10 - Conclusion

Summary

Quantitative modelling of patterns of variation and change in corpus data provides an important means to identify syntactically, functionally and distributionally independent stages within the Jespersen Cycle, and provides an empirical basis to establish how formal change and functional change interact in corpus data. The findings of the study fall into two broad areas: characterisation of the grammatical and functional changes affecting early English negation, and their interaction; and the role quantitative models of diachronic change can play in our understanding of the Jespersen Cycle, grammaticalisation, and morphosyntactic change.

Grammar Competition Models

Most recent work on morphosyntactic change incorporates both qualitative and quantitative data. However, these quantitative analyses do not always take the form of thoroughly worked out statistical models of change over time, and as such do not allow us to realise fully the role of particular constraints or factors within a change. The accounts of functional change described in Chapter 6 provide a good demonstration of these issues.

Here, I develop the proposals made in Kroch (1989) to bring together a formal syntactic approach and a variationist approach. The grammar competition model becomes a diagnostic tool in exactly the way Kroch (1989, 235) concludes might be possible:

Further work on historical change promises to extend the evidence linking patterns of change to grammars in competition, allowing us to understand changes better from the perspective of linguistic theory and eventually, perhaps, to refine grammatical analyses on the basis of the predictions they make about the patterning of usage in change.

Kroch (1989, 235, fn.29) elaborates:

Once the principle that contexts change together when they are surface reflexes of a single grammatical competition becomes firmly established, itmay be possible, on occasion, to choose among grammars proposed on the basis of synchronic analysis by the predictions they make as to which contexts should change together.

Our syntactic framework must be sufficiently nuanced to allow theoretically plausible pathways of change via syntactic reanalysis, and to reproduce patterns of change observed at an empirical level. Those syntactic reanalyses must be learnable during language acquisition.