A tale of two inverses

Abstract

AbstractThe functional‐typological literature distinguishes two kinds of inverse constructions: inverse voice, in which the patient becomes the subject, and inverse alignment, in which the patient is agreed with like a canonical subject. In this literature, Algonquian languages are held to be the prototypical example of a system in which the two kinds of inverses coexist: the inverse is a “deep” voice construction in clauses with two third‐person arguments and a “shallow” alignment pattern in clauses in which a third person acts on a speech‐act participant. This article argues that that conclusion is correct and attempts to reconcile it with formal models of voice and agreement. It is proposed that despite their distinct syntactic underpinnings, both inverse constructions result in a derivation in which Voice lacks phi features and Infl indexes only the patient. This shared outcome explains why the inverse appears to be a unified phenomenon from a morphological perspective even though its syntactic correlates differ in third‐person and speech‐act‐participant contexts.