INTRODUCTION

This paper focuses on teaching of - and research on English pronunciation in the West African region, with special emphasis on problems and prospects. There is one major problem that affects research on - and the teaching of - English pronunciation in West Africa. That problem is the choice of the English model to teach and to use for research. Over the years, RP has been the chosen model, with research in West Africa being limited to identifying, comparing and contrasting the salient phonetic and phonological differences between RP and each of the local English varieties in the region. This choice was influenced by renowned researchers including Prator and Quirk. Prator (1968) opted for a single, native-speaker model for the English language education of non-native speakers across the world. Similarly, Quirk (1988) considered variations and innovations as performance errors that needed to be corrected and as deviations from the native-speaker norm that needed to be avoided. As a result of the use of RP as target model to be attained by all learners of English, the findings of all research works highlighted the acquisitional deficiencies of each of the West African local accents. This paper proposes a completely different approach which would have an impact on both teaching and research. The paper first describes what strategies can be used to choose a model of English to teach (1), and then it highlights other ways of conducting research on English pronunciations in West Africa (2).

I. CHOOSING A MODEL OF ENGLISH TO TEACH IN WEST AFRICA

This section proposes a strategy for choosing a model for teaching oral English in the West African region. This region comprises some 17 countries, when Cameroon is included. Of these 17 countries, six are English-speaking. These are: Cameroon, Gambia, Ghana, Liberia, Nigeria, and Sierra Leone. Over the years, research on the pronunciation of English in this region consisted in comparing the features of one model of English, i.e., RP and those of each of the local English varieties. As a result, the findings of these studies highlighted the acquisitional failures of each of the West African nation-state accents, which were labelled in turn as: Cameroon English (Simo Bobda 1994; Kouega 2013), Gambian English (Wolf 2003), Ghanaian English (Sey 1973), Liberian English (Hancock 1974; Singler 1997), Nigerian English (Jibril 1979; Banjo 1993; Bamgbose, 1995; Igboanusi 2002), and Sierra Leone English (Pemagbi, 1989).

Reliance on RP was justified by the fact that each local variety had to be intelligible to the native speaker; the belief at the time was that anyone who learnt English did so in order to be able to communicate with the native speaker of English, who was assumed to speak RP. Little was it known that the RP accent in question was not the dominant accent in Britain. Actually, it was and it still is the most documented accent of English. Because of its availability in textbooks, Oral English teachers found it difficult to discard it. Unfortunately, as Soneye (2008:194) pointed out, these teachers continue "to labour, most times fruitlessly, to teach students to pronounce sounds the British way, they themselves not having the necessary input, thereby sounding bookish and the students most of the time passive." Today many researchers have realised that there is a need for an international model that can be used for pedagogical purposes. MacArthur

(2004: 416) for example, reports that: "at the moment, there are demands for a standardised international variety so that it can be taught". Such a standard will be hard to construct as there is no authority that is in a position to determine how other users of the language would speak.

To choose a model of English to teach today, it would be a good idea for English-using countries to first discard the belief that all learners of English learn this language in order to be able to communicate with native speakers of English. This old belief can be referred to here as the one-model approach, which was recommended by many influential figures including Prator and Quirk. When this mono-model approach is applied to RP and any other native accent of English, it reveals a number of significant deviations and divergences (Wells, 1982), just as divergences are observed between RP and any non-native variety of English.

The present study proposes a multi-model approach, which stipulates that each country should adopt the position that its citizens learn English in order to communicate with the citizens of other English-speaking countries, especially those countries with which they do business. If an English-using country does business exclusively with Britain, then it would focus on RP. If on the contrary, an English-using country does business exclusively with another English-using country, it goes without saying, the two countries would need to know the characteristic features of their Englishes. Simply put, it is important for each country to integrate into its classrooms and research laboratories, those salient elements of the English of the country or countries with which it does business. To be more specific, the choice of the target model of English to use for the teaching of oral English and research on English pronunciation in a country should be determined by the transactions done in English with this country's external trade partner states. This new way of choosing a target model for teaching English is referred to above as the multi-model approach.

