Abstract
Speech reports of the form ‘A said that p’ are sometimes used by a speaker S as a reason in support of S’s own claim to know that p – in particular, when S’s claim to know is made on the basis of A’s testimony. In this paper I appeal to intuitions regarding the epistemology of testimony to argue that such ‘testimonial’ uses of speech reports ought to be ascribed their strict de dicto truth conditions. This result is then used as the basis for the claim that, no matter how they are used, all speech reports of this form ought to be ascribed their strict de dicto truth conditions. I conclude by offering a characterization of the content of the notions of saying and what is said, and by making some programmatic remarks regarding the role of these notions in semantic theory.