Autor: Pavlína Josisová (Ústav filosofie a religionistiky, FF UK)

The following text is devoted to the discrepancy between the concept of definite descriptions of Donnellan and Kripke. It explains Donnellan’s distinction between the attributive and referential use of definite descriptions which has been motivated by the perceived insufficiency of traditional approaches. It follows up by explaining Kripke’s critique which is based on showing inconsistency in Donnellan’s approach, as he understands it. Finally it provides a brief evaluation of Kripke’s critique, questioning its relevance.

1. Attributive and referential use

By a “definite description” it is usually meant a description of the form “the x”, when x is a noun or a noun-phrase. Classical examples would be definite descriptions as “the man”, “that man over there”, “the husband”, “her husband”. They should be distinguished from indefinite descriptions of the form “a x”, where the objects denoted are ambiguous, as in the case of “a man”, “a husband”, etc. This text will focus on how definite descriptions refer, and how to analyze them.

In his text “Reference and Definite Descriptions”1 Donnellan introduces a distinction between the attributive and referential use of definite descriptions. His presentation is motivated by the problematic character of Strawson’s2 and Russell’s3 accounts of definite descriptions. Donnellan’s proposal aims to solve those difficulties. We can illustrate his main point by means of the following example from Linsky.4

Let’s imagine that at a party Ms Smith meets her lover, Mr Jones, unbeknownst to her husband, Mr Smith. Mr Jones and Ms Smith are talking together and he looks to be really kind to her. A group of people at the party starts to talk about them. These people don’t know either who Mr Jones or who Ms Smith are and they don’t know anything about their lives. Given the situation, they all take Mr Jones as Ms Smith’s husband. During that conversation, one person among the observers says, referring to Mr Jones: (1) “Her husband is kind to her.” Despite their beliefs, Mr Jones is not Ms Smith’s husband, he is her lover.

According to Russell’s theory of definite descriptions, without taking into account any further investigation, he argues that (1) is false. Since the speaker was not directly acquainted with Mr Jones and does not know him she refers to him by description “her husband”. So the question is: Is there someone that makes the expression “there is a man who is the husband of the woman over there and he is kind to her” true? According to Russell, the sentence fails to be true in two cases: there is no one to which the description refers, or that person is not kind to her. There is no one who is both Ms Smith’s husband and is being kind to her thus the proposition is false. According to Strawson and Linsky, (1) is neither true nor false. Since the condition of the existence of the corresponding referent is not met we had failed to refer and the proposition has no truth value.

Those analyses are not satisfying because they are not able to capture the fact that (1) actually picks out someone, so the description appears to be a referring expression. It seems that the speaker is not referring to Mr Smith when she says “her husband” – she is clearly referring to Mr Jones. Despite the fact that there is no husband, when the speaker said “her husband” she must have expected a husband to exist. However, the communication between the speaker and the audience does not fail. How should we cope with the fact that the term “her husband”, even though it appears to be empty, still seems to be able to pick out someone?

It is to explain this phenomenon that Donnellan introduces his distinction between attributive and referential use of definite description which he defines in the following way: „A speaker who uses a definite description attributively in an assertion states something about whoever or whatever is the so-and-so. A speaker who uses a definite description referentially in an assertion, on the other hand, uses the description to enable his audience to pick out whom or what he is talking about and states something about that person or thing.“5 In the case of attributive use the speaker already has in mind someone or something he is speaking about. The attribute of being so-and-so is thus important, because “the speaker wishes to assert something about whatever or whoever fits that description,”6 while in the case of referential use it is not important – “the definite description is merely one tool for… calling attention to a person or thing.”7 We are using it to help the audience to pick out a person or a thing that we are speaking about.

Donnellan says that while Russell accepts the attributive use he neglects the referential use of definite descriptions. When using it we do not expect anything that we could assign attributes to, but rather we try to create such conditions that the entity has to fulfill to become an object of referring. Donnellan does admit that Strawson acknowledges referential use, but he does not agree with the fact that its assignment would be based solely on the type of sentence in which it would occur. Donnellan says that referential and non-referential use of descriptions can occur in one and the same sentence and the difference will only be in its usage by the speaker in each particular occasion. What exactly is being said by that sentence is impossible to recognize without the knowledge of the way in which it is being used. When we know how it is being used we can determine which of the two uses is being performed.

