Reference

6. Descriptions

Definite descriptions have the form, ‘the F‘, and indefinite descriptions the form, ‘a/an F‘. In his theory of descriptions, Russell (1905) claimed that ‘the F is G’ is equivalent to ‘there is something that is alone in being an F and it is G‘; and ‘an F is G’ is equivalent to ‘there is something that is an F and it is G‘. So the descriptions are to be understood in terms of the general term, ‘F’ and the existential quantifier, ‘there is something’.

Under the influence particularly of Donnellan (1966), many now think that a description is ‘ambiguous’, having not only this ‘attributive’ meaning captured by Russell but also a ‘referential’ meaning like that of a name or demonstrative.

It has been generally agreed that descriptions have a referential use as well as an attributive use. Used attributively, ‘the F’ conveys a thought about whatever is alone in being F; ‘an F’ conveys a thought about some F or other. Used referentially, each description conveys a thought about a particular F that the speaker has in mind, a thought about a certain F. Thus, consider:

used in the following two contexts. (a) We come upon Smith foully murdered. We have no idea who is responsible but the brutal manner of the killing leads us to utter (7). Its description is used attributively. (b) We observe Jones on trial for Smith’s murder. The oddity of his behaviour leads us to utter (7). Its description is used referentially. Next, consider:

used in the following two contexts. (c) Anne’s computer is discovered missing in the morning. We find signs that the burglar made a hasty escape dropping a red cap in the alley. This leads us to utter (8). Its description is used attributively. (d) After discovering that Anne’s computer is missing we remember noticing a man in a red cap behaving suspiciously earlier in the day. We utter (8) to report our suspicions to the boss. Its description is used referentially.

Despite agreement that there are these two uses, there is no agreement that descriptions are ambiguous. Appealing to ideas prominent in the work of H.P. Grice (1989), many have defended Russell. They argue that a speaker can use a description referentially, thus making the object in mind the ‘speaker referent’, even though that object is not the semantic referent. Whether a speaker has an object in mind or not, the truth-conditions of the sentence are as specified by Russell. The referential use is pragmatically, but not semantically, different from the attributive use (see Pragmatics §§3–4, 8). Thus, in context (b), although Jones is the speaker referent, the truth of (7) will depend on the sanity of whoever murdered Smith, whether Jones or not. And in context (d), although the man behaving suspiciously is the speaker referent, the truth of (8) will depend on whether some man or other in a red cap stole the computer.

Against this, many have found reasons for thinking that in contexts like (b) and (d) the speaker referent is also the semantic referent and hence that descriptions are semantically ambiguous after all. Some of these reasons – for example, those arising from failures to describe correctly the object in mind and from the behaviour of descriptions in opaque contexts – have not stood up well. Others seem more promising. First, we are not just able to use descriptions referentially, it seems that we regularly do so. This regularity suggests that there is a convention of so using descriptions. If there is, then it is hard to see why the convention is not semantic. Second, in their referential uses descriptions seem to have roles just like demonstratives: ‘the F’ and ‘an F’ function like ‘that F‘, and are similarly based on perception of a particular object. To try to treat these demonstratives like Russellian descriptions would be to give a description theory of them, and we have already noted problems for this (see §5). Third, definite (but not indefinite) descriptions seem to have the same range of anaphoric roles as a pronoun like ‘she’ (see §7). We might then expect them also to share the pronoun’s role as a demonstrative which, as has already been pointed out, seems not to be Russellian. Fourth, consider the utterance, ‘The book is on the table’. In the right circumstances, this will seem true and yet, according to the Russellian view, it must be false: since the world is full of books and tables the two definite descriptions fail to describe unique objects. The obvious modification to save the Russellian view is to treat these ‘incomplete’ descriptions as elliptical. But this modification has problems. A speaker may have many ways to complete the description and there may be no basis for saying that any one is the correct way. Alternatively, trying to complete the description may lead to the familiar problems of ignorance and error.

Another argument against the ambiguity thesis appeals to rigidity. If referential uses of descriptions were semantically significant then, it is claimed, they should be rigid like names and demonstratives. Yet they do not seem to be. Consider the use of ‘Smith’s murderer’ in context (b) above. ‘Smith’s murderer is insane’ does not seem to be true in a world where Smith is alive and well, even if Jones is insane. However, this argument has a problem: the referential use of the demonstrative ‘that murderer’ would equally seem to fail this rigidity test in these circumstances (hence suggesting a need to revise Kaplan’s claims about demonstratives; see §5). Yet a demonstrative surely has a semantically significant referential use. If so, then a description may have one too, despite not being rigid (see Descriptions).