Identifying reference and truth-values¹

The materials for this paper are: one familiar and fundamental speech-function; one controversy in philosophical logic; and two or three platitudes.

We are to be concerned with statements in which, at least ostensibly, some particular historical fact or event or state of affairs, past or present, notable or trivial, is reported: as that the emperor has lost a battle or the baby has lost its rattle or the emperor is dying or the baby is crying. More exactly, we are to be concerned with an important subclass of such statements, viz. those in which the task of specifying just the historical state of affairs which is being reported includes, as an essential part, the sub-task of designating some particular historical item or items which the state of affairs involves. Not all performances of the reporting task include the performance of this sub-task - the task, I shall call it, of identifying reference to a particular item. Thus, the report that it is raining now, or the report that it was raining here an hour ago, do not. But the statement that Caesar is dying, besides specifying the historical fact or situation which it is the function of the statement as a whole to report, has, as a part of this function, the sub-function of designating a particular historical item, viz. Caesar, which that situation essentially involves. And this part of the function of the whole statement is the whole of the function of part of the statement, viz. of the name ‘ Caesar’.

The speech-function we are to be concerned with, then, is the function of identifying reference to a particular historical item, when such reference occurs as a subfunction of statement. We are to be concerned with it in relation to a particular point of philosophical controversy, viz. the question whether a radical failure in the performance of this function results in a special case of falsity in statement or, rather, in what Quine calls a truth-value gap. The hope is not to show that one party to this dispute is quite right and the other quite wrong. The hope is to exhibit speech-function, controversy and one or two platitudes in a mutually illuminating way.

I introduce now my first pair, a complementary pair, of platitudes. One, perhaps the primary, but not of course the only, purpose of assertive discourse is to give information to an audience of some kind, viz. one’s listener or listeners or reader or readers. Since there is no point in, or perhaps one should say, no possibility of, informing somebody of something of which he is already apprised, the making of an assertive utterance or statement - where such an utterance has in view this primary purpose of assertion - implies a presumption (on the part of the speaker) of ignorance (on the part of the audience) of some point to be imparted in the utterance. This platitude might be called the Principle of the Presumption of Ignorance. It is honored to excess in some philosophical proposals for analysis or reconstruction of ordinary language, proposals which might appear to be based on the different and mistaken Principle of the Presumption of Total Ignorance. To guard against such excess, we need to emphasize a platitude complementary to the first. It might be called the Principle of the Presumption of Knowledge. The substance of this complementary platitude, loosely expressed, is that when an empirically assertive utterance is made with an informative intention, there is usually or at least often a presumption (on the part of the speaker) of knowledge (in the possession of the audience) of empirical facts relevant to the particular point to be imparted in the utterance. This is too loosely expressed. The connexion between the presumption of knowledge and the intention to impart just such-and-such a particular point of information may be closer than that of customary association; the connexion between the identity of the particular point it is intended to impart and the kind of knowledge presumed may be closer than that of relevance. Just as we might say that it could not be true of a speaker that he intended to inform an audience of some particular point unless he presumed their ignorance of that point, so we might often say that it could not be true of a speaker that he intended to inform the audience of just that particular point unless he presumed in his audience certain empirical knowledge. So the second principle, in which I am mainly interested, is truly complementary to the first.

Now this may sound a little mysterious. But at least there will be no difficulty felt in conceding the general and vague point that we do constantly presume knowledge as well as ignorance on the part of those who are the audiences of our assertive utterances, and that the first kind of presumption, as well as the second, bears importantly on our choice of what we say. The particular application that I want to make of this general point is to the case of identifying reference. To make it, I must introduce the not very abstruse notion of identifying knowledge of particulars.

Everyone has knowledge of the existence of various particular things each of which he is able, in one sense or another, though not necessarily in every sense, to distinguish from all other things. Thus a person may be able to pick a thing out in his current field of perception. Or he may know there is a thing (not in his current field of perception) to which a certain description applies which applies to no other thing: such a description I shall call an identifying description. Or he may know the name of a thing and be able to recognize it when he encounters it, even if he can normally give no identifying description of it other than one which incorporates its own name. If a man satisfies any one of these conditions in respect of a certain particular, I shall say he has identifying knowledge of that particular. One is bound to define such a notion in terms of its outlying cases, cases, here, of minimal and relatively isolated identifying knowledge. So it is worth emphasizing that, in contrast with cases of minimal and relatively isolated identifying knowledge, there are hosts of cases of very rich and full identifying knowledge, and that, in general, our identifying knowledge of particulars forms an immensely complex web of connexions and relations - the web, one might say, of our historical and geographical knowledge in general, granted that these adjectives are not to be construed as qualifying academic subject alone, but also knowledge of the most unpretentious kind about the particular things and people which enter into our minute-to-minute or day-to-day transactions with the world.

