Pronunciation model: South African English

View the key for South African English here.

The focus for South African English is not the African variety widely termed ‘Black South African English’, nor the distinct ‘Indian South African English’, but the variety known as ‘White South African English’. Such labels may seem uncomfortable, but as Bowerman (2008 p.168) explains, they ‘are not intended to reflect the apartheid classifications; however, owing to South Africa’s legacy, the correlations between ethnic affiliation and dialect of English remains significant’. In post-apartheid South Africa, despite English being the first language of just 9.6% of the population (2011 census, up from 8.2% in 2001) and 79.2% of the population Black African (compared with 8.9% White), English has grown to be the predominant language for formal public purposes, particularly secondary and higher education. English lies fourth behind IsiZulu, IsiXhosa and Afrikaans in terms of first language, with particular features of the OED pronunciation model for South African English catering for the influences of these languages and the ways in which they are anglicized.

Within White South African English, there are three groupings: the upper-class-associated ‘Cultivated’ form, the middle-class-associated ‘General’ form, and the working-class/Afrikaans-associated ‘Broad’. The variety of focus for the OED model is the General form, although tending towards Cultivated over Broad.

The model is loosely based on the vowels of Wells plus Bowerman’s descriptions of General and Cultivated, with the consultation of native South African members of the OED editorial staff. The Dictionary of South African English on Historical Principles (‘DSAE’, published online in 2015 but originally in 1996) has been the primary source for individual pronunciations, with revisions now being made against those from its more recent cousin (the Oxford South African Concise Dictionary, ‘OSACD’). Like the OSACD, the OED gives anglicized pronunciations of words loaned from other languages, reflecting the pronunciations of first-language English speakers using the words in English contexts. However, unlike the OSACD, clicks are not recognized in the OED and only the anglicized forms are given.


One of the more contentious decisions to be made with the South African English [SAfE] model concerns the KIT vowel. Wells notes a phonemic split, with one variant (close to the English KIT vowel) used in stressed syllables when (a) neighbouring a velar consonant (/k/, /ɡ/, /ŋ/, /x/), (b) after /h/, or (c) word-initially (it is also found with many speakers before /ʃ/ and before /tʃ/, /dʒ/). Otherwise, the sound is somewhat centralized, but Bowerman only uses the ə symbol to represent it in the  Broad variety, where a more open centralized form is noted, more appropriately equating to:


Bowerman equally notes that the Cultivated form does not have such a split. Where the seemingly complementarily-distributed distinction does exist, Wells acknowledges some debate exists as to whether these sounds should be considered different phonemes and subsequently proposes that the centralized variant should be considered merged with /ə/. This approach would mean that the /ə/ would be permitted to be stressed, giving pronunciations such as kiss /kɪs/ but tin /tən/. However, this would appear to overlook the nature of the KIT distinction as one of allophonic variation and may also reflect more of the Broad realization rather than the slightly higher General. At present, the OED model tends towards a more traditionally phonemic (and perhaps more ‘Cultivated’) approach, reflecting both variants with the symbol /ɪ/, although this policy is under ongoing review and the spoken pronunciations will inherently reflect the allophonic variation.

The nurse vowel is usually only of traditional /ɜː/ quality in Cultivated, but in General and Broad it is usually further forward with lip rounding; to represent this particular value would necessitate an additional symbol and render the OED’s set of vowel symbols less economical, hence it is covered under its nearest counterpart. Such rounding is therefore also not reflected at the start of the GOAT vowel.

There is a tendency in General to make diphthong vowels such as those of PRIDE and MOUTH into monophthongs such as /aː/ and /ɑː/, but the more contrasting forms as given by Wells are preferred for these vowels at present (this is also subject to ongoing review), which equate more to Cultivated. The monophthong version of square is included. Similarly, the General THOUGHT/NORTH/FORCE vowels are acknowledged by Wells and Bowerman to be rising and now nearer /oː/ (while Cultivated remains more open).

The HAPPY vowel is noted by Bowerman and his sources to be longer than in many other varieties, but is typically still shorter than the FLEECE vowel so is represented as /i/.

The DSAE transcriptions do not straightforwardly transfer into the system used by the OED, and editors are provided with an extensive conversion table. This covers areas such as the use of short /ɔ/ in DSAE, which is not deemed phonemically distinct but subsumed under /ɒ/ (the precise quality may be dependent on etymology). Recorded pronunciations will have a vowel of short, back, rounded and fairly open quality, with the short length critical for the distinction with THOUGHT/FORCE/NORTH.

DSAE’s /a/ is rendered as OED SAfE /ʌ, ə/ (stress-dependent; this is a further area subject to ongoing review), while /æ/ remains such. The stereotypical raising of the trap vowel to /ɛ/ is deemed a current feature of Broad (see Bowerman’s chapter) and is not reflected here. The SAfE conversion does not consistently predict the British/U.S. vowels though, which may be TRAP, BATH, PALM, STRUT or AGO equivalents and may differ between British and American. DSAE’s long /eː/ is rendered primarily as OED SAfE /e/, some as /eɪ/. There are specific conversions relating to vowels including /iː/ (sometimes becoming /ɪə/), final /e/, and y-spellings (in words such as outstryder, converted to /eɪ/ in SAfE but not necessarily in British/U.S.).

SAfE is transcribed non-rhotically throughout, although some speakers may retain some postvocalic (after-vowel) /r/ sounds in words of Afrikaans origin, and a ‘linking’ /r/ is often found between words as in British English. There is no flapping of /t/ in General or Cultivated (though it is noted in Broad), and occasions where syllabic consonants may be found are largely indicated by bracketed /(ə)/, particularly word-initially (e.g. Ndebele). Straddling phonetic descriptions and DSAE’s system, syllable-final voiced consonants /b/, /d/, /ɡ/, /ð/, /v/, /z/, /ʒ/, /dʒ/ are usually given with both voiced and voiceless variants (baardman /d, t/, baartman /t/), although exceptions exist, e.g. final /ɡ/ in berg. Similarly, a word with a v spelling in syllable-initial position is devoiced, e.g. veld.

DSAE uses a symbol for a voiceless palatal stop (make a voiceless /j/ but push the tongue higher until it completely closes against the roof of the mouth, like /k/ but further forward) which is context-dependently represented as /tʃ/ or /k/. The OED SAfE transcriptions largely retain the /x/ for words borrowed from Afrikaans (e.g. gogga), although it is described by Wells as a uvular sound (even further back).