As we saw in the last chapter, Bernard Williams, who views physics as giving us the ultimate metaphysical truth, thinks that ethical statements and normative statements generally have truth values which presuppose the perspective of “some social world or other”. Outright relativists about truth would agree with this conclusion, though not with Bernard Williams’ metaphysics. As we have seen, however, neither physicalism nor relativism has been successful. Attempts to define reference in terms of “causal attachment” have failed, and relativist attempts to define truth, which I discussed in Chapter 4, lead to a tangle of self-contradictory or solipsistic or otherwise unacceptable consequences. Bernard Williams’ move, which is to say that talk of the “content” of a belief (that is, talk of its reference and truth) is itself “perspectival”, lacks any clear sense. This state of affairs has been grist for the mills of deconstruction. Deconstructionists think that the whole idea of representing reality, indeed the whole idea of “reality”, needs to be deconstructed. The tangles in which scientific realists and relativists involve themselves are only manifestation of the incoherence of the very idea of truth. Indeed, Rorty himself, as I have mentioned, has moved from a relativist to a deconstructionist position (Rorty cites the father of deconstruction, Jacques Derrida, as one of his heroes).1 Unlike Rorty, however, Derrida has never suggested that we should try to get along without the notion of truth; the notion may be incoherent, but we have no way of doing without it, according to Derrida.

In the opinion of most analytic philosophers, trying to criticize deconstruction is like trying to have a fistfight with a fog. Indeed, although Derrida does not disdain argument, some of his followers seem to scorn it. The very habit of arguing in a close analytic fashion is seen by many deconstructionists as a sign that one is “out of it”.

But there are arguments to be found in Derrida's writing, although they are usually alluded to rather than actually given. Although Derrida complains about the “assertoric style” in philosophy, his own writing tends to consist of one assertion after another. I shall discuss some of those arguments a little later. But first I want to discuss the views of a philosopher who does delight in argument, an American analytic philosopher, not a French deconstructionist, who has reached conclusions in some ways perilously close to Derrida's: Nelson Goodman.

Nelson Goodman's “Irrealism”

Goodman first announced his “irrealism” in a challenging and unquestionably path-breaking little book provocatively titled Ways of Worldmaking. The title captures two of the most important claims of the book: that we inhabit not one world, but many simultaneously; and that these worlds are worlds of our own making.

The idea of a plurality of worlds is connected with an idea I shall look at shortly, the idea that there is not one unique “right version” of the world, but rather a number of different “right versions” of it. This is an idea that I agree with. An example—mine, not Goodman's—is that there is no unique right version of the relation between ordinary objects (tables and trees and animals) and scientific objects. We can speak as if such ordinary objects were identical with scientific objects, or as if they were distinct from the physical systems which constitute their matter,2 or we can say that which physical system a given common sense object is identical with is to some degree vague (as I would urge) but that there are some physical systems that this chair, or whatever the example may be, is definitely not identical with. Moreover, there are many possible choices as to what we should take the physical system to be, if we want to identify chairs and trees with physical systems: space-time regions (or the gravitational, electromagnetic, and other fields that occupy those regions), or aggregates of portions of the histories of various molecules. Each of these ways of speaking can be formalized, and each of the resulting formalisms3 represents a perfectly admissible way of speaking; but Goodman would say (and I would agree) none of them can claim to be “the way things are independent of experience”. There is no one uniquely true description of reality.

The idea that the facts admit of more than one picture has been around for over a century, however. It is anticipated by Herz's talk of equally good “world pictures” in the introduction to his Principles of Mechanics, and it is referred to by William James.4 Goodman's innovation is to attack the claim that our conceptual schemes are just different “descriptions” of what are in some sense “the same facts”. Goodman regards this idea as empty. For him it is immaterial whether we speak of versions as descriptions of worlds or say that there are no worlds and only versions. What Goodman is adamant about, however, is that if we do choose to speak of worlds as distinct from versions, then we must say that incompatible versions refer to different worlds. It cannot be true of one and the same world that the space-time points are individual things and that they are abstractions. Thus we ought to say—if we keep the concept of a world at all—not that we describe the world (as philosophers) sometimes using a language in which tables and chairs are talked of as aggregates of “time-slices” of molecules and sometimes using a language in which those aggregates of “time-slices” of molecules are regarded as the matter of the tables and chairs (and the matter is spoken of as something distinct from the table or chair); but rather that we sometimes choose to make a world in which tables and chair are aggregates of “time-slices” of molecules and sometimes choose to make a world in which tables and chairs are distinct from those aggregates of “time-slices” of molecules. Goodman confronts us with a choice between saying that there are many worlds or that world-talk is nonsense.5 True to his own pluralism, Goodman sometimes speaks as if there were no world(s) at all, and sometimes speaks as if there were many.

