View
2
Download
0
Category
Preview:
Citation preview
OLUSANMI BABARINDE
NASALISATION IN YORUBA: THE ONKO DIALECT PERSPECTIVE
FACULTY OF ARTS
THE DEPARTMENT OF LINGUISTICS, IGBO AND OTHER NIGERIAN LANGUAGES
Paul Okeke
Digitally Signed by: Content manager’s Name DN : CN = Webmaster’s name O= University of Nigeria, Nsukka OU = Innovation Centre
2
TITLE PAGE
NASALISATION IN YORUBA: THE ONKO DIALECT PERSPECTIVE
i
3
CERTIFICATION Mr. Olusanmi Babarinde, a postgraduate student in the Department of Linguistics, Igbo and other Nigerian Languages, University of Nigeria, Nsukka, has satisfactorily completed the requirements necessary for the award of degree of Ph.D in Linguistics. The work embodied in this project report is original and has not been submitted in part or full for any diploma or degree of this or any other University. ____________________________ ___________________________ PROF. (MRS.) C. I. IKEKEONWU PROF. C. N. OKEBALAMA (SUPERVISOR) HEAD OF DEPARTMENT ____________________________________
PROF. F.O. OYEBADE EXTERNAL EXAMINER
ii
4
DEDICATION TO the apple of my eyes,
ENIọLA AJọKẸ ABIGAIL my lovely daughter and my parents who taught me to appreciate language.
iii
5
TONE MARKING COVENTIONS High tone ( ́ ) Low tone ( ˋ ) Mid tone ( ˉ ) ‘always overtly left unmarked’
iv
6
LIST OF ABBREVIATIONS ASP = Autosegmental Phonology CY = Central Yoruba IAP = Initial Association Principle IPA = International Phonetic Alphabet MSC = Morpheme Structure Condition NBU = Nasal Bearing Unit NWY = Northwestern Yoruba SEY = Southeastern Yoruba SPE = Sound Pattern of English Son = Sonorant TBU = Tone Bearing Unit TG = Transformational Grammar WFC = Well Formedness Condition
v
7
LIST OF PHONETIC SYMBOLS AND NOTATIONAL CONVENTIONS
ε = ẹ j = y k ͡p = p ʤ = j ͻ = ọ ʃ = ṣ
Standard Yoruba Onko Dialect ã = ε̃ ῦ = ͻ ̃ ῖ = ῖ ͻ̃ = õ ε̃ = ẽ ʃ = ʧ
˜ = nasality [ ] = phonetic transcription // = phonemic transcription ( ) = parenthesis V = vowel C = consonant = becomes = in an environment of… Ø = Deletion
vi
8
ACKNOWLEDGEMENTS
No one ever does a research ‘all by oneself’. Every researcher owes debts of
gratitude to some people. This research work has come to be not through my ideas alone
but because of many other people who gave of their time, talent, and ideas. It is my
pleasant duty to express my profound gratitude to them.
First and foremost, I am particularly grateful to my supervisor, Professor (Mrs.)
Clara Ikekeonwu for reading through the work and making very insightful and revealing
comments. Her diligent and excellent supervision can not be divorced from whatever
merits this project achieves. May the Lord Jesus continue to strengthen you in your
service to humanity professor.
It is impossible to mention more than a fraction of those whose ideas I have drawn
on over the years. In relation to this academic programme, I should like to acknowledge a
special debt to members of the Departmental Postgraduate Board for very thoroughly
working over the seminars and thesis proposal and commenting on them in such detail.
I would particularly like to thank my respondents for their warm reception
whenever I visited and for the attention given to me even at their inconvenience. May
your children find favour in the sight of God and man.
Grateful acknowledgement is also given to Dr. Gideon Omachonu. He has been
involved throughout the work, as a friend and a constant source of encouragement (and of
pressure when required).
To my teachers at the University of Nigeria and the University of Ilorin such as
Prof. Awobuluyi, Prof. Yiwola Awoyale, Prof. (Mrs.) C. I. Ikekeonwu, Prof. (Mrs.) G. I.
Nwaozuzu. Prof. Inno U. Nwadike, Prof. C. N. Okebalama, Prof. Franny Oyebade,
vii
9
Dr. Ore Yusuf, Dr. B. M. Mbah, Dr. (Mrs.) Okorji, Dr. Kris Agbedo, Mr. Anasiudu, Dr.
Diipo Ajiboye, Dr. Peter Adesola, Dr. (Mrs) Mbah, Dr. E. S. Ikeokwu among others,
your ideas have brought me this far. Thank you all.
To my family, you are so wonderful. I am very grateful.
viii
10
TABLE OF CONTENTS Title Page…………………………………………………………………………………..i Certification……………………………………………………………………………….ii Dedication………………………………………………………………………………...iii Tone Marking Convention………………………………………………………………..iv List of Abbreviations……………………………………………………………………...v List of Phonetic and Notational Conventions…………………………………………….vi Acknowledgements………………………………………………………………………vii Table of Contents…………………………………………………………………………ix Abstract………………………………………………………………………..………....xii CHAPTER ONE INTRODUCTION
1.0 Background to the Study…………………………………………………………..1
1.1 Statement of the Problem………………………………………………………….4
1.2 Research Objectives……………………………………………………………….5
1.3 Research Questions………………………………………………………………..6
1.4 Significance of the Study………………………………………………………….6
1.5 Scope and Limitation………………………………………………………..…….8
1.6 Theoretical Framework……………………………………………………………8 CHAPTER TWO LITERATURE REVIEW
2.0 Introduction…………………………………………………………………….…13
2.1 Theoretical Studies………………………………………………………………...15
2.2 Empirical Studies……………………………………………………………..…..30
2.3 Summary………………………………………………………...………………..54
ix
11
CHAPTER THREE RESEARCH METHODOLOGY
3.0 Introduction…………………………………………..……………………………55
3.1 Area of Study…………………………………………………………………...…55
3.2 Research Design…………………………………………………………………...56
3.3 Sampling Technique…………………………………………………………….....56
3.4 Selection of Respondents……………………………………………………….….57
3.5 Instrumentation…………………………………………………………………….57
3.6 Administration of Instrument………………………………………………………57
3.7 Method of Data Collection…………………………………………………………58
3.8 Method of Data Analysis…………………………………………………………..58
CHAPTER FOUR
DATA PRESENTATION AND ANALYSIS
4.0 Introduction...………………………………………………………………..…..59
4.1 Onko Dialect Nasalization: Data Presentation…………………………………...60
4.1.1 Words with Prefixes……………………………………………..……….60
4.1.2 Disyllabic Phrases without Prefixes…………………………...…………60
4.1.3 Disyllabic Words with Prefixes…………………………………….……61
4.2 Data Analysis……………………………………………………………….……61
4.2.1 Analysis of Words with Prefixes……………………………………...…61
4.2.2 Analysis of Phrases without Prefixes…………………………………….65
4.2.3 Analysis of Disyllabic Words with Prefixes……………………………..72
4.2.4 Nasal Stability……………………………………………………..……..76
4.2.5 Denasalization Under Deletion……………………………………...……80
4.2.6 Nasal Effacement in Onko………………………………………...……..85
4.2.7 Nasalization of e/o…………………………………………………...…..85
4.3 Autosegmental Account of Words with Multiple Nasalized Sounds………..….87
x
12
CHAPTER FIVE Summary of Findings and Conclusion…………………………………...………98
REFERENCES…………..…………………………………….………………102
Appendix A……………………………………………………………….……111
Appendix B………………………………………………………...…………..114
Appendix C………………………………………………………...…………..115
xi
13
ABSTRACT
This research work examines nasalization in Yoruba with particular reference to Onko dialect of the language. Nasalization in Onko is pervasive by virtue of nasal vocal quality to the extent that it has become an identification mark of any native speaker amidst speakers of other dialects of Yoruba. This is because nasality is realised on virtually all the segments in the dialect as may be noticed when one listens perceptively to a speaker during a speech event. There has neither been any rigorous research work of any kind on Onko generally nor has any indepth study done in this dialect using the non-linear approach. As such, autosegmental phonological framework was used in the analysis of nasalization in Onko. We used the Ibadan 400 List of Basic Words and unstructured oral interview method to elicit data from the native speakers of Onko. Random sampling method was used in the selection of respondents. Respondents were selected using based on their occupation. For data analysis, we adopted non-linear method of phonological analysis which is in consonance with the theories used in the research. From the analysis, the following findings were made: apart from the fact that nasalized vowels were found word initially as against their restriction to medial and final positions in standard Yoruba, vowels /e/ and /o/ earlier found incompatible with nasalization in the New Benue-Congo languages were among the five nasalized vowels in Onko. Similarly, denasalization often realized through deletion in standard Yoruba is non-existent in Onko. Besides, the direction of nasality spread which was interestingly bidirectional was accounted for by the location of the consonants which were susceptible to nasalization within a syllable as long as such consonants were not blocked by plosives or affricates. The domain of nasalization is the syllable. In addition, it is discovered that the nasalized vowels could be found word initially in the dialect as against what obtains in the standard variety of the Yoruba language. Through the bidirectional nasality spreading and a dual process of copying and spreading, we were able to account for how nasality could be derived from words having more than one nasalized sound. Suffice it to say here that the adoption of the autosegmental phonology enables us to give an objective analysis of nasality and nasalization in the Onko dialect of Yoruba. Judging from the data presented and the accompanying explicit analyses, the work indeed achieved its stated objectives by providing convincing answers to the research questions raised.
