I want the game
for my wonderful essay on utter nonsense:
Wittgenstein's Tractatus Logico-Philosophicus is a baffling little book. Through 526 numbered propositions, Wittgenstein addresses the most crucial questions in logic and language, and—as he put it in the Preface to the text—finds the “final solution” to all philosophical problems.1 In the seventy-five years since its initial publication, there have been many interpretations of the text. The three main readings of the text are, what I will call: the positivist reading, the standard reading, and the resolute reading. The chronologically earliest of these three is the interpretation put forth by members of the Vienna Circle, many of whom were directly influenced by Wittgenstein's work—namely Schlick, Waissmann, Carnap and Ayer. They credit Wittgenstein with, as Carnap puts it, developing “the logical and epistemological conception which underlies” the verification principle—the position that any nonanalytic proposition has cognitive meaning if and only if it can be in principle sense-verified.2 Of course, this interpretation of the text was not without its problems. Central to it is the so-called “picture theory of meaning,” which specifies that the isomorphism between the proposition and the world—in which simple logical objects in the world map on one-to-one with simple names in the proposition—as Wittgenstein's sole criteria for a proposition to attain sense. They interpret this as a sort of verificationist position, and through this the Vienna Circle understood the Tractatus in harmony with their own views. The positivists read this as a verificationist position since, for Wittgenstein, a meaningful proposition can be decomposed by logical analysis down to elementary propositions, which 1 Ludwig Wittgenstein, Tractatus Logico-Philosophicus tr. David Pears and Brian McGuinness (New York: Routledge, 1961). Hereafter cited as parenthetical references to the specific numbered proposition. All references will be to the Pears-McGuinness translation, unless otherwise noted. 2 Carnap, Logical Syntax of Language, 65 Brommage, “An Essay on Nonsense” 2 the logical empiricists incorrectly identified as sense-datum sentences. This hasty equivocation between Wittgenstein's elementary propositions and sense-datum sentences cannot be textually maintained.3 This is not by any means the only interpretive error which the Vienna Circle made, but since my goal is not to demonstrate what is largely already accepted, I will not rehearse all the arguments here. Let me summarize by saying that the positivist reading took Wittgenstein's remarks on the nature of the proposition as primary importance, and downgraded certain other remarks—such as those concerning “the mystical.”4 I will have more to say about why Carnap's theory of meaning cannot be conflated with Wittgenstein's theory of meaning below. Following the renunciation of the positivist interpretation, a new interpretation emerged which took the distinction between saying and showing as the central doctrine of the text. The standard reading, as I will call it, is the position that most specialists and nearly all non-specialists understand to be Wittgenstein's early philosophy. These commentators argue, although it is true that for Wittgenstein that the only truly meaningful statements are those which are fact-stating, that is not to exclude certain other “truths”—which can only be shown [zeigen], although they cannot properly be said [sagen]. They cite two separate categories of sentences that do not have meaning to support their case. Wittgenstein uses the term nonsense [unsinnich] for those that the positivists might say have no cognitive meaning, such as statements with undefined terms. But he also uses a separate term— senseless [sinnlos]—for a class of certain propositions which, although they are not fact-stating, are not nonsense either. Tautologies and contradictions, he tells us “lack sense,” but they “are not, however, nonsensical” (4.461-4.4611). 3 See G. E. M. Anscombe, An Introducton to Wittgenstein's Tractatus (St. Augustine Press, 2001) p. 26-28 4 Carnap tells us that “[e]arlier, when we were reading Wittgenstein's book in the Circle, I had erroneously believed that his attitude towards metaphysics was similar to ours. I had not paid sufficient attention to the statements in his book about the mystical, because his feelings and thoughts in this area were too divergent from mine. Only personal contact with him helped me see more clearly his attitude at this point. I had the impression that his ambivalence with respect to metaphysics was only a special aspect of a more basic internal conflict in his personality from which he suffered deeply and painfully.” Rudolf Carnap “Intellectual Autobiography,” in The Philosophy of Rudolf Carnap edited by Paul Arthur Schlipp. (La Salle, IL: Open Court Press, 1963), 27. Brommage, “An Essay on Nonsense” 3 For the standard readings of the Tractatus, Wittgenstein meant that are some truths which, although not fact-stating and thus unable to be said, might still somehow be shown. That is, they rely upon the structure of language to show us ineffable features of language and the world. Standard readers take various things discussed in the Tractatus to populate this category of redeemable nonsense, including but not limited to the propositions of logic, and the “form of representation” between language and the world—and in stronger cases, even existential, ethical and metaphysical insights—all of which Wittgenstein alludes to (although he cannot express outright) throughout the course of the text. The second category of nonsense however, for which Wittgenstein uses the term unsinnig, is just gibberish or plain nonsense. Included in this category are statements which contain terms whose linguistic role has not been defined within a sentence. The nonsense utterance “Ab sur ah,” on the other hand, is just pure nonsense; it means nothing to us.5 Recently a new interpretation has emerged, initially put forth by Cora Diamond, and expanded considerably by James Conant, Michael Kremer, and others. These interpreters argue that the distinction between saying and showing is not intended by Wittgenstein as a legitimate doctrine of the text; but rather it, and many of the other supposed doctrines of the text, should be thrown away with the ladder. This includes the distinction between saying and showing, and thus the parallel distinction between senselessness and nonsense. Indeed, the resolute readers accuse both the standard and positivist readings of falling into the same error, of attributing to Wittgenstein a “substantial” view of nonsense. Instead they interpret Wittgenstein as only putting forth a single, austere notion of nonsense: on the one hand, there's sense, and on the other there's just plain nonsense—and there's nothing in between. That is to say, anything which does not enter into the category of fact-stating propositions are strictly speaking nonsensical. In what follows, I intend to argue in favor of the doctrine of austere nonsense as the resolute readers understand