Competent speakers are normally authoritative as to the meaning of their words. Increasingly students of religion seem to be discounting this "first person authority", claiming that religious persons are not - as they seem to be - talking about what Hume called "invisible intelligent powers", gods, goddesses, ancestors, angels, and the like. This interpretative stance has understandably drawn criticism. I argue that the force of first person authority is diminished in religious discourse, and that this explains why it is so widely discounted. My argument turns on a broadly externalist account of linguistic meaning. When we view linguistic meaning as somehow "in the head" of the speaker, the interpreter will be set the incoherent task of trying to extract it with a minimum of distortion. But by viewing the meaning of words as emerging from the circumstances, mental and physical, and of their use, we may uncover the basis of first person authority and come to understand why its place in religious discourse has proved elusive.
This paper focuses on two contrasting approaches to the theory of linguistic meaning and asks how they color a range of issues of interest to scholars of religion. The so-called truth-conditional approach makes truth basic. It trades on the thought that we sometimes or perhaps often know what someone has said when we know what it would be for what she has said to be true. The other approach pegs meaning to how expressions and sentences are used in communicative situations. Dummett and Davidson are front and center. Davidson is of course in one sense a champion of truth-conditional semantics, but, over the issues I have in view, his case is instructively mixed. This discussion leads us toward an account of linguistic meaning which elevates over truth a family of concepts associated with use, including verification, justification, and pragmatic success.