Pronunciation model: West African English

View the key for West African English here.

English is considered a national or official language in seven parts of West Africa: Nigeria, Ghana, Sierra Leone, Liberia, Cameroon’s anglophone provinces, the Gambia, and the island of Saint Helena (listed in descending population order). Each of these can and has been studied in isolation, but there are several common tendencies which give value to a West African English (WAfE) set of pronunciations.

The OED WAfE pronunciation model gives priority to Nigerian and Ghanaian Englishes as being the two West African Englishes with the largest number of speakers. Within Nigerian English, there are greater similarities between Yoruba English and Igbo English than between either of these with Hausa English, so it is the commonalities of the former two that are adopted in the main. It should be noted that there are different attitudes towards ‘owning’ a distinct variety of English which has seemingly affected the development of some West African Englishes (see e.g. Asabea Anderson, 2009).

Readers and OED users familiar with the World English pronunciation models will notice a subtly different approach to the WAfE model compared with others. A common set of sound qualities and symbols is established, but in several cases these are combined in different ways depending on the history and usage of each entry. For example, a word exclusive to Ghanaian English will be represented using the WAfE model but the phonotactics (permitted sound combinations) will be tailored more towards those of Ghanaian English than to those of a more generic WAfE pronunciation.

The OED WAfE pronunciation model is based primarily on the work of Ulrike Gut (2015), with extensive reference to the descriptions of Jowitt (2019), Bobda (2008), Huber (2008), Gut (2008), Singler (2008), and Wilson (2008).

 DRESS ɛ GOOSE u SQUARE ea / ɛ / ia LETTER a
 BATH a START a FACE e ADDED e / i / ɛ
 LOT ɔ NURSE ɔ-ɛ-a / ɛ / a PRIDE

FLEECE and KIT are merged, as are FOOT and GOOSE, LOT/CLOTH/THOUGHT and PALM/TRAP/START. Vowel length is only distinctive in Gambian English (and Nigerian Hausa English) and consequently is not represented in the WAfE model. WAfE is also non-rhotic, more similar to British than American English in respect of having no /r/ representations in START, NURSE, NORTH, FORCE, NEAR, SQUARE, or LETTER lexical sets.

STRUT: The absence of a centralized /ʌ/ vowel (in all but Liberian and Saint Helenian Englishes, and some acrolectal varieties) renders this vowel /ɔ/ in WAfE except where a word is specifically of Ghanaian or Nigerian Hausa Englishes, in which case /a/ may be given (though even in Hausa English the quality is moving towards [ɔ], while [ɔ] and [a] are in free variation in Ghanaian English – see Huber, 2008, for a discussion of the factors affecting this). It may also be given as a variant pronunciation if usage spans Ghana and/or the Hausa-speaking region of Nigeria. Consequently, in Ghanaian English, there is an apparent merger between vowels of the BATH/TRAP/STRUT sets. The specifically Sierra Leonean pattern of /aʊ/ in words such as country is not represented in the OED WAfE model.

NURSE: This vowel is affected by the spelling in most West African Englishes, e.g. work /ɔ/, girl /ɛ/, perch /a/ (though the latter two would both be /a/ in Sierra Leonean English). As with the STRUT pattern, if a pronunciation pertains to Ghana, the uniquely Ghanaian blanket /ɛ/ pronunciation for NURSE vowel words will be given, while words of Nigerian Hausa English or Saint Helena English will be given with /a/ (note that length is not being represented, unlike the [aː] in the descriptions of Gut, 2008, and Wilson, 2008; neither is the Liberian more central [ʌ] quality).

FACE and GOAT are given in monophthongal form, while the Nigerian English pronunciation of NEAR is prioritised (though it is acknowledged that in Ghana and Cameroon the second element is likely to be less open). SQUARE is given as /ea/ except for when the word is of Ghanaian, Cameroonian, or Liberian English (in which cases a monophthongal /ɛ/ is found), though in Nigerian English words may be /ia/ as in NEAR. CURE, by default, is given with both most common forms found in West African Englishes (/ua/ and /ɔ/) as they do not neatly split along variety lines. Non-finally this vowel may be monophthongal /u/ (in e.g. Euro). However, GOOSE words spelled with u, ue, eu or ew – which are pronounced BrE /uː/ – are transcribed /(i)u/ in WAfE words spanning Ghana, reflecting the /iu/ of Ghanaian English pronunciations such as new /niu/ (see Huber, 2008). If the word is solely Ghanaian, the transcription is /iu/ with no brackets. The ADDED quality by default is represented as /e/, but can be higher (/i/) or lower (/ɛ/) depending on the phonological environment and/or the variety.

The combined WAfE pronunciation model is intentionally broad and does not aim to reflect all subtleties in vowel quality. In particular, [ʊ] and [ɪ] are both qualities heard in WAfE pronunciation (reflected in the audio recordings) but are not regarded as meaningfully distinct from /u/ and /i/, respectively.

Epenthetic glides in diphthongs (documented in the speech of a range of speakers for NEAR and CURE) are not represented here, but Nigerian English glide formation rules are applied for triphthongs, resulting in fire as [ˈfaja], power as [ˈpawa]. A similar pattern may be noted with diphthongs before dark-/l/, e.g. oil as [ˈɔjil].

Unstressed vowels are not represented by schwa, even in weak forms of function words. Full vowel qualities are given, largely dependent on the spelling. Similarly, there is no scope for word-final syllabic consonants, the consonant preceded by /u/ (able), /e/ (written), /a/ (legal), or /i/ (pencil).

Consonantally, WAfE has no phonemic /ŋ/ (commonly /n/, with /ŋ/ occurring only in otherwise /nɡ/ contexts to create /ŋɡ/ or occasionally in /nk/ contexts to create /ŋk/). /n/ before bilabial sounds (/p/, /b/, /m/) is likely to assimilate to /m/ (in e.g. unmotorable). Although stigmatized (particularly amongst more acrolectal Nigerian English speakers), dental fricatives are commonly ‘stopped’ to /t/, /d/, and for descriptive accuracy /θ/ and /ð/ are not reflected in this model. /ʃ/, /dʒ/ or /z/ are sometimes heard where /ʒ/ may be expected (e.g. occasion). Word-final voiced obstruent (plosive, fricative, and affricate) sounds may be devoiced (e.g. peas as [pis]), but this is reflected only where warranted on a case-by-case basis.

Consonant clusters are often (but not always) simplified, particularly in word-final position and often involving either the elision of a word-final /t/ or /d/ (e.g. hand, post, also hundreds /hɔndrɛs/) or an epenthetic /u/ or /i/ (e.g. bottle /ˈbɔtul/, silk /ˈsilik/). According to Jowitt, elision is less stigmatized than epenthesis amongst acrolectal speakers of Nigerian English. Where elision occurs, it is most likely to be the final consonant of the coda cluster, and shown in OED as a bracketed sound if the cluster reduction is variable. Yet there are also a number of spelling pronunciations in Nigerian and Ghanaian Englishes resulting in pronunciations such as bomb /bɔmb/, song /sɔŋɡ/. Post-initial /j/ is also retained in clusters at the starts of words such as blue /blju/ and grew /ɡrju/. Also note that the sequence /ɡb/ will often be treated as a single sound rather than a cluster. It is similar to e.g. /dʒ/ in having two symbols representing one phoneme, but while /dʒ/ is an affricate, phonetically /gb/ is a ‘double articulation’, [ɡ͡b].

As noted above, apparent ‘spelling pronunciations’ are noted in a wide range of words, for example Nigerian English cater as /ˈkata/. However, this is not an environment-predictable phenomenon, since words such as able are pronounced with /e/ not /a/ (see Jowitt, 2019). The likelihood of a more spelling-based pronunciation is assessed on an individual basis.

The processes Gut (2015) documents as less widespread are not included here, such as variable voicing/devoicing of medial /s/ and /z/, and /t/ affrication. Several other processes affecting BrE /ʃ/ and /ʒ/ are noted in West African Englishes (with knock-on implications for /tʃ/ and /dʒ/) but no approach is adopted over others. This includes particularly salient Ghanaian English alveolo-palatal [ɕ ʑ] realizations, which do not cross phonemic boundaries and are therefore deemed to be allophonic detail (potentially realized on audio recordings but not influencing the transcriptions).

Word stress is quite different from that found in British or American Englishes. Stress is typically shifted to the right in words such as hygiene, challenge but to the left in extent and despite; both phonetics and morphology influence stress placement. With no vowel reduction and seemingly more equal syllable durations, most West African Englishes appear to be more syllable-timed than stress-timed. Gut (2015) directs readers to several additional sources on the interaction between tone and intonation for those speakers of tone languages. Source language tones are one of several factors considered in determining the stressed syllable(s) of OED WAfE transcriptions, but lexical stress is more nuanced than tone alone and is assessed case-by-case based on all-round prominence. However, unlike some of the other more syllable-timed OED World English models, the OED WAfE model does not represent all non-primary-stressed syllables as having secondary stresses. Despite the relatively full vowel qualities in the unstressed syllables, they are still fairly consistently described by WAfE scholars as ‘unstressed’.

Key Sources