Peter Culicover

Aspects of a Theory of Grammatical Change

Abstract The question as originally posed for this workshop is “How do opposing trends towards simplicity and communicability generate language change?” To address such a question productively it is important to be as clear as possible about what actually is changing, and what the conditions are that lead to change. So I will situate my remarks in a broader theoretical context, and argue that the notion of ‘construction’ plays a central role in grammatical description I briefly review the Simpler Syntax (SS) perspective on grammar and contrast it with mainstream generative grammar (MGG) (§1). I illustrate the difference with some examples of constructions and of what I have called ‘syntactic nuts.’ (§2) It is possible to model the contribution of grammatical complexity to language change with very idealized computational simulations (§3). The key factors are those that have been used to model the dynamics of social attitudes in networks of agents under various parameters settings (Latane and Nowak 1997, Culicover and Nowak 2002, Culicover et al. 2003, Culicover and Nowak 2003). But such modeling ignores the foundations of grammatical complexity. This is not a simple notion – it has to do not only with the formal complexity of linguistic expressions but with the transparency of their correspondence with meaning. With this as background, we come to the central point, which is that the reduction of grammatical complexity is a central goal of grammatical change (§4). I will suggest some plausible measures of complexity with SS as background. The two forces of complexity and transparency, with attendant mechanisms, lead to a view of change as cyclic, where some constructions are generalizing over time towards rules, while some rules are contracting over time towards constructions and even syntactic nuts. I illustrate this point with some observations about the history of English do-support and related phenomena (§5).