The problem of proper names¹

At first sight nothing seems easier to understand in the philosophy of language than our use of proper names: here is the name, that is the object. The name stands for the object.

Although this account is obviously true, it explains nothing. What is meant by ‘ stands for ’ ? And how is the relation indicated by ‘ stands for ’ ever set up in the first place? Do proper names ‘ stand for ’ in the same way that definite descriptions ‘ stand for ’ ? These and other questions which I wish to attack can be summed up in the question, ‘ Do proper names have senses ? ’ What this question asks, as a start, is what, if any, similarity is there between the way a definite description picks out its referent and the way a proper name picks out its referent. Is a proper name really a shorthand description? We shall see that the two opposing answers given to this question arise from the tension between, on the one hand, the almost exclusive use of proper names to perform the speech act of reference, and, on the other hand, the means and preconditions for performing this speech act especially the condition expressed in the principle of identification. The principle of identification may be stated as follows:

A necessary condition for the successful performance of a definite reference in the utterance of an expression is that either the expression must be an identifying description or the speaker must be able to produce an identifying description on demand.

For an explanation of, arguments for, and qualifications to this principle, see: John A. Searle, Speech Acts, Cambridge, 1969, pp. 77 ff.

The first answer goes something like this: proper names do not have senses, they are meaningless marks; they have denotation but no connotation (Mill).² The argument for this view is that whereas a definite description refers to an object only in virtue of the fact that it describes some aspect of that object, a proper name does not describe the object at all. To know that a definite description fits an object is to know a fact about that object, but to know its name is not so far to know any facts about it. This difference between proper names and definite descriptions is further illustrated by the fact that we can often turn a definite description (a referring expression) into an ordinary predicative expression by simply substituting an indefinite article for the definite, e.g., ‘a man’ for ‘the man’. No such shift is in general possible with proper names. When we do put the indefinite article in front of a proper name it is either a shorthand way of expressing well-known characteristics of the bearer of the name (e.g., ‘He is a Napoleon’ means ‘He is like Napoleon in many respects ’), or it is a shorthand form of a formal-mode expression about the name itself (e.g., ‘He is a Robert’ means ‘He is named Robert’). In short we use a proper name to refer and not to describe; a proper name predicates nothing and consequently does not have a sense.

Our robust common sense leads us to think that this answer must be right, but though it has enormous plausibility, we shall see that it cannot be right, at least not as it stands, for too many facts militate against it. First, let us look at some of the metaphysical traps that an uncritical acceptance of such a view is likely to lead us into. The proper name, we are inclined to say, is not connected with any aspects of the object as descriptions are, it is tied to the object itself. Descriptions stand for aspects or properties of an object, proper names for the real thing. This is the first step on the road that leads to substance, for it fastens on to what is supposed to be a basic metaphysical distinction between objects and properties or aspects of objects, and it derives this distinction from an alleged difference between proper names and definite descriptions. Such a muddle is to be found in the Tractatus, ‘The name means the object. The object is its meaning ’ (3.203).³ But notice to what interesting paradoxes this leads immediately: the meaning of words, it seems, cannot depend on any contingent facts in the world, for we can still describe the world even if the facts alter. Yet the existence of ordinary objects - people, cities, etc. - is contingent, and hence the existence of any meaning for their names is contingent. So their names are not the real names at all! There must exist a class of objects whose existence is not a contingent fact, and it is their names which are the real names.⁴ And what does this mean? Here we see another good illustration of the original sin of all metaphysics, the attempt to read real or alleged features of language into the world.

The usual rejoinder to the thesis that there is a basic metaphysical distinction between objects and properties is that objects are just collections of properties.⁵ The first thesis is derived from the distinction between referring and predicating, the second thesis is derived from the tautology that everything that can be said about an object can be said in descriptions of that object. But both theses are equally nonsensical. It is nonsense to suppose that an object is a combination of its propertyless self and its properties, and it is nonsense to suppose that an object is a heap or collection of properties. Again, both views have a common origin in the metaphysical mistake of deriving ontological conclusions from linguistic theses.

There are three objections to the view that proper names do not have senses:

1. We use proper names in existential propositions, e.g., ‘there is such a place as Africa ’, ‘ Cerberus dees not exist ’. Here proper names cannot be said to refer, for no such subject of an existential statement can refer. If it did, the precondition of its having a truth value would guarantee its truth, if it were in the affirmative, and its falsity, if it were in the negative. (This is just another way of saying that ‘ exists ’ is not a predicate.) Every existential statement states that a certain predicate is instantiated. (As Frege put it, existence is a second order concept.)⁶ An existential statement does not refer to an object and state that it exists, rather it expresses a concept and states that that concept is instantiated.

