William Croft

What drives language change?

Abstract What drives language change? This is actually an ambiguous question, since language change, like other evolutionary phenomena, is a two-step process: the generation of variation and its propagation through a population. In this presentation, I describe some recent proposals for mechanisms for both steps.

Traditionally, mechanisms for the generation of variation in language have been assumed to be intentional: that is, speakers change their language for various reasons, including expressivity, economy, and to be understood. However, in the realm of sound change, nonintentional mechanisms have also been proposed. Ohala (2003) argues for listener-driven processes by which the mapping from the complex phonetic signal to the phonological structure that produced it is reanalyzed. There is also evidence for a speaker-driven process. Many phoneticians, including Ohala and Pierrehumbert (2001), argue linguistic production at the phonological level is highly variable, and that variation can be the source of sound change.

I argue that the same nonintentional mechanisms can be observed at the grammatical level. Utterances are the instantiations of multiple interacting grammatical constructions, and their meaning in context is an integrated whole. Listeners may reanalyze the mapping from the meaning in context to the constructions, in one of several types of form-function reanalysis (Croft 2000). On the speaker's side, an examination of parallel productions reveals that verbalization of an experience at the grammatical level is also highly variable, and that variation can be the source of grammatical change (Croft, to appear).

Traditionally, the mechanisms for propagation have been the social valuation of particular linguistic variants, which leads to their differential replication (this is the classical sociohistorical linguistic model). Recently, Trudgill (2004) has argued that in the special case of the emergence of a new dialect via dialect mixture in an isolated community (specifically, New Zealand English), does not require the invocation of any social valuation of the variants: variants are propagated solely through exposure to their use in the speaker's social network, and the majority variant wins out. However, in a mathematical model of the evolutionary framework for language change in Croft (2000; see Baxter et al. 2006), we demonstrate that New Zealand English could not have emerged in Trudgill's model (Baxter et al., to be submitted). It appears that some sort of fitness such as social valuation must be associated with linguistic variants in order to account for observed patterns of the propagation of language change.

Baxter, Gareth J., Richard A. Blythe, William Croft and Alan J. McKane. 2006. Utterance selection model of linguistic change. Physical Review E 73.046118.

Baxter, Gareth J., Richard A. Blythe, William Croft and Alan J. McKane. To be submitted. Modeling language change: An evaluation of Trudgill’s theory of the emergence of New Zealand English. (Also to be presented at the Annual Meeting of the Linguistic Society of America, Chicago, Illinois.)

Croft, William. 2000. Explaining language change: an evolutionary approach. Harlow, Essex: Longman.

Croft, William. To appear. The origins of grammaticalization in the verbalization of experience. Linguistics.

Ohala, John J. 2003. Phonetics and historical phonology. Handbook of Historical Linguistics, ed. Brian Joseph & Richard Janda. 669-86. Oxford: Blackwell.

Pierrehumbert, Janet B. 2003. Probabilistic phonology: discrimination and robustness. Probabilistic linguistics, ed. Rens Bod, Jennifer Hay and Stefanie Jannedy, 177-228. Cambridge, Mass.: MIT Press.

Trudgill, Peter. 2004. New-dialect formation: the inevitability of colonial Englishes. Edinburgh: Edinburgh University Press.