Human beings talk to themselves; sometimes out-loud, other times in inner speech. In this paper, I present a resolution to the following dilemma that arises from self-talk. If self-talk exists then either, (i) we know what we are going to say and self-talk serves no communicative purpose, and must serve some other purpose, or (ii) we don’t know what we are going to say, and self-talk does serve a communicative purpose, namely, it is an instance of us communicating with ourselves. Adopting (i) was the strategy taken by Bart Geurts, who claims that the primary purpose of self-talk is to entrain commitments, and is not (primarily) communicative. While accepting that self-talk can usefully play this role, I criticise the view that entraining commitments is self-talk’s fundamental role. I argue that adopting the view that we are self-blind, at least to a significant degree, means that we can accept that self-talk does play a communicative role.