EkeGusii augmentation
a) non ndash pejorative
o- mo- onto o- mo- nene a- ba- anto- a ndash ba- nene)
aug - 1- person aug-1 - big aug-2 -person ndash aug-2- big
[omoonto omonene] [abaanto abanene]
lsquoa big person lsquobig personsrsquo
(b) non ndash pejorative or more usually pejorative
rii ndash mo-Nto a- ma-mo-Nto
5 ndash 1 ndash person aug-6-1 persons
[riimoonto] [amamoonto]
rii- here means lsquobigrsquo just like ma ndash
c) abusive
rii- ke-mo-Nto a -ma-ke-mo-Nto
5- 7- 1 ndash person aug- 6- 7 ndash 1- person
[riikemoonto] [amakemoonto]
d) pejorative
rii-mo-oNto rii ndash nene a ndash ma- mo-oNto- a- manene
5 1 peson 5 ndash big aug- 6 ndash 1 person aug ndash big
[riimoonto riinene] [amamoonto amanene]
lsquovery big personrsquo lsquovery big personsrsquo
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 rsquobig small
personrsquo 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
42 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 (43) are carried out within the standard Optimality Theory (McCarthy amp Prince 1993
Prince amp Smolensky 19932004) 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 (421) the focus is on the consonants and the vowels phonotactic nativization
focuses on syllable structure in (422) prosodic nativization (423) focuses on tone while
phonological processes nativization (424) focuses on a number of processes
421 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 4211 and consonant
phoneme nativization in 4212 respectively
4211 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 lsquoexcessrsquo
vowels from English are collapsed into the few EkeGusii vowels In other words while
English has twenty- five vowels (Cruntenden 2011 OrsquoConnor 2011 and Roach 1983
among others) EkeGusii has fourteen as has already been observed in this study Sub-
section 42111 analyzes nativization of English pure vowels 42112 with English
diphthongs while 42113 analyzes thriphtongs
42111 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] ( ɪrarr[i])
The English vowel ɪ shares almost similar features with the EkeGusii vowel [i] They are both
[+HIGH -ROUND AND ndashBACK] However while the English ɪ is [+LAX] EkeGusii [i] is
[ndashLAX] which explains why they are acoustically different as illustrated by tables (3) and (4) in
section (411) 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 pronunciationChristmas krɪmǝs ekirisimasi [ekirisimasi]guitar gɪtɑ egiita [egiita]kitchen kɪtintǝn ekicheni [ekitintɛnicabbage kǝbɪdʒ ekabichi [ekaitinti]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 ndash
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] (OrsquoConnor 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 lsquochrismasrsquo is presented in figure (26) Oslashk r ɪ s m ǝ s Oslash English k Oslash r Oslash s Oslashm Oslash 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 amp 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 OrsquoConnor 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 Theoryrsquos 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) gtgt 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
lsquoChristmasrsquo krɪmǝs and ekirisimasi [ekirisimasi] respectively are analyzed in tableaux
(1) and (2) respectively English realizationInput 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 (FEATUREPLACE) V gtgt 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 [krɪsmǝs]
b [kirisimasi]
Tableau (41) English realization of the input krɪsmǝsThe 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) belowInput kirisimasi lsquoChristmasrsquoThis realization re-ranks the constraints as follows LAX (V) gtgt 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 [krɪsmǝs]
b[kirisimasi]
Tableau (42) 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 ɪEnglish word pronunciation Hmong realizationMcKinley mǝkkɪnli [mekiŋli]Mitsubishi mɪtsǝbiinti [miintimbiinti]Adapted from Goldstone and Yang (2001)(56) shows that the English lax vowel ɪ (bold) is realized as White Hmong tense [i] (bold)
Golstonersquos and Yangrsquos 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 aelig into EkeGusii [a] ( aeligrarr[a])
The vowels [a] and [aelig] differ only in one respect while the English aelig is lax EkeGusii [a]
is tensed accoustically They are the same in all other aspects they are [front non-
rounded low]
The English vowel aelig is realized as [a] by EkeGusii speakers as in (57)
57) Nativization of English aelig to EkeGusii [a]English noun pronunciation EkeGusii nativized form pronunciationtaxi taeligksi etagisi [etaγisi]glass glᴂs ekerasi [ekerasi]tank tᴂŋk etanki [etaŋgi] bathroom bᴂethrum ebaturumu [eaturumu] In (57) the English vowel aelig 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 aelig is [ndashTENSE]There are three possible explanations for the occurrence in (57) The first one is phonetic as
explained by Yip (2002) This provides that since aelig and [a] both have a lowered jaw in
their production the speakers find [a] a better perceptual match for the English aelig 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 aelig which is lax is realized as [a] which is tensed
in EkeGusii (see section411) 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 childrenEnglish Grapheme French Realization (Pronunciation) Example of word
ltugt œ butltoogt ltugt 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 aelig as [a] in EkeGusii besides closeness in terms of phonological features
discussed in sub section 411 above (57) above indeed shows that all the noun loans the
vowel aelig is realized as [a] Optimality Theory account of this realization is the same as
that discussed in section 42111 above
Nativization of English ᴧ into EkeGusii [a] ( ᴧrarr[a])
