Printer Friendly

Meaning's role in truth.

What words mean plays a role in determining when they would be true; but not an exhaustive one. For that role leaves room for variation in truth conditions, with meanings fixed, from one speaking of words to another. What role meaning plays depends on what truth is; on what words, by virtue of meaning what they do, are required to have done (as spoken) in order to have said what is true. There is a deflationist position on what truth is: the notion is exhausted by a given, specified, mass of "platitudes", each to the effect that if words said (say) things to be thus, things must be that way. (The thought that thus-and-so is true iff thus-and-so.) These platitudes, and so deflationism, miss that aspect of truth that determines meaning's role. Truth requires words to have the uses which, given what they mean, they should have in the circumstances of their speaking. Through this link with use, when words would be true is a factor fixing what it is they said.

What is truth?(1) A notorious question, tempting many, for millennia, to dismiss it. But even if the answer cannot be a definition, there is much to say about what truth is.(2) Part of what truth is, one might think, is the way the truth of words depends on what they mean. But there is a widespread view, tracing back at least to Frege, on which we may say all there is to say about what truth is without so much as mentioning words. I hope to show this view wrong. Here is my plan. What words mean plays a role in fixing when they would be true; but not an exhaustive one. Meaning leaves room for variation in truth conditions from one speaking to another. What that non-exhaustive role is depends on what it is to have said what is true. Identify the aspect of truth which fixes this role, and the widespread view collapses. So, step by step, I will argue.

Paul Horwich (1990, pp. 4-6) answers the question "What is truth?" with a version of the widespread view:

The proposition that quarks really exist is true if and only if quarks really exist, the proposition that lying is bad is true if and only if lying is bad, ... and so on; but nothing more about truth need be assumed. The entire conceptual and theoretical role of truth may be explained on this basis.

That is certainly one way with notoriety. What exactly is Horwich's answer? First, he evidently supposes that when he said "The proposition that quarks really exist is true iff quarks really exist" he stated a condition for the truth of something, or at least specified a condition under which something or other would be true (and, moreover, identified what that something is). Second, with his "and so on", he suggests that that is just an example of something; and that he could easily have produced examples other than the two he did. I take that suggestion this way. Suppose "S" is a sentence in which Horwich might then have stated something, or at least might then have spoken of something (in particular) being so. Then instead of saying what he did, he might have said "The proposition that S is true iff S". That would be an example as good as either of the two he in fact gave. Let us call any fact anyone might ever state in words of this form an H-equivalence. Now consider the corpus of all the H-equivalences Horwich either stated, or might have, had he chosen different examples. I take Horwich to claim two things about this corpus: first, that it captures all that it is for a proposition to be true; second, that when we have said what it is for a proposition to be true, we have (as good as) said all that need be said as to what truth is, full stop.

There is an Aristotelian platitude which, paraphrasing freely, is to this effect: to state truth is to say things to be some way or other, and, in doing so, to say things to be none other than the way they are. That fits speakers as well as words. There is a parallel platitude about propositions: a proposition is true just in case things are as they are according to it. Must Horwich deny such platitudes, or reject them as somehow misbegotten? Nothing so far requires this. He need only hold that such generalizations hold precisely in case his collection of H-equivalences do. Since Horwich need not reject such platitudes, I will not hold him to doing so. So reading him, I will call his answer to the question "What is truth?" deflationism.

Horwich has said nothing so far about what it would be for words to be true, or to state truth: nothing about conditions for their truth, or conditions under which they would have stated truth, or about just what they would have achieved in being true.(3) At first blush that seems a lacuna. But there is a simple, and widely accepted, story on which that lacuna, such as it is, is not one in an account of what truth is. On the story, for words to be true is (nothing other than) for them to express a proposition that is true. (Let us not haggle about whether that says more about truth than deflationism allows.) Once it has been determined which, if any, proposition words expressed (or would express), it has been determined when they would be true: when that proposition would be true. If they were not true under precisely those conditions, they would not have expressed that proposition. To see when that proposition would be true (the thought is), just consult the relevant H-equivalence.(4) Moreover, whatever determines which proposition words expressed, that does not depend on what it is for words, or anything, to be true; or on no more about this than has been mentioned.

