What to say on what is said

Discussions in philosophy of language, semantics, and pragmatics, often make crucial use of the notion of what is said. It is held that in order to account for our intuitions on what is said, we need a distinguished semantic level. A tripartite distinction is made among what the sentence means independently from the context of utterance, what it means (or "says") within the context, and what the speaker means (or "conveys"). I will challenge the need for that intermediate level of meaning, and argue that the enterprise of drawing a neat distinction between meaning and what is said is pretty hopeless. My main point is that our intuitions on what is said cannot be detached from the ways in which we talk about it, and from the semantics of speech-reports and attitude-reports in general.