Indexicals as token-reflexives

Reichenbachian approaches to indexicality contend that indexicals are 'token-reflexives': semantic rules associated with any given indexical-type determine the truth-conditional import of properly produced tokens of that type relative to certain relational properties of those tokens. Such a view may be understood as sharing the main tenets of Kaplan's well-known theory regarding content, or truth-conditions, but differs from it regarding the bearers of truth-conditional import and truth-conditions. Kaplan has criticized these approaches on different counts, the most damaging of which is that they make impossible a 'logic of demonstratives'. The reason for this is that the token-reflective approach entails that not two tokens of the same sentential type including indexicals are guaranteed to have the same truth-conditions. In this paper I rebut this and other criticisms of the Reichenbachian approach. Additionally, I point out that Kaplan's original theory of 'true demonstratives' is empirically inadequate, and claim that any modification capable of accurately handling the linguistic data would have similar problems to those attributed to the Reichenbachian approach. This is intended to show that the difficulties, no matter how real, are not caused by idiosyncrasies of the 'token-reflexive' view, but by deep facts about indexicality.