The purpose of this paper will be to critically evaluate Donnellan’s assertion that the referential use of definite descriptions is “merely one tool for doing a certain job – calling attention to a person or thing.” (Ostertag, p.176) In order to do this I will first describe the referential use of language after which I will explore a number of variables which possibly contribute to the unique reference of a definite description. I will then contrast this account with Kripke’s account of speaker reference. In conclusion, I will argue that Donnellan’s account of referential use is superior to Kripke’s account of speaker’s reference, since it does not conflict with many of the intuitions with the latter does.
In his paper “Reference and Definite Descriptions” Donnellan draws a distinction between the attributive and referential uses of definite descriptions. A case of the former would be a detective who, judging by the horrific nature of a particular murder scene, exclaims
(1) “The murderer of Smith is insane.”
In such a case, the detective is clearly assuming that some individual murdered Smith and that this individual is also insane. Just as clearly, however, the detective is not assuming that some particular individual who he has in mind fits this description. Pragmatically speaking, (1) is roughly equivalent with
(2) “The murderer of Smith, whoever he is, is insane.”
An example of the referential use of a definite description, in contrast, can be seen in the case of a reporter at a trial in which Jones is being tried for the murder of Smith. Due to the unruly conduct of Jones, the reporter exclaims
(3) “The murderer of Smith is insane.”
What is important to Donnellan is that while (3) is semantically as well as linguistically equivalent to (1), it differs pragmatically from both (1) and (2). While in the case of (1) the speaker is referring to Smith’s murderer “whoever he is” with no particular person in mind, the speaker in (3) clearly has a particular individual in mind who he is referring to. Thus, the reporter in the case of (3) is assuming not only that some person in general fits the description, but also that some person in particular (Jones) fits the description.
The primary difference between the two uses lies in the functions which the definite description serves in each case. In the attributive use, “The murderer of Smith” serves to denote a category which either must be uniquely satisfied or “the entire linguistic purpose of the speech act will be thwarted.” (p.182) Sentence (1) is either about the actual murderer of Smith or it is about nobody at all; if nobody murdered Smith, then (1) will not be felicitous in any way.
The same cannot be said of the referential use in (3). “The murderer of Smith” serves as a tool or as helpful material to pick out or draw attention to a particular individual that the speaker has in mind and which the sentence is supposed to be about. Thus, sentence (3) is about Jones, whether he actually murdered Smith or not. Even if the definite description in (3) did not fit Jones, it would still be about him, for it is Jones that the reporter is calling insane, not anybody else. In other words, in the case of referential use, “there is a right thing to be picked out by the audience and its being the right thing is not simply a function of its fitting the description.” (191)
What else could an object “being the right thing” be a function of? Let us consider the case of a woman who has been driven by the cruelty of her husband into the arms of another man. One evening, this woman attends a party and is accompanied by her lover who throughout the evening is very kind and attentive to her. So attentive is he that an onlooker remarks to another,
(4) “Her husband is very kind to her.”
This statement is an example of the referential use in that it is clearly about the lover and not the woman’s actual husband. Thus, not only is it assumed by the onlooker that the woman has some husband, but it is also assumed that a particular individual (her attentive lover) is that husband.
In virtue of what, however, is the woman’s lover “the right thing” which (4) is about? Let individually consider three possible variables: context, predicate and description. If we systematically vary these elements in (4), do any of them effect the possibility and/or target of referential use and if so, to what degree? It modifying sentence (4), it will be important that we do not alter it too drastically. It must still be possible for the speaker to take each modified sentence to refer, at minimum, to both the woman’s lover as well as her husband.
In order to vary the context, let us suppose that rather than making the remark during the party, the same person makes the remark to the same listener after having seen the woman alone at the store a week later. Thus, the speaker and listener are still privy to the same information about the woman as they were at the party.
(5) (At the store) “Her husband is kind to her.”
It is unclear whether (5) is still about the woman’s lover or not. Indeed, given the revised context, it is difficult to see (5) as an unambiguous case of referential use anymore. It does seem possible, however, that (5) could still be used to refer to the woman’s lover in a referential manner, especially if (5) is uttered within the context of a conversation dealing with the previous week’s party. Consequently, (4)’s being about the lover does seem to be a function of the context in which it is uttered, at least to some degree.
It is also possible that the reference of (4) could also be a function of its predicate. Could it be that (3) was about Jones in virtue of his being the only crazy man in the courtroom? Could (4) be about the woman’s lover simply because he is the only one being kind to her at the party? What happens when we slightly vary the predicate of (4):
(6) (At the party) “Her husband is a scoundrel.”
What is surprising about (6) is that even though it is literally true (for the woman’s actual husband is indeed a scoundrel), intuition still suggests that (6) is about the woman’s lover rather than the unseen, and presumably unknown husband. If the predicate of a sentence does function as helpful material for drawing attention to the subject of a sentence, it seems to do so to a relatively insignificant degree. Indeed, our ability to say anything that is not contextually obvious with the referential use of language depends upon the predicate not contributing too much to the reference. (In (3) the reporter was surely saying something more than “The insane guy is insane.”)
