Abstract
Kayardild and the other Tangkic languages of Northern Australia are well known for their typologically unusual and complex case systems (Evans (1995, 2003a), Dench and Evans (1988)). In this paper we discuss the phenomenon of 'verbal case' (Evans 1995, 2003b), by which nominals are inflected with an alternative set of semantic case markers causing them to inflect like verbs, while still functioning syntactically as nominals. Verbal cases are similar in function to the regular cases, but differ in taking morphologically verbal endings agreeing with the main verb in tense/aspect/mood, rather than the modal case inflections usually found in this function with NPs inflected with regular cases. The phenomenon of verbal case poses a number of challenges for theories of morphology and the morphology-syntax interface. In particular it argues very strongly for a model that assumes a strict separation of morphology and syntax, such as LFG. Following recent work in LFG morphosyntax arguing for a distinction between morphological features (m-features) and syntactic features (s-features) (e.g. Sadler and Spencer 2001, Ackerman and Stump (in press), Sells (in press)), we propose that such a distinction is required at the categorical level also: verbal case converts a nominal stem into a morphological verb, while maintaining its syntactic category of noun. We show how this approach interacts with the constructive case model of Nordlinger (1998) to provide a unified account of Kayardild case at the morphosyntactic level, despite the substantial differences in morphological structure.