Abstract

This thesis examines procedures for ascribing ambiguity to particular sentences and words of a language. My discussion focuses on theories advanced by Keith Donnellan, Saul Kripke and David Kaplan regarding the alleged referential/attributive ambiguity of definite descriptions, and on arguments offered by Paul Ziff, David Wiggins and Jerrold Katz concerning the ambiguity of the word 'good'. I distinguish two kinds of semantic ambiguity, which I call "strong" and "weak", and develop a theoretical model of the former. Using this model, I devise and evaluate a number of theoretically related practical tests designed to detect the presence of ambiguity in concrete cases. I argue against direct appeals to our intuitions of meaning, to Occam's Razor, to grammatical analyses and to foreign languages in settling disputed ascriptions of ambiguity. I favor procedures which indirectly test our intuitions of meaning by isolating the linguistic effects of mixing and shifting senses. My tests rely on the assumption that the effects which accompany the use of truly ambiguous expressions are distinguishable from the mere pragmatic uncertainty which permeates general language use. Semantically ambiguous terms are viewed as awkward linguistic devices whose use, for example, will occasionally produce humor, as in the case of a pun, or deceptively bad reasoning. The account I develop rests heavily on the traditional theory of meaning , and to the extent that the standards I erect furnish better evidence for an ascription of semantic ambiguity, to that extent they ratify the classical analysis of meaning