. ...:..;..;
v .3 "a“:
«an; {'33

r

1 FM.

.M LA
:T

957 4.

2.3.7 ..(.

_ 3: 3:7: .

if
w >:.~"?i

 

 

 

 

 

M)

‘ V
5 1...:

,. 2} I:

F. $5.
43$

11v

‘ In
(‘4.

.Kac.

.
anti ‘3? 7L

 

5..

1 $0.?

53%»..6‘“;

THESIS

/
4...

__~’"’(‘( 1-(

This is to certify that the

thesis entitled

Vowel Systems of African Englishes:
and Perceptual Analysis

Acoustic
presented by

Mungai Mutonya

has been accepted towards fulfillment
of the requirements for

Ph.D. Linguistics

 

degree in

 

 

 

 

, L/ ,
0- Major professor

Date June: is, 2000

MS U is an Affirmative Action/Equal Opportunity Institution

 

LIBRARY
Michigan State
University

 

 

 

PLACE lN REIURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

‘ DATE DUE

MAY 2 32002

   
 
 

 

 

 

 

 

 

 

| DATE DUE DATE DUE

4 ocrobzum
Dawn

i‘[‘-‘ i} 8 EOCS

1004

 

  
    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

JAN 8
0629 06

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11m C‘JCIRC/Datth-pas-pjd

#_fl/

VOWEL SYSTEMS OF AFRICAN ENGLISHES: ACOUSTIC AND
PERCEPTUAL ANALYSIS

By

Mungai Mutonya

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY

Department of Linguistics, Germanic, Slavic, Asian and African Languages

2000

ABSTRACT

VOWELS SYSTEMS OF AFRICAN ENGLISHES: ACOUSTIC AND
PERCEPTUAL ANALYSIS

MUNGAIIIBVIIUTONYA

This study analyzes production and perception of African English (AfrE) vowels
by university students in Ghana, Kenya and Zimbabwe. Acoustic and perception tests are
carried out and subsequently correlated in order to determine the characteristic of vowel
systems in each sample. In order to determine the acoustic qualities of vowels, data was
initially elicited from language groups with relatively homogeneous social and linguistic
background. F1 and F2 scores of onset (steady state) vowels contained in eleven English
monosyllabic were calculated and subsequently plotted using computer software;
Signalyze (version 3.12) and Plotnik (version04) respectively. To account for vowel
perception, data was elicited from AfrE listeners using the minimal pair and identification
test methods.

Both acoustic and perception results challenge previously held assumptions
regarding vowel variance in west, east and southern Africa. Ghanaian respondents in this

study did not back RP central vowels [3] and [A] to [D ], a position so strongly held

in earlier studies. Moreover, vowel system identity among ‘Bantu English’ speakers in

eastern and southern Africa is not corroborated by Kenyan and Zimbabwean respondents

in this study. Gender distinction within and across varieties is also identified.

Copyright by
MUNGAI MUTONYA
2000

DEDICATION

I dedicate this work to my wife Laurah, my sons Mwariri and
Mutonya, our parents, brothers and sisters.
Could not have made it without your love and support.

ACKNOWLEDGEMENTS

I am greatly indebted to many people who made substantive contributions,
directly or indirectly, to this research project. Their inspiration, support and guidance has
influenced the completion of this study. Needless to say, there have been numerous tough
hurdles to clear, many sacrifices to make, short and long term goals to meet, numerous
responsibilities to bear, and of course, a dissertation to write. Such challenges would have
been insurmountable without the valuable insights, guidance, love and support from my
professors, colleagues, friends and family members. To them all, I extend my grateful
acknowledgements.

I have been lucky to work with a great teacher and sociolinguist, Dr. Dennis R.
Preston, my mentor and advisor. From the formative stages of this study, through data
collection, data analysis, to the completion of this research, Dennis has always been
available to answer any questions, address any research concerns, and to read the drafts
whenever I had one ready for him. 1 thanks him most sincerely.

Other members of my dissertation committee have equally informed the study in
numerous ways: Dr. David Dwyer has enriched the study through his expertise on
African linguistics and through his very constructive suggestions; Dr. Yen-Hwei Lin

provided valuable insights on phonology and language variation; Dr. Denise Troutman

vi

painstakingly read my draft and gave very thoughtful and valuable insights. I am equally
grateful to Dr. Geneva Smitherrnan, who agreed to participate in my dissertation defense
as an outside reader. Her comments and suggestions helped ameliorate this final draft. I
thank them all for their wise guidance.

Dr. Harry Akussa of University of Accra, Ghana; Dr. Patrick Kambewa of
Malawi; Dr. Abraham Gitau Ndungu of Kenyatta University, Kenya; Maina Mutonya,
currently a graduate student at the University of Witwatersrand, South Africa; and
Mwangi Mutonya, assisted me with data collection. Without their assistance and the
cooperation of all the respondents they interviewed, this research would not have taken
its current form. I thank them most sincerely.

The vicissitudes and sacrifices encountered through Graduate school could not
have been easily borne without the incredible support I received from my beloved family.
My greatest appreciation and gratitude goes to my loving wife Laurah for the love,
patience, support and friendship through the years; my precious boys, Mwariri and
Mutonya for giving me a cherished learning, and the motivation to accomplish this task;
my loving parents Jeremiah Mutonya and Rebecca Njoki for their inspiration and a firm
educational foundation; the Mutonya family - Njuguna, N joroge, Mari, Wangui, Wanjiru,
Kamau, Muchiri, Maina, Mwangi, Mukami, Machira, and all the youngsters for their
support, and for daring to dream. Thanks to my in-laws for the support, especially my late
father—in-law, Jotham Mwariri Gichura, who supported and encouraged me. May his soul
rest in peace. Thanks to Beatrice Wambui Cege for sharing her life with us.

I thank Michigan State University and in particular the chair of Department of

Linguistics, Dr. George Peters, for the Teaching Assistant Fellowship that financed my

vii

graduate study. Laurie Koehler was more than a Graduate Secretary to us; she was a
friend, a dear friend with a warm smile, an attentive ear, and a loving heart that radiated
warmth and sunshine to our department. I am grateful to Saisunnee Visanyoongoon and
Li Qing, for sharing the trials and rigors of comprehensive exams. I am thankful to fellow
sociolinguists who shared in the joy of Preston’s classes, the companionships in the
language lab, and the inevitable sleepless nights during comps and dissertation writing.
Thanks to Cege, Kuria, Smucker, Wangui, Walter, and Jamila for the solidarity, and to all
my Kiswahili students for making it memorable.

The faculty and staff of the African and Afro-American Studies program at
Washington University in Saint Louis provided me with the necessary support and
conducive environment that expedited the completion of my dissertation work. I am
especially grateful to Professors Gerald Early, Rafia Zafar, Priscilla Stone, Tim Parsons
and Richard Kisiara for their continued support. Adele Tuchler and Raye Riggins have
made my life at Wash U and AFAS in particular more exciting. Thank you all.

I am very grateful to the University of Nairobi, Kenya and the Department of
Linguistics and African Languages for granting me study leave to pursue my doctoral
studies at Michigan State University.

I am grateful to Mzalendo Kibunja and Carolyn Harford for facilitating the opportunity.

To all my teachers, classmates, friends and relatives who have positively

impacted my educational pursuits, I say, ASANTE SANA. THAAAAIIII.

viii

TABLE OF CONTENTS

List of Tables ..................................................................... xiii

List of Figures ............................................................... xiv

Chapter 1: Introduction

1. 0. An Overview of the Study ................................................... l
l. 1 Statement of the Problem ................................................... . 3
1. 2 Background to the study ................................................... 5
1. 3 Theoretical Framework ................................................... 7
l. 4 Scope and Limitations ................................................... 11
1. 5 Data Collection ............................................... 13
l. 6. Data Analysis ................................................... ' 16

Chapter 2: Background to the Study

2.1. 0 Introduction .................................................................. 16
2. 2. 0 Background to the study ................................................... l6
2. 2. 1 African English? .......................................................... 16
2. 2. 2 AfrE’s Regional Vowel Systems .......................................... 19
2. 2. 2. 1 West African English (WAfrE) .......................................... 23
2. 2. 2. 2 East African English (EAfrE) .......................................... 29

ix

2. 2. 2. 3 Southern African English (SAfrE) ............................... 31

Chapter 32Theoretical Framework

3. l. 0 Introduction ..................................................................... 33
3. 2. 0 Theoretical Framework ...................................................... 33
3. 2.1 Some General Theoretical Assumptions .................................... 34
3. 3. 0 Theoretical Assumptions on Non-Native Speech Perception ........... 39
3. 4. 0 The Native Language Magnet Theory .................................... 42
3. 5. 0 Acoustic Theory of Speech Production ................................... 44
3.5. l Vowels ........................................................................ 45
3. 6. 0 Assumptions About Phonemic Identity ................................... 46
3. 6. I Introduction ................................................................... 46
3. 6. 2 On Defining a Phoneme: A Brief Historical Overview ............... 46
3. 6. 2. l Phonetic and Phonemic identity ......................................... 47

Chapter 4: Methodology

4. 1 Introduction .................................................................... 53
4. 2 Collection of Acoustic Data ................................................ 53
4. 2. l The Production- Sample ................................................... 53
4. 2. 2 Data Collection ............................................................. 54
4. 3. 0 Data Analysis ................................................................ 54
4. 3. 1 Acoustic Measurement ...................................................... 54
4. 4. 0 Vowel Perception Data ...................................................... 56

4. 4. 1 Data Collection Methods .................................................

4. 4. 2 The Perception-Sample ......................................................

Chapter 5: Data Analysis

5. l. 0 Data Presentation and Discussion ...........................................
5. l. l Schmied's Analysis of AfrE Vowel Systems ..............................
5. 2. 0 An Empirical Analysis of AfrE vowel systems ...........................
5.2. l Kenyan speakers ...............................................................
5. 2 .2 Male respondents ..............................................................
5. 2. 3 Female Respondents ..........................................................
5. 3. O Zimbabwean Speakers .......................................................
5. 3. 1 Male Respondents ............................................................
5. 3. 2 Female Respondents .........................................................
5. 4. 0 Ghanaian Speakers ............................................................
5. 4. 1 Male Respondents ............................................................
5. 4. 2 Female Respondents .................................................... ’ .....
5. 5. O A Comparative Analysis of Acoustic Results ...........................
5. 6. 1 Kenyans .....................................................................
5. 6. 2 Ghanaians ....................................................................
5. 6. 3 Zimbabweans ................................................................
5. 7. O AfrE Listeners’ Vowel Perception ..........................................
5.7. 1 Data Analysis ..................................................................

5. 8. 0 Identification Results and Discussion ......................................

xi

56

57

59

6O

64

64

66

69

72

74

76

80

8]

83

86

88

88

89

9O

9O

92

5. 9. 0 Results of the MPT ............................................................ 108

Chapter 6: Conclusions

6. 1 Introduction .................................................................... 112
6. 2 Correlating Acoustic and Perceptual Results ................................. l 13
6. 3 Theoretical Assumptions Revisited ............................................ l 18
Appendix A’ ............................................................................. 125
Appendix A ............................................................................. 126
Appendix B ............................................................................. 129
Appendix C ............................................................................. 141
Appendix D ............................................................................. 143
Appendix E ............................................................................. 148
Appendix F ............................................................................. 156
References ............................................................................. 159

xii

LIST OF TABLES

Table 1 — Identification Results of Vowel [i] .............................. 91

xiii

LIST OF FIGURES

Figure 1 - Adapted Representation of Schmied's Classic Comparison of RP,
WAfrE, and EAfrE ...........................................................

Figure 2 — WAfrE Vowel System as Presented by Trudgill and Hannah .......
Figure 3 - Todd’s Representation of the Vowel System of Cameroon English

Figure 4 — Sey’s Comparison of RP and Ghanaian Vowel Systems ............
Figure 5 — A Comparison of RP, Nigerian and Cameroonian Vowel systems
Figure 6 — The Realization of RP Schwa in Nigerian and Cameroonian Englishes
Figure 7 — A Representation of the RP Vowel system as Illustrated in Figure 1
Figure 8 — A Representation of WAfrE as Illustrated in Figure 1 .................
Figure 9 — A Representation of EAfrE as Illustrated in Figure l ..................
Figure 10 — Vowel Plot ofAll Kenyan Respondents........................ .......
Figure 10.1 - Njenga’s Vowel System ( a Kenyan Male) .......................
Figure 10.2 - Njogu’s Vowel system ( a Kenyan Male) ...........................
Figure 10.3 - Mukami’s Vowel System ( a Kenyan female) ......................
Figure 10.4 - Kamba’s Vowel System ( a Kenyan Female) .......................
Figure 11 - Vowel Plot of All Zimbabwean Respondents ........................
Figure 1 1.1 - Jones’ Vowel System (a Zimbabwean Male) .......................

Figure 11.2 - Ndanga’s Vowel System (a Zimbabwean Male) ..................

xiv

26

28

29

62

63

65

67

68

70

71

73

74

76

Figure 11.3 - Kapinda’s Vowel System ( a Zimbabwean Female) ............... 77

Figure 11.4 - Tamika’s Vowel System ( a Zimbabwean Female) ................. 78
Figure 12 - Vowel Plot of All Ghanaian Respondents .............................. 80
Figure 12.1- Kwame’s Vowel System ( a Ghanaian Male) ........................ 81
Figure 12.2 - Kojo’s Vowel System ( a Ghanaian Male) .......................... 82
Figure 12.3 - Akosua’s Vowel System ( a Ghanaian Female) ..................... 84
Figure 12.4 - Efua’s Vowel System ( a Ghanaian Female) ......................... 85
Figure 13.1— AfrE Listeners’ Identification of [i], [u], [e] ........................... 93
Figure 13.2- AfrE Listeners’ Identification of [i], [u], [e] ........................... 94
Figure 14— AfrE Listeners’ Identification of [ie], [0i][ua], [ju] ..................... 94
Figure 15.1- AfrE Listeners’ Identification of [a] .................................... 95
Figure 15.2- AfrE Listeners’ Identification of [a] ................................... 96
Figure 16.1- AfrE Listeners’ Identification Of[A] ........ 97
Figure 16.2- AfrE Listeners’ Identification of [A] .................................... 97
Figure 17.1- AfrE Listeners’ Identification of [3] ................................... 99
Figure 17.2- AfrE Listeners’ Identification of [3] ................................... 99
Figure 18.1- AfrE Listeners’ Identification of [I] ................................... 101
Figure 18.2- AfrE Listeners’ Identification of [I] ................................... 101
Figure 19.1- AfrE Listeners’ Identification of [D] ................................... 102

XV

Figure 19.2 - AfrE Listeners’ Identification of [D] ................................. 103

Figure 20.1- AfrE Listeners’ Identification of [e] .................................... 104
Figure 20.2- AfrE Listeners’ Identification of [8] ..................................... 105
Figure 21- AfrE Listeners’ Identification of [ai] ..................................... 105
Figure 22 - AfrE Listeners’ Identification of [au] ..................................... 106
Figure 23- AfrE Listeners’ Identification of [ou] ...................................... 106
Figure 24 — Ghanaians Discrimination of Minimal Pairs .............................. 109
Figure 25 — Kenyans Discrimination of Minimal Pairs ................................. 110
Figure 26 — Zimbabweans Discrimination of Minimal Pairs ........................... 111

xvi

Chapter One: Introduction

Introduction

In an earlier study, Mutonya (1997) observed that university students in three
African countries positively identified regional varieties of African Engish (AfrE), and
they assigned contrasting attitudinal values towards local and non-local varieties based on
degree of accentedness. The scope of the study, however, did not allow for the
measurement of physical and nonphysical qualities of the sounds that respondents reacted
to in their judgments. Consequently, the current study focuses on the measurement of
acoustic and perceptual characteristics of AfrE vowel systems.

In general, the current study further investigates the nature of documented
varieties of AfrE; West African English (WAfrE), East African English (EAfrE), and
Southern African English (SAfrE). Specifically, the study seeks to determine what the
vowel system of each regional variety looks like based on acoustic and perceptual
analysis of data elicited from university students from Ghana, Kenya and Zimbabwe.

However, unlike previous research in this field, the current study has paid
considerable attention to empirical methods of sampling, data collection, acoustic and
perceptual analysis, in an effort to tease out the most probable vowel systems in each

region.

1. 0 An Overview of the Study
The present study investigates the production and perception of English (L2)
vowels by university students in three African countries. While holding L1 dialectal

factors as homogeneous as possible, F1 and F2 scores of African English (AfrE) vowels

produced by multilingual speakers from Ghana, Zimbabwe and Kenya were measured
and analyzed. Furthermore, AfrE listeners' perception of Received Pronunciation (RP)
vowels were analyzed and subsequently correlated with acoustic results. The probable
vowel system for each group was determined by correlating the more physical attributes
of the vowels (production) with the more abstract ones (perception). Consequently,
conclusions drawn in this study challenges some widely held assumptions about AfrE,
particularly, in regard to the nature of AfrE vowel systems.

Although the primary objective of this study is to define and interpret the phonetic
and phonemic identities of AfrE vowel systems, defining a methodological framework
for future studies is an equally significant goal. This study accentuates the need for more
empirical investigation in the study of English (as well as other languages) in the densely
multilingual sub-Saharan Africa. The lack of information about the techniques and
instruments used in the study of AfrE vowels has been a major shortcoming of previous
studies. The lack of empirical evidence pertaining to Lls and L25 in this region, has,
needless to say, created impediments and challenges to this study.

In order to provide an empirical investigation of AfrE vowel systems, this study
has adapted tools of investigation and analysis applied in fields of sociolinguistics,
acoustic phonetics, and the social psychology of language.

Towards this end, sociolinguistic methods of sampling and data collection have
been applied in identifying a relatively homogeneous sample. Moreover, the physical
characteristics of vowels realized by AfrE speakers of each regional variety have been
quantified using tools adapted from acoustic phonetics. Likewise, vowel perception

methods, primarily used in field of the social psychology of language, have been

utilized in quantifying vowel perception. Collectively, the theories and methods derived
from these fields of study have informed this study immensely.

Chapter one defines the parameters of this study. Chapter two provides a
background to the study while chapter three delineates the guiding theoretical framework.
Chapter four is a discussion on methodology while in chapter five, an analysis of

acoustic and perception data is discussed. Conclusions are drawn thereafter in chapter six.

1. 1 Statement of the Problem

The existing literature characterizes AfrE vowel systems as homogeneous
varieties with regionally distinct qualities. The widely documented comparison of RP and
AfrE vowel systems (see Figure 1) suggests a seven-vowel system for West African
English (WAfrE), and a five-vowel system for East African English (EAfrE). Hardly any
empirical evidence is provided to account for the vowel system variation.

Research in AfrE vowel systems is not only limited in scope and methodology but
also replete with conflicting generalizations and conclusions. The literature exhibits
significant flaws in methodology such that reported findings and conclusions require
verification. Taking advantage of available tools of investigation and analysis, this study
seeks to accomplish that goal.

Adegbija (1994: 53) outlines existing shortcomings in current research:

Deficiency in methodology appears to be the principal weakness of most language
...[variation] studies hitherto carried out in sub-Saharan Africa. Some of such studies...
appear to be largely impressionistic in approach and sampling techniques and
instruments, when indicated are often weak.

Adegbija's observation echoes Schmied's (1989) earlier challenge to African
English researchers to employ new approaches in data collection and analysis, in order to
ensure that studies of English in Africa provide a systematic empirical account.

In the study of variant vowel systems of African English, many researchers have
failed to acknowledge the dense multilingual complexities of sub-Saharan Africa. The
assumed homogeneity of speech communities has led to overgeneralizations that seem to
outlook the linguistic complexities prevailing in many sub—Saharan countries where tens
or hundreds of diverse indigenous languages are spoken, and English is acquired as a
second or third language by a great majority of its speakers. Moreover, the use and
function of English in these countries, is limited to official settings.

In such dynamic language contact situations, complex patterns of influences and
transfer of features from one language to another may be easy to find. However,
although the exposition of such patterns and transfers are not within the purview of the
current study, we should note that the realization of English vowels in such dense
multilingual setting may depend on speaker’s L1, level of education, social network,
social motivation among other factors. These factors, especially ethnicity, may vary
considering that speakers in a given country may speak related languages or distinctively
different ones. Previous studies seem to have overlooked such factors as they generalized
regional variations of AfrE. Angogo and Hancock (1984: 74-5) adequately represent this

common misperception in the field:

Phonology is perhaps the most distinctive characteristic of [AfrE]. .. it is quite

possible to generalize about the existence of a ‘West African English’ and ‘East African
English’ despite their respective geographical variant forms, in the same way that we
speak of an American English’ which in fact consists of many regional and social
varieties. Even though a Gambian will say “bot” for “but” when a Ghanaian will say
“bet”, it does not mean that there are many [WAfrE’s] rather than one; there is a
phonological cohesiveness to all these kinds of English which identify them as African

rather than, say, West Indian English, and specifically as West African rather than East
African.
Considering WAfrE as a homogenous language variety overlooks even the

widespread pidgin varieties that many scholars argue may constitute a major input in the
formation of Englishes in that region. A similar generalization is echoed by Schmied
(1991), who considers the Eastern and Southern African English varieties to be so
indistinct as to constitute a common variety referred to as Bantu English.

Most studies identify three broad regional varieties of African English: West
African English (WAfrE), East African English (EAfi'E), and Southern Afiican English
(SAfrE). National varieties such as Nigerian, Kenyan, Ghanaian, Cameroon and
Zambian Englishes are also discussed. Only a few studies have discussed the variation
of the vowel systems in an attempt to identify the phonological distinctions among these
AfrE varieties, but sampling techniques, methods of data collection, and the analysis
applied in the research are under described. It is uncertain whether the findings reported
in these studies represent distinctions exhibited by a particular social class, age group,
gender, or by an entire speech community in the densely multilingual speech
communities in sub-Saharan Africa.

The aim of this exploratory study is to remedy such methodological flaws by
presenting a methodological framework that will at least partly answer the need for more
empirical research in multilingual Africa.

Such a task can be accomplished by maximizing the advantages of the empirical
tools of acoustic phonetics. and sociolinguistic research. In using computer software
designed to extract, measure and plot vowels, we seek to verify the widely documented

findings regarding the AfrE vowel systems .

1. 2 Background to the study

The English language enjoys institutional support in most African countries that
were under British colonial rule. Ngugi wa Thiong'o (1984:13) points out that in the
colonial and post-colonial era in these countries, “English was the official vehicle and
the magic formula to colonial [and neo-colonial] elitedom.” It is still the official
language and medium of instruction in schools, colleges and universities. However,
English remains a minority language in regard to number of users and frequency of use.
African languages remain the dominant languages of informal interactions, even among
well-educated AfrE speakers.

In language contact situations, features of native languages are bound to be
manifested in the realization of the acquired languages. Non-native speakers of English
in Africa may perceive and produce English vowels based on the phonological schema
of their first language. Consequently, a new form of English, distinctive to that speech
community, grows and over a period of time is institutionalized and accepted. However,
although the foregoing assertion assumes that speakers of related Lls may have
relatively similar perceptions and production of English, it should be stressed that social
factors may also influence production and perception of non-native sounds.

The regular usage and acceptance of such new varieties of English eventually
leads to institutionalization of the Africanized variety and a gradual change of linguisitc
roles and attitudes in the speech communities. The Africanized English that was a
stigmatized non-standard variety at one time has become a prestigious standard variety

and a target language for many new learners of English in the speech community.

In an earlier attitudinal study, Mutonya (1997) reported that AfrE speakers readily
identify regional varieties of AfrE and attach attitudinal values to sub-varieties of

regional AfrE varieties on the basis of degree of accentedness. Kachru (1982245)

reported similar attitudinal judgments among Ghanaian speakers of AfrE:

educated Ghanaian English is acceptable; but as Sey (1973: 1) warns us, it does not
entail competence in speaking RP since in Ghana the type that strives too obviously to
approximate to RP is frowned upon as distasteful and pedantic. ln Nigeria the situation is
not different from Ghana or India... Many Nigerians will consider as affected or even
snobbish any Nigerian who speaks like a native speaker of English.

However, there is hardly any empirical evidence that teases out the distinctive
linguistic characteristics of each one of these varieties. Besides such attitudinal research,
detailed linguistic analysis of the varieties will help in teasing out the defining features of
those varieties. By utilizing tools of research that will guarantee greater precision, this
study begins that work by initially measuring acoustic qualities of vowel production in
each regional variety, followed by a methodical measurement of listeners’ abilities to
identify and discriminate those vowels, and finally, a correlation of production and
perception results. Correlating speech production and perception will help determine the

most probable vowel system in each region.

1. 3 Theoretical Framework
This study seeks to identify the nature of vowel systems of a homogeneous
sample of university students from Kenya, Ghana and Zimbabwe. Measurements of
vowel production and perception are correlated to determine the probable vowel system
of each sample.
Peterson and Barney (1952) observed that speakers’ pronunciations of vowels

within words are influenced by their particular dialectal background. Furthermore, they

noted, pronunciation of vowels may differ both in phonetic quality and in measurable

characteristics from that produced speakers with other backgrounds. Listeners, likewise,

are influenced in their identification of a sound by their past linguistic experience.
Lieberman and Blumstein (1993:153) accentuate the importance of

psychoacoustic tests in speech perception research:

Although it is possible to perform precise analyses of speech signals using electronic
instruments and computer programs that effect various mathematical transformations of
the signal, these analyses are, in themselves, meaningless. We can never be certain that
we have actually isolated the acoustic cues that people use to transmit information to
each other unless we run psychoacoustic studies in which human listeners respond to
acoustic signals that differ with respect to the acoustic cues that we think are relevant

This research is further informed by Lieberman and Blumstein’s (1993:170...)
phonetic theory (physiological theory for vowels) that is structured in terms of
biological mechanisms that are involved in the production and perception of speech.

Contrary to traditional "articulatory" vowel theory that postulate that the phonetic
quality of vowels is derived from the position and the height of point of constriction of
the tongue, this physiological theory, as demonstrated by Steven and House (1955),
postulates that most vowels can be generated by means of many different articulatory
patterns, e.g., adjustments of lip opening and total vocal tract length.‘ The only vowels
that need particular tongue contours are [i], [u] and [a]. Since that is the case, Jacobson
and Ladefoged (1972:93) prefer perception to production in analyzing and synthesizing

a speech signal:

The nature of some vowel targets is much more likely to be auditory than articulatory.
The particular articulatory mechanism that a speaker makes use of to attain a vowel
target is of secondary importance only

The physiological approach to phonetic theory for vowels recognizes the fact that
the shape of the supralaryngeal vocal tract determines the particular acoustic signal, and
different speech sounds are specified by different acoustic signals. Since different

8

speakers have different supralyngaeal vocal lengths, human speakers do not attempt to
produce the same absolute formant frequency values for the "same" vowel; instead, they
produce a set of format frequencies that is frequency-scaled to their approximate
supralaryngeal vocal tract length.

In vowel perception, Lieberman argues, a listener approximates the length of a
speaker's supralaryngeal vocal tract and normalizes it within the parameters of an

appropriate vowel space.

A human listener has to determine the probable length of the supralaryngeal vocal tract of
the speakers he is listening to in order to determine the frequency parameters of the
appropriate "vowel" space. Psychoacoustic experiments show that human listeners can
make use of various acoustic cues and strategies to effect tract normalization. (1993: 179)

Phonetic information produced by the speaker and the phonological knowledge of
the listener are inter-linked in a process that reveals to us the authenticity or relevance
of the acoustic cues we analyze, in this case, the similarities and differences that we seek
to define in the variation of AfrE vowel systems.

The physiological approach further states that, since the quantal vowels, ([i], [u],
[a]), are maximally distinct, they help define the vowel space within which speakers
differentiate the other vowels of English. The quantal vowels have well-defined spectral

peaks because of the convergence of two formant frequencies:

F1 and F2 converge to yield a central spectral peak at about 1 kHz for [a]. F2 and F3
converge to yield a high frequency peak for [i], F1 and F2 converge to yield a low
frequency spectral peak for [u]. (Lieberman and Blumstein 1993: 175)

Several psycho-acoustic tests have determined that [i] and [u] produce the lowest

errors of all vowels when listeners are asked to identify the vowels. In contrast, the

Peterson and Barney (1952) study shows a high rate of confusion occurs between [Ct]

and [o ] due to the instability of phonemic /a / and /0 / distinctions in many dialects of

English . Are these findings manifested in the study of AfrE vowel systems?

