Throughout his career, Wittgenstein propounded two main claims concerning nonsense. First, all metaphysics is nonsensical. The proper task of philosophy is not to answer metaphysical questions by producing theses, doctrines or theories, but to show that they rest on conceptual confusions. Second, ‘the negation of an a priori proposition’, e.g. ‘7 + 5 = 12’ or ‘Some objects are red and green all over at the same time’, is not a necessarily false proposition, but a nonsensical combination of signs (see §§251–2). By the same token, the a priori or necessary propositions themselves do not exclude a genuine possibility, since in their case no such possibility can meaningfully be specified (TLP 3.03–3.05; AWL 139–40, 165–6). Instead, the later Wittgenstein argued, such propositions ban a certain combination of words as meaningless from our language. Necessary propositions are not necessary truths, but norms that exclude certain sign combinations from our language. Both of these ideas have been vigorously contested by mainstream analytic philosophers. In recent years, Wittgenstein’s conception of nonsense has also been the central focus of the proponents of a ‘New Wittgenstein’ (Creary and Read 2000). Two claims set the New Wittgensteinians apart. The first is a reading of the Tractatus. In the final sections, Wittgenstein condemns the propositions of the Tractatus itself as nonsensical (6.54–7). According to a standard interpretation, his reason was that these propositions try to express truths about the essence of language which, by Wittgenstein’s own lights, cannot be expressed in philosophical propositions, but which manifest themselves in non-philosophical propositions properly analysed. According to the New Wittgensteinians, by contrast, the Tractatus does not consist of illuminating nonsense, nonsense that vainly tries to hint at ineffable truths, but of ‘plain nonsense’ (Diamond 1991: 181; Conant 1992: 198), nonsense in the same drastic sense as gibberish like ‘ab sur ah’ or ‘piggly tiggle wiggle’. The purpose of the exercise is therapeutic. By producing such sheer nonsense, Wittgenstein tries to unmask the absurd nature of philosophy and to wean us off the temptation to engage in it.
The second claim is an interpretation of Wittgenstein’s conception of nonsense. According to the New Wittgensteinians, his conception of nonsense, both early and late, was ‘austere’ rather than ‘substantial’ (Creary and Read 2000: 12–13; Diamond 1991: 111–12; 2000: 153, 165; Conant 2002: 380–3). There are two aspects to this contrast. First, according to the austere conception, nonsense is always a matter of privation. Whereas the substantial conception of nonsense allows for ‘positive nonsense’, nonsense that results from combining meaningful expressions in illegitimate ways, the austere view allows only for ‘negative nonsense’, nonsense which results from our not having assigned a meaning to expressions in a certain context.
Second, the austere conception of nonsense is monistic. Whereas the substantial conception distinguishes between different types of nonsense, the austere view insists that from a philosophical or logical point of view there is only one kind of nonsense: ‘all nonsense is just nonsense’ (Diamond 2000: 165); logically or semantically speaking there is no difference between the statements of metaphysicians and the babblings of a drunkard.
Finally, it should be noted that the New Wittgensteinians not only ascribe these views to Wittgenstein, they also subscribe to them. They think that the statements of the Tractatus are, indeed, nothing but plain nonsense. They also endorse the austere conception of nonsense, with its exclusive emphasis on privation and its monism.
In this chapter, I explore Wittgenstein’s conception of nonsense with particular emphasis on the later work. Two ideas provide succour for the austere conception. The strong contextualism of the Tractatus according to which a word has a meaning only in the context of a proposition can be used to support the privation view. And several passages in the later work endorse a monistic conception of nonsense. Nevertheless, my account differs sharply from that of the New Wittgensteinians, both exegetically and substantially. Wittgenstein never actually used contextualism to support the privation view. On the contrary, both early and late he allowed that nonsense can result not just from failure to assign a meaning, but also from combining meaningful expressions in a way that is prohibited by the rules for the use of these expressions. Furthermore, the monistic conception of nonsense is off-set and superseded by a pluralistic one, especially in the Philosophical Investigations. More importantly, the two ideas that favour the austere conception are mistaken, and they are incompatible with some central insights of the Investigations. Strong contextualism is ruled out by Wittgenstein’s insistence that the meaning of a word is determined by its use in the language. Monism is incompatible with reductio ad absurdum arguments, including Wittgenstein’s own, but also with his recognition that there are different degrees of linguistic understanding. There are many kinds of nonsense, and one of them results from the illicit combination of meaningful words.
The structure of my chapter is as follows. I shall first touch on the relation between combinatorial nonsense and contextualism. In section 2, I shall argue that the later Wittgenstein allows for combinatorial nonsense. He rejected a certain explanation of such nonsense, the ‘meaning-body’ conception, but this rejection is compatible with the idea that a string of words can be nonsense because of what its components mean. The third section tackles the question of how many kinds of nonsense there are. It explores the tension between monistic and pluralistic views in the later Wittgenstein and maintains that, by contrast to gibberish, we understand not just the syntactic construction of philosophical nonsense, but also its components. Otherwise we could not operate with latent nonsense in order to transform it into patent nonsense. What unites latent with patent nonsense is that there is, ultimately, no way of explaining it coherently. But this does not prevent us from handling such constructions in a reductio, no matter whether it occurs in mathematics or in the elenctic arguments of the Investigations. The final section raises two unresolved issues: How should we classify different types of nonsense? And how can latent nonsense feature in a reductio, given that it is incapable of being either true or false?
Both the German Unsinn and the English ‘nonsense’ can mean at least three different things. In one sense, they apply to statements or beliefs and mean something like patently false or unreasonable. In this capacity, the term was widely used in the enlightenment, to brand superstition and religious dogma. At present, it is popular among politicians, who tend to disparage uncomfortable allegations for example as ‘absolute and utter nonsense’. I shall refer to this as absurdity.
In a second sense, the term applies to actions of all sorts, including linguistic utterances, and means something like ‘obviously pointless’ or futile. I shall refer to this as futility.
Both these usages retain a connection to the original meaning of Unsinn, in that a nonsensical statement or action betokens lack of sense or even sanity. In a third and historically most recent sense, the term applies to linguistic expressions or utterances and means something like ‘meaningless’ or ‘unintelligible’. I shall refer to this as linguistic nonsense or simply as nonsense.
Wittgenstein seems to have been the first major thinker to link philosophy in general to linguistic nonsense in particular. Many philosophers of the past have disparaged the theories of their predecessors as false, unfounded or futile. But according to Wittgenstein metaphysical theories suffer from a more basic defect, namely that of being nonsensical. It is not just that they provide wrong answers, but that the questions they address are misguided questions to begin with, what the logical positivists later called ‘pseudo-problems’ (see §§109, 119; TLP 4.003, 3.323f.).
Kant already held that many of the questions that give rise to metaphysical theories are misguided to begin with (Critique of Pure Reason B82–3).
The idea that these questions are not just futile but linguistically nonsensical, however, seems to have entered Wittgenstein’s thought through Frege and Russell. Russell’s theory of types introduced a systematic dichotomy between propositions which are true or false and statements which are meaningless, although they may be impeccable as regards vocabulary and syntax.