it. I will do so by arguing that Wittgenstein adopts a 5 The example is taken from Lectures 1932-5, ed. A. Ambrose (Prometheus 2001), p. 64 Brommage, “An Essay on Nonsense” 4 Fregean, not a Russellian, theory of meaning. Failure to understand the indebtedness of Wittgenstein on Frege, specifically his context principle from the Grundlagen der Arithmetik (which appears as 3.3 in the TLP) leads to misunderstanding of Wittgenstein's understanding of linguistic meaning. In the first section, I will explain the received view of Wittgenstein's theory of meaning when integrated with the say/show distinction, and point to some difficulties with it. Many commentators which correspond to the “standard reading,” as I have been calling it, believe that for the early Wittgenstein there are a series of 'ineffable 'truths' which transcend the bounds of language. In the second section, I will expand upon what a Fregean theory of meaning constitutes, with special attention to the concept/object distinction and the context principle from his Grundlagen der Arithmetik. In the third section I will contrast this with Russell's theory of meaning during his logical atomist period. I believe that the theory of meaning that most commentators take to be Wittgenstein's is really Russell's. And in the fourth section, I will provide evidence that Wittgenstein's view of nonsense in the Tractatus was Fregean. I conclude by pointing to some benefits that follow from the interpretation that I will provide. I should note that although I am sympathetic to the resolute reading of the Tractatus, I do not buy it wholesale. There are some aspects in some stronger proponents of this interpretation (such as Rupert Read) that I feel are both textually and contextually mistaken. Although the austere notion of nonsense is not one of them. My own reading of the text “splits the difference,” as it were, between the standard and resolute readings. I accept the austere notion of nonsense while maintaining a minimal doctrine of showing. Although the argument for this interpretation of the text this lies outside the scope of this paper. I One of the most characteristic features of the resolute reading is its starting place. At 6.54 Brommage, “An Essay on Nonsense” 5 Wittgenstein writes, in conclusion of the text: My propositions serve as elucidations in the following way: anyone who understands me eventually recognizes them as nonsensical [unsinnig], when he has used them—as steps—to climb up beyond them. (He must, so to speak, throw away the ladder after he has climbed up it.) He must transcend these propositions, and then he will see the world aright. There are two important features of this passage which deserve mention. First, Wittgenstein specifically distances understanding him (the author), from understanding the text. And secondly, when one properly understands the author, the reader comes to recognize that the remarks in the text are nonsense. That is, not senseless [sinnloss] in that they surpass the representational capacity of language, but rather plain nonsense [unsinnig]. 6 The resolute interpretation takes him seriously at 6.54, and understands the text exactly as he demands—as nonsense. The resolute readers take this remark as their hermeneutic anchor, insisting that the proper interpretation of the text is to understand that the propositions of the Tractatus themselves are in fact nonsense, and thus in the end must be thrown away. But several commentators who endorse the standard reading have tried to downplay or interpret this remark at 6.54 such that it denies the obvious. Many standard readers contend that the propositions of the Tractatus are literally meaningless, because in the course of the book he needed to speak about things which are, by the lesson of the text, ineffable. Such a defense is put forth by Max Black in his Commentary on Wittgenstein's Tractatus.7 Black rejects the suggestion that the remark at 6.54 implies that the remarks in the book are nonsense, noting that Wittgenstein's conclusion in 6.54 is “profoundly unsatisfactory” and suggests that “this ladder need not be thrown away.”8 Arguing that while most of Wittgenstein's remarks in the text cannot be literally meaningful since they cannot be modeled on the 6 Standard readers will be tempted to interpret this remark at 6.54 to mean that his propositions are nonsense in the weaker connotation; that is, although they have no sense, they can still inform us about 'truths' which transcend the proper bounds of language. Indeed, the original translation of the text by Ogden and Ramsey used the word 'senseless' in this passage, instead of 'nonsense'—leading an entire generation of readers to gloss over the significance of this remark. However, even a cursory look at the original German of 6.54 above reveals that the word used is not sinnlos, meaning 'without sense' or 'senseless,' but rather unsinnig—'nonsense'! 7 Max Black, A Commentary on Wittgenstein's Tractatus (Ithaca, NY: Cornell University Press, 1964). 8 Ibid., 376-7. Brommage, “An Essay on Nonsense” 6 picture theory, he suggests that Wittgenstein really meant to say they are “senseless,” and not “nonsense.” Black argues that, “f we take 'seriously' the conclusion that metaphysics is nonsense, it seems our confidence that we succeed in understanding the book must have been a self-delusion. It is one thing to say we must throw away the ladder away after we have used it; it is another to maintain that there never was a ladder at all.”9 Black seeks to put forth a line of defense which thwarts the temptation to read Wittgenstein as equating “'nonsense' with gibberish,” and “what 'cannot be said' with what cannot be rationally communicated.”10 Instead he understands 6.54 to mean that his propositions are meaningless since they illicitly employ what Wittgenstein calls “internal” or “formal” concepts, rather than and legitimate “external” or “material” concepts. Wittgenstein explicitly calls the former “pseudo-concepts,” as they stand for terms which belong to the symbolism itself. The terms 'object,' 'complex,' 'fact,' 'function,' and 'number,' are examples that he provides—all of which Wittgenstein thinks are properly symbolized through an adequate Begriffsschrift, but are properly ineffable in language (4.1272). It is nonsense by Wittgenstein's own philosophy to say, for example, 'There are objects.' Although this is constructed in a similar grammatical style to the expression 'There are books,' it is only by analogy that we feel the former expression is legitimate.11 Objects in the symbolism espoused by the Tractatus instead can only be shown in symbolic form by the use of a bound variable. Although the latter statement actually says something about the world, the former is an attempt to treat the concept which is itself internal to the symbolism as if it were a sign within it. In this way a formal concept will 'show' us its logical role, through the way it is symbolized. But they are not properly concepts which can be spoken of, but rather show themselves in the use of language. Black explains that “[l]ogical syntax cannot include such remarks as 'The world is everything 9 Black, Commentary, 379. 10 Ibid., 379. 11 “So one cannot say, for example, 'There are objects', as one might say, 'There are books'” (4.1272). Brommage, “An Essay on Nonsense” 7 that is the case' or 'The world is the totality of facts, not of things.'” Since these remarks use the pseudo-concepts 'world,' and 'fact,' which are “used by Wittgenstein in invented or stretched senses.”12 In each case of these remarks, terms which stand for formal concepts are employed, thereby giving us the illusion that they have legitimate usages in language. Sentences which use these words, just as with tautologies and contradictions, are not fact-stating. They are neither bivalent, nor do they express propositions according to the picture theory of meaning. On Black's rendering, when we reach 6.54 at the end of the book, we come to realize the point he was trying to make concerning internal and external concepts. But in order to make such a distinction it required him to appear to speak about things which could not be spoken about; throwing the ladder away is constitutive of this realization. For Black, “Wittgenstein is trying out a new way of looking a the world, which forces him to twist and bend language to the expression of his new thoughts . . . A negative metaphysics, such as that of the Tractatus, has its own rules of procedure: the ladder must be used before it can be thrown away.”13 Peter Hacker argues that Black's line of reasoning is “mistaken,”14 but nonetheless mounts a line of defense similar to Black's. Hacker rightfully points out that Black confuses the categories of senselessness and nonsense. Senseless propositions, like tautologies and contradictions, can have legitimate uses outside language—but do not have meaning in a fact-stating way. Sentences which use formal concepts, unlike tautologies and contradictions, are always illegitimate for Wittgenstein. Hacker readily admits that Wittgenstein's propositions, such as the ones quoted above, are nonsense— but not pure nonsense. Rather they are what he calls “illuminating nonsense.”15 He tells us that it “will guide the attentive reader to apprehend what is shown by other propositions which do not purport to be philosophical; moreover it will intimate, to those who grasp what is meant, its own illegitimacy.”16 12 Black, Commentary, 382. The quoted expressions are remarks 1 and 1.1, respectively. 13 Ibid., 386. 14 P. M. S. Hacker, Insight and Illusion: Themes in the Philosophy of Wittgenstein Revised edition (New York: Clarendon Press, 1986), 25. 15 Ibid., 18. 16 Ibid., 18. Brommage, “An Essay on Nonsense” 8 When we reach the conclusion, we must realize that they are nonsense. But they still do have the power to enlighten us about the structure of logic and language, which is the conclusion we come to understand after the ladder has been tossed. One can't help but be struck by the vagueness of Hacker's explanation here. He might respond to this charge by pointing out that, of course, it must be vague—he's trying to express what is ineffable, after all. But this points to exactly what is so philosophically troublesome about this doctrine attributed to Wittgenstein. Both Black and Hacker seem to presuppose, strangely, that there really is a sense to be understood underlying nonsensical sentences. As Wittgenstein reminds us later in the Philosophical Investigations, “When a sentence is called senseless, it is not as it were its sense that is senseless.”17 And this is reflected in how daunting a task it is to write about their commitment to communicate the ineffable. To do so, commentators must use troublesome phrases such as “illuminating nonsense,” and assorted grammatical tricks in order to somehow talk about what Wittgenstein supposedly thinks is shown but cannot be spoken. While they try to claim that these are 'truths' of a sort, they are not 'truths' in the sense that they correspond to an existing fact; and then somehow they attempt to pull an ontological moral out of the Tractatus, to articulate these pseudo-truths on Wittgenstein's behalf—all the while denying that they are really saying anything. As James Conant points out: Some of [the standard readers] think it helps to call what is at issue here a 'fact,' placing the word 'fact' in quotes to mark the difference between such facts and garden-variety facts. Facts are what can be spoken of, what can be depicted by meaningful propositions. What is at issue here is not that sort of fact, but rather something much deeper. Something? Well not some thing. It is something much deeper than a fact or a thing. It is like a fact, in that we can, in our thought about it, get it right or wrong; but it lies at too deep a level—deeper than any ordinary fact—to be a mere fact.18 This type of “backpedaling” as he calls it, is common to ineffability interpretations of the Tractatus. And yet still the standard readers have composed volumes on a philosophical doctrine which they claim 17 PI §500 18 James Conant, “Why Worry about the Tractatus?” in Post-Analytic Tractatus ed. Barry Stocker (Burlington, VT: Ashgate, 2004), 169-70. Brommage, “An Essay on Nonsense” 9 cannot be articulated, and yet still think that we can understand what Wittgenstein meant to say all along. Not only is the category of “illuminating nonsense” philosophically troubling and difficult to understand, but it also leaves Wittgenstein in a strange position. Throughout the literature, standard readers cite many things which can only be shown by language—which by the lesson of the text itself, are ineffable. Yet, nonetheless, they feel confident interpreting Wittgenstein not only as attempting to speak about these things of which one cannot speak through the course of the book, but moreover they assume that—somehow—these lessons are able to be communicated to the reader. II It has been suggested by P. T. Geach that Wittgenstein's distinction between what can be said and what can only be shown was borrowed from Frege.19 Geach cites Frege's argument in “Concept and Object”20 as the single most important influence on Wittgenstein's distinction between saying and showing; thus, he argues, it is by looking to Frege that we can attain a clear view of this distinction. Frege's article is a response to an attack on his Grundlagen der Arithmetik by Benno Kerry, but also contains important insights into Frege's formula language of Begriffsschrift itself. Kerry attacks Frege's sharp separation of concepts and objects by arguing that sentences such as “The concept 'horse' is easily obtained,” show that concepts can also serve the function of objects in sentences.21 Kerry was, of course, missing Frege's point. But it is instructive to understand why Kerry's criticism is a misunderstanding of Frege's point. 19 Peter Geach, “Saying and Showing in Frege and Wittgenstein” in Essays in Honor of G. H. von Wright, ed. Jakko Hintikka, Acta Philosophica Fennica 28 (1975): 54-70 20 Frege “On Concept and Object” in The Frege Reader ed. Michael Beaney (Malden, MA: Blackwell, 1997), 181-193. 21 Frege in fact allows for a given word to serve both as a concept and as an object in differing contexts. The example that he provides in the Grundlagen is the word “moon.” In the sentence 'The moon is bright tonight,' it is used as an object word; in the sentence 'Titan is a moon,' it is used as a concept word. See Gottlob Frege, Foundations of Arithmetic trans. J. L. Austin (Oxford: Basil Blackwell, 1980), §64. Brommage, “An Essay on Nonsense” 10 In the Introduction to the Grundlagen, Frege lays out three principles to which he would adhere in the text: [1.] There must be a sharp separation of the psychological from the logical, the subjective from the objective; [2.] The meaning of word must be asked for in the context of a proposition, not in isolation; [3.] The distinction between concept and object must be kept in mind.22 He comments on this third principle, that “it is an mere illusion to suppose that a concept can be made into an object without altering it.”23 Adding the words 'the concept' in front of an concept term does not somehow preform some sort of semantic magic trick, which changes a concept term into an object term. To the contrary, it is not the way the words are compounded that is important, but instead the logical role of the term in a Begriffsschrift formula that determines whether it is a concept or object.24 It is only in their combinatorial role—that is, only in the context of a proposition—that its logical role can be truly understood. For Frege, the role of a object term is fundamentally different than a concept, since objects or arguments are saturated, whereas concepts are not; Frege tells us that the concept expression is “unsaturated,” but are only functions from arguments to truth-values. Perhaps an example might help. Consider the concept expression, “Socrates is a man.” This meaningful expression is a function of two parts. First, the object (”Socrates”) and the concept under which the object falls (“x is a man”). It is essential to “x is a man” that it has not meaning on its own. Rather it is “unsaturated” because it can 'absorb' an object term. There is no 'gap' to be filled in the object. It is an argument that 'fits gaps' in the functions, the concept expressions. By substituting objects, the completed formula then points (or refers) to the True when, e.g., 'Martin Heidegger' is plugged in for the variable; likewise it would point 22 Frege, The Foundations of Arithmetic, x. See also The Frege Reader, op. cit., 90 23 Ibid. 24 As Kwasi Wiredu has pointed out to me, there is a deeper, ontological distinction that Frege may have omitted between concept and object. On this reading, 'the concept “horse”' is indeed a concept. But further, this would also imply that concepts (or 'senses,' as in his “On Sense and Reference”) also must be objects for Frege. Brommage, “An Essay on Nonsense” 11 to the False when 'Hannah Arendt' is. The proper logical role of an unsaturated predicate term, as a concept, is to take an object in its variable place; the proper logical role of an object is to fill that variable place in the concept term. Although concepts can sometimes serve as argument places for other concepts, this amounts to nothing more than confusing a first-order predicate under which objects fall, and a second-level concept under which first-level objects fall.25,26 Thus Frege concludes, with a seemingly paradoxical air, that “The concept 'horse' is not a concept.”27 To Frege, this is not paradoxical at all, but merely a consequence of the 'grammar' of these terms. “The concept 'horse'” is not a predicate at all—that is to say, not a concept; but rather, it is a name (since objects can only be named). Frege shows that Kerry's alleged counterexample amounts to nothing more than “an awkwardness of language,”28 which shows the insufficiencies of ordinary language and the necessity of the clarity generated by the Begriffsschrift notation. Of Kerry's criticism of Frege's alleged 'definition' of concepts and objects, Frege says that it was not intended to be a definition at all: . . . my explanation is not meant as a proper definition, One cannot require that everything be defined, any more than one cannot require that a chemist decompose every substance. What is simple cannot be decomposed, and what is logically simple cannot have a proper definition . . . On the introduction of a name for something logically simple, a definition is not possible; there is nothing for it but to lead the reader or hearer, by means of hints, to understand the words as it intended.29 Although such a distinction can be drawn perspicuously in an adequate Begriffsschrift sentence, it cannot be adequately drawn in language. For in order to speak about the distinctive logical roles of concepts and objects, one would need to use these same linguistic items to talk about them—thus 25 Beaney, A Frege Reader, 189 26 For example, “All humans are mortal.” Here this is a relation between concepts “x is human” and “x is mortal,” under the second-order relation of universality: “All.” 27 Ibid., 185. 28 Ibid., 185. 29 Ibid., 182. Brommage, “An Essay on Nonsense” 12 employing them as both as explanans and explanandum. Simply put, if we were to use words which denote concepts in language, then one must refer to them—and that would be to treat them as objects, not as concepts, in order to draw the distinction between then in language. In an adequate formula language of logic, no such confusions could be possible, since it can be shown that such sentences contain logical infelicities, even if one is unable to say so in any precise way. Put differently, logical analysis will clarify semantic confusions, or reveal certain constructions as outright nonsensical in ordinary discourse. Indeed, such a notation should allow one to see that Kerry's example is no example at all: it illicitly uses a certain type of word in a logically inappropriate role, because he mistakes the role taken by 'the concept “horse”' to be that of a concept. The resulting confusion of Kerry is not solely due to failing to pay attention to Frege's foundational third principle, the distinction between concept and object. Kerry not only transgresses Frege's third principle, but in his confusion he also sins against the second. For Geach's Wittgenstein, the distinction between saying and showing allows one, from a scrutiny of the structure of language, to gain insight into the nature of the world. This is because Kerry attempted to set the role of “the concept 'horse'” outside the context of a proposition, and then smuggled it back into the proposition with its logical role as a concept and not an object pre-determined —a violation of the context principle.30 This is because, for Frege, the distinction is an ontological one —which, he tells us, is “founded deep in the nature of things.”31 Thus in a way to understanding this distinction drawn in language is to understand something about the world. “Paradoxical as is the doctrine of aspects of reality that come out but cannot be propositionally expressed,” Geach tells us, “it is hard to see any viable alternative to it so long as we confine ourselves to philosophy of logic: and in 30 And as Frege points out, a transgression of the second principle is also a violation of the first. “If the second principle is not observed,” he tells us, “then one is almost forced to take as the meaning of words mental images or acts of individual mind, and thereby to offend against the first as well.” So Kerry can be interpreted as rejecting the distinction between concept and object, by simultaneously violating the context principle and lapsing into psychologism. 31 Gottlob Frege, “Function and Concept,” in Translations from the Philosophical Writings of Gottlob Frege, ed. P. Geach and M. Black (Oxford: Basil Blackwell, 1970), 42-3 Brommage, “An Essay on Nonsense” 13 this domain Wittgenstein revised Frege's views without unfaithfulness to Frege's spirit.”32 Yet these ineffable 'truths,' to Geach and other standard readers, do somehow succeed 'gesturing at' these features of reality nonetheless. Although they are not strictly speaking 'truths' in a fact-stating way, and although language has no capacity to represent them, standard readers believe them to be conveyed by language. Just as the distinction between concept and object is real, although it transcends the representational capacity of language—so too, they think, Wittgenstein meant to hint at the fact that there is more to reality than language can represent. III The Fregean understanding of nonsense can be revealingly illuminated in contrast with the philosophy of Bertrand Russell. Russell's theory of meaning, as I hope to show in this section, is fundamentally different than Frege's—and, as I will argue, also Wittgenstein's. Recall that Frege held a contextual theory of meaning, that the meaning of a term is only given in the context of a given proposition. Russell instead held a compositional theory of meaning. A compositional theory of meaning holds that terms have meaning in isolation, and each contribute to the proposition when patched together in meaningful ways. A compositional theory of meaning is important to Russell (especially in his atomist period) because the thought of the proposition as a truth-function of its simple logical constituents. But he is by no means its only proponent throughout 20th century philosophy of language.33 For a compositional theorist like Russell, meaning is primarily a property of terms; the meaning of a proposition is merely derivative of the meaning of its simple parts. Since a compositional theory of meaning specifies that the meaning of an expression is merely a combination of the meanings of the terms that are contained it it, compositional nonsense results from an inappropriate combination 32 Geach, “Saying and Showing,” 68. 33 A compositional theory of meaning has, for example, been endorsed perhaps most prominently by Donald Davidson, who thought that every syntactic item should be associated with a clause of a T-schema with an operator in the semantics which describes how the constituents of an expression should be syntactically built. Brommage, “An Essay on Nonsense” 14 of meanings, perhaps due to an impermissible 'clash' between the logical categories of two terms. A contextualist theory of meaning takes the meaning of a proposition or judgment to be primary, and says that the meaning of the terms is derivative of their use in propositions. Thus, contextualist nonsense is generated when there is no role defined for a given word, relative to the context in which the term appears in a sentence. In the Lectures on Philosophy of Logical Atomism, which date to 1917-1918, Russell explicitly denies a version of the context principle. He tells us that “propositions concerning the particular are not necessary to be known in order that you may know what the particular itself is. It is rather the other way around. In order to understand the proposition in which the name of a particular34 occurs, you must already be acquainted with that particular.”35 The reason he provides for this is the usual argument for compositionalism: “that one can understand a proposition when you understand the words of which it is composed even though you never heard the proposition before.”36 It is a deeply held conviction for Russell that the logical syntactic role of a term plays a role in how it can combine with other terms to form significant propositions. In essence, this is the central idea of the theory of types. And it should not escape notice that Russell's ramified theory of types in Principia Mathematica was robustly criticized in the Tractatus; Wittgenstein refers to it as an “error” (3.331), and he thinks that it “vanishes” (3.333) in a proper logical symbolism. In the 1924 essay “Logical Atomism,” he describes the importance of his type-theoretic analysis: n its technical form, this doctrine [of types] states merely that a word or symbol may form part of a significant proposition, and in this sense have meaning, without being always able to be substituted for another word or symbol in the same or some other proposition without producing nonsense. State in this way, the doctrine may seem like a truism. 'Brutus killed Caesar' is significant, 'Killed killed Caesar' is nonsense, so that we cannot replace 'Brutus' by 'killed,' although both words have meaning. This is plain common sense, but 34 By “particular” here, he means “terms of relations in atomic facts.” See PLA p. 60. 35 PLA p. 65 36 PLA p. 53 Brommage, “An Essay on Nonsense” 15 unfortunately almost all philosophy consists in an attempt to forget it.37 What Russell is working with here is what might be called a subsentential notion of meaning. That is, he requires that semantic content is 'pre-loaded' into a term (here, to indicate in which role a term can be substituted) so as to avoid nonsense. Contrast this with a sentential notion of meaning, in which the semantic properties of a term are really only defined in the context of a proposition. Compare this with Wittgenstein's explicit statement on the same problem, in a 1913 letter to Russell: I have changed my views on “atomic” complexes: I now think that Qualities, Relations, (like Love), etc. are all copulae! That means for instance analyze a subject-predicate prop[osition], say “Socrates is human” into “Socrates” and “something is human” (which I think is not complex). The reason for this, I s a very fundamental one: I think that there cannot be different Types of things! In other words whatever can be symbolized by a proper name must belong to one type. And further: every theory of types must be rendered superfluous by a proper theory of symbolism . . . What I am most certain of is not however the correctness of my analysis, but of the fact that all theory of types must be done away with by a theory of symbolism showing that what seem to be different kinds of things are symbolized by different kinds of symbols which cannot possibly be substituted in one another's places.38 This is a legitimate change in Wittgenstein's thinking about matters of logic. And I believe it is not coincidental that this letter to Russell was written within days of traveling to Jena to visit Frege in December of 1912. Briefly put, what Wittgenstein is endorsing here is the Fregean distinction between concept and object. Wittgenstein's solution offered in this letter is largely due to way that his expression is symbolized. If we analyze the expression into the following: , (∃x)φx Rather than in Russell's symbolism: 37 PLA p. 170 38 Wittgenstein in Cambridge, 38. Brommage, “An Essay on Nonsense” 16 , ∃(x y) ε1 (x, y) It is clear that Russell needed the theory of types in order to prevent substituting the wrong way around, due to the defects in his logical symbolism. Both “Humanity” and “Socrates” are simple terms which will both 'fit' on either side of the monadic relation of predication. Given this, Russell needed to rely upon extra-logical semantic principles such that the combination “Humanity is Socrates” could not be allowed.39 Rather, Wittgenstein's insight was to change the symbolism such that the syntactic role of the term was perspicuous from the propositional sign. That is, as he says, “what seem to be different kinds of things are symbolized by different kinds of symbols which cannot possibly be substituted in one another's places.” The compositional theory of meaning was not unique to Russell. In fact, many of those who followed in the analytic tradition adopted Russell's theory of meaning—including many commentators on Wittgenstein, as mentioned above. Perhaps the most notorious example is Rudolf Carnap, who can also be seen as adopting Russell's compositional theory of meaning, and the notion that subsentential semantic content exists for a term. In “Elimination of Metaphysics through Logical Analysis of Language,” Carnap describes two types of nonsense pseudo-statements: “either they contain a word which is erroneously believed to have meaning, or the constituent words are meaningful, yet are put together in a counter-syntactic way.”40 It is of course, this latter type of nonsense which Russell endorses, and it is my aim to show that Wittgenstein rejects. Carnap provides several examples of compositional nonsense in this essay. (1) Caesar is and (2) Caesar is a prime number 39 Russell has reasons for this. He thought that universals or concepts must be able to stand in the subject places of propositions. See The Principles of Mathematics, §49. 40 Carnap, “Elimination of Metaphysics” p. 64 Brommage, “An Essay on Nonsense” 17 He glosses on these examples that “word sequence (1) is formed countersyntactically; the rules of syntax require that the third position be occupied, not by a conjunction, but by a predicate . . . But, now, word sequence (2) is likewise syntactically correct, for it has the same grammatical form as the sentence just mentioned. Nevertheless, (2) is meaningless. 'Prime number' is a predicate of numbers; it can neither be affirmed nor denied of a person.”41 This reveals a difficult interpretive question, which remains unanswered by Russell, Carnap, and the rest of the standard readers of the Tractatus: that is, how are we to discern that the word 'Caesar' as it appears in this sentence is the name for a general, and not an adjective to describe salad? Or even, just a secret-code that means “37.” If the proposition truly is nonsense, then we should not be able to discern the linguistic roles of its constituent parts? Simply put, nonsensical utterances have no grammatical or logical parts. Cora Diamond makes this argument in a celebrated series of papers concerning the role of nonsense in Frege and Wittgenstein. She makes two key points about nonsense sentences: first, that nonsense sentences have no discernible logical parts; and secondly, that pieces of nonsense cannot play a role in a sensible sentence. As to the first point, we are no more able to understand the logical role of the items in the sentence 'Socrates is identical,' than we are able to fix the logical role of a term in plain nonsense utterances, such as 'Ab sur ah.' But on the Fregean theory of meaning, there is no such thing as fixing the role of words in nonsense sentences because, simply put, nonsense has no sense. If we cannot diagnose the meaning of a word from within the context of an expression, then it has no meaning. It is only in the context of a sensible sentence, for both Frege and Wittgenstein, that a word has a logical role. To think otherwise is to draw an analogy to other similarly constructed sentences, such as 'Socrates is happy,' and to presume that only a term with that logical role that can satisfy the construction 'Socrates is ( ).' On the compositional view, it is nonsense because the relational term 'identical' does not fit where the adjective should be that nonsense results. That is, 41 Ibid., p. 67-8 Brommage, “An Essay on Nonsense” 18 “Socrates is identical” is constructed in a similar fashion to “Socrates is wise,” but trying to force a relational term where a predicate should be. But how, one might wonder, can this be recognized as a predicate rather than a relational term which accidentally omits the second term in the relation, modeled on something like “Socrates and the teacher of Plato are identical”? To recognize the word 'identical' as a relational term in “Socrates is identical” would then be to fix the logical role of the word outside the context of the sentence, and then smuggle it back into the nonsense utterance with its logical role fixed. But words can only have meaning in the context of a well-formed sentence. Just as Benno Kerry misunderstood Frege's distinction between concept and object due to the failure to understand the logical role of the word within the context of the sentence, so too do the standard readers fall prey to the same error in advocating a compositional theory of nonsense. In this sense, as Diamond puts it elsewhere, “[w]e are all Benno Kerrys through and through.”42 IV Wittgenstein did adopt a lot from Frege, but not what Geach thinks. Rather, it is the context principle (and, to a lesser extent, the distinction between concept and object) and that plays a fundamental role in all of Wittgenstein's philosophy—both early and late. Wittgenstein only adopts a version of Frege's context principle, he copies is outright. “Only propositions have sense; only in the context of a proposition does a name have meaning,” reads proposition 3.3 of the Tractatus. Compare what Frege says in the Grundlagen: “it is only in the context of a proposition that words have any meaning.”43 Ignoring the difference between “word” and “name” here,44 the German text of the second 42 Diamond, “Throwing Away the Ladder,” 184. 43 Frege, Foundations §62. J. L. Austin's translation contains a stylistic difference, but the German is almost identical. 44 For Frege a proposition was a name for its truth value. Not so for Wittgenstein. Brommage, “An Essay on Nonsense” 19 clause is nearly identical. In this section, I hope to demonstrate that it is Frege's theory of meaning that Wittgenstein adopts, not Russell's. But first, a bit of background is in order. The distinction between sign and symbol is crucial for understanding Wittgenstein's views on meaning, and indeed—failure to provide meaning. Wittgenstein defines a symbol as “any part of a proposition that characterizes its sense” (3.31) and a sign as “what can be perceived of a symbol” (3.32). In a set of lectures from 1930, Wittgenstein is recorded as saying that “[T]he sign is the written scratch or the noise.”45 Thus, the sign is the written or spoken form of the word or phrase. They are waves in the air, or empty marks on a page. The symbol, on the other hand, is the sign used with meaning. In 3.326 Wittgenstein tells us that “in order to recognize a symbol by its sign we must observe how it is used with a sense.” That is, words are empty words or marks on a page until one assigns them a sense, a meaning. It is important that he specifies their use is what creates the possibility of understanding the meaning of a sign. Likewise, as Wittgenstein puts it in 3.327, “A sign does not determine a logical form unless it is taken together with its logic-syntactic application.” Wittgenstein is clear that “in logical syntax the meaning of a sign should never play a role” (3.33).46 But it is important to note that in natural language, it often happens that “the same word has different modes of signification—and so belongs to different symbols—or that two words that have different modes of signification are employed in propositions in what is superficially the same way” (3.323). Wittgenstein provides the example of the verb 'is.' [T]he word 'is' figures as the copula, as a sign for identity, and as an expression for existence . . . (In the proposition, 'Green is green”—where the first word is the proper name of a person and the last an adjective—these words do not merely have different meanings: they are different symbols.) (TLP 3.323) The phrase 'Green is green' can have three separate interpretations, in all of which the word 'is' has a 45 Wittgenstein, Lectures Wittgenstein's Lectures, Cambridge 1930-1932, ed. D. Lee (Oxford: Blackwell, 1980), p. 26 46 This remark is linked to his criticism of Russell's theory of types, which adjoins this proposition in 3.331-3.334/ Brommage, “An Essay on Nonsense” 20 different logical function. This is revealed to us by making it explicit through an adequate symbolism. In the first case, we might read it as an instance of the predicative 'is,' a relation between concepts and objects, with the first occurrence of the word 'green' as a proper name (Mr. Green), with the latter as an adjective: 'Mr. Green is green.' Secondly, as a relation of objects, an 'is' of identity with proper names on either side: 'Mr. Green is Mr. Green.' Lastly, we can read the 'is' in this sentence as a claim to coextensionality, a relation between concepts: 'Every green thing is colored green.' In each case, the translation into symbolism (where the predicate Gx is 'x is green' and 'g' stands for the proper name 'Mr. Green') shows that each variant uses a different sense of the word 'is.'47 Likwewise, it is possible that different signs stand for the same symbol. For example, in Russell's symbolism: p ~ ~p ~ ~ ~ ~ p . . . There are, in fact, an infinite number of propositional signs that stand for the proposition p—that is, prefixed with any even number of negations. With that background in place, there are several remarks in the Tractatus that are especially applicable to Wittgenstein's understanding of nonsense, particularly 5.473-4733: Logic must look after itself. If a sign is possible, then it is also capable of signifying. Whatever is possible in logic is also permitted. (The reason why 'Socrates is identical' means nothing is that there is no property called 'identical.' The proposition is nonsensical because we have failed to make an arbitrary determination, and not because the symbol, in itself, would be illegitimate. (5.473) Frege says any legitimately constructed proposition must have sense. 47 This point is originally made by James Conant in “Two Conceptions of Die Uberwindung der Metaphysik: Carnap and the Early Wittgenstein” in Wittgenstein in America, edited by Timothy McCarthy and Peter Winch (New York: Oxford University Press, 2001), 27-8. Brommage, “An Essay on Nonsense” 21 And I say that any possible proposition is legitimately constructed, and, if it has no sense, that can only be because we have failed to give meaning to some of its constituents. (Even if we think we have done so). Thus the reason why 'Socrates is identical' says nothing is that we have not given any adjectival meaning to the word 'identical.' For when it appears as a sign for identity, it symbolizes in an entirely different way—the signifying relation is a different one—therefore the symbols are entirely different in the two cases: the two symbols have only the sign in common, and that is an accident. (5.4733) In 5.473 Wittgenstein uses the example “Socrates is identical” to help us understand what he means by nonsense. He tells us that the reason it is nonsense “is that there is no property called 'identical.' The proposition is nonsensical because we have failed to make an arbitrary determination, and not because the symbol, in itself, would be illegitimate” (5.473). There us nothing wrong with treating 'identical' as an adjective—perhaps in the case of a new slang word. The sign itself is arbitrary. Here we just have not identified this sign with the new “mode of signification,” as he puts it (3.322). Rather, “the reason why 'Socrates is identical' says nothing is that we have not given any adjectival meaning to the word 'identical.'” The text at 5.4733 above seems very clear-cut, and from it one is led to believe that the expression “Socrates is identical” is nonsense because it illicitly uses a two-place relational term where a one-place predicate should be used. The statement is nonsense, on this reading, because there are not two terms to fill this relation. It is this type of nonsense that the standard readers rely upon to “show” us something about the logical forms of the terms, and why such a combination is linguistically or logically impermissible. It might be said that this type of illicit grammatical combination shows us something about the logical syntax of language itself—the logical role of the relation “identical” as a two place predicate term. In this way, standard readers will consider this piece of nonsense to be a case of illuminating nonsense, which indicates to us that the term 'identical' only has meaning as a relation, not as a singular predicate term. Due to a violation of logical grammar, this statement means nothing; Brommage, “An Essay on Nonsense” 22 but insofar as it is nonsense, it can “show” us something about the impossibility of such a linguistic combination. But recall that Wittgenstein tells us “any possible proposition is legitimately constructed, and, if it has no sense, that can only be because we have failed to give a meaning to some of its constituents” (5.4733).48 It is important not to lose sight of the word “only” in this passage. This indicates that there are not two ways that nonsense is generated. Rather, there is only one type of nonsense. Russell holds that that nonsense is generated through an impermissible combination of symbols (not signs). Put in Tractarian terms, this type of nonsense emerges when a proposition is composed of symbols whose logical (or perhaps, ontological) categories of the symbols clash in their composition. In Tractarian terms, contextualist nonsense emerges when one cannot understand the symbol from the sign—just like with 'green is green' above. It is the former type of nonsense that is rejected by Wittgenstein here. Nonsense results from a failure to recognize the symbol in the sign, not from the result of taking two logically incompatible words and trying to out them together. The obvious reason why this is the case is Wittgenstein's commitment to the context principle. If a word is understood outside of the context of the proposition, there is no clear way to discern exactly what logical role it has. Confusions result (as they did with Benno Kerry) when one attempts to fix the logical role of a given word outside of the context of the proposition.49 There is no set logical role of a word outside the context of its significant use in a sentence. Thus it is not the case, for either Frege or for Wittgenstein, that there is a special type of nonsense generated by category errors; that is, when otherwise meaningful signs are put together in illicit ways. Rather it is only when a sign has no 48 Emphasis mine. 49 In fact, Frege is careful to point out that concept words can, at times, be used as object words; an example that Frege provides is “moon,” such as in the sentence “The moon is bright tonight” and “Titan is a moon.” In the former case it is being used as a object word, and the latter case as a concept word. “We should not be deceived by the fact that language makes use of proper names . . . as concept words, and vice versa,” he tells us. “This does not affect the distinction between the two.” To recognize the logical role of the word 'moon,' one must look to see how it is being used in the context of the sentence. Cf: Frege, Foundations of Arithmetic §64 Brommage, “An Essay on Nonsense” 23 clearly defined meaning that we arrive at nonsense. V In this last section, I wish to briefly point to a few things that are important for our understanding of Wittgentein from the preceding analysis. To summarize, the austere notion of nonsense that I attribute to Wittgenstein does not result from a violation of logical syntax, or similar category mistake—but rather when a symbol cannot be perceived in the sign (or group of signs) which compose the proposition. Rather the Fregean view of nonsense that Wittgenstein adopts implies that it is only when there is no symbol that corresponds to the sign that nonsense results. In order to understand what symbol corresponds to this sign, what meaning it has, one must understand how it is used in a proposition. Understanding Wittgenstein on meaning and nonsense has application to specific debates surrounding the Tractatus. One of these is the realism/anti-realism debate. David Pears and Max Black, for example, read Wittgenstein as a semantic realist who espouses a very specific set of theses about the world in 1-2.063, upon which his theory of language is then generated. On this model of Tractarian language, the meaning of linguistic expression is not dependent upon our linguistic capacities; rather according to this interpretation, language's ability to represent the world is grounded upon these the links forged between names and simple objects. Other interpreters, such as Brian McGuinness, Hide Ishiguro and Rush Rhees all read Wittgenstein as a linguistic anti-realist, who believed that the structure of the world only shows up through language. To them, the postulation of simple objects in the opening passages merely follows from the structure of our language. In more radical cases, the early Wittgenstein has been even labeled a linguistic idealist.50 The dispute here is a 50 G. E. M. Anscombe, “The Question of Linguistic Idealism,” in Essays in Honor of G. H. von Wright, ed. Jakko Hintikka, Acta Philosophica Fennica 28 (1975). Brommage, “An Essay on Nonsense” 24 matter of the direction of the causal arrow: is our awareness of the world as it shows up to us dependent upon the way that language works, or is the structure of language derived from the nature of the world? Which side Wittgenstein falls on this issue is still an open question, and evidence can be garnered for both sides. If I am right, and the proposition (not the term) is the primary unit of meaning, then the realist reading fails. Starting with the assignment of names to simple objects, and building language up from there, will never succeed in “fixing” meaning. Instead if we realize that the proposition is the smallest unit of meaning, then it follows that we pick out objects when we have the capacity to use the names of objects in meaningful sentences. Further this means that he was not providing a metaphysical account of language (again, sadly, like Russell did in his atomist period). The “ontological myth” (as McGuinness puts it) at the opening of the text is exactly that—an explanatory myth. Another aspect for which this point on meaning has particular relevance is the question of how Wittgenstein's corpus is to be interpreted, particularly in relation to the continuity of Wittgenstein's philosophy. The standard potted history of analytic philosophy often interprets Wittgenstein as having two discrete phases of philosophical development. In his early years, including up to the publication of the Tractatus, he was mostly inspired by Russellian logical atomism, and was a type of reluctant metaphysician. After he returned to philosophy in 1929, he gradually began changing his mind about the views expressed in the Tractatus, and started taking a quite different standpoint to analyze the philosophical problems of language. His later methodology, exemplified by the Philosophical Investigations, is often considered to be a renunciation of his earlier work. Instead of espousing a view which ascribes the metaphysical conditions underlying logic and language, his later work shows the insufficiencies of this view by renouncing philosophical theories about language. Instead, the goal of his later philosophy is therapeutic, 51 to rid us of our philosophical confusions, and the urge to wax 51 Throughout the Philosophical Investigations, he often likens philosophy to a type of “therapy” (PI §133) or “treatment” Brommage, “An Essay on Nonsense” 25 metaphysical. If what I am arguing is correct, then it follows that there might not be such incongruity at all between the position of the Tractatus and the Philosophical Investigations. The notions of context and use that are often thought of as characteristic of Wittgenstein's later thought are already there in rudimentary form in the early work. This means that there is much more continuity in his philosophy than is often thought. The contextualism of his later thought—through the notion that meaning must be understood in terms of the language game or form of life—is largely a difference of degree, not of type.