xii
14
CHAPTER ONE
INTRODUCTION
1.0 Background to the Study
The subject of language according to Crystal (1982) has justifiably constituted an
object of fascination and a subject of serious enquiry among scholars and researchers for
years because of its functional dynamics. What Crystal implies from the foregoing is that
the complexity of language can not be captured by a single definition; otherwise, one can
very easily fall into the trap of thinking that we know all about it. But there is much more
than we think. Though all speakers of a certain language can talk to each other and
understand one another, yet no two speakers speak exactly alike. Beyond these individual
differences, the dialect of a group of people often shows some systematic differences
from that used by other groups. Speaking to one another is so much part of normal life
that they often seem unremarkable. Yet, as in any scientific field, the curious investigator
finds rich complexity beneath the surface.
During speech event, the standard variety of the Yoruba language will invariably
depict the Onko (Oke-Ogun) speaker’s version as deviating from the rule of the standard
variety. Universality of language notwithstanding, generative linguists believe that some
aspects of language are not universal. It is against this backdrop that this work seeks to
bring to light some salient phonological attributes like nasality and nasalization process,
and their manifestation in the speech pattern of Onko speakers of the Yoruba language.
A language is composed of its dialects. Oyelaran (1978) and Adetugbo (1967 and
1982) made an important contribution to the dialectology of Yoruba. They both used
sociological and linguistic evidence to classify Yoruba dialects into three major linguistic
15
groups viz: Northwestern Yoruba (NWY), Southeastern Yoruba (SEY) and Central
Yoruba (CY). By their classification, the Onko (Oke-Ogun) dialect is classified as
Northern part and the dialect is spoken in well over ten speech communities like
Saki, Okeho, Iganna, Tede, Sepeteri, Ago-Are, Irawo, Igboho, Ago-Amodu and Iseyin all
in Oyo state. Oyetade (1995) notes that the whole area of Oyo north division should
constitute a linguistically homogeneous area because of certain peculiarities. For
instance, the area is characterized by a peculiar nasalization. The [ã] and [ũ] forms in
other areas of Southwestern and Central Yoruba correspond with [ε ̃] and [ͻ ̃] respectively
as:
nnkã kã nnkε ̃ kε ̃ ‘something’
nãã nε̃ε ̃ ‘the’
dũ dͻ̃ ‘tasty’
rũ rͻ̃ ‘to smell’
Beside nasalization, there is the presence of the voiceless palato-alveolar affricate [ʧ ] in
the speech of the people from these areas unlike in other areas:
iʃέ iʧέ ‘work’
ʃe ʧe ‘to do’
Every speaker unconsciously has some knowledge of the sound pattern of his language.
Fromkin and Rodman (1981) say it is the phonological knowledge of a speaker that
enables him to produce meaningful sounds of his language and to recognize immediately
the different phonetic sounds or strings of sounds that do not and cannot form any
meaningful unit in his language; it further allows them to appreciate that there is a
difference between the function of sounds of their language and other languages.
16
However, there has neither been any rigorous research work of any kind on Onko
generally nor any indepth study done on this dialect using the non-linear approach.
Using autosegmental phonological (a non-linear approach) is advantageous in accounting
for the nasalization in Onko because it shows explicitly how nasality is realized and
spread to other vocalic and transparent consonantal segments. Autosegmental phonology
reveals clearly the domain of nasality and when/how its spread can be blocked.
Additionally, adopting a linear approach would not do justice to the phenomena of
nasality stability and relinking in Onko since any phonological model which adopts a
linear approach treats speech as only a linear sequence of individual segment with arrays
of phonological processes taking place at a one single level. Such linear approach is
grossly inadequate to analyse nasalization as a phonological process.
Nasalization may be a general property of speech according to Clark et al (2007), for
reasons of individual articulatory habit, or dialect type. Following this insight, it is
perceived that nasalization in Onko is pervasive to the extent that it has become a
badge/identity of some sort in identifying any native speaker amidst speakers of other
dialects of Yoruba. This is because nasality seems to be realized on almost all the
segments in the dialect as may be noticed when one listens perceptively to a speaker at
any speech situation. The speech situation in Onko is aptly captured by the term ‘nasal
twang’ being first used by Abercrombie (1967). According to him, it is the result of
keeping the velum lowered almost all the time one is speaking. This nasal voice quality,
according to him, could be an institutionalized feature common to a group of speakers.
There is no doubt that a special voice quality is recognizable as characteristic of certain
languages or dialects. The term voice quality refers to those characteristics which feature
17
more when a person is talking: it is a quasi-permanent quality running through all the
sound segments that issue from his mouth. These characteristics do, naturally, include
some that have their origin in the anatomy of the larynx.
Hyman (1985) proposes that we should regard nasality as an autosegment capable
of being mapped unto a lexical morpheme or stem. Within the autosegmental model, the
occurrence of a particular feature on one tier does not preclude it from occurring on
another tier. A particular feature may appear on different tiers within phonological
representation, and since the tiers are independent of one another, what affects the feature
on one tier may not affect it on another tier.
The autosegmental theory is used in accounting for the phenomenon of nasality in
Onko. This is because these theoretical models represent a considerable advancement
over the limited strength of generative phonology (SPE) apparatus, and they equally have
certain salient empirical advantages, hence their relevance in nasalization in Onko.
1.1 Statement of the Problem
There is not much variation in the nasal manifestation in the various Onko dialect
speech communities speaking Onko (Oke-Ogun) dialect speech comm. The nature of
nasalized vowels in the dialect further makes the nasalization in the dialect more
interesting. Unless one listens very perceptively, one may find it quite difficult to
distinguish between oral and nasalized vowels as it appears to be that there is no single
utterance in the speech of any speaker of this dialect that does not somehow exhibit some
trace of nasality. The foregoing inspires the desire to make a linguistic inquiry into
what/which linguistic feature, possibly phonological, account for this seemingly rare and
18
unique phenomenon in all its manifestations. Since linguistics is an empirical science
which does not rely on assumption, there is the need for thorough inquiry into the
analysis of intricacies of nasalization in the Onko dialect leaning on theoretical
framework of autosegmental theory.
Some of the principles of autosegmental phonology are about an attempt to
supply a more adequate understanding of the phonetic side of the linguistic
representation. The autosegmental theory is a non-linear approach. It is a theory of
suprasegmental representation, hence its appropriateness in accounting for issues that
border on nasalization.
1.2 Research Objectives
The goal of the theoretical study of a language is the construction of a grammar.