Confidence in this last assertion is generally fortified by a view of what does determine which proposition words did, or would, express: bracketing ambiguities words may have in their language,(5) and modulo the objects (people, etc.) words spoke of on an occasion, which proposition, if any, they expressed is determined by what they mean. Part of their meaning what they do is that (modulo referents) there is a particular proposition which is the one they would express.(6) The words "Pigs grunt", e.g., meaning what they do, express a certain proposition which we may refer to as "the proposition that pigs grunt". The words "Fred is fat", spoken on an occasion, expressed the proposition, of a particular person who they called "Fred", that that person was, at the time of their speaking, fat. And so on. (On this view, if we know what words mean, we can always identify the proposition they expressed on an occasion, though if we don't know their referents, our identification might not be in a very helpful form.)

I will try to show that this is a wrong view of what meaning does, and that, given the right view, deflationism is a mistake. Not that what words mean is irrelevant to when they would be true. Meaning fixes something words would do (and say) wherever spoken meaning what they do; something they are for, so also something about what they ought to do. Truth requires that they do all that sufficiently well, that is, up to the standard truth imposes. But all that meaning fixes allows for words to state truth, but also falsehood, of given items in given conditions. What meaning fixes often enough leaves both possibilities open. That means, I will argue, that these requirements for truth cannot be captured in the form "If words expressed the proposition P, then they are true only where the condition for the truth of P is satisfied." A given proposition is true just where the world is thus and so (or so the deflationist picture asks us to suppose). But there is no one way the world must be to supply what is required for the truth of words with given meaning. On the contrary, for different speakings of words alike in meaning, there are different ways the world must be. To capture this sensitivity of truth's demands to speakings, we must see how those demands make the way the world is matter differently to the truth of different such speakings. To do that we must depart from the forms deflationism allows.

On one model, words, meaning what they do, say, or said, thus and so, and then are simply true, or not, according as that is so or not; no further factors need be mentioned in the story. On the model I recommend, truth depends on what words mean, the way the world is, and further factors: aspects of the circumstances in which words were produced. Whether, and how, any given such factor matters to the truth of particular words must depend on what truth is; on the particular way speaking it requires words to relate to those factors (or vice-versa). Further factors can matter at all, and then as they do, only if truth is a notion which demands, case by case, precisely those factors to be arranged in just those ways. Deflationism cannot allow truth to make such demands.


How does what words mean relate to what is said in speaking them? Consider the sentence "The ball is round", and two cases of its use. Case A: What shape do squash balls assume on rebound? Pia hits a decent stroke; Jones watches. "The ball is round", she says at the crucial moment. Wrong. It has deformed into an ovoid. Jones did not say the ball to be as it was, so spoke falsely. Case B: Fiona has never seen squash played. From her present vantage point the ball seems a constant blur. "What shape is that ball?", she asks. "The ball is round", Alf replies; truly, since that it is the sort of ball a squash ball (and this one) is. It is not, e.g., like a very small rugby ball.

So there are both true things and false things--thus a variety of different things--to be said of a given ball, and of the way it is at a given time, in the words "The ball is round", used so as to have meant (as used) what they mean (in English). What those words mean, or their meaning what they do, makes no one of these things "that which those words say" (in English). What those words mean leaves it open for them to be used (in suitable circumstances) to say any of various things, each true under, and on, different conditions. There is no one set of conditions under which those English words, spoken of a given ball and time, would be, or say what is, true. Nor even one condition which is the condition for them to be true. If differences in truth condition make for different propositions, then what those words mean makes no one proposition the one (modulo referents) they express.(7)

The example is not special. Most English sentences behave the same. Supplied with referents, there is still no one fact, or falsehood, they would state. That is not what meaning makes them do. Though it might be nice to make a fuller case, I will not pause to do so here.(8)

We can see why there should be this variety of things to say if we ask what meaning does do. The words "is round", in meaning what they do, speak of being round. In fact, I suggest, for them to speak of that is just for them to mean what they do.(9) For English words to speak of being round comes to just this.(10) If you use them as meaning what they do, you will thereby speak of being round. At least, on any occasion of your so speaking, that is something you would then be doing. So if you want to speak of being round, e.g., so as to call something round, or describe it as round, a way of achieving your aim in speaking normal English is to speak the words "is round" (in a suitable construction).

