Definite descriptions (e.g. 'The king of France in 1997', 'The teacher of Aristotle') do not stand for particulars. Or so I will assume. The semantic alternative has seemed to be that descriptions only have meaning within sentences: i.e., that their semantic contribution is given syncategorimatically. This doesn't seem right, however, because descriptions can be used and understood outside the context of any sentence. Nor is this use simply a matter of "ellipsis." Since descriptions do not denote particulars, but seem to have a meaning in isolation, I propose that they be assigned generalized quantifiers as denotations — i.e. a kind of function, from sets/properties to propositions. I then defend the pragmatic plausibility of this proposal, using Relevance Theory. Specifically, I argue that, even taken as standing for generalized quantifiers, descriptions could still be used and understood in interpersonal communication.