Why does meaning matter? It is common to think that communication depends on it: were I to fail to know what your words mean, I would be unable to draw any conclusions either about what you are trying to convey to me when you talk, or about the world about which you talk, to the extent that I take your talking to express your knowledge. I will argue that this is mistaken: communication and the gathering of knowledge by testimony depend not on knowledge of the meanings of words, but rather on knowledge of what speakers mean by their words. I intend by this more than a bland emphasis on speaker's meaning as opposed to meaning in a "public" language: communication and knowledge by testimony go just fi ne even if the speaker's words mean nothing at all in any language. I'll develop these ideas, relate them to work on the semantic paradoxes, and discuss the role of the human language faculty in the facilitation of communication. To have a label, one could call the view I will present "non-factive cognitivism" about semantic theory: semantic competence is a psychologically real relation to a semantic theory, but the theory in question need not be true.Why believe that the words and sentences we use to communicate with one another have the meanings they appear to have? 1 The answers seem to be two: because they just seem to have these meanings and because if they weren't meaningful our ability to communicate would be compromised. In what follows I will dispatch both responses, focusing mostly on the second. Communication and even the transmission of knowledge by testimony do not require that the sentence used means anything. This being so, if there are widely shared impressions of meaning, we have no reason to accept 1 This article was written during the term of a Forschungsstipendium für erfahrene Wissenschaftler from the Alexander von Humboldt Foundation for research done at the