(1) The class of lions is a lion
is not just obviously false, it is ‘meaningless’ or ‘nonsense’, since it predicates of a class what can only be predicated of individuals. Russell’s theory of types blocks such nonsense by prohibiting sentences which predicate of a thing of one logical type–e.g. classes–properties which can only be predicated of things of a different type–e.g. individuals.
The Tractatus, for its part, rejects as nonsensical not just sentences like (1), but also Russell’s attempt to prevent such sentences by type-restrictions like
(2) The class of lions is not a lion.
Indeed, the famous finale condemns the propositions of the Tractatus itself as ‘nonsensical’. This is not the place to go into this extraordinary conclusion (see Hacker 2000; Glock 2004 and 1996b: 258–64, 330–6). In order to become clear about Wittgenstein’s later views on nonsense, however, we must attend to the question of whether the Tractatus allows for combinatorial nonsense. There are two arguments in favour of the view that the book only admits of negative nonsense. The first is provided by the following two passages (Diamond 1991: 196–7; Conant 2002: 404, n. 92, 411).
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 (heißt nichts) is that there is no property called ‘identical’. The sentence is nonsensical because we have failed to make an arbitrary determination, and not because the symbol, in itself, would be illegitimate.) In a certain sense we cannot make mistakes in logic.
(TLP 5.473)
Frege says: every legitimately formed proposition must have a sense; and I say: every possible proposition is legitimately formed, and if it lacks a sense, this can only be (my emph.) because we have given no meaning to some of its constituents. Hence ‘Socrates is identical’ means nothing (sagt nichts) because we have not given a meaning to ‘identical’ as an adjective. For if it appears as identity-sign, it symbolizes in a completely different manner and way–the signifying relation is a different one,–hence, also the symbol is entirely different in the two cases; the two symbols only accidentally share the sign.
(TLP 5.4733)
Contrast
3 Socrates is identical
and
4 Socrates is identical with the husband of Xanthippe.
According to our passages, that (3) ‘means nothing’ or ‘says nothing’ is exclusively due to the fact that we have failed to give ‘identical’ any adjectival meaning. Of course, we could lay down such a meaning by an ‘arbitrary determination’, stipulating, for example, that ‘identical’ is to mean ‘human’. In that case, however, in (3) and (4) the tokens of ‘identical’ symbolize in an entirely different manner; the sentences merely have a sign in common. The failure of (3) to say something is due not to the meaning ‘identical’ has in statements like (4), since there is no sign with the same meaning in those two propositions. Instead, it is due to ‘identical’ not having been assigned a meaning in (3). This, in turn, suggests that lack of sense in a propositional sign can only result from privation, from failure to assign a meaning to one of its constituents.
According to the second argument, this restriction to negative nonsense is no coincidence, but explained by the contextualism shared by Frege and Wittgenstein (Diamond 1991: 98–100; Conant 2002: 398–405; Whitherspoon 2000: 321–5). In the Foundations of Arithmetic, Frege adopted a strong, restrictive form of contextualism: ‘Only in the context of a proposition do words mean something’ (1953: §62). Frege himself never adapted this context-principle to his later division of content into sense and meaning. The early Wittgenstein did just that, although it is essential to his distinction between sense and meaning that propositions do not have a meaning, and that names do not have a sense. Within these altered parameters, the Tractatus repeats Frege’s restrictive principle: ‘Only the proposition has sense. Only in the context of a proposition has a name meaning’ (3.3, see 3.314).
The restrictive principle that a word has meaning only in the context of a proposition seems to clinch the case for an austere conception of nonsense. It provides the crucial premise for the following argument:
P1 A word (name) has meaning only in the context of a proposition
P2 A proposition is a sentence with a sense
C No component of a sequence of signs that lacks a sense can have a meaning.
The TLP is definitely committed to P2: a proposition is, by definition, a propositional sign with a sense, one that has been projected onto reality (see 3.1ff., 3.31ff., 3.5, 4). Furthermore, the argument is valid. Finally, from (C) it follows that no part of (3) has a meaning, since (3) does not have a sense. It also follows that no part of (3) could mean what it does in a meaningful proposition like (4).
In the wake of Frege and Wittgenstein, the restrictive context-principle (P1) has been repeated by countless philosophers of language, Quine and Davidson pre-eminent among them. Nevertheless, it has one major shortcoming: it is wrong! Consider the following two columns:
to be to abide
to have to arise
to do to awake.
Two things about these words are indisputable. First, they are not part of a proposition. Second, they are neither meaningless nor nonsensical, but do have a meaning. The words in the first column are the auxiliary verbs of English, the words in the second are the first three items from the dreaded list of irregular English verbs.
Next, consider the following extract from the Merriam Webster’s Collegiate Dictionary:
nonsense n 1 a: words or language having no meaning or conveying no intelligible ideas b (1): language, conduct, or an idea that is absurd or contrary to good sense (2): an instance of absurd action 2 a: things of no importance or value: trifles b: affected or impudent conduct.
It would be absurd to maintain that the words printed in bold at the beginning of dictionary entries are meaningless, all the more so since the text that follows specifies what they mean.
These objections cannot be defused by invoking the distinction between type and token. Not only is it obvious that, for example, the type ‘to awake’ has a meaning in English, it is equally obvious that the particular token printed above is a meaningful word of English, rather than a mere scribble or collage of letters, even though it occurs on its own, outside the context of a proposition.
In ordinary parlance, we ascribe meaning primarily, though not exclusively, to words. Dictionary definitions, the very paradigm of explanations of meaning, paraphrase words or phrases rather than sentences. This is no coincidence. Far from being the ‘primary vehicle of meaning’ or the ‘units of significance’, as certain contextualists have it, most sentences are complex signs. Their meaning depends on the meaning of their constituents (Dummett 1981: 3, 593; cf. Quine 1980: 38–9; Diamond 1991: 108–11). By and large, we master sentences by learning how to construct them from familiar words. Furthermore, understanding the components and mode of composition of a sentence is a necessary condition for a genuine understanding of the whole sentence, even if, as the later Wittgenstein argued, it is not always sufficient (see §§350–1; BB 105–6; RPP II §§93–4).
It may be objected that my animadversions against the restrictive principle are based on our ordinary notion of meaning, whereas Frege and the Tractatus were operating with theoretical notions to which our standard criteria for meaningfulness do not apply. This response is not available to those who maintain that the Tractatus consists of plain nonsense, since such nonsense cannot constitute a theory. But Frege and Wittgenstein did operate with notions of meaning that differ from the ordinary one and are part of philosophical systems. Ye t this does not make them immune to my objection. For they are also committed to claims that involve our ordinary semantic notions. Otherwise, their pronouncements would simply lack the relevance that they and their followers have attached to them.
