At least two philosophers have raised the question of the semantic significance of Donnellan's distinction between attributive and referential definite descriptions. 1 Kripke doubts that it has any, 2 whereas Devitt thinks it does, and tries to explain what it is. 3 I wish to explore the issue further in this paper, taking a closer look at what semantic significance Donnellan claimed for his distinction, and what significance it might actually have.
Devitt grants Donnellan's distinction "some significance at the level of intuitive semantics", but attacks the 'more substantial' claims he finds Donnellan making when we move to the notion of truth. 4 He characterizes these claims as follows.
According to Donnellan, a [referentially used] description 'refers' to the object the speaker had in mind even when it does not correctly describe that object. Further, the sentence containing the.., description is true or false according as the predicate in it is true or false of that object which the speaker had in mind. In all Donnellan's examples the.., description does not correctly describe anything; so the choice is between reference to what the speaker had in mind and reference failure. Donnellan plumps for the former. He does not discuss any example where the speaker has one object in mind but the.., description he uses correctly describes another. However, the implication of his discussion is clear: in such a case the.., description refers to the first object and the truth value of the sentence depends on its characteristics. (Devitt, p. 193) 5
Devitt goes on to criticize these claims, but the criticism is idle, since Donnellan does not make them. At the level of so-called intuitive semantics (that is in talking about reference) Donnellan speaks exclusively of what speakers refer to. More crucially -for Devitt at least -when Donnellan talks about truth in Section VIII of his paper, he very carefully avoids claiming that the sentences used are true. Indeed, he very nearly disavows this view explicitly. He fastens instead on the notion of 'saying something true of something', a notion which he thinks needs investigation and is isolated as a by-product of the referential/attributive distinction (Donnellan, p. 298).
His distance from the view Devitt attributes to him begins to become especially