The notion of identifying reference is to be understood in close relation to the notion of identifying knowledge. When people talk to each other they commonly and rightly assume a large community of identifying knowledge of particular items. Very often a speaker knows or assumes that a thing of which he has such knowledge is also a thing of which his audience has such knowledge. Knowing or assuming this, he may wish to state some particular fact regarding such a thing, e.g. that it is thus-and-so; and he will then normally include in this utterance an expression which he regards as adequate, in the circumstances of utterance, to indicate to the audience which thing it is, of all the things in the scope of the audience’s identifying knowledge, that he is declaring to be thus-and-so. The language contains expressions of several celebrated kinds which are peculiarly well adapted, in different ways, for use with this purpose. These kinds include proper names, definite and possessive and demonstrative descriptions, demonstrative and personal pronouns. I do not say that all expression of these kinds are well adapted for use with this purpose; nor do I say, of those that are, that they are not regularly used in other ways, with other purposes.

When an expression of one of these classes is used in this way, I shall say that it is used to invoke identifying knowledge known or presumed to be in possession of an audience. It would now be easy to define identifying reference so that only when an expression is used to invoke identifying knowledge is it used to perform the function of identifying reference. But though it would simplify exposition thus to restrict attention to what we shall in any case count as the central cases of identifying reference, it would not be wholly desirable. For there are cases which cannot exactly be described as cases of invoking identifying knowledge, but which are nevertheless sufficiently like cases which can be so described to be worth classifying with them as cases of identifying reference. For instance, there may be within a man’s current field of possible perception something which he has not noticed and cannot be said actually to have discriminated there, but to which his attention may be intentionally drawn simply by the use, on the part of a speaker, of an expression of one of the kinds I mentioned, as part of a statement of some fact regarding the particular item in question. In so far as the speaker’s intention, in using the expression in question, is not so much to inform the audience of the existence of some particular item unique in a certain respect as to bring it about that the audience sees for itself that there is such an item, we may think this case worth classifying with the central cases of identifying reference. Again, there are cases in which an audience cannot exactly be credited with knowledge of the existence of a certain item unique in a certain respect, but can be credited with a strong presumption to this effect, can be credited, we might say, with identifying presumption rather than identifying knowledge. Such presumed presumption can be invoked in the same style as such knowledge can be invoked.

So we may allow the notion of identifying reference to a particular item to extend beyond the cases of invoking identifying knowledge. We must then face the unsurprising consequence that if, as we do, we wish to contrast cases in which a speaker uses an expression to perform the function of identifying reference with cases in which the intention and effect of a speaker’s use of an expression is to inform the audience of the existence of a particular item unique in a certain respect, then we shall encounter some cases which do not clearly belong to either of these contrasting classes, which seem more or less dubious candidates for both. But this is not a situation which should cause us embarrassment, in philosophy; and, having made the point, I shall for simplicity’s sake speak in what follows as if all cases of identifying reference were, at least in intention, cases of invoking identifying knowledge.

What I have said so far, in describing the function of identifying reference, is I think, uncontroversial in the sense that the description has proceeded without my having to take up a position on any well-worn point of controversy. It has a consequence, just alluded .to, which should be equally uncontroversial, and which I shall labour a little now, partly in order to distinguish it from any prise de position on a matter which is undoubtedly one of controversy.

I have explained identifying reference - or the central case of identifying reference - as essentially involving a presumption, on the speaker’s part, of the possession by the audience of identifying knowledge of a particular item. Identifying knowledge is knowledge of the existence of a particular item distinguished, in one or another sense, by the audience from any other. The appropriate stretch of identifying knowledge is to be invoked by the use of an expression deemed adequate by the speaker, in the total circumstances of utterance, to indicate to the audience which, of all the items within the scope of the audience’s identifying knowledge, is being declared, in the utterance as a whole, to be thus-and-so. Depending upon the nature of the item and the situation of utterance, the expression used may be a name or a pronoun or, a definite or demonstrative description; and it is of course not necessary that either name or description should in general be uniquely applicable to the item in question, so long as its choice, in the total circumstances of utterance, is deemed adequate to indicate to the audience which, of all the particular items within the scope of his identifying knowledge, is being declared, in the utterance as a whole, to be thus-and-so.

Now one thing that is absolutely clear is that it can be no part of the speaker’s intention in the case of such utterances to inform the audience of the existence of a particular item bearing the name or answering to the description and distinguished by that fact, or by that fact plus something else known to the audience, from any other. On the contrary, the very task of identifying reference, as described, can be undertaken only by a speaker who knows or presumes his audience to be already in possession of such knowledge of existence and uniqueness as this. The task of identifying reference is defined in terms of a type of speaker-intention which rules out ascription to the speaker of the intention to impart the existence-and-uniqueness information in question. All this can be put, perfectly naturally, in other ways. Thus, that there exists a particular item to which the name or description is applicable and which, if not unique in this respect, satisfies some uniqueness-condition known to the hearer (and satisfies some uniqueness-condition known to the speaker) is no part of what the speaker asserts in an utterance in which the name or description is used tc perform the function of identifying reference; it is, rather, a presupposition of his asserting what he asserts.

