OF

Sentences like the following entail puzzles for standard systematic theories about language: (1) Bertrand Russell smoked a pipe. (2) Sherlock Holmes smoked a pipe. Prima facie, these sentences have the same semantic structure and contain expressions of the same semantic type; the only difference between them is that they contain different proper names. Intuitively, (1) and (2) are true, but they are made true and false, respectively, in different ways. Presumably (1) is true because the individual, Bertrand Russell, has or had the property of being a pipe smoker. In contrast, (2) is true for a reason something like this: the sentence ‘Holmes smokes a pipe ’ or an equivalent thereof, or a sentence entailing this sentence, was inscribed in the Holmes novels by Arthur Conan Doyle (2002). I show that the existence of fictional names, and the truths uttered using them, are not adequately explained by any extant account of fictional discourse. A proper explanation involves giving a semantics for names that can account for both referential and fictional uses of proper names. To this end, I argue that names should not be understood as expressions that immediately refer to objects. Rather, names should be understood as expressions that encode information about a speaker’s act of introducing novel uses for them. Names are not linked to objects, but to what I call “contexts of introduction”. I explain how this allows room for an explanation of fictional names, and how it also accommodates Kripkean uses of proper names. One consequence of my account is that one must think differently about analyticity, since truths uttered containing fictional names turn out to be language-dependent truths