But if we choose to speak of worlds, where do these worlds come from? Goodman's answer is unequivocal: they are made by us. They are not made ex nihilo, but out of previous worlds—or out of previous versions, since the distinction between a world and a version is of no moment. Springing full-blown within contemporary analytic philosophy, a form of idealism as extreme as Hegel's or Fichte's!

In addition to Goodman's more technical arguments, there is a more accessible argument that he first used in commenting on the papers in a symposium on Ways of Worldmaking at the December meeting of the American Philosophical Association in 1979.6 Goodman discussed the question, raised by Israel Scheffler, “Is it a consequence of Goodman's philosophy that we made the stars?” Goodman answered that while there is a sense in which we did not make the stars (we don't make stars in the way in which a brickmaker makes a brick), there is indeed a sense in which we did make the stars. Goodman illustrated this by asking us to consider a constellation, say the Big Dipper. Did we make the Big Dipper? There is an obvious sense in which the answer is no. All right, we didn't make it in the way in which a carpenter makes a table, but did we make it a constellation? Did we make it the Big Dipper? At this point, perhaps many of us might say yes, there is a sense in which we made “it” the Big Dipper. After all, it is hard to think of the fact that a group of stars is a “dipper” as one which is mind independent or language independent. Perhaps we should give Goodman this much, that we didn't “make” the Big Dipper as a carpenter makes a table, but we did make it by constructing a version in which that group of stars is seen as exhibiting a dipper shape, and by giving it a name, thus, as it were, institutionalizing the fact that that group of stars is metaphorically a big dipper. Nowadays, there is a Big Dipper up there in the sky, and we, so to speak, “put” a Big Dipper up there in the sky by constructing that version. But—and Goodman is, of course, waiting for this objection—we didn't make the stars of which that constellation consists. Stars are a “natural kind”, whereas constellations are an “artificial kind”.7

But let us take a look at this so-called natural kind. Natural kinds, when we examine them, almost always turn out to have boundaries which are to some degree arbitrary, even if the degree of arbitrariness is much less than in the case of a completely conventional kind like “constellation”.8 Stars are clouds of glowing gas, glowing because of thermonuclear reactions which are caused by the gravitational field of the star itself, but not every cloud of glowing gas is considered a star; some such clouds fall into other astronomical categories, and some stars do not glow at all. Is it not we who group together all these different objects into a single category “star” with our inclusions and exclusions? It is true that we did not make the stars as a carpenter makes a table, but didn't we, after all, make them stars?

Now Goodman makes a daring extrapolation. He proposes that in the sense illustrated by these examples, the sense in which we “make” certain things the Big Dipper and make certain things stars, there is nothing that we did not make to be what it is. (Theologically, one might say that Goodman makes man the Creator.) If, for example, you say that we didn't make the elementary particles, Goodman can point to the present situation in quantum mechanics and ask whether you really want to view elementary particles as a mind-independent reality.9 It is clear that if we try to beat Goodman at his own game, by trying to name some “mind-independent stuff”, we shall be in deep trouble.

In spite of its elegance, it seems to me that this little argument of Goodman's is easily defused. There is a fundamental difference between the terms “constellation” and “Big Dipper”, on the one hand, and a term like “star” on the other. The extension of the term “Big Dipper” is fixed by linguistic convention. The term applies to a finite group of stars, and one learns which stars are in the group and how they are arranged when one learns the meaning of the term. In this respect, “Big Dipper” is a typical proper name.

We know which stars belong to the Big Dipper by knowing what it is we call “the Big Dipper”. I would not say that it is “analytic” that the Big Dipper contains all of those stars, because if one of those stars “went out” or was totally removed by aliens with vast superscientific powers we would undoubtedly go on speaking of the Big Dipper and just say that the Big Dipper didn't have as many stars as it used to have. In the same way, we will continue to refer to John Smith as “John Smith” even if he loses his hair. (If a new star appeared “in” the Big Dipper, however, it would not automatically count as a part of the Big Dipper. Whether it came to count as a part of the Big Dipper would depend entirely on subsequent linguistic practice; which stars are part of the Big Dipper is a question for an anthropologist or a linguist, not a question for an astrophysicist.)

In contrast to the term “Big Dipper”, the term “star” has an extension that cannot be fixed by giving a list. And no particular object is in the extension of “star” simply by virtue of being called a star; it might be crazy to doubt that Sirius is really a star, but someone who thought that Sirius is really a giant light bulb or a glowing spaceship wouldn't thereby show an inability to use “star” in the way in which someone who doubted that that constellation is really the Big Dipper would show an inability to use “Big Dipper”.

