Indexicality and The Puzzle of the Answering Machine

On the other hand, tokens of (1), recorded on millions of answering machines and voicemail systems worldwide, apparently express something true. It appears, then, that our best theory makes predictions at odds with the (quite banal) facts, and so must be amended. But how? In this paper I’ll defend a solution that (i) collapses onto Kaplan’s classic semantics for non-answering-machine cases, (ii) departs from (and is superior to) Kaplan’s view for answering-machine instances of (1), and (iii) extends smoothly to the variations of the answering machine cases that have arisen in the literature. Specifically, I’ll defend a semantics according to which indexicals are interpreted relative to the contexts in which they are tokened. I’ll compare this view — what I’ll call the context of tokening view — to competitors, and urge that it succeeds where they fail. Moreover, I’ll suggest that getting clear about the puzzle and the resources required for its solution holds important methodological lessons.