Draft written in occasion of the talk at the University of Genoa, October 23, 2001
Please, do not quote


Relative Truth

In England such concepts as justice, liberty and objective truth are still believed in. (George Orwell, 'England, your England', 1941)


Some years ago I talked over lunch at a college not my own to a young historian who was studying the history of mediaeval philosophy; I remarked that the attitudes of philosophers and of historians to the history of philosophy differed greatly. The historians were interested in influences of one philosopher upon another, I said, whereas the philosophers were interested in whether what they said was true. "You don't mean that you believe in absolute truth?", she exclaimed in horror.

In denying the existence of absolute truth, she was surely denying the value of the work of those whose writings she was studying: for they conceived themselves to be engaged in the pursuit of unqualified truth. One can indeed study the history of something one believes to be without value: a rampant antagonist of religion can still study the history of religion, for example. Still, philosophy, if it is without value, is the occupation of only a tiny minority, not, like religion, something affecting whole societies. Moreover, she might well be thought to be denying the value of any intellectual endeavour, including history; for are not historians engaged in the pursuit of truth concerning a different subject-matter? In coming to conclusions about the history of mediaeval philosophy, was she not deciding that those conclusions were true? Is not to believe something to believe that it is true, to assert something to assert that it is true? Does not someone who denies the existence of absolute truth thereby forswear all beliefs, disqualify himself from asserting anything whatever?

It may be pleaded that to think all truth relative is not to devalue all intellectual endeavour. An artist does not have to believe that aesthetic value is absolute. If you were to convince a painter or sculptor that his works would be meaningless to people in China or India, or to everyone in 600 years' time, he would not conclude that his artistic activity was worthless. It would still be worth while if his work was meaningful to people of his own culture and his own time (and perhaps for two or three centuries to come); after all, a ballet dancer does not despair if his or her performance is not recorded on film. So perhaps a philosopher, scientist or mathematician who is persuaded that the propositions he believes himself to have established cannot be true for ever or for everyone, but only for some minority of people, may yet think that he has not wasted his life in trying to establish what is true for that minority.

It is obvious that the thesis that all truth is relative (to a person, a time, a culture or whatever) will be self-defeating if maintained in total generality. It says that truth tout court never attaches to a statement or proposition: it attaches to it only as relativised to whatever truth is thought to be relative to (a person, a time, a culture or the like); let us call that to which truth is thought to be relative the 'truth-frame'. If the thesis is advanced in full generality, then it must hold for a statement or proposition ascribing truth to another statement or proposition, including one ascribing truth to that other statement or proposition relatively to some given truthframe. Let us take as an example the statement "Relatively to the culture of Victorian England, it is true to say that Shakespeare wrote several great tragedies". If absolutely all truth is relative, it will be senseless to ascribe truth to this statement save when we explicitly or implicitly relativise the ascription to some truth-frame, say to the culture of the early Soviet Union: we shall thus arrive at the statement "Relatively to the culture of the early Soviet Union, it is true to say that, relatively to the culture of Victorian England, it is true to say that Shakespeare wrote several great tragedies". Plainly, our principle will demand yet further relativisation. In fact, the relativisation will be required indefinitely, and we shall never reach a statement that it makes sense to estimate as it stands as true or as false. If this were right, we should never be able to say anything at all, that is, to assert anything that was capable of being assessed as right or wrong; nor could we think anything that could rank as true or false. Our mouths would be stopped; and so would our processes of thought.

Obviously, this cannot be what is meant by those who proclaim that there is no absolute truth. What they mean must be that it is only relatively to some truth-frame that truth can properly be ascribed to a statement or proposition that does not itself incorporate a relativisation to some suitable truth-frame. The most plausible truthframe is a culture, understood not as something that has a history or may evolve, but as it exists at a certain period. Now it is easy to understand why the effect intended by an artist or poet may be produced only on people of a particular culture. They will recognise what is represented, while others may not; they will catch allusions that will pass others by; they have been habituated to react in certain ways to particular shapes, combinations of colours, words, expressions and turns of phrase, to which others may react quite differently. But truth? How can truth vary with culture? How can it be that what was true for XVI-century Italians may be false for XIX-century inhabitants of the Netherlands, or IV-century B.C. Persians, for that matter. Certainly, what people of these different places and times took to be true differed greatly; but surely it does not literally hold good that whatever a group of people take to be true is true for them. Can there not be such a thing as widespread error?