Let’s imagine the situation described before, let’s imagine seeing Ms Smith in a lively conversation with Mr Jones who is, at least from the speaker’s point of view, kind to her. Based on that she assumes he is her husband. The sentence “Her husband is kind to her”, will thus be used in a referential way. Indeed, when uttering this sentence, she wants to refer to someone in particular: someone she believes fits the description she is using. The question “Who is kind to her?” could be answered via other possible descriptions which lead to that man – e.g. “the man over there”, “the man drinking Martini” – or even by using his name – e.g. – “Mr Jones”. In this case they can refer to the person they want to refer to, even though there is no husband. Because when using the referential use, her aim is to enable the audience to pick out the right person, Mr Jones, about whom she was talking about. If audience succeed in picking out the right person, than they can answer this question even if this person does not fit the given description. E.g. should Mr Jones only be her lover, not her husband.

Let’s imagine another situation similar to the previous one: Let’s imagine seeing a woman at a party looking happy and satisfied. On the basis of her appearance, the speaker assumes that her husband is kind to her. The sentence “Her husband is kind to her”, is then used attributively. She is not referring to any specific person, she is just saying that her husband, whoever he is, is kind to her. The question “Who is kind to her?” could be answered by saying “Someone who is her husband”. Thus the audience repeats the condition to be satisfied in order for an entity to be the object of their speech and in order for them to be able to assign it a spoken attribute.

Once we acknowledge that there are two possible uses of definite descriptions (attributive and referential) we can also identify two possible ways of affecting truth conditions of any given sentence. The assumptions of what we are predicating, what we are denoting or referring to, are different in each case. The statement “Her husband is kind to her” is thus ambiguous because it allows two different readings. Donnellan says that this ambiguity is not semantic or syntactic, but rather pragmatic. The meanings of words are not ambiguous but the ambiguity is rather in the way the description is used, which depends on the momentary intentions of the speaker.

Now we can compare the differences between particular truth values of the statement “Her husband is kind to her”. Let’s use “F” instead of “her husband” and “G” instead of “to be kind to her”, the whole sentence is then “F is G”.

Russell would say that if there is one and only one x such that it is (i) an F and (ii) a G at the same time, then the proposition is true. Otherwise, the proposition is false. Strawson would say that the sentence contains no existential assertion, but would try to use “her husband” as a referential phrase. The proposition is true when we are able to refer to x with F and at the same time x will be G. The proposition is false if x does not have the property of G. However, if we cannot refer to an x at all, the proposition has no truth value.

Let’s imagine that Ms Smith is a married woman whose husband is in the next door room. Then we have a problem. In this case both Strawson and Russell would evaluate this sentence as false should Mr Smith, currently in the next door room, be cruel to her, but they would evaluate it as true should he be kind to her. The problem is that the speaker was not at all speaking about the man who actually is Ms Smith’s husband, Mr Smith. She did not see him in the next door room and when saying “Her husband is kind to her”, was referring to someone else, namely, Ms Smith’s lover, Mr Jones.

This is when Donnellan presents attributive and referential use, as the truth values will be different in both the cases. In the case of attributive use the sentence is true if x exists and is F and G at the same time. If x does not exist, nothing true has been said, the statement is neither true nor false. In the case of referential use we say that x is G and the proposition is true if and only if it is G. In the case that x does not exist it is, according to Donnellan, possible that something true has been said about the person to which is being referred to. He does not openly say that the proposition is true. For Donnellan, contrary to Russell or Strawson, one refers to the person the speaker is thinking of – in our case this person is the lover, whether or not the husband exists. The person marked “x” does not have to be her husband even though we are referring to him that way – the person who actually is the husband of the woman, lets mark him “y”, is not the object of the utterance. In reply they can thus say “It is true that her husband is kind to her”, and they can again be referring to her lover. The proposition would be false if the lover were cruel to Ms Smith, and not if the husband were cruel to her.

Besides that, Donnellan refuses Russell’s opposition of definite descriptions and proper names. He puts names in the same function category as referentially used definite descriptions – both are used to allow the audience to pick out the right object of which the utterance is about, so they are mere tools to serve their function, without any categorical difference. This object always already exists, the speaker, contrary to the attributive use, has it already on mind.

2. Speaker’s reference and semantic reference

In his text “Speaker’s Reference and Semantic Reference”8 Kripke, per his own words, tries to make a methodological critique of Donnellan’s theory and disprove that Donnellan refutes Russell. We will focus on the substantial critique that shows up together with the methodological critique.

