This essay attempts to give substance to the claim that the liar''sparadox shows the truth predicate to be context sensitive. The aim ismodest: to provide an account of the truth predicate''s contextsensitivity (1) that derives from a more general understanding ofcontext sensitivity, (2) that does not depend upon a hierarchy ofpredicates and (3) that is able to address the liar''s paradox. Theconsequences of achieving this goal are not modest, though. Perhapssurprisingly, for reasons that will be discussed in the last section (...) ofthis essay, a natural account of the truth predicate''s contextsensitivity appears to lead naturally to a version of the correspondencetheory of truth according to which the truth predicate can be understoodas a relation holding between a sentence and a salient set of contexts.The plan of this essay is as follows. Section 1 contains a generalaccount of context sensitivity. The purpose of this section is toisolate certain features of context sensitivity and formal methods oftreating them, which we will then apply to the truth predicate. Section 2then outlines two minimal conditions to be satisfied by a truthpredicate. In Section 3, I present a version of the liar paradoxthat results from these conditions and the assumption that the truthpredicate is not context sensitive in the sense described in sectionone. Finally, in section four, I provide what appear to be naturalconsequences of a truth predicate''s context sensitivity. Section 4 isadmittedly speculative and points in the direction for future research. (shrink)

This paper examines the ontological commitments of the second-order language of arithmetic and argues that they do not extend beyond the first-order language. Then, building on an argument by George Boolos, we develop a Tarski-style definition of a truth predicate for the second-order language of arithmetic that does not involve the assignment of sets to second-order variables but rather uses the same class of assignments standardly used in a definition for the first-order language.

This paper develops a classical model for our ordinary use of the truth predicate (1) that is able to address the liar's paradox and (2) that satisfies a very strong version of deflationism. Since the model is a classical in the sense that it has no truth value gaps, the model is able to address Tarski's indictment of our ordinary use of the predicate as inconsistent. Moreover, since it is able to address the liar's paradox, it responds to arguments against (...) deflationism based upon that paradox alone. The model is based upon a notion of the complexity of propositions that a fixed set of speakers might express. A context-sensitive definition of the truth predicate is then provided based upon a class of possible worlds defined in terms of these speakers. Reasonable constraints on the memories and lifetimes of ordinary speakers are used to limit the set of propositions that they might express so that deflationist requirements are satisfied. (shrink)