According to some scholars, the stressed allophone of /ə/ is actually closer than mid ([ɪ̈]). However, other scholars do not distinguish between stressed and unstressed schwas. This article uses the symbol [ə] regardless of the exact height of the vowel.
The central /ə, əː/, not the front /ɛ, ɛː/ are the unrounded counterparts of /œ, œː/. Phonetically, /ə, əː, œ, œː/ have been variously described as mid [ə, əː, ɞ̝, ɞ̝ː] and open-mid [ɜ, ɜː, ɞ, ɞː].
/œ, œː/ are rather weakly rounded, and many speakers merge /œ/ with /ə/ into [ə], even in formal speech. The merger has been noted in colloquial speech since the 1920s.
In some words such as vanaand/faˈnɑːnt/ 'this evening', unstressed ⟨a⟩ is actually a schwa [ə], not [a].
/a/ is open near-front [a̠], but older sources describe it as near-open central [ɐ] and open central [ä].
/ɑː/ is either open near-back [ɑ̟ː] or open back [ɑː]. Especially in stressed positions, the back realization may be rounded [ɒː], and sometimes it may be even as high as the /ɔː/ phoneme. The rounded realization is associated with younger white speakers, especially female speakers of northern accents.
As phonemes, /iː/ and /uː/ occur only in the words spieël/spiːl/ 'mirror' and koeël/kuːl/ 'bullet', which used to be pronounced with sequences /i.ə/ and /u.ə/ respectively. In other cases, [iː] and [uː] occur as allophones of /i/ and /u/ respectively before /r/.
Like /i/ and /u/, /y/ is phonetically long [yː] before /r/.
/ɛ/ contrasts with /ɛː/ only in the minimal pair pers/pɛrs/ 'press' – pers/pɛːrs/ 'purple'.
Before the sequences /rt, rd, rs/, the /ɛ–ɛː/ and /ɔ–ɔː/ contrasts are neutralized in favour of the long variants /ɛː/ and /ɔː/, respectively.
/əː/ occurs only in the word wîe 'wedges', which is realized as either [ˈvəːə] or [ˈvəːɦə] (with a weak [ɦ]).
The sequence /œː.ə/ is realised as either [œː.ə] or [œː.ɦə] (with a weak [ɦ]).
As a phoneme, /æ/ occurs only in some loanwords from English, such as pêl/pæl/ 'pal', as well as in some words such as vertrek/fərˈtræk/ 'departure'. As an allophone of /ɛ/ before /k, χ, l, r/, [æ] occurs dialectally, most commonly in the former Transvaal and Free State provinces.
/a/ has been variously transcribed with ⟨a⟩, ⟨ɐ⟩ and ⟨ɑ⟩. This article uses ⟨a⟩.
/ɑː/ has been variously transcribed with ⟨ɑː⟩ and ⟨aː⟩. This article uses the former symbol.
In some words, such as hamer, short /a/ is in free variation with long /ɑː/ despite the fact that the spelling suggests the latter. In some words, such as laat, the pronunciation with short /a/ occurs only in colloquial language. In some other words, such as aambeeld/ˈambɪəlt/ 'anvil', the pronunciation with short /a/ is already a part of the standard language. The shortening of /ɑː/ has been noted as early as 1927.
The orthographic sequence ⟨ae⟩ can be pronounced as either [ɑː] or [ɑːɦə] (with a weak [ɦ]).
In some instances of the postvocalic sequence /ns/, /n/ is realized as nasalisation (and lengthening, if the vowel is short) of the preceding monophthong, which is stronger in some speakers than others, but there also are speakers retaining [n] as well as the original length of the preceding vowel.
The sequence /ans/ in words such as dans is realised as [ãːs]. In monosyllabic words, that is the norm.
The sequence /ɑːns/ in more common words (such as Afrikaans) is realized as either [ɑ̃ːs] or [ɑːns]. In less common words (such as Italiaans, meaning Italian), [ɑːns] is the usual pronunciation.
The sequence /ɛns/ in words such as mens (meaning human is realized as [ɛ̃ːs].
The sequence /œns/ in words such as guns (meaning favour) is realised more often as [œns] than as [œ̃ːs]. For speakers with the /œ–ə/ merger, these transcriptions are to be read as [əns] and [ə̃ːs], respectively.
The sequence /ɔns/ in words such as spons is realised as [ɔ̃ːs].
Collins & Mees (2003) analyze the pre-/s/ sequences /an, ɛn, ɔn/ as phonemic short vowels /ɑ̃, ɛ̃, ɔ̃/ and note that this process of nasalising the vowel and deleting the nasal occurs in many dialects of Dutch as well, such as The Hague dialect.
Some sources prescribe monophthongal [øː, eː, oː] realizations of these; that is at least partially outdated:
There is not a complete agreement about the realisation of /ɪø/:
According to Lass (1987), it is realised as either rising [ɪ̯ø] or falling [ɪø̯], with the former being more common. The unrounded onset is a rather recent development and is not described by older sources. The monophthongal realisation [øː] is virtually nonexistent.
According to Donaldson (1993), it is realised as [øə]. Its onset is sometimes unrounded, which can cause it to merge with /eə/.
There is not a complete agreement about the realisation of /ɪə, ʊə/
According to Lass (1987), they may be realised in four ways:
Falling diphthongs. Their first element may be short [ɪə̯, ʊə̯] or somewhat lengthened [ɪˑə̯, ʊˑə̯].