This way of putting it is still uncontroversial. For it is a natural way of putting what is itself uncontroversial. But it introduces a contrast, between the asserted and the presupposed, in words which are associated with an issue of controversy.

We can come at this issue by considering some of the ways in which an attempt to perform the function of identifying reference can either fail altogether or at least fall short of complete success and satisfactoriness. There are several ways in which such an attempt can fail or be flawed. For instance, it may be that, though the speaker possesses, the audience does not possess, identifying knowledge of the particular historical item to which the speaker intends to make an identifying reference; that the speaker credits the audience with identifying knowledge the audience does not possess. It may be that though the audience possesses identifying knowledge of the particular item in question, the expression chosen by the speaker fails to invoke the appropriate stretch of identifying knowledge and leaves the audience uncertain, or even misleads the audience, as to who or what is meant. Failures of this kind may be, though they need not be, due to flaws of another kind. For it may be that the speaker’s choice of expression reflects mistakes of fact or language on his part; and such mistake reflecting choices are still flaws, even where they do not mislead, as, for example, references to Great Britain as ‘England’ or to President Kennedy as ‘the U.S. Premier’ are not likely to mislead.

Though these are all cases of flawed or failed reference, they are not cases of the most radical possible kind of failure. For my descriptions of these cases imply that at least one fundamental condition of success is fulfilled, even if others are not fulfilled. They imply at least that there is a particular historical item within the scope of the speaker’s identifying knowledge - even if not all his beliefs about it are true - such that he intends, by suitable choice of expression, to invoke identifying knowledge, presumed by him to be in possession of the audience, of that item. But this condition might fail too; and that in various ways. It might be that there just is no such particular item at all as the speaker takes himself to be referring to, that what he, and perhaps the audience too, take to be identifying knowledge of a particular item is not knowledge at all, but completely false belief. This is but one case of what might uncontroversially be called radical failure of the existence presupposition of identifying reference. It involves no moral turpitude on the part of the speaker. Different would be the case in which the speaker uses an expression, by way of apparently intended identifying reference, to invoke what he knows or thinks the audience thinks to be identifying knowledge, though he, the speaker, knows it to be false belief. The speaker in this case can have no intention actually to refer to a particular historical item, and so cannot strictly fail to carry out that intention. He can have the intention to be taken to have the former intention; and in this intention he may succeed. A full treatment of the subject would call for careful consideration of such differences. For simplicity’s sake, I shall ignore the case of pretence, and concentrate on that case of radical reference-failure in which the failure is not a moral one.

Our point of controversy concerns the following question: given an utterance which suffers from radical reference-failure, are we to say that what we have here is just one special case of false statement or are we to say that our statement suffers from a deficiency so radical as to deprive it of the chance of being either true or false? Of philsophers who have discussed this question in recent years some have plumped uncompromisingly for the first answer, some uncompromisingly for the second; some have been eclectic about it, choosing the first answer for some cases and the second for others; and some have simply contented themselves with sniping at any doctrine that offered, while wisely refraining from exposing any target themselves. In virtue of his Theory of Descriptions and his views on ordinary proper names as being condensed descriptions, Russell might be said to be the patron of the first party - the ‘special case of false statement’ party. One recent explicit adherent of that party is Mr Dummett in his interesting article on Truth (P.A.S. 58-9). The second party - the ‘ neither true nor false ’ party - might be said, with some reservations, to have included Quine, Austin and myself. Quine invented the excellent phrase ‘truth-value gap ’ to characterize what we have in these cases. Austin contrasts this sort of deficiency or, as he calls it, ‘ infelicity ’ in statement with straight¬ forward falsity and prefers to say that a statement suffering from this sort of deficiency is void - ‘ void for lack of reference ’.

Let us ignore the eclectics and the snipers and confine our attention, at least for the moment, to the two uncompromising parties. I do not think there is any question of demonstrating that one party is quite right and the other quite wrong. What we have here is the familiar philosophical situation of one party being attracted by one simplified, theoretical - or ‘ straightened out - concept of truth and falsity, and the other by another. It might be asked: How does ordinary usage speak on the point? And this, as always, is a question which it is instructive to ask. But ordinary usage does not deliver a clear verdict for one party or the other. Why should it? The interests which ordinary usage reflects are too complicated and various for it to provide overwhelming support for either way of simplifying the picture. The fact that ordinary usage does not deliver a clear verdict does not mean, of course, that there can be no other way of demonstrating, at least, that one view is quite wrong. It might be shown, for example, to be inconsistent, or incoherent in some other way. But this is not the case with either of these views. Each would have a certain amount of explaining and adjusting to do, but each could perfectly consistently be carried through. More important, each is reasonable. Instead of trying to demonstrate that one is quite right and the other quite wrong, it is more instructive to see how both are reasonable, how both represent different ways of being impressed by the facts.

