Reconciling older and newer approaches to grammaticalization

The grammaticalization literature has seen three different strands of grammaticalization research: research inspired (i) by Lehmann’s unidirectional reduction view, in which a grammaticalizing item loses integrity and becomes increasingly dependent; (ii) by Himmelmann’s context-expansion view; and (iii) research that is integrated into a construction grammar approach to language. With each (new) domain of inquiry seems to have come a distinct set of criteria. This may create the impression that grammaticalization simply adjusts its criteria - in an ad-hoc fashion - to the domain at hand. Informed by the discussion in Traugott and Trousdale (2013), I argue in this paper that what look like distinct sets of grammaticalization criteria may not be as distinct as first meets the eye.