Therefore, among the objectives of this study is that after a careful examination of
nasalization in the said dialect, it is considered essential that readers should be able to
evaluate critically the factual claims about nasality in the Yoruba language with particular
reference to Onko dialect. The foregoing notwithstanding, the main objective of this work
is to do an indepth analysis of nasalization in Onko dialect within the scope of
autosegmental theory. Specifically, however, the objectives are to:
(i) present areas of differences between nasalization in the standard Yoruba and
Onko dialect;
(ii) account for the autosegmental analysis of nasalization in Onko dialect;
(iii) account for nasality spread in Onko dialect;
19
(iv) explain how nasality is derived on lexical items with more than one nasalized
sound;
(v) explain the realization of denasalization and nasal effacement in Onko dialect.
1.3 Research Questions
To achieve the set objectives, the following leading questions shall be answered.
This will certainly go a long way in making the task of this research work worthwhile:
(i) To what extent does nasalization in Onko differ from that of the standard
Yoruba?
(ii) To what extent does autosegmental phonology capture the process of
nasalization in Onko dialect?
(iii) How can the nasality spread be objectively explained in Onko?
(iv) How do we derive nasality on lexical items that have more than one
nasalized sound?
(v) To what extent are denasalisation and nasal effacement realizable in Onko?
1.4 Significance of the Study
This study is aimed at extending the problem-solving capacity of autosegmental
phonology to analyzing nasalization; and also to reveal how traces of nasality perceived
on every utterance could be explained using the theory.
This research however is important especially when one considers the significant
role that phonological investigation plays in the study and understanding of language. It
is this aspect of linguistic study that analyzes the speech activities which go on in
20
everyday speaking, resolving the integrated speech complexity into its different aspects,
breaking down the overall activity into its component details, explaining how the
deceptive simplicity of everyday (speech) is achieved.
This work will undoubtedly bring to light some unnoticed phonological treasures
embedded in the Onko dialect for the overall update of available data in the Yoruba
language thereby expanding the scope of the language theoretically. Besides, a study such
as this will hopefully repair some of the damage that may have been done in language-
teaching situation, through its objective descriptive analysis of issues. The study will
equally reveal to us the changes that are taking place in language from area to area and
among different categories of speakers of the language.
Additionally, a proper understanding of nasalization in the Yoruba language may
not only add to our insights about nasalization in the language but also to our
understanding on nasalization as a general linguistic phenomenon. It is therefore not an
overstatement that findings in this study will have implications for theoretical matters
such as the nature of language itself, language learning and language-using abilities of
human beings.
Moreover, the study will contribute immensely to the formulation of theories in
the study of Nigerian languages more so that these languages have revealed a lot about
the phenomenon of language as human specific attribute.
Furthermore, this kind of research work will or may serve as a point of reference to
other researchers or students who may perhaps pick interest in this area of linguistic
investigation for the purpose of filling some gap in their knowledge.
21
1.5 Scope and Limitation
This research work focuses on nasalization in the Yoruba language with particular
reference to Onko (Oke-Ogun) dialect using the autosegmental framework. Topical
issues to be examined in this regard include an explicit analysis of nasalization as a
concept, its manifestation in Onko dialect with a view to revealing those salient attributes
that account for the perceived nasality in almost every utterance in the speech
community.
The work gives an analysis of the orientation, assumption and strength of
autosegmental framework in the objective analysis of nasalization. The work equally
explored the phonological processes like elision and assimilation for their great insight
and input to nasalization process. Had our primary concern been the Yoruba
phonology, we would have investigated some other aspects of phonology such as tone
and syllable structure in a considerable detail for their theoretical importance to the work.
Their exclusion, however, is essentially because this work limits itself to a particular
phonological phenomenon within the study area. We equally felt that incorporating those
other complex aspects here would unnecessarily over-labour this work since they can not
be effectively handled at least to avoid haphazard analysis. Time equally poses some
constraints during the course of the study.
1.6 Theoretical Framework
Autosegmental Phonology is a theory of non-linear phonological
representation. It was developed out of research in Generative Phonology at MIT in the
mid and late 1970s, as a response to certain problems in the phonological theory of that
22
time. Autosegmental phonology was initially developed in response to the challenge of
developing an adequate theory of tone. Its immediate source of inspiration was the work
of Williams 1971 and Leben 1973; these were the first to introduce non-linear structures
into generative phonology in their treatments of tone systems in West African languages
such as Margi, Igbo and Mende.
Autosegmental Phonology received its name from the doctoral thesis of John
Goldsmith in 1976 in his work entitled Autosegmental Phonology. According to
Goldsmith, the Autosegmental phonology arose out of certain inadequacies of the SPE
model that were brought to light explicitly and implicitly by William’s (1971) and
Leben’s (1973) work. The most glaring problem was the nature of ‘contour-toned’
vowels – that is, vowels whose surface tone is rising or falling, a situation that can often
be shown to be the result of a concatenation of Low and High level tones. How can a
single segment bear or carry two tonal specifications in sequence? The principal
innovation of autosegmental phonology, as presented in Goldsmith 1976, was the idea
that tone mapping rules do not merge tonal and segmental representations, but associate
their elements by means of formal entities known as Association Lines. In this
framework, phonological representations consist of parallel tiers of phonological
segments, both tonal and segmental.
Goldsmith made suggestions on modifications of the theory of generative
phonology with the introduction of parallel tiers of segments or autosegments to solve
some formal problems in the current theory (generative phonology). To show the
effectiveness of the theory, a detailed analysis of Igbo, a tone language, is used to show
that phonological representations are no longer seen as simple rows of segments, with
23
arrays of phonological processes taking place at one single level. Rather, phonological
representations are seen as arrays of items organized on different tiers. Goldsmith’s
understanding of what he refers to as ‘absolute slicing hypothesis’ led him to what he
describes as a multilinear phonological analysis in which different features may be placed
on separate tiers. The tiers are connected to each other by association lines which allow
for the fact that there may not always be a neat one-to-one mapping between tiers. The
lines provide flexible way of associating autosegment with segmental feature such that
the autosegment is not swallowed up within a strictly segmental notation. The notion of
tiers is reminiscent of Firthian prosodic phonology. It is also found in CV phonology of
Kahn (1980) on syllabic organization and McCarthy on Semitic languages (McCarthy
1981).
Some of the principles of autosegmental phonology are about an attempt to supply a
more adequate understanding of the phonetic side of the linguistic representation. Viewed
in this light, it is a framework which aims at the same logical level as the idea that
phonetic representation is a linear sequence of atomic units – which could be called
segments; it is at the same level the proclamation that these atomic units are cross-
classified by distinctive features. Autosegmental phonology is a particular claim, then,
about the geometry of phonetic representations; it suggests that the phonetic
representation is composed of a set of several simultaneous sequences of these segments,
with certain elementary constraints on how the various levels of sequences can be
interrelated or associated.
Autosegmental framework is known for the analysis of tone, like Goldsmith
himself remarks that the system of analysis was originally suited to fit the intricacies of
24
African tone languages (Goldsmith, 1979). Various papers developing the autosegmental
perspective (Clement 1977, 1981, 1984, 1985) show how similar notation can be applied
to other phenomena like vowel harmony and nasality where their spread across segmental
boundaries can be treated analogously.
Nasality has a special feature in Onko. It is one of the phonological features of the
Onko dialect. The work examines the process of nasalization in Onko. It claims that
nasality is most insightfully captured within the autosegmental framework given its
principles and its efficiencies as regards topical issues like stability, re-linking, spreading,
deletion and effacement. All these can most conveniently be captured within this chosen
framework. It also provides further evidence for the segmentalization of the feature
(Nasal) onto a separate autosegmental tier, independent of other segmental features.
