Questions of standardization

Ian Dale

Sat. 9:00-10:40 A

The sociolinguistic variable, a linguistic feature which can be shown to vary with certain socially or demographically defined characteristics of speakers, is a central sociolinguistic concept. In cases where such a variable tends to be realized differently in a local vernacular from the prevailing standard speech, it can serve as an indicator of a speaker's implicit orientation towards the local or the wider community.

This paper goes beyond the individual sociolinguistic variable and discusses the problems of constructing, from a set of variables, a general measure of speech standardization (or vernacularization). It draws on a data set of some 44,694 variant tokens distributed over eighteen variables and sixty-three informants, and discusses, in particular, the ways in which these are variously manifested in speech tasks designed to reflect different levels of discourse formality, levels which are here termed 'free conversation,' 'questionnaire style,' 'reading passage,' 'word list,' and 'minimal pair.'

Some of the questions that concern us are:

- "What to do when variables collide?' (that is, when the presumed vernacular variant of one variable seems to pattern together with the presumed standard variant of another),

- "Is the standard spoken or written?" (that is, the potential paradox set up when the presumed vernacular variant is closer to a spelling pronunciation than is the presumed standard), and

- "Is an overall standardization measure even feasible?" (or should variables be examined individually/atomistically, and only combined, if at all, when they are consistent with one another?).

This latter appears to be the crux of the matter. Given enough material, individual variables, functioning as specialized indicators, can give, as it were, a close-up of specific socio-linguistic interactions. A broader measure, on the other hand, is less sensitive to variations in specific factors, can function with a smaller total data set, and may reveal underlying consistencies masked by the surface interplay of individual variables.