19.1 Introduction
This chapter focuses on providing, through the analysis of early recorded data, a window on past varieties of English used in South Africa, particularly by those born and raised in that country. The data and speakers analysed for this chapter form part and parcel of a larger collection of recordings obtained from the sound archives of the South African Broadcasting Corporation (SABC) in Auckland Park, Johannesburg, South Africa.Footnote 1 These recordings cover a range of genres, including news broadcasts, recorded speeches, interviews etc.Footnote 2 While there are a number of recordings dated earlier than the three selected for analysis below, they were deemed unsuitable for analysis for a number of different but not mutually exclusive reasons, e.g. poor sound quality as well as uncertainty as to the birthplace or first language of the subject concerned. Thus, by way of example, the earliest data available is from a 1922 recording of Sir Percy Fitzpatrick, born in 1862 in King William's Town in the Eastern Cape (see Map 19.1), author of Jock of the Bushveld and prominent South African financier and politician.Footnote 3 The recording is, however, of a rather poor quality and, in addition, it is clear that Sir Fitzpatrick spent much of his early schooling career in England (to be specific at Downside School near Bath, Somerset) and is thus not a particularly useful subject in terms of trying to gain a perspective on early varieties of South African English (SAfE). Other recordings, particularly during the 1930s and 1940s, are of interest, but it is often unclear whether the individuals concerned were British or South African born or, in some cases, L1-English or L1-Afrikaans. The choice of the three eventually selected recordings was thus based on a number of practical considerations and one theoretical assumption. The considerations were those of a standard of sound quality high enough to allow for acoustic analysis, as well as clear evidence that the subject was South-African born and L1-English. The theoretical assumption was that which lies behind the so-called ‘apparent-time hypothesis’, common in Labovian variationist sociolinguistics as well as in traditional dialectology, and which is explained in, for example, Chambers (Reference Chambers2003: 202–203) in the following way:
For the stages of life beyond young adulthood, our best evidence indicates that … people's speech preserves markers, some subtle and some blatant, that indicate where they have been. For most people, these markers include tell-tale signs of the home-dialect where they spent their childhood, the fossilized slang of a faded adolescence, and the fine adjustments of maturity. Having worked their way through these formative periods, people reach a point where the range of styles and the inventory of socially significant variants are deemed sufficient, at least subconsciously, for all practical purposes in the situations they find themselves in.
Given the above assumption, it was decided to focus on three recordings, all of subjects born in South Africa in the second half of the nineteenth century and thus, in different ways, providing a window to the linguistic past of the English-speaking community of this country. These three speakers were also chosen on the basis that, impressionistically, they appear to represent three different ‘levels’ of speech in early SAfE, each individual using speech that is further or closer to a ‘standard’ or Received Pronunciation (RP) accent of the time, similar in this respect to the three-way distinction provided in, for example, Lanham and Traill (Reference Lanham and Traill1962), i.e. South African Received Pronunciation (SARP) ‘A’, SARP ‘B’ and non-SARP SAfE, echoed in turn by the three modern SAfE sociolects: Cultivated/Conservative SAfE, General/Respectable SAfE and Extreme/Broad SAfE. Naturally, the links between English-spe-=laking South Africa and Britain were particularly strong during the historical period in question, so any analysis of early SAfE speech needs to keep this factor in mind.
Map 19.1 shows Southern Africa during this period. Two areas were under British control at the time: the Cape Colony (in which one of the subjects was born) and Natal (in which the other two were born). The Transvaal and the Orange Free State (now part of South Africa but with different names) were at the time independent republics, subsequently overthrown by the British and eventually incorporated into the Union of South Africa as a result of the Second Anglo-Boer War (1899–1902).
In Section 19.2 below the individual subjects are first introduced in broad biographical terms, and then a broad impressionistic analysis is provided, including information on consonantal features. The emphasis of the main (acoustic) analysis is, however, on the qualities of the vowels of each of the speakers. In each case details concerning the analytical procedure are given and then a full vowel chart is provided for consideration (Figures 19.1–19.3). Each vowel chart is then briefly analysed in terms of what it ‘says’ about the vowel qualities of the speaker concerned. In what follows Wells’ (Reference Wells1982) well-known lexical sets are used as a convenient basis for description and analysis.