The vowelsᴧ 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 [ndashTENSE] 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 pronunciationcut kᴧt ekati [ekati]brush brʌint eburasi [eurasi]cupboard kʌbǝd ekabati [ekaati] 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 backHigh Mid ʌLow a Chart (6) English ʌ and EkeGusii aChart (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 rarr[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 ndashround])
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 pronunciationskirt skɜt esikati [esikati]shirt intɜt esati [esati]breakfast brekfɜst burekibasiti [urekiasitinurse nɜs omonasi [omonasi]
142
(61) Nativization of English ǝ and to EkeGusii [a]English noun pronunciation EkeGusii nativized form pronunciationfather fɑethǝ omobaata [omoaata] christmas krɪmǝs ekirisimasi [ekirisimasi]pastor pɑstǝ omobasita [omoasita]computer kǝmpjutǝ 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 intɜt for exampleEnglish input intɜt
Input intɜt IDENT IO (F) V CENTRAL (V)
a [intɜt]
b esati
Tableau (43) English realization of the input intɜt 143
EkeGusii input [esati]
Input esati CENTRAL (V) IDENT IO (FEATURE) (V)
a [intɜt]
b [esati]
Tableau (44) 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) gtgt 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 rarr[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 pronunciationcar kɑ ekaa [ekaa]card kɑd ekati [ekaati]glass glɑs ekerasi [kerasi]garage gaeligrɑʒ egarachi [γaratinti]
In (62) the open low back tense English vowel ɑ is realized as EkeGusii [aaa] 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 sEkeGusii input ekerasi
Input ekerasi BACK (V) IDENT IO (F) V
a [glɑs]
b [ekerasi]
Tableau (45) EkeGusii realization of the input ekerasi English input glɑs
Input glɑs IDENT IO (F) V BACK (V)
a [glɑs]
b [ekerasi]
Tableau (46) English realization of the input intɜt
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 pronunciationgarage gaeligrɑdʒ garach [garatint]glass glɑs gilas [gilas]card kɑd 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 rarr[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 pronunciationCotton cɒtn ekotini [ɛkͻtoni]Box bɒks epogisi [ɛpͻγisi]Bolt bɒlt eboriti [ɛͻriti]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 ] ( ɪrarr[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 pronunciationcollege kɒlɪdʒ ekorechi [ɛkͻrɛtintisenate sɪneɪt eseneti [seneti]elephant elɪfǝnt erebanti [ɛrɛanti]
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 41111 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 41112 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 ltegt 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 respectivelyEnglish realizationInput sɪneɪt lsquosenatersquoConstraints ranking IDENT IO (F) V gtgt ASSIM (F)
Input sɪneɪt ASSIM (F) IDENT IO (F) V
a [sɪneɪt]
b [eseneti]
Tableau (47) English realization of the input sɪneɪtEkeGusii realizationInput esenetiThis realization is presented in tableau (8)Constraint ranking IDENT IO (F) gtgtASSIM (F)
Input eseneti IDENT IO (V) ASSIM (F)
a [sɪneɪt]
148
b [eseneti]
Tableau (48) 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 pronunciationmission mɪintn misen [misen]television telɪvɪintn 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]
ʌFigure (27) EkeGusii nativization of the English central vowels SourceBright (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 EkeGusiiEnglish vowel EkeGusii realization
i ɪ [i]ʌaelig ǝ ɑ [e ɛ] e aelig 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 phonemesEnglish vowel Approximate African languages vowel
i ɪ [i]ʌ ǝ ɜ ɑ [a]e aelig [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 ampYang 2001 Owino 2003 Zivenge
2009 Hussain 2011 Kang 2011and Chebanne amp 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 amp 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 amp 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
42112 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] ( ǝʊrarr[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 pronunciationLocation lǝʊkeɪintn erookeseni [ero keseni]Sofa sǝʊfǝ esooba [esoa] Cocoa kɒkǝʊ ekooko [ekoko]Pawpaw pǝʊpǝʊ ripoopo [ripopo]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 amp Phili 2015)
Orthographic influence according to Golstone amp 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 gtgt
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] lsquolocationrsquo for example is presented in tableau
(9)
EkeGusii input ero keseni Constraint ranking COMPLEX V gtgt MAX OI IDENT IO
Input erokeseni COMPLEX V MAX OI IDENT IO (F)
153
a [lǝʊkeɪintn]
b [erokeseni]
Tableau (49) 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 lǝʊkeɪintn locationOutput [lǝʊkeɪintn]Constraint ranking IDENT IO(F) MAX OI gtgt COMPLEX (V)
Input lǝʊkeɪintn IDENT IO(F) MAX OI (F) COMPLEX (V)
a [lǝʊkeɪintn]
b [erokeseni]
Tableau (410) English realization of the input lǝʊkeɪintn
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 ǝʊa) Urdu realizationEnglish noun pronunciation Urdu realization pronunciationHotel hǝʊtǝl hootel [hotǝl]Coach kǝʊtint kooch [kotint]
b) Punjabi realizationEnglish noun pronunciation Urdu realization pronunciationRoad rǝʊd rood [rod]Coach kǝʊtint kooch [kotint]
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 amp Phili (2015)
among others These studies differ with the present one in two crucial ways some employ
different theoretical approaches (Owino 2003 amp Zivenge 2009) Chebanne amp 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] ( aɪrarr[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 pronunciationfile faɪl ebaeri [eaeri]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 198320)
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 lsquotilersquo there is no diphthong in EkeGusii