Thus, if a proper name occurs in an existential statement it mast have some conceptual or descriptive content. Attempts such as Russell’s⁷ to evade this point have taken the form of saying that such expressions are not really proper names, a desperate maneuver which shows that something must be wrong with the assumptions which drive one to it.

2. Sentences containing proper names can be used to make identity statements which convey factual and not merely linguistic information. Thus the sentence, ‘ Everest is Chomolungma ’ can be used to make an assertion which has geographical and not merely lexicographical import. Yet if proper names were without senses, then the assertion could convey no more information than does an assertion made with the sentence ‘ Everest is Everest ’. Thus it seems that proper names must have descriptive content, they must have a sense. This is substantially Frege’s argument that proper names have senses.⁸

3. The principle of identification requires that an utterance of a proper name must convey a description just as the utterance of a definite description must if the reference is to be consummated. And from this it seems to follow that a proper name is a kind of shorthand description.

All three objections point to the same conclusion, namely, that proper names are shorthand definite descriptions.

But it seems that this conclusion cannot be right, for, aside from its grotesque unplausibility, it is inconsistent with too many obvious truths. First, if it were the case that a proper name is a shorthand description, then descriptions should be available as definitional equivalents for proper names; but we do not, in general, have definitions of proper names. In so called dictionaries of proper names, one finds descriptions of the bearers of the names, but in most cases these descriptions are not definitional equivalents for the names, since they are only contingently true of the bearers.

Not only do we not have definitional equivalents, but it is not clear how we could go about getting them to substitute in all cases for proper names. If we try to present a complete description of the object as the sense of the name, odd consequences would ensue, e.g., any true statement about the object using the name as subject would be analytic, any false one self-contradictory, the meaning of the name (and perhaps the identity of the object) would change every time there was any change at all in the object, the name would have different meanings for different people, etc. So it seems that the view that proper names are descriptions cannot be true either.

Here we have a beautiful example of a philosophical problem: on the one hand common sense drives us to the conclusion that a proper name is not a species of description, that it is sui generis, but against this a series of theoretical considerations drive us to the conclusion that it must be a shorthand definite description. But against this too we can adduce serious arguments. This antinomy admits of a solution toward which I shall now argue.

We might rephrase our original question, ‘ Do proper names have senses? ’ as ‘ Do referring uses of proper names entail any descriptive predicates? ’ or simply ‘ Are any propositions where the subject is a proper name and the predicate a descriptive expression analytic?’⁹ But this question has a weaker and a stronger form: (a) the weaker: ‘ Are any such statements at all analytic ? ’ and (b) the stronger: ‘ Are any statements where the subject is a proper name and the predicate an identifying description analytic?’

Consider the first question. It is characteristic of a proper name that it is used to refer to the same object on different occasions. The use of the same name at different times in the history of the object presupposes that the object is the same; a necessary condition of identity of reference is identity of the object referred to. But to presuppose that the object is the same in turn presupposes a criterion of identity: that is, it presupposes an ability on the part of the speaker to answer the question, ‘ In virtue of what is the object at time t. ɪ, referred by name N, identical with the object at time t. 2, referred to by the same name?’ or, put more simply, ‘The object at time t. ɪ is the same what as the object at time t. 2? ’ and the gap indicated by ‘what ’ is to be filled by a descriptive general term; it is the same mountain, the same person, the same river, the general term providing in each case a temporal criterion of identity. This gives us an affirmative answer to the weaker question. Some general term is analytically tied to any proper name: Everest is a mountain, the Mississippi is a river, de Gaulle is a person. Anything which was not a mountain could not be Everest, etc., for to secure continuity of reference we need a criterion of identity, and the general term associated with the name provides the criterion. Even for those people who would want to assert that de Gaulle could turn into a tree or horse and still be de Gaulle, there must be some identity criterion. De Gaulle could not turn into anything whatever, e.g., a prime number, and still remain de Gaulle, and to say this is to say that some term or range of terms is analytically tied to the name ‘ de Gaulle ’.

To forestall an objection: one temptation is to say that if we continue to call an object ‘ Everest ’, the property of being called ‘ Everest ’ is sufficient to guarantee that it is the same. But the point of the above analysis is that we are only justified in calling it ‘ Everest ’ if we can give a reason for supposing it to be identical with what we used to call ‘ Everest ’ and to give as the reason that it is called ‘ Everest ’ would be circular. In this sense at least, proper names do have ‘ connotations ’.

