Abstract

This dissertation is concerned with the problem of giving a correct account of the semantics of vague predicates such as '...is tall', '...is bald' and '...is near...'. ;In Chapter 1 I present a definition of vagueness that aims to capture, in a useful form, all our fundamental intuitions about the vagueness of predicates such as those mentioned above; such a definition is lacking in the literature. I also present an abstract characterisation of the Sorites paradox: one that is independent of the particular forms in which the paradox may be presented, and brings to light the essence of the paradox. ;In Chapter 2 I examine existing theories of vagueness. In light of the definition of vagueness defended in Chapter 1, I argue that we need a semantics for vague language that countenances degrees of truth. I distinguish two sorts of degree theory. One sort---which includes the degree form of supervaluation semantics---sees vagueness as an essentially semantic matter. The other sort---which includes accounts based on fuzzy set theory---accounts for vagueness in language in terms of vagueness in the world. I argue that we need a degree theory of the latter, worldly sort. ;In Chapter 3 I examine the fuzzy theory in detail. I show that many of the objections to the fuzzy view that have been raised in the literature do not carry weight. In particular, I defend the coherence of the idea of degrees of truth. However, I isolate a problem for the fuzzy view---the problem of higher-order vagueness---that is serious enough to render the view unacceptable. ;In Chapter 4 I present a new theory of vagueness: one that is intended to share the advantages of the fuzzy view, while avoiding its disadvantages. In particular, this theory accommodates the phenomenon of higher-order vagueness. The theory involves degrees of truth, but they are not the same as the degrees of truth involved in the fuzzy theory. Although the theory involves a non-classical semantics for vague language, this semantics validates classical logic