Figure 19.1 Mr Flemming's vowel chart (Lobanov-normalised).
Figure 19.2 Ms Murchie's vowel chart (Lobanov-normalised).
Figure 19.3 Ms Gibson's vowel chart (Lobanov-normalised).
Section 19.3, in contrast, focuses on a comparison between the three speakers, and Figures 19.4 to 19.7 provide a useful basis for the relevant comparisons. These place the acoustic results of all three speakers on the same plot, each plot reserved for one of Wells’ (Reference Wells1982) four so-called part systems, the details of which will be explained further below.
Figure 19.4 Part-System A cross-subject comparison.
Section 19.4 details a number of tentative conclusions and generalisations gleaned from the above-mentioned analysis, hopefully opening up avenues for future research.
19.2 The Individual Recordings
As mentioned above, each of these subjects appears to have been differentially influenced by the standard British English of the time, with the first subject showing the least influence, the third subject the most influence, while the second subject seems to lie somewhere in-between. The various analyses below will provide more evidence for this broad assertion.
19.2.1 Mr Flemming
The first recording is of one Mr Charles Flemming (henceforth CF), who was 104 at the time of recording, 1958. He was thus born in 1854. From the interview it is clear that CF is the son of an 1820 SettlerFootnote 4 and born in South Africa in the Eastern Cape, in all likelihood in Port Elizabeth (see Map 19.1). Given the short duration of the recording (2 minutes) it is possible to provide the full transcription:
We didn't stop too long in P.E. [Port Elizabeth], we shifted to Grahamstown, from Grahamstown we shifted to Bedford … my father was an 1820 wagon builder by trade … and of course timber they couldn't buy in those days … but in Bedford there was bush … that he could go and cut his own timber to build wagons … and carriages … and the country was very rough yet … I remember down in the Cape elephants were still knocking about … even after they built the railway … used to stop the train sometimes from going along … once going down to P.E. … the elephant was standing on the line and they had to stop the train until he moved on … coz they didn't worry about shooting them you see … but the lions … used to be about and … wolves … we had to be very careful … lot of wolves about in those days …
I'm quite proud of a … lot of big buildings I put up … and schools … I did build bridges and that was on the Cape-Natal railway … [undecipherable] in Seaview … the little Catholic church there, we put it up just the two of us … just me and my son, my son was still a [ap]prentice under me … me and my [undecipherable] built the government building [undecipherable] … that was in the early days, we had to transport the stuff up from Ladysmith … by wagons … we hired transport … we took a short cut through the Free State … they were old transport wagons … [undecipherable] that way too, it was the best way for them to go, but there were no roads … sometimes I had to walk fifteen miles to my job … you got ten bob a day [undecipherable] you were getting good pay … and in those days we had to work from sunrise till sunset … not like today, hey [laughter], only eight hours … you were so dead-beat that you … glad to have, get something to eat and get into bed.
On an impressionistic level, the first thing one notices is the pronunciation of the word church as [kjəts]. The word prentice is also used instead of apprentice and pronounced as [pɹəntəs], while the town Bedford is pronounced as [bɪtfəd] instead of [bədfəd]. Voiceless stops in CF's speech have a peculiar quality, which might be described as unaspirated,Footnote 5 but which some listeners have even described as ‘Indian-sounding’, i.e. retroflex. The possible influence of CF's advanced age at the time of recording complicates this issue somewhat however.