realization [etaeri] instead the English diphthong is split into two partseach 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 pronunciationFile faɪl fael faelStyle staɪl stael staelMile maɪl mael maeltie taɪl tael taelSource 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 rarr [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 Pronunciationcake keɪk ekeeki [ekɛki]lsquocakersquocase keɪs ekeesi [ekesi]lsquocasersquobasin beɪsn ebeeseni [ɛɛseni]lsquobasinrsquostation steɪintn esiteseeni [esiteseni]lsquostationrsquo
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 rarr [e] and changing the elements of the diphthong altogether and
adding length to the new element as in eɪ rarr [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 eɪEnglish noun pronunciation KiKamba nativized form pronunciationframe freɪm bulemu [ulemu]crane kreɪn keleni [keleni]crate kreɪt keleti [keleti]Adapted from Mutua (2007)
42113 Nativization of English triphthongs
A triphthong is defined as a vowel made up of three short vowels produced as one
(OrsquoConnor 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 4111 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 realized75)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] lsquowhoir kwǝɪǝ ekwaya [ekwaja]ii) Realization of eɪǝ as EkeGusii [aja]English noun pronunciation EkeGusii nativized form pronunciationplayer pleɪǝ epureya [epureja]layer leɪǝ ereya [ereja]iii) Realization of ͻɪǝ as EkeGusii [aja]English noun pronunciation EkeGusii nativized form pronunciationemployer emplͻɪǝ eemburoya [eemburoja]iv) Realization of aʊǝ and ǝʊǝ as EkeGusii [awa]English noun pronunciation EkeGusii nativized form pronunciationflower flaʊǝ eburawa [eurawa]shower intǝʊǝ 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 lsquowirersquo for example illustrates
how triphthongs are handled by EkeGusii OslashOslash w ǝ ɪ ǝ English
OslashOslash OslashOslashOslashOslash EkeGusii OslashOslashw 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 wǝɪǝEnglish noun Pronunciation Urdu Realization Pronunciationwire 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)UrduEnglish noun pronunciation Urdu realization Pronunciationfire faɪǝ faer [faer]wires waɪǝs waeles [waeles]
160
ii) PunjabiEnglish noun pronunciation Punjabi realization pronunciationfire 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)ndash (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 gtgt 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 gtgt COMPLEX Given these rankings analyses of EkeGusii and English
realizations of the English triphthong aɪǝ for example are given in tableaux (11) and (12)
respectivelyEkeGusii realizationEkeGusii input waja
Input waja COMPLEX V MAX OI IDENT IO (F)
a [waɪǝ]
b [waja]
Tableau (411) 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 constraintEnglish realizationInput waɪǝ lsquowirersquoOutput waɪǝ
Input waɪǝ IDENT IO (F) MAX OI COMPLEX V
a [waɪǝ]
b [waja]
Tableau (412) 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 amp Phili2015) 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
Diphthongtriphthong 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
4212 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 eth ʒ int 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 ldquoincreasingly noticeable in the
speech of the younger generation which has had contact with Swahili and Englishrdquo 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 gtgt 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
42121 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 [ndashDENTAL] 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 andv by EkeGusii [β] [ɸ] English noun pronunciation EkeGusii nativized form pronunciationverandah vǝraeligndǝ ebaranda [eβaranda]tv tivi etibii [etiβii]fashion faeligintn ebaasoni [eβaasoni]fridge frɪdʒ eburichi [euritinti]
(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 [euritinti]
Oslashf Oslash r ɪ dʒOslash English OslashOslashOslash r Oslash OslashOslash EkeGusii
Oslash Oslash r i tintOslash Phonemic substitution
e u r i tint 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 [eβuritinti] lsquofridgersquo
Constraint ranking [STRIDENT] (FRIC) gtgt IDENT IO (SEG) C
Input eβuritinti [STRIDENT] (FRIC) IDENT IO (SEG) C
a) frɪdʒ
b) eβuritinti
Tableau (413) EkeGusii realization of the input eβuritinti
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 lsquofridgersquo
Constraint ranking IDENT IO (SEG) C gtgt [STRIDENT] (FRIC)
Input frɪdʒ IDENT IO (SEG) C [STRIDENT] (FRIC)
167
a) frɪdʒ
b) eβuritinti
Tableau (414) 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 [euritinti] 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 observationTonga input [sikava]Constraint ranking in Tonga VTV gtgt IDENT IO (SEG) C
Input skava VTV IDENT IO (SEG) C
a) skafu
b) sikava
Tableau (415) 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 constraintReranking the given constraints will yield English output of the given word since the
languages rank constraints differently
42122 Nativization of English eth and
The consonants [Ө] and [eth] are characterized by the feature values [+interdental] [+
continuant] [+coronal] and [+ anterior] They are distinguished by the feature [voice]
While [Ө] is voiceless [eth] 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 eth as EkeGusii [t] English noun pronunciation EkeGusii nativized form pronunciationthermos Өɜməs etamosi [tamosi]thief θif etiβu [ tiβu]father f ɜethə omoβata [omoβata]The substitution process involved in these realizations is illustrated by figure (30) for the