We may put this by saying: there is something the English words "is round" are for; something they are for doing in speaking English. They are for speaking of being round; in suitable constructions, describing things as round, calling things round, etc. That describes something they would do, spoken meaning what they do. There is another way in which those words are for something (in speaking English). At first approximation, they are for describing round things. Roughly that is what they ought to be used for; used otherwise they would not state truth.(11)

If this is what meaning does, we can see why there should be contrasts of the sort with which this section began--both true and false things to be said, e.g., of a given item in using "is round" to mean what it does, in calling the item round. The reason is that the concept of being round does not by itself settle how an object must be to be correctly describable as round. The concept of shape, e.g., does not decide under just what conditions an object must assume a certain shape in order for that to be its shape. If what "is round" means settled such issues where the concept does not, those words would not express it, and so would not mean what they do. There are various reasonable views of what it is for a ball to be a given shape, e.g., round. Words which say it to be round may be rightly understood as doing so on one or another of these views. That makes for various things to be said in so speaking. So if we are to judge whether words were used for what they ought to have been used for--to describe the things they are, or were then, for describing (things rightly and truly described as round), we will have to look beyond merely what the words used mean, on the one hand, and the way the described item in fact is, on the other. That raises a question: to what else must we look? I suggest that to answer that question we must look to what truth is.

The occasion-sensitivity of what words with given meanings say--the plasticity, as it were, of meaning--might well make one wonder just what H-equivalences might say, and what they could possibly say that would make some collection of them exhaust what truth is. But I will not pursue those questions here. Rather, I will stick to the question what it is for words to state truth, and how their doing so depends on what they mean.


Call words which, rightly understood, say, or aim to say, how things are statements. A statement purports to satisfy a certain general condition: however it says things to be, whatever it says to be so, things are that way. Meeting that standard is a condition on its truth. If we ask when a statement would meet the standard, or what it must do to accomplish this, one move would be to break an answer down into particular cases: words which would describe a thing as round should be used in conformity with the rule: use them only to describe round things; words which would describe a fireplace as having a fire in it should be used in conformity with the rule: use them only to describe fireplaces with fires in them; and so on. If, in making a statement, you conform to all such applicable rules, your statement will satisfy the general condition.

One hitch in using this answer may come in deciding which such particular rules do apply, in the circumstances, to your chosen words. But leave that worry aside. What would it be to have conformed to a particular such rule on a particular occasion--to have done what the rule then demanded? For example, you described a ball as round. On one understanding of its being round, it is so; on another it is not. so was the relevant rule (properly understood) conformed to? What decides this? There are three possible replies.

The first is this. Given the way the ball is, nothing about what it is to be round determines, by itself, whether the rule was followed; for nothing about that makes the ball either count as round, or count as not. So nothing decides whether it was followed. So nothing decides either that such words met the general condition or that they did not. There is just no fact of the matter.

If this line of thought is right, it would apply equally to any instance of "The ball is round", spoken of a ball like the ball in the present case. So no such instance would be a case of speaking truth (or falsehood).(12) But that conflicts with the fact that there are various things to be said of a ball in describing it as round; in the case of a ball like the present one, some true and some false. So this response must be rejected.

A second reply is this. Depending on how you view the matter of (a ball's) being round, you might count that ball as (then) round, or you might not. If you do, then count the rule as followed. If you don't, then count it as not. Count the words in question as satisfying the general condition, or as not, accordingly. In so doing, you will be judging the demands of truth just as they ought to be judged.

This cannot be right either. For one might perfectly well sometimes view being round in one way and sometimes in another. There are occasions on, or purposes for, which I would happily describe the ball as round, and others on which I would equally happily say that it wasn't. (If you are too inflexible ever to depart from one view of what it would be to be round, there are still always contrasts between you and other, perhaps equally inflexible, but also equally reasonable, people who take a different view.) The second response would accordingly, commit us to count a given statement sometimes as true, other times as not. We would need also to allow that some people who held the statement true, and others who held it false were equally, and perfectly, right, so in no genuine conflict.