Another consonantal feature is the use of [w] or even [β] for /v/.Footnote 6 Turning to the vowels, the use of an [æ]-quality in one of the tokens of elephant is quite apparent on an impressionistic level, as well as an occasional fronted quality to the bath vowel, which in the case of two out of three instances of the word transport, has an Australian [a] or even American-like [æ] quality. The two other instances of bath (which could not be analysed acoustically) were, however, rendered with a typically SAfE backed [ɑː]: father and after. David Britain (p.c.) also mentions that ‘two instances of the nurse vowel, “church”, and “work” were short and quite open and, with respect to frontness, central. Trudgill reports this for Norfolk.’Footnote 7
While the sound quality of the recording was, understandably, not up to modern standards, it did allow for the acoustic analysis of 165 vowel tokens, the results of which are represented in Figure 19.1. In Figure 19.1 (as well as Figures 19.2 and 19.3) the traditionally short vowels of English are bolded, italicised and the name of the relevant lexical set is abbreviated. Although for most of Wells’ (Reference Wells1982) lexical sets there was enough data to determine a mean value, there were a few lexical sets that were not represented at all (e.g. cure and near) or where there were only one or two tokens; thus in the case of square there were two measurable tokens of the word there (averaged out for the purposes of Figure 19.1), and in the case of nurse there was one analysable token of early. The remaining lexical sets were represented by a minimum of three (i.e. foot and bath, with three tokens of transport) to a maximum of twenty-three tokens, i.e. face. In the case of all tokens, F1 and F2 measurements were taken, using Praat (Boersma and Weenink Reference Boersma and Weenink2013), in the case of the traditionally short vowels at the temporal mid-point, while in the case of the traditionally long English vowels (both monophthongal and diphthongal) two measurement points were used, roughly at the 25% temporal point and the 75% temporal point. The formant data for CF was also subject to normalisation (Lobanov Reference Lobanov1971) to render it comparable with data from the other two female subjects.
Turning to Figure 19.1, a brief perusal of this subject's vowel chart alerts one to the following main features:
1. The concentration of vowels in the front-low area of the vowel space. Within this area, the close approximation of dress (dr in Figure 19.1) to trap (tr) is particularly noticeable;
2. Of particular interest too is the relatively fronted bath (ba) vowel, confirming the original impressionistic analysis, and which overlaps in quality with the strut (st) vowel in a very Australasian-like fashion. This, in fact, provides some evidence for the claim made in Bekker (Reference Bekker2012) that the backing of bath in SAfE is perhaps more the result of later developments in the formation of SAfE rather than due to inheritance from the 1820-input British dialects; this assessment should, however, be tempered by the observation that the unanalysed tokens of bath (father and after) were, impressionistically, fully backed, thus pointing to variability in this regard in the speech of CF;Footnote 8
3. Apparent too are the rather narrow price (pr) and mouth (mo) vowels; in the case of the latter we have a somewhat fronted onset with an (unusual) slight off-glide to a lower position; and
4. CF's kit (ki) shows a classic SAfE kit-Split (Wells Reference Wells1982: 612–613), with clear polarisation in phonetic space between kit in high-front position (kitV in Figure 19.1Footnote 9), kit before tautosyllabic /l/ (kitL) and unmarked kit (kitU).
19.2.2 Ms Murchie
The second subject is one Ms Murchie (henceforth MM), recorded in Durban in 1970, when the subject was ninety-seven years old (i.e. born in 1873). It is clear from the recording that she was born in Durban, the daughter of a settler.Footnote 10 There is some evidence in the recording to suggest that her mother arrived in Durban in 1863. While it is clear that the subject spent most of her childhood in South Africa and received her schooling there, it is equally clear that time was also spent abroad in England, particularly in early adulthood.
On an impressionistic level, MM sounds much more ‘standard’ than CF, although it is very clear that she is nonetheless South African, echoing Lanham and Traill's (Reference Lanham and Traill1962) later SARP ‘B’. Some features which stand out as ‘standard’ are the use of [ɪ], as opposed to [ə ∼ ɤ], in words such as did and milk and a clearly unrounded goat nucleus. The occasional nurse token also sounds unrounded, unlike the traditionally rounded SAfE nurse, i.e. [ɸ̈ː]. Less ‘respectable’ (and more typically SAfE) features, however, include a clearly fronted mouth vowel.
