Differential Argument Marking in Two-term Case Systems and its Implications for the General Theory of Case Marking

Abstract

INTRODUCTION In this paper I present a view of case marking that explicitly rejects a commonly assumed position that its primary function is to merely distinguish arguments from one another (cf. Comrie 1978, 1989; Dixon 1979, 1994; Kibrik 1979, 1997), while marking them according to their specific semantic or pragmatic functions is a secondary phenomenon. In order to show that such a view (which has already been challenged by many linguists, see section 1) is untenable, I will investigate data from argument-encoding variations in the languages which possess only two cases, and will compare them with similar phenomena from languages with richer case systems. As it will be seen, ‘nondiscriminative’ coding strategies found in two-term case systems, though typologically unusual, can be easily accounted for under the assumption that case marking of a particular argument is subject to ‘local’ ‘indexing’ rules and constraints dealing rather with this particular argument, than with the overall ‘global’ relational structure of the clause. ‘Discriminatory’ function, though retaining its importance, is, in this view, no more than just one of the constraints relevant for argument marking, whose ranking with regards to other such constraints in not always and not necessarily high. Also, I am going to argue that, contrary to some recent Optimality-theoretic proposals (see e.g. Woolford 2001), the case inventory found in a particular language cannot be always derived from a universal set of constraints (see Wunderlich & Lakamper 2001 for a similar proposal). As it will be shown, in order to account for case marking patterns in the languages with two-term case systems it is inevitable to regard the case inventory as a part of the input, and not as a feature of the candidates. In section 1 I will briefly outline the ‘discriminatory’ theory of case marking and summarize some arguments against it which have already been discussed in the literature. In section 2 I will discuss data from argument encoding variations (in