Abstract

This dissertation is concerned with two problems in semantics that have attracted a great deal of attention: negative polarity items and donkey sentences. "Any", as it occurs in is a negative polarity item, while the sentence in is a donkey sentence: Mary doesn't have any apples. Every farmer who owns a donkey beats it. ;The problem posed by negative polarity items is to account for their distribution, while that posed by donkey sentences is to explain the interpretations they get in a systematic way. The two topics may seem unrelated, but I argue that, in fact, they have a great deal in common. In both, the notion of a general statement plays a role. Roughly speaking, negative polarity items occur only in sentences that make general statements, and donkey sentences choose between two possible readings based on a preference for expressing general statements. Making these claims precise and arguing for their empirical adequacy is the main object of the dissertation. ;The definition of "general statement" I propose involves a condition I call the "negative witnessing" condition. This condition formalizes the idea that a general statement is one that can easily be falsified, in the sense that it is easy to find counterexamples to such a statement. The condition can easily be tailored to predict either the distribution of the so-called "strong" negative polarity items , or the distribution of the so-called "weak" negative polarity items . ;With regard to donkey sentences, I argue that they are, in fact, just a special case of a more general problem of the interpretation of definites under quantification. This generalization is made possible by treating the pronouns as so-called "E-type" pronouns . One of the main conclusions of this half of the dissertation is that donkey sentences do not motivate much of the special machinery that has been thought to be necessary for them. In particular, donkey sentences do not force us to abandon the traditional view that indefinites and definites are quantificational