This dissertation investigates constituent order in Latin AcI clauses in order to understand how Latin authors avoided grammatical ambiguity in AcI clauses with two accusatives. Through a careful analysis of AcI data gathered from two Republican prose authors, Caesar and Cicero, this study aims to provide a more comprehensive description of the mechanisms by which the linear order of accusative constituents in AcI clauses is determined and through which the syntactic roles of the accusative constituents are disambiguated.

Ultimately, I show that the linear order of the accusative constituents is motivated by the information structure of the clause, with divergences from the statistically more common SOV order resulting from the object constituents raising to a pragmatic function slot, primarily through the two pragmatic movement processes Topicalization and Focus raising. The nuclear constituents’ syntactic functions, however, are determined by the semantic status of the constituents, with higher levels of agency and animacy positively correlating to subject role and lower levels of agency correlating to object role. In fact, we find a markedly strong preference for pairing high-agency, animate, individuated entities with low-agency or inanimate entities, and the prevalence of this asymmetric relationship largely counteracts any potential for grammatical ambiguity in such clauses. In other words, since semantic status, specifically agency and animacy, is such a reliable proxy for syntactic function, pragmatic movement processes like Topicalization can alter the SOV linear order without impairing the syntactic analysis.