Approaches to reanalysis

The study of grammaticalization has shown that new grammatical markers and constructions come into being gradually over time, through the operation of various reductive processes, as well as processes of inference in context. These gradual changes have the effect of creating new grammatical categories or new members of categories, and what could be viewed as new ‘rules’ or conventions of grammar. The creation of new grammatical structures is often called ‘reanalysis’. Reanalysis is usually viewed as an abrupt, but covert, change in the grammar that may not have any immediate overt effects (Langacker 1978, Lightfoot 1979, Kroch 1989a, Harris and Campbell 1995, Haspelmath 1998, Janda 2001, Roberts and Roussou 2003, Van Gelderen 2004). It is this view that requires a discontinuity in the transmission of language across generations (Andersen 1973); the child has access only to the surface forms of the language and not to the adult grammar and may therefore formulate a grammar that differs in structure from the adult grammar. As mentioned in Chapter 6, Janda 2001 argues that the discontinuity between generations makes some aspects of grammaticalization – such as unidirectionality – implausible and unexplainable.

The gradual nature of grammaticalization and the lack of evidence for abrupt reanalysis suggest that rather than postulate covert, inherently unobservable changes, we revise our conception of synchronic grammar so that it is more in line with the facts of grammatical change.