Claim: Both the directly referential semantics and the more recent anaphoric accounts of 1st and 2nd person indexicals offer a picture of indexicality which is empirically and conceptually inadequate. They fail to capture this fact: Indexicals are essentially perspectival, as reflected in the fact that 1st and 2nd person indexicals are always de se.
Why hasn’t that been evident before?
Here is something important that compositional semantics has taught us: You cannot properly assess the meaning of an expression without considering its use and meaning in embedded contexts. But, as Kaplan drove home, the English 1st and 2nd person pronouns never seem to vary in interpretation in embedded contexts. However, recent work in linguistics has uncovered a wide variety of unrelated languages where the 1st and 2nd person pronouns can be shifted under attitudes. Careful consideration of their shifted meanings offers a new perspective on indexicality. Accordingly, I offer arguments for a de se account of indexicality.