The analysis of the acoustic data generally followed the same principles as in the case of CF (formant analysis using Praat and Lobanov normalisation), although in this case there was much more data. In the case of MM, therefore, a target of twenty tokens per lexical set was aimed for, reached in all cases except near (sixteen tokens), choice (ten tokens) and cure (one token of sure). In the case of kit, forty tokens were analysed, given the special interest in the SAfE literature in the allophonic distribution of this lexical set, i.e. the kit-Split. An attempt was made to not analyse more than two tokens of the same word, and this was mostly achieved, except in one or two lexical sets (e.g. square, the analysis of which had to depend on numerous cases of the word there). The overall number of tokens analysed for MM was 369. Figure 19.2 provides a graphic representation of the relevant acoustic data. Some of the more obvious features of MM's vowel space include the following:
1. Clearly diphthongal price (pr in Figure 19.2) and face (fa) vowels;
2. A fronted (and raised) mouth (mo) vowel, thus suggestive of Diphthong Shift (Wells Reference Wells1982) and the so-called price-mouth Crossover;Footnote 11
3. A backed bath (ba) vowel (identical in quality with lot (lo));
4. A narrow goat (goa) vowel;
5. A lack of phonetic polarization with respect to the kit (ki) vowel,Footnote 12 i.e. no kit-Split; and
6. A raised dress (dr) vowel.
19.2.3 The Lady in White (Ms Gibson)
This speaker has the most (old-fashioned) RP-like dialect of the three South-African born speakers, and from the interview it is clear that she, and her family, had close ties with England. Her speech is most likely characterisable as Wells’ (Reference Wells1982) ‘near-RP’.Footnote 13
The exact date of the interview is unclear, but the subject (LW henceforth) was born in Durban in April 1888 and schooled in that city as well. Her grandfather arrived in Natal in 1851 (from King's Lynn, Norfolk) and her father appears to have joined him in 1881 (from Woolwich, London). The title ‘Lady in White’ appears to relate to her singing career and, from her own account, she seems to have achieved a fair degree of international success in this regard. Her father was a prominent public figure of the time with, for example, ties to the Prime Minister of Natal in the late nineteenth century, and much of the interview is taken up with a description of her and her family's ties with numerous luminaries both in South Africa and abroad. She is decidedly upper class, an assessment clearly reflected in her speech. It should also be mentioned that the interview was of a prepared nature, thus lending to the use of a particularly formal style of speech: the interviewer had clearly sent the subject a set of questions before the interview and it is also clear that the subject had spent time constructing notes of some sort for herself – at some junctures in the interview it seems in fact as if she is reading.
On a consonantal level, two outstanding features of her speech include the clear, virtually constant, maintenance of a /hw ∼ w/ distinction (e.g. which vs. witch) and the use of an occasional (highly affected) tapped ‘r’ (e.g. in inspi[ɾ]ation), which Wells (Reference Wells1982: 282) characterises as ‘typical of some varieties of U-RP’, see also Fabricius, this volume. On a purely impressionistic level, her goat and face vowels stand out as being particularly front, close and narrow and her mouth vowel is fully backed. Her square vowel often has an in-glide to [ə]. The price vowel is generally fronted and occasionally glide weakened and there is no evidence of a kit-Split in her speech.
In terms of the acoustic analysis of vowel quality, the same procedure was followed as in the case of MM, though the slightly longer interview allowed for the collection of a little more data; while the same lexical sets as in the case of MM were relatively poorly represented, more tokens could be analysed: nineteen for near, fifteen for choice and six for cure. The total number of tokens analysed was 381. Figure 19.3 provides a graphic illustration of the relevant results. Some of the features which stand out and which generally confirm the broad impressionistic analysis above are as follows:
1. Very high, front and narrow face (fa) and goat (goa) vowels;
2. A backed nucleus for mouth (mo), which is also glide weakened;
3. A clear in-gliding diphthong in the case of square (sq); and, less clearly, for nurse (nu);
4. A raised trap (tr) and dress (dr) vowel; and
5. No evidence of a Second-Force Merger (Wells Reference Wells1982: 237)Footnote 14 with respect to cure, i.e. [ʊə ∼ ʊɐ], not [ɔː]-like.
