Published online by Cambridge University Press: 18 December 2008
A recent issue of this journal carried an analysis of the theoretical status of variable rules by the present authors (Kay & McDaniel 1979) together with a response by D. Sankoff and W. Labov (1979). The central issue with which Kay and McDaniel dealt was the following: which (if any) of the mathematical assumptions embodied in variable rules may reasonably be attributed to the linguistic abilities of speakers? The Sankoff and Labov response has gone some distance toward clarifying their position in this matter, though problematical issues remain. The basic question divides naturally into two related, though distinct, parts: (1) which (if any) of the mathematical assumptions comprised in the variable rule method are to be taken as substantive claims (versus conveniences for the data analyst)? (2) of the mathematical assumptions which are held to make substantive claims, what is the empirical locus of these claims – that is, what are the claims about (e.g., speakers' linguistic abilities)?