If Wittgenstein’s notions of sense and meaning were simply stipulations, his notions of senselessness and nonsensicality would be equally free-floating. In that case, his claim that philosophical propositions and questions are nonsensical because they contain words without meaning would not have to worry metaphysicians. Even if true, it would not show that these propositions and questions lacked the kind of sense that metaphysicians thought they possessed. Similarly, one cannot invoke the restrictive principle to condemn ‘identical’ in (3) as meaningless, unless that principle is based on the established notions of sense and meaning. But if we use the word ‘meaning’ ‘as it is normally used (and how else are we to use it?)’ (see §246), words can and do have a meaning outside the context of a sentence.
The reason why the early Wittgenstein’s restrictive principle fails to heed this point lies indeed in his extraordinary use of the word ‘meaning’. According to the Tractatus, meaning is something possessed by ‘simple signs’ or (logically proper) ‘names’. The meaning of such a name is the (simple, unanalysable) object it stands for (3.203, 3.22). The sense of an elementary proposition is a function of its constituent names, i.e. of their meanings–the objects they represent–and of their form, their combinatorial possibilities (3.318). The flipside of this compositionalism is that the role of names is to contribute to the determination of the sense of the elementary proposition. Outside the context of a sentence, Wittgenstein seems to have held, they cannot have such a role.
This line of reasoning rests on two assumptions, namely, first, that the ultimate constituents of propositions are all names and, second, that the meaning of a word is the object it stands for. Both assumptions were conclusively criticized in the Investigations (§§1–8, 40). Even if we accept them, however, the Tractatus is entitled at most to hold that names must be capable of occurring in propositions. It is not entitled to hold that they only have meaning when they actually occur in propositions, as the restrictive principle requires. Like a person, an expression can have a role without actually fulfilling that role at every given instant.
The early Wittgenstein may have overlooked this fact because of an important feature of the picture theory. The picture theory conceives of propositions as a kind of picture or model. A paradigmatic example of such a model is provided by the initial inspiration for the picture theory, namely, the law-court reconstruction of traffic accidents with the aid of dolls and toy-cars (NB 29.9.14). It is, indeed, the case that the individual elements of such a model do not have what the Tractatus calls a meaning outside of the model. It is only within a particular reconstruction that a doll or toy-car has a meaning, i.e. deputizes or goes proxy for a particular person or object. Outside of such a reconstruction, for example in a box, it certainly does not represent, e.g. Ms Smith or the lorry of Mr Anderson.
Words of a natural language, however, are different in this respect. Adjectives and verbs have a meaning even outside the context of a proposition. This even holds for proper names, the use of which is more flexible. They have a denotation generally, because of their role in a persisting practice, not merely on a case-by-case basis. It is not just on a particular occasion, through a fiat relating to its occurrence in a specific token-sentence, that tokens of ‘Socrates’ stand either for the Athenian philosopher or the Byzantine theologian or the Brazilian footballer. At most, it is possible to alter the denotation of proper names on a specific occasion.
In the Tractatus we already find ideas that are at odds with the restrictive principle, namely its compositionalism. One understands a proposition ‘by understanding its components’, and in translation from one language into another we translate not each individual proposition but ‘the constituents of propositions’. ‘The meanings of simple signs (words) must be explained to us if we are to understand them. With propositions, however, we make ourselves understood’ (4.024–4.026). This suggests that the early Wittgenstein did not take TLP 3.3 as literally as the proponents of the austere conception suppose, and as they themselves need to do in order to construct the aforementioned argument against combinatorial nonsense.
Wittgenstein’s later conception of meaning militates even more against the restrictive principle. According to the Investigations, the meaning of a word is its use. ‘For a large class of cases–though not for all–in which we employ the word “meaning” it can be defined thus: the meaning of a word is its use in the language’ (§43, see §30; BB 69). Alternatively, the later Wittgenstein declared that the meaning of a word is determined by its use, or by the rules for its correct use. There are important differences between these formulations, and the second one is superior (see Glock 1996a). But both concur on two points: whether a sign is meaningful depends on whether there is an established use, whether it can be employed to perform meaningful linguistic acts; and what meaning it has depends on how it can be used. For us, the crucial point is this: it is individual words and phrases rather than whole sentences that have a rule-governed use in the language. By parity of reason, it is individual words rather than whole sentences that have a meaning. The same moral emerges from the dictum ‘The meaning of a word is what is explained by the explanation of meaning’ (§560). As TLP 4.026 already recognized, we typically explain the use of individual words or phrases rather than of whole sentences.
To be sure, in the Investigations Wittgenstein quotes Frege’s restrictive principle with approval ‘A word has meaning only as part of a sentence’ (PI §49). But, with occasional exceptions, Wittgenstein explains the context-principle in a non-restrictive way, one that is compatible with the idea that individual words can mean something without actually occurring in a proposition (see M 54; BT 1; PG 1, 44; cf. PR 59). What he insists on is that they must be capable of occurring in propositions. He also suggests that such occurrences of words are semantically primary in the following sense. A proposition is the minimal unit by which a move is made in the language-game: only propositions can say something. There are no half-propositions in the sense in which there is half a loaf of bread. By the same token, naming an object is no more a move in a language-game than putting a piece on the board is a move in chess. Naming alludes to a sentential context in that it is essentially a ‘preparation’ for sentential use (Waismann 1965 13–4, 199, 318–20). There is a general dependency of words on sentences in that the practice of explaining words is a preparation for their employment in sentences. On the other hand, any particular sentential employment presupposes that the component words have a meaning in advance, on account of an antecedent practice.
The kernel of truth in contextualism is that the meaning of a word is determined by how it can be used within sentences. But it does not follow that the word has meaning only in the context of a sentence. On the contrary, it is the individual word which has such a use. If it is clear what role tokens of a type-word would play in a proposition, tokens of that type have a meaning, whether or not they actually fulfil that role. There is a difference between having a use in the language and being actually used on a particular occasion. There is also a difference between a word having a meaning and a sentence being used to say something. Wittgenstein was well aware of this. He distinguished sharply between meaning and sense in the Tractatus, and even in his later work he tended to speak of the Bedeutung of words and the Sinn of sentences.1
Hence, the privation conception of nonsense cannot be defended by appeal to contextualism. But perhaps it has intrinsic plausibility. One might concede that ‘Socrates’ might have a meaning outside the context of a meaningful sentence, while insisting that it lacks meaning when it occurs within the context of a nonsensical sentence like (3). One problem with this rejoinder is that it is at odds with the privation view. If in (3) ‘Socrates’ is meaningless not because we have failed to assign a meaning to it but because of its context, then its lack of meaning is not just a matter of privation, it is a matter of being in inappropriate company, just as the combinatorial view has it.
Furthermore, the claim that ‘Socrates’ lacks meaning in (3) is implausible. One can establish who ‘Socrates’ stands for by consulting an encyclopaedia, and establish what the term ‘identical’ means by consulting a dictionary (Hacker 2003: 10, 20). One can also translate (in the case of ‘identical’) or transpose (in the case of ‘Socrates’) the constituents of (3) into another language. Consequently, there is no obstacle to ‘Socrates’ meaning the same in (3) as it does in (4); in both cases, it is probably the name of the snub-nosed Athenian philosopher.