Let us illustrate this position with the cases of Nigeria and Ghana, two leading West African countries. In the year 2017, to take just that year, Nigeria's main export partners were: India (18%), United States of America (14%), Spain (9.7%), France (6%), Netherlands (4.9%), and Germany (3.6%); that same year, her main import partner countries were: China (28%), Belgium (8.9%), Netherlands (8.3%), South Korea (6.4%), United States (6%), and India (4.6%) (Source: https://oec.world/en/profile/country/nga/). Trade with all these countries is negotiated in English and, as one can guess, none of the negotiators speaks RP. This means that the target accents that need to enter Nigerian classrooms and research laboratories are: RP, General American, Indian English, and Chinese English. RP is chosen because it is the most documented accent of

English and it cannot therefore be left out. General American is chosen by virtue of the USA being Nigeria's second export partner country (14%). Indian English must be on the list because India is Nigeria's first export partner country (18%); finally, Chinese English, though a performance variety, is chosen because China is Nigeria's top import partner country. In short, research on English pronunciation in Nigeria should focus at the same time on: RP, American English (GenAm), Indian English (IndE), and Chinese English (ChinE). The same reasoning can be made for Ghana, whose main export partners are India ( $5.09B),China($ 1.9B), Switzerland ( $1.84B),SouthAfrica($ 918M) and the Netherlands ( $911M),withitsmainimportpartnersbeingChina($ 3.08B), the United States ( $1.1B),India($ 660M), Belgium-Luxembourg ( $637M)andtheUnitedKingdom($ 587M). (Source: https://oec.world/en/profile/country/gha/). This means that Ghana's target accents for both teaching and research should be RP, GenAm, IndE, and ChinE. The point this paper is putting across is that pronunciation teaching and phonological analysis in a given country should focus on the accents of English used in its major trade partner countries. As far as Nigeria is concerned, the focus should be on RP, GenAm, NigE, IndE and ChinE, because these five accents are likely to be heard more frequently by most speakers of English within the Nigerian territory. At this point, there is a need to define each of these five accents.

a) RP (Received Pronunciation)

As many researchers including Jones (1917), Wells (1982), Gimson (1989) and MacArthur (1992) pointed out, it is an accent that developed in the Southeast of Britain and that has usually been referred to by various names including: BBC English, Standard British pronunciation, Southern British pronunciation, or Public School English. Although used by a very tiny proportion of British citizens, it is regarded as the prestige standard in the country. What makes it prestigious is the fact that it is used by people of wealth and power like the royal family, high-ranking army officers, graduates from Oxford and Cambridge universities, politicians, successful actors and, more importantly, BBC newsreaders. It is the accent that was taught in all English as a Second and Foreign language classes in the rest of the world. As such, all spoken pedagogic materials used in the world were produced in RP. Strangely enough, when visitors get to Britain, they hardly meet anyone to speak RP with. What they hear is usually a multiplicity of regional accents of English. In short, RP is an accent that is associated with education, as it is used by journalists and university graduates.

b) GenAm (General American)

This accent, which may be abbreviated to GA or GenAm, is associated with the speech of educated people. It started up in major business centres in New

York and St Louis and, with time, it imposed itself as the form of speech to be heard over the radio and television, at the expense of many regional accents used in the south and the east of the country. Later on, it became the standard pronunciation used in scientific and business interactions throughout the country. Like RP, this accent is associated with education, as it is used by journalists and scientists.

c) NigE (Nigerian English)

Nigeria is a country where three dominant communities co-exist; these are Hausa, Igbo and Yoruba. Typical Nigerians learn Standard British English after having acquired their respective L1s in the home context. As Jowitt (1991) pointed out, they do not acquire English in the first few years of life, as is the case with native English children; they learn it as an L2 over the years, mainly through formal instruction in school. To describe their output, this researcher proposed a number of terms: Popular Nigerian English, Standard Nigerian English, and Nigerian English (p. 46). Popular Nigerian English (PNE) is the mesolectal variety of English used mainly by primary school certificate holders and high school leavers. Standard Nigerian English (SNE) is what may be referred to as educated English, that is, the acrolectal variety. Nigerian English (NE) is therefore the sum total of both Popular Nigerian English and Standard Nigerian English. In short, NE=PNE+SNE meaning: Nigerian English is Popular Nigerian English plus Standard Nigerian English. For the purpose of this study, Nigerian English is considered as the English produced in formal contexts by educated speakers like journalists, lawyers, medical doctors, teachers, and other learned people.