It matters a great deal whether we are considering truth as ascribed to propositions or to statements. A statement, as I am here using the term, is something verbal; it is made in a particular language, and constituted by a declarative sentence in that language. A proposition, on the other hand, in a usage familiar to philosophers which I am following here, is what is expressed by a statement, regarded as actually or hypothetically uttered by a particular person on a particular occasion: it is what the statement says. A proposition may therefore be expressed in different languages, or, often, by different forms of words in the same language. The truth-value of a statement, as I am understanding the term, may change from one time to another. The most trivial reason for this is that we have words whose reference changes with the occasion of utterance: the statement "It is very hot today" may be true on one day and false on the next, because the days referred to are different. This is not so with propositions: the proposition expressed by making that statement on the one day differs from that expressed by making it on the other. More interestingly, a statement made at one time may differ in truth-value from the same statement made a century later because the meanings of the words have changed. A language is part of a culture, and the meanings of its words may change with time: meaning is certainly relative to culture. Again, this is not so with propositions: when the meaning of a sentence has changed, it now expresses a different proposition from that it expressed before the change. If the thesis that all truth is relative means only that the truthvalues of statements may change with their meanings, it is banal. It is not meant to be banal: it is meant to be radical, contemporary, revolutionary, exciting. So we must take it that what is intended is that the truth-values of propositions are relative to cultures (or whatever the correct truth-frames are taken to be).

But how can this be? Have we not characterised the identity of propositions in such a way as to make it quite independent of culture? In the passage from a statement to the proposition it expresses, was not everything allowed for to which the truth-value might be relativised? A favourite criterion among philosophers for two statements' expressing the same proposition is that the condition for their truth is the same. Clearly, this is a logical, not a literary, criterion: under it, "Margaret's father has just died", "Margaret's male parent is recently deceased", "The one who begot Margaret has just passed away" and "Margaret's old man has just kicked the bucket" all express the same proposition. More importantly, in the present context, it is a criterion devised by those who hold that a proposition, if true at all, is true tout court. It cannot therefore be adopted by anyone who takes a relativist view of truth.

The relativist can adapt the criterion, however. He can say that two statements express the same proposition just in case, relatively to every culture, the condition for their truth is the same. If no short cut is available, it may be objected, it would be practically impossible to apply this criterion: in order to decide whether, for example, the statements "The first human beings are dead" and "The first human beings are deceased" express the same proposition, it would be necessary to review in turn every culture that has ever existed, which no one is capable of doing. The relativist may reply that he was not proposing a practicable method of decision: the notion of a proposition is a purely theoretical one, he may plead, and there is no practical necessity to determine whether or not any two given statements express the same proposition.

We are now at the heart of the matter, and at this point I shall forsake the dialogue format, in order to speak for myself and not just for a generic opponent of relativism. I cannot argue in support of the assumptions made by the objector and repudiated by the relativist, because I too reject them. At the same time I cannot side with the relativist, because I think that the objector was arguing along fundamentally correct lines, even if from shaky assumptions. I do not think that there is any room for the relativisation of the truth-values of propositions. In fact, I think that the relativist subverts, not merely truth, but meaning with it; he underrnines the foundations of language, and therewith the foundations of thought itself.