Of course, on some occasions words might be used with a meaning that differs from their normal one. In that case, their literal meaning will differ from their speaker’s meaning, their meaning on occasion of utterance. That meaning is to be gleaned from the speaker’s explanations. For instance, if someone utters both (3) and (4), and explains the term ‘identical’ in both as meaning human, that is what he means by it. And then his utterance of (4) is nonsensical, because of the meaning he has given to ‘identical’.The explanations speakers give of their words cannot be discarded as being of merely psychological importance (cf. Diamond 1991: 99, 106). What a speaker means by a word in a specific utterance is not determined by what words or images happen to cross her mind; but it is determined by how the speaker would explain her utterance when challenged. As the later Wittgenstein realized, the proper way to avoid psychologism is not to sweep the topic of understanding and meaning something by a word under the carpet, as Frege and the Tractatus had done. Rather, it is to realize that understanding and meaning something depend not on mental accompaniments of words, but on speaker’s explanations (Glock and Preston 1995).
The privation conception entails that all components of (3) are totally and equally bereft of meaning. But when we approach (3) in the spirit of the Investigations, namely by considering how we would react to it, we reach a different conclusion. If someone uttered (3), our first response would be to treat it as an incomplete sentence. We would be liable to ask ‘Identical with whom?’, precisely because ‘is identical’ is not a meaningless sign, but part of a two-place predicate. Imagine that our interlocutor responds by insisting ‘Not identical with anybody; he is just identical period!’. Being philosophers, we might then suspect that by ‘identical’ he means ‘self-identical’. If he denied this, however, we would be at a loss. The trouble with (3) concerns ‘identical’ rather than ‘Socrates’. No substitution for ‘Socrates’ would render (3) meaningful, given the literal meaning of ‘identical’. By contrast, replacing ‘identical’ by any number of one-place predicates restores sense. It is ‘identical’ rather than ‘Socrates’ that is being misused in (3), and it is the combination of ‘Socrates is’ with ‘identical’ that makes for nonsense.
Once we take to heart the shortcomings of strong contextualism and the difference between literal and speaker’s meaning, the insistence that lack of sense is always the result of privation is either false or trivial. It is false if it means that none of the components of (3) have a literal meaning, and the same meaning as in (4), or that the nonsensicality of (3) cannot result from combining ‘identical’ with ‘Socrates is’. It is trivial and inconsequential if it means that we could always give a meaning to any combination of words, by assigning a new meaning to one or more of its constituents, or to the combination as a whole.
Of course, one cannot rule out that 5.473 and 5.4733 were meant to preclude combinatorial nonsense simply by noting that such a preclusion would be unwarranted. But it is clear from other passages as well as from Wittgenstein’s later comments that the Tractatus condoned combinatorial as well as negative nonsense (see Glock 1996b: 214–16, 260; Hacker 2003: 12). Even more pertinent to my present concerns is evidence from the later period, which suggests that the import of 5.473 and 5.4733 does not lie in the rejection of combinatorial nonsense. In his record of Wittgenstein’s lectures of 1930–1, G.E. Moore reports:
Later on [Wittgenstein] said that for any sign whatever there could be a method of projection such that it makes sense, but that when he said of a particular expression ‘That means nothing’ or is ‘nonsense’, what he meant was ‘With the common method of projection that means nothing’, giving as an instance that when he called the sentence ‘It is due to human weakness that we can’t write down all the cardinal numbers’ ‘meaningless’, he meant that it is meaningless if the person who says it is using ‘due to human weakness’, as in ‘It is due to human weakness that we can’t write down a billion cardinal numbers’.
(M 58)
Here Wittgenstein starts out by making the point of 5.473. (3) is not ‘illegitimate in itself’, because the sign can be projected onto reality by making an appropriate stipulation. Nevertheless, (3) can be condemned as nonsense, because, in the parlance of the Tractatus, there is a common or standard method of projecting the sign onto reality. In the parlance of the later Wittgenstein, (3) is nonsense because there is a common or standard way of using ‘identical’, and in this standard way the sign cannot be meaningfully combined with ‘Socrates is…’. This squares well with the fact that 5.473 declares (3) meaningless on the grounds that ‘there is no property called “identical”’, i.e. by reference to the fact that ‘identical’ is not standardly used to signify a property. A sign that has a method of projection, that is used to signify something, is a symbol. Accordingly, nonsense here results precisely from an illegitimate combination of symbols rather than from failure to assign a meaning to signs. Further confirmation for this interpretation comes from the Philosophical Remarks:
The question whether philosophers have always so far spoken nonsense might be answered: No, but they have not noticed that they use a word in quite different meanings. In this sense it is not unconditionally nonsense to say that one thing is as identical as another, for somebody who has this conviction means something by the word ‘identical’ at the moment (perhaps ‘large’). But he does not know that he is here using the word in a meaning other than that in which it is used in 2 + 2 = 4.
(PR 55–6)
What makes a proposition like
(5) Socrates is as identical as Plato
nonsensical is the way in which its components are standardly used: ‘identical’ is not standardly used as a scaling adjective. By a similar token, a philosopher who utters
(6) It is due to human weakness that we can’t write down all the cardinal numbers
is speaking nonsense if he is thinking of human frailty, but not if he means that nothing a human being could do would count as writing down all the cardinal numbers. Far from regarding the explanations that speaker’s give of what they mean by individual words as irrelevant, Wittgenstein regarded them as crucial to establishing whether a particular utterance is nonsensical.
Seen in this light, our passages do not rule out combinatorial nonsense. Rather, they anticipate an important point of the later work. Meaning can be a feature of both type-expressions and token-expressions, whether words or sentences. But making sense is a feature of token-sentences. For Wittgenstein, as for Ryle, it is uses of words on a particular occasion which have, or lack, sense. Furthermore, both were especially interested in the uses that philosophers make of words in a specific question, statement or argument. Whether an utterance makes sense, and what sense it makes, is not simply determined by the linguistic form of the sentence uttered, its constituents and mode of combination (as compositionalism implies). Instead, it depends on the circumstances under which the utterance is made, including the prior interaction between speaker and hearer (see §489, II 221; OC §§348–50, 433; AWL 21; BB 9–10; Z §328; Ryle 1961).
This marks an important difference between Wittgenstein’s earlier idea of logical syntax and his later idea of grammar. Wittgenstein used ‘logical syntax’ (till 1931) and thereafter ‘grammar’ as label for the constitutive rules of a language. Furthermore, like logical syntax, the grammar of a language is a ‘system of rules for the combination of symbols’. These rules determine ‘which combinations make sense and which don’t, which are allowed and which are not allowed’, and nonsense results from transgressing them (LWL 46–17, my emph.; see Hacker 2002: 12–17). What changes is the conception of how these rules operate. Unlike logical syntax, grammar is not a comprehensive calculus of rigid and precise rules, but more akin to the rules of a loosely structured and flexible game. Some combinations of signs make sense and others don’t; yet this distinction applies not to type-combinations, but to token-combinations in a specific context of utterance.