As a point of departure here, it is reasonable to take related cases of what are indisputably false statements and then set beside them the disputed case, so that we can see how one party is more impressed by the resemblances, the other by the differences, between the disputed case and the undisputed cases. The relevant undisputed cases are obviously of two kinds. One is that of an utterance in which an identifying reference is successfully made, and all the conditions of a satisfactory or allround successful act of empirical assertion are fulfilled except that the particular item identifying referred to and declared to be thus-and-so is, as a matter of fact, not thus-and-so. It is said of Mr Smith, the new tenant of the Grange, that he is single, when he is in fact married: a statement satisfactory in all respects except that it is, indisputably, false. The other relevant case is that in which an explicit assertion of existence and uniqueness is made. It is said, say, that there is one and not more than one island in Beatitude Bay. And this is false because there is none at all or because there are several.

Now, it might be said on the one side, how vastly different from the ways in which things go wrong in either of these undisputed cases of falsity is the way in which things go wrong in the case of radical reference-failure. A judgment as to truth or falsity is a judgment on what the speaker asserts. But we have already noted the uncontroversial point that the existence-condition which fails in the case of radical reference-failure is not something asserted, but something presupposed, by the speaker’s utterance. So his statement cannot be judged as a false existential assertion. Nor, evidently, can it be judged as an assertion false in the same way as the first undisputed example, i.e. false as being a mischaracterisation of the particular item referred to. For there is no such item for it to be a mischaracterisation of. In general, where there is such an item as the speaker refers to, and the speaker asserts, with regard to that item, that it is thus-and-so, his assertion is rightly assessed as true if the item is thus-and-so, as false if it is not. In the case of radical reference failure, the speaker, speaking in good faith, means his statement to be up for assessment in this way; he takes himself similarly to be asserting with regard to a particular item, that it is thus-and-so. But in fact the conditions of his making an assertion such as he takes himself to be making are not fulfilled. We can acknowledge the character of his intentions and the nature of his speech-performance by saying that he makes a statement; but we must acknowledge, too, the radical character of the way in which his intentions are frustrated by saying that his statement does not qualify as such an assertion as he takes it to be, and hence does not qualify for assessment as such an assertion. But then it does not qualify for truth-or-falsity assessment at all. The whole assertive enterprise is wrecked by the failure of its presupposition.

But now, on the other side, it could be said that what the disputed case has in common with the undisputed cases of falsity is far more important than the differences between them. In all the cases alike, we may take it a genuine empirical statement is made; a form of words is used such that if there were as a matter of fact in the world (in Space and Time) a certain item, or certain items, with certain characteristics - if, to put it differently, certain complex circumstances did as a matter of fact obtain in the world (in Space and Time) - then the statement would be true. The important distinction is between the case in which those complex circumstances do obtain and the case in which they don’t. This distinction is the distinction we should use the words ‘ true ’ and ‘ false ’ (of statements) to mark, even if we do not consistently do so. And this distinction can be drawn equally in the disputed and the undisputed cases. A false empirical statement is simply any empirical statement whatever which fails for factual reasons, i.e. on account of circumstances in the world being as they are and not otherwise, to be a true one. Cases of radical reference-failure are simply one class of false statements.

It no longer seems to me important to come down on one side or the other in this dispute. Both conceptions are tailored, in the ways I have just indicated, to emphasize different kinds of interest in statement; and each has its own merits. My motives in bringing up the issue are three, two of which have already partially shown themselves. First, I want to disentangle this issue of controversy from other questions with which it is sometimes confused. Second, I want to dispel the illusion that the issue of controversy can be speedily settled, one way or the other, by a brisk little formal argument. Third, I want to indicate one way - no doubt there are more - in which, without positive commitment to either rival theory, we may find the issues they raise worth pursuing and refining. I shall say something on all three points, though most on the third.