The notion of truth does not work like that. Perhaps being true is an occasion-sensitive notion in the same sense that being round is. So there might be a statement which sometimes counted as true and sometimes did not. I might sometimes speak truth, and other times speak falsehood, in calling given words true. But truth is surely not occasion-sensitive in the way the second response demands. One might consistently hold that what Fiona said in calling the ball round is true, while what Pia said in calling it round is not, all the while with one's own view of what one would then call round oneself. That is the whole point about occasion-sensitivity. That is to say that I might judge words which described the ball as round to do (or fail to do) what they need to do for accuracy independent of whether I, at that particular moment, would count that ball as (then) round. I need not change my view of Pia's words simply on the basis of how I myself, at that moment, would describe the ball. Similarly, if Jones judges Pia's statement true, while Fiona judges it false, Jones and Fiona are not equally right--or right at all--merely on the grounds of the view each takes, at the moment, of what being round would be. Again, this cannot be how accuracy, in the present sense, is to be assessed.

Only one option remains. Whether words were spoken in conformity with a specific particular rule, so whether with all applicable ones--thus whether they count as satisfying the general condition--depends on the speaking in question, and not just on the state of things spoken of. Given the way the ball is, some words which described it as round may count as satisfying the condition, while others which so described it do not. What would count as satisfying it--the way the ball must be--is relative to the speaking, or describing, being assessed. Truth, that is, requires different things of different such describings.

To assess correctly the truth of words which described a ball as round, one needs to take the right view of what being round would come to--one view among many possible ones. The general condition, and the relevant particular rule, yield results only on an understanding, and the right results only on the right understanding. But what view is right for this purpose? That is a matter of the sorts of merits truth involves. For if, as we have seen, the circumstances of the speaking matter to what truth requires, something about what truth is is needed to make such circumstances matter in one way rather than another, so in the way they do. So there must be something about what truth is which fills that role. Truth is sensitive to circumstances; so there is a way words must relate to their circumstances in order to be true.

We have reached a major change of perspective. The first two responses saw truth as the product of exactly two factors: first, which descriptions words gave of things--how they described things as being; second, the way those things in fact are, or were. From this perspective, a judgement of truth is simply one as to the match between these factors. Truth, from our present vantage point, is a product of three factors: the above two; and the character of the occasion on which given words were used so to describe things. For a statement to be true, it must relate in the right way to a pair of factors: the way the world it describes is; and the circumstances of its making. So a substantial question arises. How must the statement relate to the second of these given the first (the way the world is)? Or, the same question put differently, how does the second (the circumstances) matter to how the statement must relate to the first (the way the world described is)? The answer must depend on something about just what sort of an achievement truth is. Without relevant substance to that notion, there is no answer.

The relevant aspects of what truth is, whatever they are, are not fixed by H-equivalences. For those, at best, relate but two factors: descriptions (individuated somehow) for words to give of things; and the world thus described. They are silent on how the circumstances of giving a description matter to its truth; that is, on the relation about which we are now inquiring.

Meaning, while it does not decide when words would be true, does not do nothing either. What words mean imposes definite conditions on their truth, but ones leaving a residual question: what would it be for those conditions to be satisfied in the right way--in the way, that is, that truth, in a given case, demands. It does not contribute to an answer to insist that if words say things to be thus, then things must be thus.

Words which satisfy the general condition, so say things to be only as they are, are ipso facto true. That does not mean that that condition is the only one truth imposes. If it were, what truth required could not choose between standards for assessing satisfaction of that condition, as we have seen it does and must. Rather, given what words mean, so how they describe things, they must satisfy the condition in a certain way: the way things are must count as things fitting those descriptions by those standards in achieving which words would relate to the occasion of their speaking as they must in order to be true. That selects between standards only in so far as there is a way truth demands that words relate to their occasions.