Let’s continue to use the example “Her husband is kind to her”. As we have shown Donnellan would not hesitate to put proper names and referentially used definite descriptions in the same category. The utterance “Mr Jones”, should, based on his point of view, function the same way as “her husband”. Kripke objects that they do not belong to the same category; while proper names are necessarily rigid, definite descriptions can be rigid but they are not necessarily rigid, even when used referentially.

Kripke at middle-age, in undated photo, circa 1980's (source: http://faculty.washington.edu/smcohen/453/KripkePhotos.html)

Saul Kripke at middle-age, in undated photo, circa 1980’s (source: http://faculty.washington.edu/smcohen/453/KripkePhotos.html)

Let’s assume that the people at the party are speaking about Mr Jones and have for some time been using the term “her husband” to describe him and believe that he is her husband. As soon as they find out about their mistake they will stop referring to him in that way. Some of them have however been using his proper name – “Mr Jones”. We can say that for some time the referential definite description “her husband” has been used to refer to Mr Jones and could have, at first sight, played the same role of the name “Mr Jones”. After finding about their mistake they will stop using the description “her husband” to refer to Mr Jones. That does not mean that they will stop using “Mr Jones” in the same way to refer to Mr Jones. Her husband in the next door room will get the right to be referred to him by the description “her husband” but he will not get the right to be referred to by “Mr. Jones”.

The following step is the critique to Donnellan’s conviction that his distinction between attributive and referential use is neither semantic nor syntactic but just pragmatic. Kripke objects that this position is untenable. If the ambiguity of a sentence were just pragmatic, the utterance would admit just one analysis, namely, the Russellian one: the proposition expressed by the utterance is true if there is one and only x such that it is F and G at the same time. The difference between attributive and referential use would then really have to be solely pragmatic, it would depend on what the speaker intended to say with his words, but not what the words themselves mean. Since Donnellan apart from Russell’s analysis also presents an alternative analysis, the analysis of referential use, he moves the difference to the semantic level.

If Donnellan wants to pursue this alternative analysis he should admit that the difference is indeed a semantic one. On the contrary, if he wants to deny that, he should give the same analysis of the statement under all the circumstances. In that case, the difference would be purely pragmatic and it would amount just to a matter of different use, tertium non datur, according to Kripke.

Now, we should present Grice’s9 distinction to which Kripke appeals – the distinction between what the speaker’s words meant and what she meant when she was saying these words in some situation. Speaker’s reference and semantic reference are special cases of Gricean notions. Kripke defines them in the following way: “If speaker has a designator in his idiolect, certain conventions of his idiolect (given various facts about the world) determine the referent in the idiolect: That I call the semantic referent of the designator… The speaker’s referent of a designator is that object which the speaker wishes to talk about, on a given occasion, and believes fulfills the conditions for being the semantic referent of the designator… The speaker’s referent is the thing the speaker referred to by the designator, though it may not be the referent of the designator, in his idiolect.”10 In our case the referent of “her husband” was Mr Jones by speaker’s reference while he was Mr Smith by semantic reference.

According to Kripke, it is inconsistent that Donnellan analyzes speaker’s referent and does not analyze the semantic referent, when presenting two different semantic analyses of (1). For his differing semantic analyses refer to semantic referent, not speaker’s referent. In other words, Donnellan ends up with two different propositions when analyzing one sentence – therefore, the ambiguity should be semantic. If Donnellan wants it to stay on the pragmatic level, he should drop the different analyses of (1). But it seems that the two different semantic analyses of (1) depend on the pragmatic distinction between speaker’s referent and semantic referent.

Kripke wants to complete his theory in the way that he would be able to cope with a sort of situations in which the semantic referent of our statement differs from the speaker’s referent and despite that, our conversation do not collapse. However, he still wants to hold the distinction between semantic and pragmatic level of the analyses. He presents two types of situations: a simple case and a complex case. The simple case should be parallel to attributive use and the complex case should be parallel to the referential use. In the simple case the goal of the speaker is to refer to the semantic referent thus his goal is the same as his general intention which is given conventionally. In the complex case his goal is different than his general intention, however, he still believes that he will be able to determine the same object as he would determine with general intention. In the simple case speaker’s referent and semantic referent are one and the same while in the complex case they may coincide, but not necessarily. It depends on whether or not the speaker’s belief is correct.