d) IndE (Indian English)

India is a big country of over three million square kilometres where several languages co-habit. Hindi is the dominant indigenous language with the greatest number of speakers in the country, but there are a number of regional languages which are widespread in certain states to the point that some inhabitants have limited knowledge of Hindi, the official language used for domestic affairs. It shares this official status with English, which is used mainly for interstate and federal transactions as well as international dealings. India is an English-using Outer Circle country. This term is drawn from Kachru (1985, 1992) who used three concentric circles to distinguish three groups of English-using countries: the Inner Circle which brings together the old English-speaking countries where English is a native language like Britain, the Outer Circle where English was taken to during the years of colonization and has to co-exist with other languages of various statuses like Ghana, and the Expanding Circle which groups an increasing number of countries which have adopted English as a foreign or trade language like China. In the literature on the distribution of English- using countries in the world, three types of territories are distinguished. These are: English as a Native Language (ENL) territories like Britain, English as a Second Language (ESL) countries like Ghana, where English is used for various purposes and usually has official status, and English as a Foreign Language (EFL) countries like Japan, where English is learned and used mainly for business or education (Kachru, ibid; MacArthur, 1992: 353). In this categorisation, India falls into the group of English as a Second Language (ESL) countries, where English has official status and where the preferred spoken model is usually the RP accent. English in India is used as a natural medium of communication by many people. The focus in this study is on the English of educated people (Bansal 1990; Kachru 1983), who may be journalists, lawyers, and businessmen.

e) ChinE (Chinese English)

China is a huge country where several languages or distant dialects of the same language coexist. Northern Chinese, which is the native language of over $70\mathrm{\%}$ of the population, has been accepted as the written standard language and the national and official language for all Chinese. In other words, all Chinese learn Northern Chinese also known as Mandarin, in school. This means that Mandarin is an L2 for some people like Mongolians and Tibetans, and an L1 for the vast majority of Chinese. English in China is a non-native variety which falls under Kachru's Expanding Circle countries; it is a foreign language used mainly for international trade and commerce. In primary school, Chinese pupils learn Mandarin and English, with the model of English chosen being British English. In secondary school, they switch to American English and in high school they switch back to British English. This means that, right from the start, Chinese pupils and their teachers can hardly make any distinction between American and British English. In other words, Chinese English (Feifei Han 2013; Zhang and Yin 2009), whose characteristic features have been referred to as Chinglish (MacArthur 1992: 214), is a learner English variety which is made up of a mix of features of Mandarin, British English, and American English. For the purpose of this study, Chinese English is regarded as the English used in formal contexts by educated speakers like journalists, lawyers, and businessmen dealing with international trade and commerce.

Varieties of English are grouped on pedagogical terms above into Inner, Outer and Expanding Circle varieties (Kachru 2001). They can also be grouped on geoeconomic terms into "centre" and "periphery", drawing from world systems economic theory (Wallerstein, 1991). Centre countries are those with a high level of economic development like the USA and Britain; periphery countries, on the other hand, are countries with a relatively low level of economic development, like Nigeria and India. This geoeconomic antinomy has been transposed into the geolinguistic domain, where "centre" countries refer to Kachru's Inner Circle English-speaking countries while "periphery" countries refer to postcolonial Englishes (Canagarajah 2002, Pennycook 1998, Phillipson 1992). This geolinguistic centre-periphery dichotomy seems to overlook some countries like China which have embraced the learning of English, i.e., the Expanding Circle countries; this gap was filled by Souza Santos (1994), who coined the term "semi-periferico" or "semiperiphery" to refer to such countries. In short, varieties of English can additionally be grouped on geolinguistic terms into Centre (GenAm or RP), Periphery (IndE and NigE) and Semi-periphery (ChinE) varieties. This geolinguistic categorisation will be useful in the multi-model descriptions presented in this study.

In brief, the multi-model approach outlined above dwells on international trade, as English is essentially used in most non-native countries for international transactions.