The arrival of autosegmental theory has reduced the dependence on the rules in
describing the phonological structure due to their advancement on SPE model coupled
with the objective and explicit account of suprasegmentals which were earlier
haphazardly handled in SPE. Being a theory of suprasegmental representations, there is a
good reason to agree that autosegmental theory is more adequate in analyzing
nasalization. Not only because the goal of autosegmental phonology according to
Goldsmith (1990) is to formulate general principles of phonological descriptions which
can be sufficiently applied to a greater number of languages but its use especially in this
work objectively accounts for the detailed analysis of nasality as a phonological feature
and nasalization process in Onko speech dialect beyond what a linear approach such as
SPE can explicitly explain. For example, the use of autosegmental phonology throws
25
more light on the nasality spread, nasality stability and nasality re-linking. All these
attributes of nasality could not be handled using linear approach.
26
CHAPTER TWO
LITERATURE REVIEW
2.0 Introduction
A nasal consonant is a type of consonant segment which like a stop, is produced by
a stricture of complete closure; a nasal segment however, unlike a stop, has no
simultaneous velic closure. The airstream, therefore, though prevented from passing
through the mouth, is not dammed up; it is entirely diverted through the nose. We refer to
nasalization as a phenomenon which arises when a sound (oral) is pronounced with some
degree of nasality.
One result of autosegmental phonology according to Clements (1997) is to
provide an avenue that allowed linguists to express their insights in a revealing way. An
example of this is how autosegmental phonology brought an illuminating approach for
expressing views field researchers had about tone, for example, the representation of tone
spreading. In the Aghem language, a prefixal high tone (H) spreads onto a following low
tone (L) root, thereby creating a HL falling tone:
/é-zù/ [é-zû] ‘to skin
In pre-autosegmental phonology, Hyman and Schuh (1974) express such a rule roughly
as:
H L H H͡L
Goldsmith’s (1976) autosegemental representation, on the other hand, is shown
below:
27
V C V
H L
Goldsmith’s representation clearly indicates that there is a single H feature involved in
tone spreading, Hyman and Schuh’s formulation implies that a H feature is being copied
onto the following vowel as in the SPE generative model’s conception of assimilation.
Comparatively, there is no gain saying that the application of autosegmental theory
has some explanatory force over and above the SPE generative model. For one, it shows
convincingly how to account for the tone spreading in Aghem.
2.1 Theoretical Studies
The linguist’s concern with all languages and dialects derives from the proclaimed aim
of his subject: the construction of a scientific theory of the structure of human language.
Theory is the backbone of any academic inquiry. Phonological theories have contributed
a lot to the development of the study of African languages. Some phonological issues in
African languages like tone, nasalization, vowel harmony to mention but a few have
provided grist for theoretical mills showing how theories benefit the study of languages
by leading to more insightful discoveries.
However, linguistic theory has provided part of the training necessary to do good
work. In the most successful situation, the result has been a give-and-take between theory
and description; and between general and African Linguistics. Considering what the
potential contribution of linguistic theory should be to language specific work, Steel and
28
Threadgold (1987) note that theory serves both as a guide to discovery and as a tool for
expressing insights. That is, it can deal with the unknown as well as the known
phonological issues. In other words, a theory can help direct the research towards
important discoveries. It might lead us to ask questions we might not have otherwise
asked and data we might otherwise not have considered. In addition, theoretical
awareness can help the researcher see connections that might otherwise not be made, as
well as pinpointing problems that might have been overlooked. A particularly clear
example of this is how autosegmenal phonological theory provide an explicit framework
for expressing intuitions about nasalization.
Language structure is amenable to three different levels. These are: sound level
(phonetics and phonology), form and structure level (morphology and syntax), and
meaning level (semantics). The study of the substance of sound is, therefore, of more
central concern to the phonologist than is the investigation of graphic substance.
Language, according to Radford (1981), at every level of organization, is highly
structured in a highly complex and highly systematic way. However, in a very general
term, we will see that the analysis of human language at each structural level is stated in
terms of:
(i) a set of discrete units of some sort, and
(ii) rules combining those units.
Interest in pronunciation is far older than the pursuit of phonetics and phonology
as academic subjects. Several centuries before Christ, Indian scholars were devoting
themselves to the description of Sanskrit and achieving remarkable accuracy in
articulatory phonetics. Although their primary concern seems to have been to maintain
29
the correct pronunciation of what was already becoming a classical language, their
observations about points and manners of articulation and other aspect of pronunciation
reveal an interest that qualifies as scientific in the best sense of the term (Allen 1953).
Phonology according to Oyebade (1998) is the scientific study of the arbitrary
vocal symbols used by man in speech for communication and systems by which vocal
sounds produce intelligent and meaningful utterances. Ikekeonwu (1996) observes that
for some the practice of phonology has revealed important variations. We have shifted
from classical phonology to generative, and from classical generative period to some
more recent autosegmental phonology, metrical phonology, prosodic phonology and
optimality theory. Each of these theories, on the basis of evidence from different
languages has modified, redefined and clarified issues on our understanding and
description of language.
The foregoing knowledge of the aim of phonology led to its maiden theoretical
orientation of the scientific analysis of speech sounds which was prominent in the 1920’s
to the mid 1960’s. In line with Sommerstein’s (1977) submission, this orientation seeks
to answer these questions:
“what phonic features
(a) serve in the language under investigation, or
(b) are capable of serving in natural language, to distinguish one utterance
from another’’.
Phoneme has been defined variously by many scholars like Daniel Jones (1967) who sees
phoneme as a family of sounds in a given language which are related in character and are
used in such a way that no member ever occurs in a word in the same phonetic context as
30
any other member without bringing about semantic change. This class of phonemes will
have as members a group of allophones that appear in complementary distribution based
on one phonetic distribution or another. The above phonemic account, notes Lightner
(1980), stems from the insights and orientation of a school of linguists called
structuralists. Their model is also rather referred to as taxonomic grammar because of its
crucial emphasis on classification of sounds into families of phonemes and their
allophones and stating their distribution.
One of the ironic tenets of classical phonemics is the biuniqueness principle. This
principle, according to Bybee (2001) insists that once a sound has been identified as a
phoneme in a language, it can not later be said to be an allophone in that language. This
condition has created a lot of problems. For example, the sounds [s] and [k] can be
established as phonemes in English since they occur in minimal pairs ([sIk]: [k Ik] ‘sick,
kick’). However, in some environment they may be described as variants of the same
phoneme:
[kritik] ‘critic’
[krItIsIzәm] ‘criticism’
It can be argued that criticism is derived from the collection of two morphemes [kritik]
and [-Izәm]. This is not an isolated case as the same thing occurs in such pairs of words
as ‘fanatic: fanaticism’ electric: electricity: if the two sounds [k] and [s] occur in
complementary distribution such that one can predict the environment in which each
occurs, then [k] and [s] must be considered allophones of the same phoneme. However,
the two sounds had earlier been established as discrete phonemes in the same English
language. This proposal, of course violates the be-uniqueness principle. Oyebade (1998)
31
notes that classical phonemics avoided this problem by regarding the level at which [k]
and [s] are in mutually exclusive environment as the morphophonemic level and not the
phonemic level which the bi-uniqueness condition adopts. This problem made Morris
Halle and other generative phonologists to feel that maintaining the bi-uniqueness
condition will lead to establishing an artificial separation of levels into phonemic and
morphophonemic which has no empirical basis for actual language. They contended that
an ideal theory of phonology must not only explain the type of issue above but also
account for other revealing phenomena about language. They rejected bi-uniqueness
condition and suggested that phonemic analysis be carried out that will be faithful to the
data by stating when a sound acts like a phoneme and when the same sound is a variant
form in another instance. This position was called systematic phonemic position.
Another point of modification was in terms of what the smallest level of
phonemic analysis should be. Taxonomic phonemics considered the sound as the minimal
unit of phonemic analysis. One of the major flaws in such position, according to
Langacker (1972), is the fact that it ignores a crucial linguistic fact that pairs of sounds
are different from each other in different ways. For instance, the minimal pairs ‘glow:
grow’ and ‘pit: bit’ in English differ in terms of a pair of sound each [I]: [r] and [p]: [b].