First, then, the issue between the truth-value gap theory and the falsity theory, which has loomed so large in this whole area of discussion, has done so in a way which might be misleading, which might give a false impression. The impression might be given that the issue between these two theoretical accounts was the crucial issue in the whole area - the key, as it were, to all positions. Thus it might be sup¬ posed that anyone who rejected the view that the Theory of Descriptions gives an adequate general analysis, or account of the functioning, of definite descriptions was committed, by that rejection, to uncompromising adherence to the truth-value gap theory and uncompromising rejection of the falsity theory for the case of radical reference-failure. But this is not so at all. The distinction between identifying reference and uniquely existential assertion is something quite undeniable. The sense in which the existence of something answering to a definite description used for the purpose of identifying reference, and its distinguishability by an audience from anything else, is presupposed and not asserted in an utterance containing such an expression, so used, stands absolutely firm, whether or not one opts for the view that radical failure of the presupposition would deprive the statement of a truthvalue. It remains a decisive objection to the Theory of Descriptions, regarded as embodying a generally correct analysis of statements containing definite descriptions, that, so regarded, it amounts to a denial of these undeniable distinctions. I feel bound to labour the point a little, since I may be partly responsible for the confusion of these two issues by making the word ‘ presupposition ’ carry simultaneously the burden both of the functional distinction and of the truth-value gap theory. Only, at most, partly responsible; for the line-up is natural enough, though not inevitable; and though there is no logical compulsion one way, there is logical compulsion the other. One who accepts the Theory of Descriptions as a correct analysis is bound to accept the falsity theory for certain cases and reject the truth-value gap theory. One who accepts the truth-value gap theory is bound to reject the Theory of Descriptions as a generally correct analysis. But it is perfectly consistent to reject the view that the Theory of Descriptions is a generally correct analysis, on the grounds I have indicated, and also to withhold assent to the truth-value gap theory.

Now for my second point. I have denied that either of the two theories can be decisively refuted by short arguments, and I shall support this by citing and commenting on some specimen arguments which are sometimes thought to have this power. First, some arguments for the truth-value gap theory and against the falsity theory:

A (1) Let Fa represent a statement of the kind in question. If the falsity theory is correct, then the contradictory of Fa is not -Fa, but the disjunction of -Fa with a negative existential statement. But the contradictory of Fa is -Fa. Therefore the falsity theory is false.

(2) If ‘false’ is used normally, then from It is false that S is P it is correct to infer S is not P. But it is agreed on both theories that S is not P is true only if there is such a thing as S. Hence, if ‘false’ is used normally, It is false that S is P is true only if there is such a thing as S. Hence, if ‘ false ’ is used normally in the statement of the falsity theory, that theory is false.

(3) The question Is S P? and the command Bring it about that S is P may suffer from exactly the same radical reference-failure as the statement S is P. But if an utterance which suffers from this radical reference-failure is thereby rendered false, the question and command must be said to be false. But this is absurd. So the falsity theory is false.

Now arguments on the other side:

B (1) Let Fa represent a statement of the kind in question (e.g. The king of France is bald). Then there may be an equivalent statement, Gb, (e.g. France has a bald king) which is obviously false if there is no such thing as a. But the two statements are equivalent. So Fa is false if there is no such thing as a. So the truth-value gap theory is false.

(2) Let P be a statement which, on the truth-value gap theory, is neither true nor false. Then the statement that P is true is itself false. But if it is false that P is true, then P is false. In the same way we can derive from the hypothesis the conclusion that P is true, hence the conclusion that P is both true and false. This is self¬ contradictory, hence the original hypothesis is so too.

The defender of either view will have little difficulty in countering these arguments against it. Thus to B2 the reply is that if a statement lacks a truth-value, any statement assessing it as true simpliciter or as false simpliciter similarly lacks a truth-value. So no contradiction is derivable. To B 1 the reply is that the argument is either inconclusive or question-begging. If ‘ equivalent ’ means simply ‘ such that if either is true, then necessarily the other is true ’, it is inconclusive. If it also means ‘ such that if either is false, then necessarily the other is false’, it is question-begging. To A3 the reply is that there is no reason why what holds for statements should hold also for questions and commands. To A 2 the reply is that the inference is not strictly correct, though it is perfectly natural that we should normally make it. To A1 the reply is that it is question-begging, though again it is perfectly intelligible that we should be prone to think of contradictories in this way.

It is just an illusion to think that either side’s position can be carried by such swift little sallies as these. What we have, in the enthusiastic defence of one theory or the other, is a symptom of difference of direction of interest. One who has an interest in actual speech-situations, in the part that stating plays in communication between human beings, will tend to find the simpler falsity theory inadequate and feel sympathy with - though, as I say, he is under no compulsion, exclusively or at all, to embrace - its rival. One who takes a more impersonal view of statement, who has a picture in which the actual needs, purposes and presumptions of speakers and hearers are of slight significance - in which, as it were, there are just statements on the one side and, on the other, the world they should reflect - he will naturally tend to brush aside the truth-value gap theory and embrace its simpler rival. For him, one might say, the subject of every statement is just the world in general. For his opponent, it is now this item, now that; and perhaps sometimes - rarely and disconcertingly enough - nothing at all.

