Grammatically citing your sources via ‘say’-complementation

Abstract

Across the world’s languages, one of the most common clausal complementation strategies involves a form of the verb ‘say.’ Most research in this area assumes ‘say’ to be equivalent to the English complementizer ‘that.’ The present paper argues that an element in Lubukusu (Bantu, Kenya) that was formerly treated as an agreeing complementizer is actually a stative verb meaning ‘say.’ This ‘say’ predicate is the head of a clausal adjunct that merges at the VP level of the matrix clause, reasonably mimicking a complementizer. However, we demonstrate a number of reasons to think it is a verb. As a consequence of this analysis, an otherwise complex agreement system is reduced to run-of-the-mill subject-verb agreement both within the matrix clause and within a clausal adjunct. A central part of the argument is claiming that the verb ‘say’ introduces two arguments, a source argument and a communicated content argument; this argument structure explains a broad range of properties of ‘say’ cross-linguistically and correctly predicts a range of otherwise unexpected patterns in Lubukusu. If correct, this proposal advances our understanding of complementation structures and solves a standing agreement puzzle. We introduce a broad range of new Lubukusu patterns relevant to the argument structure of this verbal element, with implications for complementation, multiple-verb constructions, and agreement. In the course of the argument, we also address common objections to this kind of approach by showing that ‘say’ in English has a number of previously under-appreciated properties that parallel the Lubukusu facts. We conclude with some brief comments on the implications for a theory of selection.