INTRODUCTION

A traditional separation of pragmatics and semantics has it that speech acts are computed at the level of the entire sentence and that their illocutionary force cannot be embedded under various truth conditional operators. To quote from David Lewis (1972: 209):

Since speech acts other than assertions do not classically have truth values as extensions (Searle 1969; Vanderveken 1990), they cannot possibly combine with truth-functional operators like the propositional connectives in standard logic. This much is right. But when we look at the behavior of natural language conjunctions such as and, or, or if… then, speech acts such as directives and questions clearly embed under some of these operators. As Krifka (2001, 2002) has argued, considerable evidence has amassed over the years that most types of speech acts do embed within conjunctions and conditionals, and imperatives clearly embed under disjunctions. The problematic question for truth conditional semantics is then: what is the uniform account of the meaning of natural language sentential connectives that captures their behavior with interrogatives, imperatives, and indicatives?

Several decades of research have shown that this question is difficult if not impossible to answer within the confines of standard truth conditional semantics.