In these respects, the term “constellation” lies somewhere in between “Big Dipper” and “star”. If we discovered that all the stars in the Big Dipper are really giant fakes installed to fool us by those superscientific aliens (giant light bulbs in the sky, so to speak), we would say “they aren't really stars”, but we wouldn't say “that isn't really the Big Dipper”. Would we cease to regard the Big Dipper as a constellation? Perhaps we would, but I am completely unsure.

The upshot is very simple. One perfectly good answer to Goodman's rhetorical question “Can you tell me something that we didn't make?” is that we didn't make Sirius a star. Not only didn't we make Sirius a star in the sense in which a carpenter makes a table, we didn't make it a star. Our ancestors and our contemporaries (including astrophysicists), in shaping and creating our language, created the concept star, with its partly conventional boundaries, with its partly indeterminate boundaries, and so on. And that concept applies to Sirius. The fact that the concept star has conventional elements doesn't mean that we make it the case that that concept applies to any particular thing, in the way in which we made it the case that the concept “Big Dipper” applies to a particular group of stars.

The concept bachelor is far more strongly conventional than the concept star, and that concept applies to Joseph Ullian, but our linguistic practices didn't make Joe a bachelor. (They did make him “Joe Ullian”.)10 General names like “star” and “bachelor” are very different from proper names like “the Big Dipper” and “Joe Ullian”, and Goodman's argument depends upon our not noticing the difference.

Irrealism and Conceptual Relativity

Goodman has far more serious arguments for his theses, and these arguments contain many real insights (even if they lead to wrong conclusions). We have to discuss them with some delicacy. Those arguments depend on a phenomenon I have called “conceptual relativity”.11 Here is an example. Points in space (or nowadays one often refers instead to points in space-time) can be regarded as concrete12 particulars of which space consists (the ultimate parts of space) or, alternatively, as “mere limits”.13 Geometrical discourse can be adequately formalized from either point of view;14 so can all of physics. And whether formalized or left unformalized, both ways of speaking will do perfectly well for all the purposes of geometry and physics.

Goodman regards these two versions of space-time theory as “incompatible”. At the same time, he regards them as both right. And since incompatible versions cannot be true of the same world, he concludes that they are true of different worlds “if true of any”.

One well-known objection to views like Goodman's has been advanced by Donald Davidson15 (and Quine has recently been converted to it as well). Davidson's argument is as quick and dirty as Goodman's: Davidson simply accepts Goodman's view that the two versions are incompatible, points out that according to standard logic incompatible statements cannot both be true, and concludes that it is unintelligible to maintain that both versions are true. Even if both versions are equally good for practical purposes, I cannot say that they are both true, according to Davidson and Quine. Quine does say16 that I may pick one of these versions some of the time (say Mondays, Wednesdays, and Fridays?) and the other one at other times, but at any given time I must say that the version I am using then is true and the version I am not using then is false, on pain of self-contradiction.

Goodman's view, as we have seen, is to say that logic does indeed tell us that incompatible statements cannot both be true of the same world, but, in his view, the equal “rightness” of both of these incompatible versions shows that they are true of different worlds. Davidson does not explicitly discuss this move, but Quine rejects it on the ground that to split reality into a number of different worlds is to violate the principle of parsimony. (The possibility of saying that there are no worlds at all, that is, no extralinguistic reality, is one that Quine apparently does not deem worthy of discussion.)

Goodman and Davidson seem to me to be making the same mistake—although, as often happens in philosophy, it leads them into opposite camps. Davidson and Goodman both accept without question the idea that statements which appear to be incompatible, taken according to their surface grammar, really are incompatible, even in cases like these. If the sentence, “points are mere limits” is a contrary of the sentence “points are not limits but parts of space”, even when the first sentence occurs in a systematic scheme for describing physical reality and the second occurs in another systematic scheme for describing physical reality even though the two schemes are in practice thoroughly equivalent,17 then we are in trouble indeed. But the whole point of saying that the two schemes are in practice thoroughly equivalent is that, for from leading us to incompatible predictions or incompatible actions, it makes no difference to our predictions or actions which of the two schemes we use. Nor are the two schemes “equivalent” only in the weak sense of what is sometimes called “empirical equivalence” (that is, leading to the same predictions); rather, each sentence in one of them, say the scheme in which points are concrete particulars, can be correlated in an effective way with a “translation” in the other scheme, and the sentence and its translation will be used to describe the same states of affairs.

