Bronwyn Bjorkman (Toronto): Ergative as Perfective Oblique

It is well known that many languages with ergative systems of case or agreement exhibit splits in their alignment. Viewpoint aspect is a common factor in such splits, with perfective aspect being associated with ergative alignment, and imperfective (or specifically progressive) aspect being associated with its absence (Moravcsik, 1978; Silverstein, 1976). Recent work on aspect-driven splits has focused on properties of the imperfective, arguing that it is associated with structures that disrupt otherwise-available mechanisms of ergative alignment (Laka, 2006; Coon, 2010, 2013).

This talk focuses instead on the syntax of the perfective, arguing that in some languages it is the perfective aspectual head itself that licenses ergative case. I argue specifically that ergative alignment in Hindi-Urdu arises from the intersection of two different ways of expressing perfective aspect, each attested independently in other languages. The first is the use of oblique case to mark perfect or perfective subjects, found in languages such as North Russian (Jung, 2011; Seržant, 2012), Estonian (Lindström and Tragel, 2010), and the Kartvelian dialect Mingrelian (Tuite, 1998). The second is a morphosyntactic sensitivity to transitivity, a hallmark of auxiliary selection in Germanic and Romance languages, whose parallels to ergativity in Hindi-Urdu were noted by Mahajan (1997). Ergativity of the type found in Hindi-Urdu fits naturally into this typological picture - but only if the licensing of ergative case is tied directly to perfective aspect, rather than disrupted by a structurally complex imperfective.

The result is a more unified view of the morphosyntax of perfective aspect, at the cost of a unified account of aspectually split ergativity: the proposal cannot extend to languages like Basque, for example, where both imperfective and perfective aspect show ergative alignment, with only progressive contexts being non-ergative (Laka, 2006). If the Basque type of split is better accounted for by properties of the imperfective, this again opens the question of how to account for crosslinguistic similarities in alignment splits: if imperfective and perfective aspects can independently influence ergativity, why do they nonetheless result in the same direction of splits?