English noun lsquothermosrsquo Өɜməs OslashӨ ɜ m ə s Oslash English OslashOslash Oslash m Oslash s Oslash EkeGusii Oslasht a m o s Oslash 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 eth 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 ldquothough 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 obstruentrdquo 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 eth 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 rarr [sijori] while eth is realized as [dʒ] as in leather
leethǝ rarr [ledʒa] in Japanese Kay (1996) English Ө is realized as [s] as in thrill Өrɪl rarr
[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 Өɜməs
lsquothermosrsquo
English input Өɜməs lsquothermosrsquoConstraint ranking in English IDENT IO (SEG) C gtgt [INTERD] (FRIC)
Input Өɜməs IDENT IO (SEG) C [INTERD] (FRIC)
a) Өɜməs
b) tamosi
171
c) ethɜməs
Tableau (416) English realization of the input Өɜməs In tableau (416) (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 followsEkeGusii input tamosi lsquothermosrsquoConstraint ranking in EkeGusii [INTERD] (FRIC) gtgt IDENT IO (SEG) C
Input tamosi [INTERD] (FRIC) IDENT IO (SEG) C
a) Өamosi
b) tamosi
c) ethɜməs
Tableau (417) 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
42123 Nativization of English lThe 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 rarr [r]) English noun pronunciation EkeGusii nativized form pronunciationtelevision tεlεviintn eterebisoni [tereβisoni]class klaeligs ekerasi [ekerasi]
172
lunch lʌndʒ ranchi [rantinti]glucose glukǝʊz gurukosi [γurukosi]In (98) the English consonant l is substituted for [r] in EkeGusii The substitution process
involved is illustrated by the English word klaeligss lsquoclassrsquo in figure (31) Oslash k l aelig s Oslash English
Oslash k OslashOslash Oslash s Oslash EkeGusii Oslashk Oslash r a s Oslash 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 pronunciationhustle hʌsl hassuru [hasuru]whistle hwɪsl hoissuru [hoisuru]last laeligst 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 amp 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 amp Yang 2001) KiKamba
(Mutua 2007) and Tonga (Zivenda 2009) among others for example English r is
realized as [l] Thus in Hawaiian merɪ merry rarr [mele] in Tonga rulǝ ruler rarr [lula]
while in KiKamba krim cream rarr [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 and173
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 rEnglish noun pronunciation Kalanga nativized form pronunciation
driver draɪvǝ dirayivara [dirajivara]plastic plaeligstɪk pulasitiki [pulasitiki]plate pleɪt puleyiti [pulejiti]film fɪlm filimu
Adapted from Chebanne amp Phili (2015)
[filimu]
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 gtgt IDENT IO (SEG) C
Input ekerasi LATERAL IDENT IO (SEG) C
a) claeligs
b) ekerasi
c) ekelasi
Tableau (418) EkeGusii realization of the input ekerasi Candidate (b) is the output even when it violates the faithfulness constraint Itrsquos 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
lsquoschoolrsquo showsConstraint ranking in KiKamba TRILL gtgt IDENT IO (SEG) C
Input skulu TRILL IDENT IO (SEG) C
a) sukuru
b) sukulu
Tableau (419) 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
42124 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 pronunciationzero 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 lsquozerorsquo in figure (32) Oslash z i r ǝʊ English OslashOslash i r Oslash EkeGusii Oslashs Oslash 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 fiz rarr viisi [iisi] A similar occurrence is witnessed in Dholuo (Owino 2003) as in
lsquogazettersquo gǝszet rarr [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 lsquozerorsquo for EkeGusii and
English respectively Constraint ranking in EkeGusii VOICEOBS gtgt IDENT IO (SEG)
Input esiro VOICEOBS IDENT IO (SEG) C
a) zirəʊ
b) esiro
Tableau (420) 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) gtgt VOICEOBS
Input zirəʊ IDENT IO (SEG) VOICEOBS
a) esiro
b) zirəʊ
Tableau (421) 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
42125 Nativization of English int and ʒ
These consonants share the feature values [+CORONAL] and [+CONTINUANT] among
others But while int 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 int as EkeGusii [s] English noun pronunciation EkeGusii nativized form pronunciationshow intǝʊ esoo [esoo]shirt intɜt esati [esati]shock intͻk esoki [esͻki]bishop bɪintͻp ebisobu [eisͻu]The substitution process involved in the realizations in these data set is illustrated by figure
(33) for the English word intɜt lsquoesatirsquo Oslash int ɜt Oslash English
OslashOslash Oslash r Oslash EkeGusii Oslashs a t Oslash Phonemic substitution
e s a t i (Nativization Borrowing complete)Figure (33) Substitution of the English int for EkeGusii [s]
(186) indicates that the English consonantal segment int 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] [int]
is [-ANTERIOR] This points to the fact that the feature [anterior] determines the choice of
[s] and not [int] 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 intɜt
lsquoshirtrsquo for EkeGusii and English respectivelyEkeGusii realization
Constraint ranking NON-ANTERCORLFRIC gtgt IDENT IO (SEG)
Input esati NON-ANTERCORLFRIC IDENT IO (SEG)
a) intɜt
b) esati
Tableau (422) EkeGusii realization of the input esati English RealizationConstraint ranking IDENT IO (SEG) gtgt NON-ANTERCORLFRIC
Input intɜt IDENT IO (SEG) NON-ANTERCORLFRIC
a) intɜt
b) esati
Tableau (423) English output and input of intɜtIn 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 int loaned words in other
languages such as in Dholuo as in (87)
87) Nativization of English int in Dholuo English noun pronunciation EkeGusii nativized form pronunciationshirt intɜt sati [sati]fashion faeligintn fason [fason]bishop bɪintͻp pisopu [pisopu]
179
Source Owino (2003)In (87) the English consonant int is realized as [s] in Dholuo Thus like in EkeGusii the
consonant is equally marked in Dholuo