Wittgenstein thought, or at least frequently wrote as if he thonght, that the meaning of a word or sentence is constituted by its use, that is to say, its role in our linguistic exchanges and the behaviour connected with those exchanges. The 'use' of an expression plainly involves more than what I earlier called the truth-relevant ingredient of its meaning; it involves, for instance, conventions about when it is appropriate to employ that form of words. But the most important features of the use of a declarative sentence are twofold: what we count as establishing that statement as true, that is, as definitively justifying its assertion; and what we take to be the consequences of asserting it -- what the speaker commits himself to by asserting it, and how a hearer's subsequent behaviour will be affected by his accepting it. These notions must be understood in the light of the interconnectedness of language. Most statements cannot be established by a bare confrontation with a fact, as if they could be employed as straightforward reports of observation: inference, more or less extended, plays an essential role in establishing them. Likewise, the acceptance of a statement by speaker or hearer may later be manifested by what he says -- what he infers from it -- rather than by his non-linguistic behaviour. But into an account of 'use, the notion of a statement's being true, independently of our taking it as true, does not enter: only the conditions for our recognising it as true and the consequences of our accepting it as true.

It might be objected to this account that it is not a theory of meaning but of linguistic competence, of what someone needs to be able to do to be accepted as understanding a language. The criticism should be rejected: a theory of meaning is fruitless unless it is possible to derive from it an explanation of what understanding -grasp of meaning -- consists in. The theory of meaning most popular among philosophers is that the meaning of a declarative sentence is constituted by the condition that must hold for it to be true (and the meanings of expressions other than whole sentences by the contribution that they make to determining the truthconditions of sentences in which they occur). The correlative account of understanding will then be that the understanding of a declarative sentence is constituted by a grasp of the condition that must hold for it to be true. In my view, this account must be rejected because it cannot explain in what such an interior mental grasp of the truth-condition consists. Obviously, the whole account will be circular if the truth-condition is conceived as grasped by an expression of it in language. It will also be circular if the intention had been to explain our capacity for thought in terms of our capacity for expressing it linguistically. The truth-conditional account of linguistic understanding requires the attribution to a competent speaker of a non- or pre-linguistic grasp of thoughts as complex as those we can express in language; the hypothesis is both occult and superfluous.

A defender of the truth-conditional theory must meet two challenges: to show that it is adequate and to show that it is needed (both sufficient and necessary). For it to be adequate, it must be possible to show how the use of a declarative sentence can be derived from the condition for its truth: how one who grasps that condition is guided to a knowledge of what is required to establish it as true and of the consequences that follow from accepting it as true. Truth-conditional theorists have paid little attention to this requirement. It nevertheless seems plausible that they could satisfy it if they

The derivation whose possibility the truth-conditional theorist needs to demonstrate is not solely in that direction, however: he must also sbow how anyone who attains a general competence in a language can come to form those conceptions of the conditions for the truth of statements in that language from the practice of using it that he has acquired. It plainly is not necessary that, in learning the language, the speaker is introduced to the word for "true" in that language. Even if such a word exists, he might learn the greater part of the language without ever being given that word. The truth-conditional theorist must nevertheless maintain that, in acquiring the language, he must come to grasp the concept of truth, even if he never learns a word for it; and must come, too, to form a conception, for each statement he understands, of the condition that must obtain for that statement to be true. The theorist must aim to show, not merely that he can come to form such conceptions, but that he must. For if it were possible that he shonld attain competence in the language without forming those conceptions, what reason should we have for supposing that anyone had any such conceptions? If that were possible, then it could happen that someone should master the use of words and sentences of the language without possessing that which, according to the truth-conditional theory, constitutes an understanding of those words and sentences; but how should we know he lacked it? All that we ever have to go on, in deciding whether someone knows a language, is the use that he makes of it. The truth-conditional theorist must undertake to show that no one can master the use of the words and sentences of a language without forming a conception of the conditions for the truth of statements made in it. And how is he to do that?

