Morphological and phonological sources of split ergative agreement

A prevalent assumption is that alternations between distinct paradigms of verbal agreement directly reflect the grammatical functions and/or case of the arguments with which the verb agrees. This proves to be too simplistic a view of case and agreement, as shown by the various case/agreement mismatches attested cross-linguistically, and by evidence that morphological and phonological factors can sometimes influence the alternation between distinct agreement paradigms. This result suggests that the phenomena known as “ergativity” in case systems and in agreement systems, respectively, may sometimes be distinct, with a possible source of ergativity in agreement being post-syntactic morphological and phonological factors. I present evidence from several languages with unusual “split” ergative agreement systems, in which the factor conditioning the split arises only in the morphological structure. Although these systems would be exceedingly problematic to generate by means of case assignment in the syntax, I show that they follow quite straightforwardly from an Optimality Theoretic approach to morphology in which lexical insertion and the ordering of affixes and clitics are achieved in parallel through a single optimization.