Chapter 11

11.1 Proposal In analysing the Tongan data, we proposed the following hypothesis. First, based on the fact that ergativity is manifested also at the level of syntax, we argue that ERG is a structural case. Second, following Bobaljik (1993), we assume that the difference between ergative case marking and accusative case marking is fundamentally the choice of active Agr. If a language chooses Agrs as the active Agr, it will have an accusative case marking. If Agro is selected, then, the language will show an ergative case marking. In other words, the difference can be reduced to the question of which case is assigned to S. We then propose the following modification of the standard feature-checking theory (Chomsky 1991, 1993). Instead of assuming that case features are intrinsic to T and V, we propose that they are intrinsic to Agr(s). Specifically, Agrs