19.3 Comparative Analysis
Figures 19.4 to 19.7 reconfigure the data in Figures 19.1 to 19.3, in order to allow for an easier comparison of this data. Figure 19.4 focuses on Wells’ (Reference Wells1982) Part-System A (i.e. the traditionally short English vowels: kit, dress, trap, strut, lot and foot); Figure 19.5 deals with Part-System B (fleece, face, price and choice), Figure 19.6 with Part-System C (goose, goat and mouth) and Figure 19.7 with the traditional Part-System D glides-to-[ə] (i.e. near, square, cure, bath, thought and nurseFootnote 15). In all these figures CF's data is in caps, bolded and italicised (e.g. his trap is ‘TR(CF)’), MM's data is in small-letters, bolded and italicised (e.g. ‘tr(mm)’), while LW's data is in caps but not bolded or italicized (e.g. ‘TR(LW)’).
Figure 19.5 Part-System B cross-subject comparison.
Figure 19.6 Part-System C cross-subject comparison.
Figure 19.7 Part-System D cross-subject comparison.
19.3.1 Part-System A
Beginning with Part-System A and with foot (fo) in Figure 19.4, the first difference we note is that CF's foot is far more back than either LW's or MM's. We also note, from Figure 19.1, that in CF's case foot is nearly identical in quality to his kit vowel before final, tautosyllabic /l/ (kitL), a feature CF shares with modern-day SAfE. While foot in both MM and LW is more front than in CF's case, there is no evidence for an overlap with pre-/l/ kit (Figures 19.2 and 19.3), mainly due to a lack of a kit-Split in the idiolects of MM and LW.
With respect to lot (lo), MM's vowel is separate from the other two speakers and appears to be discernably lower. The subjects’ strut (st) vowels are virtually identical. The same applies to trap which is quite raised across the various subjects, e.g. if we compare trap to the position of strut or to the nuclei of price for LW and MM as illustrated in Figure 19.5. A useful further basis for comparison are the relatively modern RP mean acoustic values provided in Cruttenden (Reference Cruttenden2001: 99) for trap in citation-form and connected speech for female speakers, i.e. 1,011 Hz and 1,018 Hz respectively. In comparison, the two female speakers in this study (LW and MM) have mean values of 602 Hz and 710 Hz respectively. As confirmed by Britain (p.c.),Footnote 16 trap was often rather raised in both the traditional RP of the time as well as in many rural southern English counties, a fact which no doubts accounts for the similarity in this regard across subjects.
The similarity across subjects ends with dress with CF having a clearly lowered quality in comparison with the other two subjects. As was the case with trap, dress had a particularly raised quality in the RP of the time as well as in dialects from the south-east of England (Torgersen and Kerswill Reference Torgersen and Kerswill2004; Cruttenden Reference Cruttenden2014; Wells Reference Wells1982). CF's kit vowel also clearly has a more central quality, pointing to the already mentioned polarisation in phonetic space, i.e. the kit-Split. The kit-Split has generally been either attributed to the influence of Afrikaans on South African English (Lanham and Macdonald Reference Lanham and Macdonald1979) or to a ‘nineteenth century vowel shift … in which raising of /æ/ towards [ɛ] and raising of original /ɛ/ to [e] seem to have forced (most of) original /ɪ/ to centralise’ (Lass Reference Lass and Mesthrie1995: 96).
19.3.2 Part-System B
Turning to Figure 19.5, while all the subjects have relatively monophthongised fleece (fl) vowels it is interesting to note that CF's fleece vowel is slightly closer, a particularly high-front (and monophthongal) fleece vowel being characteristic of modern Broad SAfE. A monophthongal fleece is characteristic of SAfE more generally (contrasting this southern hemisphere variety with the Australasian English dialects). From Figure 19.5 the narrow, close nature of LW's face (fa) vowel is apparent, a feature echoed in modern prestigious varieties of General SAfE (Lass Reference Lass and Mesthrie1995: 99). Neither CF's nor MM's face vowels seem particularly lowered, although it would appear that CF's face is somewhat glide weakened in comparison.