In the Investigations the term ‘rules of grammar’ and its cognates appear only rarely. Rumours to the contrary notwithstanding, however, the book does not abandon the idea of grammar as a system of constitutive rules (see §§371–3, 496–7, 558). Nor did Wittgenstein abandon the idea that these rules draw the bounds between correct and incorrect uses of words. ‘It is correct to say “I know what you are thinking” and wrong to say “I know what I am thinking” (A whole cloud of philosophy condensed into a drop of grammar.)’ Furthermore, such incorrect usage can amount to ‘philosophers’ nonsense’ (PI II 221–2). Finally, the book continues to uphold the idea that a ‘combination of words’ can fail to make sense, and that in excluding it ‘from the sphere of language’ we set limits to ‘the domain of language’ (§§498–9). It is combinations of words that make or lack sense, and whether the occurrence of a word on a particular occasion results in nonsense depends at least partly on what other words it is combined with.
Proponents of austerity characterize the positive conception of nonsense they reject not just as maintaining that nonsense can result from combining expressions that are individually meaningful in an illegitimate way. They also associate it with a more specific idea. It is the claim that such combinations are nonsensical ‘because of the incompatible meanings of the words involved’, because the meanings of the parts ‘clash with one another’ and hence ‘fail to fit together so as to make sense’ (Conant 2001: 23). This is a claim that Wittgenstein would indeed reject as misguided. It amounts to the idea that behind each word there is a ‘meaning-body’, a non-linguistic entity which determines how it is to be used correctly. According to this view, a word is analogous to a single painted surface of an otherwise invisible glass-body with a certain geometrical shape, e.g. a cube or pyramid. The combinatorial possibilities of the visible surface depend on the shape of the body behind it. Similarly, we can derive the rules for the use of a word from its meaning, since the latter is a (concrete, abstract or mental) entity which determines the combinatorial possibilities of the word (PG 52–8; AWL 50–1).
The meaning-body conception is definitely mistaken. The meaning of an expression is not an object associated with it. Consequently, there can be no question of combinatorial nonsense resulting from a ‘clash’ of meaning entities, whatever that metaphor might amount to. By the same token, it is at the very least misleading to suggest that combinatorial nonsense is the result of combining expressions with incompatible meanings. At the same time, it is legitimate to hold that a combination of signs may be infelicitous because of what its constituents mean. Compare
(7) A ant is crawling along the floor
and
(8) The no it blue
with Carnap’s (1959) example
(9) Caesar is a prime number.
In (7) there is a phonetic or morphological incompatibility between ‘a’ and ‘ant’. (8) is garbled because it combines words of the wrong syntactic category. Grammaticality is restored if we replace ‘no’ by any number of nouns and ‘it’ by the copula. By contrast, what makes the components of (9) incompatible is neither their phonetic character nor their syntactic category. Sense can be restored if we replace ‘Caesar’ by another name–e.g. ‘5’–or ‘prime number’ by another count noun, e.g. ‘general’. Rather, (9) is nonsense because of what its components mean, because of the fact that the role of ‘Caesar’ is to refer to a person while the role of ‘is a prime number’ is to classify numbers according to their divisibility. One need not reify ‘meanings’ in order to speak about what an expression means. The ‘what’ here is not a relative pronoun, as in ‘what the expression is written on’, but an interrogative pronoun (Latin quid rather than quod). It indirectly introduces a question, namely: ‘What does the expression mean?’. That an expression has a meaning simply means that there is an answer to this question (see Rundle 1979: §47).
Some passages suggest that Wittgenstein might object even to this position. One of his criticisms of the meaning-body conception is that the rules for the use of a word do not follow from the meanings of signs, they partly constitute them. The rule of negation elimination, for example, according to which ‘p’ follows from ‘~~p’, does not follow from the meaning of ‘~’. Without that rule the sign would not have the meaning it has. And if the rule were changed, if we accepted instead ‘~~p = ~p’ the meaning of ‘~’ would change correspondingly (see 147n; PG 184; AWL 4; RFM 42, 398). Proponents of austerity might extrapolate this line of reasoning as follows. It is not the case that (9) is nonsense because of what the constituent expressions mean. On the contrary, these constituent expressions mean what they do partly because competent speakers of English reject (9) as nonsense.
Unfortunately, this extrapolation misapplies an important insight. It is correct that words do not have a meaning independently of the overall use that a linguistic community makes of them. This is perfectly compatible, however, with the idea that individual uses of words can be illegitimate because of what their components mean, because of how they are standardly applied and explained. Thus, (9) is excluded by the explanations of its constituents, for example:
‘Julius Caesar’ is the name of a Roman general
‘is a prime number’ is a predicate that applies to any number that
is divisible only by 1 and by itself.
Wittgenstein reminded us that the meaning of a word is what the explanation of meaning explains. He also stressed that explanations of meaning are not simply certain forms of words, regarded in isolation. They are part of a whole practice. But as part of such a practice, they can preclude certain uses of words as nonsensical. Anyone who responds to the above explanations by uttering or accepting (9) will be deemed to have misunderstood these explanations. Of course, speakers of English could suddenly regard (9) as compatible with these explanations. In that case, however, they would have changed the meaning either of the components of (9), or of the components of the explanans. To acknowledge the priority of the general practice over the particular utterance is in no way to reify meanings or to accept the picture of clashing abstract entities.
Both early and late Wittgenstein allowed not just bare nonsense, the uttering of signs that have no meaning in a certain language, but also combinatorial nonsense, the nonsensical combination of meaningful signs. Moreover, he was right to do so. This does not yet tell us, however, precisely what distinctions between kinds of nonsense Wittgenstein, in fact, drew or should have drawn. In the later work we encounter a tension between a monistic and a pluralistic tendency. Wittgenstein famously remarked that ‘the results of philosophy are the uncovering of one or another piece of plain (schlichten) nonsense’ (§119). But the term schlicht here need signify no more than that we are dealing with linguistic nonsense, and the passage does not address the question of how many kinds of linguistic nonsense there are. On that question, the Investigations features one explicit and well-known distinction, namely between patent and disguised nonsense.
What I want to teach is: to pass from a piece of disguised nonsense to something that is patent nonsense.
(§464, see §524)
Ye t another passage is at odds both with the privation view and with monism:
To say ‘This combination of words makes no sense’ excludes it from the sphere of language and thereby sets limits to (umgrenzt) the domain of language. But when one draws a boundary it may be for various kinds of reason.
(§499; my emphasis)
On the other hand, in the ‘Yellow Book’ the same admission is followed by a qualification:
The word ‘nonsense’ is used to exclude certain things, and for different reasons. But it cannot be the case that an expression is excluded and yet not quite excluded–excluded because it stands for the impossible, and not quite excluded because in excluding it we have to think the impossible.
(AWL 64; my emphasis)
A similar idea occurs in Investigations §500:
When it is said that a sentence is senseless, it is not, as it were, its sense that is senseless. Rather, a combination of words is excluded from language, withdrawn from circulation (my emphasis).
Wittgenstein was anxious to avoid the suggestion that there is a half-way house between sense and nonsense, an impossibility that can be thought or a sentence that has a sense that is senseless. In a lengthy passage from the ‘Yellow Book’ that precedes the one just quoted, this anxiety leads him to downplay the differences between kinds of nonsense.