Lieberman (1993:182-3) further states that languages do not have the same
inventory of peripheral, non-quantal vowels. Speakers of a language may be unable to
identify non-quantal vowels of another language with certainty, because vowels seem to
be perceived by means of neural acoustic property detectors that respond to particular
acoustic signals. Moreover, Lieberman argues, the range of formant frequencies to
which these property detectors can potentially respond is delimited by the quantal
vowels [i], [u] and [a]. Our response to particular vowel sounds is made possible by
neural devices that are "tuned" to particular acoustic signals that reflect the constraints of
our speech producing mechanisms. Furthermore, humans partition the possible range of
formant frequencies that the human vocal tract can generate differently as they grow up
in different linguistic environments. Therefore, the different linguistic and social
background of different AfrE speakers will condition the nature of acoustic signals that
their neural acoustic devices detect in speech perception.

Other studies have contributed to a better understanding of a physiological
approach to phonetic theory. Niedzielski (1999: 63) suggests that listeners use a variety
of different factors in their perception and interpretation of a speaker's dialect, including
social expectations that are influenced by socially constructed beliefs and stereotypes

Strand (1999:96) states that, beyond the language-specific phonetics, socially
constructed information also affects how we perceive and categorize speech sounds.
Some of those influencing factors include dialect background or nationality of the

speaker. We expect our speakers and listeners from each region to exhibit such socially

10

determined influences in their speech production and perception, and the results of this
present study will be interpreted in light of earlier attitudinal work to begin an account

of such influences.

l. 4 Scope and Limitations

In this exploratory study, only a representative sample will be chosen from each
region. Speech samples will be recorded from a population that is controlled for level of
education, mother tongue, region and linguistic evidence to be analyzed. A major
criticism leveled against previous research efforts in the region is the glaring
overgeneralization that seems oblivious to the multilingual complexities of sub-Saharan
Africa. Controlling our sample for homogeneity seeks to avoid such a methodological
flaw.

Respondents are chosen from a population of male and female university students
from Kenya (EAfrE), Zimbabwe (SAfrE), and Ghana (WAfrE). A comparative study of
production and perception of such a sample will enable us to make concrete claims about
the vowel systems of educated speakers in those regions.

The implicit target sample of earlier studies was educated AfrE speakers.
Angogo and Hancock (1982) categorize the following four types of Englishes spoken
Sub-Saharan Africa: Type 1 is the native English of whites and expatriates born or living
in Africa; Type 2 is the native English of Africans of racially or linguistically mixed
marriages; Type 3 is the non-native English spoken as a second language by well-

educated Africans who, by virtue of their academic or professional life, speak English

11

fluently as a second language; Type 4 is pidgin/Creole English, such as Nigerian Pidgin
English. Angogo and Hancock conclude that Type 3 is prototypical AfrE. This study
focuses on educated AfrE speakers. Although all other varieties mentioned above are
relevant in understanding AfrE, the scope of the current study limits us from discussing
them comprehensively.

Gfirlach ( 1991 :24), argues that pronunciation of AfrE speakers is 'the most reliable

test for localizing a speaker.’ Schmied (1991 :57) concurs:

the pronunciation of English in Africa is of particular importance because (non-
standard) pronunciation features seem to be the most persisitent in African varieties i.e.,
they are retained even in the speech of the most educated speakers.

It is for this reason that we chose phonological evidence over syntactic, lexical, or
morphological variation for this study.

Kenya, Zimbabwe, and Ghana were chosen specifically because of their almost
identical history of colonization by Britain and for the dominant role that English
continues to play in the social relations of the citizens of these countries. The Bantu and
Kwa language groups were chosen because they belong to the larger Niger-Congo
family group. However, we should note that even among these language group there are
subtle dialectal differences, but have minimized the differences among Kwa speakers by
choosing respondents from the Akan-dialect cluster.

Kwa languages are a branch of the Niger-Congo language family spoken by the
inhabitants of an area extending along the Atlantic coast of Africa from Céte d'Ivoire to
the Nigerian border and including the southern parts of Cote d'Ivoire, Ghana, Togo, and
Benin. The Kwa languages, Stewart (1984) argue, include the Akan cluster, with 7

million speakers. The principal members of the Akan cluster are Asante Twi, Akuapem

12

(Akuapim), and F ante in Ghana; Anyi and Baule (2 million) in Cote d'Ivoire; and Ewe (2
million) in southeast Ghana and southern Togo.

Our Ghanaian respondents are L1 speakers of the Akan dialect cluster, which
comprises Twi , Asante and Fante languages. Some respondents spoke fluently two or
three of these languages.

Bantu languages are spoken by millions of people in sub-Saharan Africa below a
line roughly demarcated by the southern boundaries of Nigeria, Chad, Central African
republic, Sudan, Ethiopia and Somalia. 15 of the 37 languages of Africa which have a
million or more speakers are Bantu languages (Hinnebuschl989: 451)

Kenyan respondents speak the following Bantu languages: Gikuyu, Kikamba,
Kitaita and Ekegusii . Most Zimbabweans are Shona and Ndebele speakers although
some speak Chinyanja and Chichewa.

Suprasegmental variations such as pitch, tone, Advanced Tongue Root (ATR), are
not within the scope of the current study. The study focuses only on the onset (steady
state) characteristics of vowels. F1 and F2 frequency measurement is the only criteria
used in this study to determine acoustic quality of the vowels. The voWels were elicited
by requesting respondents to read stimulus word-list adapted from previous studies.

1. 5 Data Collection

Two sets of data were collected for this study: acoustic data from speakers and
psychoacoustic data from listeners.
Acoustic Data:

A questionnaire and interview method was used to collect data. The questionnaire

(see Appendix A) comprised four sections: biographical information, word list, reading

13

passage and respondent's data on language use. Every interview was tape-recorded.
Having provided the interviewer with biographical information, the respondent was then
asked to read a list of nineteen words . The word list was adapted from the lexical items
used in previous studies (see Figure 1 below), but other lexical items were added to
represent vowels and diphthongs missing from Figure 1. The respondent then read a
passage that was constructed to incorporate the lexical items in the word list, in order to
give each respondent the chance to produce the lexical item in two styles: formal style for
the word list, and less formal for the reading passage. For this particular study, word list
productions were heavily relied upon, although in a few cases we relied on reading
passage productions in cases of audio problems or misreading of words. Such cases were
very rare and did not influence our findings.
Psychoacoustic data

After results of the acoustic analysis were obtained, a perceptual test was carried
out. A sample of subjects with the same demographic characteristics as those who read
the word list was identified. Speech perception of fifty university students from each
region were elicited using Identification Test (IT) and Minimal pair Test (MPT). In each
of these two tasks, listeners heard fifty seven minimal pairs of carefully selected
monosyllabic words, recorded by an RP speaker. In the IT task, listeners identified each
word in every pair, while the MPT entailed determining whether the words in each pair
were “same” or “different.” (see Appendix D)

1. 6 Data Analysis
The steady state of vowels were extracted from the recordings and frequencies of

the first two formants (F 1 and F2) calculated using a computer program (SIGNALYZE,

l4

version 3.12) designed for acoustic analysis. The F1 and F2 scores were calculated using
linear predictive coding (LPC). Labov (1994) argues that LPC increases accuracy and
reliability achieved in acoustic analyses through the use of software methods that apply
LPC to the digitized speech wave. The formant measurements were subsequently entered
into a computer program (PLOTNIK, version 04) designed for plotting and displaying
vowel systems.

Results of the psychoacoustic test was calculated to determine the frequency of
correct identification of each vowel and discriminations between them in a minimal pair

test. Comparisons were made and conclusions drawn.

15

Chapter 2: Background to the Study

2. l. 0 Introduction

The background to the study of AfrE regional vowel systems is presented in the
form of a review of relevant literature. Previous studies have provided little or no
empirical evidence to corroborate observations and conclusions, and subsequently
subjecting the studies to accusations of overgeneralizations, methodological flaws and
implicit assumption in representation of AfrE vowel systems.
2. 2. 0 Background to the study

2. 2. 1 African English?

The term ‘African English’ has been used as a theoretical abstraction by the few
scholars who have done research on the forms and functions of English in Anglophone
Africa. The most notable publications in this field are Todd (1982), Ladham (1982),

Angogo and Hancock (1982), Pride (1982), Kachru (1986), Cheshire (1991), Schmied

(1991), Gérlach (1991), and Adegbija (1994).

Is it presumptuous to conceive of such an idea as ‘African English,’ taking into
account that English-speaking sub-Saharan African countries are densely multilingual
and that English is spoken by an educated minority as a second language? Some scholars
have attempted to answer this question as follows.

Gorlach (1991: 123) carried out tests among native English speakers of British and
American origins. He found that these students had no great difficulty in identifying a
speaker from Nigeria as African. Although this observation does not provide much

evidence regarding the existence of an African English, it vaguely points to the fact there

are some distinct linguistic features that are uniquely African, at least in the speech
perception of other non-Afiican speakers of English.

Schmied (1991 :2) defines African English as :

forms of English spoken by African speakers; this does not imply that there is an
acknowledged variety or that there are several distinct varieties of the language, nor that
these forms are already standardized and codified in any way. '

Schmied’s definition uses the geographical criterion to identify the English
variety. It also groups together all forms of English, from the English-based Creoles and
Pidgins in West Africa to other Africanized and non-Africanized English varieties
spoken in most English speaking regions. In view of the linguistic diversity of Africa, this
is a misleading definition.

Bokamba (1982:78) reiterates Gorlach’s argument that native speakers of English
are able to positively identify an African English utterance. He alludes to distinctive

linguistic characteristics that define AfrE:

these Englishes share certain properties that can be identified as Africanisms, in that
they reflect structural characteristics of African languages. Specifically, these properties
can be discovered at all linguistic levels: phonological, morphological, semantic and
syntactic.

Bokamba’s study focuses on the syntactic and semantic properties of sentences produced

by educated Africans. Our area of concern in this study is phonology.
The Africanisms defined by Bokamba are, in Lanham’s (19651198) words,

‘deviations from the authentic English’ which arise as a result of the transfer of features
and strategies previously known from the mother tongue to the target language. In every
major linguistic area of the world where English is learnt as a second or foreign language,
Lanham argues, there is a characteristic set of deviations from authentic English, much of

which is as a result of transfer from the mother tongue to English. With the passage of

17

time these deviations become institutionalized and give specific identity to the English
variety, as is the case for Indian English, African English, Singapore English, and so on.

These early descriptive studies played a very important role in laying a
foundation for the study of English in Africa. Although evidence does not point to careful
empirical analysis of data, such pioneering studies generated a research interest that has
culminated in more studies on varieties of African English. In a much more recent
development, the first international dictionary of World Englishes (1999) classifies
African English as one of the major varieties of English.

These studies have presented phonological, syntactic, lexical and morphological
evidence in support of the distinctness of an African English variety in relation to RP and
other Englishes. However, this study is limited only to phonological evidence. Why focus
on phonology?

Pronunciation is the ‘most reliable test for localizing a speaker’ (Gorlach
1991224); Schmied's ( 1991:57) agrees:

the pronunciation of English in Africa is of particular importance because (non-standard)
pronunciation features seem to be the most persistent in African varieties i.e., they are
retained even in the speech of the most educated speakers.

Gimson (1989:3180 notes that the British phoneme system may pose problems
for foreign learners of English. The main problems are identified as: the tense-lax
opposition of the close vowels /i/-/I/, /u/-/u/; the existence of a long central vowel / 3/;
and phonemic length.

Schmied, a widely cited authority in the field, makes three contrasting

generalizations about the AfrE and RP vowel systems.

(i) There is a length difference in vowels. The African English short vowels are longer

and more peripheral than in RP, especially the closed /I/, /U/ or /A/.

18

(ii) The central vowels /A/ (but), and /3/ (bird) are avoided and tend towards open
positions [ i, a].
(iii) Diphthongs tend to be monophthongized.

Schmied's characterization of the AfrE vowel system is broadly acknowledged
in the literature, and many studies have followed with similar generalized descriptions of
AfrE’s regional vowel systems. However, such generalizations are yet to be corrobated

by compelling empirical evidence as proposed in this study.

2. 2. 2 AfrE’s Regional Vowel Systems

 

 

Example RP WafrE EafrE
Bead i i i
Bid 1/ /
Bade e e

Bed ' e e \

Bird 3

Bad a\ a a

Bud A

Bod o 0

Board 0 > o
Bode o ............. 0

Pull 0

pool u;u\u

Figure 1 - Adaptedl representation of Schmied's classical comparison of RP, WAfrE,

and EAfrE

 

' Phonetic symbols have been adapted to reflect a more consistent phonetic transcription.

19

Although not included in Schmied’s illustration of vowel systems, the following

diphthongs have been investigated; [aI]buy, [au]bout, [3i]boy, [ia]year, [03] sure and
[89] hair. Schmied’s (1991 :59) characterization of AfrE diphthongs states with a sense

of finality that, “all the centring diphthongs (lial, leel, lua/ tend to be pronounced as

opening diphthongs or double monophthongs (lia, ea, ual)”

Figure 1 contains background information that is essential to this study. The
primary literature on AfrE vowel variation utilizes information contained in this
illustration.

Figure 1 compares and contrasts WAfrE and EAfrE vowel systems with those of
the RP system. AfrE researchers, including Schmied, have made general claims that the
SAfrE vowel system is identical to EAfrE, considering the common Bantu substratum of
L2 English speakers of these varieties. Whether such conclusions represent every group
in this ethnically and linguistically diverse region has yet to be proven empirically.

A quick glance at Figure 1 shows that WAfrE is portrayed as a seven-vowel
system whereas EAfrE is a five—vowel system. Previous studies attribute variation in the
AfrE vowel systems to such phonological processes as simplification leading to loss of
quantity distinction, reduction leading to mergers, and decentralization of the central
vowels. Some major claims that have been made in the literature concerning the nature
of the vowel systems of AfrE are stated below.

Schmied (1991) attributes a major source of AfrE variation to the ‘deviation’ in

the production of the RP long central vowel [3] (as in bird) by speakers of each regional

20

speech community. Schmied claims that [3] backs to [o] in WAfrE, fronts and lowers to

[a] or fronts to [e] in EAfrE, while in SAfrE it fronts to [e].
Todd (1984) provides further examples of backing and fronting of the central

vowels in the realignment of RP vowels in the WAfrE vowel system, e.g., the /A/

phoneme which occurs in the word but is realized as [0]. Additionally, the la/ phoneme
as in hat, rat, bad is realized as [(1]. Both these claims are consistent with what Schmied

presents in Figure 1.

Bokamba (1992:21) also discusses the reduction in the AfrE vowel system to

five or seven vowels; /i, e, a, o, u / and perhaps /0/ and /8/ resulting from vowel mergers

in such pairs and triplets as bit/beat, had/hard,full/fool, and cut/court/caught. Bokamba

further argues that individual items may be realized variously: bed as [bed] or [bed], bird

as [bed] or [bad]. These claims are also consistent with Schmied's, as shown in Figure l.

Angogo and Hancock (1980: 72f) contend that, in discussing regional varieties of
African English, it is important to note that each variety consists of several levels of style
and acceptability. Nevertheless, like any other regional varieties of English, they contend,
AfrE varieties exhibit a homogeneity between English as spoken in West Africa and
English as spoken in East Africa. Furthermore, in their judgment, the model for AfrE is
the non-native English spoken fluently as a second language by African people who have
learnt an African language first in life, but, because of their education or profession, have
grown up hearing and using English daily. These social groups speak English as well as,
or maybe even better than, their mother tongues.

Angogo and Hancock (1980: 72) are explicit about the phonology:

21

[AfrE] phonology is already becoming well established. This is evident. from the fact that
speakers whose native languages have a wide range of vowel sounds nevertheless retain

[AfrE] phonology; Igbo distinguishes between tense and lax [u] and [u], and [i] and [I]
but Igbo speakers of [WAfrE] will not as a rule differentiate between the pairs fool/full
or sheep/ship. Temne speakers contrast [a], [a] in their language, but for them, faster will

still be pronounced [fasta], not [fa:sto].

These claims about WAfrE, are also consistent with Schmied's.

In short, Angogo and Hancock (74-5) argue specifically that regional varieties
exist in spite of considerable internal variation. They argue that phonology is perhaps the
most distinctive characteristic of AfrE. Furthermore, as stated earlier in this study, they
insist that it is quite possible to generalize about the existence of a ‘West African
English’ and ‘East African English,’ in spite of their respective geographical variant
forms, in the same way that we speak of an ‘American English’ that consists of many
regional and social varieties.

Whether 'phonological cohesiveness' condenses varieties and subvarieties of
Englishes spoken in particular regions into one geographical variety (WAfrE, EAfrE, or
SAfrE) is an issue that continues to be debated. This paper seeks not to be embroiled in
the debate, but proposes extensive empirical studies (analogous to the one carried out in
this study), focusing on well-defined homogeneous populations in’the regions as a
beginning "phonetic" approach to the problem.

Banda (1996) also laments the methodological flaws in research on AfrE, and the
paucity of phonological indices of educated Africans, whose speech is the target language
of most learners of English as a second language in sub-Saharan Africa. Banda concurs
with Jibril’s (1986) claim that educated AfrE with a fairly African accent is usually the
accepted target language norm for most English learners in Africa, while heavily

ethnically marked accents are not.

22

Consequently, Banda (1996:68) further argues that describing AfrE from the
perspective of English phoneme systems, as previous research has done, is not only

erroneous, but it also distorts data.

thorough and useful descriptions of New Englishes are unavailable, particularly the
phonological indices of the [Target Language] Educated AfrE. Studies that have
attempted to do comparative analyses of vowel and consonant systems in African
Englishes are usually based on [Non-native English varieties], but even lack the
phonological data and sociocultural awareness that is crucial for the sort of analysis
required... researchers have yet to identify, let alone describe the distinguishing acoustic
correlates in AfrE .

Banda may be criticized for generalizations, but certainly not for identifying the
problem that confronts research in AfrE. Our exploratory study addresses the
phonological uncertainities and gaps discussed by Angogo and Banda.

Even Josef Schmied, a leading authority in this field, acknowledges these
shortcomings when discussing variant forms of the RP long central vowel, which in his
opinion, ‘is the primary parameter that defines regional variation in AfrE.’ He cautions:

but their [RP I3 / realizations] tendencies are not uniform in a region, neither across all

ethnic groups (Igbo speakers tend towards le/ and Yoruba towards /a/ ...) nor across the
lexicon.

Note that he builds this variation into Figure 1.
The present study begins with the system described in Figure 1 as a basis for
reexamining the regional variation of vowel systems of Bantu (in Kenya and Zimbabwe)

and Kwa (in Ghana) speakers.

2. 2. 2. 1 West African English (WAfrE)

English is the official language in six West African countries.2 Todd (1982)

identifies the main types of English spoken in the region as:

 

2 English is the official language in Gambia, Sierra Leone, Liberia, Ghana, Nigeria and
Cameroon.

23

(i) Pidginized and Creolized variety: other-tongues creoles of Liberia (Merico) and
Sierra Leone (Krio) and of the Krio—speaking settlers in Gambia, Nigeria, Cameroon
and Equatorial Guinea. Pidgin Englishes of the coastal regions and of many urban
communities.
(ii) Second language English that is acquired in schools and strongly influenced by
the mother tongue(s) of speakers, also known as “ broken English.”
(iii) Standard WAfrE, which with the exception of Liberia " is equated with
British (RP) norms. It is spoken as a second language by educated African speakers.
(iv) the English of expatriates ( mainly American, British, Indian and Lebanese).
Our interest is in type (iii).
Like others, Todd argues that there are distinct characteristics of WAfrE:

in spite of regional and educational differences, certain generalizations can be

made about pronunciations of WAE largely because West African languages are

fairly similar in structure.

Similarity of the structure of the West African languages, as claimed by Todd
above, is a highly contentious issue. Even among the Niger-Congo languages there are
structural differences; some have SOV word order whereas others have SVO; moreover,
vowel systems are not uniform in this family group (Williamson .1989). Here we
summarize Todd's description of the WAfrE vowel system:

(a) WAfrE has fewer vowel contrasts than RP. It utilizes 7 vowels and 3 or 4 diphthongs.
(b) Central vowels and centring diphthongs are virtually nonexistent. Narrow diphthongs

in bade and goat tend to be monophthongized to [bed] and /got/.
Words which end in [a] in RP have [a] in WAfrE, e.g. hair [hea].

(c) Long central vowel fronting: bird is realized as [bed].
((1) [A ] is replaced by [0] : [bAd] is realized as [bod]

(e) [a] is realized as [a]

24

Trudgill and Hannah (1985), identify a WAfrE3 vowel system that contains seven
vowels and three diphthongs. This system has no central vowels and lacks tense/ lax

contrasts especially among high and low vowels.

 

 

 

 

 

 

 

 

/i/ bid, bead4
/e/ bade

la / bed, burn
/cr/ bad, bard
ID/ bod, bud
/o/ boat

/u/ pool, pull
lai/ buy

ID i /———— boy

/au/ ———- bout

Figure 2 - WAfrE Vowel System as presented by Trudgill and Hannah
This analysis is similar to other findings reported in other studies; it can be read as Figure
1 “backwards” - from phonemes to “word classes”

Writing about varieties of English in Cameroon, Todd (1982) argues that vocalic
contrasts of the seven-vowel system are reflected in Cameroon English. Consequently,
many Cameroonians will hardly distinguish between /i/ and /I/ as in bead and bid,

between lo/ and /D/ as in board and bad, or between /a/ and / a/ as in bard and

bad. Central vowels and centring dipthongs are avoided with the result that words like

year and hair are realized as lia/ and /he/. The seven-vowel system for the Cameroonian
vernaculars (Lamso and Bulu), and in effect Cameroon English, is shown in the vowel

chart in Figure 3.

 

3 By WafrE, the writers stress that they refer to ‘ varieties that are unarnbigously English, particularly those
spoken in Ghana, Nigeria and Sierra Leone.’ (102)

" Trudgill and Hannah’s lexical items are substituted with Schmied’s (Figure 1) without altering the vowels
intended. Similar changes have been made whenever different lexical items are used to represent vowels
already represented in Shmied’s figure.

25

The vowel chart also reflects the vowel system of West African Pidgin English,

as documented by Dwyer (1969 ) and Schneider (1966).

r u /

 

 

Figure 3 - Todd's representation of the vowel system of Cameroon English

Comparing RP and educated Ghanaian English, Sey (1973) provides the
following contrastive evidence:

RP Ghanaian English

Bad Ia/

 

Bard /CI/ /(1/

Bade /e/

 

Bod /D/ /D/
Bird /3/ /8/

Pool /u/ \

 

 

Pull /u/ ‘ /u/
Beat /i/ \
Bid /I/ /1/

Figure 4 - Sey's Comparison of RP and Ghanaian Vowel Systems

Two claims made by Sey are particularly important to our study :

(a) Sey claims the vowel /3 / as in bird is fronted and realized as /8/ contrary to

claims made in other studies

(b) Sey claims that the vowel /A/ as in the words bud is realised as either /D/ or
/8/. However, Sey qualifies this observation with another baffling claim:

/A/ does not occur in L1, but the most likely substitute for it [in L2]
would be /Ct/ and not /0/ or /8/ (147).
But he does not speculate on why, in fact, [D] and [8] actually occur.

In a comparative study of normative English pronunciations in Cameroon and
Nigeria by educated speakers, Bobda (1995) claims that educated Nigerian and
Cameroonian Englishes are homogeneous across regional and ethnic boundaries. He
argues that speakers of sub-varieties of English usually exhibit norms of the national
variety; for instance, speakers of a language which may have vowel contrasts similar to
RP's will neutralize the contrasts in their English productions in conformity with the
national norm. To support his claim, Bobda gives an example of Hausa language which
has contrastive [i] and [I] vowels, and yet speakers will not make a contrast in their
production of English beat and bit.

Bobda illustrates the similarities between Nigerian and Cameroonian Englishes as

shown in Figures 5 and 6.

27

RP Ni gE and CamE

Beat/i/ / /i/
Bid /I/
Pool/u/ /u/

Pull /u/

Bad /a/ /a/

 

Bard / CI/
Bod /D/
/ 0/
board/ 0/
bud /A/
goat Ion /

sure lue/

Figure 5 - A comparison of RP, Nigerian and Cameroonian Vowel Systems.
A spelling pronunciation of RP schwa into different segments in both varieties is
represented in Figure 6.

Bobda criticizes the overgeneralizations of studies of West African Englishes for

apparently misrepresenting the facts. He rejects the claim by Hancock and Angogo,

echoed by Gorlach (1991:135), Schmied (1991 :61) and others, that the pronunciation of
bird as [bod] is a feature of WAfrE. He contends that bird has the vowel [8] as in the

words girl and shirt, presumably distinct from [c], as in made.
Bobda claims that most of features identified as West Africanisms are indeed

Nigerianisms, which have different realizations in CamE. Although Bobda’s study

28

distances itself from others, the empirical basis of his findings is as doubtful as the others

and continues to illustrate the murky state of research in AfrE.

RP Ni gE and CamE

a sofa

C statement

i visible

 

e O docth

Ll status

Figure 6 - The Realization of RP Schwa in Nigerian and Cameroonian Englishes

Bobda’s assumptions are illustrative of the confusion that non-empirical studies
pose to the field. For instance, the claim that Cameroonian English is a “ some variety of
Nigerianisms” poses more questions than answers. Which Nigerianisms is Bobda
referring to? Do these characteristics transcend ethnic, social and linguistic boundaries?
Such concerns mitigate for a more empirical investigation of languages in densely
multilingual sub-Saharan region. ’

2. 2. 2. 2 East African English (EAfrE)

Angogo and Hancock (1982) state that while West Africa had little exposure to
native speakers of English, East Africa had a considerable number of native speakers who
had a greater influence in government and teaching during the colonial and the post-
independence era. Due to the continued reinforcement of British standards in the schools,
EAfrE never strayed far from the prestige dialect of England . They identify four types of

English in the region:

29

(i) Native English (NE) of whites and expatriates.
(ii) NE of Africans of racially or linguistically mixed marriages.
(iii) Non native English (NNE) spoken fluently as a second language. The speech variety
of the educated, mid and higher status Africans, ‘...it is the speech of this group, which
may be taken as the norm of the varieties of EAE.’ (308)
(iv) NNE imperfectly used as the foreign language by speakers 'who have little
knowledge of English.

The samples in this study were drawn from type (iii) speakers who are educated
and have a higher rate of contact with native speakers of English. The distinct vowel

characteristics of EAfrE are:

(a) Vowel raising : ran is realized as [ran]

(b) Contrast of /a/ and / A/ is collapsed to [a]

(c) Tensed and lax vowels are not distinguished.

Hancock and Todd (1987) provide additional cases of vowel mergers in educated EafrE:
(a) Little distinction is made between vowels in:
bear and bid (usually realized as [I] )
bad and bed (usually realized as [e] )
Far and for and fore (realized as [a] )
Pool and pull ( realized as [u] )
(b) Diphthongs are monophthongised lei/ is realised as [e], so that bade and bed are
often indistinguishable.

As stated above, there is no empirical evidence to support the above analysis.

30

2. 2. 2. 3 Southern African English (SAfrE)

Todd and Hancock (1987) identify nine Southern African English speaking
countries: Angola, Zambia, Malawi, Lesotho, Zimbabwe, Mozambique, Namibia,
Botswana and Swaziland. However, Angola and Mozambique have a much stronger
contact with Portuguese than English to be proper members of the set. Namibia's contact

with Afrikaans also disqualifies it.

Todd and Hancock (1987:431) further argue that:

There are many similarities between the English of east and southern Africa, the most
marked of which is the tendency to raise the vowel in back so that back and beg differ
mainly in terms of the final consonant.

They also note that SAfrE vowel length distinctions are rarely preserved, leading to such

mergers as:

/i/ and /I/ so that bead and bid are both realized as [bid]
/a/ and / 8/ so that bad and bed are realized as [ bad]

/D/ and/A/ so that bod and bud are realized as [de]

/u/ and /U/ so that pool and pull are realized as [pul]

Moreover, central vowels are avoided. Schwa is replaced by /a/ and / 3/ by /e/.

The absence of the SAfrE vowels in Figure 1 is due to the fact that the available
literature analyzes the phonological characteristics of native speakers of English in the
region. Literature on Black non-native speakers is scarce and sketchy.

Angogo and Hancock (1982) argue that there is a phonological similarity
between SAfrE and EAfrE varieties since both varieties are influenced by indigenous

languages that share a common Bantu substratum.

31

This study seeks to verify the assumptions about these broad varieties against
vowel production and perception results of relatively homogeneous samples from Ghana,

Kenya and Zimbabwe.

32

Chapter Three: Theoretical Framework
3. 1. 0 Introduction

This chapter discusses some basic theoretical premises that inform this study on
issues pertaining to vowel perception and production, and on the definition of phoneme.
By adapting some of the theoretical positions that are principally derived from empirical
research on monolingual and bilingual populations, this exploratory study seeks to
determine whether such assumptions apply to multilingual speakers. The Native
Language Magnet Theory, the principal guiding theory of the current research, is
particularly relevant, for it considers the perception of non-native speech.