In saying that they are used to describe the same states of affairs, I am not introducing a transcendent ontology of states of affairs. By a “state of affairs” I mean something like a particle's being at a point, or a place X's being between a place Y and a place Z; in short, I assume a familiar language to be already in place. I am not saying that Noumenal Reality consists of states of affairs. (I could have spoken of “situations”, or “physical events”, or in many other ways. The relation of language to the world is also something that can be described in more than one way.) Nor am I assuming a one-to-one correspondence between sentences and states of affairs, thus populating the world with “sentence-shaped objects”, in a phrase taken from Richard Rorty. The whole point of what I just said is that very different sentences can describe the very same state of affairs. In short, what I meant by a “state of affairs” when I said that a sentence in one formalization of physical geometry and its “translation” in the alternative formalization will be used to describe the same “states of affairs” is just what anyone would mean by that phrase who was not giving it a metaphysical emphasis.

That the sentences in one such scheme that we use in practice can be correlated with the sentences in another such scheme was pointed out long ago by Goodman himself. The nature of such correlations was studied by Goodman,18 who referred to this kind of correlation as “extensional isomorphism”. Am I am saying, then, that the sentence “points are mere limits” and the sentence “points are not mere limits but parts of space” actually have the same meaning? Not at all. Some sentences function virtually as tautologies within their own “version”. If we identify points with limits by definition, then, in our version, “points are mere limits” will be a conventional truth. And we do not generally translate such conventional truths in one version into the other version (although we could; we could, for example, translate every quasi-logical truth in the one version by any fixed tautology in the other). But the more interesting, less conventional, truths in the one version—say, theorems of ordinary Euclidean geometry, or the statement that there is a particle with a certain mass at a certain place—can be “translated” from one version into another version; these are the statements for which we provide correlates.

Should we say that any such statement in the version in which points are treated as concrete particulars—say, the statement that between any two points on a line there is a third point—has exactly the same meaning as its “translation” into the version in which points are identified with limits (say, with convergent sets of concentric spheres)? I would say that in the context of real-life physics and real-life mathematics it makes no difference which of these two ways one talks and thinks. I am saying that if a sentence in one version is true in that version, then its correlate in the other version is true in the other version. But to ask if these two sentences have the same meaning is to try to force the ordinary-language notion of meaning to do a job for which it was never designed.

The phenomenon of conceptual relativity is a mind-boggling one. To suppose that questions like “Do S₁ and S₂ have the same meaning or a different meaning?” make any sense in this case seems to me precisely the assumption that we should not make. The sentence “Between any two points on a line there is a third point” and its “translation” into the version in which points are identified with convergent sets of concentric spheres have the same truth conditions in the sense that they are mathematically equivalent. The answer to the question “Do the two sentences have the same meaning?” is that the ordinary notion of meaning simply crumbles in the face of such a question. It was never meant to do that job.19

Now to the question of “incompatibility”, which exercises both Goodman and Davidson so: “point”, “line”, “limit”, and so on are used in different ways in the two versions. To say that the sentence “points are convergent sets of concentric spheres”,20 as used in the one version, is incompatible with “points are not sets but individuals”, as used in the other version, is much too simple. Rather than conclude with Goodman that either there is no world at all or else we live in more than one world, or to conclude with Davidson that the phenomenon of equivalent descriptions, which we have recognized in science since the end of the nineteenth century, somehow involves a logical contradiction, we should simply give up the idea that the sentences we have been discussing preserve something called their “meaning” when we go from one such version into another such version.

Am I not, then, saying that the sentence has a different meaning in the two versions? (If a sentence doesn't preserve its meaning, it must change it, right?) I repeat that the answer is that the notion of “meaning”, and the ordinary practices of translation and paraphrase to which it is linked, crumble when confronted with such cases. We can say that the words “point”, “limit”, and so forth have different “uses” in these two versions, if we like. In view of that difference in use, one should not treat a sentence in one version as though it contradicted what the same physicists or the same mathematicians might say on another day when they are employing the other version. But whether such a change of use is or is not a change of “meaning” is not a question that need have an answer.

The Significance of Conceptual Relativity

The significance of conceptual relativity might come out more clearly if we consider a somewhat different case. In The Many Faces of Realism I described in detail a case in which the same situation, in a perfectly commonsensical sense of “the same situation”, can be described as involving entirely different numbers and kinds of objects (colored “atoms” alone, versus colored atoms plus “aggregates” of atoms). If you have a world in which there are two black “atoms” and one red one, you can either say that there are three objects (the atoms), or that there are seven objects (the atoms and the various aggregates of two or more atoms). How many objects are there “really” in such a world? I suggest that either way of describing it is equally “true”. The idea that “object” has some sense which is independent of how we are counting objects and what we are counting as an “object” in a given situation is an illusion. I do not mean by this that there “really” are “aggregates”, and there really are atoms and there really are sets and there really are numbers, and so on, and it is just that sometimes “object” does not refer to “all objects”. I mean that the metaphysical notion of “all objects” has no sense.