However, the fact that the difference between the first pair and the second pair is not the
same is ignored. The difference can not be captured if only the sound is seen as the
minimal unit of phonemic analysis.
Inadequacies such as these and lots more led to the proposal for major
modifications to the phonemic principle of classical phonemics. These modifications had
significantly led to the formulation of a new theoretical framework with different
32
orientation and goals. This framework was called Transformational Generative
Phonology. According to Oyebade (1998), classical phonemics could not do justice to the
phenomenon of speech. Rather, the model only attempts the phenomenon slightly leaving
the entire bulk of empirically interesting puzzles unsolved. Riemsdijk (1985) notes that
the development and characteristics of generative grammar also include a theory of sound
structure or phonology. Chomsky and Halle (1968) observe that a universal grammar is a
system of conditions that characterize any human language. Generative phonology is a
theory which builds on the insights of taxonomic phonemics even while remodeling the
focus of phonological analysis. It was given its full status in Chomsky and Halle’s book
entitled The Sound Pattern of English (SPE) in 1968. It is a part of theory of language
called generative grammar. For Ikekeonwu, (1996), the theory is the introduction of a
new framework to phonological theory and analysis whose goal is to explore and further
understand the nature of linguistic knowledge.
Yul-Ifode (1998) asserts that generative phonology belongs to a new school of
linguistics - Transformational Generative Theory (TG). This theory focuses on the
linguistic competence, in which the underlying representations are converted into surface
representations by the application of rules. The major motivation for this theoretical
framework was the clash between theoretical assumption and linguistic data under theory
of classical phonemics. This theoretical model according to Chomsky and Halle (1968)
sees speech as sequences of discrete segments that are complexes of a particular set of
phonetic features, and that the simultaneous and sequential combinations of these features
are subject to a specific constraint. This framework characterizes the utterance as a
bundle of unordered features arranged in an ordered sequence. Goldsmith (1976) refers to
33
this as the absolute slicing hypothesis. According to him, this is the claim that speech can
be exhaustively sliced into segments which consist of unordered bundles of features
which are linearly ordered. For instance, the English word ‘cat’ is made up of the
segments k-æ-t and these segments in turn are made up of the features:
k = voiceless, plosive
æ = low, front, unrounded vowel
t = voiceless, alveolar plosive
The utterance can be represented in the following ways:
k æ t
voiceless low voiceless
velar front alveolar
plosive unrounded plosive
vowel
The features are unordered in that any of the features for each sound segment can be first
written without any restriction:
k æ t
plosive low alveolar
voiceless unrounded plosive
velar vowel voiceless
front
34
However, if the sequence is restructured in a way that the bundle of features for [t] is
followed by [æ] and finally by [k], the output will surely yield another word different
from ‘cat’. This, then, justifies why the sequence is so unordered. So, features are a
significant part of SPE generative model. These features must be distinctive and must be
specifiable phonetically. This is to say that any feature proposed as linguistically
distinctive must be relatable to the way the sound {or class of sounds} to which it is
referred is produced. Brame (1974) adds that this is an important check on the porosity of
feature because generative phonologists argue that all human languages draw from a pool
of highly limited number of features in the construction of their sound systems. This
phonetic specifiability of the distinctive features must be able to distinguish between any
two phonetically related sounds. Morphemic relevance is yet another condition for the
distinctive feature. By this, the acceptable features must be functionally relevant in one
language or the other. The main preoccupation of the SPE model was, according to
Katamaba (1986), with rules that modify feature specifications in various ways; with the
manners in which rules mapping deep (or underlying) onto surface representations relate
and with the extent to which underlying surface representations may differ from surface
representations. Ikekeonwu (1996) submits that in generative phonology, the emphasis is
shifted from identifying the phonic properties that bring about contrast in different
languages to describing those phonic elements that may be used distinctively across
languages. She points out that generative phonology appears with two levels of analysis –
the surface level or systematic phonemic level and the underlying or systematic phonetic
level. Whereas the underlying level captures the details of pronunciation of segments and
their modifications whether or not they are distinctive, the surface level contains all the
35
distinctive segments and whatever modifications that are non-distinctive. The two levels
are related. The surface level is derived from the underlying through phonological rules.
However, the age-long general opinion among phonology scholars was that
phonological representations consist of both segmental and suprasegmental
representations. The former were taken to be made up of consonant and vowel sounds as
well as empty segments which are referred to as syllable, morpheme, word Katamba
(1989) and Dinnsen (1976). In the SPE generative phonology, it was held that some
phonological phenomena like tone, vowel harmony, stress, nasalization were super-
imposed on the segmental layer. Then, both segmental and suprasegmental elements were
thought to be organized in a row one after the other. And, for a long time in phonological
history, the validity of the assumption that phonological representations consist of linear
segmental and suprasegmental layers was overlooked. Generative phonology, notes
Fudge (1973) characterizes the utterance as bundles of unordered features that were
organized in an ordered pattern. But the effect of this description is that all of the
articulatory and acoustic signals that represent an utterance are organized in linear order.
So, Goldsmith (1976) notes that ever since segments have been part of phonology, there
have been phonological phenomena that transcend segmental classification or description
and as a result there have been suprasegmentals. Besides, the question of how two levels
of segmental and suprasegmentals got related to each other never occurred to anyone or
probably was lackadaisically treated.
Due to all these cloudy and unsettled phonological issues, in the 1970’s a number
of studies and researches centered on the connection between segmental and
suprasegmental representations. The outcome of such inquiries showed that the
36
assumptions that SPE phonological model was based on were rather shady and
questionable. Katamba (1989) notes that such inadequacies first occurred in the
discussion of tonal representation. An important question which was brought up was
whether features of tone like [high], [low] should be seen as features attributable to
vowel, in the same token [front] or [rounded] are seen as vowel features. Or, should we
regard tonal features as distinct from the segmental representation of vowel?
Furthermore, should tone still be indicated with the conventional marks of ( ̀ - ʹ ) for
high, mid and low tones respectively in order to show its (tone) status? These and more
were the findings of various studies undertaken by phonologists. So, it was observed that
the SPE generative theory and even the structuralist phonological theory before it were
not right in assuming that phonological representations are linear especially with regard
to sound segments some of which bear suprasegmental properties arranged in a neat
sequence.
The generative phonology model that seeks to analyze the suprasegmentals (that
is, those phonological features that transcend beyond segmental level like tone, stress,
nasalization, vowel harmony) has been labeled a non-linear generative phonology theory.
Hyman (1975) observes that the issue of whether certain phonological phenomena should
be analysed segmentally or suprasegmentally has been a major issue to the generative
phonologists. Infact, some of the main issues in phonological model have been argued on
the basis of suprasegmental phenomena. The term suprasegmental according to Hawkins
(1984), is used to refer to both phonological and grammatical units larger than the
segment. Phonological suprasegmentals are those which are defined in terms of the sound
segments of which they are comprised. To the above observation raised by Hyman,
37
Goldsmith (1976) proposed some revealing answers. He said the division of the speech
continuum into segments may proceed in different ways in different languages.
Buttressing this proposal, he cites an instance of nasalization which is usually a feature of
nasal consonants only, still in some languages it can be a property of the syllable or even
the word as a whole. Ikekeonweu (1996) notes that a tone may not necessarily apply to a
single segment. It may apply to a whole syllable or to sequence of two or more segments.
It then becomes rather difficult to “slice out’ a segment, as it were, and tie down the tone
to it, just as distinctive features are specified under the affected segment.