And now for the third matter, which we shall find not unconnected with this last thought. It seems to be a fact which advocates of either, or of neither, theory can equally safely acknowledge, that the intuitive appeal, or prima facie plausibility, of the truth-value gap theory is not constant for all example-cases of radical reference failure which can be produced or imagined. We can, without commitment to either theory, set ourselves to explain this variation in the intuitive appeal of one of them - which is also an inverse variation in the intuitive appeal of the other. The attempt to explain this fact may bring into prominence other facts which bear in interesting ways upon speech-situations in general, and those involving identifying reference in particular. I shall draw attention to but one factor - no doubt there are more - which may contribute to the explanation of this fact in some cases. In doing so, I shall invoke another platitude to set beside, and connect with, the platitudes we already have.

First, we may note that the truth-value gap theory can be expressed, in terms of the familiar idea of predication, in such a way as to secure for it a certain flexibility in application. Let us call an expression as and when used in a statement with the role of identifying reference - whether or not it suffers in that use from radical reference-failure - a referring expression. Then any statement containing a referring expression, E, can be regarded as consisting of two expression-parts, one the expression E itself, to be called the subject-expression or subject-term, and the other the remainder of the statement, to be called the predicate-expression or predicate-term. In the case of a statement containing more than one, say two, referring expressions, it is to be open to us to cast one of these for the role of subject-expression, while the other is regarded as absorbed into the predicate-term which is attached to the subject-term to yield the statement as a whole. The adherent of the truth-value gap view can then state his view as follows.² The statement or predication as a whole is true just in the case in which the predicate-term does in fact apply to (is in fact ‘true of’) the object which the subject-term (identifyingly) refers to. The statement or predication as a whole is false just in the case where the negation of the predicate-term applies to that object, i.e. the ease where the predicate-term can be truthfully denied of that object. The case of radical reference-failure on the part of the subject-term is of neither of these two kinds. It is the case of the truth-value gap.

Now consider a statement consisting of two referring expressions one of which is guilty of reference-failure while the other is not. Then it is open to us to carve up the statement in two different ways; and different decisions as to carving-up procedure may be allowed to result in different assessments of the statement for truth-value. Thus (i) we can see the guilty referring expression as absorbed into a predicate-term which is attached to the innocent referring expression to make up the statement as a whole; or (2) we can see the innocent referring expression as absorbed into a predicate-expression which is attached to the guilty referring expression to make up the statement as a whole. Now if we carve up the statement in the second way, we must say - according to our current statement of the truth-value gap theory - that the statement lacks a truth-value. But if we carve it up in the first way, we may say that it is false (or, sometimes - when negative in form - that it is true). For to carve it up in the first way is to think of the statement as made up of the satisfactory or innocent referring expression together with one general term or predicate into which the guilty referring expression has been absorbed. The question whether that predicate does or does not apply to the object referred to by the satisfactory referring expression remains a perfectly answerable question; and the fact that the predicate has absorbed a guilty referring expression will, for most predicates affirmative in form, merely have the consequence that the right answer is ‘ No ’. Thus, if we look at such a statement in this way, we can naturally enough declare it false or untrue, and naturally enough affirm its negation as true, on the strength of the reference-failure of the guilty referring expression.

In this way, it might seem, the truth-value gap theory can readily modify itself to take account of certain examples which may seem intuitively unfavourable to it. For example, there is no king of France; and there is, let us say, no swimming-pool locally. But there is, let us say, an Exhibition in town; and there is, let us say, no doubt of Jones’ existence. If we consider the statements

(1) that Jones spent the morning at the local swimming-pool

and

(2) that the Exhibition was visited yesterday by the king of France

it may seem natural enough to say (1) that it is quite untrue, or is false, that Jones spent the morning at the local swimming-pool, since there isn’t one; that, however Jones spent the morning, he did not spend it at the local swimming-pool, since there’s no such place; and similarly (2) that it is quite untrue, or is false, that the Exhibition was visited yesterday by the king of France; that, whoever the Exhibition was visited by yesterday, it was not visited by the king of France, since there is no such person. And the modified truth-value gap theory accommodates these intuitions by allowing the guilty referring expressions, ‘the local swimming-pool’, ‘the king of France’, to be absorbed into the predicate in each case.

This modification to the truth-value gap theory, though easy and graceful, will scarcely seem adequate. For one thing, it will not be available for all intuitively unfavourable examples, but only for those which contain more than one referring expression. For another, it will remain incomplete inside its own domain unless some principle of choice between alternative ways of carving up a statement is supplied. The theory might resolve the latter question self-sacrificially, by declaring that the carving-up operation was always to be so conducted as to permit the assignment of a truth-value whenever possible. But this move might be too self-sacrificial, turning friends into enemies, turning intuitively favourable cases into unfavourable ones.