There are two ways of thinking of the result just reached. To approach the first consider the words "Mary had a little lamb", a means in English for speaking of any of several things--eating ovine, or keeping an ovine, for example. English provides a definite stock of readings for these words. On a given occurrence they may bear one or another of these. We might try to use readings as a model of ways of understanding words, thinking of a stock of understandings there are for words to bear. There is, e.g., a stock of understandings of "The ball is round" for each of many understandings of what it would be to be so shaped. One reading words have in English may allow for any of many understandings of a speaking of them. But just as choosing the right reading fixes what words meant as used, so choosing the right understanding fixes what they said. For words which said something in particular, there is always some one understanding in the stock which is the one they bore.

One might further think: which understanding words bear depends on the circumstances of their speaking; when things would be as said to be on a given understanding does not. Understandings, so conceived, extract content from circumstances. Circumstances play no further role in determining conditions for truth. Deflationism, and its use of "proposition", depend on exactly that idea.

Though I think this a bad picture, this is not the occasion to argue the point. Given the picture, the present result takes the form: truth is a determinant of content. Which understanding given words bore is fixed, in part, by how, for them to be true, the description they gave must mesh with the circumstances of its giving.

But here is another picture. Understanding requires sensitivity. Understanding words consists, in part, in sensitivity to how they fit with the circumstances of their speaking. Part of that is sensitivity to how they need to fit in order to be true. So adequate sensitivity requires grasping what truth is, and how that notion applies in particular cases. The sensitivity to words that grasp permits is, to coin a phrase, an Auffassung which is not a Deutung, and may constitute one's understanding of them.

Understanding a statement, we may agree, entails grasping well enough how it, and specifically its truth, are to be assessed. One might suppose that such a grasp, or what follows from it, could be made entirely explicit by some specification of an understanding there is anyway for words to bear, and by what follows from bearing that one (e.g., that they expressed "the proposition thus-and-so"). But perhaps, as Wittgenstein suggested, a proper appreciation of words is not exhaustible in that way. If we do not assume so, then we may see truth, not as selecting from some stock of items each of which relates to the world in an entirely fixed way, but rather as guiding the treatment of words with given content in given circumstances, forming our perceptions of when to be satisfied with the descriptions they gave.


The main result of this essay has been reached. What words mean imposes a condition on their saying, on a speaking, what is so. Different occasions impose different standards for satisfying that condition. Something about what truth is makes occasions matter to such standards. Deflationism cannot recognize such elements in truth. This section explores, tentatively, what they might be.

What truth demands of a description is properly measured by different standards for different givings of it. What fixes the right standards for judging given words? One way to see what might do that work is by looking, not at anything anyone actually said, or might say, but rather just at descriptions to be given, and situations in which they might be. What, in given circumstances, would a given description describe (be for describing)? What is then so describable? Perhaps different circumstances for describing things yield different answers. To see whether that is so, we need to look at pairs of a description and a thing to be described.

For relevant pairs we might consider "The ball is round"(13) and a ball just off the racquet, or "Mary ate a little lamb" and Mary's meal of mioleira. Or, again, "She is locked in" and Pia in a room with the doors locked, the window open, and a long drop; "There's a fire in the fireplace" and a fireplace into which a lit ball of paper has been tossed; "The oven is hot" and an oven at 140[degrees] C; "The cat is puce" and a cat dipped in puce dye.

Let us begin with a lemma. One would not count a 140[degrees](C) oven as hot for purposes of baking pizza. A hot oven, for that purpose, is much hotter. By contrast, for purposes of removing a rack bare-handed a 140[degrees] oven would count as hot. Something may be describable in a given way for some purposes, but not for others. There is such a thing as describability for a purpose. That is the lemma.

Why such facts, and why the particular facts there are? In one case the answer could go like this. One scheme for classifying ovens as hot or not reserves "hot" for the upper part of a normal oven's working range. Ovens being as they are, that makes (say) 220[degrees] a fair dividing line. Nothing makes that scheme right, wrong, or implausible in any absolute sense. Choose it and the description "hot" has a clear use in pizza baking: if an oven is so describable, the pizza may go in. For that purpose, no other scheme recommends itself (to us) so highly; which is enough for that scheme to decide, for that purpose, what is hot.

