Abstract

In the fourth chapter, I show how the semantics developed in the third chapter can be applied to the seemingly unrelated problem of discourse concerning fictional objects. Using the theory of nonexistent objects developed by Parsons as a point of departure, I show how this theory can be both simplified and enriched by the use of a truth-value gap semantics. ;In the third chapter, I develop a truth-value gap semantics for a fragment of English which includes both sortally correct and sortally incorrect sentences using the formal devices of partial functions, supervaluations, and sorted quantification which evaluates simple sortally incorrect sentences as being neither true nor false, retains the classical truths of logic, and avoids the problems had by Thomason's system concerning quantification. ;In the first chapter, I argue that the deviance of these sentences cannot be based on purely syntactic characteristics ) or merely by appeal to Grician maxims of conversation, and that any account of such sentences must be given in the semantic component of a grammar for a language. ;In the second chapter, a number of alternative semantic accounts of such sentences are investigated, including the bivalent system proposed by Bergmann , the four-valued system developed by Martin , and the truth-value gap semantics advocated by Thomason . After showing that all of these systems fail to adequately treat quantification, I argue that, while the differences in the approaches can be traced to disagreements so fundemental that no one of them can be chosen on the grounds of adequately accounting for the data, the type of approach taken by Thomason is preferable to the others on the theoretical grounds of simplicity and generality. ;The bulk of this work is an attempt to deal with sentences such as "The theory of relativity is shiny," which are called category mistakes by philosophers and sortally incorrect sentences by linguists