Object agreement and topicality hierarchies in Kiyaka

This investigation of object agreement in .Kiyaka raises a theoretical question common to Bantu but which has been answered only partially in previous studies: what accounts for the control and sequential order of object agreement markers (OAMs) in languages where the verb complex can take more than one such marker? Individual parameters of topicality hierarchies (TH) have been shown to be responsible for the behaviour of the OAMs, especially in constructions where single hierarchies are active. For TH to account for conflicting predictions by different individual hierarchies, I propose a Trans-Hierarchical Constraint (THCC) that can be claimed to be cross-Bantu at least. This constraint predicts that OAMs can be determined by a single hierarchy (the predominant in the language), or by two hierarchies, one of which must be the semantic one. It follows from the nature of the THCC that there are two types of languages in terms of the strategies they use to resolve conflicting cases: (1) those where the semantic hierarchy is always active and (2) those where the semantic hierarchy is sometimes inactive. The study also predicts that ambiguity resulting from multiple agreement occurs only when the semantic hierarchy is inactive.