How should we treat 'Sherlock Holmes'?

Abstract

Discussions about fictional names provide us an interesting perspective when we try to make clear important notions in the philosophy of language, such as sense and reference, because, so far, few theories have been successful in treating fictional names.
In this paper, I first point out that fictional names raise problems for two influential views in the philosophy of language, namely, Mill's view that proper names do not have meaning but only refer to their object, and Frege's view that proper names do have both reference and meaning(or sense). And then we shall examine three options to overcome the difficulties, namely, the predicate view, Walton-Evans' make-believe theory, and Thomasson's theory of dependent abstracta. Each view has its own merit and demerit when sentences/statements containing fictional names are divided into two types: internal and external. Internal statements express what is the case in a story(say, 'Sherlock Holmes is a famous detective)and external statements express what is the case from the real-world point of view(say, 'Sherlock Holmes is created by Conan Doyle'). I would like to propose that the make-believe theory would be the best option for internal statements and Thomasson's view for external statements. But what about statements that are both internal and external at the same time, such as 'Sherlock Holmes is a detective created by Conan Doyle'? This problem is discussed in§4 and one option is proposed there.