Reanalysis vs metaphor: what grammaticalisation CAN tell us about language evolution

Abstract

Language evolution has a notorious data problem: its object of study is simply too far remote in the pre-historic past for any direct observation to be possible. In such situations, Ockham’s razor recommends the assumption of the uniformity of process: that the mechanisms operating in the past are the ones still operating in the present. This would lead to the assumption that we should be able to learn something about the evolution of language from the study of language change, and in particular of semantic change leading to grammaticalisation (Heine & Kuteva, 2002a; Hurford, 2003). Grammaticalisation denotes the (unidirectional) process by which a discourse strategy, syntactic construction, or word, loses some of its independence of use and becomes more functional. It is usually accompanied by phonetic reduction and semantic bleaching and generalisation. There is disagreement over whether the study of grammaticalisation can give useful insights into language evolution. Newmeyer (2006), for instance, criticises the assumption that the unidirectionality of grammaticalisation provides sufficient evidence that early human language contained only nouns and verbs. We argue that grammaticalisation is indeed worthy of evolutionary linguists’ study, if one abstracts away from linguistic descriptions of individual phenomena to underlying psychological mechanisms. We thus support calls for a more cognition-oriented study of grammaticalisation (Heine, 1997; Kuteva, 2001; Tomasello, 2003):