Turning to price (pr), a prominent social variable in modern-day SAfE, with backed glide-weakened varieties carrying the least and fronted (optionally glide weakened) varieties carrying the most prestige (Lass Reference Lass and Mesthrie1995: 99), it is interesting to note that both LW's and MM's price vowels are relatively fronted, with LW's price showing the most fronting and most extensive glide weakening of the two. While CF's price is glide weakened, it is certainly not backed and, in fact, shows a somewhat centralised quality reminiscent of Labov's (Reference Labov1963) values for price in Martha's Vineyard. This reflects a variant that has not survived into modern-day SAfE and is perhaps attributable to a number of (non-south-eastern) British dialects which had not at that time undergone the full-force of the Great Vowel Shift, thus retaining an [əɪ]-like quality for the price vowel.Footnote 17 There was no data for choice (ch) for CF. In the case of the other two subjects, there is a definite glide to a higher and/or fronter position, although it is difficult to account for the difference in glide direction.
19.3.3 Part-System C
Turning to Figure 19.6, we note that goose across the three subjects is virtually identical and, while all show some evidence for slight diphthongisation in the acoustic record, this is certainly not apparent on an impressionistic level. Furthermore, while MM and CF's goat begin at the same point in phonetic space, MM's goat appears prone to glide weakening. LW's goat vowel is substantially different to the other two in the sense of being very much glide weakened and having an extremely close (RP-like) nucleus: this is no doubt the Conservative RP variant mentioned by Wells (Reference Wells1982: 294) who also claims that it is now ‘widely considered “affected”, and has ceased to be fashionable among younger speakers’. Modern SAfE, however, still places positive indexical value on goat values with a relatively close (and generally fronted) nucleus (whether rounded or unrounded) (Lass Reference Lass and Mesthrie1995: 100). We note the relatively fronted mouth-variant in the case of MM (echoing many broader variants of modern-day SAfE), the fully backed conservative mouth in the case of LW and, as with price, a relatively centralised variant in the case of CF.
19.3.4 Part-System D
The relatively fronted bath vowel in the case of CF is clearly apparent from Figure 19.7, although, again, this needs to be qualified by the fact that it is based on an acoustic analysis of two tokens of the same word (i.e. transport), as well as by the fact that the other (unanalysed) bath tokens (after, father and one token of transport) were fully backed in CF's recording. thought is difficult to analyse, although it appears that CF has a slightly retracted variant in comparison to the other two subjects. MM seems to show some evidence for the Second-Force Merger with respect to cure (i.e. [ɔː] and not [ʊə]), although this is based on only one token of sure. In the case of LW there is clear evidence for the old-fashioned RP [ʊə]-quality. nurse has a fronted (and, impressionistically, occasionally unrounded) quality in the case of both MM and LW, with LW's nurse showing some evidence of an in-glide. Echoing Britain's (p.c.) impressionistic analysis of CF's speech (see Section 19.2.1 above), the one token of early is substantially lowered. As mentioned above, there were no near tokens in CF's speech; MM's and LW's near are, however, substantially different. Both are in-gliding, but in the case of LW substantially lowered. MM's near echoes modern SAfE values, basically [ɪə], while LW's has a [eɐ] quality.Footnote 18 Both CF's and MM's square is based on limited data, but the difference between the three subjects does echo the modern distinction between Broad SAfE [eː], General SAfE [ɛː] and Cultivated SAfE [ɛə], with LW's square having a clear RP-like in-glide (Wells Reference Wells1982: 293).
19.4 Conclusion
While the features of each subject's speech, as well as their differences, are interesting in their own right, and provide a tantalising ‘window’ on the varieties of English prevalent during the second half of the nineteenth century, are there any broader generalisations or conclusions that can be drawn from the above analysis? I believe there are, although these are necessarily tentative and speculative, based as they are on the speech of only three subjects (one of which ‘provides’ only limited data), but they do provide a number of possibly fruitful avenues for further research.