3. 2. 0 Theoretical Framework

Theoretical assumptions discussed in this section are derived from previous
research in linguistics and psycholinguistics. Of particular interest to this study are
attempts to correlate speech production and perception (the phonetics/phonology
interface).

Taking into account the nonexistence of similar empirical research in AfrE vowel
systems, this study cautiously adapts theoretical assumptions based on the relationship
between speech production and perception which are drawn from studies that have
primarily focused on monolingual and bilingual speakers. This exploratory study, as
stated above, targets multilingual speakers of English as a non-native language.

Two particular theories are central to our analysis and correlation of AfrE vowel
production and perception. The Acoustic Theory of Speech Production provides an
overview of some basic assumptions in acoustic analysis as presented in chapter five; the

Native Language Magnet Theory, as outlined below, is our principle guiding theory in

33

our understanding and interpretation vowel perception as discussed in chapter four. Prior
to outlining these theories, some general theoretical assumptions about speech production
and perception are discussed.
3. 2.1 Some General Theoretical Assumptions
(a) Assumptions about AfrE
Central to this study is the assumption that African English is a very general

term that refers to diverse social, regional and ethnic sub-varieties of English in East,
West and Southern Africa.

Unlike previous studies that assumed homogeneity of broad regional varieties of
AfrE, this study assumes those varieties are diversely influenced by such social and
linguistic factors as level of education and ethnicity of these multilingual speakers.

Although RP vowel production is used as stimuli in listeners’ speech perception
tests in this exploratory study, it is assumed that RP is no longer the primary source of
input in the formation of varieties of African English. African Englishes are the
predominant target languages for new English speakers in the continent as educational
institutions have been Africanized. Unlike in the colonial and early pre-independence era
when RP speakers dominated the teaching core in teacher institutions of learning, and RP
was the target language for new English learners, AfrE speaking teachers have dominated
the classrooms , and consequently varieties of AfrE are target language for English
learners. Moreover, RP continues to loose its dominant and prestigious role, as more
communities place more emphasis in the growth of indigenous national languages. The
expanding roles and social acceptance of Pidgin continues to diminish the dominance of

European languages in Africa. For instance, West African Pidgin English (WAPE) has

34

been so much intertwined with English in daily usage that its impact on WAfrE has long
been overlooked. Although the impact of WAPE on WAfrE is not within the scope of
the current study, Dwyer (1969) and Schneider (1966) have reported that WAPE has a
seven vowel system. The similarity of WAPE and WAfrE's seven-vowel system, is an
interesting topic worth of serious consideration.

(b) Assumptions about Vowels Systems in Some African Languages

Having conducted extensive research in African languages in western and
southern African, including a reputable collaboration with numerous linguists in sub-
Saharan Africa, David Westermann and Ian Ward wrote a classic phonetic guide
designed for students interested in the study of African languages. Using x-rays to
determine tongue position during vowel production, and a Cardinal vowel schema to
compare and analysis vowel systems, these phoneticians observed that many African
languages have seven-vowel systems, a few have more than seven, while others have
five or six vowels only.

Ladefoged (1993) observes that phoneticians have extensively used the arbitrary
reference points of cardinal vowel system as a scale for describing a wide variety of
languages. He points out that the system consists of eight cardinal vowels, “evenly
spaced around the outside of the possible vowel area and designed to act as a fixed
reference point for phoneticians” (219). Although critics have noted the seemingly
arbitrary nature of the vowel representation and the confusion over whether vowels are
described on the chart in terms of tongue height or in terms of acoustic properties,

Ladefoged ( 221) concludes:

35

Despite all these problems, the cardinal vowel system works fairly successfully. It has allowed the
vowels of a large number of languages and dialects to be described with far greater precision than
has any other method.

The methods used by these phoneticians may have been the most appropriate
tools to ensure precision in vowel comparisons and characterization, which may not
withstand the test of time in modern studies, but may still represent assumptions worth
considering, in a field lacking in empirical research. However, these phoneticians have
emphasized more on the physical characteristics of the vowel. The non-physical attributes
are not investigated as the current study has attempted to in the perception tests.

There are eight cardinal vowel reference points commonly used by phoneticians
(see Appendix A’(i)) and Westermann and Ward have shown how the RP vowels fit into
the scheme ( See Appendix A’(ii)) . The vowel chart points 1 through 8 indicate the
limits of possible vowel quality. For example, point 1/i/, represents the highest and most
front possible vowel and its relation to the other front vowels.

Westermann and Ward make the following observations about vowels of African
languages:

(a) Cardinal vowel No. 1 /i/: They claim that every language contains a vowel of i-type
vowel.

(b) /e/ and / 8 / : A large number of African languages contain two /e/ sounds ; a close one

(Cardinal no. 2) and an open one (Cardinal no. 3). They claim that the English vowel in
the word bed lies between the two. In some languages, for instances Zulu, the two

vowels are allophones.

36

(0) Most languages have /a/ type vowel; in some languages, they further argue, it is

nearer to the front /a/ (Cardinal no. 4) and in others it is nearer the back /13/ (Cardinal

No.5) African /a/ is rarely like English /a/ in man.

(d) Like /e/ and /a/, /o/ and /o/ are two ./0/ sounds; close and open /0/ closer to Cardinal

No. 6 and /o/ Cardinal No. 7.

(e) Cardinal No. 8 /u/ can be long or short without changing quality.

These assumptions apply to a wide range of diverse languages, and only empirical
research will verifying the conclusions made. In the meantime, they can assumed to be a
rough depiction of vowels of Lls discussed in this study. Note that Westermann and
Ward explicitly illustrate a peripheral vowel system lacking in central vowels, consistent
with the acoustic analysis discussed later in the study. Do the perception results discussed

in Chapter five point towards a Native Language Magnet effect?

(c) Assumptions About Speech Production and Perception

Some of the general assumptions concerning the correlation between speech
production and perception are outlined below.

Kenstowicz (1994) argues that finding a proper balance between production and
perception of speech sounds of any language remains a central concern of linguistic
theory. Such a theoretical balance is essential, Kenstowicz further argues, considering the

limitations of the vocal and auditory organs in speech production and perception:

the phonological categories we do find empirically attested are constrained by the vocal
tract and the human auditory system anatomical apparatus not specifically evolved for the
articulation and perception of language. Phonological distinctions and categorizations
display gaps that appear arbitrary from a purely abstract, classificatory point of view but
seem to reflect contingencies of the articulatory and acoustic systems that realize

language in speech. (1994:136)

37

Furthermore, generative phonologists have developed a model for the
representation of speech sounds that is based on a close relation between phonetics and
phonology. The Articulator Model, Kenstowicz argues, postulates that articulators play a
central role not only in the production but also in the perception of speech sounds. The
underlying assumption is that phonetics represents the physical realization of abstract
linguistic categories.

A similar theoretical assumption is contained in Lieberman and Mattingly’s
(1985) Motor Theory of Speech Perception, which hypothesizes that listeners’
interpretations of acoustic signals are guided by articulatory gestures in the sense that a
variety of acoustic cues for a given feature point to a particular articulatory gesture. In
Lieberman’s ( 1993) term, this is vocal tract normalization, a case in which human
listeners (in the case of vowel perception) have to determine the probable length of the
supralaryngeal vocal tract of the speakers they are listening to in order to determine the
frequency parameters of the appropriate vowel space.

Experiments conducted by speech scientists studying speech perceptual
developments among infants have indicated that human beings are innately endowed with
special devices for the perception of speech that differ from the neural and cognitive
equipment used to perceive other sounds.

Moreover, Lieberman (1990) reports that experiments point to the fact that
listeners perceive speech categorically; that is , listeners’ ability to discriminate sounds
from two different categories, such as /i/ and /e/, should be relatively easy, but

discriminating two different tokens of /i/ from the same speaker should be difficult.

38

Listeners perceive the former as two distinct categories, but as a single sound category in
the latter case.

Another theoretical assumption that echos a principal premise of sociolonguistic
studies is the fact that perception and production of speech are influenced by linguistic
and nonlinguistic factors. Speech perception research has shown that adults have
difficulties perceiving and discriminating many nonnative contrasts. Best et al (1988),
for instance, argue that language environment influences developmental speech
perception, and that adult listeners perceive non-native sound contrasts based on a
schema developed from their native language (see the Native Language Magnet theory
below).

Language perception and production are intimately related and are, therefore,
difficult to separate operationally, considering that every speaker is simultaneously a

listener, and every listener is at least potentially a speaker.

3. 3 Assumptions About Non-Native Speech Perception.

Early theories that sought to define the relationship between speech perception
and production devoted considerable attention in seeking to explain where in the human
brain production and perception of speech took place. While some theorists argued that
the functions were completely separated, other theorists, motivated by Broca’s and
Wemicke’s discovery (of the distinct types of aphasia caused by lesions at different sites
of the brain), held that production is localized on one part of the brain and perception in
another, although the functions were interconnected. However, later studies showed the

relationship was more complex than that presented by the classical theorists.

39

In this section theoretical assumptions relating to the perception of non-native
sounds are discussed. Relevant to the current research is the perception of vowels of an
acquired second language that are diverse from the first language of multilingual
subjects. Identifying a correlation between speakers’ production and hearers’ perception
of vowel variants, will help us determine the phonetic and phonemic properties of AfrE
vowel systems.

In discussing perception of non-native speech contrasts, Best et al (1988) propose
that any explanation of language-specific effects on speech perception should take into
account the relation of phonetic properties to phonemic contrasts. Best and colleagues
postulate several hypotheses to explain non-native speech perception. Unlike the general
assumptions outlined in the previous section, the following hypotheses provide a more
linguistically-oriented explanation of the nature of non-native perceptions.

The central premise of the following hypotheses is that the native phonological
system of the listener plays a prominent role in perception of non-native sound contrasts.
These hypotheses are reported in Best et al (1988).

The Specific Phonological Relevance Hypothesis, reported in Werker et al
(1984), postulates that the more competent infants become in their native language, the
less their ability to discriminate contrasting non-native sounds becomes, especially if
those sounds do not have corresponding native phonological contrasts.

The Allophonic Experience Hypothesis, presented by Tees and Werker (1984),
argues that listeners are able to discriminate contrasting non-native phones if their native
language contains such similar allophonic variants. For instance, English-speaking adults

are able to discriminate, especially after perceptual training, nonnative contrasts in which

40

members occur as allophonic variants in English, e.g., Hindi [d"]-[t"], Spanish utterance
initial prevoiced versus voiceless unaspirated [b]-[p]. In contrast, listeners have persistent
difficulties with many nonnative contrasts in which one or both members fail to occur
allophonically in the (English) native language (Tees & Werker 1984).

Werker and Logan (1985) present the Phonemic Perception Hypothesis, which
assumes that during speech perception listeners who have acquired a phonological system
of their native language perceive incoming sounds phonemically and assimilate them to

phonemic categories of their native language. Best et al argue:

assimilation may take place regardless of whether those sounds are native or normative and
regardless of whether they actually occur allophonically or are simply phonetically similar to some
native category ( 1988: 347).

Such non-native speech perceptions, they further contend, reveals that when
phonemic perception (assimilation) occurs, it may fall into the following three
different categories:

(a) Single-category assimilation: Contrasting phones are assimilated as variants of a
single native category.

(b) Opposing-category assimilation: Phones are assimilated as opposing members of a

native phonological contrast.

(c) Category-goodness difference assimilation: one member is better assimilated to a
more phonetically similar category than the other.

(d) Non-assimilation: both members are phonetically dissimilar from any native

categories and therefore not assimilated.

41

Categories a, b, and c involve perception of phonemically relevant information. Non-
assimilated contrasts should be perceived in terms of their auditory or phonetic
characteristics. Most contrasts in the world’s languages fall under the first three classes.
AfrE speakers learn English as a second or third language after the phonologies of
their first languages are well developed. It will interesting to test whether the phonemic
perceptions and assimilations outlined in the above hypotheses and assumptions apply to

our sample.

3. 4 The Native Language Magnet Theory

The Native Language Magnet theory claims that vowels of the native language
are stored as prototypes or ideal exemplars of a given category. Further, these prototypes
are said to hold a special status in speech perception in that they act as perceptual
magnets, thus minimizing differences between the prototype and surrounding stimuli.

Kuhl (1993) reports on various studies that indicated that infants are innately
predisposed with language-general abilities that initially enable them to perceptually
partition a series of sounds at the places that the world’s languages divide the series of
stimuli into phonetic categories, rather than at arbitrary places. Furthermore, these initial
abilities do not depend on linguistic experience, since infants have shown the same effect
for foreign sounds they have never heard. On the other hand, research shows that
linguistic experience transforms the language-general perceptual abilities into language-
specific ones, as infants begin to acquire words, or as adults encounter difficulty in
discriminating certain foreign-language contrasts. Infants are thus born with an excellent

capacity to resolve the acoustic differences between speech sounds.

42

Studies from laboratory experiments conducted by Kuhl (1993: 125-6) reveal that:

Prototypes (which, for these purposes, 1 am defining as exceptionally “good” instances
of members of categories, ideal exemplars) play a unique role in speech perception.
Phonetic category prototypes function like perceptual magnets for other stimuli... they
attract nearby members of the category, rendering them perceptually similar to the
category prototype. .. the magnet effect is obtained only for the prototypes of native-
language categories.

The results have led to the development of the Native Language Magnet (NLM)
theory, which describes how innate factors and experience with a specific language
interact in the development of speech perception. Linguistic experiences in different
cultures (linguistic environments) result in magnets that differ in number and location of
vowel space for speakers and listeners growing up listening to different languages. Kuhl
shows that 3 six-month old infants growing up in an English, Swedish, and Japanese
environment will have language specific magnets that differ in location and number of
vowels in the vowel space.

Of central interest to the current study is the NLM‘s explanation of adults’

acquisition of a second language. Kuhl (19932131) argues:

The native-language categories of the listener interfere with the ability to perceive certain
phonetic distinctions in the new language. The proximity principle again holds: the
nearer a new sound is to a nativeolanguage magnet the more it will be assimilated by it,
making the new sound indistinguishable from the native-language sound. Phonologists
have argued that the phonetic categories of ones’ native language form some sort of
“sieve” through which phonetic units of newly acquired language must pass (Trubetzkoy,
1939). NLM provides a potential mechanism by which this could come about.

Werker and Polka (1993) report that research with adults has shown that
experience with a particular language leads to a decreased perception of at least some
non-native phonetic contrasts and an enhanced perception of native phonetic contrasts.
Therefore, adults will have difficulties in discriminating phonetic contrasts that are not

used in their native language, although they have some ability to do so. Adult speech

43

perception is organized to process the native language with the least effort and greatest
efficiency.

In cross-language adult vowel perception, Werker and Polka report, evidence
shows that effects of language experience are more apparent in the identification rather

than in the discrimination of sounds.

3. 5 Acoustic Theory of Speech Production

Having discussed our guiding theory in speech perception, a theoretical
framework for analysis of speech production is presented in this section. Jusczyk’s
(1986) articulation of the theory is our primary reference in this section.

The human vocal apparatus is a complex system involving many different sets of
muscles to control the actions of variousarticulators involved in the production of
speech. Different sets of articulators determine the nature of sound produced. The vocal
apparatus generates complex acoustic waveforms. Each speech sound that starts as a
pressure wave generated by the lungs has a source that generates an acoustic wave form
filtered by the vocal tract as voiced or voiceless.

The supralaryngeal tract varies in size and shape, resulting in different acoustic
filtering characteristics. For instance, one can narrow the vocal tract in one location while
widening it in another as when producing a vowel resulting in changes in acoustic
filtering characteristics. Therefore, the physical sound that is realized is a product of
several factors.

Jusczyk (1986:27-3f) defines those factors such as source spectrum S(f), vocal

tract transfer function T(f), and the radiation characteristic R(f); all interact in the

production of any given speech sound. Source spectrum consists of components at
multiples of the fundamental frequency. The amplitude of the components decreases by
about 12 dB per octave at high frequencies. T(f) relates to the filtering characteristics of
the vocal tract. It changes with the shape of the vocal tract from one articulatory position
to another. The shape of the vocal tract will determine which components of the source
spectrum will be reinforced and which ones will be suppressed. The resonant frequencies
favored by a particular vocal tract configuration are called formants, and they will appear
as formant peaks in the transfer function. R (f) describes the relation between the
volume velocity at month opening and sound pressure at distance from the lips. The
radiated sound pressure P(f) is a product of all the three components: P(f) = S(f) x T(f) x
R(f).

The acoustic filtering characteristics of the vocal tract are critical in determining
which speech sound emerges during an utterance. As the vocal tract is narrowed at some
location during articulation, its resonance characteristic changes.

3. 5. 1 Vowels

Vowels can be distinguished by reference to the frequency values for the first two
formants. The shared articulatory features of members of a class can be another criteria
for distinguishing vowels. Such a division can be made based on the point of the

narrowest constriction in the vocal tract, back vowels such as [u], [o], [o ]and [a] are

produced with the tongue raised or lowered at the back of the oral cavity. For central
vowels, tongue height is modified in the midpalate region, whereas for front vowels the

tongue is raised or lowered front of the palate.

45

Degree of closure gives the distinction between open and closed vowels; amount
of lip rounding distinguishes rounded and unrounded vowels. There are also
monophthongs and diphthongs.

Acoustic analysis of AfrE vowel systems presented in Chapter five is defined and
interpreted on the basis of Juscyzk’s analysis presented above.

3. 6. 0 Assumptions About Phonemic Identity
3. 6. 1 Introduction

While the previous sub-sections discussed AfrE vowel production, this sub-
section attempts to illuminate the phonemic shape of variant vowel systems based on
listeners’ perception of RP vowels.

Although AfrE vowel perception data is the main concern of this chapter, a
sketchy history of efforts towards identifying and analyzing phonemes may help
contextualize this study relative to earlier AfrE vowel studies discussed in Chapter Two.
Conclusions presented later in this chapter are based on a correlation of empirically
derived vowel production and perception results. The overview presented here is far
from being a comprehensive account of the development of phonemic theory, but it tries
to provide a description of some defining stages in the ongoing attempts to present a

precise analysis of variation in both physical and abstract terms.

3. 6. 2 On Defining a phoneme: A brief historical overview
A historical overview of descriptive and analytic attempts over the years at
interpreting phonetic and phonological (or phonemic) identity in language studies may

provide a better understanding of the phonemic investigation discussed in this chapter.

46

Outlined below are studies that have significantly contributed to the formulation of
phonemic theory and the refinement of methods of phonemic identification.

The relevance of such an overview to this study arises from shared objectives, in a
very general sense, that is, interest in the interpretation of phonetic and phonological
identity. Moreover, a significant contribution of this study lies in methodology: attempts
at identifying L2 (or even L3) phonemic variables from multilingual speakers by adapting
methods of analysis traditionally used in the investigation of monolingual speech
communities.

3. 6. 2. l Phonetic and Phonemic Identity

Trubetzskoy and the Prague school phonologists were instrumental in defining the
phoneme as a complex phonological unit realized by the sounds of speech. Earlier
scholars had viewed the phoneme as a transcriptional device. A phoneme, argued the
Prague school, was composed of a number of separate distinctive features with each
distinctive feature standing in opposition to another feature or its absence in at least one
other phoneme in the language. Adapting Saussurean theoretical concepts, these scholars
defined speech sounds as belonging to the physical parole while the phoneme belonged
to the abstract langue. As a result of the work of the Prague school, R.H. Robins argues,
the phoneme concept “became one of the fundamental elements of linguistic theory as a
whole, and of the scientific description and analysis of language” (1990: 226).

Ideas postulated by the American structuralists marked the next landmark in the
study of phonemes. Leonard Bloomfield’s physicalist view of language analysis and
Edward Sapir’s psychological school marked the early structuralists’ attempts at

determining the nature of a phoneme. While the psychological school viewed phonemes

47

as abstract concepts; ‘ideal sounds;’ ‘mental equivalents of a speech sound’ and
‘percepts,’ Bloomfieldians considered phonemes as physical constructs of speech sound-
‘ overt aspects of physical speech event.’ Bloomfield’s perspective dominated this
particular linguistic era.

In his widely cited definition of the phoneme as ‘a minimal unit of distinctive
sound’ (Language: 1933: 79), Bloomfield, as Stephen Anderson observes, implies that a
phoneme has a phonetic identity composed of distinctive and non-distinctive physical
(phonetic) prOperties. The first mentioned properties, Bloomfield contented, belong to the
phoneme while the latter properties, though present in actual realization of a phoneme,
are not significant for the linguistic study of speech.

Phoneme features are present in sound waves and have linguistic significance in
that speakers are trained to produce and respond to these features while ignoring the rest
of gross acoustic mass reaching the ear (1933: 79)

Bloomfield’s analysis lacked a method that would unequivocally recover
phonemic representations (distinctive properties) from phonetic data as defined.

In the 19405, while linguists seemed to have reached a consensus on the definition
of a phoneme, there were still disagreements on methods of identifying them, Anderson
reports. Consequently, more linguists concentrated on formulating methods of extracting
phonemes from language. Morris Swadesh’s formulation of basic principles of phonemic
analysis (the inductive procedure) is widely cited as the groundbreaking analytical
procedure whose refinement was the crux of subsequent statements on phonemic analysis
in the post-Bloomfieldian era. The following two principles were central to post-

Bloomfieldians’ analysis of the phoneme.

48

Post—Bloomfieldians made the assignment of phones to phonemes subject to what
is now generally referred to as the principle of bi-uniqueness. The principle stipulates that
if two words or utterances are pronounced alike, then they must receive the same
phonemic description; conversely, two words or utterances that have been given the same
phonemic analysis must be pronounced alike. The principle of bi-uniqueness was also
held to imply that, if a given phone was assigned to a particular phoneme in one position
of occurrence, then it must be assigned to the same phoneme in all its other positions of
occurrence; it could not be the allophone of one phoneme in one context and of another
phoneme in other contexts.

A second important principle of the post- Bloomfieldian approach was its
insistence that phonemic analysis should be carried out prior to and independently of
grammatical analysis. Neither this principle nor that of bi-uniqueness was at all widely
accepted outside the post-Bloomfieldian school, and they have been abandoned by
generative phonologists In short, the behaviorist approach of American structuralism
contended that phonemic structure could be adduced from information present in overt
speech.

By the 19503, Anderson further reports:

Science in general was becoming more concerned with the extent to which theories taken as a
whole had explanatory and predictive power within a given domain, bringing coherence and
clarity to it, rather than with the mannner in which individual statements within a theory can be
operationally verified. With this turn, much of the philosophical rationale for the specific
conceptual foundations of structuralism crumbled. (312)

Morris Halle and Noam Chomsky’s generative phonology revolutionalized
phonemic analysis by showing the inadequacies of the structuralist school. They rejected

the behaviorist approach in favor of a mentalist perspective similar to the one that Sapir

49

advocated. The bi-uniqueness principle, the cornerstone of structuralism, was shown to
be counterintuitive. Halle emphasized the centrality of rules in phonological description.
Halle’s presentation marked a major shift from a concentration on the properties of
phonological representation and their elements to a much greater stress on rules of
grammar. Chomsky, Halle and Lukoff (1956) had shown that phonological structure was
not independent from grammar as structuralists proposed.

Generative phonologists, Kenstowicz (1994) argues, interpret phonetic and
phonemic identity as two different conceptualizations or representations for phonological
information. On the one hand is surface (phonetic) representations which indicate how
sounds are actualized in speech-‘the instructions sent to the vocal apparatus to articulate
the sounds and acoustic properties that are isolated in order to decode the speech signal’
(7). Underlying (phonological) representation, on the other hand, contain the more
abstract information structures that an individual speaker has stored in the brain after
acquiring a language. The two representation are however ‘systematically related by
phonological rules’ (7).

Interest in establishing an interface between phonetics and phonology has been
widespread in recent years. In seeking to identify social and linguistic factors that account
for language variation, sociolinguists are increasingly seeking more precise methods of
extracting and analyzing phonetic and phonemic data. Such concerns for precision have
for instance led to development of computer software (as detailed in the previous chapter)
that provide detailed acoustic characteristics of sounds to enable researchers to identify
even subtle variations in speech production. Researchers interested in the American

Northern Cities Vowel Shift have made great inroads in refining methodology in

50

phonetics. Phonemic analysis continue to attract interest and generate more refinements
from interdisciplinary research.

Labov (1994) has noted that the Minimal Pair Test (MPT) for the
identification of phonemes continues to be a preferred method of analysis. This method
of determining whether a single sound difference distinguishes the meanings of two
words and classifies as separate phonemes the sounds that are responsible for a difference
in meaning in such a minimal pair. In other words, sounds are separate phonemes if they
contrast in identical environments; that is, if the choice of one sound over another in a
particular environment alters the meaning in a word then the two sounds are different
phonemes.

Labov argues that the MPT as an empirical method for determining the
contrastive status of phonemes in a speech community takes into account that listeners
perceive and respond to sounds in categorical terms. The test requires a native speaker of
a language variety to determine whether two utterances are ‘the same’ or ‘different.’ The
pairs of words are identical except for one element, which is represented as one type of
sound in one member of the pair and another type in the other member. This method has
been widely used in the extraction of phonemes discussed in this chapter.

Psycholinguists have shown evidence that points to the fact that the way in which
people speak greatly influences their perception of what is said to them. For example,
experiments have shown that speakers are unable (prior to training) to pronounce lax and
tense vowels if such a distinction is not realized in their language. It is further reported
that such people also have difficulty in hearing the difference between such distinctly

defined vowels. The basic premise of this motor theory of speech perception is that the

51

perception of speech is structured in terms of linguistic categories. Such a presumption is
made in this study.

Language acquisition research among bilingual adults and children has shown that
native language vowels are stored as prototypes. During speech perception such
prototypes act as magnets that ‘pull’ the non-native sound closer to the native prototype,
and as a result, spectral differences between the native and non-native sounds are
minimized. This is the central premise of the Native Language Magnet Theory; the

guiding theoretical framework of this study

52

Chapter Four: Methodology
4.] Introduction

This chapter describes the methods used in the collection and analysis of acoustic
and perception data. The study focuses only on the onset (steady state) characteristics of
vowels, while at the same time being cognizant of the fact that other phonetic features
distinguish the vowels within and across varieties.

The first half of this chapter describes the respondents and the procedures of
collection and analysis of acoustic data.
4. 2 Collection of Acoustic Data
4. 2. 1 The Production-Sample

All respondents were carefully selected based on their country of origin (i.e., sub-
Saharan African countries with a long tradition of English language usage), level of
education (i.e., university or college students) and their first language (i.e., speakers of
Kwa and Bantu languages). Male respondents were initially selected for purposes of our
preliminary study, and female respondents were later incorporated into the schema of this
broader study. Appropriate respondents for this study were identified by the ‘friend of a
friend’ network method, as described in Milroy (1980). Initially, I contacted my friends at
three African universities‘ who identified a network of friends that made up our sample.

Data was collected from male and female university students in Ghana, Kenya
and Zimbabwe. The age range, of these native speakers of Kwa and Bantu languages,
was between 19-45 years. (See Appendix A for respondents’ profiles). Speakers of the

Akan dialect cluster languages were chosen in Ghana (These include speakers of Fanti,

53

Twi, Ashanti and Ewe languages). Shona and Ndebele speakers in Zimbabwe, and the
largest Bantu language groups in Kenya: Gikuyu, Gikamba, Lubukusu, and Ekegusii.
4. 2. 2 Data Collection
Speech samples were elicited using the questionnaire method. In the first section
of the questionnaire, respondents were asked to answer brief biographical questions that
elicited age, gender, residence and language background information. Subsequently, the
subjects were requested to read a randomized list of 21 monosyllabic words that
contained vowels and diphthongs that are relevant to this study. The list included the
following words: bead, here/year, hair, hid, caught, bout, good, hard, sod/nod ,boat/goat,
tour, buy/pride, sure, board/saw, boy, mud, bad, bird, mood, bet/bed, hue made/name.
The word list was adapted from the lexical items used in previous studies on AfrE (see
Figure 1). A reading passage specifically constructed to incorporate lexical items already
presented in the wordlist was also presented to the subjects. The procedure allowed us to
analyze both casual and formal styles of production, on the stylistic continuum, for each
vowel under investigation. Each interview was tape recorded.
4. 3. 0 Data Analysis

4. 3. 1 Acoustic Measurement

The collected data was subjected to two kinds of analysis: quantitative and
qualitative. The most fundamental data, the vowels contained in the word-list, was
extracted and frequencies of the first two formants (F1 and F2) calculated using a
computer program (SIGNALYZE version 3.12) designed for acoustic analysis. The

extraction procedure involved digitizing sound by way of sound input from tape to

 

1 Collection of data in Africa was conducted by Dr. Harry Akussah of University of Ghana, Dr.
Sawasawa Kambewa of Malawi, Maina Mutonya of University of Witwatersrand, S.A.and Dr.

