Syntactic conditions on accusative to ergative alignment change in Austronesian languages

This paper develops the proposal put forth by Aldridge (2015, 2016) for the emergence of ergative alignment in a first-order subgroup of the Austronesian family. I first provide new evidence for reconstructing Proto-Austronesian (PAn) as accusative rather than ergative. I then propose a significantly revised approach to Aldridge’s proposed reanalysis. On the basis of evidence from Tsou, I propose that the reanalysis took place in biclausal constructions embedded under motion or locative verbs. Since such biclausal constructions are contexts for restructuring, no accusative case is available for an object. This forced objects which needed structural licensing to value nominative case with T. I additionally show that subjects were assigned inherent non-nominative case in PAn when objects needed to enter into Agree with T, as when valuing nominative case. These conditions yielded a new ergative clause type in a daughter of PAn, which Aldridge (2015, 2016) calls “Proto-Ergative Austronesian”. No change took place in clauses lacking an object needing structural licensing. Consequently, subjects in intransitive clauses and transitive clauses with indefinite objects continued to surface with nominative case, yielding the type of ergative alignment prevalent in Formosan and Philippine languages today.