So let us consider further. Confronted with the classical example, ‘ The king of France is bald’, we may well feel it natural to say, straight off, that the question whether the statement is true or false doesn’t arise because there is no king of France. But suppose the statement occurring in the context of a set of answers to the question : ‘ What examples, if any, are there of famous contemporary figures who are bald? ’ Or think of someone compiling a list in answer to the question, ‘ Who has died recently?’ and including in it the term ‘the king of France’. Or think of someone including the statement ‘The king of France married again’ in a set of statements compiled in reply to the question: ‘ What outstanding events, if any, have occurred recently in the social and political fields?’ In the first two cases the king of France appears to be cited as an instance or example of an antecedently introduced class. In the last case the statement as a whole claims to report an event as an instance of an antecedently introduced class. The question in each case represents the antecedent centre of interest as a certain class - the class of bald notables, the class of recently deceased notables, the class of notable recent events in a certain field - and the question is as to what items, if any, the classes include. Since it is certainly false that the classes, in each case, include any such items as our answers claim they do, those answers can, without too much squeamishness, be simply marked as wrong answers. So to mark them is not to reject them as answers to questions which don’t arise, but to reject them as wrong answers to questions which do arise. Yet the answers need included only one referring expression for a particular item, viz. the one guilty of reference failure; and the questions need not contain any at all.

This suggests a direction in which we might look for the missing principle of choice in the case of our previous examples, those about the swimming pool and the Exhibition, which contained two referring expressions. The point was not, or was not solely, that each contained an extra and satisfactory referring expression. It was rather that we could easily see the centre of interest in each case as being the question, e.g. how Jones spent the morning or what notable visitors the Exhibition has had, or how the Exhibition is getting on. And the naturalness of taking them in this way was increased by the device of putting the satisfactory referring expression first, as grammatical subject of the sentence, and the unsatisfactory one last. We might, for example, have felt a shade more squeamish if we had written ‘The king of France visited the Exhibition yesterday’ instead of ‘The Exhibition was visited yesterday by the king of France ’. We feel very squeamish indeed about ‘ The king of France is bald’ presented abruptly, out of context, just because we don’t naturally and immediately think of a context in which interest is centred, say, on the question What bald notables are there ? rather than on the question What is the king of France like ? or Is the king of France bald? Of course, to either of these two questions the statement would not be just an incorrect answer. These questions have no correct answer and hence, in a sense, no incorrect answer either. They are questions which do not arise. This does not mean there is no correct reply to them. The correct reply is: ‘ There is no king of France.’ But this reply is not an answer to, but a rejection of, the question. The question about bald notables, on the other hand, can be answered, rightly or wrongly. Any answer which purports to mention someone included in the class, and fails to do so, is wrong; and it is still wrong even if there is no such person at all as it purports to mention.

I should like to state the considerations I have been hinting at a little more generally, and with less dependence upon the notion of a question. Summarily my suggestions are as follows.

(1) First comes the additional platitude I promised. Statements, or the pieces of discourse to which they belong, have subjects, not only in the relatively precise senses of logic and grammar, but in a vaguer sense with which I shall associate the words ‘ topic ’ and ‘ about ’. just now I used the hypothesis of a question to bring out, with somewhat unnatural sharpness, the idea of the topic or centre of interest of a statement, the idea of what a statement could be said, in this sense, to be about. But even where there is no actual first-order question to pinpoint for us with this degree of sharpness the answer to the higher-order question, ‘ What is the statement, in this sense, about? ’, it may nevertheless often be possible to give a fairly definite answer to this question. For stating is not a gratuitous and random human activity. We do not, except in social desperation, direct isolated and unconnected pieces of information at each other, but on the contrary intend in general to give or add information about what is a matter of standing or current interest or concern. There is a great variety of possible types of answer to the question what the topic of a statement is, what a statement is ‘ about ’, - about baldness, about what great men are bald, about which countries have bald rulers, about France, about the king, etc. - and not every such answer excludes every other in a given case. This platitude we might dignify with the title, the Principle of Relevance.

(2) It comes to stand beside that other general platitude which I announced earlier under the title, the Principle of the Presumption of Knowledge. This principle, it will be remembered, is that statements, in respect of their informativeness, are not generally self-sufficient units, free of any reliance upon what the audience is assumed to know or to assume already, but commonly depend for their effect upon knowledge assumed to be already in the audience’s possession. The particular application I made of this principle was to the case of identifying reference, in so far as the performance of this function rests on the presumption of identifying knowledge in the possession of the audience. When I say that the new platitude comes to stand beside the old one, I mean that the spheres of (a) what a statement addressed to an audience is about and (b) what, in the making of that statement, the audience is assumed to have some knowledge of already, are spheres that will often, and naturally, overlap.

But (3) they need not be co-extensive. Thus, given a statement which contains a referring expression, the specification of that statement’s topic, what it is about, would very often involve mentioning, or seeming to mention, the object which that expression was intended to refer to; but sometimes the topic of a statement containing such an expression could be specified without mentioning such an object. Let us call the first type of case Type 1 and the second type of case Type 2. (Evidently a statement could be of Type 1 relative to one referring expression it contained and of Type 2 relative to another.)