42126 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 pronunciationbag bᴂg epagi [epaγi] glue glu eguru [eγuruu] goal gǝʊl egori [eγorigroup grup egurubu [eγuruu] 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 Dahlrsquos Law of feature dissimilation described
elsewhere in this study which affects EkeGusii is in operation here Figure (34) for the
English word bᴂg lsquobagrsquo is an illustration of this observation Oslash baelig g Oslash English OslashOslash OslashOslashOslash EkeGusii Oslashp a γ Oslash 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 baeligg lsquobagrsquo is analyzed in tableaux (24) and
(24) for EkeGusii and English rankings respectively EkeGusii input epeγi lsquobagrsquoconstraint ranking [ASSIM] (F) gtgt FAITH (F)
Input epaγi [ASSIM] (F) FAITH (F)
a) baeligg
b) epaγi
Tableau (424) 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 EkeGusiiEnglish input baeligg lsquobagrsquoConstraint Ranking FAITH (F) gtgt [ASSIM] (F)
Input baeligg FAITH (F) [ASSIM] (F)
a) baeligg
b) epaγi
Tableau (425) English realization of the input of baeliggCandidate (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 gand z respectively and substitution of the voiceless English stop k
with EkeGusii voiced fricative [γ] This is Dahlrsquos law in operation (Bickmore 1998) This
process is discussed in detail under phonological processes in sub-section 4241
422 Phonotactic nativization
In section 421 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 (199991) 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 consonantvowel 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 413 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 mama mother V CV omote omote tree
b) English syllable structure
Syllable structure word pronunciation
V owe ǝʊ CVC take teɪk CCVCC stains steɪns CCCVC structure strʌktintǝ CVCCC facts fᴂkts CCVCCCC glimpsed glɪmpstThese syllable types can be presented on syllable nodes as figures (35) and (36)
demonstratei) CV as in mama lsquomotherrsquo
σ σ
O N O N
C V C V
m a m a
ii) VCV as in omote lsquotreersquoσ 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 lsquoowersquo
σ
N
V
ǝʊ ii) CCVCC as in steɪns lsquostainsrsquo
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 IO184
(SEG) which is presented as ONSET gtgt DEP-IO Tableau (26) for the EkeGusii input
ma ma [mama] lsquomother lsquoillustrates
Input mama
Constraint ranking ONSET gtgt DEP-IO
Input mama ONSET DEP-IO
a) mama
b) a ama
c) maa
Tableau (426) EkeGusii realization of the input mama
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)
gtgt ONSET (ii) MAX IO (SEG) gtgt NO CODA (iii) MAX- I0 (SEG) gtgt COMPLEX
This is analyzed in tableaux (27-29)
i) Input ǝʊ lsquoowersquo
Constraint ranking in English DEP-I0 gtgt ONSET
Input ǝʊ MAX IO(SEG) ONSET
a) ǝʊ
b) tǝʊ
c) ǝʊǝ
Tableau (427) English realization of the input ǝʊ
185
ii) Input teɪk take
Constraint ranking MAX(seg) -IO gtgtNOCODA
Input teɪk MAX-IO(SEG) NO CODA
a) teɪ
b) teɪk
c) teɪki
Tableau (428) English realization of the input teɪk
iii) Input steɪnz lsquosteinsrsquo
constraint ranking MAX I0 (SEG) gtgt COMPLEX
Input steɪnz MAX-IO(seg) COMPLEX
a) seɪn
b) siteɪnsɪ
c) steɪns
Tableau (429) 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 gtgt
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 4221 below
4221 Nativization by vowel epenthesis
Epenthesis according Katamba (1989) Eichoff (1990) Uffmann (2001 amp 2004) and Rose
amp 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 42211 and 42212 respectively
42211 Nativization of complex syllable margins
In section 4131 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 anaptyxisEnglish noun pronunciation EkeGusii nativized form pronunciation store stͻ sitoo [sitoo]
brush brʌint burasi [βurasi]school skʊl sukuru [sukuru] glass glɑs kerasi [kerasi]
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 lsquostorersquo
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 DEPndashI0 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 ndashI0 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 ndashIO 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
lsquoschoolrsquo for example is analyzed by tableaux (30) and (31) for English and EkeGusii
ranking of the constraints respectively
English input skul lsquoschoolrsquo
English constraint ranking DEP-I0 gtgt COMPLEX
Input skul DEP-IO COMPLEX
a) sukuru
c) skul Tableau (430) English realization of the input skul
EkeGusii input sukuru lsquoschoolrsquo
EkeGusii ranking COMPLEX gtgt DEP IO
189
Input sukuru COMPLEX DEP IO
a) skuru
c) sukuru
Tableau (431) 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 lsquostorersquo 1 [sitoo] 2
brʌint lsquobrushrsquo 1 [βurasi] 3
skul lsquoschoolrsquo 1 [sukuru] 3
glɑs lsquoglassrsquo 1 [kerasi] 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 skul lsquoschoolrsquo
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 uFigure (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 margincluster 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
42212 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 βainitorch tͻ tint tochi tͻͻtinti machine mǝintin masini masi ni suit sut suti suti(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 sut lsquosuitrsquo
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 gtgt CODA
Ekegusii ranking is CODA gtgt DEP-IO The English word torchrsquo tͻtint and its nativized
form in EkeGusii lsquotochirsquo [tͻͻtinti] are analysed by tableaux (31) and (32) respectively
English input tͻtint lsquotorchrsquo
Constraint ranking DEP IO gtgt CODA
Input tͻtint DEP-IO NOCODA
a) tͻtint
b) tͻtinti
Tableau (432) English realization of the input tͻtint
EkeGusii input [tͻͻtinti]
constraint ranking NO CODA gtgt DEP-IO
Input tͻͻtinti NO CODA DEP IO
a) tͻtint
b) tͻͻtinti
Tableau (433) EkeGusii realization for the input tͻͻtinti
