This dissertation examines a few dimensions of morphosyntactic complexity in Georgian. Central are the language's split-ergative case system, whereby clausal arguments are assigned different case morphology across different tense–aspect–mood categories, and its verbal agreement paradigm, in which phi-agreement morphemes interact in complex but systematic ways. Three pairs of self-paced reading experiments probe the ramifications of Georgian split ergativity for online sentence processing, in ordinary transitive root clauses and also relative clauses. The Georgian comprehender is often faced with arguments whose case morphology does not unambiguously indicate their syntactic roles. Results show that comprehenders navigate these temporary ambiguities by harmonically aligning animacy and syntactic role — all else being equal, human arguments are parsed as transitive subjects, and inanimates as direct objects. Interpreted as garden path effects, the specific distribution of reading-time disruptions add nuance to this parsing heuristic, suggesting that comprehenders have fine-grained by not unlimited access to their abstract grammatical knowledge.
Moving to relative clauses, I find that priorities are subtly readjusted during the comprehension of filler--gap dependencies. With both Accusative- and Ergative-aligned relative clauses that can either precede or follow their head nouns, Georgian is uniquely well equipped to disentangle theories of filler–gap processing. And in relative clauses of all stripes, cues that eliminate the possibility of a subject-gap parse regularly lead to processing difficulty. This observation lends support to theories in which the structural distance between fillers and gaps — rather than the linear distance, or the informativity of ambient morphological cues — is the primary predictor of relative-clause processing difficulty.
The thesis is rounded off with a detailed formal investigation of argument–verb agreement in Georgian and its sibling languages. I identify a few generalizations that reveal key systematicities within superficially complex paradigms. These motivate an analysis deriving agreement patterns from the interaction of narrow-syntactic and post-syntactic mechanisms. Specifically, a syntactic principle independently motivated by non-agreement phenomena in other languages permits syntactic locality constraints to be loosened in certain circumstances; resulting derivational indeterminacy is obscured by morphological constraints that filter out all but the most expressive and economical combinations of agreement morphemes.