Ergativity in Austronesian Languages

Within the Austronesian family, many languages are classified as ergative (e.g., Samoan) or as having some ergative properties (e.g., Tagalog). For one particular language to exhibit some but not all of the characteristics of ergativity is problematic for an ergativity macroparameter. The same issue arises when looking at these languages from an accusative perspective: how do we account for the ergative properties (e.g., Schachter’s (1976) two subjects: Topic and Actor). One current proposal is that these languages have A-bar subjects (Richards 2000; Pearson 2001). By classifying a language as ergative or as A-bar, what do we predict? We compare the two approaches as they apply to Malagasy (and to a lesser degree, Tagalog) as a first step in answering this question. We begin with the ergative analysis, according to which the so-called Topic is the absolutive NP and the Actor is the ergative NP (see, e.g., Bittner and Hale 1996 and Maclachlan 1996). To make the ergative approach fit with the Malagasy and Tagalog facts, it is best supplemented with the exceptional accusative case, rather than a true antipassive construction. The advantages of this approach are many: we can account for the syntactic prominence of the Ergative-Actor (for control, binding and imperatives) and the famous restriction on extraction to Absolutive-Topics. There are, however, problems with the ergative approach, beginning with the nature of the antipassive (verbal morphology, the term status of the object). Setting aside this problem (resolved with the exceptional accusative case), there remains the lack of weak crossover. Unlike in ergative languages, a quantified NP in the Absolutive-Topic position can bind a pronoun embedded in the Ergative-Actor position.