1 Decomposing Imperatives*

There is relatively little formal work on general issues concerning the syntax/semantics of clause types and concerning the nature of clause type systems, despite the fact that: • They are basic descriptive categories which pervade the way we talk and think about language; and • There appear to be significant universals in how clause type systems work. Sadock & Zwicky (1985) identify three key features of clause type systems. Clause types are universal, form a closed system, and are associated with a specific force. The concept of force most frequently invoked is that of illocutionary force, but this is not accurate. A better concept for purposes of explicating the concept of clause type is that of sentential force (Chierchia & McConnell-Ginet 1990) or sentence mood (e.g. Reis 1999). Our aim is to outline some general considerations and specific ideas about the status of clause types in syntactic and semantic theory. The general question I’d like to address is: How is the sentential force of a given clause type arrived at? Two main approaches are found in the literature: 1. Clauses are “typed ” (i.e. marked as belonging to a certain type) through the presence of some specialized element that encodes the force of the clause. This element is viewed as residing high in the clausal structure, typically in the CP domain (e.g. Baker 1970, Chen