In the simple case, by saying “Mr Jones” we would simply refer to Mr. Jones who is also the object designated by the proper name Mr Jones. In the complex case we intend to refer to a certain person as “Mr Smith” because we believe that he is Mr Smith – we know the surname of his wife and assume that he has the same surname. However, the name “Mr Smith” can have – and in this case does have – a different semantic referent. If the man we referred to by “Mr Smith” was Mr Smith it would be the compliance of speaker’s referent and semantic referent – but if that person is Mr Jones, the semantic referent and the speaker’s referent differ.

Kripke again shows that the description used referentially in Donnellan’s way cannot be seen as of the same usage with the proper name. If we follow Kripke’s account, instead of Donnellan’s system of attributively used definite descriptions and referentially used definite descriptions counter to proper names, we get a different categorization: definite descriptions counter to proper names where both categories are related to pragmatic distinction between speaker’s reference and semantic reference. According to this different categorization, the semantic referent of the statement does not change across different circumstances and contexts of utterance; the difference would occur just at the pragmatic level depending on whether or not the speaker’s referent coincides with the semantic referent and also whether it is a simple or a complex case.

3. The relevance of Kripke’s critique

Kripke does not deny the existence of the attributive and the referential use, and the distinction between them. He just doubts its importance. As he himself declared, he would like to methodologically refute that Donnellan refuted Russell. However, in this attempt he refutes Donnellan’s account. He does not think that Donnellan’s distinction was important for our communication, as we can cope without it. On the contrary, we need the distinction between speaker’s reference and sematic reference, and the distinction between complex cases and simple cases. His argumentation is based primarily on the critique of Donnellan’s inconsistency: he shows that Donnellan implicitly assumes a semantic distinction between two different types of descriptions, which he thinks does not obtain. Kripke continues with showing the necessity to distinguish proper names and definite descriptions and then separate semantic and pragmatic analyses could be applied to both categories.

Thus it depends on whether or not Kripke’s strategy of critique based on the semantic analysis of truth values of propositions is appropriate for Donnellan’s distinction. For Donnellan refuses semantic or syntactic ambiguity for a certain reason – he says that the meanings of words stay the same and it is just the usage that differs. He does not refute that truth conditions of propositions are affected by certain usages, but that does not have to change the literal meaning of the sentence. In Kripke’s words, the semantic referent would stay the same. The difference would then only be at the pragmatic level and the evaluation of a sentence would still depend on the way it has been used by the speaker. Thus Kripke would provide a critique of a different concept than Donnellan had.

The fact that Kripke says that we do not need Donnellan’s distinction does not mean it cannot be useful to clarify some of the features of our communication. Apart from opening the discussion of the problematic concept of having something or someone in mind, it also points to the not so sharp division of the pragmatic level from the semantic or syntactic levels.

Bibliography

Donnellan, Keith, Reference and Definite Descriptions, in: The Philosophical Review, 75, 3, 1966, pp. 281-304.

Grice, H. P., Meaning, Philosophical Review, 66, 1957, pp. 377-388.

Kripke, Saul, Speaker’s Reference and Semantic Reference, in: Midwest Studies in Philosophy, 2, 1, 1977, pp. 255–276.

Linsky, Leonard, Referring, Routledge & K. Paul, 1967.

Russell, Bertrand, On Denoting, in: Mind, New Series, 14, 56. 1905, pp. 479–493.

Strawson, P. F., On Referring, in: Mind, New Series, 59, 235, 1950, pp. 320-344.

1 Donnellan, Keith, Reference and Definite Descriptions, in: The Philosophical Review, 75, 3, 1966, pp. 281-304.

2Strawson, P. F., On Referring, in:Mind, New Series, 59, 235, 1950, pp. 320-344.

3Russell, Bertrand, On Denoting, in: Mind, New Series, 14, 56. 1905, pp. 479–493.

4Linsky, Leonard, Referring, Routledge & K. Paul, 1967.

5 Donnellan, Keith (1966), pp. 285.

6 Ibid.

7 Ibid.

8 Kripke, Saul, Speaker’s Reference and Semantic Reference, Midwest Studies in Philosophy, Vol. 2, 1977, pp. 255–276.

9 Grice, H. P., Meaning, Philosophical Review, 66, 1957, pp. 377-388.

10Kripke, Saul (1977), pp. 264.