The culinary facts here are familiar. Familiarity breeds expectations. If an oven is described as hot, and we take that to be for purposes of pizza baking, ceteris paribus we will expect the information we are thus given to be usable as just described.(14) We will expect this in two senses. First, if we trust those words, we will be surprised if the oven is not pizza-ready. Second, we will (normally) expect that much of the words if we are not to count ourselves as misinformed by (in) them, so count them as stating what is incorrect. Our expectations thus form our perception of just what was said in giving that description.

We (often) perceive a description, as given, as for some given purpose(s), where, for those purposes there are uses that its fitting ought to have; uses for the information it ought to have provided. We count the description as giving correct information only where it did what it should have done to serve the relevant purpose(s). Such perceptions thus exhibit two ingredients--a purpose, and uses words would have in serving it--which fix a standard for the truth of the words perceived.

We can move from the lemma to facts about what descriptions would describe on an occasion if we can find occasions on which certain purposes are the ones a given description ought to serve, in the sense that it would be right to expect that much of them. Such is certainly part of our perception of some occasions. We are, e.g., plainly engaged in making pizza. We would expect a description of an oven as hot to be reserved for that purpose. We would, correspondingly, perceive it as one that would give correct information only if it served that purpose. (We do sometimes make judgements as to what would, and would not, count as hot. These are conspicuously not judgements as to what is hot on such and such understanding of being hot. They are non-trivial in ways that the latter sort of judgement is not. Nor are these judgements as to what one could ever correctly call hot. Reminded that they could not be that if correct, we freely appeal to purposes for which something is being called hot; ones we initially took as read.)

It would be useless to be told that the oven is hot if we had no idea of the standard by which that was to be judged. Our perceptions of purposes to be served, and uses words would have in serving them, provide just such a standard. They thus allow descriptions of things to be the sort of good to us they are. These perceptions of occasions are perceptions of what it would be, on them, for a given description to describe truly, or for words which give it to state truth; to provide information which is correct. Their structure thus reveals some ingredients in truth, or what we are prepared to recognize about it. Part of the idea of truth is that a description (of something), to be true, must satisfy a general condition different in kind from conditions to the effect that what is described as thus must be as thus described: it must serve all the purposes that must be served (for truth) on that occasion, by having all the uses it ought in serving them. Part of this idea is that, for a description, and an occasion (on which there are facts as to what that description would describe truly), there are definite purposes truth demands be served, and uses which truth demands the description have in serving them. Another part of the idea consists in particular perceptions of how the concept of truth would apply in particular cases; what ought to count as, or be demanded for, describing truly on the concrete occasions on which the description might be offered--which purposes in particular, the description, so offered, would have to serve if true.

Our perceptions of descriptions on occasions thus link truth with use, that is, when words would state truth with the uses they ought to have to be true. Such is our idea of what truth, or stating it, is. Without that idea, it would be indeterminate when words had said things to be as they are. The idea of truth, we have seen, cannot exhaust itself in a cloud of ideas to the effect that words which say things to be so are true only if things are so.


So far we have not considered how to treat anything anyone actually says. We have only looked at what descriptions would be for on occasions. If, in that way, we can identify ingredients in truth--demands that words serve purposes, be appropriately useful--then, from one point of view, tant mieux. But we began by asking how to judge the truth of what is said. I suggest that the standards for that are fixed by just those ingredients in truth which we have so far identified. To see that we must see past a complicating factor.

To begin, our perceptions of a speaker's words, where we take ourselves to understand them, contain the same ingredients as our perceptions of descriptions on occasions. We see words as taking responsibility for serving certain purposes, in that we will count them as having said what is correct, so true, only where we count these purposes as (adequately) served. Jones says "The oven is hot"; Pia inserts the pizza. On learning, shortly, that the oven is at 140[degrees], she will take herself to have been misinformed. On her understanding of the words, what they said is incorrect. Such are the sorts of understandings we take words to bear. Again, without such perceptions, we would have no standards by which to judge the accuracy of what is said; so there would be no judging it.

Pia may perceive wrongly. She may not grasp what the circumstances of the speaking are, or may miss some crucial fact as to what one could do if the oven were, in the right sense, hot (perhaps it is a trick oven), or may be plain unreasonable. Sometimes, though, we do no such thing. Sometimes we understand words. If Pia did no such thing, then her perception is correct.