Also, the rate of vibration of the vocal cords which determines the pitch of a sound can
be a property of an individual segment, of a syllable or even an entire word. According to
Katamba (1989) a central claim of the new framework formulated by Goldsmith is that,
in the deep sense of it, the various articulatory criteria like aspiration, nasalization and
tones are autonomous and the articulation that result from them are, in principle,
independent. In an SPE inspired model, a word like ‘mad’ would be represented as:
While the assignment of the various features to discrete segments above might
look plausible, the same procedure could not be extended to the feature [+ voice] in this
word since [+ voice] is a property of the entire word. Nor could it be extended to the
+ cons –cons + cons
+nas -nas - nas
+lab -lab - lab
-cont +cont - cont
- cor - cor + cor
m æ d
38
analysis of pitch here because it too could not be vertically sliced and allocated to a single
segment, without any leakage into adjacent segments. This evidence, according to Lehiste
(1970), undermines the fundamental claim of the slicing hypothesis. Infact, segmental
features can often extend over more than one segment. In autosegmental phonology,
notes Clements (1976) phonological representations are no longer seen as simple rows of
segments, with arrays of phonological processes taking place at one single level. Rather,
phonological representations are seen as arrays of items organized on different levels or
tiers using association lines which link items on different tiers. Leben (1973) in a work
entitled Suprasegmental Phonology notes the problem inherent in the use of the SPE
approach to account for features that transcend individual segments. SPE cannot handle
these features because it adopts a linear method.
One of the major functions of phonological theory is to establish the language-
specific features as well as universal principles with the linking of these autonomous
parameters. Autosegmental phonology is a name intended to highlight the fact that
potential autonomy of the various phonological parameters is taken as significant. The
aim of autosegmental phonology is to deal with the consequences of generative
phonology of multi-linear phonological analysis and representation. That is we let go of
the assumption that phonological and phonetic representations consist of a single string of
segments. Instead we set up underlying and surface forms consisting of parallel strings of
segments arranged in two or more tiers. Features are distributed over the various tiers in
the sense that no feature may appear on more than one tier. The earliest model of
autosegmental phonology focused on the relationship between tones and segmental
properties. According to Harris (1985), this accounts for the reason why the earliest
39
models were represented in terms of tonal tier and segmental tiers. Non-linear phonology
assumes that the features that represent each sound in an utterance are situated on
different independent autosegmental tiers. These independent layers are connected
together by a universal convention called Well- Formedness Condition (WFC). These
constraints were severally modified between 1976 and 1990. Eventually, it was agreed
that the association convention that links different tiers together can be stated thus
according to Durand (1990):
(a) Mapping: Associate vowels with tones in a one-to-one correspondence from
left to right until we exhaust tones or vowels.
(b) Dumping: If after mapping some tones are left free or unassociated, link them
to the last vowel which are still free,
(c) Association lines are not allowed to cross.
This could be illustrated with the sketch below:
The convention could add or delete association lines as appropriate at any particular point
in time during derivation. For instance, in the above sketch, a tone-bearing segment is
deleted but the high tone on it remains stable and then re-links to the succeeding vowel to
the right. This further justifies the principle of autosegmental phonology that what affects
L
өa e
L H L L L L
L L
Ø e ε
ε am ea mε a
H
40
the segmental tier does not affect the tonal tier since each tier is independent of the other.
Besides, the function of WFC according to Goldsmith (1976) is not to police
phonological representations rather, it is to be seen as a statement of the unmarked,
neutral, normal state of affairs.
In the above sketch, the inherent floating tone relinks itself to an adjacent tone
bearing unit (TBU) rightward following the elision of the vowel segment bearing it. The
tonemic movement by floating tone in search of where to hang is called docking. This
phenomenon of floating tone justifies one of the telling evidence from non-linear
representation known as tonal stability. Tone does not get deleted when the segment is
elided, rather it shifts its location and shows up on some other vowel. However, for non-
linear framework to be seen as a true linguistic insight, it must prove itself capable of
solving other phonological issues beyond the scope of tones. This is what motivated
Clements (1981) and Oyebade (1998) to extend the model to explain the interesting
linguistic issue of vowel harmony. Beside vowel harmony, the non-linear phonological
framework has equally been adjudged more effective in the analysis of reduplication and
nasalization (see Clements 1976, 1977, 1984; Leben, 1973; Durand, 1986; Oyebade,
1998).
However, the autosegmental model has undergone some modifications due to
some new emerging facts. For one, the Initial Association Principle (IAP) of this
framework proposes that tone association starts from the left to right on TBU. However,
this proposal cannot work for languages such as Hausa and some other languages which
start tones application from the right to left. However, this principle is suitable for the
Yoruba language as evident in this work. In addition, the inclusion of the skeletal tier,
41
that is, to sketch the nature of the individual sound segments, is another modification.
Oyebade (1998) observes that the skeletal tier is good for languages that have consonant
gemination and vowel length. The introduction of this tier provided strong opportunity to
adjust and accommodate more facts from new languages that are just being discovered. It
is so called skeletal tier because it is a foundation upon which any utterance is built.
The concept of tiers is also found in CV phonology, which arose from work by
Kahn on syllabic organization and by McCarthy on Semitic languages (Kahn, 1980,
McCarthy, 1981, Clement and Keyser 1983, Kenstowicz 1994). The original contribution
of CV phonology is the postulation of a CV tier, a tier of C and V ‘slots’ which are filled
by segments, Often segments can be mapped directly onto these CV positions. In this
regard, CV phonology is a contribution autosegmental notation. We can thus represent
English man and landed as;
but just as autosegmental phonology allows other kinds of mapping, so CV phonology
offers the possibility of capturing the special nature of complex segments that
traditionally require structural interpretation.
Suffice it to note that analysis which shall be done in this work at the later stage
will not be unconnected with the feature geometry. This is to allow us look at the phonic
features of sounds in Onko that make them different from those of the standard variety of
the Yoruba language. Crystal (2003) notes that feature geometry in non-linear phonology
C V C C V C C V C
m æ n l æ n d ә d
42
is a model of the ways in which features are organized in phonological representations.
The term, according to Clark, Yallop and Fletcher (2007) has become common in
discussion of the way in which phonological features are grouped or structured. In the
light of this concept, bundles of distinctive features are internally structured and that the
behaviour of segments according to Kenstowicz (1994) can be understood from the
elucidation of this internal feature structure. Various approaches to feature geometry look
at the non-linear relationship between features, and at the way they can be grouped into a
hierarchical array of functional classes. By feature we mean an element which plays the
role corresponding to phonological rule and to true suprasegmentals. The feature values
are arrayed on separate tiers, where they may enter into non-linear relations with each
other. Features are organized into hierarchical arrays, in which each constituent functions
as a single unit in phonological rules.
It is sufficient to say here that the works on autosegmental phonology is a true
reflection and continuation of the traditional work of generative phonology that was
published in Chomsky and Halle’s Sound Pattern of English (SPE) in 1968. This is
because the justifications for the theoretical changes in the framework of phonology
which resulted into autosegmental phonology were placed on contention that made, and
continues to make sense within the very theoretical heart of generative phonology. No
important shift in theoretical goals needs to be made to see why the framework is
superior to the analysis made by classical generative phonology. There is a good reason
however, to agree that autosegmental phonological theory is more adequate, explicit and
objective than classical generative phonology. The basic goal of autosegmental model
according to Goldsmith (1990) is to formulate general principles of phonological
43
descriptions which can be objectively and sufficiently applied to a greater number of
languages. According to him:
‘… success may be hard to measure, but it consists largely in the ability the analysis
grants us to see connections in various ways’
2.2 Empirical Studies
In traditional phonetic terminology, a nasalized segment is one whose production
involves flow of air through both the mouth and the nose. The oral cavity still acts as a
resonance chamber for the sound. Resonance according to Goldsmith (1989) refers to a
sound being deep, clear, and continuing to echo. Leben (1973); Cauty (1978) and
Valenzuela (2003) suggest that nasalization may be considered as a suprasegmental
feature in some languages. Both vowel harmony and nasalization were seen to be
prosodic in the British School of Linguistics.
Akamatsu (1991) shows that nasal sounds are produced when the egressive airstream
meets obstruction at a given point in the oral cavity and is channeled into the nasal cavity
through which it goes out following the lowering of the soft palate. Katamba (1989)
opines that it is the neat synchronization of velic closure along with the other articulatory
parameters like phonation, manner and place of articulation that ensures a clear cut
distinction between the oral and nasal consonants. He notes that any slight leakage of air
past the velum will cause some nasalization. Abercrombie (1967) notes that nasalization
is not a secondary articulation either in the case of vowels or consonants.