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 ndash 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 nagͻl you (sing) will do it
ii) ark-ark argar ikh to growl
iii) k m-n-man-n kamn i man i n 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 gtgt DEP IO (SEG)
Input t-n-ak-ol COPMLEX ONS DEP IO SEG)
194
a) t-n-ak-ol
b) t i nagͻl
Tableau (434) 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-tinthik-ako-aa-i-ro noɳtinthikakotaatiro 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) gtgt DEP-IO (SEG)
Input no-n-koma-i COMPLX (V) DEP IO (SEG)
a) noŋkomati
b) no-n-koma-i
Tableau (435) 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
423 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 4141 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 skul esukuru [e-s u k u r u]
L H L H
blanket blaeligŋkɪt oboranketi [oβo-r a ŋ k e t i]
L L H L H
chief t int if omochibu [o m o-t int 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) YorubaEnglish noun pronunciation (stressed) Yoruba nativized form pronunciation (toned) liberty lɪbǝt libati [liHbati]
tomato tǝmatǝʊ tomato [tomaHto]
guarantee gaeligrǝnti garanti [γarantiH]
b) shonaEnglish noun pronunciation (stressed) Shona nativized form pronunciation (toned) Recipe resǝpi resipi [reHsipi]
Philosophy fǝlɒsǝfi firosofi [firoHsofi]
c) Dholuo
English noun pronunciation (stressed) Dholuo nativized form pronunciation (toned) battery baeligtri betiri [beHtiri]
location lǝʊkeɪintn lokesen [lokeHsen]
degree digri 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 underlying198
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 gtgt MAX IO (T) gtgt IDENT (H) gtgt DEP ndash IO (T) gtgt 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 ooraŋgeti lsquoblanketrsquo from the input blaeligŋkɪt
Input blaeligŋkɪt
output o β o r a ŋ g e t i
L L H L HThis tone structure presupposes the following tone constraints FLOAT gtgt MAX-IO gtgt
IDENT-IO gtgt DEP-IO
Input oo-raŋget-i
H
FLOAT MAX-IO
(T)
IDENT-IO
(H)
DEP-IO SPEC
(T)
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 (436) EkeGusii realization of the input oo-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
424 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
4241 Nativization by voice dissimilation (Dahlrsquos Law)
According to Bickmore (1998) Dahlrsquos 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 Dahlrsquos law are found within his
zones E20 - E50 ndash 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 amp
Nurse 1982) According to Bickmore (1998) there is evidence that Dahlrsquos Law affects
EkeGusii as in (99)99) Voice dissimilation (Dahlrsquos law) process in EkeGusii(a) ͻ-kͻ-rͻͻt-a [ͻkͻrͻͻta] lsquoto dreamrsquo o-ko-kor-o [okoγoro] lsquofootrsquo o-ko-nyu-a [okoɳwa] lsquoto drink (b) o-ko-kana [oγokana] lsquoto denyrsquo o-ko-tuua [oγotuua] lsquo to be bluntrsquo ͻkͻsɛka [ͻγͻsɛka] lsquoto laughrsquoSource Bickmore (1998)
202
In (99a) in the word lsquo[okoγoro]rsquo for example the voiceless velar obstruent (k) in the
prefix oko- demands the voiced velar obstruent (γ) in the root -γor Thus the [ndashvoice]
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 lsquolegrsquo and [oγokana] ogokana lsquoto denyrsquo |okokoro| rarr [okoγoro]
o k o k o r o underlying form
o k o γ o r o surface formFigure (40) EkeGusii dissimilation from |k| rarr [γ] |okokana| rarr [oγokana]
o k o k a n a underlying form
o γ o k a n a surface formFigure (40) EkeGusii dissimilation from |γ| rarr [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 Principle203
(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 lsquolegrsquoConstraints OCP (dorstop) gtgt DEP IO (Seg)
Input okokoro OCP (dorstop) DEP IO (Seg)
a) okokoro
b) okoγoro
Tableau (437) 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
Dahlrsquos law seems to affect English loan words into Ekegusii as (100) below shows100) Dahlrsquos law in English nouns borrowed into EkeGusii(a) EkeGusii t s features dissimilate with [g] featuresEnglish noun pronunciation EkeGusii nativized form pronunciation
bank baeligŋk 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 pronunciationTransport trəensp tכ eturansiboti [e-turanziכt-i]
Transfer traelignsfə 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] lsquosinkrsquo and [e-turanzi-a] lsquotransfer respectively in figures (41) and (42)
Oslashs ɪ ŋ k Oslash English Oslash s Oslashŋ kOslash EkeGusii Oslashs i ŋ g Oslash Phonemic substitution
e s i ŋ g i (Nativization Borrowing complete)Figure (41) Substitution of the English k for EkeGusii [g] Oslash d Oslash r ɪ ŋ kOslash English OslashOslash Oslash r Oslashŋk Oslash EkeGusii Oslash t Oslash 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 ([ts]) 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
words [e-turanziכt-i] lsquoeturansibotirsquo (transport) and [e-turanzi-a] lsquoeturansibarsquo
(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 baeligŋk lsquobankrsquo showInput baeligŋkConstraints IDENT (F) gtgt OCP (voice)
Input baeligŋk IDENT (F) OCP (voice)
a) baeligŋk
b) eŋgi
Tableau (438) English output of the input baeligŋk
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 EnglishEkeGusii input eeŋgi
206
Constraint ranking OCP (voice) gtgt IDENT (F)
Input eeŋgi OCP (feature) IDENT (F)
a) eeŋgi b) ebaŋki
Tableau (439) EkeGusii output of the input eeŋgi
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 (Dahlrsquos 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) demonstrates101) EkeGusii bi-morphemic prefix structure a) o - mo - mur - a b) a - ba - mur- a
augment prefix root FV augment prefix root FV
c) e - ki - ar - a d) e - bi - ar- a
augment prefix root FV augment prefix root FV
The augment and the prefix in (101) mark the class and the number of the noun For
