2005a, ‘Exhaustive Imperatives

Before talking about conditionalized imperatives, I want to ensure a common un-derstanding of ‘imperative’, namely as a clause type in the sense of Sadock and Zwicky (1985). That means that for a given language L, imperative will be taken to denote one of L ’s form types at sentence level, namely the one tokens of which are prototypically used for ORDERs or COMMANDs. As such they are paralleled by declaratives (used for ASSERTIONs), interrogatives (for information-seeking QUESTIONs) and for some languages also exclamatives and perhaps more. A natural choice for explaining this clause type system is to assume that it is encoded semantically, in the sense that the semantic object assigned to a particular sentence form determines its prototypical usage. The actual speech act performed by uttering a token of that type is then determined by the interaction of the semantic object expressed with the context of utterance. Imperatives are well-known to pose problems for a semantic treatment because they lack a straightforward link to the anchors of both static and dynamic semantics, namely the concepts of truth/falsity and of information growth. In that, they differ from interrogatives that have bee