For nasals there is always an obstruction of complete stricture in the mouth.
Acoustically, nasal stops are sonorants in that they do not restrict the escape of air and
44
cross-linguistically are nearly always voiced. However, nasals are also stops in their
articulation because the flow of air through the mouth is blocked completely. This
duality, a sonorant airflow through the nose along with an obstruction in the mouth,
means that the nasal stops behave both like a sonorant and like an obstruent.
Generally, nasalization has been noted to obtain as a result of the position of the
velum or soft palate (Abercrombie, 1967). And further on the action of the velum,
Gimson (1980) highlights that the palate can either be held in its raised position to effect
one channel escape of air through the mouth; be lowered, as in normal breathing, to allow
a dual outlet of airstream – the oral and the nasal cavities – or be lowered (with an
obstruction in the mouth) to afford a nasal outlet; which brings about oral, nasalized, or
nasal speech sound respectively.
Nasal consonants with a complete mouth closure are common in almost all
languages, and may be exemplified by English /m/, /n/, and /ŋ/, the nasal counterparts of
the oral /b/, /d/, and /g/ (mob /mɔb/; bomb /bɔm/; nod /nɔd/; don /dɔn/; gang /gæŋ/). /ŋ/ is
not found at initial position in English words. All nasal consonants are continuants as the
egress of air is never completely cut off. They are commonly voiced, though voiceless
nasals have been reported in Burmese (see Ladefoged, 2000)
Nasal stop cluster such as mb and nd often referred to as characteristic of African
languages do not occur at all in most Nilotic languages like Dinka-Nuer and Shilluk
(Gregerson, 1972). But they do occur in a few such as the Luo and Alor languages. As a
matter of fact, in Luo some nouns show an alternation of stem-final nasal with nasal-stop
cluster, as in the following:
45
Nominative singular Appreciative singular Plural.
boum boumb buombe ‘wing’
ţuno ţund ţunde ‘breast’
bɔɲ bɔɲj bɔɲjɛ ‘ring’
A nasalized sound may be adjacent to a pure nasal or it may occur in an environment
devoid of a nasal. According to Uguru, (2005), the distinction between pure nasals and
nasalized sounds is their manner of production. While air escape is through the nose
during the production of pure nasals, air escapes through both the mouth and nasal
cavities during the production of nasalized sounds. That languages do not have vowel
systems in which there are more distinctive nasalized than oral vowel qualities has long
been recognized (see Issatschenko, 1937; Ferguson, 1963; Ruhlen, 1978; Hayek, 2005). It
is well known that languages with distinctively nasalized vowels have an equal or lower
number of nasalized vowel qualities than of oral vowel qualities Haspelmath, (2005). For
almost all the languages, each nasalized vowel can readily be considered to be the
nasalized counterpart of one of the oral vowels in the language’s inventory.
However, based on the description in Bender and Harris (1946), Cherokee is
interpreted as an exception, being reported to have five oral vowel qualities [i, e, o, a, u]
plus nasalized vowel /ɐ̃/ which does not have an oral counterpart, so that in this case the
nasalized vowel adds one to the total of basic vowel qualities in the language. However,
different accounts of the language disagree on whether this vowel is nasalized in all its
occurrences, or is only nasalized next to certain consonants, such as /h/. The range of oral
consonants and vowels that can be nasalized depends on the language under
46
consideration. In some languages nasalized vowels feature while in some, they never
occur.
O’Connor (1973) emphasizes that nasalization is another feature with which
vowels can be distinguished. Usually, nasalized vowels are fewer than their purely oral
counterpart. According to him, Burmese has seven oral vowels /i, u, e, o, a,ɛ,ɔ / and five
nasalized vowels /ã, ẽ, õ, ĩ, ũ/.
Portuguese, has eight oral vowels and five nasalized vowels which are; [ã, ẽ, õ, ĩ, ũ].
French has nine oral vowels with three nasalized vowels, thus; [ẽ, õ, ã,]
The acoustic features of vowel, according to Beddor, (1993) have shown that it is
more difficult for a listener to distinguish between different nasalized vowels than to
distinguish between their oral counterparts. This observation probably helps to explain
why the number of nasalized vowels in a language is so often less than the number of oral
vowels. More specifically, this factor can be expected to operate with particular force
when the number of vowels is larger. However, nasal vowels do not seem to be as
frequent in occurrence as nasal consonants. Nasal consonants as contained in the I.P.A.
chart of 1996 are as follows:
/m/ voiced bilabial nasal – seen in most languages like English, Yoruba, Igbo, Igala etc
/ɱ/ labio-dental nasal seen in Igbo.
/n/ alveolar nasal seen in most languages of the world like /n/
/ɲ/ palatal nasal seen in Igbo, French, Yoruba
/ɳ/ retroflex nasal seen in Wangkatya (Australia)
/ŋ/ velar nasal in French, Igbo, Ga, English, Yoruba
/ŋw/ labialized velar nasal in Igbo
47
/ŋm/ labial velar nasal in Idoma, Yoruba
In the majority of languages with nasalized vowels the historical origin of these
vowels can be traced to an earlier sequence of a nasal consonant followed by a vowel or a
vowel followed by a nasal consonant fusing together, (Ferguson (1963) and Maddieson
(2005)). Such processes are quite likely to affect all vowels and thus produce nasalized
counterparts to each oral vowel. However, if the number of vowels is large, maintaining
the distinction between all the nasalized counterparts is difficult. Nasalization can also
occur on consonant. According to Chomsky (1968), nasality can be superimposed on
glides as well as liquid articulation in some languages like Yoruba, Nupe, and other
African languages. The language exhibits nasal cognates of the non-nasal [j], [w] and [r]
contextual variants as [ȷ]̃, [w̃], and [r̃]. According to Ohala (2006), some consonants,
those called sonorants, allow their normal oral articulation to have simultaneous nasality
imposed on them. This set includes the approximants.
Stahlke (1971) notes that it is impossible to produce an implosive sonorant. In
addition, an implosive /m/ or /n/ is apparently a phonetic impossibility because what is
involved in the production of an implosive is the rarefaction of the air pressure inside the
mouth by a downward movement of the whole glottis. This lowering of the air pressure
would be impossible if the nasal passage were coupled to the oral passage. Besides, a
nasally released implosive would be phonetically impossible because since the air
pressure is lower within the mouth, it is not possible for air to be released through the
nose. Instead, air rushes into the mouth after implosive has been released. This phonetic
impossibility explains why no African language, especially in the New-Benue-Congo
sub-family, has been observed with phonetic sequences such as [ɓṽ] and [ɗṽ]. In Ebirie,
48
for instance, [ɓ] and [ɗ] do not occur before nasalized vowels, but only before oral
vowels while [m] and [n] are found before nasalized vowels. Nasalized vowels are
usually derived from the effect of a preceding nasally realized consonant. So, [ɓ] and [ɗ]
are in complementary distribution with [m] and [n] in Ebirie. In other words, if we
represent [mṽ] and [nṽ] as /ɓṽ/ and /ɗṽ/, respectively, then Ebirie, notes Dumestre (1970)
will have no underlying nasal consonants.
These analyses would have been accepted in classical phonemics and generative
phonology but for some reservations by some phonologists. One of such reservations
centres on the pronounceability of underlying forms. This argument has both a weak and
a strong version. In the weak form, it is argued that underlying forms must be
pronounceable at least in some human language. Chomsky and Halle (1968) have, for
example been criticized according to Kiparsky (1968) for setting up an ‘epsilon glide’,
the pronunciation of which is not certain. Since the native speaker will find the
underlying forms [ɓNv] and [ɗNv] hard to internalize for their unpronounceability, it
would be ruled out by a Weak Condition on the basis of pronounceability. A Strong
Pronounciability Condition would state that no underlying form can be set up in a
language that is not pronounceable by speakers of the language. This view looks
awkward. The strong pronounceability condition in conjunction with the Alternation
Condition, Kiparsky (1968), represents the view of phonology presented by Venneman
(1972b). For Vennemann, he would recognize [mṽ] and [nṽ] in Ebirie as /mṽ/ and /nṽ/,
respectively, and not as /ɓṽ/ and /ɗṽ/, since there are no alternations. This creates
phonemes in complementary distribution. /ɓ/ and /ɗ/ would occur only before oral vowels
and /m/ and /n/ only before nasalized vowels in the same way /1/ is recognized before
49
oral vowels in Yoruba and /n/ before nasalized vowels with denasalization rule which
changes /n/ to [l] in certain morphemes.
