Against mechanisms

Towards a minimal theory of change

Abstract

A widespread view in the literature on language change is that there exist a finite number of mechanisms of change to which attested instances of change can be assigned. In this paper I argue against reifying ‘mechanisms’ as primitives of the theory of language change, on multiple grounds. The mechanisms in question, such as reanalysis and analogy, are commonly invoked in multiply ambiguous ways: cause, process, event, result, and more. This is related to the fact that the ontological status of mechanisms is extremely suspect: where do they reside, and/or what are they properties of? I defend the position that a theory of change should be entirely derivative of i) a theory of language in the individual (cognition, acquisition, and use) and ii) a theory of (human) populations, with at least the latter containing no principles or stipulations specific to language. From this it follows that mechanisms, insofar as they have a role in our diachronic narratives, are epiphenomenal. If so, debates around the status of notions such as reanalysis and grammaticalization may both turn out to hinge on less contentful issues than previously thought.