In speaking of that which is impossible it seems as though we are conceiving the inconceivable. When we say that a thing cannot be green and yellow at the same time we are excluding something, but what? Were we to find something which we described as green and yellow we would immediately say this was not an excluded case. We have not excluded any case at all, but rather the use of an expression. And what we exclude has no semblance of sense. Most of us think that there is nonsense which makes sense and nonsense which does not–that it is nonsense in a different way to say ‘This is green and yellow at the same time’ from saying ‘Ab sur ah’. But these are nonsense in the same sense, the only difference being in the jingle of the words. Rules for the use of words can exclude certain combinations, and this in two ways: (1) when what is excluded is recognized as nonsense as soon as it is heard, (2) where operations are required to enable us to recognize it as nonsense
(AWL 63–4; my emphasis)
Wittgenstein here denies that there is a logical difference between gibberish like
(10) Ab sur ah
and philosophically relevant nonsense like
(11) This object is green and yellow all over at the same time.
The only difference between patent and latent nonsense lies in the ‘jingle of words’. The former causes no confusion, whereas the latter does, since ‘operations are required to enable us to recognize it as nonsense’. This seems to support nonsense monism, according to which any distinction between types of nonsense is of merely psychological relevance.
However, this monistic strand is incompatible with other aspects of Wittgenstein’s later work, and untenable in its own right. For Wittgenstein, rightly, insists that operations are required to transform latent nonsense into patent nonsense. To someone like Frege, who thinks that the meaning of a word is the object it stands for, he points out that in that case it would have to make sense for the meaning of an expression to die (§40); to someone like Russell, who holds that the object of a desire is whatever removes the desire, he points out that in that case ‘if I wanted to eat an apple, and someone punched me in the stomach taking away my appetite, then it was this punch that I originally wanted’ (PR 64; see PI §440). In this way, it is shown that apparently intelligible questions and statements have implications that are absurd and literally unintelligible, like ‘the question of whether the Good is more or less identical than the Beautiful’ (TLP 4.003).
The trouble is, we cannot operate on or with mere gibberish. If Wittgenstein genuinely denied that there is a difference between philosophical nonsense and gibberish, he would undermine his own procedure. He would also remove the possibility of making sense of reductio ad absurdum arguments and indirect proofs. In the indirect proof that 2 is irrational, we reduce to absurdity, i.e. contradiction, the assumption that 2 = p/q. If there were no difference between that assumption and ‘Ab sur ah’ such a reductio would be impossible. And as Quine wryly observed, this would surely be a reductio ad absur-dum of the idea that the negation of a necessary proposition is nonsensical rather than merely false (1980: 5).
In so far as Wittgenstein balked at drawing a distinction, he fell prey to his own rhetoric. In the quotation above, he suggests that a sentence that is nonsensical without being mere gibberish would have to have a sense that is senseless, or to express a thought, albeit an impossible or senseless one (similarly Diamond 1991: 3, 31; Conant 2001: 21–2). Now, plenty of philosophers would accept that one can think the impossible. They hold, contrary to Wittgenstein, that one can think that not p, where p is a necessary proposition, simply because the negation of a necessary proposition is false rather than nonsensical. To my knowledge, however, no philosopher has ever maintained that there are sentences that have a ‘senseless’ or ‘nonsensical’ sense, or express an ‘impossible’ or ‘senseless’ thought. And for good reasons, since these ideas are patently incoherent. But Wittgenstein was simply wrong, if he believed that a pluralistic conception of nonsense is committed to such incoherence. When it is said that (11) is nonsense of a different kind from (10), this only means that in certain respects it is part of the English language. Unlike (10), we understand (11) in certain respects. We can specify, for example, what these sentences are about. But we fail when we try to explicate them fully, notably when we try to specify what would have to be the case for them to be true or to explain them in a way that avoids absurd consequences.
Monists are wont to protest that any such difference must be merely psychological, a matter of the things we associate with different forms of words (Diamond 1991: 107; Creary and Read 2000: 12–3). There is no doubt that (11) is more prone to conjure up specific mental images or phrases than (10). But this is not the only contrast. The more important difference lies in what we can do with these locutions. This difference is neither merely psychological nor merely pragmatic, as contemporary formal semanticists might insist. At least by the lights of the later Wittgenstein, it is a difference of philosophical (logical or semantic) import. He would certainly insist that what we are able to do with an expression is no more irrelevant to its meaning or grammar than how we explain and understand it.
In fact, there is a close connection between linguistic nonsense on the one hand, explanation and understanding on the other. After all, linguistic nonsense must, strictly speaking, defy both, at least ultimately. Wittgenstein appreciated this connection (§513). For this reason, a logico-semantic distinction between different types of nonsense is required by Wittgenstein’s own insight that there are different degrees and types of understanding. Just like ‘making sense’, ‘being unintelligible’ and ‘understanding’ have ‘different senses in different cases’ (PO 66–7). One explicit attempt to distinguish between types of understanding features in Waismann’s Principles of Linguistic Philosophy, which lists various ways in which understanding may break down (347).
I cannot understand you, you must speak louder
I cannot understand you, that is sheer nonsense
I cannot understand you, I don’t speak German
I cannot understand you, that was too complicated to follow
I cannot understand you, I can’t see why you want…
I want to introduce a more elaborate list, one which combines Wittgen-steinian ideas with more recent distinctions (drawn in Künne 1983: 196–202). We can distinguish at least the following levels of understanding:
Perceptual understanding: This is achieved once we have grasped the sign, in the idiom of the Tractatus. If we have understood perceptually, we won’t need to say things like ‘Speak louder please, we can’t understand you’. Even gibberish like (11) can satisfy this requirement.
Recognition of a language: This is something that we can do as soon as we recognize words as hailing from a particular language, even if we cannot make out a sentence, as in garbled sequences like (9). Once this is possible, we can also translate the sequence, at least into languages with a similar morphology.
Recognition of a sentence structure: This is something we can do in the case of Carnap’s example, and of the notorious semantic anomaly devised by Chomsky.
(12) Colourless green ideas sleep furiously.
Because of their syntactic structure, we can report such sentences not just in direct speech, but also in indirect speech.2 We can, answer questions as to case, tense, and number. Where appropriate, we can transform such sentences from the active into the passive voice and vice versa. We can specify particular rules that have been violated. By the same token, in some cases, such as
(13) Augustine is my favourite paint,
we can conjecture that they result from a (typographic or acoustic) slip rather than from semantic confusion.
Understanding of potential sense: This is the understanding we have of, e.g. ‘Old men and women went to the bank’. What we require to understand the utterance is not linguistic instruction, but disambiguation of lexical and structural ambiguities.
Understanding of actualized sense: this is achieved on removal of the ambiguities, e.g. once we are told that the utterance means that old men and old women went to the river bank.
Understanding of what is said: Such understanding differs from the understanding of actualized sense in the case of indexical utterances like ‘You’ve lost something’ or ‘Today is Monday’. What we need to know here concerns not the sense of the type-sentence, but the conditions of utterance of specific tokens.
