ABSTRACT
Trom the fact that language is a social institution, it follows that linguistics is a social science, and the only variable to which we can turn to account for linguistic change is social change, of which linguistic variations are only consequences.’ These are the words of a great pioneer of French sociolinguistics, Antoine Meillet (1921: 16–17), cited half a century later by one of the leading methodologists of contemporary American sociolinguistics, William Labov (1972: 263). Unfortunately, it can hardly escape the attention even of a first-year student that this imposing pronouncement contains not just one glaring non sequitur but two; and it is arguable that these twin non sequiturs have continued to flaw socio-linguistic theorizing ever since Meillet first lent his authority to them. From the fact that language is a social institution, it no more follows that linguistics is a social science or that linguistic changes are only consequences of social changes, than it follows from the fact that only in human communities are crops cultivated that agriculture is a social science or that changes in cultivation are consequences of changes in society. Whatever the status of linguistics vis-à-vis the social sciences may be, it is not a status somehow guaranteed in advance by the premiss that language is a social institution (whatever that rather controversial phrase is taken to mean).