Beginning with CF, it seems reasonably clear that his speech is coloured by the original British input and contains features that were eventually ‘ironed out’ once SAfE become a fully focused variety. Assuming that a koine developed in the Eastern Cape during the mid-nineteenth century and given Trudgill's (Reference Trudgill2004: 23) fifty-year yardstick for the focusing of such a new dialect, it would have come into its own around about 1870, almost twenty years after CF was born, meaning that he in all probability spent his childhood in a linguistic environment characterised by a substantial degree of inter- and intra-individual variability. One provisional, though tantalising, conclusion, relating specifically to the variability of CF's bath vowel, is that start-Backing (a characteristic of modern SAfE, but not Australasian-English) only really completed itself during the late nineteenth-century, a fact reflected by the unexceptionless use of a backed bath in the case of MM and LW and also providing some support for the hypothesis contained in Bekker (Reference Bekker2012), i.e. that modern SAfE is, in a substantial sense, a product of late nineteenth-century developments, rather than being directly traceable to the original 1820 input. Another tantalising ‘lead’ relates to the fact that CF's speech shows a fully polarised kit vowel, which means that this feature of modern-day SAfE is perhaps traceable to an original input into SAfE and is not an endogenous development, as for example argued for in Lass and Wright (Reference Lass, Wright, Eaton, Fischer, Koopman and Van der Leek1985, cf. Lass Reference Lass and Mesthrie1995), who claim that the centralisation of kit in SAfE is most likely the result of pressure from the other two short front vowels, i.e. trap and dress, the raising of which, in terms of Lass and Wright's (Reference Lass, Wright, Eaton, Fischer, Koopman and Van der Leek1985) scheme, led to a front short-vowel chain shift in SAfE. We note, in this regard, CF's substantially lowered dress vowel. It is perhaps possible, therefore, that the SAfE kit-Split was perhaps a direct inheritance and, in fact, the initiator of the relevant chain-shift.Footnote 19
As intimated above, almost all descriptions of SAfE point to a three-way division of this colonial variety, mainly in terms of its relationship with British-based standard varieties (i.e. the various forms of early RP). Thus the early Lanham and Traill (Reference Lanham and Traill1962) note a still dominant exonormative orientation in the speech of many English-speaking South Africans and propose a division between SARP ‘A’, SARP ‘B’ and non-SARP SAfE, echoing later divisions of SAfE into Cultivated, General and Broad sociolects, with the first sociolect still generally modelled on British-based standard varieties. The speech of the three subjects seems to loosely reflect this trichotomy and, more importantly, many of the prestige values of modern-day SAfE (e.g. narrow and close face; fronted price; close, front and narrow goat) appear to be reflected in the speech of LW, which raises the question of how much SAfE has to this day been influenced by prestige varieties of the past. We note too that a raised trap and dress vowel, often associated with broader idiolects in modern-day SAfE, are in fact present in both MM and LW's speech, both of which were clearly influenced by the British standard of the time and LW's speech in fact having the most raised trap and dress vowels of all three subjects. It is clear, therefore, that, at the time, a raised dress and trap did not have the negative indexicality current in modern-day SAfE. It is, thus, perhaps worthwhile considering the possibility that raised SAfE trap and dress were not endogenous developments in SAfE, but are, in fact, (at least partly) residues of previously prestigious variants that have now, following modern varieties in the south-east of England (cf. Torgerson and Kerswill Reference Torgersen and Kerswill2004), been reanalysed on an indexical level and, in fact, show reversal in certain prestigious subvarieties (Bekker and Eley Reference Bekker and Eley2007; Bekker Reference Bekker2009: 204; Chevalier 2016) – a form of colonial lag in Trudgill's (Reference Trudgill2004) sense.
The above possibilities, however, remain speculative, and naturally further research will have to be conducted in order to determine if they have any true merit. Thus, while this chapter has hopefully provided an interesting ‘window’ on the past of SAfE, much more work needs to be done in order to unravel the processes and forces operative in the establishment and development of this southern hemisphere variety of English.