What would Pia have to get right about the circumstances for her perception of Jones' words to be right? In the normal case, I suggest, the purposes someone's words should serve to be true are just the purposes the words that person used should serve, in the circumstances in which he spoke, to be true. So what Pia would have to perceive is what the words Jones used (the English "the oven is hot") would be for on that occasion. But so much importance has been attached in the last forty years to a speaker's intentions, and there has been such a strong temptation to suppose that a speaker may fix the standards by which the accuracy of her words is to be judged merely by intending that they express "the proposition (such-and-such)", or by making this evident enough, that it is perhaps worth distinguishing two sorts of case: a default case; and exceptional ones. I will explain that idea with an analogy.

In the normal course of events, in speaking English, we mean our words to mean what they do mean, and rely heavily, and rightly, on that fact, and on the facts as to what words mean in English, for fixing what they did mean (and say) as used by us. Jones asks, "Could I have a brill?", meaning her "brill" to mean brill. Ceteris paribus, that is what it did mean, which means that, in so speaking, she asked for a brill. If you want to know what a brill is, consult a dictionary. For further details, consult an ichthyologist or chef. In matters of what words meant as used, that is the default case.

Exceptionally, Jones may have special ideas about what a brill is. She may think, e.g., that "brill" means bream. Moreover, she may make it clear that she means her "brill" to speak of bream. Perhaps in some such case "brill" as she used it did mean bream, so that what she asked for was a bream, and not a brill. Suppose so. Then there is a default case--what her "brill" meant in the absence of special reasons provided by her for thinking otherwise; and there is the exceptional case where she supplied such special reasons. It remains so that in the default case what we have for working out what her "brill" meant is just what "brill" means. That is often just the way the speaker, and everyone else, wanted things to be.

Similarly, in the normal course of events we just talk, meaning to use descriptions for what they are then for--for what is then so describable--and taking ourselves to be doing just that. We rely on the facts as to what our descriptions would then describe for fixing when things would be as we described them--though, unlike fixing what our words meant, the relevent facts are not to be found in dictionaries.(15) (Just this reliance makes the occasion-sensitivity in what words with given meanings say normally invisible.) That is the default case. In it the ingredients in truth I have identified work in just the way indicated at the start.

Exceptionally, we manifest special intentions as to the understandings our descriptions are to bear, regardless of what they would otherwise be for. Perhaps we sometimes thus get them to bear such understandings. So, e.g., we call the oven hot on some special understanding of its being so, or for some special purpose. Call an oven which is hot on that understanding, or for that purpose, shmot. Then when we said the oven was hot, we described it as shmot. The question now arises: By what standards should the accuracy of a description, on that occasion, of an oven as shmot be judged? What, on that occasion, would be describable as shmot? The story I have told provides the answer.

In any event, however complex exceptional cases may be, default cases are enough to exhibit the ingredients in truth which I meant to highlight here. Such cases are the way we normally get on.


I have discussed ingredients in truth that deflationism misses. Those ingredients are needed if there is to be any assessing whether the world words speak of is the way their truth demands it must be. What words mean constrains the ways they may speak truth, and the truth that they might speak. It imposes some condition on their truth as spoken: the world would have to be, by appropriate standards, as they (in meaning what they do) describe it. The nature of that condition depends on what it would be for a standard to be right. We cannot capture that by speaking of truth in deflationist terms, without referring to occasions for describing, much less saying how what truth is makes them work in fixing standards.

I have not suggested that the work of these ingredients is capturable in a definition; certainly not one which spells out the conditions under which a description would describe truly. In so far as a deflationist's aim is to reject substantive definitions of truth, that aim may stand, for all I have said.

Deflationists are prone to emphasize that the word "true" gets much of its importance as a device for "semantic ascent". With it, we may characterize propositions as true, or not, without actually expressing them. So, for example, we can quantify, as in "Everything Jones said is true". Semantic ascent is important. Deflationists have missed, I think, one of the main reasons why it is. To call Pia's words true is to credit her with having described things correctly, as being as they are. That is not to commit oneself to the (present) describability of things by the descriptions Pia used. Pia, say, described a ball as round. But there are many standards for being round. The right ones for her describing need not be the right ones for some describing I might do now. So to say that what Pia said in describing the ball as round is true is not yet to describe the ball as round, or say it to be round. It is not to assert what one would in so describing it.