54

computer. Consequently, each word is saved as an acoustic waveform on a Signalyze
screen, from which the relevant vowel is identified, isolated, and F1/F 2 scores calculated.
Formant measurements were calculated by locating the cursor at an appropriate point of a
waveform, using linear predictive coding (LPC). The F1 frequencies are inversely
associated with the height dimension, and the F2 frequencies are associated with the
front-back dimension.

A data file for each respondent was subsequently prepared showing F 1, F2 scores,
vowel class, stress and word (see Appendix B). These scores were subsequently loaded
into a computer program (PLOTNIK (version 04)), a vowel system analysis program
developed by William Labov for plotting vowel systems from formant measurements
(see Appendix C 1). The data is recoded, a procedure that reads the orthography of the
word and adds phonetic information of the vowel, indicating, in code form, the manner,
place of articulation, and voicing of preceding segments and following sequences.

A preliminary accuracy evaluation task was performed after tokens were plotted
on a PLOTNIK vowel chart. When formant values of the same vowel and subject showed
wide variation, we double-checked them by listening to and comparing the spectrograms
of the tokens. In cases where a respondent clearly misread a word (e.g. read the word
boat as boot, made as mad, or sod as sad) on the word list, the correct word was
extracted from the reading passage and measurements of the correctly read vowel
replaced the misread one. After data from each group and region is recoded, a group
mean is calculated, and all files are normalized. Labov (1998223) explains the need to

normalize data:

 

Abraham Ndungu of Kenyatta University, Kenya.

55

[Normalization transforms] all measurements into a single reference grid which will show vowels
which sound the same with the same formant values. Ideally, this would duplicate the normalizing
process of the human ear and neural network...it has proven to be more successful than others in
eliminating effects due to differences in vocal tract length, while preserving those social
differences of age and sex that are inherent characteristics of the speech community .

Having reduced linguistically irrelevant differences in the production of speech
signals through normalization, the resultant acoustic vowel chart becomes an accurate
representation of the linguistic aspects of the vowels, facilitating both across-speaker and
across-language comparison as B.G. Yang, (1996) demonstrates in his comparative

study of American English and Korean vowels.

4. 4 Vowel Perception Data

Two perception tests were conducted to determinate listeners ability to identify
and discriminate RP vowels: The minimal pair test (MPT) and the identification test (IT).
4. 4. 1 Data Collection Methods

Data was collected using a tape recorded stimulus voice and questionnaire as the
main data elicitation tools. A minimal pair test (MPT) and identification test (IT) were
carefully designed to elicit perceptual responses of lexical items (and vowels indirectly)
used in the acoustic analysis presented in the previous chapter (see questionnaire and
minimal pair list in Appendix D). In the MPT, respondents were requested to mark
whether the each of the fifty-seven pairs of English words recorded on tape were the
same words or not. The IT entailed listening to the same fifty -seven pairs and writing on
the questionnaire provided what each word was.

The minimal pair list was adapted from a randomized list of 21 monosyllables
used earlier in the elicitation of acoustic data (see 3.2.2). However, in this case other

monosyllabic words were carefully added in order to create minimal pairs, while in some

56

instances words were simply reduplicated to test for a particular vowel. The list included
the following pairs: beat/bit, weed/wed, bid/bird, hid/hid, wade/word, bit/bet, meet/meet,
mad/mud, hay/hair, bid/beard, made/made, he/hue, bed/bad, bay/beer, bet/bet, bird/bird,
beat/bait, don/dawn, bird/bad, bird/bud, wade/wed, bed/bird, bat/bat, had/hard,
heard/hoed, bait/bat, bird/board, dam/dime, bat/bout, rug/rug, mud/mad, cat/cot,
hut/hurt, gut/got, mud/mode, land/lend, bud/board, got/goat, wooed/wood, law/lure,
sod/soy, call/call, but/bought, hot/hot, bought/bout, call/coil, walk/work, boot/boat,
caught/cot, cod/cured, cook/coke, good/good, got/gout, he/here/book/book, who/hue,
cot/caught.

An RP speaker carefully read the list for recording. All respondents listened to the
same list read by the same speaker. Subsequently, a correlation of acoustic and

perception results was done to determine the vowel systems.

4. 4. 2 The Perception-Sample

Responses were elicited from one hundred and sixty students (undergraduate and
graduate) from universities in Ghana, Malawi and Kenya. The demographic
characteristics of this perception-sample is similar to the production-sample discussed
above. In a nutshell, the age range of the sample is 19-45, both male and female
respondents participated, Kwa and Bantu speakers were identified to participate in the
survey. Respondents were identified using Leslie Milroy’s friend of friend network

method.

57

The RP speaker who read the stimulus words is a thirty year old graduate student
at Michigan State University. By the time of the recording, he had lived in the United
States for less than two years.

Acoustic and perception data is discussed in the following chapter.

58

Chapter 5: Data Analysis

5. 1 Data Presentation and Discussion
In this section, vowel production among male and female speakers of AfrE is
analyzed using Si gnalyze and Plotnik computer software.

As stated above, the current study focuses on vowel production among educated
male and female members of Kwa (in Ghana) and Bantu (in Kenya and Zimbabwe)
language groups, who have acquired English as a second or third language. This
narrowed focus takes cognizance of social and linguistic factors that are bound to affect
a non-native speaker’s production and perception of foreign language phones (i.e.
English).

English in sub-Saharan Africa has varied degrees of contact with diverse
indigenous African languages in different language settings. In many settings, dominant
indigenous languages compete strongly with English in societal perception and
assignment of roles. Apart from the dense multlingualism of the speech communities in
question, our study takes account of the diversity of multilingual backgrounds among
individual AfrE speakers and its assumed effect on production and perception. As
postulated in our guiding theoretical assumption, the NLM theory (see chapter 3), non-
native speakers will perceive, and presumably produce, foreign vowels in terms of native
phonological categories and as exemplars of native vowel types. Considering further that
many sub-Saharan countries have a diversity of L1 speakers (Nilotic, Semitic, Cushitic,
Bantu, Kwa and others), it was necessary to seek a homogeneous sample. Earlier studies

seemed to overlook this linguistic diversity.

59

The following analysis and discussion is based on acoustic measurements of
the physical qualities of vowels. In the next section, the non-physical qualities of the
vowels are measured using the vowel perception tests.

AfrE vowel variation is illustrated in Schmied’s comparison of RP vowel
realization with those of WAfrE and EAfrE speakers (Figure l). The illustration shows
how the RP thirteen-vowel system is realized as a seven-vowel system among WAfrE
speakers and a five-vowel system among EAfrE speakers through such phonological
processes such as mergers, lowering and fronting. The reduction in the AfrE vowel
systems may be a clear testimony of the influence of phonological systems of indigenous
African languages in the production of RP vowels by AfrE speakers. Such a claim
would reflect the primary tenet of the Native Language Magnet Theory (see Chapter 3)

According to earlier studies in the field, the WAfrE vowel system contrasts lax

and tense front and back central vowels /e/, /8/ and /o/, lo /, while EAfrE does not. Both

systems merge their lax and tense high vowels /i/, /I/ and /u/, /U/.

5. 1. l Schmied's Analysis of AfrE Vowel Systems

Figures 7, 8, and 8 recreate RP and AfrE vowel systems presented in Figure l in
a vowel chart configuration, highlighting suggested phonological processes that occur in
the production of AfrE vowels. Lowering, backing and fronting, are the major processes
that result from decentralization of RP central vowels.

Since F 1 and F2 scores are generally unavailable in the literature, Figures 7, 8,
and 9 are not acoustic representations of the vowel systems, but extracts intended to

display relations among the vowel systems as represented in Figure 1. Therefore, unlike

60

F2

 

n 0001
. i bead
I Did UDUII
e bade 3 bird 0 bode
8 bed Abud
9 board
1 a bed a bard

 

Figure 7 - representation of the RP vowel system as illustrated in Figure 1
our later vowel charts, which are based on instrumental measurements of F1 and F2
scores of each vowel nucleus, such scores are not indicated.

At this juncture, we should note that throughout this study vowel charts are
presented in a two-formant illustration , irrespective of whether the presentation is based
on approximations (Figures 7, 8, 9) or precise instrumental measurements of vowel
nuclei. First formant frequencies (F1) are plotted on the ordinate (vertical axis),
indicating the height of the tongue in the oral cavity. Second formant frequencies (F 2) are
plotted on the abscissa (horizontal axis) indicating the position of the vowel along the
front/back dimension of the oral cavity.

Figure 7 shows the configuration of an RP vowel system. On the front/back (F2)

dimension, three types of vowels are identified: front vowels [ i, I, e, e, a], central [3, /\ ],

61

and back vowels [ u:, u, D, o, a ]. On the height dimension (Fl), RP vowels are

distinguished as high [ i: I, u, 0], mid [ e, 8, 3, A, o, D] or low [ a, a ]. A tense/lax

 

 

 

 

 

 

F 2
band
. u
' 010 '
B “9””.
94.
"LEI
[El
‘ bad

Figure 8 - A representation of WAfrE as illustrated in Figure l.

(close/open) distinction is evident in the vowels [e/ 8] [ill], [u:/ u] [b /o]. Moreover, RP

has a vowel length distinction in [ i:/I, u:/ u, and D /o ]. In the discussion to follow, it is

necessary to consider the arrangement of the phonological space among AfrE speakers,
particularly in regard to RP's central vowels, vowel length, and tense/lax distinction.

In the WAfrE seven-vowel system that Schmied suggests (Figure 8), RP central

vowels [3 and A] are backed to [D]. [a] is lowered to [D]. The tense/lax distinction is

maintained in the front mid vowels [e/e].

62

Figure 9 illustrates Schmied’s EAfrE five-vowel system which (like WAfrE) has

no central vowels. RP's central vowels [3 and A] are lowered to [a]. In the discussion to

follow, our dissension with Schmied on this issue is not in the characterization of the

EAfrE system, but in the assumption that the EAfrE and SAfrE vowel systems are

 

 

 

 

 

 

 

F 2
1
p001
Dead 11
1
bid pull
l bade f . bode
e bed ’5 I; board 0

  
   
 
 

bird bud
bad bard

    
 

 

 

Figure 9 - A representation of EAfrE as illustrated in Figure 1

identical. RP's tense/lax distinction, evident in the WAfrE mid vowel system above, is

lost in the mergers of high, mid, and low vowels in EAfrE, and, by inference, SAfrE.

Schmied’s discussion of diphthongs is limited to the following observation that

other studies have echoed throughout the literature:

Diphthongs tend to have only marginal status and to be monophthongized. This is certainly true
for the closing diphthongs /ei/ and /'u /, where the second element is hardly heard in many African

varieties, thus almost coinciding with the /e/ and /o/ phonemes. Diphthongs with a longer glide are

preserved, but they are not really pronounced as falling diphthongs, i.e., with less emphasis on the
second element than the first, but rather as double monophthongs (e.g. /oi/, /au/). All the centering
diphthongs (/i'/ , /e'/, /u'/ tend to be pronounced as opening diphthongs or double monophthongs

(/ia, ea, ua/) (Schmied, 1991:59).

63

In the following subsections, we compare our findings with Schmied's and, by

implication, the vast majority of others who have commented on AfrE vowel systems.

5. 2 An Empirical Analysis of AfrE vowel systems

In this section, we present and discuss vowel production among respondents from
Kenya, Zimbabwe and Ghana, based on the results derived from Signalyze and Plotnik
analysis. The results for each region are presented in three phases. First, an overview of
vowel production by all respondents in each region is presented by combining all data
files into one. The file is then analyzed using Plotnik, resulting into vowel charts in
Figures 10, 11, and 12. Secondly, each general vowel plot is followed by individual
sample charts that exemplify the general trend in the majority of cases among male and
female respondents. Finally, observations are made and conclusions drawn.

It is important to ma at this juncture that the plot for “all tokens” in each region is
supposed to confirm the general picture and to justify the token groupings made in the
illustrative individual plots. For the instance, the token groupings shown in the
individual plots in each region and gender, is based on the emerging clustering of
tokens when all individual vowel plots in each region and gender are considered. The
two individual plots were chosen as the exemplars of the entire group. All F1/F2 scores
are shown in Appendix B2.

5. 2. l Kenyan Speakers

64

F2

3000 2900 2600 2400 2200 2030 I80) I 600 I 400 12(1) I 000 300
200 l l l l l l L l L l l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F V f "
hid
m“ bead "W good
i gear . mood }
400a ‘ —- ‘ .
F——_
‘ hai r sure a
500. e made bet I 50d board
name '
93" caught
L - J
6004 fi * f 4]
[E] II bad pride bout
7”“ bad bird hard
- mud
aoo- bl rd ‘1? ‘ J
k
900- All Kenyans (20 IMP)

Figure 10 - Vowel Plot of all Kenyan Respondents

Figure 10I represents the overall vowel production of 20 male and female
university students from Kenya. Based on the evidence presented, the vowel system is
evidently a peripheral five-vowel system that lacks central vowels. Vowel clustering in

Figure 10 can be summarized as follows:

65

/i/ bead,[hid, year]
/e/ made, bet, hair, name, [hid, year]
/a/ bad, bird , hard, pride, bout, mud

/o/ sod, goat , board, caught, boy, sure.
/u/ good, mood

Considering that such general plots may obscure finer details in the vowel
production, this study analyzes each vowel production in each individual system, and
assumes that patterns recurring in a majority of cases represent vowel production of the
entire sample.

5. 2. 2 Kenyan Male Respondents

The following two normalized vowel charts are representative of the vowel
production among eleven Kenyan male respondents. The collection, analysis and acoustic
measurements of the vowels is explained above. The normalized F1/F2 scores of all
respondents are shown in Appendix C2.

Figure 10.1 illustrates a plot of the F1 and F2 frequencies for the 22 tokens
produced by Njenga-a 24 year old Kenyan male university student. Njenga’s vowel
system, like all other AfrE vowel systems analyzed in this study, is a peripheral system

that, unlike the RP system, contains no central vowels. Central vowels are fronted,

 

1 Figures 10, 11 and 12 were carefully edited in order to reduce dense clustering of the 400
tokens and labels that would otherwise have obscured the details in the chart.

66

 

 

FZ-FI

3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800
200 I I l L l l I l l I l

 

 

300 -

 

 

  

 

 

400-
sod fopen
500] ' 0 J 9M 0
board 'b gang“
boy
600-1 E
[:Z] Ilbad
(bout
7001 1"” d
pride
1| bud
800- Emmi
E
900-
1000-

 

 

 

| Nl'engaglwormawralgsenza |

Figure 10.1 - Njenga’s Vowel Plot

backed or even lowered to merge with non central vowels. In Figure 10.1, vowels

contained in the tokens bead and hid are realized as [i]; vowels in the words made and

bet as [6]; year, hair, and name , are produced as [8]; bad, hard, pride, bout, bird and

mud [A] are pronounced as [a]; sad, open, goat, caught, board and boy as [0]; while

67

FZ-Fl
3000 28100 26100 2100 22100 2000 18100 isoo 1100 nice 1000 900

 

 

 

 

 

 

 

 

 

 

 

 

 

200
300-1 [ 'bb’“ I ] d‘mood u
Ohid 09m
400-
[ bmade CObeiJ 05m caught
0 97'0”?
500- [ 9w onair board '5‘; t 0
go:
‘0 name 8
6004 .
ED Iblrd I]? pride
Brim
a W“ v‘
700. bout
300.
900-
1000-
| Nl'ogugwlformallvrbanéenya |

Figure 10.2 - Njogu’s Vowel system ( a Kenyan Male)
mood , good and sure are realized as [u].

Njogu’s vowel production (Figure 10.2) is almost identical to Njenga’s (Figure
10.1).The only discernible difference is that while Njenga realizes the vowel in sure as
[u], Njogu realizes it as [0]. Few other Kenyan male respondents realize the vowel in

question as [o].

68

Based on acoustic data analyzed from our Kenya male sample (as represented in
the two vowel plots above), AfrE vowel production among Kenyan males may be
summarized in the representation shown below. Tokens in parentheses indicate across-
speaker variations identified in the plots. For instance, parenthesized [hair, bet] indicate
that there exist a difference in the respondents pronunciation of vowels in those tokens,
such that some respondents will realize vowels contained in these words in an identical
manner to other vowels of the cluster that they belong to. For example [hair and bet]
are pronounced as either /e/ or lal.

/i/ bead, hid

/e/ made, [bet, hair]

/8/ year, name, [hair, bet]

/a/ bad, bird , hard, pride, bout, mud

/0/ so, open, goat , board, caught, boy, [sure]

/u/ good, mood, [sure]

5. 2. 3 Kenyan Female Respondents
Now let us examine the vowel production among educated female respondents

from Kenya. Figures 10.3 and 10. 4 are representative of other individual vowel charts of
this homogenous sample of 11 respondents. The variation in the production of [on] as in
the word boat is the only distinct difference in production among these speakers.

Based on acoustic measurements of F1 and F2 frequencies, Mukami (Figure 10.3)

realizes the English vowels under investigation as follows: Vowels contained in the

tokens bead and hid, hue and here are realized as [i]; made, hair, bed are produced as

69

[e]; hard, bird ,bad, buy , mud, and bout are realized as [a]; while saw, nod, boy, and

boat are pronounced as [o] and good, mood, tour, and sure are produced as [u]. Note,

however, how close some of the “0” (or mid-back) tokens are to the “low” set (e.g. nod,

 

 

 

 

 

 

 

 

 

 

 

 

hard).

3000 2800 2500 2400 2200 2000 1800 1600 1400 1200 1000 800 t
200 I I g I I I l l I I I
300-

. $0?!
01110 good
0 here i 'bbead ll 0 mood
400. d‘hue 1 0 0"
our
' . l
500. “a"
0
'Rmade
e bed
600-
[El
700-
300+
9004
1000-
[ Mukami,21,i,'ionnal,0rban,xenya J

 

 

Figure 10.3 - Mukami’s Vowel System ( a Kenyan female)

7f)

Kamba’s vowel system (Figure 10.4) is almost identical to Mukami’s (Figure 10.3)
except for the production of [au] as in bout, which in Kamba’s system is realized as [o].

Kamba’s low vowels are also more compact (in a back position).

 

 

 

 

 

 

 

 

 

 

 

 

F2 - i1
2003000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800
300~ ' Ohid ' I
11
head mood sun
I he re I
400‘ l hue goodo our
8W boat
500«
I l bout
0
mode bed 00],]
600- hair e W
[E 5" b . I bod
“U
700. El bird
1-
a hard
H
800

 

  

Kamba,30,F,Urban,Kenga

  

 

Figure 10.4 - Kamba’s Vowel System ( a Kenyan Female)
Acoustic analysis of vowel production among educated female

respondents in Kenya can be summarized as follows:

71

/i/ hue, bead, hid, here
/e/ mad, hair, bed
/a/ hard, bird , bad, buy , mud, bout

/0/ saw, nod , boy, boat, bout

/u/ good, mood, tour , sure

An analysis of both male and female respondents from Kenya shows a clear
pattern of vowel clustering which varies only in the respondents’ realization of the front
mid vowel as both tense and lax. It should be noted that only a few of the respondents do

not make this distinction. The only other significant variation is that while male
respondents realize [Is] as /e/, female speakers produce it as /i/. As might be expected,
there is more range to the targets in low-vowel territory. The gender distinction noted

here is worth of serious investigation in future studies. Previous studies have lumped the

speaker together and no gender distinctions have been reported.

5. 3 Zimbabwe Respondents
Figure 1 1 represents an edited plot of F l and f2 frequencies for over 400 tokens produced

by 20 respondents in Zimbabwe. Like the EAfrE vowel chart in Figure 10, this is also a

peripheral system with no central vowels (although it can be argued that /8/ is fairly

central relative EAfrE and WAfrE systems).

72

F2

3000 28.00 26100 2100 2290 2000 18100 1600 1400 1200 100) 800

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

200
f ' V
hue
300a U
mood J0. 900d
400. tour
1 - l
‘ sov sure board

500‘ '0 mode 0 v nod 0"

bed name 9001

0’ boot
, bed

600- b1rdD D

3 bad D l I
E ‘ D 1 mod bill]
700‘ bout

‘ mud
pride
800~ '0 :31 hard
11
’00 h .1000 ‘ A
All Zimbabweans (20 M&F)

 

 

Figure l 1 - A vowel plot for all Zimbabwean Respondents
However, unlike EAfrE, this is a six-vowel system. The results can be

summarized as follows:

/i/ bead, hid, here, [hue]
/e/ hair, [bird, name, made, bed]

/8/ bad,[bird, name, made, bed]
/a/ hard, pride, bout, mud

/o/ sod, nod , goat, boa

/u/ good, mood, tour, [hue]

73

5. 3. l Zimbabwean Male Respondents

3000 2000 26100 2490 2200 2000 1000 16100 1100 1200 1000 000

 

 

 

 

 

 

 

 

 

 

 

 

200 ‘
fl
mm _me
l 0 good
400- hid 0 mod 0‘
6 09h gear 1*
500_ 0" Sure % 001]
sod goat
0 € board
6004 :00 caught
E
700- Amud '17
' a pride
800-
Jard +
I Jones,33,F0rm01,Rurai,Zimbobwe

 

Figure l 1.1 - Jones’ Vowel System (a Zimbabwean Male)
Figure 11.1 illustrates a plot of the F l and F2 frequencies for the 22 tokens

produced by Jones-a 35 year old Zimbabwean male university student. In this chart ,

vowels contained in the tokens bead and hid are realized as [i]; year and hair as [e];

74

made, bet, name, bad , bird as [8]; hard, pride, bout, and mud as [a]; sod, open, goat,

caught, board, sure, and boy as [0]; while mood and good are realized as [u].

Ndanga’s vowel systems (Figure 11.2) shows minor variations from Jones’

(Figure 1 1.1). Inversely, Ndanga realizes the onset of the diphthong in name as [e], and

the vowel onset in hair as [8], unlike Jones who realize them as [a] and[e] respectively.

Furthermore, Ndanga realizes the vowel in caught as [a], while the other Jones realizes it
as [0]. These variations are evident among other respondents in our Zimbabwe sample.

Zimbabwean male respondents English vowel production may be summed up as
follows: .

/i/ bead , hid

/e/ year, name, hair

/8/ made, bed, bad, bird

/a/ hard, pride, bout, mud, [caught]

/0/ sod, open , goat, board, boy, sure, [caught]

/u/ good, mood

75

3000 28100 26100 2100 2200 2000 18100 16100 1100 1200 1000 800

 

 

 

 

 

 

 

 

 

 

 

 

 

 

200
300 7
u moo
400 1 0 9°“
hot;
500 - so sod ‘7' goat
V obet bird “board
6 hair sure
’7' open 0
600 8
‘ ‘0 made
I bad
[E L f M" Vcauqht
. ard
70°“ fiyprid'va mud
a

800 -
900-

1000-

] Ndang alarmaIIUrban. Zimbabwe |

 

Figure l 1.2 - Ndanga’s Vowel System (a Zimbabwean Male)

5. 3. 2 Zimbabwean Female Respondents

Turning our attention to vowel production among educated AfrE female speakers
from Zimbabwe, another peripheral vowel system is evident. Figures 11.3 and 11.4 are

representative charts of all individual speakers in a sample of eleven speakers.

76

 

3000 28100 26100 2100 2200 2000 18100 16100 1100 1200 1000 800

 

 

 

 

 

 

 

 

   

 

 
 

 

 

   

 

 

 

 

 

 

200
300-
0 good
O’hue “no
0" 0 tour II
400-- mood
. bou
. hair made '2)
50° 6° e bird saw eh“
nod . 0
500. had
I I hard
+'Ind
700 - mud
'wa
soo-
900~
1000‘
l Kapindarfilfiormallwbanthnbabwe I

Figure l 1.3 - Kapinda’s Vowel System ( a Zimbabwean Female)
Kapinda’s normalized F1 and F2 scores are presented in Figure 11.3. Like all

other Kapinda realizes the English vowels presented in word list form as follows: Vowels

contained in the tokens bead and hid and here are realized as [i]; made, hair, bed and
bird are produced as [e], although bird seems fairly central; hard, bad, buy, mud, and
bout are realized as [a]; while saw, nod, boy, and boat are pronounced as [o]; and good,

mood, tour, sure, and hue are produced as [u].

77

Tamika’s (Figure 11.4) vowel production is similar to Kapinda’s (Fig. 11.3)

except for bead which is realized as [e], and sure and tour, which are produced as [0].

Most female respondents realize tour as [0]. Again, bird is fairly central.

 

300 —

400-

500 —

600 —

700 -

800 -

900 -

1000—t

 

 

Figure

“SC 00 28100 26100 24:00 2200 2000 18.00 16100 1‘100 1200 1000 800

 

h . 11 9°“ 0 mood
re
. hid l “"9 0’ 0"

ma 9 e
% hair e El bird sunI 9
'b bead 0
9 tour

boy 2)“ ‘I boat

 

 

 

 

 

 

 

 

 

 

em «I
I fig bzd
(NM 1.0::
a

 

 

A mud

 

 

l Tamika,32,r,rormaI,UrbanTzimbabwe ]

l 1.4 - Tamika’s Vowel System ( a Zimbabwean Female)

78

Vowel production among Zimbabwean female respondents may be summarized
as follows:

/i/ hid, here,[ bead]

/e/ made, hair, bed, bird, [bead]
/a/ hard, bad, buy , mud, bout

/0/ saw, nod , boy, boat, tour ,sure

/u/ hue, good, mood,[sure, tour

A gender distinction similar to the one already identified in discussion of the
Kenyan sample is also evident among the Zimbabwean AfrE speakers. Male respondents

have a lax and tense distinction for the front mid vowel while female speakers do not.

Moreover, while male respondents front RP’s central vowel [3] as in bird to [8], and

raise the low back vowel [a] to [8], female respondents barely front [3] as in bird to [8],

but they do not raise [a]. It is worth emphasizing that gender distinction is not confined to
the Kenyan sample as the Zimbabwean data shows. Further research into the variation
should illuminate more characteristics of vowel variance within and across English

varieties spoken in sub-Saharan Africa.

It is worth mentioning at this juncture that the fronting of the central vowel [3]
to [8] seems to be the major variation between Bantu speakers of Eastern Africa (Kenya)
and Southern Africa (Zimbabwe). Moreover, [us] as in tour and sure is consistently

realized as [u] by Kenyan respondents, but some Zimbabweans realize the

monophthongized diphthong as either [0] or [u]. This finding casts doubt on earlier

claims that AfrE speakers consistently produced [09] as [u].

79

5. 4 Ghanaian Respondents

Figure 12 represents vowel frequencies of all the token produced by 20 Ghanaian

university students. This five-vowel system is peripheral and contains no central vowels.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F2
2003000 2W0 16100 2000 2200 2000 1§30 |§00 14100 12(1) 1900 89)
r , ‘

30°. hue ll 0 0’ mood

‘o bead J made good
400- gear 0 good

i ' 9 here 6 sure boat
5004 L o hid J ' 9 board

bog a saw
bird ‘9 b d ‘ but
0 e
600- II
ca ht
[E] bird “9
El
9 0 pride
700- k 9hair J i; 'b D - bout
(i' buy
300-
bad El
900- ‘ mud
'3' hard
‘9 J
L
l llll Ghanaians (20 M81?) |

 

 

Figure 12- Vowel Plot of All Ghanaian Respondents
The results may be summarized as follows:

/i/ hue, made, bead , hid , here

/e/ hair, bird, bed
/a/ hard, bad, buy, mud, [bout], pride

80

/0/ saw, nod, boy, tour, [boat], caught

/u/ good, mood, sure, [boat],[ bout]

5. 4. 1 Ghanaian Male Respondents
Figure 12.] illustrates a plot of the F 1 and F2 frequencies for the 22 tokens produced by

Kwame , a 27 year old male university student from Ghana. In this chart,

IE

 

   

 

3t00 2930 2600 2400 2200 200) 1800 1600 1400 1200 1000 800
200 J l 1 l l 1 l 1 1 l 1
300-
. bead ll 09°“
Ill) ohtd d‘mood
400-
goat open
Sure
500-
600-
700-
800-
Kasem,27,M,Formal,Urban,Ghana

 

 

 

Figure 12.1 — Kwame Vowel System ( a Ghanaian Male)

81

vowels contained in the tokens bead and hid are realized as [i]; year and name

 

 

 

 

 

 

   

   

     

 

    
 

 

 

 

 

3C00 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800
200 J l l l l l l 1 l l l
E
300 - . {
I 0 bid good u
bead R0 0 r? mood
400—
open 0
‘gbo
500 - , *
gear 9 ha" 0 E bird
bet
8
600-
700 -

800-

900-

1000-

 

 

 

 

| Twi,42,M,Formal,Urhan,Ghana |

Figure 12.2 - Kojo’s Vowel System ( a Ghanaian Male)

82

as [e]; and hair,, bet, made and bird as [8]; hard, pride, bout, bad and mud as [a]; sod,

caught and board are pronounced as [0]; open, goat, sure, and boy are realized as [0];

while mood and good are realized as [u].