II. POSSIBLE WAYS OF DOING RESEARCH ON ENGLISH PRONUNCIATIONS IN WEST AFRICA

Thus far, the issue of the choice of a model of English for research and pronunciation teaching has been outlined. This section discusses how segmental features can be described using the multi-model approach. Vowels are considered first (2.1) and consonants are taken up next (2.2). Finally, consonant cluster and syllable structure simplification processes (2.3) are dealt with.

a) Vowels

Under this heading, vowel length (2.1.1), monophthongs (2.1.2), and diphthongs and triphthongs (2.1.3) are considered.

i. Vowel length

Vowel length is characteristically phonemic in RP, as can be illustrated by the FLEECE and KIT vowels, where /i:/ and /ɪ/ are perceptibly distinct. The /i:/ and /ɪ/ distinction is attested in Indian and Chinese Englishes, but not in American and Nigerian Englishes. The START vowel is replaced by /a/ in most varieties except American English where pre-consonantal /r/ is added, hence /αr/. This pre-consonantal /r/ is also observed in CLERK, NORTH, FORCE, and NURSE. In the word CLERK, /ɑ:/ tends to be rendered as /ε/ in the non-native varieties; this is an obvious case of spelling-induced pronunciation, with the -ER- letter sequence being pronounced as it is spelt. The THOUGHT vowel is usually replaced by a shorter /ɔ/ sound, which is however not as short as RP /ɒ/. In brief, vowel length is not realised in the targeted varieties of English; long vowels are generally replaced by shorter vowels which are not as short as their RP equivalents. From these observations, it can be concluded that there is convergence in the rendering of long vowels in the speech of the non-native speakers considered in this study. In short, the long and short vowel pairs /i:, i/, /ɑ:, x/, /ɔ/, d/, and /u:, u/ tend to be conflated to /i, a, o, u/ respectively. These findings are summarised in Table 1 below.

ii. Monophthongs

The FLEECE and KIT vowels are distinct in the two native varieties under consideration; conversely they are rendered in the non-native varieties under consideration by an /i/ vowel which is not as long as the FLEECE vowel nor as short as the KIT vowel.

Interestingly, this /i/ vowel is heard in final position in GenAm as the word HAPPY in Table 2 attests.

The DRESS vowel is rendered as $/\epsilon /$ in all varieties. However, it may occasionally be replaced by $/\mathrm{e}/$ in the Indian and Chinese varieties, as Table 3 shows. This means that, in an interaction with these people, one should not be worried when /drεs/ is occasionally rendered as /dres/, as these two forms may occur in the speech of the same speaker.

The TRAP vowel tends to be rendered as /æ/ in the native varieties. However, it is systematically replaced by /a/ in Nigerian English and it alternates with /ε/ in Chinese English. The START vowel is replaced by /a/ in the non-native models and it is accompanied with the pre-consonantal /r/ in GenAm. The letter A in the word TOMATO is rendered by the monophthong /a/ in the non-native varieties and by the diphthong /eI/ in GenAm. The implication here is that non-native speakers can more readily understand this word when it is pronounced by an RP speaker than by an American English speaker. The BATH vowel is consistently rendered as /a/ in the non-native models, which cannot affect a conversation with Americans who tend to pronounce it as /æ/ nor RP speakers who say /α:/ instead. Finally the CLERK vowel is replaced by /ε/ in the non-native varieties; this rendering seems to be influenced by the spelling of the word, as the sequence of letters –ER– in medial position may be rendered as /ε/ in words like "clergy" and "cleric". This rendering may therefore be attributed to spelling pronunciation. In Nigerian English, a few purists use /a/ instead of /ε/; this rendering is therefore consistent with the rule of conflation of long and short RP vowels mentioned earlier. Finally, in GenAm, the pre-consonantal /r/ element follows the vowel /3/, yielding /kl3rk/. Both the RP and GenAm renderings of this word can be a real challenge for the non-native speakers, whereas the reverse is not true, as these native speakers can be inspired by the presence of the –ER– sequence in the spelling of the word. This is summarised in Table 4.

The LOT vowel on the one hand, and that of THOUGHT, NORTH, and FORCE are conflated to /ə/ in the non-native varieties. This conflation may affect the perception of isolated words but not words used in context. However, the rendering of the vowel /ɒ/ in the word LOT as /ɑ/ in GenAm is likely to confuse both unaccustomed RP speakers and non-native users. These findings are presented in Table 5.

The FOOT vowel, which is pronounced the same in the two native varieties considered here, is systematically rendered as /u/ in the non-native varieties.