It wonld be wrong to argue that, if an agreement on the meanings of the words of a language entails a common recognition of what establishes statements made in that language as true, it is paradoxical that we should so radically disagree with one another in our opinions, and on that basis to accuse one who subscribes to the conception of meaning as use with inability to resolve this paradox. We hold much to be true whose truth we do not claim to have been established; our judgements depend in large part on estimates of probability that cannot readily be quantified and may therefore differ from one subject to another. They also depend on reasoning which we take to be valid, but may be mistaken in so taking it, even though the canons of valid reasoning are agreed by all: even mathematicians have published fallacious proofs. The equation of meaning with use still leaves much room for divergence over what each of us decides to treat as true.

If we reject the truth-conditional theory of meaning and understanding in favour of the Wittgensteinian equation of meaning with use, we appear to be able to dispense with the notion of a statement's being true independently of our knowledge or means of knowledge, confining our application of the word "true" to the context "establish as true" and "take as true". I suspect this appearance to be illusory, but shall not argue this difficult point here. However this may be, it is plain that any notion of the truth of a statement compatible with such a conception of meaning will have to be explained in terms of our means of establishing the statement as true. The advocate of this conception of meaning must therefore agree with the relativist in eschewing a notion of truth as wholly independent of our means of knowledge. Must he not therefore also agree with the relativist that the truth of a proposition is relative to the culture that judges it true or false?

If we are concerned with the truth of a proposition as expressed in our own language, we are in the same position as before: we must consider how the same proposition will be or would have been expressed by the people of another culture. Since we are concerned with truth, rather than with any other feature of a proposition, we must take the proposition expressed by a statement to be that determined by the truth-relevant ingredient of the statement's meaning. On the conception of meaning as use, this will be that feature of its meaning that determines what is to be taken as establishing the statement as true. So the proposition will be expressed by those of the other culture by means of some statement which matches that by means of which we express it in respect of what is treated as establishing it as true; more exactly, that member of a small segment of their language which is isomorphic in this respect to a corresponding segment of ours. When the term "proposition" is so understood, we are again in the position of our original objector to relativism: there is simply no room within which to relativise the notion of a proposition's being established as true. Of course, there is plenty of room for differences of judgement. The people of the other culture may generally agree, or have agreed, in taking as true propositions which we generally agree in taking as false. But they cannot have established as true, by the correct application of criteria on which they concur, a proposition which we, by the correct application of strictly analogous criteria on which we concur, have established as false. The brunt of the objector's argument was perfectly sound.

What can the relativist say? A feeble reply would be I was not concerned only with what can properly said to have been established as true, but with what is generally believed to be true.
This reduces his claim to banality: everyone knows that prevalent beliefs vary from culture to culture. Something generally thought to be true may later be shown, by the criteria of demonstration that have all along been accepted, to be false. This is not a case of what was formerly true becoming false, or of what was true for them being false for us: it is simply the rectification of a mistake, of what was thought true proving to be false. Or perhaps the relativist will say:

That is what he believes and what he wanted to show: but he has not shown it. For a language to operate as a language, that is, as an instrument of communication, there must be agreement on whatever gives meaning to its sentences: on truth-conditions or on use, on what makes them true or on how they may be recognised as true. Without such agreement, there is no saying that what the hearer understands is what the speaker meant: no real communication occurs. Often this agreement is imperfect; when we stumble against such uncertainties, we must patch them up if we are to progress towards agreement on truth, since until we have done so we do not know what we are agreeing on or whether we are really disagreeing about anything. The invention of writing made it possible for people of the past to communicate with us, and for us to communicate with each other over great distances. Certainly differences between languages and change within a language place obstacles to this communication; but there must be a criterion for deciding whether we are interpreting another, still living or long dead, aright. A language only serves to express definite thonghts if there are criteria accepted by all its speakers for whether statements made in that language are right or wrong; in the absence of such criteria, there is no meaning. The relativist spurns such criteria, and, in so doing perverts language by which we convey our thoughts to one another. He calls down a curse upon us worse than that which God called down on the builders of Babel; rather than our speaking different languages, not to be speaking a genuine language at all.

Michael Dummett




(*) Letteres X, Y and Z are in the original Greek letters.