Understanding of illocutionary force: We may understand what is said by ‘I’ll be back’, without yet knowing whether it is, for example, a conversational flourish, a prediction, a promise or a threat.
Understanding of conversational implicatures: This kind of understanding is absent in the case of the novice academic who fails to spot that in a reference for a Ph.D. candidate, the sentence ‘He turns his essays in on time’ does not count as a recommendation. Such a novice might, for example, say to the referee ‘I cannot understand you, I don’t see why you write…’.
Understanding of implications: Holistic thinkers, including Wittgenstein and Davidson, have maintained or implied that one cannot understand a sentence, or entertain the thought it expresses, without understanding all of its logical and conceptual implications. In my view, this is an exaggeration. Someone can understand a statement, at least up to a certain point, without understanding all of its indefinitely many consequences (Glock 2003: ch. 9). One of the features that sets latent nonsense apart is that we fail to appreciate their patently nonsensical or absurd implications.
Aesthetic understanding: One basic criterion for understanding a linguistic expression is the ability to explain or paraphrase it. As Wittgenstein pointed out, however, this ability is not sufficient for understanding a work of art. For example, someone who understands a poem will not just be able to paraphrase expressions occurring in it, but also to say why they cannot be replaced by a paraphrase in this context. In this respect, understanding a poem involves a higher degree of linguistic understanding (§§522–35; PG 69; M 105).
Aetiological understanding: Wittgenstein was keen to pinpoint the sources of the errors he diagnosed, and he detected various similarities between his philosophical therapy and psychoanalysis. Freud sought to discover the Sinn im Unsinn, the rationale behind the apparent nonsense of jokes, dreams and psycho-pathologies, and Wittgenstein tried to do something similar for philosophical confusions.
Against correlating nonsense and understanding, it might be held that we can understand (13), even though it clearly violates semantic rules. But what we understand in such cases is what the speaker meant to say, not what he actually said. On the other hand, it is true that utterances may defy understanding in some respects without amounting to nonsense, for example utterances that defy perceptual understanding or understanding of conversational implicatures. Still, there is some correlation between types and degrees of understanding and types and degrees of nonsense. As we have seen, there are numerous differences between the various linguistic misfits we have encountered. What holds them together is not that they defy thinkability or imaginability, at least according to Wittgenstein (§§511–7). Rather, it is that they cannot be explained coherently. Ultimately, the attempt to explicate even the most latent piece of linguistic nonsense fails. We cannot coherently specify what it would be for 2 to be a ratio of two natural numbers, or what it would be for sentences to be names. Ye t explanations and understanding can fail at different points and for different reasons.
The pluralistic strand in the later work takes these lesson to heart. Vis-à-vis sentences like ‘Two colours are in the same place at the same time’, Wittgenstein writes: ‘It sounds English, or German, etc., all right’ (BB 55, my emphasis). With respect to more latent nonsense he advisedly made further concessions.
‘But in a fairy tale, the pot too can see and hear!’ (Certainly; but it can also talk.) ‘But the fairy tale only invents what is not the case: it does not talk nonsense’.–It is not as simple as that. Is it false or nonsensical to say that a pot talks? Have we a clear picture of the circumstances in which we should say of a pot that it talked? (Even a nonsense-poem is not nonsense in the same way as the babbling of a child.)
(PI §282; my emphasis)
This passage implies that there are at least three different types of nonsense: the kind exemplified by the babbling of a child; the kind exemplified by a nonsense-poem like Carroll’s ‘The Hunting of the Snark’; and the kind exemplified by the statement that a pot talks. In §348 we read:
‘These deaf-mutes have learned only a gesture-language, but each of them talks to himself inwardly in a vocal language.’–Now, don’t you understand that?–… I do not know whether I am to say I understand it or I don’t understand it. I might answer, ‘It’s an English sentence; apparently quite in order–that is, until one wants to do something with it; it has a connection with other sentences which makes it difficult for us to say that nobody really knows what it tells us; but everyone who has not become calloused by doing philosophy notices that there is something wrong here’.
In this passage, Wittgenstein definitely allows for a distinction between philosopher’s nonsense and gibberish. His answer to the interlocutor’s question grants that:
(14) These deaf-mutes have learned only a gesture-language, but each of them talks to himself inwardly in a vocal language.
To be sure, according to Wittgenstein there is something wrong with this sentence. It sounds ‘queer’ (PI II p. 174). But what is wrong with it can only be brought out because it has certain connections with other sentences. Only in this way is it possible to transform latent into patent nonsense.
In yet another passage, Wittgenstein explicitly links intelligibility with explicability, and suggests that there are degrees of both. Our understanding of the question ‘Does 7777 occur in the expansion of π?’ reaches ‘just so far’ as our explanations of that question (§516).
It emerges that the austere conception of nonsense is untenable both as an interpretation of Wittgenstein and in its own right. At the same time there are exegetical and substantive difficulties for the emerging picture. Let me mention two areas. First, while Wittgenstein accepted that latent philosophical nonsense results from the violation of rules, he resisted the idea that the infringed rules are more fundamental or essential to language than those violated by patent nonsense (see M 69–70; LWL 97–8; BT 413). He was also suspicious of the weaker claim that philosophical nonsense like (11) or semantic anomalies like (12) violate rules of a different, namely semantic, kind from those violated by garbled sequences like (8). The classic division of semiotics into syntax, semantics and pragmatics would be anathema to Wittgenstein. One cannot distinguish between syntactic and semantic rules on the grounds of whether they merely regulate the relations between signs or link signs to the world, if the meaning of an expression is not an object in the world but depends on its use, including its combinatorial possibilities.
It is also problematic to separate syntax and semantics on the grounds that the former classifies parts of speech exclusively on the basis of distribution, i.e. of whether they can be substituted for one another in a given context (see Lyons 1981: ch. 4.4). For substitutability is subject to the proviso that the result is an English sentence. Yet this is precisely what is in question with respect to semantic anomalies. As mentioned above, for Wittgenstein, a sentence is the smallest linguistic unit which can be used to perform a complete linguistic act. But it is at least arguable that so-called semantic anomalies fail to meet this standard. Chomskians ignore the fact that there is no uncontested criterion by which semantic anomalies amount to a sentence.
On the other hand, consider Waismann’s example
(15) London is north-east of the South Pole
(1964: 136)
It is plausible to hold that no claim is made by this statement, that it does not amount to a ‘move in the language-game’. Yet Wittgenstein tended to underestimate our intuitions of grammaticality, the difference which the grammatical ‘jingle’ of cases like (15) makes to their role in our language, e.g. with respect to indirect speech. (15) also nicely illustrates that there is a continuum of cases between patent nonsense and latent nonsense of a philosophical kind. It requires some reflection on our geographic concepts to tell why (15) is nonsensical, and even more elaborate reflection to establish what, if anything, is wrong with (14).