Umeh (1987) maintains that nasalization impinges directly only on vowels and
approximants. According to him, the nasal effect is postponed at the point of closure for
stops, until the closure opens and nasal release continues on the following vowel. In other
words, all Igbo vowels are capable of being nasalized in the environment of nasals and
nasalized oral sounds, especially approximants like ‘r’ and also ‘h’ whose vocalic
features help to enhance nasal effects.
Nasals have a very special place in the Sotho group of languages. Nasal
homogeneity consists of two points:
(a) when a consonant is preceded by a nasal it will undergo nasal permutation,
(b) when a nasal is immediately followed by another consonant with no vowel
between them, the nasal agrees in place of articulation of the following consonant,
after the consonant has undergone nasal permutation. If the consonant is already a
nasal then the preceding nasal will simply change to the same.
The general Bantu pronoun ‘I’ and ‘you’ are ‘mi’ and ‘we’, respectively. The Bantu
languages have general aversion towards monosyllabic words and use different ways of
making pronouns disyllabic:
Kiswahili uses doubling – ‘mimi’ and ‘wewe’
Shona uses prefixes – ‘imi-’ and ‘iwe-’
Zulu uses suffixes – ‘mina’ and ‘wena’
Sesotho and Xhosa also use the suffix ‘-na’, but the I in ‘mina’ has been elided to ‘mna’.
However, in Sesotho this construction contradicts the second principle of nasal
50
homogeneity, so the m changes to the nasal in the same position as n, given the Sotho
word ‘na’ for ‘I’.
Nasalization is marked by a tilde [ ͂ ] over vowels, for example, [ã] only when the
nasalization represents a phonological feature of a particular language. For instance, in all
languages, a vowel that is adjacent to a nasal consonant must become at least slightly
nasalized. Ohala (2000) however observes that we should not run away with such
impression that all languages have a phonological rule that nasalizes vowels before and
after nasal consonants. Instead, what we should do is to determine whether a language
has, according to him, ‘phonologized’ this phonetic phenomenon which is due to the
timing factors between the movement of the soft palate and the other articulators.
For instance, in Nupe, [nã´] ‘to shine’ is realized as [na´] in Gwari (a language
closely related to Nupe). Although the vowel in ‘to shine’ in Gwari cannot completely
escape the nasalizing effect of the proceeding /n/, this factor cannot be attributed to the
phonetic nature of all human speech. According to Flanagan (1972) a universal phonetic
tendency is said to become phonologized when language-specific reference is made to it,
as in a phonological rule. A phonetic feature becomes phonemicized when contrastive
reference must be made to it, as in underlying phonemic forms. In other words,
nasalization is said to be phonologized if a rule is required in the language that nasalizes
vowels in the contexts of nasal consonants. Nasalized vowels are phonemicized when
they cannot be predicted by rule, but rather must appear in the lexicon.
It is accepted by most phonologists that nasalized vowels are derived from the
earlier state of the oral vowels in proximity with nasal consonants. According to Lunt
(1973) and Herbert (1986), the most common origin of contrastive nasalized vowels
51
seems to be syllable final sequences of oral vowel followed by nasal consonant. This
source which has been relied on for years first appeared in Stahlke’s (1971a) publication
and has in fact been the only one considered for New Benue-Congo languages. Gilbert
and Wyman (1975) reveal that when nasalized vowels are contained in a phonetic system,
they are expected to be learnt much later than their non-nasalized equivalents. In a test
they conducted on children aged between five and seven years on the acquisition of four
nasalized and four non-nasalized vowels, it was seen that the children of the three ages,
five, six, and seven years, learned the nasalized vowels later and slower than the non-
nasalized ones.
In determining whether nasalization should be viewed as segmental or
suprasegmental for any language, several factors must be considered. One of such factors
is whether a direction for nasal spreading can be established or not? Like in the Terena
language below nasalization clearly spreads from left to right. Besides, the following
oral-nasal opposition is found in comparing the third person singular and first person
singular forms:
emo?u ‘his word’ ẽmõ?ũ ‘my word’
ayo ‘his brother’ ãỹõ ‘my brother’
owoku ‘his house’ õwõŋgu ‘my house’
It is quite clear from these forms that nasalization is used as a parameter in differentiating
between 3rd and 1st person singular. Besides, nasalization (or orality) is realized on
several syllables. It is this feature that suggests it as a suprasegmental feature in Terena. It
is possible to identify a nasal item ‘1st pers. sing.’ that is prefixed to nouns and verbs
52
which cause the preservative spreading of nasalization. A latter rule deletes /N/. So,
Leben (1973b) proposes Terena vowel nasalization as follows:
(i) nasalize all vowels and semivowels in the word up to the first stop or fricative,
(ii) nasalize the first stop or fricative in the word as follows: mb replaces p, nd
replaces t, ŋg replaces k, nz replaces h.
Umeh (1996) says nasalization like aspiration is not articulation type par
excellence. It is borne out of the oro-nasal process. While making a specific case for the
Igbo language, Carrel (1970) claims that nasalization in Igbo is a feature of the syllable.
In other words, some morphemic rules distribute the nasal feature throughout the syllable.
Nasalization can be seen as an example of suprasemental feature because it extends its
feature over units which can encompass more than one segment. Normally, note Dickson
and Dickson (1982) the specification [+nasal] is a fractional part of nasal consonants and
probably nasal vowels. However, in some languages, the [+nasal] feature is got from the
segmental tier and, through spreading, put on the suprasegmental tier so that it can affect
some syllables or morphemes or the entire word.
So, in a suprasegmental analysis, according to Leben (1973a) a nasal exponent can
be fashioned out which, by a ‘mapping rule’ is assigned to each segment within the
suprasegmental unit like a syllable. A segmental analysis on the other hand would
endeavour to give an underlying [+nasal] feature specification to one segment within
each suprasemental unit and will provide a rule by which neighbouring segments
assimilate to that feature specification. In the underlying form, /bã/, nasality is assigned to
underlying vowels. A rule is therefore recognized to nasalize oral consonants in the
context of a following nasalized vowel:
53
V
C [ +nasal] / [+ nasal]
An equally plausible segmental analysis would recognize the underlying form /ma/,
where nasality is assigned to the consonant. In this case, a rule is needed to nasalize an
oral vowel following a nasal consonant, as seen below:
C
V [+ nasal] / [+nasal]
Hyman (1982) notes that those languages that use nasality situate it either
syntagmatically or paradigmatically. So, nasality can be analyzed using non-linear
approach only if the latter option (paradigmatically) is adopted. Here, nasality is
manifested over many syllables phonetically. Many languages (especially African
languages) have contributed to our understanding of nasals and nasalization process in
several ways. A large number of African languages have nasal consonants. Many of these
nasal consonants are in near or total complementary distribution unlike their oral
counterparts. Thus quite early in the history of generative phonology, Schachter and
Fromkin (1968) had reported derivations such as the following from dialects of Akan,
that there are no underlying nasal consonants in the language. That is, no voiced oral
consonant appears before nasalized vowels in surface forms:
/bã/ [mã] ‘give’
/dã/ [nã] ‘and’
/jã/ [jã] ‘receive’
/wãdĩ/ [w̃ãnĩ] ‘scrape’
Rather than representing nasal where it is contrastive, they also could quite easily have
abstracted the feature away as prosod
Recommended