Deconstructing switch-reference

Abstract

This paper develops a new view on switch-reference, a phenomenon commonly taken to involve a morphological marker on a verb indicating whether the subject of this verb is coreferent with or disjoint from the subject of another verb. I propose a new structural source of switch-reference marking, which centers around coordination at different heights of the clausal structure, coupled with distinct morphological realizations of the syntactic coordination head. Conjunction of two VPs has two independent consequences: First, only a single external argument is projected; second, the coordinator head is realized by some marker A (the ‘same subject’ marker). Conjunction of two vPs, by contrast, leads to projection of two independent external arguments and a different realization of the coordination by a marker B (the ‘different subject’ marker). The hallmark properties of this analysis are that (i) subject identity or disjointness is only indirectly tied to the switch-reference markers, furnishing a straightforward account of cases where this correlation breaks down; (ii) switch-reference does not operate across fully developed clauses, which accounts for the widely observed featural defectiveness of switch-reference clauses; (iii) ‘same subject’ and ‘different subject’ constructions differ in their syntactic structure, thus accommodating cases where the choice of the switch-reference markers has an impact on event structure. The analysis is mainly developed on the basis of evidence from the Mexican language Seri, the Papuan language Amele, and the North-American language Kiowa.