On South African English vowel system

Abstract

In practice, phonetic analyses are very often characterized by two features which could, potentially, limit their validity. The first of these features relates to the data used. Two aspects of this feature can be mentioned. First, linguists tend to describe speech sounds as they appear in words seen in isolation of their context. Labov (1971a, 1971b, 1972, 1975) and Labov, Yaeger and Steiner (1972:ch. 2, henceforth LYS) have clearly shown that, due to the large degree of self-monitoring present ~n situations like these, such data are easily slanted in the direction of the assumed prestige norms of pronunciation in the cOIlUllunity and thus represent the (ultra-) formal standard rather than the knowledge underlying the vernacular of the community which is, presumably, what the linguist wants to investigate. What one needs, therefore, is data from free and spontaneous speech. The second aspect, which is closely relctted to the first, is the practice of linguists to rely (almost) solely on introspective data. In attempting to describe the vowel segments of a language, linguists often resort to articulating what they intuitively judge the vowel to be, and then set about describing its supposed features. As Labov has pointed out quite convincingly, some of the problems attending the use of such techniques are that the resulting data may be artefacts of the linguist's theoretical position, that the linguist's intuitions commonly reflect the prestige norms of the cOIlUllunity, and that differences regarding the data may be difficult to resolve. So here too, what one needs is data from actual language use.