Title

Fantasy, fiction, forecast

Date of Completion

January 2009

Keywords

Language, Linguistics|Philosophy

Degree

Ph.D.

Abstract

My dissertation addresses three questions in conditional semantics. What is the correct way to taxonomize conditionals? What does tense contribute to conditionals where it does not seem to impart temporal information? And how should we address the problem of similarity/cotenability? ^ I argue that the dominant taxonomy of conditionals is a vestige of an abandoned research program in Philosophy of Science. Though the distinction was relatively well motivated when it was introduced, the theoretical background against which the distinction labours has changed enough to show us that we need a replacement. There are good things about the distinction: the counterfactuals, when properly delineated, are in fact a subclass of an interesting natural class of conditional sentences. My first chapter outlines, criticizes, and develops an alternative taxonomy of conditionals first introduced by V.H. Dudman. ^ Under this new taxonomy, an interesting question jumps to the fore. In one interesting variety of conditionals—the class that includes the counterfactuals (I call them D-conditionals)–tense shows a markedly queer behaviour. The third chapter of my dissertation sketches a theory of the role of tense in D-conditional if-clauses. I argue that tense provides an argument for the restriction on an accessibility relation. That is, all accessible worlds are required to be identical with the evaluation world up to the time identified by the tense of the if-clause. This allows us to explain the obvious relation between these sentences as a standard relationship between present tense and past-past sentences: (1) If you go, you'll see Sue (said earlier); (2) If you had gone, you would have seen Sue (said later). ^ In the last chapter I attempt to bring Millikan's resources to bear on the problem of similarity/cotenability. The result is a theory of what is good about the conditionals that we like and what is bad about the conditionals that we do not. While this does not amount to truth conditions, it is perhaps the most we should expect. ^

COinS