Again, in quantum mechanics, any two states of a system can be in a “superposition”; that is to say, any particular state of a system, involving having a particular number of particles or a particular energy or a particular momentum, can be represented by a kind of “vector” in an abstract space, and the superposition of any two such states can be represented by forming a vector sum. These vector sums are sometimes classically very difficult to interpret: what do we make of a state in which the answer to the question “How many electrons are there in this box?” is “Well, there is a superposition of there being three electrons in the box and there being seventeen”? But we can represent such unthinkable states mathematically, and we know how to derive predictions and formulate explanations using them. This principle of superposition applies to field theory as well as to particle theory; the “field states” of the quantum field theorist are not the field states of the classical field theorist; they are typically superpositions of the field states of the classical theory. We may say, then (and here I leave out entirely the puzzling role of the observer in quantum mechanics), that from the point of view of quantum mechanics, the world consists of fields in “funny” states. But—and this was the discovery of Richard Feynman—it is also possible to think in a very different way. We can think of the world as consisting of particles (although we have to vastly increase the number of particles we postulate in order to carry this through) and we can think of any situation that we describe in field physics as a superposition of an infinite number of different particle situations. In short, there are two different ways of thinking in quantum field theory. In one way of thinking, the way the physicist thinks when performing the usual field calculations, the system is in a superposition of field states. In the other way of thinking, the way of thinking when drawing “Feynman diagrams”, the system is in a superposition of particle states. In short, the system may be thought of as consisting either of fields or of particles, but it cannot be thought of as consisting of either classical fields or classical particles.

Consider a given physical system which the physicist represents twice over, once in the language of fields and once in the language of particles (say, by drawing Feynman diagrams). What I am saying is that this is a real system, and that these are two legitimate ways of talking about that real system. The fact that the real system allows itself to be talked about in these two very different ways does not mean either that there is no real physical system being talked about, or that there are two different physical systems in two different Goodmanian worlds being talked about.

The point is even clearer in the case of the first example, the example in which “the same situation” was described as involving entirely different numbers and kinds of objects. It is absolutely clear, it seems to me, that the two descriptions are descriptions of one and the same world, not two different worlds.

Part of Goodman's challenge—as it was part of the challenge of German idealists like Hegel and Fichte in the beginning of the nineteenth century—is to say, “Well, if you say that these two ways of talking are both descriptions of the same reality, then describe that reality as it is apart from those ways of talking.” But why should one suppose that reality can be described independent of our descriptions? And why should the fact that reality cannot be described independent of our descriptions lead us to suppose that there are only the descriptions? After all, according to our descriptions themselves, the word “quark” is one thing and a quark is quite a different thing.

Nevertheless, the phenomenon of conceptual relativity does have real philosophical importance. As long as we think of the world as consisting of objects and properties in some one, philosophically preferred sense of “object” and “property”—as long as we think that reality itself, if viewed with enough metaphysical seriousness, will determine for us how we are to use the words “object” and “property”—then we will not see how the number and kind of objects and their properties can vary from one correct description of a situation21 to another correct description of that same situation. Although our sentences do “correspond to reality” in the sense of describing it, they are not simply copies of reality. To revert for a second to Bernard Williams’ book, the idea that some descriptions are “descriptions of reality as it is independent of perspective” is a chimera. Our language cannot be divided up into two parts, a part that describes the world “as it is anyway” and a part that describes our conceptual contribution. This does not mean that reality is hidden or noumenal; it simply means that you can't describe the world without describing it.

Irrealism and Deconstruction

We may now begin to appraise the frequently heard claim that “the problematique of representation has collapsed”.22 What people who talk like this mean is that the notion of reference to an objective world has collapsed. Goodman's work has the virtue of setting forth an argument for this position which is much clearer than any that one can extract from the work of Derrida. Goodman's argument does, I have claimed, destroy one traditional version of “realism”, the version I like to call metaphysical realism.23 According to that version, the notions of an object and a property each have just one philosophically serious “meaning”, and the world divides itself up into objects and properties in one definite unique way. This is the myth of the ready-made world. (This is also one form of what Derrida would call the “metaphysics of presence”.)

