ABSTRACT
According to Perlmutter (1989), the idea that intransitive verbs do not form a unified class dates back at least to Sapir (1917): many Native American languages morphologically register the existence of two classes by distinct sets of cross-referencing pronominal affixes which traditionally are assumed to correlate with a semantic difference in volitionality. ' Various scholars working within the standard Transformational Grammar framework noted that the argument of certain intransitive verbs was best analyzed as a deep direct object (e.g. Postal, 1962, Hall-Partee, 1965, Chvany, 1975, Babby, 1975).~ This idea became formalized as the Unaccusative Hypothesis in Perlmutter (1978), marking the beginning of some of the most empirically rich and fruitful investigations across languages. Since then, the original RG formulation has been translated into other frameworks, including Government-Binding (Burzio, 1986) and Lexical-Functional Grammar (Bresnan and Zaenen, 1990).