Having considered the context and predicate of a definite description, let us now vary the description itself to see how significant the role it plays in determining reference is. While the partygoer was able to refer to the woman’s lover with an incorrect description, it seems absurd to think that any description would function equally well to draw attention to that same individual. Consider the following:
(7) (At the party) “Her plumber is kind to her.”
or
(8) (At the party) “Her uncle is kind to her.”
All things being equal, neither of these sentences seems to draw attention to the woman’s lover. Thus, while the exact fit of a description does not seem to be necessary for picking a referent out, the perceived degree of fit seems to be of some significance.
Thus, the effectiveness of definite descriptions as tools for drawing attention to a person is determined primarily by the description which is employed as well as the context in which it is employed. What is of particular interest for the topic at hand is the public nature of both description and context. Both the speaker and the listener at the party know what it means to be a husband and are thus able to pick from their shared context the individual which is most likely to be the woman’s husband and thereby say something about him. This public nature of referential use, as we shall see, is what sets it apart from Kripke’s notion of speaker reference. Let us now turn to Kripke’s account.
Kripke draws a distinction between what he calls semantic reference and speaker’s reference. Semantic reference is that which a sentence literally refers to as given by one’s conventional language. Sentences (1) and (3) both take Smith’s actual murderer as their semantic referent, just as sentences (4), (5) and (6) all take the woman’s actual husband as their semantic referent. Speaker’s reference, in contrast, is that individual object or person who the speaker of a sentence believes to satisfy the category which is denoted by the definite description. Thus, while the presupposition that somebody in general fits the description is clearly built into the semantic reference of a description, the presupposition that some person in particular fits the description is isolated to the beliefs of the speaker.
There are a number of differences which exist between speaker’s reference and the referential use of language. The essential difference which I see between the two, the difference which the remainder of the paper will be dedicated to exploring, is that the former is primarily private, just as the latter is primarily public. Speaker’s reference is primarily a question of epistemology and correctness in personal belief while the referential use of language is an example of cooperative communication among two or more parties.
To this, Kripke would object, for he explicitly acknowledges that “it may be clear which thing [the speaker] has in mind as satisfying [semantic reference].” (241) But how, exactly, is speaker’s reference made clear? Donnellan can simply say that the description within a context is what draws our attention to the object or person in question. The Russellian account which Kripke defends, however, seems to be more limited in this respect. Perhaps a description within its particular context can serve to indicate to the listener the beliefs which the speaker holds, but Kripke cannot allow that the speaker actually says anything at all about his intended referent. This seems overly restrictive at best. It seems counterintuitive to claim that while a speaker and a listener both take themselves to be describing the woman’s lover, the truth of the matter is that they are really are not saying anything about him at all.
This difference between referential use and speaker’s reference can be clearly illustrated by the case of what I will call informed misdescriptions. Donnellan suggests that in the referential use of language one can refer to a person who the speaker believes or even knows does not fit the stated description. Such could be the case for any number of reasons. The speaker may know that the listener incorrectly identifies the person by the misdescription, and therefore uses the misdescription out of convenience. The speaker might refer to the person by way of misdescription out of a sense of sarcasm or as a part of a joke between him and his listener. The speaker might simply be in the habit of referring to the person using the misdescription. In all of these cases a speaker seems quite capable of referring to a particular individual by way of knowingly misdescribing him.
Kripke’s account of speaker’s reference seems able to deal with such informed misdescriptions in one of two possible ways. The first approach is to view cases of misdescriptions which have become widespread habits within a community as being proper names rather than definite descriptions. This, however, seems to be exactly what Donnellan is arguing for in his paper. He asserts both that the gap between proper names and definite descriptions is not as wide as Russell had thought and that the referential use of language is very similar to the Russellian account of proper names.
The second approach open to Kripke is to view all such cases of informed misdescriptions which cannot be viewed as proper names not as different uses of language so much as misuses of language altogether. This response, however, seems too extreme. There does not seem to be any positive reason to actively assume that any non-denoting use of definite descriptions must be a misuse of definite descriptions altogether. At minimum, the burden of proof in upon the Russellian to show why such must be the case.
Another objection worth considering is that Donnellan’s own account of referential use seems itself to be too private in nature:
“I do not fail to refer merely because my audience does not correctly pick out what I am referring to. I can be referring to a particular man when I use the description … even thought the people to whom I speak fail to pick out the right person or any person at all.” (184)
Donnellan seems to suggest that individuals can refer to objects without any concern for constraints from public rules of language use. In other words, his account sounds suspiciously similar to Wittgenstein’s beetle-in-a-box thought experiment.
Such, however, is too strong of a reading. The beetle-in-a-box thought experiment differs from the referential use of language in two important aspects. First, Wittgenstein’s thought experiment was aimed at showing how the public availability of linguistic meaning and rules is a logical necessity. Far from denying the public nature of meaning and rules of use, the referential use of language depends upon such. Second, the individual boxes in which Wittgenstein’s beetles are kept are intended to prevent a public context between speaker and listener. Again, the referential use of language presupposes the exact opposite. Indeed, it could be argued that referential use is only possible between speaker and listener to the extent that a relevant context is shared between them. Donnellan is not asserting that a speaker can refer to whatever they what, by way of any description whatsoever. Rather he is simply suggesting that the listener’s being able to pick out the speaker’s intended referent is not an absolutely necessary condition for the speaker to make any reference at all.