In comparison to Kwame’s vowel production (Figure 12.1), Kojo’s system

(Figure 12.2) represents other individual systems in, the sample that vary from those that

resemble Kasem’s in their inverse realizations of made and year as either [e] or [8], and
goat and board as [0] or [o]. This calls into question, of course, the idea of a seven

(rather than five) vowel system. The clustering of [i, u, a] is also identical in a majority

of cases. AfrE vowel production among educated male Ghanaians may be summarized

as follows:

/i/ bead, hid

/e/ name, [year, made]

/8/ hair, bet, bird, [year, made]
/a/ hard, bad, pride, bout, mud
/o/ caught, sod , [board, goat]

/0/ open , boy, sure, [board, goat]

/u/ good, mood

5. 4. 2 Ghanaian Female Respondents
Let us examine AfrE vowel production among educated Ghanaian women, based
on a controlled sample of eleven Kwa language speakers. Figures 12.3 and 12.4 were

identified as representative Ghanaian speakers.

83

 

200

3000 28100 26100 2100 2200 2000 1000 MPG 1100 1200 1000 800

 

300 -

400-

500 -

600 -

700 -

800 -

900 -

1000-

 

 

 

 

 

 

 

 

 

 

’ 1
0 good
ll
‘7' boat
0 sure mood
L O, 1
nod
tour saw
r V
a bout
v M
I bad+ hard
A mud
x 1

 

 

 

 

I Twi,42,F,Formal,Urban,5hana j

Figure 12.3 - Akosua’s Vowel System ( a Ghanaian Female)

Akosua (Figure 12.3) realizes the English vowels presented in tokens shown in

the chart as follows: Vowels contained in the tokens hue and here are realized as [i];

bead , hid and made are produced as [e]; hair, bed and bird are realized as [a]; hard,

bad. buy, mud, and bout are realized as [a]; while saw, nod, boy and tour are pronounced

as [0], while vowels in good, mood, sure and boat are produced as [u].

84

In contrast to Akosua (Figure 12.3) Efua (Figure 12.4) produces vowel in hid and
head as /i/, bed as /e/ and boat as [o]. The differences across female speakers seem to be
in vowel height in such tokens as hid, bead, and here. There seem to be differences in

the tense/lax distinction of the vowel in bed. Some respondents pronounceboat with an

 

n3C00 2800 26'00 2400 2200 2000 1800 16IOO 14100 1200 1000 800

20..
300 -
400 -

500 —

 

600 -

   

700 -

 

800 -

{' bout

07 bug + hard
900 -‘ I had A mud

 

1000—

 

 

| Wasa;36,FnFormaIIUrbanIGhana |

Figure 12.4 - Efua’s Vowel System ( a Ghanaian Female)
[0] and some with an [u]. Such variations merit further detailed acoustic
measurements. A summary of Ghanaian female vowel production contained in charts

discussed above is as follows:

85

/i/ hue, [bead , hid , here]
lei made, [bead, hid, here], [bed]

/8/ hair, bird, [bed]
/a/ hard, bad, buy, mud, [bout]

/0/ saw, nod, boy, tour, [boat]

/u/ good, mood, sure, [boat], [bout]

An overview of Ghanaian sample shows that both male and female vowels

systems have a lax and tense front middle vowel, front RP’s central vowel [3] as in bird

to [a], but differ in the realization of [ua] as in sure. While male respondents realize the

monophthongized diphthong as [0], females pronounce it as [u].

5. 5 A Comparative Analysis of Acoustic Results

The primary aim of this study is to verify conclusions drawn in previous studies
pertaining to variation in AfrE vowel systems. Conclusions drawn in the current study are
based on empirical analysis of data collected from a relatively homogeneous population
of well-educated male and female AfrE speakers from Ghana, Kenya and Zimbabwe. Our
Ghanaian sample is made up of native speakers of Kwa languages, while Kenyan and
Zimbabwean AfrE speakers are native speakers of Bantu languages. All respondents
speak English as their second or even third language.

Let us compare results of our acoustic analyses presented in this chapter with
some primary results documented in earlier studies in the field. The results are

discussed in more detail in the literature review section ( 1.4.0.).

86

Schmied (1991) attributes AfrE variation to the ‘deviation’ in the production of

the RP long central vowel [3] (as in bird) by speakers of each regional speech

community. He concludes that [3] backs to [o] in WAfrE, fronts and lowers to [a] or

fronts to [e] in EAfrE, while in SAfrE it fronts to [e]. The two possibilities in EAfiE are
not elaborated. This variation is presented in Schmied's illustration as shown in Figure 1.
This primary hypothesis, which is widely echoed in the literature, is not

corroborated in the evidence from the acoustic analysis presented in this study. Ghanaian

and Zimbabwean speakers front the RP central vowel /3/ to/ 8/ while Kenyans lower it
to /a/. The central vowel //\l as in bud is lowered by Kenyans and Ghanaians speakers

while Zimbabweans front it to /e/.

Furthermore, evidence from our male and female respondents from Ghana clearly

shows that /I\/ is realized as [a] and not as [o], as Todd (1984) claims (see 2. 2. 2. 1).

Moreover, other previous claims not supported by our evidence from acoustic analyses

include Bokamba’s (1992:21) observation that //\/ as in bud is produced as [o], and [3]
as inbird is realized as [0] and Schmied’s (1991259) claim that all the centring

diphthongs (lial, /ee/, lua/ tend to be pronounced as opening diphthongs (lia, ea, ual).

Our analysis, however, shows variation in realization of lia/ and /ea/ as either /i/ or /e/
and /ua/ as /0/ or /u/, while /au/ is realized as /a/ or /o/. These variations occur in all three

varieties.

87

5. 6. l Kenyans
Acoustic analyses of Kenyans English vowel production generally support previous
characterizations of the EAfrE vowel system as shown in Figures 1 and 4.

However, contrary to earlier claims, acoustic evidence presented above shows that

some male respondents make a tense and lax distinction in the front mid vowel area.

5. 6. 2 Ghanaians

Acoustic analysis of vowel production among our Ghanaian respondents does not

concur with Todd’s claim that [A] backs to [o ] in WAfrE .

Trudgill and Hannah’s (1985) illustration (Figure 5), makes identical claims to
Schmied’s (Figure 1), and therefore disagrees with ours in the same way.

Sey (1973) compares RP and educated Ghanaian English (Figure 4) and claims

that the vowel /3/ as in bird is fronted and realized as /e/ contrary to Schmied's claims.

Sey claims that the vowel /A/ as in the word bud is realised as either /0/ or /8/.

The significance of Sey’s observations to this study lies in the fact that it is the only

study we know of in the previous literature that claims that RP’s central vowel /3/ is

fronted to [8] by WAfrE speakers. All other studies claim that the vowel is backed to [0].

Our analysis concurs with Sey’s observation. However, like other studies mentioned in

this study, Sey claims that educated Ghanaians realize /A/ as /0/. Sey goes a step further

and notes that /A/ is realized as /0/ or /8/. Empirical evidence from a diverse population

88

of male and female Kwa language speakers at the University of Ghana unequivocally

shows, however, that /l\/ is realized as [ 3].

Based on our analysis, it is clear that our sample of educated Ghanaian speakers

does not realize the vowel in bird as [0], but rather as [8]. Moreover, the RP central

vowel /A/ as in the word mud is not realized as [0], but as [a].

Therefore, that broad generalization from earlier research regarding the nature of

WAfrE vowel system does not hold, particularly among our Ghanaian male respondents

5. 6. 3 Zimbabweans

Based on our analysis, Zimbabwean speakers do not make a vowel quality
contrast; the only discernible lax/tense distinctions is among male respondents who make
a distinction between lax and tense front mid vowel. The vowels in the words
bet/bird/bad/made are realized as /e/. The vowels in hard/mud are realized as [a], while
those in board/sod/caught are realized as [o].

Suffice it to say, the phonological systems proposed in earlier studies fairly
represent the phonetic reality, but the problem lies in grouping.

In conclusion, the strength and significance of our findings lies in the evidence
we presented to contradict major claims presented in primary studies in the field.
Generalizations about the realignment of RP’s central vowels by diverse multilingual
AfrE speakers residing in broadly defined linguistic boundaries may be true in some
cases, but certainly not wholly representative of the homogeneous samples that we

analyzed.

89

This section has focused on variation in the production of AfrE vowels. Although
empirical methods of analysis have carefully been applied to ensure a precise
presentation of data, our study will be more meaningful if we correlate acoustic cues with
phonemic representations through perception tests. As Lieberman and Blumstein (1993:

170) argues:

Although it is possible to perform precise analyses of speech signals using electronic instruments
and computer programs that effect various mathematical transformations of signal, these analyses
are in themselves meaningless. we can never be certain that we have actually isolated the acoustic
cues that people use to transmit information to each other unless we run psychoacoustic studies in
which human listeners respond to acoustic signals that differ with respect to the acoustic cues that
we think are relevant.

This is the task we seek to accomplish in the following section.

5. 7 AfrE Listeners’ Vowel Perception
5. 7. 1 Data Analysis

Data from each region was calculated separately in order to determine how each
vowel was perceived in the IT and MPT. The former entailed calculation of correct and
incorrect identifications of every vowel by each respondent. After accounting for all
responses in each region, frequency scores were calculated to establish any trends or
patterns of identification. Similarly, MPT analysis sought to calculated and subsequently
identify any patterns in the correct and erroneous discrimination of the pairs.

Suffice it to say, the focus in the analysis of IT data is the onset (steady state)

characteristics of the vowels and not the word. For instance, if a stimulus pair bed/bad
was identified as bent/ bird, the first vowel in the pair was marked as correctly identified

in spite of the error in matching the word, while the error in identifying the vowel of the

90

second word was counted as such. Although the entire raw data is shown in Appendix E,
Table 1 is an example of such vowel perception in each region.

The first column contains the pairs of words that listeners heard on the tape, and
vowels tested. The other columns show the percentage of respondents who identified the
intended vowel, or an alternative one. The former is represented in the first line of each
cell, while any additional scores in each cell represent the latter.

For example, the first score box in the last column shows that 84 % of listeners
from the EAfrE speaking region correctly identified in their questionnaires that the first
word of the stimulus pair meat/meat contained the vowel [i]. 63% of those listeners also
identified the second word as containing [i]. The second line in the same box shows that
5 % and 21% of EAfrE listeners misidentified the first and second words respectively,
as containing [1] instead of [i]. Data in the table should be interpreted in that manner
unless otherwise stated.

Table 1- Identification Results of Vowel [i]

 

 

 

 

 

 

 

 

Stimulus Ghanaians Zimbabweans Kenyan respondents
minimal pairs respondents respondents perception of stimulus
words played to perception of stimulus perception of stimulus vowels (in %)
i ~ i i ~ i 94-94 i ~ i 98-100 i ~ i 84—63
meat/meat a ~ 6 06-06 I ~I 05-21
i~I i~I 68-48 i~I 41-24 i~I 75-21
beat/bit I ~ i 32-38 I ~ i 59-41 I ~ i 21- 28

e 26 e 49
i ~ 6 i ~ 6 64-43 i ~ 6 43-59 i ~ 6 57-14
beat/bait 1~A 31-17 I~ai 47-13 a~a 60-60

a 13

 

91

 

 

 

 

 

 

 

 

i ~ 8 i ~ 8 100-22 i ~ 8 98-67 i ~ 8 93-65
weed/wed 3 61 la ~O 02-19 ai ~ ai 07-19
i ~ ia i ~ ia 74-100 i ~ ia 79-76 i ~ ia 68-100
he/here ia l7 ia~i 21-24 ia 32

i ~ ju i ~ ju 85—73 i ~ ju 94-58 i ~ ju 82-69
he/hue ju ~ i 15-17 ia ~ i 06-39 ia ~ i 09-32

 

Having calculated the frequency scores for each vowel in each pair as shown
above, an effort was made to determine the overall perception of each vowel in each
region. Towards this end, mean values of identification responses to each vowel were
calculated by adding up frequency scores and dividing them by the total number of
responses. For instance, to determine the mean value [i], all frequency scores for

correct identification and misidentifications are added up separately and divided

appropriately.

5. 8 Identification Results and Discussion
Two types of charts are presented in the discussion of some vowels or groups of
vowels. The bar charts seeks to represent visually respondents’ identification of stimulus

vowel in percentages, while the vowel charts reflect the same information on a simulated

vowel space.

Identification of i

e oandu

92

 

Figure 13.1 shows that over 60% of respondents in each region correctly

identified vowels [i], [e], [o], and [u]. It is worth noting the high intelligibility score of

 

 

   

  
  

   
  
  

   

    

  
 

 
  
  

   
   
     

 
 

 

 

          

 

 

     

  
   

100 goo-“i -3.
do; 8 _, __
80; g sgggs NEE: —. as madam

J. 53555: :Nss; \g
285 '§ 3 fii§= ammo]

. u \.... \...: III\
50! I §sgga Kass; EN-
oof \ a as: " “11M
:2; ii % 50$- oooooo
fgiifl be gag

Gh Zim Ken

 

 

 

Figure 13.1 - AfrE Listeners’ Identification of [i], [o], [u], and [e]
middle vowels [e] and [o] in all three regions.

Earlier perception studies have concluded that listeners easily identify high
vowels [i] and [u], and in many cases the low vowel [a]. For instance, Lieberman and
Blumstein’s (1993) physiological theory for vowels (see section 1.3), argue that quantal
vowels [i] and [u] are easy to identify because the F1 and F2 of each of these vowels
converge to yield well-defined spectral peaks, making them maximally distinct.
Furthermore, Bladon and Lindblom (1981) report that experiments in vowel perception
have shown that listeners use spectral information to access vowel quality. They argue
that listener’s are capable of approximating the length of a speaker’s supralaryngeal vocal
based on spectral information, and to consequently normalize the vocal tract within the

parameters of an appropriate vowel space. The detailed mechanism of vowel perception

93

perception are not within the purview of this study, but it worth noting the high

intelligibility rate of these vowels may point towards a maximal distinctness of these

ull

1 bad u p

>70°/ >80%
0

e bade a bought

>80% >65%

Figure 13.1: AfrE Listeners’ Identification of [i], [o], [u], and [e]

vowels.

 

 

 

 

    

       
 
 
     

    
   
 

   

   

 

 
  

I

IIIIIIIL'

    

iflTear [ia]—
Iboy [ oi]
II sure [Ua]
1 fl hue [ju]

  

 

 

 

, .

.W mfiuuunuu. i
, I f f f I f I f l

g I r r r r r r“

iiiiiiiiiiiiihi

   

 

Figure 13.2 - AfrE Listeners’ Identification of [i], [o], [u], and [e]

94

However, this study shows middle vowels [e] and [o], and not the lower vowel

[a], as the other easily identified vowel, besides the documented [i] and [11]. Moreover,
Figure 14 - AfrE Listeners’ Identification of [ia], [oi], [ua] and [ju]

Over 50% of listeners correctly identified diphthongs containing vowels [ i], [ o] , and

[u] as Figure 14 shows. The data here, seems to corroborate earlier evidence in Figure

13.1 that the vowels [i], [o], and [u] are easily identified for AfrE listeners; but not [e

 

 

     

22:3

   

 

B’bad [a]

 
 

{liltzllilllbliliiElIE

 

 

a bud[A]

 

 

 

 

 

 

Figure 15.1 - AfrE Listeners’ Identification of [a]

Identification of [al

Figures 15 and 16 show that AfrE listeners often confuse [a] with [ A], and [A ] with
[a]-

Figure 15.1 shows that 55% of Ghanaian respondents correctly identified [a], but

45% misidentified the vowel as [A] . Eighty percent of Zimbabwean respondents

correctly identified [a], while 20% thought it was[A]. 55% of Kenyans had the correct

95

identification, but 35% erred. Figure 15.2 illustrate the vowel identification pattern in

Figure 15.1

45 %
A bud

55 %
a bad

 

(a) Ghanaians (b) Zimbabweans (C) Kenyans

Figure 15.2 - AfrE Listeners’ Identification of [a]
Identification of IA I

It is hardly surprising that IT results shown in Figure 16 are almost identical to the
ones in Figure 15. After all , these are results of an identical pair of words tested twice

with the only difference being in the ordering of the pair that is, in the previous results,

96

 

«3195' -

 

W, |
1 I: bud [A]

 

 

Ibad [a]

1 _

 

 

 

 

 

 

 

 

(c) Kenyans

(a) Ghanaians (b) Zimbabwans

Figure 16.2 -- AfrE Listeners’ Identification of [A]
the tested minimal pair was bad, bud , while in this case the listeners heard the word bud

first. In spite of that slight change identification of [A] as low as in the first pair.

97

Only 45%, 25%, and 35% of Ghanaians, Zimbabweans, and Kenyans respectively

correctly identify the vowel. 30%, 50%, and 40% of respondents, in the same order,
confuse [A] with [a].

Considering the peripheral nature of L1 vowel systems of the current sample, and
in particular the non-existence of central vowels, two plausible arguments can be made to

account for the misperception of the two vowels. First, [A] is confiised with [a] when
the native-language magnet “pulls” the non-native RP vowel [A] towards the native

phoneme prototype [a]. Secondly, language contact and exposure to RP in the course of

educational pursuit makes these university students more conscious of such a non-native

sound as [A] . It would be interesting to test whether respondents with a lower level of

education would exhibit similar vowel perceptions.
Identification of I3 I

A similar native language effect may be anticipated in the perception of RP’s

central vowel [3], another non-existent vowel in the Lls of the present sample.

However, an overview of the results presented in Figure 17 seem to invalidate such

expectations in view of the fact that a majority of Ghanaians and Zimbabweans seem to

98

 

 

1:1 bird[3]
El bed[e]
I bad [a]

 

 

 

1
I
I
I
I
I
I
II
VI
II
9-
VI
II
I
I
II
I
I
III
I
-.
r

 

 

 

 

 

Figure 17.1 - AfrE Listeners’ Identification of [3]

correctly identify the vowel. The bar chart shows that over 55% of respondents in west
and southern Africa correctly identify the vowel. 20-28% of respondents in these two

regions confuse the vowel with either [8 ]or [a]. However, upon closer scrutiny, it is
evident that respondents who confuse [3 ]with either [8 ] or [a], are almost as many as

those who correctly identify it, particularly among Zimbabweans. For instance, note that

while 52% of Zimbabweans correctly identify [3 ] , the total percentage of Zimbabweans
who confuse the vowel with [8] and [a] is 50%.

Kenyan respondents present particularly interesting results in this case. While

over 50% respondents confuse [3] with [a], no respondent confused [3] with [8].
Similar to the observation given above regarding the perception of [A], NLM

may account for the variants in this case, but the identification of the non-native vowel

99

[3] may be as a result of perceptual experience through language exposure, or through

6:“ bed 20%
013m
8:332

(081mm

 

(t) itenyans

(U) Zimbabweans

Figure 17.2 - AfrE Listeners’ Identification of [3]

an association of the sound with similar L1 allophonic variants. [A correlation of
production and perceptual results will verify these assumptions. Figure 17.2 shows the
strong identification of the vowel among Ghanaians and Zimbabweans, and lowering of

the vowel to /a/ by a majority of Kenyans.

11 I'll I' [[1]

’3“
.0

100

In another case of misperception, the RP vowel [I] is confused with [i] and [e],

 

 

a Bid [1]
I Bead [I]
ll Bade [e]

 

 

 

 

G1 Zim Ken

 

 

 

Figure 18.1 - AfrE Listeners’ Identification of [I]
more evidently by Kenyans and Zimbabweans than Ghanaians as Figure 18 shows.

Among Kenyan listeners, 38% correctly identify [I], 42% confuse it with [e],

e bade
10%

 

(a) Ginnaians (b) Zimbabwans

Figure 18.2- AfrE’s Listeners’ Identification of [I]
while 35% identify it as [i]. 45% of Zimbabwean listeners correctly identify [1], 45%

mistake it for [i], and 10% heard it as [e]. On the other hand, over 60% of Ghanaian

101

speakers easily identified the vowel, 30% confused it with [i], and no respondent

incorrectly identified [1] as [e], as respondents from other regions did.
Identification of ID |

In all three regions, a majority of listeners confuse RP vowel [D] with [o], as

illustrated in Figure 19.1. A low score of 36% (Ghanaians), 38% (Zimbabweans), and

38%(Kenyans) correctly identified [D], compared to 45%, 60%, and 42% (in the same
order as above) who confused the vowel with [ 9].

Considering results of some earlier perception tests conducted in other English-

 

 

 

 

 

 

   
   

 

 

 

 

 

 

 

 

 

 

 

 

 

l
l 60 g‘
|
j 50 g p
i 40 E “fl .. r-
; i i i i — 7
l 30 o 3:3: 3;; Ilbod [n]
l '3': 7 “a“ .
‘ 20 1 2:2: . , iii Iboard [o]
i o 3:3: egg
1 ': iaii iii
' i i i i
l . iii: ‘3'
I , 3:3: ° "‘5
i G1 Zim Ken
1
| a -L

 

Figure 19.1 - AfrE Listeners’ Identification of [D]

speaking settings, it may hardly be surprising to find such a confusion among AfrE
speakers. But Peterson and Barney (1953), Preston (1994) have reported such confusion

only minimally among American respondents. Lieberman and Blumstein ( 1988: 179)

102

have argued that the confusion of these two vowels even in languages or dialects where
the vowels are distinct in production, is due to close proximity of these vowels in the
vowel space. Such a confusion among our multilingual sample may be considered a case

of categorial perception of non-native sounds. The low percentage of respondents who

 

(a) Ghanaians (b) Zimbabweans (c) Kenyans

Figure 19.2 - AfrE Listeners’ Identification of [D]
correctly identify RP’s [D], and the relatively higher number of the highly recognized

quantal vowel [o], is discussed above.

103

Identification of | 8|

 

 

B bed [e]
1: Bad [a]

 

 

 

 

en Zim I Ken

 

i
l
l
l
l
l
l
l

—— —_l

 

Figure 20.1 - AfrE Listeners’ Identification of [8]

Figure 20 shows over 60% of respondents from eastern and southern Africa correctly

identify [8], while less than 30% of respondents in both regions confuse the vowel with
[3]. Less than 40% of the West African sample correctly identified [8], while 38%

confused it with [a]. [8] appears to be quite distinct among Kenyans and Zimbabweans.

104

   

(c) Kenyans

(b) Zimbabweans
(a) Ghanaians

Figure 20.2 — AfrE Listeners’ Identification of [8]

Identification of ai au and 00

Although a majority of Ghanaians confused [ai] with [a], and [au] with [a] as did

 

 

' I:lbuy[ai]

l[a]

 

 

 

 

 

 

 

 

Figure 21 — AfrE Listeners’ Identification of [ai]

the Zimbabweans, diphthong [0g] was, however, correctly identified by most of

105

 

 

..._ ...

 

 

Zim Gt Ken

 

I _

Figure 22 - AfrE Listeners’ Identification of [au]

 

Ghanaian respondents as shown in Figures 21, 22 and 23.

On the other hand, Kenyan respondents confused [ai] and [au] with [a], and a
majority mistook [ou] for [o] as shown in Figure 21 through Figure 23.

Schmied’s (1991:59) conclusion that all AfrE’s centring diphthongs tend to be

pronounced as opening diphthongs or double monophthongs (/ia,ua) (see 2.2.2) is not

 

EIboat __
[0U]
’10]

 

 

Zim Gh Ken

 

 

 

Figure 23 - AfrE’s Listeners’ Identification of [ou]

supported by these perceptual results.

106

Conclusion

In conclusion, vowel identification results from the MPT show that vowels [ i],

[e], [o] and [u] are reliably identified by all AfrE respondents, including diphthongs

containing these vowel onsets.

In general, all respondents had difficulties identifying the RP vowels [A], [3 ],
[D], and [I]. Regional variations are particularly evident in the perception of [3 ] , [I ] ,
[8]. For example, a majority of Kenyans identify [3] as [a], while among Ghanaians and
Zimbabweans, [3] is correctly identified by a majority of respondents. Likewise, many
Kenyans and Zimbabweans correctly identify [8], although a few confuse it with [a], but

among Ghanaians, only a small percentage of respondents identified RP vowel [8].

Other variations are discussed above, but it is worth noting that these vowels are non-
native to the Lls of the respondents. This is further proof that the generalized vowel
systems in earlier studies are not accurate representations of the systems of the present
sample.

Errors evident in the perception of non-native vowels may be attributed to the
native language magnetic effect postulated in the NLM theory. Social factors such as

ethnicity, level of education, and age, may also account for the variation. For instance,

results in the perception of RP vowels [A], [3] , [D], and [I], vowels that are generally

absent in vowel systems of Kwa and Bantu language are perceived variously in the tests.

107

Suffice it to say, IT results reflect acoustic analysis discussed in chapter 3. Note

for instance that RP vowel [3 ]as in bird is produced as [a] by Kenyan respondents, but

Zimbabweans and Ghanaians produce it as [e].

However, we should hasten to reiterate that level of education is an important
factor in the production and perception of an L2. A sample of well-educated respondents,
like the present sample, may produce and perceive L2 sounds differently than the less-
educated considering the inclination to RP exposure.

Results of MPT task, discussed below, will provide us with a more complete

picture of vowel perception.

5. 9 Results of the MPT

Respondents were asked to listen to fifty-eight minimal pairs and determine
whether the words were similar or not. Responses were marked on questionnaire (see
Appendix D). Results were tallied and percentages calculated to determine which pairs
were easily discriminated, and which ones were not (see the discrimination results in
Appendix B).

At this juncture, a highlight of some presuppositions of the Native Language
Magnet theory will help with the analysis. Liberman (1990) observes that sounds from
distinct categories are relatively easier to discriminate in MPT than non-distinct ones.
The latter may be perceived as one sound. Moreover, L1 vowel categories may interfere
with multilingual listeners’ ability to perceive certain phonetic distinctions of non-native

L2 sounds.

108

Consequently, it is assumed in the following analysis that AfrE listeners will have
difficulties discriminating vowels that are non-distinct and non-native to their L1. These
adult respondents, we presume, will use their already developed L1 vowel schema in
discriminating RP vowels present in the MPT. However, we hasten to add that these
university students have had ample exposure to RP, but needless to say, not as much as
exposure as to either AfrE or their respective Lls. This study focuses on vowels that were
difficult to discriminate.

Results of the MP task are illustrated in Figures 24, 25 and 26. Each figure shows

the vowel pairs erroneously identified by approximately 40-50% of respondents as

F2

 

 

 

 

 

 

 

Figure 24 - Ghanaians Discrimination of Minimal Pairs
“same”, and correctly identified as “different” by a similar percentage of respondents.
Shaded areas represent cases in which over 55% of respondents thought different vowel

pairs were the same.

109

Figure 24 shows that Ghanaians had difficulty discriminating pairs involving

vowel [A] and [3] (bud/bird); [A] and [a] (hut/hurt); [A] and [o](gut/got); [a]and [au]

(bat/bout); and [u] and [U] (wood/wooed)

Figure 25 shows that Kenyans, on the other hand, had difficulties discriminating

vowel pairs [e] and [8], (wade/wed); [3] and [A] (bird/bud); [A] and [a] (mud/mad); [A]
and [O] (gut/got); [a] and [au] (bat/bout); [O] and [o](col/caught), [au], [ua], and [on],

(got/goat, bought/bout, caught/cot, law/lure).

 

 

 

 

 

 

 

 

 

A A 4A F2 A A I
i
I
{—5
‘ u:
e i 3 r0
2:
. A
IE] 0 a
‘ I

 

 

 

 

 

Figure 25 - Kenyans Discrimination of Minimal Pairs

Lastly, as shown in Figure 26, Zimbabweans had difficulties with pairs

containing vowels [3] and [A] [bird/bud]; [A] and[a] (mad/mud); [o], (mud/mode); [e]
and [8], (wade/wed); [a], [o],[au], [ua], and [ou], (hot/hot, gut/got, got/goat, cot/caught,

call/call, bought/bout, law/lure).

llO

Beyond the apparent obscurity of discrimination results, there appears a clear

pattern of perceptions which mirrors identification results to some degree. Consider for

instance the recurrence of [a] and [o] in the lists of confused pairs in all regions, and

recall the misidentification of these two vowels in the IT task.