Now (4) assessments of statements as true or untrue are commonly, though not only, topic-centred in the same way as the statements assessed; and when, as commonly, this is so, we may say that the statement is assessed as putative information about its topic.

Hence (5), given a case of radical reference-failure on the part of a referring expression, the truth-value gap account of the consequences of this failure will seem more naturally applicable if the statement in question is of Type 1 (relative to that referring expression) than if it is of Type 2. For if it is of Type 2, the failure of reference does not affect the topic of the statement, it merely affects what purports to be information about its topic. We may still judge the statement as putative information about its topic and say, perhaps, that the failure of reference has the consequence that it is misinformative about its topic. But we cannot say this if it is a case of Type 1. If it is a case of Type 1, the failure of reference affects the topic itself and not merely the putative information about the topic. If we know of the reference-failure, we know that the statement cannot really have the topic it is intended to have and hence cannot be assessed as putative information about that topic. It can be seen neither as correct, nor as incorrect, information about its topic.

But, it might be said, this account is self-contradictory. For it implies that in a Type 1 case of radical reference-failure the statement does not really have the topic which by hypothesis it does have; it implies that a statement which, by hypothesis, is about something is really about nothing. To this objection we must reply with a distinction. If I believe that the legend of King Arthur is historical truth, when there was in fact no such person, I may in one sense make statements about King Arthur, describe King Arthur and make King Arthur my topic. But there is another sense in which I cannot make statements about King Arthur, describe him or make him my topic. This second sense is stronger than the first. I may suppose myself to be making statements about him in the second, stronger sense; but I am really only making statements about him in the first and weaker sense. If, however, my belief in King Arthur were true and I really was making statements about him in the second sense, it would still be true that I was making statements about him in the first sense. This is why the first is a weaker (i.e. more comprehensive) sense than the second and not merely different from it.

Bearing this distinction of sense in mind, we can now frame a recipe for distinguishing those cases of reference-failure which are relatively favorable to the truth-value gap theory from those cases which are relatively unfavourable to it. The recipe is as follows. Consider in its context the statement suffering from reference-failure and frame a certain kind of description of the speech-episode of making it. The description is to begin with some phrase like ‘ He (i.e. the speaker) was saying (describing). . . ’ and is to continue with an interrogative pronoun, adjective or adverb, introducing a dependent clause. The clause, with its introductory conjunction, specifies the topic of the statement, what it can be said (at least in the weaker, and, if there is no reference-failure, also in the stronger, sense) to be about; while what is said about its topic is eliminated from the description in favor of the interrogative expression. Examples of such descriptions based on cases already mentioned would be:

He was describing how Jones spent the morning

He was saying which notable contemporaries are bald

He was saying what the king of France is like.

If the peccant referring expression survives in the clause introduced by the interrogative, the clause which specifies what the original statement was about, then we have a case relatively favourable to the truth-value gap theory. If the peccant referring expression is eliminated, and thus belongs to what purports to be information about the topic of the original statement, then we have a case relatively unfavourable to the truth-value gap theory. There can be no true or false, right or wrong, descriptions-of-what-the-king-of-France-is-like, because there is no king of France. But there can be right or wrong descriptions-of-how-Jones-spent-the-morning, and the description of him as having spent it at the local swimming-pool is wrong because there is no such place.

It is easy to see why the relevance of these factors should have been overlooked by those philosophers, including myself, who, considering a few example sentences in isolation from possible contexts of their use, have been tempted to embrace, and to generalize, the truth-value gap theory. For, first, it often is the case that the topic of a statement is, or includes, something referred to by a referring expression; for such an expression invokes the knowledge or current perceptions of an audience, and what is of concern to an audience is often what it already knows something about or is currently perceiving. And, second, it often is the case that the placing of an expression at the beginning of a sentence, in the position of grammatical subject, serves, as it were, to announce the statement’stopic. The philosopher, thinking about reference-failure in terms of one or two short and isolated example sentences beginning with referring expressions, will tend to be influenced by these facts without noticing all of what is influencing him. So he will tend to attribute his sense of something more radically wrong than falsity to the presence alone of what is alone obvious, viz. a referring expression which fails of reference; and thus will overlook altogether these considerations about aboutness or topic which I have been discussing.

Let me remark that I do not claim to have done more than mention one factor which may sometimes bear on the fact that a truth-value gap theory for the case of radical reference-failure is apt to seem more intuitively attractive in some instances than it does in others.

¹ This paper is reprinted from Theoria, xxx, 1964, pp. 96-118.

² This way of stating it is in fact implicit in the fundamental definition of predication which Quine gives on p. 96 of Word and Object.