ABSTRACT In recent years the field of historical syntax has become polarized into two more or less separate research communities. One, inspired by various versions of the Principles & Parameters programme, emphasises the necessary discontinuity in the transmission of language from generation to generation, and sees change as abrupt, reanalytic, and mediated through the process of language acquisition. The other, centring on a revival of the traditional notion of grammaticalization, points to apparent continuities in change, emphasises the gradualness of syntactic change and its semantic basis, and finds the motivation for change in discourse and the communicative needs of speakers. In its most extreme versions the latter approach even denies the relevance of linguistic structure, and hence of the formal models that have sought to characterise structure. There has been very little dialogue between these two camps. Perhaps the most common generative response has been to deny that grammaticalization is a real historical process. A notable exception is Ian Roberts, who has sought in recent years to develop a Minimalist model of the process of grammaticalization (Roberts 1993, 2001; Roberts & Roussou 1999, 2001).
In my paper, I will argue that, contrary to some suggestions (Janda 2001, Newmeyer 2001), grammaticalization is a genuine phenomenon, and that it has an inherent directionality which is in need of explanation. At the same time I will seek to show that there is no necessary link between the semantic 'bleaching' effect commonly noted in grammaticalization and the tendency towards phonological reduction which is also a common concomitant of this type of change. I will then briefly criticise the model proposed by Roberts & Roussou before developing an alternative LFG account which capitalises on the inherently lexicalist basis of LFG and on its clear separation of different types of structure (f-structure, c-structure, m-structure, etc). I will further suggest, following Bresnan (1998) and Bresnan & Deo (2001), that the most appropriate version of LFG for this purpose is one that incorporates the ideas of constraints and constraint re-reanking derived from Optimality Theory (cf also Vincent 1999, 2000, 2001). The data will be drawn from a cross-linguistic study of the evolution of markers of complementation and subordination (Bojars 2001). These arguments reinforce the conclusions reached by the contributors to Butt & King (in press) - Cynthia Allen, Julia Barron, Miriam Butt, Christoph Schwarze, Jane Simpson, Ida Toivonen - that LFG offers a natural basis for insightful accounts into morphosyntactic changes in a variety of languages and language families. All this work taken together - and earlier work such as Allen (1995) - suggests that historical data provide a new argument in favour of LFG as an approach to the characterization of natural languages, since:
REFERENCES
. . Back to Abstracts index |
