Abstract

In case-marking languages with nominative-accusative alignment the subject of a sentence is usually marked by nominative case. In some of these languages, however, the subject of a number of verbs is either consistently or alternately marked by another, non-nominative case. Such non-canonical case marking has often been approached in the linguistic literature as a phenomenon at the interface between syntax and semantics. Yet the predictions of this kind of approach seem more probabilistic than regular. This paper offers a new perspective to analyse the phenomenon, which encompasses the role of information structure in case marking. Drawing on Silverstein’s theory of differential subject marking and Dalrymple & Nikolaeva’s approach to differential object marking, it is argued that non-canonically case-marked subjects can be better analysed as instances of either non-topical subjects or subjects lacking one or more semantic features typical of topicality. The approach outlined in the paper is tested on a number of constructions in Russian and Lithuanian. It is shown how, in both languages, the analysed instances of non-canonically case-marked subjects exhibit a complex interplay among grammatical, semantic and discourse-pragmatic factors.