F2

 

 

 

 

 

 

 

 

fill

 

Figure 26 - Zimbabweans Discrimination of Minimal Pairs

Furthermore, RP central vowels [3] and [A] are also prominent

recurrences in various pair combinations. The fact that minimal pairs containing these
vowels (bird/bud, hut/hurt) are confused by many respondents in all regions, and taking
cognizant of results of identification task and acoustic analysis, reaffirm that AfrE
speakers and listeners realign RP central vowels in accordance with their respective L1
systems. The Westermann and Ward illustration of African Lls (see Appendix A’)

affirms these observation.

lll

Chapter 6: Conclusions

6. 1 Introduction

This study has utilized empirical tools of investigation and analysis to identify
phonetic and phonemic qualities of AfrE vowel systems. Acoustic analysis has been done
to determine speakers’ production of AfrE vowels. Speech perception tests have been
conducted to determine listeners’ abstract realization of those vowels. Several theoretical
assumptions guided us in the present analysis of the data.

This concluding chapter seeks to tease out a probably phonemic interpretation of
these vowel systems by correlating vowel production and perception results presented in
the previous chapters. Although it is difficult to state categorically that the results
represent the actual phonemes of the vowel systems of a given multilingual region,
acoustic and perceptual analysis indicate, among other things, a variance in vowel
production and perception across regions and ethnic groups studied. The results
contradict some of the regional variations widely documented in previous studies. For
example, Schmied (1991) and other scholars, have consistently argued that the major

source of AfrE variation is due to regional variance in the production of the RP central

vowel [3] (as in bird). As stated earlier, these scholars claims that [:3] backs to [0] in

WAfrE, fronts and lowers to [a] or fronts to [e] in EAfrE, while in SAfrE it fronts to [e].
As discussed in the following sections, data from this sample does not corroborate this
theoretical premise, especially in regard to what the acoustic and perceptual realizations
of the central vowels are in each region. Furthermore the identical vowel production
among eastern and southern Africa speakers, that is claimed to constitute Bantu English,

is also not corroborated .

112

This study also found a distinction between male and female vowel system,
particularly in the acoustic analysis. Such a distinction merits further investigation with a
view to understanding vowel variance within and across social groups in each region. For
instance, what are the social factors that account for these gender distinctions? Is the
distinction indicative of any changes in progress in the vowel production of the varieties
we examined? And what is direction of the change, if any?

On the theoretical front, the current study has corroborated assumptions that there
indeed is an underlying AfrE vowel system, as well as distinct regional variations.
Furthermore, comparing Westermann and Ward’s illustration of L1 vowel system and
the results derived in this study, it is apparent that the tenets of the Native Language
Magnet theory apply in our respondents’ perception of non-native vowels.

Let us attempt to correlate production and perception results, and determine how
we reached these conclusions.

6. 2 Correlating Acoustic and Perceptual Results

Vowel production and perceptual data has unequivocally shown that vowels / il,
/o/, and /u/ and diphthongs that contain similar onsets are contrastive, easily identified
and discriminated without difficulties across the board. AfrE speakers and listeners
realized these phones as allophonic variants; [i]/[I], [u]/[U], [Al/[a] and [D]/[o].
Applying Lieberman’s physiological theory for vowels, we can assume that these vowels
have well defined spectral properties, and are therefore, AfrE’s quantal vowels which

define the vowel space within which AfrE speakers and listeners (represented in this

study) differentiate other English vowels. Moreover, based on principles of Native

113

Language Magnet theory, we can further argue that these and other identified regional
phonemes (discussed below), are in effect influenced by the L1 of these multilingual

speakers. However, we should hasten to add regional differences are evident in the

production and/or perception of [3 ], [I], [e], [8] and other diphthongs .

In the following sections, we will seek to correlate production and perception

results by restating the main results and subsequently discussing them.

Ghanaians

Although there were gender variations in vowel quality, vowel production among

Ghanaians was such that vowels in tokens bead , hid, hue, and here were realized as [i];

year and hair as [e]; and bed, name, bird as [8]; hard, pride, bout, bad and mud as [a];
sad, caught and board are pronounced as [a]; open, goat, sure, and boy were realized as

[0]; while mood and good were realized as [u]. The only apparent gender distinction in

vowel production among our sample lies in the realization of [ua] as in sure. While male
respondents realize the monophthongized diphthong as [0], females pronounce it as [u].

Speech perception tests reflect production results, considering in particular the

high identification rate of vowels [i], [o], and [u], and diphthongs that contain similar

onsets. For example, in the acoustic analysis, F1 and F2 scores of diphthongs contained

in hue, year load in the phonetic space of /i/.

As stated above allophonic variants [a]/[A] and [a]/[o] pairs are not clearly

identified, and are not easily discriminated, because, as the production results indicate,

they are not distinct vowels. [A]is produced as [a] while [a] is produced as [o]. The first

114

mentioned change is particularly interesting considering that earlier studies claimed that

WAfrE speakers realized RP’s [A] is as [a].

RP central vowel [3], is produced as [8], identified as such by 23% of
respondents, perceive as [a] by 20%, and difficulties were evident in discriminating
[3]/[A] pair. We have already established that [A] is indeed realized as [a], so in the last

mentioned pair it should be read as such. Again, what is particularly interesting in this

finding is that, production and perceptual evidence does not corroborate widely

documented claim that WAfrE respondents back the central vowel to [0], at least not

with this sample.

Based on production and perceptual evidence, RP / 3/ may be considered an

allophone of /8/ among our Ghanaian respondents.

Monophthongized lai/ and /au/ are allophones of /a/.

Zimbabweans

Zimbabweans produce vowels contained in the tokens bead, hid as [i]; year and

hair as [e]; made, bet, name, bad , bird as [8]; hard, pride, bout, and mud as [a]; sod,
open, goat, caught, board, sure, and boy as [0]; while mood and good are realized as

[u]. Male respondents have a lax and tense distinction for the front mid vowel while

female speakers do not. Moreover, while male respondents front RP’s central vowel [3]
as in bird to [8], and raise the low back vowel [a] to [a], female respondents barely front

[3] as in bird to [8], but they do not raise [a].

115

Results of the identification task indicate that an equal number of Zimbabwean
listeners confused [c] with [ei]. Moreover, 66% of the respondents perceived made/made
pair as different words in the discrimination task. Gender distinction is also evident in

male respondents’ lax/tense distinction of the front mid vowel. Moreover, while male
respondents front RP’s central vowel [3] as in bird to [e], and raise low back vowel [a]
to [a], female respondents only front [3] as in bird to [8], but they do not raise [a]. The

distinction may account for the identification results which shows that 30% of
respondents perceive [3] as [ a] and 20% as [e]. The variation is also reflected in the
difficulties in discriminating bat/bat and bird/bud pairs. Previous studies have
categorically stated that SAfrE speakers front [3] to [e], and [e] is realized as [e]. It is

plausible that le/ has many variants in the English variety spoken by our Zimbabwean
sample, including /e/, /8/, and /ea/.
Monophthongization of [ai] and [au] to [a], and [ou] to [o], is evident in

production and discrimination results.

Kenxans

The Kenyan sample produced vowels contained in the tokens bead and hid as [i];

vowels in the words made and bet as [e]; year, hair, and name , as [8]; bad, hard, pride,
bout, bird and mud as [a]; sad, open, goat, caught, board and boy as [0]; while mood ,
good and sure are realized as [u]. It should be noted that while males make [e] — [8]

distinction, females do not. The only other significant variation is that while male

respondents realize [19] as /e/, female speakers produce it as /i/.

116

The production of [3] as [a] is reflected in identification task where 52% of

respondents mistake [3] for [a], a variation quite distinct from Ghanaians’ and

Zimbaweans’. Moreover, unlike in the other two varieties, no Kenyan respondent

identified [3] as [8]. /3/ in Kenyan English is an allophone of /a/, and not /e/ as the case

among Zimbabweans and Ghanaians.

It is also interesting to note that, unlike identification responses from the other

regions, over 63% of Kenyan respondents perceive RP [8] as [a]. The distinctions

between varieties spoken by Zimbabweans and Kenyans are worth noting considering

earlier claims that the varieties are almost identical except for variation in the realization

of [3]. Is it possible then to argue that /8/ is an allophone of la]?

Monophthongization of [ai] and [au] to [a], and [ou] to [o], is reflected in
production and discrimination results.

We should hasten to add, these results do not represent the vowel systems of
speakers of national or regional varieties of AfrE. Our sample was drawn from well-
educated, Kwa and Bantu language speakers, and can only be considered to represent that
particular sample. Whether the results represent the broader population of Kwa and Bantu
speakers, or national varieties, will be determined after similar studies have been
conducted covering other social, ethnic and language omitted in this study. Future
research should utilize empirical tools of research adapted in this study to determine AfrE
vowel production and perception among respondents of different ethnic, social and

linguistic backgrounds.

ll7

6. 3 Theoretical Assumptions Revisited
Based on our empirical findings, several observations can be made regarding
theoretical assumptions about AfrE, speech production and perception, and about the

Native Language Magnet (NLM) theory.

(a) Assumptions about AfrE

In this study African English was regarded as a very general term that refers to
divergent social, regional and ethnic sub-varieties of English in East, West and Southern
Africa. The regional varieties of AfrE (EAfrE, WAfrE, SAfrE) were also perceived in
very general terms as diverse sub-varieties that have distinct regional characteristics, as
Mutonya’s ( 1997) attitudinal study of AfrE varieties showed.

Results of acoustic measurements and perception tests discussed in this study,
show that although each regional vowel system has its own distinctive (marked)
elements, there are some basic similarities (unmarked) in all regional vowel systems that
presuppose an underlying AfrE vowel system.

The marked elements account for the individual Kenyan-Bantu, Zimbabwean-
Bantu, and Ghanaian-Kwa (Akan) vowel systems, while the unmarked ones represent the
underlying AfrE vowel system. For instance, the acoustic analyses shows that Ghanaian

and Zimbabwean speakers of English front the RP vowel [3 ] to [e] , while Kenyan
speakers lower it to [a]. Furthermore, the central vowel [A ] as in bud is lowered by

Kenyans and Ghanaian speakers, but Zimbabwean speakers realize it as [e]. These are

regional variations.

118

However, in spite of such regional variations , it is evident that all speakers have

a peripheral vowel system that consists of /i/, /e/ (and /8/ in some instances), /a/, /o/ and

/u/, similar to Westermann and Ward’s illustration of African L1 vowels. RP’s central

vowels /3/ and /A / are lowered or fronted, while RP vowel contrast is lost in /i/-/I/, /a/-
/A/, /a/-/a/, /:>/ - /o/, and /u/-/U/. These are general characteristics that transcend all three

samples in this study.
Similarly, vowel perception results in this study reflect both general and region

specific traits. For example, in the IT task all listeners confused words containing the

vowels /3, A, o, 8 , c1 /. The misidentification pattern is almost identical in all regions,

except for the specifics of whether a particular vowel was strongly or weakly. Consider

for instance that while RP /3/ is also misidentified as /e/ and /a/ in all regions, 58% of

Ghanaians and 55% of Zimbabweans positively identify the vowel, while only 32% of
Kenyan identified it correctly. 55% of Kenyans misidentified the vowel as /a/ while only
30% and 18% of Zimbabweans and Ghanaians respectively, identified it as such.

In order to avoid methodological shortcomings in analyzing AfrE varieties, use
of controlled samples and empirical tools of research is the best way of determining the

regional characteristics of AfrE .
(b) Assumptions Concerning Speech Production and Perception

Lieberman (1990) reports that experiments point to the fact that listeners perceive

speech categorically; that is , listeners’ ability to discriminate sounds from two different

119

categories, such as /i/ and /e/, should be relatively easy, but discriminating two different
tokens of /i/ from the same speaker should be difficult.

The findings in this study support the assumption that vowels are perceived
categorically, considering that results of the MPT task reflect difficulties in

discrimination of non-distinct sounds. For example, Kenyan listeners had difficulties

discriminating sounds that are non-distinct in Bantu languages [A] / [ 3 ], [A] / [a],
[A] / [Ct ] , and [a] / [ au] . Acoustic analysis suggests the Kenyan sample realize all

these phones as variants of /a/. Furthermore [e]/ [ 8] and [u]/[U] are also difficult to

discriminate because, as the acoustic analysis points out, they are variants, and
presumably allophones of /e/ and /u/ respectively. Similar examples from other
respondents can be cited. The non-distinct vowels that are difficult to discriminate in the
perception tests are also non-distinct vowels in acoustic analysis.

Correlating speech production and perception is based on the consideration that
every speaker is simultaneously a listener, and every listener is at least potentially a
speaken

Suffice it to say, the scope and design of this study does not provide us the
evidence to verify the main assumptions of the Motor Theory of Speech Perception: that
listeners’ interpretation of acoustic signals are guided by articulatory gestures . Similarly,
we cannot empirically determine whether listeners use vocal tract normalization during
vowel perception, although categorical perception of speech, discussed above, points

towards that direction.

120

(c) Assumptions about N on-native Speech Perception

Recall that the NLM theory postulates that adult listeners perceive non-native
sound contrasts based on a schema developed from their native language. The native
phonological system of the listener presumably plays a prominent role in perception of
non-native sound contrasts.

Several factors make it difficult for us to ascertain whether listeners’ L1 vowel
systems affect their perception of RP vowels. The lack of empirically defined vowel
systems of Kwa and Bantu languages denies us reliable comparative evidence. However,
based on widely documented information, including Westermann and Wards’ study (
discussed in Chapter 3), Bantu and Akan vowel systems are peripheral and symmetrical.
The L1 systems have five to seven vowels and no central vowels (Welmerszl988). Such
characteristics are reflected in the perception of RP vowels in this study. Furthermore,
results of the perceptual tests indicating RP’s contrastive vowels are perceived as non-
contrastive may be indicative of the L1 effect. The reduction in number of the RP vowels
in production and perception by these AfrE respondents may point towards the same
conclusion.

However, although it is logical to assume an L1 influence in production and
perception of non-native sounds, considering that adult speech perception is organized to
process L1 with the least effort and greatest efficiency, it is equally appropriate to argue
from a sociolinguistics standpoint that social and other linguistic factors may play
significant roles in speech perception besides L1 effect. For example, in multilingual
sub-Saharan Africa, level of education is closely related to degree of exposure and usage

of English, just as much as social status is.

121

Evaluating speech perception of multilingual respondents in sub-saharan Africa,
leaves us wondering how factors such as language contact, listener’s exposure and
attitude to RP, level of education, multi-lingualism, influence respondents’ perception
besides the L1 effect.

Since most of our respondents are fluent in other local languages, It would be
interesting to determine whether the indigenous L25 and L3s of these multilingual
listeners affect their perceptions as much as L 1 does. Would a similar sample, with a
relatively lower level of education have similar vowel perception? How about a
respondent who has a lower level of education but a higher degree of exposure and
interaction with RP speakers? Would a respondent who has a strong negative attitude
towards English perceive RP vowels any different from his counterpart who is positively
inclined towards English? Only further research can help us answer these and other
lingering questions .

Finally, this study has shown that the maximally distinct AfrE vowels are [i], [e],
[o] and [u]. Studies among English monolinguals had identified [i], [11] ad [a] as the
quantal vowels.

Future studies should build on the foundations of this research in the seeking to
determine whether empirical evidence supports the broad definitions of varieties of
African Englishes. Furthermore, the tools of investigation and analysis used in this study

can be utilized in the empirical research of African languages.

122

 

APPENDICES

123

APPENDIX A’

124

APPENDIX A’

(i): Cardinal Vowel Chart

 

 

 

 

 

 

 

Front central back
(1 i (8) U
(2) (7)0
e
(3) 8 r (6) o
(4) a. ‘* (5) 0

APPENDIX A’(ii): Westerman’s Illustration of RP vowels

Front 0 entral back

 

i u

 

e I \Ejfwuo

a D

 

 

 

125

APPENDIX A

TABLE 1: A PROFILE OF RESPONDENTS

Ll

wendwa

utia
ukami
an'

ukusu

eru
aita
an'iru

ones hona
Nenziwe hona
.Nenziwe hona
debele
afumo '

akawira hona

debele
ovele hona
hona

ukulu

hona
debele
debele

 

126

 

127

APPENDIX B

128

APPENDIX Bl

AN EXAMPLE OF A N ORMALIZED DATA FILE

Maina,20,M,Formal,Urban,
Kenya

22,6.677709
314,2289,,1.l42,1,hid
283,2207,,1 1.1421,1,bead
453,2333,,l4.00091,1,year
415,1812,,2.1411,1,bet
456,1750,,21.1422,1,made
493,2094,,21.4124,1,name
565,1731,,24.6,1,hair
663,1480,,3.1421,1,bad
683,1273,,42.1411,1,bout
678,1563,.44.6,1,hard
655,1716,,47.1428,1,pride
534,1114,,5.1423,1,sod
469,996,,53.1416,1,caught
670,1508,,6.1422,1,mud
547,1008,,61.0001,1,boy
472,915,,62. 1416,1,goat
503,996,,63.1 1 101,1,open
470,913 ,,64.6001,l,board
297,865,,7.1426,1,good

3 17,823,,72.1422,1,mood
426,915,,74.6003,1,sure
670,1334,,94.6001,1,bird

129

Ghanaians

All Ghanaians (20 M&F)
473
372,2073,,1.l42,1,hid
380,2003,,ll.l421,l,bead
387.1988,,14.00091,l,year
640,1952,,2.14ll,l,bet
566,1877,,21.1422,1,made
439,2152,,21.4124,1,name
506,1930,,24.6,1,hair
841,1564,,3.l421,l,bad
640,1449,,42.14ll,l,bout
789,119l,,44.6,1,hard
655,1580,,47.l428,1,pride
500,1213,,5.l423,l,sod
633,946,,53.l416,1,caught
767,1266,,6.l422,l,mud
484,968,,61.0001,l,boy
365,1058,,62.1416,1,goat
439,1024,,63.11101,l,open
528,789,,64.6001,1,board
314,1058,,7.l426,l,good
305,805,,72.1422,l,mood
455,1275,,74.6003,1,sure
603,l787,,94.6001,l,bird
478,2894,,l.142,l,hid
494,2694,,ll.l421,1,bead
424,2825,,l4.6,l,here
670,2331,,2.1421,1,bed
463,2725,,21.1422,l,made
617,2523,,24.6,1,hair
810,1173,,3.1421,1,bad
787,1335,,4l.0001,1,buy
525,941,,42.14ll,1,bout
988,1282,,44.6,l,hard
701,1223,,5.l424,l,nod
887,1319,,6.1422,1,mud
648,1103,,61.0001,1,boy
470,1065,,62.14ll,1,boat
602,104l,,63.0003,1,saw
455,1046,,7.1426,1,good
378,1073,,72.1422,l,mood
563,1266,,74.6003,l,sure
578,1119,,74.6003,l,tour
463,2040,,82.,l,hue
663,1976,,94.6001,l,bird
309,2724,,1.142,l,hid
426,2607,,1l.1421,1,bead
382,2430,,14.6,1,here
603,2407,,2.1421,1,bed
463,2547,,21.1422,l,made
662.2407,,24.6,l,hair
832,1077,,3.l42l,1,bad
721,1760,,4l.0001,l,buy

APPENDIX B2

ALL Fl/FZ SCORES

758,1641,,42.1411,1,bout
964,1561,,44.6,1,hard
730,967,,5.1424,1,nod
880,1308,,6.1422,l,mud
603,949,,61.0001,1,boy
522,1082,,62.1411,1,b0at
603,1178,,63.0003,1,saw
434,1031,,7.1426,1,good
476,995,,72.l422,1,mood
453.1207,,74.6003,1,sure
625,1126,,74.6003,1,tour
471,2229,,82.,1,hue
546,2371,,94.6001,l,bird
413.2797,,l.l42,l,hid
429,2859,,11.1421,1,bead
444,2629,,l4.6,l,here
581.2452,,2.1421,1,bed
406,2598,,21.1422,1,made
574,2537,,24.6,1,hair
950,1188,,3.1421,1,bad
864,1141,,41.0001,1,buy
886,1177,,42.1411,1,bout
1019,1547,,44.6,1,hard
602,1072,,5.1424,1,nod
851,1238,,6.1422,1,mud
620,1019,,61.0001,1yboy
528,989,,62.l4ll,1,boat
574,1088,,63.0003,1,saw
421,llll,,7.1426,l,good
368,1088,,72.l422,1,mood
505.1402,,74.6003,1,sure
543,1004,,74.6003,l,tour
375,2721,,82.,1,hue
628,2414,,94.6001,l,bird
515,2529,,1.l42,1,hid
393,2668,,1l.l421,1,bead
422,2885,,14.6,1,here
S97,2055,,2.l421,l,bed
437,2572,,21.1422,1,made
611,2368,,24.6,l,hair
802,1683,,3.1421,1,bad
918,1522,,41.0001,1,buy
757,1174,,42.14ll,l,bout
779,1261,,44.6,1,hard
743,1166,,5.1424,1,nod
765,1347,,6.1422,1,mud

130

611,1014,,6l.0001,l,boy
437,815,,62.14ll,l,boat
648.1137,,63.0003,1,saw
415,1003,,7.1426,1,good
453,966,,72.l422,l,mood
435,1294,,74.6003,l,sure
685,1075,,74.6003,1,tour
459,2471,,82.,1,hue
662,2318,,94.6001,1,bird
468,2352,,1.l42,1,hid
389,2583,,11.1421,1,bead
447,2294,,14.6,1,here
592,2265,,2.l421,l,bed
432,2425,,21.1422,1,made
649,2234,,24.6,l,hair
715,1060,,3.l421,1,bad
772,1082,,4l.0001,1,buy
830,1414,,42.14ll,1,bout
959,1443,,44.6,1,hard
728,1046,,5.1424,l,nod
894,1522,,6.1422,1,mud
605,880,,61.0001,1,boy
541,1002,,62.l411,1,boat
618,1024,,63.0003,1,saw
455,902,,7.l426,1,good
411,959,,72.l422,1,mood
560,1215,,74.6003,1,sure
605,1103,,74.6003,1,tour
426,2388,,82.,l,hue
605.2814,,94.6001,1,bird
435,2716,,1.142,1,hid
419,2760,,11.1421,1,bead
4SO,2851.,14.6,1,here
472,2316,,2.1421,1,bed
442,2950,,21.l422,1,made
711,2507,.24.6,1,hair
789,1180,,3.1421,1,bad
840,1180,,41.0001,1,buy
738,1335,,42.14ll,1,bout
862,1220,,44.6,l,hard
759,1141,,5.1424,1,nod
83l,1318,,6.1422,l,mud
701,106l,,6l.0001,1,boy
487,893,,62.l4ll,1,boat
612,1055,,63.0003,1,saw
408,974,,7.1426,l,good
465,975,,72.1422,l,mood
465.1299,,74.6003,1,sure
671,1055,,74.6003,l,tour
450,2583,,82.,1,hue
585,2359,,94.6001,l,bird
440,276l,.l.l42,1,hid
360,2841,,11.l421,1,bead
440.2885,,14.6,1,here
639,2452,,2.1421,1,bed
418,2836,,21.1422,l,made
647,2547,,24.6,1,hair
821,1123,,3.1421,1,bad

793,1123,,4l.0001,1,buy
785,1212,,42.l411,1,bout
895,1186,,44.6,l,hard
778.1167,,5.1424,1,nod
815,1182,,6.l422,1,mud
675.1087,,61.0001,1,boy
417,829,,62.14ll,1,boat
698,1123,,63.0003,l,saw
410,1020,,7.1426,1,good
46S,969,,72.1422,1,mood
498,1301,,74.6003,1,sure
690,1064,,74.6003,1,tour
359,2724,,82.,1,hue
682,2496,,94.6001,l,bird
493.2507,,1.142,l,hid
440,2772,,11.1421,1,bead
479,2203,,14.6,l,here
585.2187,,2.1421,1,bed
511,2630,,21.1422,1,made
673,2126,,24.6,1,hair
805,1192,,3.1421,1,bad
790,1132,,4l.0001,1,buy
766.1557,,42.l411,1,bout
896,1298,,44.6,l,hard
736,1101,,5.1424,1,nod
1010,1336,,6.l422,1,mud
587,896,,61.0001,1,boy
531.1086,,62.l4ll,l,boat
576,1033,,63.0003,1,saw
440,934,,7.l426,1,good
450,995,,72.1422,1,mood
S31,1314,,74.6003,1,sure
S42,1086,,74.6003,1,tour
448,2423,,82.,l,hue
622,2218,,94.6001,1,bird
453,2500,,1.142,1,hid
422,2832,,11.l421,1,bead
392,2569,,14.6,1,here
566,2288,,2.l421,1,bed
446,2508,,21.l422,1,made
616,2167,,24.6,1,hair
860,1812,,3.l421,l,bad
84S,1631,,41.0001,1,buy
845,1623,,42.l411,1,bout
875,1683,,44.6,1,hard
641,1004,,5.1424,1,nod
896,1563,,6.1422,1,mud
634,955,,61.0001,1,boy
430,951,,62.l4ll,l,boat
649,1004,,63.0003,1,saw
378,831,,7.1426,l,good
468,997,,72.l422,1,mood
453,1254,,74.6003,1,sure
649,1012,,74.6003,1,tour
354,2483,,82.,1,hue
581,2251,,94.6001,1,bird
424,2588,,1.l42,1,hid
427,2743,,11.1421,1,bead

440,2498,,l4.6,l,here
537,2386,,2.l421,1,bed
533,2559,,21.l422,l,made
66l,2153,,24.6,l,hair
912,1794,,3.l421,l,bad
87S,l690,,41.0001,1,buy
823,1457,,42.1411,l,b0ut
882,1129,,44.6,1,hard
508,935,,5.1424,1,n0d
905,124l,,6.1422,1,mud
569.1158,,6l.0001,1,b0y
503,965,,62.1411,l,boat

621,2421,,24.6,1,hair
953,1900,,3.1421,1,bad
873,1430,,41.0001,1,buy
408,983,,42.14ll.l,bout
907,1188,,44.6,1,hard
741,1173,,5.l424,l,nod
787,1127,,6.l422,l,mud
665,1029,,61.0001,1,boy
438,840,,62.l4ll,1,boat
711,1152,,63.0003,l,saw
446,1036,,7.l426,1,good
4S3,975,,72.l422,1,mood
453,1270,,74.6003,l,sure
680,1180,,74.6003,1,tour
370,2784,,82.,1,hue
589,2406,,94.6001,1,bird
354,226l,,1.142,1,hid
325,2255,,ll.l421,1,bead
517.2329,,14.0009l,l,year
517,1804.,2.l411,1,bet
458,2114,,21.l422,l,made
445.2047,,2l.4124,l,name
509,1942,,24.6,1,hair
739,1485,,3.1421,1,bad
694,116l,,42.14ll,1,bout
658,1471,,44.6,1,hard
576.1457,,47.l428,l,pride
531,1042,,5.1423,1,sod
557,1094,,53.l416,l,caught
820,1434,,6.1422,1,mud
463,968,,61.0001,1,boy
481,1064,,62.14l6,l,goat
435,858,,63.11101,1,open
495.1124,,64.600l,l,board
458,880,,7.1426,1,good
413,805,,72.l422,l,mood
443,1086,,74.6003,1,sure
Sl7,l779,,94.6001,l,bird
326,2193,,l.l42,l,hid
288,2382,,11.l421,l,bead
531.2028,,14.00091,1,year
531,1755,,2.14ll,l,bet
303,2481,,21.l422,1,made
4S7,2192,,21.4124,1,name
523,1966,,24.6,1,hair

647,109S,,63.0003,1,saw
395,1054,,7.l426,1,good
463,1002,,72.1422,1,mood
440,1353,,74.6003,1,sure
530,1046,,74.6003,l,tour
372,2398,,82.,1,hue
624,2393,,94.6001,1,bird
423,2883,,1.l42,l,hid
370,2989,,11.l421,1,bead
431,293S,,14.6,1,here
560,2520,,2.1421,1,bed
438,2921,,21.1422,1,made

614,1834,,3.l421,l,bad
713,1273,,42.14ll.1,bout
773,1404,,44.6,1,hard
713.1381,,47.1428,1,pride
508,956,,5.1423,1,sod
577,948,,53.1416,1,caught
827,1366,,6.1422,1,mud
493,986,,61.0001,1,boy
493,1078,,62.14l6,1,goat
429,836,,63.11101,1,open
552.1292,,64.6001,1,b0ard
448,971,,7.1426,1,good
351,873,,72.l422,1,mood
475,1079,,74.6003,l,sure
531,1548,,94.6001,1,bird
37l,2108,,1.142,1,hid
386,2282,,11.1421,l,bead
393,2375,,14.0009l,1,year
454,1835,,2.1411,1,bet
364,2214,,21.1422,1,made
447,2101,,21.4124,1,name
515.1847,,24.6,1,hair
720,1326,,3.1421,l,bad
659,1510,,42.1411,1,bout
811,1394,,44.6,l,hard
690.1342,,47.1428,l,pride
478,1213,,5.1423,l,sod
546.1206,,53.l416,1,caught
819,1228,,6.1422,1,mud
500,1008,,61.0001,l,boy
436.1067,,62.1416,1,goat