This same /u/ vowel is used in the word GOOSE in four of the five varieties under study, as Table 6 shows.

The LETTER vowel tends to be rendered as /ər/ in GenAm, IndE and ChinE. This rendering with the postvocalic /r/ may have resulted from the influence of GenAm on IndE and ChinE. This influence has not reached Nigeria and other West African countries, where RP central vowels are systematically replaced by peripheral vowels, with /ə/ rendered as /a/ as in "teacher" or as /ə/ as in "doctor". The COMMA vowel is pronounced the same in the two native varieties i.e. /ə/, and the same in the three non-native varieties i.e. /a/; the latter rendering is a good example of spelling pronunciation. Finally the STRUT vowel is pronounced the same in the two native varieties, with the /ʌ/ vowel. This vowel is rendered as /a/ in both IndE and ChinE. In West African countries including Nigeria, it is replaced by /ɔ/. In other words, the NigE pronunciation may be a problem for speakers of the other four varieties under scrutiny.

In brief, as the analysis above shows, there is a great deal of convergence in the realisations of English monophthongs in the speech of non-native speakers, which means that their citizens can easily communicate effectively, with each of them speaking its own variety of English. Needless to say, these realisations tend to be different from those of native speakers.

monophthongised in the non-native varieties. The PRICE vowel tends to be pronounced in a recognisable way in all the varieties; however the AGILE vowel is monophthongised into /3/ in GenAm and /a/ in ChinE. Like the PRICE vowel, the CHOICE vowel is readily recognisable in all the varieties under study, as Table 8 shows.

iii. Diphthongs and triphthongs

The FACE vowel is rendered as a diphthong in the native varieties; it is systematically

The GOAT vowel is realised as a monophthong in all the varieties except RP, as Table 9 reveals. The replacement of /əʊ/ by /o/ in four different varieties seems to be a mere coincidence, as there is virtually no historical or linguistic factor which justifies this convergence. The most likely factor may be spelling, as the letter sequence OA is generally rendered as /o/, as the words "coat", "approach", "toast" and "throat" show. The MOUTH vowel surfaces as a diphthong in all the varieties under consideration.

The renderings of NEAR, CURE and SQUARE are markedly different in the five varieties under consideration. This can be a serious cause of misunderstanding. The diphthong /ɪə/ is monophthongised in GenAm (/ɪr/), with the post-vocalic /r/ element replacing the schwa. In NigE, this same schwa is replaced by the DRESS vowel. The CURE vowel has the most diverse realisation in the five varieties. An extra effort has to be made by listeners to recognise this word. The SQUARE vowel is reduced to the monophthong /ɒ/ in the three non-native varieties. This reduction also occurs in GenAm, but its characteristic post-vocalic /r/ element is added.

The FIRE vowel is realised as a diphthong followed by the post-vocalic /r/ (diphthong+/r/) in GenAm. In the non-native varieties, its central element /I/ tends to surface as /j/, therefore breaking the triphthong into two different syllables, i.e., /a.ja/ in NigE and /a.jə/ in both IndE and ChinE. The GREYER vowel is systematically converted into a diphthong followed by the schwa and the postvocalic /r/, i.e., diphthong $+/\mathrm{\exists }/+/r/$ in GenAm. Elsewhere, the central component of the triphthong is realised forcefully to the point that it surfaces as the glide /j/. In short, when the first element of a triphthong is a front vowel, the central element is replaced by /j/ in the non-native varieties under consideration. When on the contrary this first element is a back vowel, the same process applies, but this time, the central element is replaced by the glide /w/ in the non-native varieties, as the words "employer", "power" and "lower" in Table 11 show. In GenAm, these triphthongs are realised as a diphthong followed by the sequence /ər/.

