Information focus in relational clause structure

Cross-linguistic studies of the recent decades have discovered a variety of split case marking systems, which mix nominative-accusative (S=A vs. P) and ergative-absolutive (A vs. S=P) groupings of core participant roles. The range of factors known to condition such splits in different languages includes the semantic nature of NP, its actual semantic role in the situation, tense/aspect/mood, and the grammatical status of the clause (Dixon 1994: 70-110). This paper discusses another type of case marking split, not mentioned in the classical overviews of split systems (Silverstein 1976; Comrie 1989: 129-135, 185199; Dixon 1994: 70-110; Harris & Campbell 1995: 240-257; Van Valin 1990; and references therein), namely, splits conditioned by focus structure (in the sense of Lambrecht (1994)). This phenomenon can be preliminary illustrated by two sets of examples from two unrelated languages, Dogon (1) and Tundra Yukaghir (2). Each of these languages has a nominal marker (ŋ in Dogon and –le(ŋ) in Tundra Yukaghir) that subsumes three functions: it marks information focus (examples (a)-(b) in both sets), P (examples (b)-(c)) and nominal predicate (examples (d)). In what follows, I use the complex functional label F|P to denote such morphological markers; here F stands for “focus”, or, to be more precise, for a language-specific grammatical counterpart of focus (see §2.1 for further details).