132

409,909,,63.11101,l,open
470,894,,64.6001,1,board
350,1106,,7.1426,1,good
345,925,,72.1422,1,mood
453,1308,,74.6003,l,sure
530.1888,,94.6001.1,bird
429,223l,,l.142,1,hid
370,2105,,11.1421,l,bead
416.2194,,14.00091,1,year
497,183l,,2.1411,1,bet
422,2105,,21.1422,1,made
441,2148,,21.4124,1,name
506,2001,,24.6,1,hair
734,1393,,3.1421,1,bad
667.1527,,42.14ll,1,bout
768,1400,,44.6,l,hard
655,1528,,47.1428,l,pride
555,1118,,5.1423,1,sod
607.1141,,53.1416,1,caught
711,1149,,6.l422,1,mud
465,1046,,61.0001,l,boy
416,989,,62.l416,1,goat
437,1019,,63.11101,l,open
540.1037,,64.6001,l,board
303,904,,7.l426,1,good
318,948,,72.l422,1,mood
451,1196,,74.6003,l,sure
440,2315,,21.1422,1,made
426,2116,,21.4124,1,name
503,2043,,24.6,1,hair
705.1266,,3.1421,1,bad
645,1484,,42.1411,1,bout
712,1294,,44.6,l,hard
687,1328,,47.1428,1,pride
494,1197,,5.1423,1,sod
588,941,,53.14l6,l,caught
66l,l455,,6.1422,1,mud
487,970,,61.0001,1,boy
481,1034,,62.1416,l,goat
536,955,,63.11101,1,open
511,992,,64.6001,1,board
413,950,,7.1426,1,good
370,1018,,72.1422,1,mood
451,1197,,74.6003,l,sure
603,l9ll,,94.6001,1,bird
36l,2064,,1.l42,1,hid
349,2220,,11.1421,1,bead
402.2086,,l4.0009l,1,year
520,187l,,2.1411,1,bet
550.1880,,21.1422,l,made
452,2095,,21.4124,1,name
497,1983,,24.6,1,hair
653,1470,,3.l421,l,bad
683,1448,,42.l411,1,bout
733,1433,,44.6,l,hard
684.1317,,47.l428,l,pride
578,1188,,5.l423,1,sod
532.1077,,53.1416,l,caught

504,1979,,94.6001,l,bird
391,2346,,1.142,1,hid
39l,2339,,ll.1421,1,bead
434.2224,,14.0009l,l,year
523.1793,,2.1411.1,bet
427,2184,,21.1422,1,made
446,2089,,21.4124,1,name
4S7,2044,,24.6,1,hair
774,1210,,3.1421,1,bad
690,1476,,42.1411,1,bout
848,1291,,44.6,1,hard
578.1458,,47.1428,1,pride
568,1070,,5.1423,1,sod
576,974,,53.14l6,1,caught
716,1188,,6.1422,l,mud
484,979,,61.0001,1,boy
460,1012,,62.1416,l,goat
442,929,,63.11101,1,open
508,855,,64.6001,1,board
413,958,,7.1426,1,good
413,807,,72.1422,1,mood
444,1263,,74.6003,1,sure
532.1904,,94.6001,1,bird
375.2227,,1.142,1,hid
395,2279,,11.1421,1,bead
381.1958,,14.00091,1,year
508,1943,,2.1411,l,bet

720,1419,,6.1422,1,mud
465,960,,61.0001,l,boy
435,1006,,62.1416,1,goat
430,928,,63.11101,1,open

556,995,,64.6001,l,board
327,854,,7.l426,l,good
354,1032,,72.l422,1,mood
437,1175,,74.6003,l,sure
542,1813,,94.6001,l,bird
346,2117,,l.l42,l,hid
323,2048,,ll.l421,l,bead
401,2052,,l4.0009l,l,year
570,1770,,2.l4ll,l,bet
392,2062,,21.1422,1,made
453,2215,,21.4124,1,name
537,2054,,24.6,l,hair
754,1424,,3.1421,l,bad
723,1532,,42.14ll,l,bout
723,1331,,44.6,1,hard
599.1362,,47.1428,l,pride
56l,1047,,5.l423,l,sod
588,862,,53.l4l6,l,caught
685,1200,,6.l422,1,mud
491,1097,,6l.0001,l,boy
490,1089,,62.l4l6.l,goat
484.1025,,63.lllOl,1,open
523,854,,64.6001,1,board
396,1009,,7.l426,l,good
339,961,,72.l422,l,mood
483,1312,,74.6003,1,sure
592,1801,,94.6001,1,bird
325,219l,,l.l42,l.hid
368,2182,,11.1421,l,bead
518.2269,,14.00091,1,year
515,1796,,2.l4ll,l,bet
386,2069,,21.l422,l,made
444,2125,,21.4124,l,name
512.1982,,24.6,l,hair
675,1342,,3.l421,1,bad
680,1558,,42.14ll,1,bout

Kenyans

All Kenyans (20 M&F)

242

27l,2616,,1.142.l,hid
244,2438,,11.1421,1,bead
542.1973,,l4.0009l,l,year
531.1877,,2.1411,l,bet
452.1728,,21.l422,1,made
548,1956,,21.4124,1,name
512,1660,,24.6,l,hair
694.1607,,3.1421,1,bad
683,1249,,42.14ll,l,bout
714,1177,,44.6,l,hard
633,1445,,47.1428,1,pride
557.1007,,5.1423,1,sod
466.1025,,53.l416,1,caught
762,1500,,6.1422,1,mud
475,918,,61.0001,l,boy
535,971,,62.1416,1,goat
465,971,,63.11101,l,open
427,789,,64.6001,1,board

682,1281,,44.6,1,hard
690.1494,,47.1428,1,pride
553.1145,,5.l423,l,sod
591.1206,,53.l4l6,1,caught
743,1326,,6.l422,1,mud
454,987,,61.0001,l.boy
494,1025,,62.1416,1,goat
439.1008,,63.11101,1,open
462,834,,64.6001,1,board
36l,1008,,7.1426,l,good
37l,939,,72.1422,1,mood
453,1292,,74.6003,l,sure
515,1706,,94.6001,1,bird
300,2203,,1.142,l,hid
354,2203,,11.1421,1,bead
508.2201,,14.00091,1,year
544.1897,,2.1411,1,bet
393,1968,,21.1422,1,made
44l,2030,,21.4124,1,name
502,1918,,24.6,1,hair
733,1459,,3.1421,1,bad
684,1458,,42.1411,1,bout
720,1459,,44.6,l,hard
641.1328,,47.1428,1,pride
541,1022,,5.1423,1,sod
583.1123,,53.1416,1,caught
662,1101,,6.1422,l,mud
486,955,,61.0001,1,boy
465,1041,,62.l416,1,goat
Sll,948,,63.11101,l,open
483.1160,,64.6001,l,board
381,1025,,7.1426,1,good
388,948,,72.1422,1,mood
4Sl,1241,,74.6003,1,sure
496.1897,,94.6001,1,bird

353,878,,7.l426,1,good
326,810,,72.l422,1,mood
434,1238,,74.6003,1,sure
S98,1518,,94.6001,1,bird
314,2289,,1.142,1,hid
283,2207,,11.1421,1,bead
453.2333,,14.00091,1,year
415,1812,,2.1411,1,bet
456,1750,,21.1422,1,made
493,2094,,21.4124,l,name
S65,l731,,24.6,1,hair
663,1480,,3.1421,1,bad
683,1273,,42.l4ll,l,bout
678,1563,,44.6,1,hard
655.1716,,47.1428,1,pride
534,1114,,5.1423,1,sod
469,996,,53.l416,1,caught
670,1508,,6.1422,1,mud
547,1008,,61.0001,1,boy
472,915,,62.14l6,1,goat
503,996,,63.11101,1,open

134

470,913,,64.6001,1,board
297,865,,7.1426,1,good
317,823,,72.1422,1,mood
426,915,,74.6003,l,sure
670,1334,,94.6001,1,bird
311,193l,,1.142,l,hid
311,1877,,ll.l421,l,bead
563.1924,,14.00091,1,year
536,187l,,2.14ll,1,bet
469,1924,,21.1422,l,made
535,1982,,21.4124,1,name
411,1657,,24.6,1,hair
717,1267,,3.142l,1,bad
674,1226,,42.l411,l,bout
707,1468,,44.6,1,hard
656.1653,,47.1428,1,pride
577,1140,,5.1423,1,sod
529,992,,53.14l6,l,caught
764,1253,,6.1422,1,mud
474,930,,61.0001,l,boy
483,966,,62.l416,1,goat
488,961,,63.11101,l,open
494,858,,64.6001,1,board
37S,972,,7.l426,1,good
390,918,,72.1422,1,mood
394,887,,74.6003,1,sure
691,1281,,94.6001,1,bird
479,2101,,1.142,l,hid
315,2239,,ll.l421,l,bead
373.2085,,l4.0009l,1,year
485,179l,,2.l411,1,bet
4S9,2219,,21.l422,1,made
484,1949,,21.4124,1,name
524.1616,,24.6,l,hair
735.1582,,3.1421,l,bad
640,1209,,42.14ll,1,bout
748,1445,,44.6,1,hard
671,1650,,47.l428,1,pride
551.1254,,5.1423,1,sod
556,932,,53.l4l6,l,caught
643,1299,,6.l422,1,mud
474,955,,61.0001,l,boy
473,887,,62.l4l6,l,goat
502,890,,63.11101,1,open
554,715,,64.6001,1,board
310,682,,7.1426,1,good
347,610,,72.1422,1,mood
432,975.,74.6003,1,sure
794,1555,,94.6001,l,bird
347,2095,,1.l42,1,hid
308,2148,,1l.1421,1,bead
353,2158,,14.00091,1,year
542,1833,,2.l4ll,l,bet
636,1854,,21.1422,1,made
474,1939,,21.4124,1,name
522.1648,,24.6,l,hair
636,1150,,3.1421,1,bad
647,1223,,42.14ll,1,bout

685,1198,,44.6,1,hard
688.1567,,47.1428,1,pride
522.1077,,5.l423,1,sod
568,1030,,53.l4l6,1,caught
702,1191,,6.1422,1,mud
543,1157,,61.0001,1,boy
471,930,,62.1416,l,goat
485,971,,63.11101,1,open
535,923,,64.6001,1,board
381,843,,7.1426,1,good
407,756,,72.1422,1,mood
450,1020,,74.6003,1,sure
648,1097,,94.6001,1,bird
369,1950,,l.142,1,hid
313,2096,,1l.1421,1,bead
492,2043,,14.00091,1,year
480,1615,,2.l411,l,bet
431,1755,,21.1422,l,made
515,2036,,21.4124,1,name
531,1714,,24.6,1,hair
682,1275,,3.1421,1,bad
1120,1306,,4.1422,1,mud
696,1281,,42.1411,1,bout
647,1072,,44.6,l,hard
672,1580,,47.1428,1,pride
515,1037,,5.l423,1,sod
487,898,,53.1416,1,caught
500,957,,61.0001,1,boy
542,1004,,62.l416,1,goat
496,1021,,63.11101,1,open
439,820,,64.6001,l,board
320,856,,7.1426,1,good
405,954,,72.1422,1,mood
418,966,,74.6003,l,sure
620,1372,,94.6001,1,bird
274,2732,,1.142,1,hid
267,2767,,1l.1421,1,bead
372.2003,,14.00091.1,year
464,2460,,2.1411,1,bet
443,2726,,21.1422,l,made
557,2090,,21.4124,1,name
S37,17S6,,24.6,1,hair
788,1570,,3.1421,l,bad
686,1284,,42.14ll,1.bout
746,1612,,44.6,1,hard
664,1270,,47.l428,l,pride
441,859,,5.l423,1,sod
517,781,,53.l416,1,caught
684,944,,6.1422,1,mud
S74,1249,,61.0001,1,boy
546,934,,62.1416,1,goat
558.1027,.63.11101,l,open
491,866,,64.6001,l,board
253,908,,7.1426,1,good
225,718,,72.l422,1,mood
406,854,,74.6003,1,sure
73l,1718,,94.6001,1,bird
294,2502,,1.142,1,hid

135

276,2425,,1l.l421,l,bead
356.2266,,14.0009l,l,year
463,1936,,2.1411,l,bet
473,2655,,2l.1422,1,made
SS7,2230,,21.4124,1,name
522.1705,,24.6,1,hair
710,1597,,3.l421,l,bad
600,1295,,42.1411,1,bout
800,1433,,44.6,1,hard
666,1387,,47.1428,1,pride
553,1148,,5.l423,1,sod
537,772,,53.l416,1,caught
684,948,,6.1422,l,mud
573,995,,61.0001,l,boy
542.1010,,62.14l6,1,goat
556,1022,,63.lllOl,l,open
455,779,,64.6001,1,board
259,906,,7.1426,1,good
259,759,,72.1422,1,mood
402,844,,74.6003,1,sure
843,1503,,94.6001,l,bird
308.1916,,1.l42,1,hid
292,2216,,1l.1421,1,bead
511.1943,,14.00091,l,year
432,1851,,2.1411,l,bet
397,1864,,21.l422,1,made
535,1963,,21.4124,1,name
505,1604,,24.6,1,hair
609,l645,,3.l421,1,bad
656,1209,,42.1411,1,bout
674,1519,,44.6,1,hard
683.1606,,47.l428,1,pride
492,1148,,5.1423,1,sod
524,982,,53.l4l6,1,caught
782,1253,,6.1422,l,mud
541,942,,61.0001,l,boy
508,950,,62.l416,1,goat
476,965,,63.11101,1,open
512,1041,,64.6001,1,board
382,815,,7.l426,1,good
327,957,,72.1422,l,mood
409,875,,74.6003,1,sure
765,1260,,94.6001,1,bird
346,2240,,1.142,l,hid

Zimbabweans

All Zimbabweans (2O M&F)
440

37l,2096,,1.l42,1,hid
298,2096,,1l.1421,1,bead
438.1783,,14.00091,1,year
553,2002,,2.1411,l,bet
495,2081,,21.1422,1,made
540,1870,,21.4124,1,name
550,1548,,24.6,1,hair
509,1830,,3.1421,1,bad
637,1309,,42.14ll,1,b0ut
808,1143,,44.6,l,hard
718,1451,,47.l428,l,pride

 

315,2271,,11.1421,1,bead
530.1915,,14.00091,1,year
428,1802,,2.1411,1,bet
443,2089,,21.1422,1,made
531,1894,,21.4124,1,name
501,1616,,24.6,1,hair
634,1170,,3.1421,1,bad
66l,1220,,42.l411,1,bout
657,1513,,44.6,1,hard
625,1400,,47.1428,1,pride
506,1008,,5.1423,1,sod
457,965,,53.1416,1,caught
647,1148,,6.1422,1,mud
464,901,,61.0001,1,boy
520,953,,62.1416,1,goat
471,986,,63.11101,1,open
474,1121,,64.6001,l,board
353,1034,,7.1426,1,good
303,1054,,72.l422,1,mood
437,1230,,74.6003,1,sure
607.1777,,94.6001,1,bird
3Sl,2374,,1.142,1,hid
318,2302,,11.1421,1,bead
516.1843,.14.00091,1,year
453,1751,,2.1411,1,bet
450,2620,,21.1422,1,made
487,1843,,21.4124,1,name
484,1577,,24.6,1,hair
675,1248,,3.1421,1,bad
649,1190,,42.1411,1,bout
705,1483,,44.6,1,hard
616.1169,,47.1428,1,pride
501,884,,5.l423,1,sod
592,1118,,53.1416,1,caught
708,1502,,6.1422,1,mud
530,917,,61.0001,1,boy
502,932,,62.1416,1,goat
557,935,,63.11101,1,open
506.1229,,64.6001,1,board
326,980,,7.1426,1,good
318,715,,72.l422,1,mood
372,781,,74.6003,1,sure
754,1483,,94.6001,1,bird

539,997,,5.1423,1,sod
597,939,,53.1416,1,Caught
734,1448,,6.l422,1,mud
519,968,,61.0001,l,boy
495,818,,62.1416,1,goat
552,946,,63.11101,1,open
487,923,,64.6001,1,b0ard
400,997,,7.1426,1,good
429,873,,72.1422,1,mood
494,1060,,74.6003,l,sure
560,1995,,94.6001,1,bird
428,2327,,1.142,l,hid
379,2714,,11.l421,1,bead

136

308,2808,,l4.6,l,here
476,2552,,2.l421,l,bed
S72,27l3,,21.l422,1,made
528,2544,,24.6,1,hair
718,1203,,3.l421,1,bad
637,1034,,41.0001,l,buy
6Sl,995,,42.l411,l,bout
592,894,,44.6,1,hard
586,989,,S.l424,l,nod
667,1056,,6.1422,1,mud
410,645,,61.0001,1,boy
549,799,,62.1411,l,boat
543,953,,63.0003,1,saw
395,968,,7.l426,1,good
413,876,,72.l422,l,mood
351,960,,74.6003,1,sure
418,902,,74.6003,l,tour
353,894,,82.,1,hue
564,2516,,94.6001,1,bird
398,2580,,1.l42,1,hid
387,2783,,11.1421,1,bead
398,2805,,14.6,l,here
444,2588,,2.1421,l,bed
407,2550,,21.1422,1,made
531,2281,,24.6,1,hair
644,983,,3.1421,1,bad
605.1056,,41.0001,l,buy
605,940,,42.1411,l,bout
663,1006,,44.6,1,hard
53l,874,,5.1424,l,nod
654,991,,6.1422,1,mud
546,874,,61.0001,l,boy
510,969,,62.1411,l,boat
546.1056,,63.0003,l,saw
388,911,,7.1426,l,good
385,697,,72.1422,1,mood
420,1027,,74.6003,l,sure
415,932,,74.6003,1,tour
401,983,,82.,l,hue
517,2340,,94.6001,l,bird
445,2681,,l.l42,1,hid
437,2563,,ll.1421,l,bead
400,1934,,14.6,l,here
430,2288,,2.1421,l,bed
408,2481,,21.1422,l,made
516,2481,,24.6,l,hair
693,946,,3.1421,1,bad
686,972,,41.0001,l,buy
605,988,,42.l4ll,1,bout
603,987,,44.6,1,hard
586,1270,,5.1424,1,nod
868,1206,,6.l422,1,mud
589,1077,,61.0001,1,boy
553,965,,62.l4ll,l,boat
SO7,1070,,63.0003,1,saw
341,713,,7.1426,1,good
422,947,,72.1422,l,mood
430,951,,74.6003,1,sure

438,876,,74.6003,l,tour
324,1031,,82.,1,hue
467.2407,,94.6001,1,bird
395,2474,,1.l42,1,hid
372,2906,,11.1421,1,bead
342,2847,,l4.6,1,here
4S4,2511,,2.1421,1,bed
496,2585,,21.1422,1,made
491,2608,,24.6,1,hair
655,1013,,3.l421,1,bad
737.1117,,41.0001,1,buy
617,993,,42.1411,1,bout
640,1043,,44.6,1,hard
573,939,,5.1424,1,nod
722,1265,,6.1422,1,mud
485,916,,61.0001,1,boy
521,947,,62.l411,l,boat
521,954,,63.0003,1,saw
320,805,,7.1426,l,good
380,1088,,72.1422,1,mood
373,975,,74.6003,1,tour
358,916,,74.6003,1,sure
358,1334,,82.,1,hue
513,1886,,94.6001,l,bird
404,2661,,1.l42,1,hid
404,2555,,11.1421,1,bead
351,2646,,14.6,1,here
526,2173,,2.1421,1,bed
549,2067,,21.1422,1,made
472,2578,,24.6,1,hair
643,1053,,3.1421,1,bad
682,976,,41.0001,1,buy
633,984,,42.1411,1,bout
580,1044,,44.6,1,hard
S49,1260,,5.1424,l,nod
686,1213,,6.1422,1,mud
442,762,,61.0001,1,boy
549,999,,62.1411,1.boat
541,1042,,63.0003,1,saw
348,828,,7.1426,1,good
395,1053,,72.1422,l,mood
401,960,,74.6003,1,sure
450,1053,,74.6003,l,tour
396,923,,82.,1,hue
556,1891,,94.6001,1,bird
402,2384,,1.142,1,hid
401,2816,,11.l421,1,bead
402.2504,,l4.6,1,here
521.2303,,2.1421,1,bed
551,2511,,21.1422,l,made
539,2616,,24.6,1,hair
618,894,,3.1421,1,bad
618,976,,41.0001,1,buy
64l,993,,42.l411,1,bout
700,1028,,44.6,1,hard
588,953,,5.1424,1,nod
886,1422,,6.l422,l,mud
566,723,,61.0001,1,boy

676,1247,,42.1411,1,b0ut
694,953,,44.6,l,hard
S48,993,,5.l424,1,n0d
865,1264,,6.1422,1,mud

513,924,,62.l4ll,l,b0at
469,990,,63.0003,l,saw
365,857,,7.l426,1,g00d

402,865,,72.l422,l,mood
343,782,,74.6003,1,sure
469,857,,74.6003,1,tour
432,990,,82.,l,hue
483,1722,,94.6001,l,bird
301,243S,,l.l42,l,hid
306,2675,,1l.l421,1,bead
360,2308,,l4.6,l,here
511,2352,,2.1421,l,bed
517,2617,,21.1422,1,made
493,2454,,24.6,l,hair
610,911,,3.1421,l,bad
595,955,,41.0001,l,buy
659,1084,,42.14ll,1,bout
793,1290,,44.6,1.hard
396,889,,5.l424,l,nod
644,1230,,6.l422,l,mud
580,1014,,6l.0001,1,boy
506,896,,62.l4ll,1,boat
521,904,,63.0003,l,saw
322,881,,7.1426,l,good
388,728,,72.1422,l,mood
407,900,,74.6003,l,sure
447,874,,74.6003,1,tour
327,2419,,82.,l,hue
435,1896,,94.6001,l,bird
412,2605,,l.l42,1,hid
381,2480,,11.l421,1,bead
358,2457,,14.6,1,here
450,2473,,2.l421,1,bed
482,2659,,21.1422,1,made
428,2551,,24.6,1,hair
664,1057,,3.l421,l,bad
637,995,,41.0001,l,buy
640,1052,,42.l4ll,1,bout
69l,972,,44.6,1,hard
428,1150,,5.1424,l,nod
660,1136,,6.l422,l,mud
524,941,,61.0001,l,boy
521.1019,,62.l411,1,boat
521.1019,,63.0003,1,saw
365,863,,7.1426,l,good
346,972,,72.1422,l,mood
365,762,,74.6003,l,sure
551,948,,74.6003,l,tour
322,972,,82.,l,hue
544,1921,,94.6001,1,bird
398,2489,,1.l42,1,hid
481,2528,,1l.l421,1,bead
382,2558,,14.6,l,here
436,2437,,2.1421,l,bed
413,2482,,21.l422,1,made
444,2415,,24.6,l,hair
614,962,,3.1421,l,bad
616,993,,41.0001,1,buy

533,985,,61.0001,l,boy
526,910,,62.1411,1,boat
556,1218,,63.0003,1,saw
330,932,,7.1426,l,good
360,752,,72.1422,l,mood
451,782,,74.6003,1,sure
496,1030,,74.6003,1,tour
360,894,,82.,1,hue
444,1956,,94.6001,1,bird
411,2460,,l.l42,1,hid
395.2857,,1l.1421,1,bead
426,2625,,14.6,1,here
433,2736,,2.1421,1,bed
414,2550,,21.1422,1,made
425,2594,,24.6,1,hair
684,950,,3.1421,1,bad
687,972,,41.0001.1,buy
657,1076,,42.14ll,1,bout
698,912,,44.6,l,hard
511,1203,,S.l424,1,nod
680,1128,,6.1422,1,mud
523,1047,,61.0001,1,boy
530,920,,62.1411,1,boat
486,1092,,63.0003,1,saw
381,935,,7.1426,1,good
366,1064,,72.1422,1,mood
386.778,,74.6003,1,sure
448,964,,74.6003,1,tour
403.1039,,82.,1,hue

486,1869,,94.6001,1,bird

416,2207,,1.l42,1,hid
364,2207,,11.l421,l,bead
459,2041,,14.00091,l,year
527,1762,,2.1411,1,bet
610,2014,,21.1422,1,made
394,1869,,21.4124,1,name
557.1957,,24.6,1,hair
639.1747,,3.l421,1,bad
647,1318,,42.1411,l,bout
698,1159,,44.6,l,hard

706,1333,,47.1428,l,pride

505,996.,5.1423,l,sod
658,1086,,53.1416,1,caught
706,1086,,6.1422,1,mud
504,1015,,61.0001,1,boy
500,822,,62.1416,1,goat
572.1033,,63.11101,l,open
512,899,,64.6001,1,board
394,959,,7.1426,1,good
378,708,,72.1422,1,mood
533,1120,,74.6003,l,sure
520,1635,,94.6001,1,bird
357,2182,,1.142,1,hid
330,2645,,11.1421,l,bead

 

659,1174,,44.6,l,hard
647,1389,,47.l428,1,pride
614,1158,,S.1423,1,sod
515,1128,,53.l4l6,l,caught
630,1367,,6.1422,l,mud
59l,952,,6l.0001,l,boy
504,860,,62.l4l6,l,goat
498,1013,,63.lllOl,l,open
491.1059,,64.6001,1,board
373,1097,,7.l426,l,good
360,767,,72.l422,l,mood
447,1204,,74.6003,1,sure
484,1419,,94.6001,1,bird
364,2200,,l.l42,1,hid
363.2177,,1l.1421,l,bead
489.1985,,14.00091,l,year
486,1168,,2.l4ll,l,bet
553.2038,,21.1422,1,made
445,1948,,21.4124,l,name
S97,l789,,24.6,1,hair
697,1517,,3.l421,1,bad
666,1353,,42.14ll,l,bout
804,1388,,44.6,l,hard
644.1371,,47.1428,l,pride
515,986,,5.l423,l,sod
682.1039,,53.l416,l,caught
811,1381,,6.1422,l,mud
584,1066,,6l.0001,1,boy
442,834,,62.14l6,1,goat
538,986,,63.11101,1,open
543,1039,,64.6001,1,board
360,1039,,7.l426,1,good

337,756,,72.l422,1,mood
465,1013,,74.6003,1,sure
553,1578,,94.6001,l,bird
349,2078,,1.142,l,hid
337,2234,,ll.1421,1,bead
469,2258,,l4.00091,1,year
418,199l,,2.1411,1,bet
416.2227,,21.l422,1,made
470,1900,,21.4124,1,name
538,2006,,24.6,1,hair
580,1812,,3.1421,1,bad
667,1380,,42.1411,1,bout
721,1223,,44.6,1,hard
635,1461,,47.l428,1,pride
572,1168,,5.1423,1,sod
563,898,,53.l4l6,l,caught
698,1529,,6.l422,l,mud
556,1115,,61.0001,1,boy
466,849,,62.1416,1,goat
454,898,,63.11101,1,open
507,917,,64.6001,1,board
4OS,1024,,7.1426,1,good
410,784,,72.1422,1,mood
510,1042,,74.6003,1,sure
470,1949,,94.6001,1,bird

APPENDIX C

140

APPENDIX C

An Example of a Plotnik Vowel Chart

3000 2:300 2600 2100 22100 2900 ratio 1600 1100 12.00 1000 090

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

200
300_ . lI) bead
I 0 good
h'd u o"
400. ‘ ° mod
0 gear .
e 0 . 7 fi
500_ “a" sure boll
sod goat
0 .,. board
600- . k sod caught J
E bout
700‘ Amud '17
a pride
800- hard :
I Jones,33,Formal,Rural,Zimbabwe

 

141

 

APPENDIX D

142

Gender:

Age:

Mother Tongue:

Please listen carefully to the following pairs of words. For each pair indicate whether the
two words you hear are the same or different by marking (‘1 ) appropriately. For example

QUESTIONNAIRE USED TO ELICIT VOWEL PERCEPTIONS

 

 

APPENDIX D

 

if words in pair 1 sound the same to you, mark:

Pair 1: (a) Same ‘1
What we mean by ‘same’ or ‘different’ is :- is the speaker pronouncing the same words in
each pair or is he pronouncing different words? You are simply judging the speaker’s

(b) different

pronunciation and your judgement is correct.

