Complex Predicates and Agreement In Polysynthetic Languages*

The term “complex predicate” in syntactic theory is still semantically transparent; it can refer to any predicate that a particular researcher finds difficult—whether due to semantic complexity, syntactic complexity, morphological complexity, or whatever. For purposes of this work, I use the term to mean a single “inflectional domain” which contains two or more distinct morphemes, each of which selects at least one argument in its argument structure (also called a theta grid). An important question is whether the types of complex predicate that a language contains are at least in part predictable from more general typological features of the language. So far there has been limited literature on this topic, and what literature there is has been somewhat inconclusive. For example, Baker 1988: ch. 4 claims that properties of the morphological causative construction in a given language are predictable from the Case marking patterns found with simple ditransitive verbs in that language. However, this claim is controversial; other theorists hold that the various causative patterns are a function of the level at which merger between the causative morpheme and the complement verb takes place (Marantz 1984) or a function of the thematic properties of the causative morpheme itself (Alsina 1992). These factors are specific to the causative construction and are not derived from other aspects of the grammar of the language. If there are any clear relationships between general typological properties and complex predicates, they remain to be established.