But the answer ‘ yes ’ to the weaker question does not entail the same answer to the stronger one, and it is the stronger form which is crucial for deciding whether or not a proper name has a sense, as Frege and I use the word. For according to Frege the sense of a proper name contains the ‘ mode of presentation ’ which identifies the referent, and of course a single descriptive predicate does not provide us with a mode of presentation; it does not provide an identifying description. That Socrates is a man may be analytically true, but the predicate ‘man’ is not an identifying description of Socrates.

So let us consider the stronger formulation of our question in the light of the principle of identification. According to this principle, anyone who uses a proper name must be prepared to substitute an identifying description (remembering that identifying descriptions include ostensive presentations) of the object referred to by a proper name. If he were unable to do this, we should say that he did not know whom or what he was talking about, and it is this consideration which inclines us, and which among other things inclined Frege, to say that a proper name must have a sense, and that the identifying description constitutes that sense. Think what it is to learn a proper name. Suppose you say to me: ‘ Consider Thaklakes, tell me what you think of Thaklakes.’ If I have never heard that name before I can only reply, ‘Who is he?’ or ‘What is it?’ And does not your next move - which according to the principle of identification consists in giving me an ostensive presentation or a set of descriptions - does this not give me the sense of the name, just as you might give me the sense of a general term? Is this not a definition of the name?

We have discussed several objections to this view already; a further one is that the description one man is prepared to substitute for the name may not be the same as the one someone else is prepared to substitute. Are we to say that what is definitionally true for one is only contingent for another? Notice what maneuvers Frege is forced to here:

Suppose further that Herbert Garner knows that Dr Gustav Lauben was born on 13 September 1875, in N. H. and this is not true of anyone else; against this suppose that he does not know where Dr Lauben now lives or indeed anything about him. On the other hand, suppose Leo Peter does not know that Dr Lauben was born on 13 September 1875, in N. H. Then as far as the proper name ‘Dr Gustav Lauben’ is concerned, Herbert Garner and Leo Peter do not speak the same language, since, although they do in fact refer to the same man with this name, they do not know that they do so.¹⁰

Thus according to Frege, unless our descriptive backing for the name is the same, we are not even speaking the same language. But, against this, notice that we seldom consider a proper name as part of one language as opposed to another at all.

Furthermore, I might discover that my identifying description was not true of the object in question and still not abandon his name. I may learn the use of ‘ Aristotle ’ by being told that it is the name of the Greek philosopher born in Stagira, but if later scholars assure me that Aristotle was not born in Stagira at all but in Thebes, I will not accuse them of self-contradiction. But let us scrutinize this more closely: scholars might discover that a particular belief commonly held about Aristotle was false. But does it make sense to suppose that everything anyone has ever believed to be true of Aristotle was in fact not true of the real Aristotle? Clearly not, and this will provide us with the germ of an answer to our question.

Suppose we ask the users of the name ‘Aristotle’ to state what they regard as certain essential and established facts about him. Their answers would constitute a set of identifying descriptions, and I wish to argue that though no single one of them is analytically true of Aristotle, their disjunction is. Put it this way: suppose we have independent means of identifying an object, what then are the conditions under which I could say of the object, ‘ This is Aristotle? ’ I wish to claim that the condition, the descriptive power of the statement, is that a sufficient but so far unspecified number of these statements (or descriptions) are true of the object. In short, if none of the identifying descriptions believed to be true of some object by the users of the name of that object proved to be true of some independently located object, then that object could not be identical with the bearer of the name. It is a necessary condition for an object to be Aristotle that it satisfy at least some of these descriptions. This is another way of saying that the disjunction of these descriptions is analytically tied to the name ‘ Aristotle ’ - which is a quasi-affirmative answer to the question, ‘ Do proper names have senses?’ in its stronger formulation.

My answer, then, to the question, ‘ Do proper names have senses?’ - if this asks whether or not proper names are used to describe or specify characteristics of objects - is ‘ No But if it asks whether or not proper names are logically connected with characteristics of the object to which they refer, the answer is ‘ Yes, in a loose sort of way ’.

Some philosophers suppose that it is an objection to this sort of account that the same word is sometimes used as a name for more than one object. But this is a totally irrelevant fact and not an objection to my account at all. That different objects are named ‘ John Smith ’ is no more relevant to the question ‘ Do proper names have senses ? ’ than the fact that both riversides and finance houses are called ‘ banks ’ is relevant to the question, ‘Do general terms have senses?’ Both ‘bank’ and ‘John Smith ’ suffer from kinds of homonymy, but one does not prove a word meaningless by pointing out that it has several meanings. I should have considered this point too obvious to need stating, were it not for the fact that almost every philosopher to whom I have presented this account makes this objection.