The second area is the operations by which Wittgenstein promises to unmask philosophical nonsense and, in particular, to transform latent into patent nonsense. The established model of reductio ad absurdum treats the process as one of deriving a contradiction from a hypothesis. This model fits indirect proofs in mathematics and some reductio arguments in philosophy. But it creates problems for Wittgenstein. For one thing, he himself regarded contradictions, e.g. statements of the form ‘p & ~p’ as senseless rather than nonsensical (TLP 4.46ff.). Following that lead, one might distinguish those reductios that derive a contradiction from a hypothesis from those which transform a latent nonsense into patent nonsense, something that literally defies comprehension. The second problem concerns this process of transformation. Logical relations are standardly defined in terms of truth and falsehood. By contrast, linguistic nonsense is supposed to fail the test of ‘truth-aptness’, of being even in the running for truth or falsehood. This problem might be solved by conceiving of logical relations in broader terms, e.g. in terms of what a claim commits one to. But there remains the question of how we should characterize logical operations that lead from something that is intelligible up to a point to something that literally defies understanding. In order to do so, it seems that we must abandon or modify another standard assumption (see Baier 1966: 521), namely that everything that stands in logical relations with something meaningful is itself meaningful. Such a modification is also required by another feature of Wittgenstein’s later work. He is committed to the idea that
(16) Nothing can be north-east of the North Pole or the South Pole
is meaningful, a grammatical proposition that expresses a linguistic rule, in spite of the fact that its function is to exclude utterances like (15) as nonsensical.
This is not the place to resolve these difficult issues. My aim was to show that some distinctions between kinds of nonsense are imperative. They are essential not just if we are to understand nonsense as a linguistic phenomenon in its own right, but also if we are to fathom the two links that interested Wittgenstein, namely, with metaphysics on the one hand, with logical necessity and impossibility on the other.3
1. Proponents of restrictive contextualism have suggested that one can only identify the meaning of a word when it occurs within the context of a sentence that has a sense (Diamond 1991: 30–1, 100–1; Creary and Reed 2000: 7, 14; Conant 2002: 384–5, 398), but this is once more based on the assumption that the meaning of a word consists in its actual contribution to the sense of a sentence in which it occurs. That assumption does not even hold for the meaning a token-word has on a given occasion of utterance, since that meaning is to be cleansed from the speaker’s explanations. It certainly does not hold for the literal or conventional meaning of a type-word. The latter depends not on the occurrence of a token in a particular sentence, but on its ‘use in the language’ (§43), on the role tokens can play within various contexts. The New Wittgensteinians purport to accept that the meaning of a word is its use in the language, yet ignore that this is incompatible with the restrictive contextualism they rely on. They also regard the contextualism of Frege and the Tractatus as inferior to the contextu-alism they ascribe to the later Wittgenstein. According to this position, even a sentence is not enough to endow a word with meaning, since ‘the minimum unit of linguistic significance is properly the whole language-game’ (Read 2000: 76). Far from solving the problems that bedevil restrictive contextualism, however, this exacerbates them, since it imposes even stronger and less plausible conditions on an expression being meaningful. Indeed, to ascribe linguistic meaning to a whole language-game makes no sense, and amounts to a confusion of the bearers of meaning (primarily words and phrases) and the source of that meaning (the use of these words in a practice that may be more or less extensive). The only viable option is to acknowledge that words have a meaning outside the context of a sentence (Read 2000: 77). But this simply removes the crucial premise for the standard argument against combinatorial nonsense.
2. Diamond (2000: 151, 161) argues that, if a grammatically well-formed sentence ‘p’ is nonsense, then so is ‘A thinks that/says that p’. But we can certainly say ‘Berkeley believed that to be is to be perceived’ and ‘Hegel wrote that the True is the Whole’. And if we say such things, what we say appears to be both meaningful and true.
3. I am grateful to Rhodes University, South Africa, for awarding me a Hugh-le-May research fellowship that enabled me to write this chapter, and to the department of philosophy at Rhodes for their generous hospitality. For comments on previous versions I should like to thank Peter Hacker, John Hyman and Daniel Whiting, as well as audiences at Grahamstown, Venice and Southampton.
Baier, A. (1966) ‘nonsense’, in P. Edwards (ed.) The Encyclopedia of Philosophy, Vol. 5, London: Macmillan.
Carnap, R. (1959) ‘The Elimination of Metaphysics through the Logical Analysis of Language’, in A.J. Ayer (ed.) Logical Positivism, Glencoe: Free Press.
Conant, J. (1992) ‘Kierkegaard, Wittgenstein and Nonsense’, in T. Cohen, P. Guyer and H. Putnam (eds) Pursuits of Reason, Lubbock: Texas Tech University Press.
——— (2001) ‘Two Conceptions of Die Überwinding der Metaphysik’, in T. McCarthy and S.C. Stidd (eds) Wittgenstein in America, New York: Oxford University Press.
——— (2002) ‘The Method of the Tractatus’, in E. Reck (ed.) Fro m Frege to Wittgenstein, Oxford: Oxford University Press.
Creary, A. and Read, R. (eds) (2000) The New Wittgenstein, London: Routledge.
Diamond, C. (1991) The Realistic Spirit, Cambridge, MA: MIT Press.
——— (2000) ‘Ethics, Imagination and the Method of Wittgenstein’s Tractatus’, in
Creary and Read (eds) The New Wittgenstein, London: Routledge.
Dummett, M. (1981) Frege: Philosophy of Language, London: Duckworth.
Frege, G. (1953) The Foundations of Arithmetic, Oxford: Blackwell.
Glock, H.J. (1996a) ‘Abusing Use’, Dialectica 50: 205–23.
——— (1996b) A Wittgenstein Dictionary, Blackwell: Oxford.
——— (2003) Quine and Davidson on Language, Thought and Reality, Cambridge: Cambridge University Press.
——— (2004) ‘Tractatus Logico-Philosophicus’, in J. Shand (ed.) The Central Works of Philosophy, vol. 4, London: Acumen.
———and Preston, J.M. (1995) ‘Externalism and First Person Authority’, The Monist 78: 515–34.
Hacker, P.M.S. (2000) ‘Was he Trying to Whistle it?’, in Creary and Read (eds).
——— (2003) ‘Wittgenstein, Carnap and the New American Wittgensteinians’, The Philosophical Quarterly 53: 1–23.
Künne, W. (1983) Abstrakte Gegenstände, Frankfurt: Suhrkamp.
Lyons, J. (1981) Semantics, vol. 1, Cambridge: Cambridge University Press.
Quine, W. V. (1980) From a Logical Point of View, Cambridge, MA: Harvard University Press.
Read, R. (2000) ‘What “There can be no Such Thing as Meaning Anything by a Word” Could Possibly Mean’, in A. Creary and R. Read (eds), The New Wittgenstein, London: Routledge.
Ryle, G. (1961) ‘Use, Usage and Meaning’, Proceedings of the Aristotelian Society, Supp. Vol. 35: 223–42.
Rundle, B. (1979) Grammar in Philosophy, Oxford: Clarendon.
Waismann, F. (1965) The Principles of Linguistic Philosophy, London: Macmillan.
Whitherspoon, E. (2000) ‘Conceptions of Nonsense in Carnap and Wittgenstein’, in Creary and Read (eds).