Pair 1 :
Pmr2:
Pair 3:
Pmr4:
Pair 5:
Pair 6:
Pair 7:
Pair 8:
Pair 9:
Pair 10:

Pair 11:
Pair 12:
Pair 13:

Pair 14:

Pair 15:
Pair 16:
Pair 17:
Pair 18:
Pair [9:
Pair 20:

Pair 21:

Pair 22:
Pair 23:
Pair 24:
Pair 25:
Pair 26:

(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same

(b) different
(b) different
(b) different
(b) different
(b) different
(b) different
(b) different
(b) different
(b) different

(a) Same
(a) Same
(a) Same
(a) Same

(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same

(a) Same
(a) Same
(a) Same
(a) Same
(a) Same
(a) Same

(b) different
(b) different
(b) different
(b) different

(b) different
(b) different
(b) different
(b) different
(b) different
(b) different
(b) different

(b) different
(b) different
(b) different
(b) different
(b) different
(b) different

143

Pair 27: (a) Same (b) different
Pair 28: (3) Same (b) different
Pair 29: (a) Same (b) different
Pair 30: (a) Same (b) different
Pair 31: (a) Same (b) different
Pair 32: (a) Same (b) different
Pair 33: (a) Same (b) different
Pair 34: (a) Same (b) different
Pair 35: (a) Same (b) different
Pair 36: (a) Same (b) different
Pair 37: (a) Same (b) different
Pair 38 : (a) Same (b) different
Pair 39: (a) Same (b) different
Pair 40: (a) Same (b) different
Pair 41: (a) Same (b) different
Pair 42: (a) Same (b) different
Pair 43: (a) Same (b) different
Pair 44: (a) Same (b) different
Pair 45: (a) Same (b) different
Pair 46: (a) Same (b) different
Pair 47: (a) Same (b) different
Pair 48: (a) Same (b) different
Pair 49: (a) Same (b) different
Pair 50: (a) Same (b) different
Pair 51: (a) Same (b) different
Pair 52: (a) Same (b) different
Pair 53: (a) Same (b) different
Pair 54: (a) Same (b) different
Pair 55: (a) Same (b) different
Pair 56: (a) Same (b) different
Pair 57: (a) Same (b) different
Pair 58: (a) Same (b) different

Please continue with section B

B. Please listen carefully to the following pairs of words and write down each word that
the speaker pronounced. This should work like a dictation exercise. Once again whatever
you hear is correct, it is an individual judgement and that is all that counts. So don’t

hesitate to write exactly what you hear.

144

Pair 1:

(a)

 

 

Pair 2:

(a)

 

 

Pair 3:

(a)

 

 

Pair 4:

(a)

 

 

Pair 5:

 

(a)

 

Pair 6:

(a)

 

 

Pair 7:

(a)

 

 

Pair 8:

(a)

 

 

Pair 9:

(a)

 

 

Fair 10:

(a)

(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)

 

 

Pair 11:

(a)

(b)

 

 

Pair 12:

(a)

(b)

 

 

Pair 13:

(a)

(b)

 

 

Pair 14:

(a)

(b)

 

 

Pair 15:

(a)

(b)

 

 

Pair 16:

(a)

(b)

 

 

Pair 17:

(a)

(b)

 

 

Pair 18:

(a)

(b)

 

 

Pair 19:

(a)

 

 

Pair 20:

(a)

(b)
(b)

 

 

Pmr2h

(a)

(b)

 

 

Fair 22:

(a)

(b)

 

 

Pair 23:

(a)

(b)

 

 

Pair 24:

(a)

 

 

Pair 25:

(a)

 

 

Pair 26:

(a)

 

 

Pair 27:

(a)

 

 

Pair 28:

(a)

 

 

Pair29: (a)

 

 

Pair 30:

(a)

 

 

Pair 31:

(a)

 

 

Pair 32:

(a)

 

 

Pair 33:

(a)

 

 

Pair 34:

(a)

 

 

Pair 35:

(a)

 

 

Pair 36:

(a)

 

 

Pair 37:

(a)

 

 

Pair 38:

(a)

 

 

Pair 39:

(a)

 

 

Pair 40:

(a)

 

 

Pair 41:

(a)

 

 

Pair 42:

(a)

 

 

Pair 43:

(a)

 

 

Pair 44:

(a)

 

 

Pair 45:

(a)

 

 

Pair 46:

(a)

 

 

145

(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)
(b)

Pair 47: (a)

 

 

Pair 48: (a)

 

 

Pair 49: (a)

 

 

Pair 50: (a)

 

 

Pair 51: (a)

 

 

Pair 52: (a)

 

 

(b)
(b)
(b)
(b)
(b)
(b)

Pair 53: (a)

 

 

Pair 54: (a)

 

 

Pair 55: (a)

 

 

Pair 56: (a)

 

 

Pair 57: (a)

 

 

Pair 58: (a)

 

 

Thank you very much for your participation.

146

(b)
(b)
(b)
(b)
(b)
(b)

APPENDIX E

147

(a) Stimulus [a]

APPENDIX E

IDENTIFICATION TASK RESULTS

 

 

 

 

 

 

 

 

 

 

 

 

Stimuli minimal pairs Gh'. Respondents' Zim. Respondents' Ken. Respondents'
words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
a ~ a a ~ a 23-33 a ~ a 47-84 a ~ a 46-50
bat/bat A ~ A 75-63 A ~ A 50-16 A ~ A 50-41
a~ A a~A 79-53 a~A 85-41 a~A 67-42
mad/mud A ~ a 15-36 A ~ 3 10-47 A ~ a 26-51
a~a a~a 74-50 a~a 81-88 a~a 56—28
a ~a 15-41 (1 ~21 19-09 a:~a 21-49
had/hard 3 ~ 3 19-16
a~ai a~ai 65-12 a~ai 84-16 a~ai 70-41
au ~ a 17-58 A ~ a 06-72 A ~ a 22-37
dam/dime
a~au A~A 74.43 a~a 50-53 a~au 51-31
a~a 29-35 A~A 47-16 A~a 42-20
bat/I301“ au=o au=0 ai ~ 31 06'18
a~D a~D 73-12 a~D 84-65 a~D' 73-00
cat/cor a ~ 0 73-54 A ~ A 10-19 A ~ 15-63
A~A 27-27 a:~a 06-17
0: ~ 0 3: ~ 0 70-28 0: ~ 0 98-56 0: ~ 0 61-41
caught/cot A ~ a 23-47 A ~ 0 02-34 on ~ 0: 24.47
0 ~ 0_ 05-16
3 ~ 3 Q ~ (F. 92-96 g: ~ a: 97-91 Q. ~ (B 98-93
walk/work
c~ua c~ua 54-20 c~ua 100-100 c~ua 15-30
law/lure 0 ~ 0 37-59 0 ~ 0 84-43
8 ~ 9 01-18

 

 

 

 

 

' Gh denotes Ghanaian, Zim , Zimbabwean , and Ken. Kenyan

148

 

 

(c) Stimulus [u]

 

Stimuli minimal
pairs
words played to

Gh’s respondents
perception of stimuli
vowels (in %)

Zim’s respondents
perception of stimuli
vowels (in %)

Ken’s respondents
perception of stimuli
vowels (in %)

 

 

 

 

 

 

 

 

 

 

 

u ~ u u ~ u 96-96 u ~ u 100-73 u ~ u 75-74
good/good u ~ 0 -do- 27 on ~ ou 15-15
u ~ U u ~ U 58-52 u ~ 6 94-76 11 ~ u 86-30
wow/“06d U ~ “ 3348 ~ 0 03-27 u ~ u 1445
u ~ju u ~ju 64-91 u ~ju 94-76 u ~ju 83-83
who/hue ju ~u 36-09 U A. u 03-15
u ~ ou u ~ ou 66-61 u ~ ou 100-49 11 ~ ou 79-26
boot/boat A ~ A 25-18 u ~ 0 -do- 35 A ~ 0 10-50
U ~ U U ~ U 96-98 U ~ U 100-100 U ~ U 93-93
book/book
U ~ ou U ~ ou 96-49 . U ~ on 98-33 U ~ on 92-35
cook/coke u ~ U 0347 u ~ 0 02-67 0 ~ 3: 08-45
((1) Stimulus [I]

 

Stimuli minimal pairs

Gh’s respondents

Zim’s respondents

Ken’s respondents

 

 

 

 

 

words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
I ~ I I ~ I 60-45 I ~ I 39-30 I ~ I 25-20
Hid/hid 1 ~ 1 22-14 1 ~i 39-42 6 ~ 8 63-60
1~ 8 1~ 8 68-36 I~ 8 61-61 I~ 8 52-63
Bit/bet g... A 30-32 i ~ a 36—16 i ~I 41-17
I ~ 3 I ~ 3 83-66 I ~ 3 39-63 I ~ 3 40-25
Bid/bird I~ 8 19-34 i ~ 6 48-20 1 ~ a 25-66

8 ~ A 21-06

 

 

 

149

 

 

 

 

 

I ~ ia I ~ ia 50-36 I ~ ia 49-61 I ~ ia 35-80
Bid-beard 1 ~ i 29-19 i ~i 46-29 i ~ i 30-14
e ~a 16-04

i~I i~I 68-48 i~I 41-24 i~I 75-21
beat/bit I ~ i 32-38 I ~ i 59-41 I ~ i 21- 28
i ~8 -do- -36 I ~ 8 ~do-—49

 

 

 

 

(e) Stimulus [e]

 

Stimuli minimal pairs

Gh’s respondents

Zim’s respondents

Ken’s respondents

 

 

 

 

 

 

 

 

words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
e ~ e e ~ e 96-95 e ~ e 97-96 e ~ e 48-49
made/made ai ~ ai 43-49
e~8 e~8 83-13 e~8 73-46 e~ 8 31-70
wade/wed .. 3 53 ~ 9 -37 ai 77-13
3 19 a1 06
e ~ 3 e ~ 3 85-78 e ~ 3 83-06 e ~ 3 23-79
wade/word 3 ... 8 10-09 ~ a: -94 ai ~ a: 77-13
~ e -08
e ~ a e ~ a 36-34 e ~ a 84-84 e ~ a 56-52
bait/bat 8 ~ A 20—61 ~ A -13 ai ~ A 3344
e~ia e~ia 37-61 e~ia 55-100 e~ia 45-100
bay/beer 8~ea 26-34 ia ~ ia 23- -do- ai ~ la 26- -do-
3 ~ ia 16- -do
e ~ ea c ~ ea 53-39 e ~ ea 83-38 e ~ ea 57-57
hay/hair a ~ 3 21-24 ~ e -25 ai ~ 3 29-18
a: ~ 3 13-22
i~e i~e 64-43 i~e 43-59 i~e 57-14
beat/bait 1~A 31-17 I~ai 47-13 a~a 60-60
1 ~ a -do- -13

 

 

 

 

(f) Stimulus [8]

150

 

 

 

Stimuli minimal pairs

Gh. respondents

Zim. respondents

Ken. respondents

 

 

 

 

 

words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
8~8 8~8 51-42 8~8 55-58 8~8 71-60
bet/bet a ~ 3 40-40 a ~ a 22-27 8 ~ e 13-18
e~e 15-10 a~a 13-11
8~3 8~3 31-38 8~3 55-62 8~3 56-16
bed/bird 3 ~ a 31-27 3 ~ a 15-21 3 ~ a 36-73
a ~ ea 20-24
8~a 8~a 26-42 8~a 44-84 8~a 72-83
bed/bad 3 ~ A 36-42 a~ A 34-16 3 ~ 3 28-07
A ~ a: 28-08 e ~ 22-
i~8 i~8 100-22 i~e 98-67 i~8 93-65
weed/wed .- 3 - 61 ia ~o 02-19 ai ~ ai 07-19

 

 

 

 

Stimulus [3]

 

Stimuli minimal pairs

Gh. respondents

Zim. respondents

Ken. respondents

 

 

 

 

 

 

words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
3 ~ 3 3 ~ 3 68-57 3 ~ 3 73-50 3 ~ 3 32-28
bird/bird a ~ 8 13-17 a ~ a 23-27 a ~ a 60-60
8~a 09-11 8 --—l4
3 ~ 8 3 ~ a 45-62 3 ~ 3 45-91 3 ~ a 27-73
bird/bad g .. A 22-31 a ~ 8 27-03 a ~ a: 56—15
A ~g 18-04 e~A 21-03 a: ~A 10-12
3 ~ A 3 ~ A 40-33 3 ~ A 42-16 3 ~ A 46-15
bird/bud 8 ~ a 22-44 a ~ a: 35-61 a ~ a 54-78
a~3 20-11 0 ~0 10-13
3 ~ 0 3 ~ 0 84-84 3 ~ 0 53-65 3 ~ 0 41-63
heard/hoed a ~ 3 09-05 a: ~ 8 38-24 a: ~ 0: 32-20
3 ~ an 17-17
3 ~ 0: 3 ~ 0: 50-32 3 ~ 3: 34-84 3 ~ 0: 39-61
a~oi 18-45 oz~oi 34-16 a~o 60-21
bird/board 8 ~ 3 16-11 a ~ 22

 

 

 

 

151

 

 

 

 

 

8~3 8~3 31-38 8~3 55-62 8~3 56-16
bed/bird 3 ~ a 31-27 3 ~ a 15-21 a ~ a 36-73
a ~ ea 20-24
I ~ 3 I ~ 3 83-66 I ~ 3 39-63 I ~ 3 40-25
Bid/bird I~8 19-34 i~e 48-20 i~a 25-66
8 ~ A 21-06

 

 

 

 

Stimulus [A]

 

Stimuli minimal pairs

Gh. respondents

Zim. respondents

Ken. respondents

 

 

 

 

 

 

 

 

 

 

words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
A~A 0~A 08-04 A~A 13-13 A~A 13-13
rug/rug 0 ~ 0 83-83 3 ~ 0 87-87 3 ~ 0 54-50
3 ~ 3 13-16
A ~a A~a 48-73 A~a 52-91 A~a 57-60
mud/mad a~A 41-25 a~e 31-07 a~A 45-38
A ~ 3 A ~ 3 07-78 33 ~ 81 49-57 A ~ 3 24-48
hut/hurt 3 ~ 3 68-11 0 ~ 0 43'2” 3 ~ A 62-26
e~A 09-11 a~a 22-26
A ~ 8 A ~ 8 92-37 A ~ 8 06-57 A ~ 8 57-50
land/lend 3 ~ A 04 -37 a ~ a 94-40 a ~ a 33-35
i ~ i 02-14
A~a A~a 04-76 A~a 4-42 A~a 20-57
gut/got 0 ~ ou 62-13 3 ~ ou 79-45 3 ~ ou 63-21
g ~3 22-07 ou~e 21-12
A~o A~o 41-68 A~o 18-64 A~o 34-20
mud/mode a ~ A 24-12 a ~ a: 36-21 a ~ on 40-48
0~3 45-15 ou~a 14-10
A~a: A~a: 45-31 o~oz 59-78 A~a: 15-40
bud/board 0 ~ oi 24-36 a ~0u 19-12 a ~ 0 30-23
3 ~23 10-26 3 ~o 23-17
3~A 3~A 40-33 3~A 42-16 3~A 46-15
bird/bud 8 ~ a 22-44 a ~ a: 35—61 3 ~ a 54-78
a~3 20-11 0 ~0 10-13
a ~ A a ~A 79-53 a ~ A 85-41 a ~ A 67-42
mad/mud A ~ a 15-36 A ~ a 10-47 A ~ a 26-51

 

 

 

 

152

 

 

 

 

3 ~ A
Bought/but

 

3~A 46-80
ou~a 30-11
A~o_ 15-07

 

0 ~A 73-42
ou~a 27- 35

 

0~A 61-59
ou~o36-3O
au~a 16-23

 

Stimulus [C1]

 

Stimuli minimal pairs

Gh. respondents

Zim. respondents

Ken. respondents

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

words played to perception of stimuli perception of stimuli perception of stimuli
respondents vowels (in %) vowels (in %) vowels (in %)
0 ~ 0 0 ~ 0 50-38 0 ~ 0 57-55 0 ~ 0 38-35
hot/hot a ~ a 15-22 a ~ a 14-39 a ~ a 24-24
3 ~3 15-22 3 ~3 24-24
0 ~ au 0 ~ au 77-63 0 ~ au 58-31 0 ~ au 75-35
got/gout 3 ~ 0 12-21 3 ~ 2 41- 50 011 ~ on 25-38
ou ~ 0 07-16
0 ~32 a~a: 14-71 a~a: 6-85 a~a: 15-70
con/caught 0 ~ ou 52-17 3 ~ A 94-06 0 ~ ou 61-19
A ~A 29-14
0 ~ 01 0 ~ oi 11-56 0 ~ oi 4-24 0 ~ oi 38-27
sod/soy 0 ~ 0: 60-29 3 ~ 3: 88-70 0 ~ 3: 23-33
0: ~o 17-21
0 ~ ua Cl ~ ua 17-95 0 ~ ua 37-50 0 ~ ua 14-91
cod/cure A ~ 0 37-05 0 ~ ju 60-47 a: ~.o 37-08
3 ... ua 10_-d0_ a: ~ ua 3l--do-
i ~ ia i ~ ia 74-100 i ~ ia 79-76 i ~ ia 68-100
he/here ia ~ ia 17- -do- ia ~ i 21-24 ia ~ ia 32- -do-
1 ~ ju 1 ~ ju 85-73 i~ju 94-58 i~ju 82-69
he/hue ju ~i 15-17 ia ~i 06-39 ia ~i 09-32

 

 

 

 

153

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 ~ oi 0 ~ oi 11-56 0 ~ oi 4-24 0 ~ oi 38-27
sod/80y 0 ~ 3: 60-29 0 ~ 3: 88-70 3 ~ 3: 23-33
3: ~ 0 17-21
0 ~ ua 0 ~ ua 17-95 0 ~ ua 37-50 0 ~ ua 14-91
cod/cure A .... 0 37-05 3 ~ ju 60.47 0: ~ 0 37-08
0 ~ ua lO--dO- 3: ~ 113 31'-dO-
0 ~ 30 0 ~ au 77-63 0 ~ an 58-31 0 ~ au 75-35
got/gout 0 ~ 6 12-21 3 ~ 3 41- 50 011 ~ on 25-38
ou ~ 0 07-16

 

 

 

 

154

 

 

 

 

APPENDIX F

155

APPENDIX F

DISCRIMINATION RESULTS

Table 1- Results of Minimal Pair Tests administered to Kenyan, Ghanaian and Malawian
University students.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ghanaian Zimbabwean Kenyan
Discrimination Discrimination Discrimination
Vowel pairs S D S D S D
tested "/0 "/o % % % %
I ~ ia 50 50 4O 60 35 65
Bid/beard
8 ~ 8 68 32 34 66 49 51
made/ made
e ~ 8 26 74 13 87 51 49
wade/wed
3 ~ A 46 54 50 50 57 43
bird/bud
21 ~ 3 88 12 48 52 78 22
bat/bat
a -. A 24 76 37 67 55 45
mad/mud
a ~ au 46 54 27 73 57 43
hat/bout
A - A 96 4 57 43 78 22
rug/rug
A .- 3 64 36 53 47 49 51
hut/hurt -
A - a so so 35 65 67 33
gut/got
A .- o 26 74 22 78 47 53
mud/mode
a .-. a 52 48 50 50 57 43
hot/hot
a - 0,, 32 68 33 67 57 43
got/goat
a - a; 36 64 37 63 53 47
cot/caught
3 ~ 3 66 34 52 48 63 37
call/call
3 ~ (Ju 12 88 8 92 57 43
bought/bout

 

 

 

 

156

 

 

 

 

 

 

3; ~ g 20 80 42 58 39 61
caught/cot
3 - ua 42 58 so 50 47 53
law/lure
u ~ u 74 26 45 55 65 35
good/good
u - U 44 56 25 75 37 63
wood/wooed
U - DU 52 48 45 55 49 51
cook/coke

 

 

 

 

 

 

 

157

 

REFERENCES

158

REFERENCES

Adegbija, BE. 1994. Language Attitudes in Sub-Saharan Africa. A Sociolinguistic
Overview. Clevedon: Multilingual Matters Ltd.

Anderson, Stephen. 1985. Phonology in the Twentieth Century. Chicago: University of
Chicago Press.
7‘ Angogo, R. and Hancock, 1. 1980. English in Africa: Emerging Standards or Diverging
Regionalisms? English World- Wide: A Journal of Varieties
of English 1(1): 67—96.

PBailey, L, and Gorlach, M. (eds) 1982. English as a World Language. Ann Arbor:
University of Michigan Press.

Baker, S. 1966. The Australian Language. Sydney: Currawong.

It Bokamba, E. 1981. The Africanization of English. In Kachru, B.B.(ed.).The Other
Tongue. English Across Cultures. Urban: Illinois University Press, 77—98.

\[ Angogo, R. & 1. Hancock. 1980. “ English in Africa: Emerging standards or diverging
regionalisms?”. English World- Wide] :6 7-96.

“f Bailey, L. and Gorlach, M. (eds.) 1982. English as a World Language. Ann Arbor:
University of Michigan Press.

Banda, Felix. 1996. “ The scope and characterization of African English: Some
sociolinguistic considerations”. English World- Wide 17:1 63-75.

Best, C.T., McRoberts, G.W., Sithole, N.M. (1988) Examination of perceptual
reorganization for nonnative speech contrasts: Zulu click discrimination by
En glish-speaking adults and infants. Journal of Experimantal Psychology: Human
Perception and Performance. Vol. 14, No. 3, 345-360

\L Bokamba, E. 1981. "The Africanization of English". In Kachru, ed. The Other Tongue:
English Across Cultures. Urbana: Illinois University Press, 77-98

Dipaolo, Marianna. 1988. Pronunciation and categorization in sound change. K. Ferrara,
et a1(eds.), Linguistic Change and Contact: NWAV-XVI. Austin: University of

159

Texas, 84-92.

Dwyer, David J. 1969. An introduction to West African Pidgin English. African Studies
Center, Michigan State University

Flege, James E., and E. Frieda. 1997. Amount of native-language (L1) use affects the
pronunciation of an L2. Journal of Phonetics 25, 169-186.

Flege, James E., and S. Jang. 1997. Effects of experience on non-native speakers’
production and perception of English vowels. Journal of Phonetics 25, 437-470

Gimson, AC. 1980. Introduction to the Pronunciation of English.London:
Longman.

Gorlach, M. 1987. "English as a world language: the state of art". English World- Wide
9: 1-32.

1991. Englishes: Studies in Varieties of English 1984-88. Amsterdam:

 

Benjamins.

1995. More Englishes: New Studies in Varieties of English, 1988-94.

 

Amsterdam:Benj amins

Hancock, 1. F. and Angogo, R. 1982. English in East Africa. In Bailey, R. and Gorlach,

M. (eds.), English as a World Language. Ann Arbor: University of Michigan
Press, 306-23.

Harms, LS. 1961. Listener’s judgements of status cues in speech. Quarterly Journal of
Speech, 47,164-68.
Heine, B. 1970. Status and Use of African Lingua F rancas. Munich: WTV.

Herbert, R.K.(ed) 1992. Language and Society in Africa. The Theory and Practice of
Sociolinguistics. Cape Town. Witwatersrand University Press.

Hinnebusch, T. 1989. “Bantu” in J. Bender-Samuel (ed.). The Niger—Congo Languages.
Lanham,MD.: UPA

160

Ingram, John C., and S. Park. 1997. Cross-language vowel perception and production by
Japanese and Korean learners of English. Journal of Phonetics 25, 343-370

Kachru, B.B.(ed) 1982.The Other Tongue: English Across Cultures. Urbana: University
of Illinois Press.
Lambert, W.E. et a1 Evaluational reaction of Jewish and Arab adolescents to dialect and
language variations. Journal of Personality and Social Psychology, 1965, 2, 84 -
90.

1960. Evaluational reaction to spoken languageJournal of Abnormal Social
Psychology, 60: 44-51.

Lanham, L.W. 1982. English in South Africa. In Bailey, R. and Gorlach, M.
(eds). English as a World Language. Ann Arbor: University of Michigan
Press, 353-83.

Jibril, Munzali. 1986. “Sociolinguistic variation in Nigerian English”. English World-
Wide 7:47-75

Jusczyk, Peter. 1986. Speech perception. In Kenneth Boff, Lloyd Kaufman, James
Thomas (eds.) Handbook of Perception and Human Perception: Volume II
Cognitive Processes and Performance.New York: John Wiley Publications.

Kenstowiscz, M. 1994. Generative Phonology. Cambridge, Ma. Blackwell

Kuhl, Patricia. 1993. Early linguistic experience and phonetic perception: implications
for theories of developmental speech perception. Journal of Phonetics, 21, 125-
139.

Labov, Williams. 1966. The Social Stratification of English in New York City.
Washington DC: Center for Applied Linguistics.

1972. Sociolinguistic Patterns. Philadeiphia: University of Pennyslavania

 

Press.

1984. Field methods of the project on linguistic change and variation. In

 

John Bauch and Joel sherzer (eds), Language in Use: Readings in

161

 

I ' . pus-"mm?
é

Sociolinguistics. New Jersey: Prentice-Hall.

1994. Principles of Linguistic Change. Cambridge, MA: Blackwell

 

Ladefoged, P. & Maddieson, I. 1990. Vowels of the world’s languages in Journal of
Phonetics 18, 93-122

Lieberman, A.M. & Mattiningly, LG. 1985. The motor theory of speech perception
revised, Cognition, 21, 1-36.

Lieberman, P. & Blumstein, S. 1993. Speech Physiology, Speech Perception, and
Acoustic PhoneticsNew York: Cambridge University Press.

Mackay, D.G. et a1. 1987. Relationships and modules within language perception and
production: An introduction. In Alan Allport, Donald Mackay, Wolfgang Prinz,
Eckart Scheerer (eds.) Language Perception and Production: Relationships
Between Listening, peaking, Reading and Writing. Academic Press: London.

Obiechina, Emmanuel. 1974. " Variety differentiation in English usage". Journal of
Nigerian Studies Association Vol. 6, no.1 74-94

Peterson, G., and H. Barney. 1952. Control methods used in a study of vowels. Journal of
Acoustical Society of America 24. 174-84.

Preston, Dennis R.(ed) 1993. American Dialect Research. Philadelphia: John Benjamins
publishing company.

1988. Methods in the study of dialect perception. In Alan Thomas (ed.),
Methods in dialectology. Clevedon, Avon and Philadelphia: Multilingual

Matters, 373-95

1989. Sociolinguistics and Second Language Acquisition. New York:
Blackwell.

1999. Folk metalanguage: Michigan State University (unpublished)

1999. The northern cities chain shift in your mind. Michigan State University

(unpublished)

162

Pride, J. (ed.). 1982. new Englishes. Rowley, Mass.: Newbury House
Schmied, J. 1991. English in Africa: An Introduction. London: Longman.

1990. " Language use, attitudes, performance and sociolinguistic background:
A study of English in Kenya, Tanzania, and Zambia". English World-Wide
1 1:2 217-238

Schneider, E. (ed.) 1997. Englishes Around the World: Studies in Honour of Manfred
Gorlach. Vol. 1&2. Amsterdam: Benjamins.

Sridhar, Kama] K. 1985.Sociolinguistic theory and non-native varieties of English,
Lingua 68, 39-58.

Strevens, Peter. 1971. World English and the world's Englishes - or whose language is it
anyway? Journal of the Royal Society of Art 130: 418-31.

Sure, K. 1991. Language functions and Language attitudes in Kenya.English World- Wide
12 (2), 245-60.

Todd, L. & W.R. O’Donnell .1980. Variety in Contemporary English. London: G. Allen
Todd, L. 1982. Cameroon. Heidelberg: Gross

1984 Modern Englishes: Pidgins and Creoles. Oxford: Blackwell

 

& 1. Hancock .1987. International English Usage. New York: NYUP

 

Thiong’o, Ngugi 1984. Decolonizing the Mind: The Politics of Language in African

Literature. Nairobi: Heinneman.

Welmers, E. 1973. African Languages Structures. Berkley: University of California

Press.

Werker, J. & Polka, L. 1993. Developmental changes in speech perception: New
challenges and directions. Journal of Phonetics, 21, 83-101.

Werker, J. &Tees, R. 1984 Cross-language speech perception: evidence for perceptual
reorganization during the first year of life, Infant Behavior and Development, &,
49-63.

Westerman, D.& Ward, 1. 1990. Practical Phonetics for Students of African Languages.
London: Kegan Paul Intern. Ltd.

163

Yang, B.G. 1996. A comparative study of American English and Korean vowels.
Journal of Phonetics 24,245-261)

164

11