A corpus analysis of Swahili’s dual-complementizer system

Abstract Tanzanian Swahili has two complementizers, kuwa and kwamba, both used to introduce finite embedded clauses. We explore whether the complementizers are in free variation, as reported in all descriptive and pedagogical work. Our study primarily relies on corpus data, which we supplement with native speaker judgments. We find that the complementizers are not in free variation, but in fact are affected by a number of factors known to affect embedded clauses cross-linguistically, including predicate class, person features of the main-clause subject, and mood in the embedded clause. We conclude that the complementizers ultimately reflect subtle, pragmatic factors concerning how the truth of the embedded clause should be evaluated. Our study expands on previous work on languages with so-called “dual-complementizer” systems.