THE EVOLUTION OF THE GREENBERGIAN WORD ORDER CORRELATIONS

Recurring traits across languages have been argued to relate directly to constraints imposed by the brain. A particular case is Greenberg's word order universals (Greenberg 1966), which have been argued to reflect the demands of language acquisition (Baker 2001, Chomsky 2010) or of processing sentences (Hawkins 1983, Kirby and Hurford 1997). This paper argues that typological universals are indeed relevant to the study of the biological basis of language, but for a different reason, that they are created by the process of grammaticalisation. Recent historical evidence for the grammaticalisation explanation will be presented, concluding that explanations invoking processing or acquisition are misguided. Grammaticalisation involves the use of metaphor and the 'chunking' of constructions similar to the chunking of motor sequences (Bybee 2002), and hence the origin of language universals is argued to be in these properties of the mind rather than those proposed by innatist or functionalist linguists.