In brief, the analysis of the realisations of vowel sounds shows that these vowels undergo various processes in the varieties under consideration. Some monophthongs are replaced by others, some diphthongs are monophthongised, and triphthongs are generally broken into two separate syllables, with the second syllable beginning with a glide like /j/ or /w/.

b) Consonants

This section dwells on the renderings of individual consonant segments. It takes up voiced consonants first (2.2.1). Then it looks into interdental consonants (2.2.2) and finally other consonants (2.2.3). These are taken up in turn.

i. Voiced consonants

A consonant is said to be voiced when, during its articulation, the vocal cords vibrate (Meyer 2009). For example, in the pair /p, b/, the first element is voiceless whereas the second is voiced. In English, voiced consonants are devoiced in certain contexts. For example, when voiced plosive, fricative, and affricate consonants precede a voiceless sound, they are systematically devoiced. Take for example the two phrases "his case" and "his box". When the word "his" occurs in isolation, it ends with the voiced consonant /z/. When it is followed by a voiced sound as in the phrase "his box", this voiced /z/ is used because the following sound /b/ is voiced: /hlz b1ks/. When on the contrary,

the following sound is voiceless as in the word "case", the sound /z/ is devoiced to /s/, hence /hIs keIs/. In other varieties of English, consonant devoicing tends to be used in different contexts. In the non-native varieties under consideration, devoicing usually occurs in word-final voiced consonants. Illustrations are provided in Table 12.

As these examples show, consonant devoicing is systematic in word-final position in NigE, IndE and ChinE.

ii. Interdental consonants

These sounds, which are pronounced the same in the native varieties, are hardly realised in non-native Englishes. In the two post-colonial Englishes under consideration, i.e. NigE and IndE, /θ/ is replaced by /t/ and /ð/ by /d/. In ChinE, these two sounds are replaced by /s/ and /z/ respectively, as Table 13 shows. At final position, /ð/ is replaced by /t/ in the post-colonial varieties and by /s/ in ChinE, as the word "with" below shows. This can be regarded as another instance of consonant devoicing that was outlined in 2.2.1 above.

iii. Other consonants

The consonant /t/ at intervocalic position is rendered the same in all varieties except GenAm, where it is pronounced /d/ as in the word "city". In RP, GenAm and NigE, the labio-dental consonant /v/ is pronounced the same in most contexts. It is systematically replaced by the bilabial /b/ in IndE and the bilabial /w/ in ChinE. The voiced palato-alveolar consonant /3/ is realised as expected in RP, GenAm and ChinE. In the post-colonial varieties, it may be devoiced as in NigE, or realised as a voiced affricate sound in IndE. The voiceless affricate sound /tʃ/ tends to be realised in all varieties as expected except ChinE where it is replaced by /tr/. The same can be said of its voiced counterpart /dʒ/, which is rendered as /ts/ in all contexts. Devoicing of consonants takes place in word-final position; /dʒ/ therefore surfaces as /tʃ/ in NigE and as /ts/ in both IndE and ChinE. The velar nasal /ŋ/ is rendered as /ŋg/ in medial position in the non-native varieties, as the word "singing" shows; this is an instance of spelling pronunciation. This velar nasal is systematically replaced by the alveolar nasal /n/ in final position in these same varieties (singing).

The alveolar lateral sound /l/ is rendered as expected in all contexts and in all the varieties; however, it is replaced at final position in ChinE by the alveolar nasal /n/, as in the word "pale", which surfaces as "pane". The alveolar roll /r/ occurs in all varieties but in different contexts. In ChinE, it is replaced by /w/ or /l/ at both initial and medial positions and is actually articulated as expected at final position. This may be an influence of GenAm, which seems to have affected IndE, as the words "rice", "derive", and "pear" below show. Needless to say, /r/ is silent at final position in RP and NigE. Finally, the bilabial semi-vowel /w/ is rendered in all varieties as expected except in ChinE, where it is systematically replaced by the labio-dental fricative /v/, as the words "win" and "underwear" show.

c) Consonant cluster and syllable structure simplification

Various consonant cluster and syllable structure simplification processes are observed in the varieties under consideration. The consonant cluster simplification processes can be grouped under two broad headings: those which declusterise by deleting some elements from clusters (consonant deletion), and those which simplify clusters by inserting a vowel in them (vowel epenthesis). These are considered in turn.

In GenAm, Yod dropping is systematic. In the context of /j/ followed by /U/ after alveolar sounds like /t/, /d/, /s/, /n/, /l/, the /j/ tends to be dropped. This can be illustrated by the words "tube", and "news", where the RP /C/ + /j/ cluster is rendered as /C/. This process seems to have entered IndE and ChinE, but not NigE, as Table 15 shows. Similarly /nt/ is rendered as /n/ in GenAm; this can be illustrated by the word "winter", which rhymes with "winner". This dropping of /t/ does not take place in NigE, IndE and ChinE. In the same vein, the cluster /kt/ as in the word "act" tends to be reduced to /t/ in NigE (Soneye and Oladunjoye, 2015), and to /k/ in IndE and ChinE. This simplification does not occur in GenAm. Both types of reduction can be a good source of misunderstanding.

