In Reference and Definite Descriptions， Keith S. Donnellan first of all puts up with his opinion that definite descriptions have two possible functions; one is to refer to what the speaker talks about and the other is different. However， he doesn’t propose the second function directly， but talks about two other theories before he discusses the distinction between the two functions. Donnellan argues that the referential theory of Russell and Strawson are both guilty of the failure to distinguish the two uses of functions and therefore result in the obscurity of the genuine referring function of definite descriptions. In Russell’s theory， if there is an entity x for which the proposition “x is identical with C” is true， then the entity x is the denotation of the denoting phrase C. But the author thinks that Russell has recognized only one function of denoting and neglected the other function of referring and therefore blames Russell for his narrow understanding about the referential use and denoting use of definite descriptions. As for Strawson， he holds the view that the use of definite descriptions seems to be the function of the sentence in which it occurs. But Donnellan is against this point of view because he insists that we cannot talk about the use in isolation from particular context. The first common assumption shared by Strawson and Russell that Donnellan "tries to refute is that the referential use of a definite description can be identified in independently particular occasion. The second common assumption is that both Strawson and Russell assume that the truth value of a proposition can be affected when the presupposition or implication of a definite description is 1. However， according to Donnellan’s theory， if there are two uses of definite descriptions， the falsity of the presupposition may affect the truth value differently in each case for which the nonreferential use may be suit whereas the referential use is unsuitable.

The two uses of definite descriptions by Donnellan’s theory are attributive use and referential use. He argues that attributive use is used by a speaker to describe whatever has so-and-so quality， while referential use is for an audience to pick out among many what the speaker is talking about. He also holds that the attributive use is essential because the speaker wants to assert something by the description while referential use is less important for it serves only as a tool to call attention to someone or something.

Donnellan tries to further illustrate the distinction between the two uses through three examples. The first example is the assertion “Smith’s murderer is insane” in which “Smith’s murderer” is the definite description. When we assume that Smith had no murderer， the two uses of the definite description then have different results. In the attributive use， if there’s no murderer， there’s no one that we attribute insanity to; in referential use， although Jones is not the murderer， we may also say that the comment “insane” we give upon him is reasonable. The conclusion of this example can be drawn into the assertion “The φ is Ψ.” In attributive use， if nothing is the φ then nothing has been said to be Ψ. In referential use， although nothing is the φ， we can say something to be Ψ. The second example is the distinction of the two uses in language use and the third example in commands or orders. We can draw similar conclusion that in the referential use of a definite description， we may succeed in picking out a person or thing although there is no such a person or thing really fits the description. However， in the attributive use， if nothing fits the description， no straightforward answer can be given or no command can be obeyed. It bears similarity with Strawson’s view that if the presupposition is 1， the assertion is neither true nor 1 which means it has no truth value. Therefore， Strawson’s view works for attributive use.

However， Donnellan opposed that the major difference between the two uses lies in the speaker’s beliefs， because he believes that a speaker can use a definite description in attributive use though he believes that a particular person fits the description; and he can also use a definite description in referential use without believing that someone fits the description. Generally speaking， attributive use and referential use of definite descriptions both carry a presupposition， but the reasons why they carry presupposition are different. When a definite description is used referentially， the speaker normally tries to give correct description so that his audience can recognize and pick out what he is referring to， therefore there is an implication in reference use of a definite description. When a definite description is used attributively， if there’s no presupposition or implication， then nothing can be attributed to the description， and the linguistic purpose of the speech then lost. However， the author thinks these two explanations both expose the default of Russell and Strawson’s theories because each of them only provides one account for the presence of presupposition. Donnellan thinks the reasons are so hard to clarify in some complicated situations that neither of the two theories seems to correctly deal with the referential use of definite description. For Russell’s theory， on the one hand， its deficiency is that it neglects the use of referring but only recognizes the use of denoting; on the other hand， it’s a pleasant result to make clear the distinction between denoting and referring. If one could not distinguish denoting and referring， it might happen that the speaker is referring to something without knowing it. For Strawson’s theory， Donnellan blames it for failing to distinguish referential use and attributive use. Donnellan concludes that it involves three propositions and each of them is either 1 or applies to one use of definite descriptions andcx "even mixes the truth value. Further explanations are as follows：

In proposition （1）， when the definite description is in referential use，something may be true about the reference; when the definite description is in attributive use， the assertion may be neither true nor 1. In proposition （2）， the statement is 1 because even though nothing fits the description when the definite description is used referentially， it is still possible to refer to something. Proposition （3） is actually the mixture of （1） and （2）. When the presupposition is 1， the attributive use fails to work because there is nothing to be attributed. But the reference may not fail to be present in referential use either when the audience can’t correctly pick out the thing or when nothing fits the description. However， in some extreme circumstances， it will fail to refer. Then the author supposes a case when he sees at some distance a man walking and asking “Is the man carrying a walk stick the professor of history？” and there are totally four possibilities of this supposition. In the first three cases， the speaker doesn’t fail to refer anything， but in possibility （d）， it is a genuine failure to refer. The failure of reference requires there be nothing to be said of， but it may result in the loss of truth value of an assertion. Therefore， Strawson’s theory mixes truth value and sometimes fails to distinguish referential use and attributive use.

On Donnellan’s view， the distinction between the two uses of definite descriptions is a function of the speaker’s intention. Therefore， due to the dual use of definite descriptions， the sentence can be said， at most， pragmatically ambiguous. To be specific， it is implausible to say either of the two uses as an ambiguity in the sentence. Donnellan believes that definite descriptions can be used either use in many sentences without ambiguous syntactically or semantically.

In the end， Donnellan gives a picture of Russell’s view on genuine referring expression compared with his own. Russell believes that proper names refer to something without ascribing any property to it while definite descriptions refer to something only satisfied the description. K.S.Donnellan， however， argues that when we use a definite description referentially to report a speaker， we may refer to something not just under a certain description. In other words， we may use any descriptions that will do the job rather than to be restricted to the description used Secondly， Russell thinks that the attributive use of definite descriptions introduce into reference generality while Donnellan believes this kind of generality is opposed to particularity shared by referential use of definite descriptions and Russell’s proper names.

References

羅素. 2003. 數(shù)理哲學(xué)導(dǎo)論[M]. 晏成書譯. 北京：商務(wù)印書館.

斯特勞森. 2004. 個(gè)體描述的形而上學(xué)[M]. 江怡譯. 北京：人民大學(xué)出版社.

Donnellan. Reference and Definite Description [J]. The Philosophical Review， 1966.

Russell. On Denoting [J]. Mind， New Series， 1950.

Strawson. On Referring [J]. Mind， New Series， 1950.