The myth of the ready-made world is a myth which has become linked with a number of other ideas. For example, there is the expectation, which we have already encountered, that the objects and properties of which the world (“in itself”) consists are the objects and properties of “finished science”, and there is the tendency—it is much more than a tendency in fact—to forget that the principle of bivalence of classical logic is simply a useful idealization, which is not conformed to fully and cannot be conformed to fully by any actual language, natural or artificial, that human beings could possibly use.

But the collapse of a certain picture of the world, and of the conceptions of representation and truth that went with that picture of the world, is very far from being a collapse of the notions of representation and truth. To identify the collapse of one philosophical picture of representation with the collapse of the idea that we represent things that we did not bring into existence is, quite simply, dotty. Deconstructionists are right in claiming that a certain philosophical tradition is bankrupt; but to identify that metaphysical tradition with our lives and our language is to give metaphysics an altogether exaggerated importance. For deconstructionists, metaphysics was the basis of our entire culture, the pedestal on which it all rested; if the pedestal has broken, the entire culture must have collapsed—indeed, our whole language must lie in ruins. But of course we can and do make sense of the idea of a reality we did not make, even though we cannot make sense of the idea of a reality that is “present” in the metaphysical sense of dictating its own unique description. As we saw, seemingly incompatible words may actually describe the same situation or event or the same physical system.

At this point we run into another source of contemporary philosophical scepticism. This source is the doctrine of incommensurability. Although that doctrine has been associated in recent years with the writings of Thomas Kuhn, it appeared in French thought decades before Kuhn's work. A version of it appeared in Ferdinand Saussure's Cours de linquistique generale,24 a work whose influence on both structuralist and poststructuralist French philosophy is unquestionable. Saussure's route to incommensurability was the following: along with other linguists of his time (for example, the Prague school) Saussure learned that the basic phonetic units of language, the phonemes, were not themselves identifiable in terms of their physical features. One can characterize a phoneme in a language only by contrast with the other phonemes in the language. It is the whole system of contrasts that determines what the phonemes of a given language are. Phonemes are not “sounds” in the physicist's sense of sound. One cannot say that the English phoneme p is the “same” as the German phoneme p; they simply belong to different systems of contrasts. Saussure assumed that something similar would have to be true of the semantic units of the language; that is, he assumed that the meanings expressible in a language could be characterized only by the ways in which they contrasted or failed to contrast with other meanings available in the same language. The idea of describing the language as a system of differences (a system of available contrasts) was to be extended from phonemics to semantics.

But different languages do not, in fact, provide the same semantical contrasts. A language which recognizes only four fundamental colors provides a different system of contrasts from one which provides seven fundamental colors, for example. The line of thinking that Saussure had embarked on leads fairly Quickly to the conclusion that meanings are parochial to languages (and from here it is not far to the thought that they may be parochial to individual “texts”). No two languages ever express the same meanings; no meaning can ever be expressed in more than one language (or even text). The very notion of a sign's meaning, as something separable from the sign, collapses.

In an interview with Kristeva,25 Derrida makes clear his wholehearted acceptance of this line of thinking; he criticizes Saussure only for not going further and abandoning talk of “signs” altogether (since the notion of a “signified” independent of the system of signs has collapsed):

To take only one example, one could show that a semiology of the Saussurean type has had a double role. On the one hand, an absolutely decisive role:

1. It has marked, against the tradition, that the signified is inseparable from the signifier, that the signified and the signifier are two sides of one and the same coin. Saussure even purposefully refused to have this opposition or this “two-sided unity” conform to the relationship between soul and body, as had always been done. “This two-sided unity has often been compared to the unity of the human person, composed of a body and a soul. The comparison is hardly satisfactory.”26
2. By emphasizing the differential and formal characteristics of semiological functioning, by showing that it is “impossible for sound, the material element, itself to belong to language”, and that “in its essence it [the linguistic signifier] is not at all phonic”;27 by desubstantializing both the signified content and the “expressive substance”—which therefore is no longer in a privileged or exclusive way phonic—by making linguistics a division of general semiology—Saussure powerfully contributed to turning against the metaphysical tradition the concept of the sign that he borrowed from it.

