In work that is now a decade old (Kroch 1989), I showed that syntactic changes evolve in accordance with the Constant Rate Effect, perhaps better called the Rate Identity Effect: at any given point, the relative frequency of the old and the new grammatical form tends to vary from context to context, but diachronically the rate of change is the same across all contexts. This effect shows that change - as manifested in historical texts - is regulated by competition in usage among alternative mutually-exclusive grammatical options. In other words, in the course of a change historical texts exhibit diglossia. The CRE/RIE was originally discovered through a statistical reanalysis of the data in A. Ellegard's rich study of the rise of periphrastic 'do' in Early Modern English, results recently confirmed by Anthony Warner (1998) in a more extensive reanalysis. Assuming the validity of the CRE/RIE, the single most important unsolved empirical problem in the study of the time course of change is the difference in the relative frequencies of the forms in the different grammatical contexts at any given point in time. In my original work on 'do', I suggested that these differences reflected processing or discourse-functional effects. In this paper, however, I will give evidence that, to a considerable extent, these differences reflect features of the changing grammar rather than being independent usage-based effects. The role of extra-grammatical factors is not eliminated in this new view, but it is more restricted than was thought in the past.
To download this paper, please return to Proceedings of the 1998 Postgraduate Conference