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
11 boy
22
boy
71
one
finger 8 more than 1
finger
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 Glossa) 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 Irsquove seen b) soma soma read dusoma du-soma I read teeka teka cook duteka du-teka I cook seka seka laugh duseka du-seka I laugh kubita kuita hit dukubita du-kuita I hitSource Katamba (1989)(102a) shows that when the prefix is voiced the stem (root) is voiceless the opposite is
true in data (101b) This is Dahlrsquos rule in operation which as illustrated by (100) above
affects English loaned nouns in EkeGusii
Dahlrsquos 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) Dahlrsquos Law in Kitharaka loaned words from English English word pronunciation Kitharaka nativized form pronunciation
UF SFsmall shool skul gasukuru |ka-sukuru| [γa-sukuru]to freeze friz gofirisi |ko-firisi| [γo-firisi]to sheet intit goshiti |ko-intiti| [γo-intiti]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 Dahlrsquos Law in operation
4242 Nativization by defricativization spirantization (fricativization) and bilabialization
Defricativization as was pointed out in sub-section 41214 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 42421 defricativization 42422 fricativizationspirantization and
42423 bilabialization
42421 nativization by defricativization
As pointed out in subsection 41214 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 taeligŋk etanki |etan-γi|rarr[etaŋ-gi]drink drɪŋk eturunki |eturun-γi|rarr[eturuŋ-gi]b) Realization of as [b] English noun pronunciation EkeGusii nativized form pronunciation
UF SF
camp caeligmp ekambi |ekan-i|rarr[ekam-bi]
remand rɪmaelignd erumande |eruman-ri|rarr[eruman-
di]
c) Realization of r as [d]UF SF
secondary sekǝndri esekenderi |ɛsɛkɛn-ri| rarr [ɛsɛkɛn-dɛri]conductor kɒndʌktǝ ekondagita |ɛk כn-raγita| rarr[ ɛk כn-daγita]
The substitution processes in (104) are further illustrated by figures (43) for the EkeGusii
γ rarr [g] (44) for EkeGusii rarr [b] and (45) for EkeGusii r rarr [d]
English word tank taeligŋk
Nativized from [etaŋgi] lsquoetankirsquo
e t a n γ i EkeGusii underlying (nativized) form
e t a ŋ g i EkeGusii surface realizationFigure (43) Realization of EkeGusii γ as [g]
English word camp kaeligmp
Nativized from [ekambi] lsquoekembirsquo
e k a n i EkeGusii underlying (nativized) form
e k am b i EkeGusii surface realizationFigure (44) Realization of EkeGusii as [b]
English word secondary sekǝndri
210
Nativized from [esekenderi] lsquoesekenderirsquo
ɛ s ɛ k ɛ n r OslashOslashi EkeGusii underlying (nativized) form
ɛ s ɛ k ɛ n d ɛ r i EkeGusii surface realizationFigure (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 taeligŋk
lsquotankrsquo sɪŋk lsquosinkrsquo and drɪŋk lsquodrinkrsquo105) Output mapping English nouns in EkeGusii by defricativization (|γ|rarr[g])
i) |e-tanγ-i| rarr [etaŋgi]ii) |e-sinγ-i|rarr [esiriŋgi ]iii) |e-turunγ-i|rarr [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
(41123) 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 (||rarr [b])
i) |e-kaan-i| rarr [ekaambi]
|ɛ-kͻͻniut-al| rarr [ɛkͻmbjuta]
ii) |r| rarr [d]
|e-rumanr-e| rarr [erumande]
211
|ɛsɛkͻnrari| rarr [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 rarr [] before being defricativized in the surface form as
shown in (107)
107) Defricativization of the English noun in EkeGusii after fricativization
camp rarr |kan -- i|rarr [kambi] lsquocamprsquo
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| gt kambi
Output [kambi]
Constraint ranking COMPLEX C gtgt IDENT IO (F)
Input kani COMPLEX C IDENT IO (F)
a kambi
b kani
c kanbi
212
Tableau (440) EkeGusii output of the input kani
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 defricativizationWord UF SF Glosskamba |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
42422 Spirantizationfricativization 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 Dahlrsquos
Law of dissimilation discussed in (4241) 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| tiu
pastor pʌstǝ basita |pasita| asita
bolt bͻlt boriti |pͻriti| ͻriti
In these data for example in the English word dɪp lsquodiprsquo which is realized as [tiu] lsquotibu
in EkeGusiirsquo 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 Dahlrsquos 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] lsquotaxirsquo 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 lsquotaxirsquo
Input eteγisiConstraint ranking OCP (voice) gtgt IDENT (F)
Input eteγisi OCP (voice) IDENT (F)
a) eteγisi
b) tᴂksi
Tableau (441) 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) Dahlrsquos 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
42423 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 labiodentalsEnglish noun pronunciation EkeGusii nativized form pronunciation
veranda vǝraeligndǝ baranda -ɸaranda
university junɪvǝsɪtɪ 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 lsquoverandarsquo in (111) is analyzed
in tableau (42)
EkeGusii input ɸaranda lsquobarandarsquo
Constraint ranking OCP(voice) gtgt IDENT IO (F)
Input ɸaranda OCP(voice) IDENT IO (F)
217
a ɸaranda
b vǝrᴂndǝ
Tableau (442) 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
4243 Nativization of nasal consonant clusters (NC) by pre-nasalization andhomorganization
In sub- section 41213 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 41131) 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 (4221) 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 staeligmp 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 lsquodrinkrsquo and ebaranda lsquoverandarsquo 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 consonantal219
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
vǝrᴂndǝ [βaranda]
NC NC