What I have said is a sort of compromise between Mill and Frege. Mill was right in thinking that proper names do not entail any particular description, that they do not have definitions, but Frege was correct in assuming that any singular term must have a mode of presentation and hence, in a way, a sense. His mistake was in taking the identifying description which we can substitute for the name as a definition.

I should point out, parenthetically, that of course the description, ‘The man called X’ will not do, or at any rate will not do by itself, as a satisfaction of the principle of identification. For if you ask me, ‘ Whom do you mean by X? ’ and I answer, ‘ The man called X even if it were true that there is only one man who is called X, I am simply saying that he is the man whom other people refer to by the name ‘ X But if they refer to him by the name ‘ X’ then they must also be prepared to substitute an identifying description for ‘X’ and if they in their turn substitute ‘the man called X’ the question is only carried a stage further and cannot go on indefinitely without circularity or infinite regress. My reference to an individual may be parasitic on someone else’s but this parasitism cannot be carried on indefinitely if there is to be any reference at all.

For this reason it is no answer at all to the question of what if anything is the sense of a proper name ‘X’ to say its sense or part of its sense is ‘called X’. One might as well say that part of the meaning of ‘ horse ’ is ‘ called a horse ’. It is really quite amazing how often this mistake is made.¹¹

My analysis of proper names enables us to account for all the apparently inconsistent views. How is it possible that a proper name can occur in an existential statement? A statement such as ‘Aristotle never existed’ states that a sufficient, but so far unspecified, number of the descriptive backings of ‘ Aristotle ’ are false. Which one of these is asserted to be false is not yet clear, for the descriptive backing of ‘ Aristotle ’ is not yet precise. Suppose that of the propositions believed to be true of Aristotle half were true of one man and half of another, would we say that Aristotle never existed ? The question is not decided for us in advance.

Similarly it is easy to explain identity statements using proper names. ‘ Everest is Chomolungma ’ states that the descriptive backing of both names is true of the same object. If the descriptive backing of the two names, for the person making the assertion, is the same, or if one contains the other, the statement is analytic, if not, synthetic. Frege’s instinct was sound in inferring from the fact that we do make factually informative identity statements using proper names that they must have a sense, but he was wrong in supposing that this sense is as straightforward as in a definite description. His famous ‘Morning Star-Evening Star’ example led him astray here, for though the sense of these names is fairly straightforward, these expressions are not paradigm proper names, but are on the boundary line between definite descriptions and proper names.

Furthermore, we now see how an utterance of a proper name satisfies the principle of identification: if both the speaker and the hearer associate some identifying description with the name, then the utterance of the name is sufficient to satisfy the principle of identification, for both the speaker and the hearer are able to substitute an identifying description. The utterance of the name communicates a proposition to the hearer. It is not necessary that both should supply the same identifying description, provided only that their descriptions are in fact true of the same object.

We have seen that insofar as proper names can be said to have a sense, it is an imprecise one. We must now explore the reasons for this imprecision. Is the imprecision as to what characteristics exactly constitute the necessary and sufficient conditions for applying a proper name a mere accident, a product of linguistic slovenliness ? Or does it derive from the functions which proper names perform for us? To ask for the criteria for applying the name ‘Aristotle’ is to ask in the formal mode what Aristotle is; it is to ask for a set of identity criteria for the object Aristotle. ‘ What is Aristotle?’ and ‘What are the criteria for applying the name “Aristotle”?’ ask the same question, the former in the material mode, and the latter in the formal mode of speech. So if, prior to using the name, we came to an agreement on the precise characteristics which constituted the identity of Aristotle, our rules for using the name would be precise. But this precision would be achieved only at the cost of entailing some specific descriptions by any use of the name. Indeed, the name itself would become logically equivalent to this set of descriptions. But if this were the case we would be in the position of being able to refer to an object solely by, in effect, describing it. Whereas in fact this is just what the institution of proper names enables us to avoid and what distinguishes proper names from definite descriptions. If the criteria for proper names were in all cases quite rigid and specific, then a proper name would be nothing more than a shorthand for these criteria, it would function exactly like an elaborate definite description. But the uniqueness and immense pragmatic convenience of proper names in our language lies precisely in the fact that they enable us to refer publicly to objects without being forced to raise issues and come to an agreement as to which descriptive characteristics exactly constitute the identity of the object. They function not as descriptions, but as pegs on which to hang descriptions. Thus the looseness of the criteria for proper names is a necessary condition for isolating the referring function from the describing function of language.

