Co-extensive theories and unembedded definite descriptions

Russell argued, famously, that definite descriptions are not logical constituents of the sentences in which they appear. In neither of the following should we suppose that the definite description picks anything out: The King of France is bald The Prince of Wales is bald Since France is a republic, nothing could be picked out by the first; and if the semantic structures of each are the same, it cannot be the function of the second to pick anything out either. On the alternative semantics developed in his 1905 article 'On Denoting', definite descriptions do not have meaning in isolation; they have meaning only in the context of a whole sentence. Andrew Botterell and Robert Stainton have pointed out that this conclusion appears to be at odds with the phenomenon of unembedded definite descriptions, in which definite descriptions are uttered, meaningfully, without accompanying predicates. For example, it is possible to utter 'The last temptation' on its own and in doing so express a proposition. Since definite descriptions can be used in this way, how can it be right to claim, with Russell, that they lack meaning in isolation? The present paper seeks to show how a Russellian semantics for definite descriptions is entirely compatible with the phenomenon of unembedded definite descriptions. In particular, Botterell and Stainton are wrong to think that generalized quantifier semantics is better able to cope with the phenomenon than a more authentically Russellian syncategorematic semantics.