As for correctness, Pia's words must be judged by the standards appropriate to the circumstances of her speaking them. Those standards need not be the right ones for judging what I would now say, or judge, in describing the ball as round. Part of grasping what it would be for words to be true, or state truth, is grasping what it would be for particular standards to be the right ones by which to judge whether given words did what truth demands; what might make given standards right for judging the truth of some words, wrong for judging the truth of others which give the same descriptions of the same things.

I need to be able to distance myself from another's words in just the way semantic ascent allows, because different occasions call for different standards by which to judge whether things are thus and so. That exposes a side of truth which deflationism misses.


Frege, Gottlob 1918: "Der Gedanke", as translated in Frege (1977).

--1977: Logical Investigations, translated and edited by Peter Geach, Oxford: Blackwell Publishers.

Horwich, Paul 1990: Truth. Oxford: Blackwell Publishers.

Travis, Charles forthcoming: "Pragmatics", in Wright and Hale forthcoming.

Wright, Crispin and Hale, Bob, ed., forthcoming: A Companion to the Philosophy of Language. Oxford: Blackwell Publishers.

(1) It would be fair to view this essay as no more than a working out of some ideas of J.L. Austin (whom this question might call to mind). I have in mind particularly his "Truth" and "How to Talk". The essay in its present shape evolved with a great deal of very patient help from my colleague Peter Sullivan, and from James Hopkins, Michael Martin, Barry Smith and Joan Weiner.

(2) This also seems to be Frege's view. For, while he held truth to be indefinable, he also said: "The meaning of the word `true' is spelled out in the laws of truth" (Frege 1918, p. 2).

(3) Horwich does say that from any declarative sentence `S' we may form a nominalization, "The proposition that S", and then "denominalize" it back into a new sentence, "The proposition that S is true", where that is, in some sense, equivalent to the original "S". But that tells us nothing about when anything would be true. Nor do we learn more about when words would be true by backing that remark with Horwich's H-equivalences.

(4) Could any deflationist account of truth provide all the H-equivalences that might be needed for this purpose? An interesting question, but not one that I will pursue here.

(5) Ambiguous English words express the proposition they would express on one of their readings in English. Similarly for other languages.

(6) How it is that they mean what they do is assumed not to be a relevant issue.

(7) If differences in truth conditions do not make for different propositions, then you can't treat the truth of propositions in the way Horwich does.

(8) See, however, Travis (forthcoming).

(9) Other words, of course, in speaking of being round, might do so in a derogatory or laudatory way, and so on. That might be part of their meaning what they do. I assume nothing like that is so of "is round"

(10) Perhaps it is still worth saying: it is not for there to be some particular set of things which are "those things the words are true of". We have already seen reason enough why there can be no such set.

(11) To miss the way in which this is only rough is to obliterate just the features of truth I mean to highlight here.

(12) Just in case this seems not really awful, note that the same sort of problem arises again for most ordinary squash balls at rest. Look carefully: there is probably an indentation where the dot is. And suppose squash balls were made with a seam. Again, such things sometimes count as being round, sometimes not. As Descartes pointed out, in our sublunary world, whether something satisfies the criteria for being round always depends on what you are going to count as satisfying them. (I will not pause to argue with people who take the heroic view that that just means that no one ever speaks truth in calling something round.)

(13) For brevity, I quote words to indicate the description they give.

(14) Virtually every sentence, here and henceforth, describing how we would perceive words, calls for a ceteris paribus. Rather than writing it, let us take it as read.

(15) Typically, we think we know what we would describe correctly. What we thus think we know is not whether we would be describing correctly on such and such special understanding of being as we described, much less on that understanding conforming to our beliefs as to what is describable as thus and so. This last would be no genuine knowledge at all.
COPYRIGHT 1996 Oxford University Press
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Travis, Charles
Date:Jul 1, 1996
Previous Article:Making room for the going beyond the call.
Next Article:Backward causation and the direction of causal processes: reply to Dowe.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters