Pronunciation model: Caribbean English

View the key for Caribbean English here.

Of all World English varieties currently addressed by the OED, delineating a ‘Caribbean English’ provides the greatest challenge. The Caribbean is defined by the OED (2b) as the region consisting of the Caribbean Sea, its islands (including Cuba, the West Indies, Puerto Rico, and others), and the surrounding coastal areas. There is vast phonetic and phonological diversity across this region, but separate models for each nation would be impractical given the small number of terms currently in the OED from some of the Caribbean nations. The features represented, therefore, are some of the commonalities between forms of English spoken in the Caribbean, with preference weighted towards slightly more widely-spoken varieties, rather than representing the pronunciations of any particular speaker per se. However, the complexities of deriving this model warrant more extended discussion than for other varieties.

The following regions are evaluated in determining a ‘Caribbean English’ variety, in descending order of influence in accordance with population and linguistic demographic data: Jamaica (both Jamaican English and Jamaican Creole), Trinidad & Tobago (English and Creole), Guyana, Belize, the Bahamas and Barbados (English and the Creole known as Bajan). These are countries and territories in which the predominant language is English and where the population is greater than 200,000. Belize is an unusual case in having an official language of English but high levels of Spanish and Creole forms which render it different in some key respects. The Englishes and/or creoles of smaller island states and territories are also considered, under Aceto’s (2008) broad category of ‘Eastern Caribbean English-derived language varieties’ (hereafter ECEDs).

Complicating matters is that many Caribbean nations have a ‘lectal’ continuum of forms of English, varying from an acrolect (a prestigious and largely ‘standard’ variety spoken in mainly formal situations) to a basilect (a low-prestige and colloquial form which differs greatly from the standard). Between the two lies the mesolect, often the most widely spoken form. The basilect may frequently be sufficiently far from the acrolect that it may be largely unrecognisable to someone only familiar with the latter, but the mesolect will both be recognisably ‘English’ whilst exhibiting some prominent creole influences. Where possible, it is the mesolectal forms that are compared across varieties for the purposes of the OED.

Although Wells (1982) gives some description of the Caribbean varieties, the vowels are generally far more similar across varieties than is indicated by the contributors to the 2008 Varieties of English series. Readers are directed to the chapters by Aceto, Blake, Devonish & Harry, Childs & Wolfram, and Youssef & James in particular.

Having compiled and reviewed the descriptions of each variety, the vowel systems of those Caribbean varieties for which sufficient data could be obtained are compared. Descriptions of Jamaican English, Jamaican Creole, Trinidad & Tobagan Creole, Guyanese English, Bajan, and a form straddling key vowel features of Anglo-Bahamian and Afro-Bahamian Englishes are cross-referenced for commonalities across vowel qualities to propose the model below. Other varieties with less data available (such as the ECEDs, which no single feature unites) are more informally considered in determining between competing representations. The vowels are discussed further below.

 KIT ɪ FLEECE iː NEAR eː (r) HAPPY i
 BATH a: START aː(r) FACE eː ADDED ɪ
 CLOTH ɔː NORTH oː(r) VOICE ɔi + ai PIANO i
 STRUT ʌ FORCE oː(r) MOUTH aʊ + ɔʊ AGO a

KIT, DRESS, TRAP, BATH and FOOT vowels demonstrate reasonable consistency across the region and are straightforward to propose an encompassing set. STRUT‘s most common phonetic realisation region-wide is rounded and slightly backed compared with the British form, but overall the variation in this sound is best considered with reference to /ʌ/ rather than /ɔ/.

LOT and CLOTH are more complex. In several of the more rhotic creoles, these are not qualitatively distinct from each other nor from TRAP, but in other varieties a clear distinction remains. In varieties such as Bahamian English, there is a qualitative difference between the three (/ɑ/, /ɔ/, and /a/ respectively). The original Caribbean English OED model adopted this distinction but with length-marking on CLOTH, for which an equivalent distinction exists in Devonish & Harry’s Jamaican varieties. This was also in keeping with the variant pronunciation quality difference described by Wells for Guyanese and Jamaican varieties. However, the quality of CLOTH as /ɔː/ does not seem reflective of the majority of Caribbean English speakers. In consultation with Jeanette Allsopp, the OED model retains the LOT-TRAP distinction, but now merges LOT and CLOTH.

FLEECE and GOOSE are consistent across the region with minimal variation. Although several varieties do not utilize central vowels as in NURSE, several more ‘Standard’ varieties retain them, if in a reduced set of environments. NEAR and SQUARE vowels are almost uniformly homophonous, while the limited references to CURE indicate common homophony with FORCE vowels. START and PALM are identical in quality and length with BATH. NORTH and THOUGHT demonstrate more notable variation, with a back rounded quality at one extreme and far more anterior unrounded quality at the other. The proposal here is to offer both, as is current common practice for US THOUGHT vowels. Rhoticity is discussed in a separate section below.

The front-closing point of PRIDE and VOICE is frequently described as [i] rather than [ɪ] (except in Wells), though these are particularly varied and many are slightly more open. The homophony between these vowels is highly variable also, hence two vowels for are given at either end of the continuum for VOICE. FACE and GOAT are monophthongs (unlike the two-quality British versions) in the vast majority of the region, but MOUTH is again highly varied.

Schwas are relatively rare in most mesolectal-basilectal Caribbean varieties, and are not proposed here. Nor are intermediate values /ᵻ, ᵿ/. Instead, LETTER and AGO are TRAP /a/ and the more central /ʌ/), HAPPY, PIANO and BECAUSE are reduced-length FLEECE /i/, BEAUTIFUL is FOOT /ʊ/, and RABBIT and ADDED are KIT /ɪ/.

The model follows Devonish & Harry’s Jamaican English description regarding nasal airflow on vowels, so there is no variable deletion of a nasal consonant (nasalisation on the vowel is therefore deemed pure allophonic detail and not indicated). Similarly, there are no ‘underspecified’ vowels.

Regarding consonants, the OED model contains 22, all as in British English but without the dental fricatives. Despite reports of both Jamaican and Bahamian English that /θ/ as /t/ is less frequent than that of /ð/ as /d/, the vast majority of the less ‘Standard’ forms have stopping of both and this is reflected here.

One of the most varied features of the Caribbean varieties is rhoticity (/r/ after vowels). The OED Caribbean English system offers a bracketed /r/ wherever it may feasibly be encountered, though this only rarely applies to LETTER even in the more rhotic varieties so is not indicated here.

Syllabic consonants may occur, but in a more limited fashion compared with British and American varieties. Nasal (/m n ŋ/) consonants can be made syllabic following consonants made in the same part of the mouth (‘homorganic consonants’, such as in the word hidden), but following Jamaican Creole syllabic /l/ can only after velar consonants (/k/, /ɡ/, /ŋ/).

Coalescence is not particularly indicated in the OED model. This follows the second of the three groups of Jamaican English speakers described by Devonish & Harry, such that /j/ is consistently retained for historical /tj, dj/ sequences such as Tuesday, dew but not over-applied (chew, junior remain with affricates). However, /sj/ is reflected as /ʃ/ (and /zj/ as /ʒ/).

Following many Caribbean creoles, /t/ and /d/ in syllable-final consonant clusters are often not pronounced after another stop or fricative sound of the same voicing (e.g. /-ft, -st, -kt). Like many West Indian ECEDs this is also applied to word-final /d/ (send, build). However, unlike some of the deeper Creoles, the consonant is retained if morphemically significant (i.e. in an inflectional suffix, so passed would retain it, but not past).

Some features considerably different from the British English are only attested in particular subvarieties across the Caribbean and are not part of the model despite extensive consideration. Following Jamaican English and most of the ECEDs, initial /h/ is retained but no /h/-insertion is indicated. While many varieties (particularly creoles) have particular rules regarding nasals after particular vowels, they too are highly variable. However, final morpheme ­–ing is transcribed /ɪn/. Several varieties within the region have a /w/-/v/ merger, but in different ways. Some more commonly have allophones of /v/ replacing /w/ in watch, waste while others vice-versa and others have both. The most logical compromise is to retain the Standard clarity regarding /w/ and /v/. Finally, ‘palatal stops’ (resembling /k g ŋ/ but made slightly further forward in the mouth, on the hard palate), a particular feature in Jamaican Creole influenced by the African Twi dialect, are not given bespoke symbols but anglicized to /kj/, /gj/ and /nj/ in parallel form to Cassidy & Le Page (2002).

 Some of the most distinctive features of Caribbean English, however, lie in the assignment of stress and intonation, with syllabic pitch separated from stress in a notable departure from British and American standard varieties. Stress must be assessed on a case-by-case basis, with the acknowledgement that several Caribbean varieties assign final stress where metropolitan varieties have initial stress, e.g. realize, celebrate, kitchen. Intonation is not indicated in the OED’s transcriptions but spoken pronunciations will invariably reflect this aspect. Readers are directed to the discussion in Allsopp (1996) for further details on this topic.