lsquoa consonant clusterrsquo lsquoa unitrsquo (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)
Input ɸaranda (unit)
Constraint ranking COMPLEX gtgt FAITH C
Input ɸaranda (unit) COMPLEX C IDENT IO (labiodental)
a ɸaranda (unit)
b vǝrᴂndǝ (cluster)
Tableau (443) EkeGusii output for the input ɸaranda (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
4244 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 disharmonya) Nativization by vowel height harmonyEnglish noun pronunciation EkeGusii nativized form pronunciation
coat kǝʊt egoti [eγoti]
basin baeligsɪn ebeseni [ɛɛsɛni]
lorry lɒri erori [erori]
cheque tintek echeki [tintɛki]
222
b) vowel height disharmony
English noun pronunciation EkeGusii nativized form pronunciation
tractor trʌktǝ ekeragita [ekeraγita]
feet fɪt ebuti [euti]
break breɪk ebureki [eureki]
(115a) shows vowel height harmony while (115b) vowel height disharmony In [ɛtͻtinti]
lsquotorchrsquo for example the prefix vowel [ɛ] is in height harmony with the root vowel [ͻ]
They are both mid vowels In [ekeraγita] lsquotractorrsquo 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 [ɛntɛrɛa] in tableau (44)
EkeGusii input [ɛntɛrɛa]
Constraint ranking HARMONY (V) gtgt FAITH V
Input ɛntɛrɛa HARMONY (V) FAITH V
a ɛntɛrɛa
b draɪvǝ
Tableau (444) EkeGusii output for the input ɛntɛrɛa
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 harmonyRoot word gloss suffixed form gloss vowelgid- stomach gidi her stomach ici(r)- head ciri her head igud- behind gudi her behind ukas- hand kasa her hand aAdapted from Pearce (2003)In (116) the height of the vowel in the root spreads to the suffix In kas- for example the
low vowel a spreads to the suffix to form kasa in the suffix form This indeed is in
support of EkeGusii occurrences observed in (115a)
43 Morphological nativization
Section 42 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 (43) 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) ndash which demands that there should be no morpheme deletion an input
morpheme should be maintained in the output
MAX- OI (MORPH) ndash 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) ndash 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 (199394) amp McCarthy (2006)
225
431 Nativization by nominal classification
As observed in 4121 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] lsquohenrsquo singular 9a
tinti - nγͻͻkͻ [tintiŋgͻͻkͻ] lsquohensrsquo plural10a
Adapted from Cammenga (2002)
The noun in (134) [nγͻͻkͻ] lsquohenrsquo 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 tinti-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 lsquohenrsquo
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) gtgt (ALIGN (AFX R RT L))
Input nkooko DEP-IO (MORPH) (ALIGN(AFX R RT L))
a nkooko
b e-nkooko
Tableau (445) 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 lsquohenrsquo or lsquohensrsquo in constructions such as
nkooko teiyo [ŋgͻͻkͻ teijo]-lsquohen not therersquo lsquothere is no hen(s)rsquo
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| gt [en-γͻͻkͻ] gt
[eŋgͻͻkͻ] is as follows
Input nkooko nγͻͻkͻ
227
Constraints and their ranking (ALIGN(AFX R RT L)) gtgt DEP IO (MORPH)
Input nkooko (ALIGN(AFX R RT L)) DEP IO (MORPH)
a e-nkooko
b nk-ooko
Tableau (446) 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 lsquoboyrsquo for example can
be analyzed in OT theoretic terms as in tableau (47) and (48) as follows
Input [bͻɪ] ndash singular form
The presupposed constraints here are ranked as follows (ALIGN (AFX L RT R)) gtgt
DEP IO (MORPH) ranked as DEP IO (MORPH) gtgt (ALIGN(AFX L RT R))
Input boy DEP IO (MORPH) (ALIGN(AFX L RT R))
228
a boy
b boy-s
Tableau (447) 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)) gtgt DEP IO (MORPH)
Input boy-s (ALIGN(AFX L RT R)) DEP IO (MORPH)
a boy-s
b boy
Tableau (448) 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 classesgroups 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] 1omo-sikaoti 2aβa-sikaoti
record records [rɛkͻti] 9e-rɛkͻti 10 tinti-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 lsquoblanketrsquo 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 lsquoblanketsrsquo
nativized as lsquoama- ranketirsquo is analyzed in tableaux (49) and (50) for English and EkeGusii
outputs respectively
English output blanket-s
Input blanket-s
Constraint ranking STRPRES gtgt (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 (449) 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)) gtgt 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 (450) 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 Dahlrsquos Law of dissimilation
Languages without Meinhoffrsquos 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 ([aelig]) 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 (4122) 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 (stembase) semantics determines prefixes choice For
example the noun -mura lsquoboyrsquo or lsquomale youthrsquo 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 rootbase lsquoscoutrsquo 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
4311 The role of semantics in morphological nativization of English nouns inEnglish
It has been observed that a noun in EkeGusii and indeed in most Bantus languages with
noun class system (Henderiks amp Poulos 1990 Givon 1972) enters into a specific class
depending on its stem gender number and size (section 4122) 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 classesprefixes 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 lsquotractorrsquo on the other
hand falls within the semantic features [-ANIMATE -HUMAN BEING +OBJECT] and
therefore enters its appropriate class- 7eke- 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
RTR)) (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 amp 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