Vowel epenthesis is another common process which characterises the varieties under consideration. The cluster /ʃn/ as in "version" is rendered as /ʒn/ in GenAm; this cluster is broken by an epenthetic vowel, which surfaces as /ɔ/ in NigE and as /ɔ/ in IndE and ChinE. Similarly the CC cluster /sp/ as in 'speech' may be rendered as /səp/, where the anaptyctic vowel /ə/ is inserted in-between the CC cluster to yield a CVC structure. On the other hand, the prosthetic vowel /i/ may be inserted at the beginning of /sp, st, sk/ clusters to turn them from a CC structure into a VCC structure, like "school" where /sk/ is rendered as /isk/.

Syllable structures tend to be simplified by one generalised process which is the insertion of a vowel or a consonant where they are not expected. The consonant /j/ may be used in fast RP speech to ease the transition of one word to the next as in "my arm". The same can be said of the consonant /w/, which may be used to link two words in fast speech, as in "go away". This process can be exploited in normal speech rate to change the structure of a syllable from a VCV pattern to a CVCV pattern. This can be illustrated by the word "every" which is rendered in Indian English as /jevəri/, with an intrusive initial /j/ (Gargesh 2008). This same process is observed in the word "only" which is rendered with an intrusive /w/ at initial position: /wnli/. Syllable structure may also be modified by the use of vowels. For example a CVC structure may be changed into a CVCV structure in ChinE when the final C is a stop consonant (Feifei Han 2013). Final alveolar and velar stops (/t/, /d/, /k/, /g/) tend to take the vowel /ə/. Hence, words like "hot" and "good" (CVC) tend to be rendered as /hətə/and /gudə/ (CVCV). Bilabial stops rather go with the vowel /u/ as in "map" and "tab", which are rendered /mapu/ and tabu/. Another process which changes syllable structure is the substitution of the vowel /ə/ for the consonant /l/ when it occurs in final position in ChinE; this process may change a CVC structure into a CV structure as can be illustrated by the word "pool" /puə/, which usually rhymes with "poor". These findings are presented in Table 18.

III. CONCLUSION

This paper has proposed a new way of teaching and doing research on - English pronunciation in Nigeria in particular and in Africa in general. Previous researchers have been highlighting the contrast between RP and other varieties, with RP being regarded as the model to approximate. This can be called the one-model approach. In this study, the multi-model approach has been outlined: the citizens of each English-using country are expected to familiarise themselves with the salient features of the English of the citizens of the country or countries with which they do business. If for example India is a country's major trade partner, then this country has to carry out research on the features of the Indian variety of English. In the one-model approach, teachers and researchers used to identify features and contrast them. Using the error analysis tool, they showed how distant the identified features were from the features of RP, the target model.

In the multi-model approach, teachers and researchers will identify features and compare them in order to facilitate convergence, with the ultimate aim being to increase intelligibility during business transactions. This model is an empowerment tool that will lead to the simplification of the phonology of the English language, and the new accent thus created will have a greater number of speakers than all the native accents' speakers put together. Ultimately, the multi-model approach will solve the issues of internal variations within a given country like Nigeria, as focus will henceforth shift from tribal, ethnic or regional features towards global features.

When research in the multi-model approach reaches full gear, this newly constructed accent of English will exhibit a number of salient features, some of which are:

• dominance of spelling pronunciation, which will cause weak vowels to be articulated fully;
• loss of phonemic distinctiveness of vowel length, which will cause "fit" and "feet" to be homophonous;
• replacement of interdental sounds with alveolar sounds /t, d. s. z/
• systematic devoicing of consonants in word-final position, which will cause "back" and "bag" to be homophonous, to name only these.

In brief, knowledge of the salient features of the Englishes of one's country business partner countries is likely to accelerate convergence, enhance intelligibility, and therefore limit the possibility of misunderstanding. Greater familiarity with salient features will eventually lead to closer convergence.