This article reports on an acoustic investigation into aspects of the vowel system of Genadendal Afrikaans (GA), an important variety of so-called Coloured Afrikaans which is spoken in the Western Cape. The aforementioned vowel system is compared with that of Standard Afrikaans (SA). The comparison is made on the basis of a review of available literature on the phonetics of spoken Afrikaans varieties.
Historically, GA is an important sub-variety of Cape Afrikaans, but its vowel system has not been described in its entirety as yet Certain aspects of its phonology are dealt with by De Villiers and Ponelis (1987) and Klopper (1983).
Non-standard varieties whose phonologies have been considered include Griqua Afrikaans, as described by Henning (1983), Nieuwoudt (1990) and Van Rensburg (1984); Namaqualand Afrikaans, which is mentioned in limited contexts (e.g. by De Villiers and Ponelis 1987); and Kharkams Afrikaans (also a variety of Namaqualand), touched on in Links (1989).
The sound system of SA is well-documented in standard works such as those by Coetzee (1982), Combrink and De Stadler (1987), De Villiers and Ponelis (1987), Le Roux and Pienaar (1928), Van Wyk (1977) and Wissing (1971; 1982). Articles by Klopper (1987), Raubenheimer (1998), Van der Merwe et al. (1993) and Wissing (2006, 2010, 2011, 2012, 2014, 2017a, 2017b) also deal with issues relating to the acoustic properties of Afrikaans.
Two sets of speakers of GA were included in this investigation. Out of 20 available recordings, four each were selected for analysis: four from the younger group (average age of 16 years) and four from the older group (average 71 years). For SA, only a single speaker was used. At the time of the recordings she was a prominent radio personality; she is taken as a representative speaker of SA.
A word list with 122 items was used as stimulus material: all Afrikaans vowels and the three true diphthongs are represented. In the case of SA, the allophone [æ] is represented in the word ek(I; first person pronoun, nominative case). Its inclusion is intended to help determine the precise distribution and nature of this particular allophone of the phoneme /ɛ/.
Two Philips DVT3500 recorders were used: one with a Sennheiser ME 3-EW microphone, the other with an AKG C1000S microphone. In most cases, recordings which are suited to the measurement of vowel quality and vowel duration were obtained. The readings by the SA speaker were done in an RSG (Radio Sonder Grense) radio station studio and are of excellent quality.
Recordings were first transcribed in Praat, following which they were reworked into their expected phonetic forms using letter-to-sound rules (i.e. grapheme-to-phoneme conversion). The Afrikaans pronunciation dictionary compiled by Davel and De Wet (2010) and Default&Refine algorithm by Davel and Barnard (2008) were used for this purpose. Statistical models (namely Hidden Markov models) were determined, given the expected phonetic sequences and sound recordings as described in Van Niekerk and Barnard (2009). Forced alignment, a limited form of speech recognition, was used so that these models could provide a suitable alignment of phone sequences on the one hand and sound sequences on the other.
Relevant acoustic parameters, mainly those of the first two vowel frequency formants, F1 and F2, were extracted automatically by standard methods. This data was subsequently analysed to serve as the basis for acoustic vowel plots.
From the respective positions of GA /i/ and /y/, relative to each other (cf. figure 4 in the main text), it can be deduced that /y/ is fairly unrounded: it lies on nearly the same vertical axis /i/ (strongly rounded variants of /y/, as in Dutch, occur far more to the right of unrounded /i/ on a vowel plot; the same is true of SA – cf. figure 2). It is striking that /y/ is so much higher than /i/ – it is usual for the rounded members of such pairs to occur toward the right of the vowel space, as compared with unrounded members, i.e. they are mostly characterised by a lower value for F2 than their unrounded counterparts.
/ɔ/ is clearly a rounded back vowel, higher than the usual mid-high position; /ә/ and /œ/ are typically neutral vowels; but, similar to /ɛ/, are mid-high, which is higher than would be expected for SA.
F2 values for rounded members of the pairs rounded: unrounded are lower than those of rounded alternatives. (Canonically round vowels are realised as such: their canonically unrounded counterparts have lower F2 values.) /œ/ does clearly lie on the same horizontal axis as /ә/, though. Phonetically, the acoustic results suggest that GA speakers typically produce their phonemically rounded /œ/ with an unrounded quality.
The fact that the short vowel /ɑ/ and the long /a/ occur so close to each other on the acoustic vowel chart is indicative of their sharing a vowel quality; the difference is only to be noticed in vowel duration (an average duration of 129 ms was measured for short /ɑ/, with 211 ms for long /a/ in this group of speakers). Due to its being at the middle of the vowel chart, /a/ is not rounded either, unlike the pronunciation of many younger speakers; long /a/ for them lies nearer to the bottom-right corner of the vowel space. GA /a/ lies to the right of short /ɑ/, as is also typical of SA. The middle position of both vowels implies a "flat" pronunciation perceptually.
GA evidences a fairly strong preference for markedly rounded vowels’ being realised with an unrounded quality; a feature which is also common to General Afrikaans – even in many formal contexts. The articulations of the high-back vowel /u/ and that of long /a/ are noticeably different from those of SA. /u/ is a classical rounded, high-back vowel, with F2 values below 1 000 Hz, as measured across both groups; values for the younger group are lower still than those for the older group (803 Hz for the former and 936 Hz for the latter). Both these measurements are considerably lower than comparable measurements for especially younger speakers of SA. Wissing (2007; 2014) reports F2 values in excess of 1 200 Hz for this group of speakers, which suggests a trend toward centralisation; which cannot be said of the vowel /u/ in GA.
Unlike in SA, the production of /ɛ/ before /r k x/ does not condition lowering, while this is the case preceding /l/ – in fact, this environment favours strong palatalisation. This is true of both the older and younger speakers of GA. The trend toward palatalisation is found in many varieties of Afrikaans, especially among so-called coloured speakers.
Seeing that the vowel systems of the two age groups largely correspond to each other, it would appear that there is a high level of homogeneity in this speech community. With the continued increase in exposure to other language varieties and SA, including on radio and television, it would not be an unreasonable assumption that further generations' speech habits will deviate from the status quo (as taken from the younger and older groups). This topic should be of utmost interest to persons working on language variation and change.
Keywords: acoustic phonetics; acoustic vowel plot; Afrikaans; articulatory phonetics; diphthong; diphthongised vowel; varieties; vowel system
Lees die volledige artikel in Afrikaans: Die vokaalstelsel van Genadendalse Afrikaans: ’n verkenning