Rising diphthongs [ɪ̯ə, ʊ̯ə]. These variants do not seem to appear word-finally. The sequence /ɦʊə/ is commonly realised as [ɦʊ̯ə] or, more often, [ɦʊ̯ə̤], with /ɦ/ realised as breathy voice on the diphthong.
Indeterminate diphthongs [ɪə, ʊə], which may occur in all environments.
Monophthongs, either short [ɪ, ʊ] or somewhat lengthened [ɪˑ, ʊˑ]. The monophthongal realisations occur in less stressed words as well as in stressed syllables in words that have more than one syllable. In the latter case, they are in free variation with all of the three diphthongal realisations. In case of /ʊə/, the monophthongal [ʊ] also appears in unstressed word-final syllables.
The scholar Daan Wissing argues that /əi/ is not a phonetically correct transcription and that /æɛ/ is more accurate. In his analysis, he found that [æɛ] makes for 65% of the realisations, the other 35% being monophthongal, [ə], [æ] and [ɛ].
Most often, /œi/ has an unrounded offset. For some speakers, the onset is also unrounded. That can cause /œi/ to merge with /əi/, which is considered non-standard.
Older sources describe /œu/ as a narrow back diphthong [ou]. However, newer sources describe its onset as more front. For example, Lass (1984), states that the onset of /œu/ is central [ɵu].
In some words, which, in English, are pronounced with /əʊ/, the Afrikaans equivalent tends to be pronounced with /œu/, rather than /ʊə/. That happens because Afrikaans /œu/ is more similar to the usual South African realization of English /əʊ/.
The long diphthongs (or 'double vowels') are phonemically sequences of a free vowel and a non-syllabic equivalent of /i/ or /u/: [iu, ui, oːi, eu, ɑːi]. Both [iu] and [eu] tend to be pronounced as [iu], but they are spelled differently: the former as ⟨ieu⟩, the latter as ⟨eeu⟩.
In diminutives ending in /ki/ formed to monosyllabic nouns, the vowels /u, ɪə, ʊə, ɛ, ə, œ, ɔ, a, ɑː/ are realised as closing diphthongs [ui, ei, oi, ɛi, əi, œi, ɔi, ai, ɑːi]. In the same environment, the sequences /ɛn, ən, œn, ɔn, an/ are realized as [ɛiɲ, əiɲ, œiɲ, ɔiɲ, aiɲ], i.e. as closing diphthongs followed by palatal nasal.
The suffixes ⟨-aad⟩ and ⟨-aat⟩ (phonemically /ɑːd/ and /ɑːt/, respectively) and the diminutive suffix /ki/ are realised as [ɑːci] (with a monophthong), rather than [ɑːici].
In practice, the diphthong [əi] is realised the same as the phonemic diphthong /əi/.
[œi], when it has arisen from diphthongisation of [œ], differs from the phonemic diphthong /œi/ by having a slightly different onset, although the exact nature of that difference is unclear. This means that puntjie 'point' sounds somewhat different than puintjie 'rubble'.
/k/ may be somewhat more front before front vowels; the fronted allophone of /k/ also occurs in diminutives ending in -djie and -tjie.
/dʒ, z/ occur only in loanwords.
/χ/ is most often uvular, either a fricative, [χ] or a voiceless trill [ʀ̥], the latter especially in initial position before a stressed vowel. The uvular fricative is also used by many speakers of White South African English as a realisation of the marginal English phoneme /x/. In Afrikaans, velar [x] may be used in a few "hyper-posh" varieties, and it may also, rarely, occur as an allophone before front vowels in speakers with otherwise uvular [χ].
/ɡ/ occurs only in loanwords. In some environments,[which?][ɡ] is an allophone of /χ/.
/n/ merges with /m/ before labial consonants. Phonetically, this merged consonant is realized as bilabial [m] before /p, b/, and labiodental [ɱ] before /f, v/.
/n/ merges with /ŋ/ before dorsals (/k, χ/). Phonetically, this merged consonant is realized as velar [ŋ] before /k/ and the [ɡ] allophone of /χ/,[can [ɡ] occur after [ŋ]?] and as uvular [ɴ] before /χ/.
/r/ is usually an alveolar trill [r] or tap [ɾ]. In some parts of the former Cape Province, it is realised uvularly, either as a trill [ʀ] or a fricative [ʁ]. The uvular trill may also be pronounced as a tap [ʀ̆].
^ abCited in Lass (1987:117–118). The preview on Google Books makes it unclear whether De Villiers' book is "Afrikaanse klankleer. Fonetiek, fonologie en woordbou" or "Nederlands en Afrikaans", as both are cited at the end of Lass's chapter.
^ abBowerman (2004:939): "White South African English is one of very few varieties to have a velar fricative phoneme /x/ (see Lass (2002:120)), but this is only in words borrowed from Afrikaans (...) and Khoisan (...). Many speakers use the Afrikaans uvular fricative [χ] rather than the velar."
Bowerman, Sean (2004). "White South African English: phonology". In Schneider, Edgar W.; Burridge, Kate; Kortmann, Bernd; Mesthrie, Rajend; Upton, Clive (eds.). A handbook of varieties of English. 1: Phonology. Mouton de Gruyter. pp. 931–942. ISBN3-11-017532-0.
van der Merwe, A.; Groenewald, E.; van Aardt, D.; Tesner, H. E.C.; Grimbeek, R. J. (2012) . "The formant patterns of Afrikaans vowels as produced by male speakers". South African Journal of Linguistics. Taylor & Francis Group. 11 (2): 71–79. doi:10.1080/10118063.1993.9723910.