Derrida fails to notice that a Utopian project lay behind Saussure's way of thinking. The hope was for a strictly scientific account of meaning, one that would exactly parallel the structure of the newly emergent phonemics. Since that hope has collapsed (it was hardly coherent to begin with), we are not forced to the bizarre view that no one can understand any language but his or her own ideolect. Nor does Derrida himself go so far; like Quine, after having denounced the notion of meaning-preserving translation, he recognizes the indispensability of translation in practice, although in a very guarded way:

In the limits to which it is possible, or at least appears possible, translation practices the difference between signified and signifier.28 But if this difference is never pure, no more so is translation, and for the notion of translation we would have to substitute the notion of transformation: a regulated transformation of one language by another, of one text by another. We will never have, and in fact never had, to do with some “transport” of pure signifieds from one language to another, or within one and the same language, that the signifying instrument would leave virgin and untouched. (Positions, p. 20; emphasis in the original)

The alternative to Saussure's view is to keep the notion of “sameness of meaning” while recognizing that it is not to be interpreted as the self-identity of objects called “meanings” or “signifieds”. When two uses of words may be regarded as “the same” and when they may be regarded as “not the same” is not a matter of some clean mathematical relation of equivalence or non-equivalence between two systems of contrasts. If people inquire about the meaning of something that someone says, we generally have some idea as to why they are asking and what they are going to do with the answer. Given the context and the interests of the people involved, we can usually come up with a pretty good answer. Can it be that in Derrida's use of Saussure we see some of the same mistakes that are made by American analytic philosophers like Jerry Fodor? Fodor would, of course, reject the idea, which is implicit in the argument I have cited, that sameness of meaning makes strict sense only in the impossible case in which the two languages or texts in question are isomorphic. But the fact that Derrida takes this idea seriously, while not even considering the possibility that the kind of “sameness of meaning” we seek in translation might be an interest-relative (but still quite real) relation, one which involves a normative judgment, a judgment as to what is reasonable in the particular case, does remind me of Fodor's scientism.

I don't want to claim that the two factors that I have discussed —the way in which metaphysical realism has gotten itself into trouble in the twentieth century, and the way in which the doctrine of incommensurability of different languages and even different texts has come to seem coercive to Derrida—are the sole reasons which shape Derrida's eventual deconstructionist position. Certainly there are many other influences, including Heidegger, Marx, Freud, and Nietzsche. But when one looks for arguments in and around Derrida's writings to support the radical claims that he repeats over and over, I think one finds that they are related to, on the one hand, Goodman's irrealism29 and, on the other hand, Saussure's form of the doctrine of incommensurability. While those doctrines are well worth reflecting on—they do show much that is of interest—they do not justify the extreme philosophical radicalisms of either Goodman or Derrida.

Differences between Goodman and Derrida

Although Goodman and Derrida might both be described as “irrealists”, the philosophical morals that they draw from their respective irrealisms are quite different. Although Goodman sees difficulties with the notion of truth, he never proposes that we should give it up. Instead, he proposes that we should widen the range of philosophical discussion. Instead of talking exclusively or primarily about language, about versions that consist of statements, we should also consider other “versions” of the world, such as paintings, musical compositions, and so on. (According to Goodman, all works of art function semantically and constitute versions/worlds.) Truth is a predicate which we apply only to statements, and statements occur only in verbal versions of the world, but non-verbal versions can also contribute to understanding and can be right or wrong. Goodman is fully aware that there are no necessary and sufficient conditions that we are at present able to state for either rightness or truth—there is certainly no “algorithm” for either rightness or truth. Moreover, he is aware that any even partial and vague standards that a philosopher might propose will always be controversial. But neither the lack of an algorithm nor the controversial character of such general statements as we are able to make should occasion dismay. Even if we do not have a general characterization of rightness, we have partial characterizations of certain kinds of rightness, as Goodman points out. For deductive validity, we have long had such a partial characterization. (That it is only partial is shown by the Gödel incompleteness theorem.) For inductive validity, Goodman has himself proposed the beginnings of an account, although he is well aware that that account does not constitute a formal inductive logic, in the sense of Rudolf Carnap. Generally Goodman's attitude towards the lack of “standards” for rightness or truth is that it is the job of the philosopher to try to devise standards, if not for truth simpliciter or for rightness simpliciter, then for rightness and truth in various areas. If those standards are not an algorithm, they can at least be the beginnings of an account. If we don't yet have even the beginnings of an account in many areas, then that shows that there is a great deal of work for philosophers to do. Goodman describes himself as a “constructionalist”; he constantly stresses the idea that the lack of preexisting standards is a challenge to philosophers, rather than a reason for dismay.