To put the same point differently, suppose we ask, ‘ Why do we have proper names at all ? ’ Obviously, to refer to individuals. ‘ Yes, but descriptions could do that for us.’ But only at the cost of specifying identity conditions every time reference is made: suppose we agree to drop ‘Aristotle’ and use, say, ‘the teacher of Alexander’, then it is an analytic truth that the man referred to is Alexander’s teacher - but it is a contingent fact that Aristotle ever went into pedagogy. (Though it is, as I have said, a necessary truth that Aristotle has the logical sum [inclusive disjunction] of the properties commonly attributed to him.)¹²

It should not be thought that the only sort of looseness of identity criteria for individuals is that which I have described as peculiar to proper names. Identity problems of quite different sorts may arise, for instance, from referring uses of definite descriptions. ‘This is the man who taught Alexander’ may be said to entail, e.g., that this object is spatio-temporally continuous with the man teaching Alexander at another point in space-time; but someone might also argue that this man’s spatio-temporal continuity is a contingent characteristic and not an identity criterion. And the logical nature of the connection of such characteristics with the man’s identity may again be loose and undecided in advance of dispute. But this is quite another dimension of looseness from that which I cited as the looseness of the criteria for applying proper names, and does not affect the distinction in function between definite descriptions and proper names, viz., that definite descriptions refer only in virtue of the fact that the criteria are not loose in the original sense, for they refer by providing an explicit description of the object. But proper names refer without providing such a description.

We might clarify some of the points by comparing paradigm proper names with degenerate proper names like ‘the Bank of England’. For these limiting cases of proper names, it seems the sense is given as straightforwardly as in a definite description; the presuppositions, as it were, rise to the surface. And a proper name may acquire a rigid use without having the verbal form of a description: God is just, omnipotent, omniscient, etc., by definition for believers. To us, ‘ Homer ’ just means ‘ the author of the Iliad and the Odyssey ’. The form may often mislead us: the Holy Roman Empire was neither holy nor Roman, etc., but it was, nonetheless, the Holy Roman Empire. Again, it may be conventional to name only girls ‘ Martha ’, but if I name my son ‘ Martha ’, I may mislead, but I do not lie. And of course not all paradigm proper names are alike with respect to the nature of their ‘descriptive content ’. There will, e.g., be a difference between the names of living people, where the capacity of the user of the name to recognize the person may be an important ‘ identifying description ’, and the names of historical figures. But the essential fact to keep in mind when dealing with these problems is that we have the institution of proper names to perform the speech act of identifying reference. The existence of these expressions derives from our need to separate the referring from the predicating functions of language. But we never get referring completely isolated from predication, for to do so would be to violate the principle of identification, without conformity to which we cannot refer at all.

¹ This paper is reprinted from J. R. Searle, Speech Acts: An Essay in the Philosophy of Language (Cambridge, 1969), pp. 162-74.

² J. S. Mill, A System of Logic (London and Colchester, 1949), book 1, chapter 2, para. 5.

³ Mill’s: proper names have no meaning, might appear to be inconsistent with Wittgenstein’s: objects are their meanings. But they are not inconsistent. (Ambiguity of ‘mean’ and ‘bedeuten’.) Both say, proper names have referents but not senses.

⁴ Cf. also Plato, Theaetetus.

⁵ E.g., Russell, An Inquiry into Meaning and Truth (London, 1940), p. 97.

⁶ Grundgesetze der Arithmetik (Jena, 1893), vol. 1, section 21.

⁷ ‘The Philosophy of Logical Atomism’, R. Marsh (ed.), Logic and Knowledge (London, 1956), pp. 200 ff.

⁸ Though, with a characteristic perversity, he did not see that this account of identity statements provides an explanation of the use of proper names in existential statements. He thought it was nonsense to use proper names in existential statements. ‘ Uber die Grundlagen der Geometrie II’, Jahresbericht der Deutschen Mathematiker-Vereinigung (1903), p. 373.

⁹ Of course, in one sense of ‘ analytic’, no such subject-predicate proposition can be analytic, since it is in general a contingent fact that the subject expression has a referent at all and hence contingent that the proposition has a truth-value. To meet this objection we can either redefine ‘ analytic’ as: ‘p is analytic = df. if p has a truth-value, it is true by definition’ or we can rephrase the original question as, ‘ Is any proposition of the form “ if anything is S it is P” analytic, where “S” is replaced by a proper name and “P” by a descriptive predicate?’

¹⁰ ‘The Thought: a logical inquiry’, trans. A. and M. Quinton, Mind (1956), p. 297.

¹¹ E.g., A. Church, Introduction to Mathematical Logic (Princeton, 1956), p. 5.

¹² Ignoring contradictory properties, p v ~ p would render the logical sum trivially true.