Abstract

If semantic paradoxes such as the Liar arise because ‘true’ and other metalinguistic expressions can change their reference with changes of linguistic context, is that due to indexicality (they have the same linguistic meaning as reference changes) or ambiguity (their linguistic meaning itself changes)? An argument from communication that appears to favour the indexicality interpretation is not compelling. This paper defends the ambiguity interpretation. It is left open whether its considerations generalize to other kinds of paradox.