Linguistic Diversity and Theoretical Assumptions

1. Introduction The morphosyntactic diversity of human languages raises problems for many commonly held theoretical assumptions. Syntactic theories typically arise out of the study of a restricted set of issues in a specific group of languages, and then after the fundamental principles of the theory and its analytic constructs have been worked out on the basis of these data, an attempt is usually made to extend the approach to new issues and language-types. Chomskyan syntax grew out of the study of English, and this is still reflected in many of the theory's crucial assumptions , e.g. the necessity of a universal phrasal category headed by V and excluding the subject. 1 Relational Grammar [RelG], on the other hand, developed out of the analysis of a wider range of languages, but all were syntactically accusative, hence its postulation of what look like traditional Indo-European grammatical relations as universal primitives. Role and Reference Grammar [RRG] took a rather different starting point from generative approaches; it asked the question, 'what would syntactic theory look like if it were based on Lakhota, Tagalog and Dyir-bal, rather than on English?' Interesting problems arise when theories expand their analytic reach beyond the initial range of data which motivated them, and the purpose of this paper is to look at the ways theories deal with morphosyntactic phenomena which appear prima facie to be incompatible with one of more of their central assumptions. Three theoretical assumptions will be examined: (1) the universal validity of X-bar theory without discontinuous constituency; (2) the consistency of grammatical relations within a language; and (3) the linked units in complex sentences are VPs or clauses. Every theory does not make all of these assumptions, but most theoretical approaches make one or more of them. In the following sections, data problematic for each of these assumptions will be presented, and then the responses to them from several theories will be summarized. Conclusions will be presented in the final section.