How should philosophers go about constructing standards for different kinds of rightness? They should look at what we already believe about various cases, and try to formulate standards that agree with those beliefs. But they should not be the slaves of their beliefs. As we try to develop standards, we often find that the very activity of trying to formulate principles leads us to change our view about particular cases. We have thus to aim at “delicate mutual adjustment” of standards and individual cases to one another, hoping for something like a Rawlsian “reflective equilibrium” at the end. But what if our own reflective equilibrium is not regarded as a reflective equilibrium by others? Then, Goodman says, we must simply try to “sell” what seems right to us. The criterion of rightness (in philosophy or anywhere else) cannot be universal consent.30 Goodman is not afraid of incompleteness, and he is not afraid of making normative judgments.

Derrida's attitudes are much harder to make out. Although this is certainly a misinterpretation,31 his attacks on the “logocentricism” of Western culture have been interpreted by some of his more left-wing followers as licensing an all-out rejection of the very idea of rational justification. These followers interpret Derrida as teaching that logic and standards of rightness are themselves repressive. Freeing ourselves from capitalism is seen as requiring that we free ourselves from notions like rightness and truth. Goodman would be seen as a hopeless reactionary by these people.

In certain ways, one can understand the reasons for this interpretation. Traditional beliefs include much that is repressive (think of traditional beliefs about various races, about women, about workers, about gays). Our “standards” require not only rational reconstruction but criticism. But criticism requires argument, not the abandonment of argument. The view that all the left has to do is tear down what is, and not discuss what might replace it, is the most dangerous politics of all, and one that could easily be borrowed by the extreme right.

Derrida himself is not guilty of this kind of thinking. He has movingly replied to the charge of nihilism: “We can easily see on which side obscurantism and nihilism are lurking when on occasion great professors or representatives of prestigious institutions lose all sense of proportion and control; on such occasions they forget the principles that they claim to defend in their work and suddenly begin to heap insults, to say whatever comes into their heads on the subject of texts that they obviously have never opened, or that they have encountered through mediocre journalism that in other circumstances they would pretend to scorn”.32

Yet the fact remains that the thrust of Derrida's work is so negative, so lacking in any sense of what and how we should construct, politically or otherwise, that it is difficult to exonerate him complete from responsibility for the effect of his teaching. He himself does not exonerate Nietzsche completely:

I do not wish to “clear” its author and neutralize or defuse either what might be troublesome in it for democratic pedagogy or “leftist” politics, or what served as “language” for the most sinister rallying cries of National Socialism. On the contrary, the greatest indecency is de rigueur in this place. One may even wonder why it is not enough to say: “Nietzsche did not think that,” “he did not want that,” or “he would have vomited this,” that there is falsification of the legacy and interpretative mystification going on here. One may wonder how and why what is so naively called a falsification was possible (one can't falsify just anything), how and why the “same” words and the “same” statements—if they are indeed the same—might several times be made to serve certain meanings and certain contexts that are said to be different, even incompatible.33

Commenting on this passage, Richard Bernstein has written:

I am not suggesting that Derrida's texts are the occasion for “the most sinister rallying cries”. It is difficult to imagine any texts which are more anti-authoritarian and subversive for any (and all) “true believers”. But I am asking whether the signatory of these texts bears some responsibility for their reception. If the desire to write “is the desire to perfect a program or a matrix having the greatest potential variability, undecidability, plurivocality, et cetera, so that each time something returns it will be as different as possible”, then doesn't the signatory bear some “responsibility” for the divergent and incompatible ways in which the texts are read and heard. One may wonder “how and why” the texts signed by J.D. can be read (or heard) as being nihilistic, obscurantist, selfindulgent logorrhea and (and I have argued) passionate, political, subversive, committed to opening the spaces of differance and respecting what is irreducibly other. What is it about the texts of Derrida that allows for, indeed invites, this double reading? After all, “one can't falsify just anything”.34

I would suggest that the bind Derrida is in is the bind those will find themselves in who do not want to be “irresponsible”, but who “problematize” the notions of reason and truth themselves, by teaching that, even if they are indispensable, nevertheless they “retain us in the logocentric circle”, they have “collapsed”, and so forth.

The problem is that notwithstanding certain moments of argument, the thrust of Derrida's writing is that the notions of “justification”, “good reason”, “warrant”, and the like are primarily repressive gestures. And that view is dangerous because it provides aid and comfort for extremists (especially extremists of a romantic bent) of all kinds, both left and right. The twentieth century has witnessed horrible events, and the extreme left and the extreme right are both responsible for its horrors. Today, as we face the twenty-first century, our task is not to repeat the mistakes of the twentieth century. Thinking of reason as just a repressive notion is certainly not going to help us to do that.

Derrida, I repeat, is not an extremist. His own political pronouncements are, in my view, generally admirable. But the philosophical irresponsibility of one decade can become the real-world political tragedy of a few decades later. And deconstruction without reconstruction is irresponsibility.