Distributional differences

Distributional differences fall into two subclasses. First, there are differences in lexical incidence: certain individual lexical items will simply have one vowel phoneme in some accents, and another in others. For example, British English speakers are quick to comment on American English /aυ/ in route, or /ε/ in lever; Americans find British English /ru:t/ and /li:və(ɹ)/ equally odd. Some Northern English English speakers have /u:/ rather than /υ/ in look and otherwords; and it is fairly well-known in Britain that words containing /ɑ:/ vary in English English, with grass, dance, bath, for instance, having /a/ for many northern speakers, but /ɑ:/ in the south, though both varieties have /ɑ:/ in palm. Similarly, in SSE, weasel has /w/, and whelk; but in Borders Scots, where these phonemes also contrast, and where indeed most of the same minimal pairs (like Wales and whales, witch and which) work equally well, the lexical distribution in these two words is reversed, within weasel and /w/ in whelk.

On the other hand, a difference in the distribution of two phonemes may depend on the phonological context rather than having to be learned as an idiosyncracy of individual lexical items. For instance, in GA there is a very productive restriction on the consonant /j/ when it occurs before /u:/. Whereas in most British English [j] surfaces in muse, use, fuse, view, duke, tube, new, assume, in GA it appears only in the first four examples, and not in the cases where the /u:/ vowel is preceded by an alveolar consonant. There is also, as we have seen, a very clear division between rhotic accents of English, where /r/ can occur in all possible positions in the word (so [ɹ], or the appropriate realization for the accent in question, will surface in red, bread, very, beer, beard, beer is), and nonrhotic ones, where /r/ is permissible only between vowels (and will be pronounced in red, bread, very, beer is, but not the other cases).

Again, vowels follow the same patterns. For instance, in many varieties of English, schwa is only available in unstressed positions, in about, father, letter; in NZE, however, its range is wider, since it appears also in stressed syllables, in the KIT lexical set. Similarly, in some varieties words like happy have a tense /i/ vowel in the second, unstressed syllable; this is true for Tyneside English, SSE, GA and NZE. In SSBE, however, only lax vowels are permitted in unstressed syllables, so that /I/ appears in happy instead. Not all these distributional restrictions have to do with stress; some are the result of other developments in the consonant or vowel systems. For instance, the presence of the centring diphthongs before historical /r/ in SSBE (and other non-rhotic accents) means that non-low monophthongs cannot appear in this context. On the other hand, in rhotic accents like SSE and GA, there are no centring diphthongs, and the non-low monophthongs consequently have a broader range, with the same vowel appearing in FLEECE and NEAR, FACE and SQUARE, GOOSE and CURE.

In defining how accents differ, then, we must consider all three types of variation: systemic, realizational, and distributional. Although some of these (notably the systemic type) may seem more important to a phonologist, since they involve differences in the phoneme system, we must remember that one of the phonologist’s tasks is to determine what speakers of a language know, and how their knowledge is structured. It follows that we must be able to deal with the lower-level realizational and distributional differences too, since these are often precisely the points native speakers notice in assessing differences between their own accent and another variety of English. In any case, all of these types of variation will work together in distinguishing the phonological systems of different accents, and as we have seen, variation at one level very frequently has further implications for other areas of the phonology.