AN INTERPRETATION AND DEFENSE OF THE ‘PROOF’ OF THE FIRST ANALOGY IN KANT’S CRITIQUE OF PURE REASON

 

Guillermo Del Pinal*

 

III.              Two Variants of the Substantial Conception of Nonsense  

Remember what a hard time children have believing (or accepting) that a word really can have two completely different meanings.

Wittgenstein[1]

 

I just said that we established the restrictive claim of the austere conception of nonsense—that austere nonsense is the only possible kind of nonsense—but I need to be a little more careful. What we have established is the restrictive principle—that one cannot give a sign the wrong sense, that there is no such thing as a proposition in which a symbol appears in the wrong logical place. If it is true that all cases of substantial nonsense are ones in which nonsense is supposed to arise because of a clash in the logical categories of symbols or meanings, then it follows that we have established the restrictive claim of the austere conception of nonsense. But how do we know that we can’t have a plausible account of nonsense in the Tractatus that does not properly fall in either category? Well, the division of kinds of (Tractarian) nonsense into cases of austere nonsense and cases of substantial nonsense certainly seems exhaustive: either the constituent signs of a proposition symbolize or they do not, so a proposition is nonsense either because there is some sort of logical incompatibility between these symbols that does not allow them to form a significant proposition, or because we have simply failed to assign symbols to its signs. In other words, if there are cases of nonsense that do not arise because we fail to determine symbols for the signs (austere nonsense), then they must be because the signs symbolize, but something goes logically awry—and whatever goes logically awry seems to have to lie in the relation between these symbols, not in the resulting combination, because if there is a resulting combination then it seems the symbols do manage to combine to form a symbolizing proposition, or to express something (as a whole), however ‘illogical’ we might take that to be. I think that within a Tractarian framework (i.e. if we agree, within certain parameters, on the paradigmatic use or definition of concepts like ‘sign’, ‘symbol’, etc.) this division is indeed exhaustive. But some commentators hold accounts of nonsense in the Tractatus that commits them to an implicit disagreement with this division. To do that they have to hold that there are cases of nonsense that arise neither because the constituent signs of a proposition lack symbols, nor because they stand for logically incompatible symbols. As far as I know, there are two accounts of nonsense in the Tractatus that hold (not altogether explicitly) that there are cases of nonsense where the constituent signs of the proposition have meanings, but their ‘logical’ problem is not one that can properly be described as a clash of meanings or logical categories of symbols (so the restrictive principle does not apply to do them, or not at least in any obvious way).

I argue below that both of these accounts are philosophically indefensible and arise from failure to take seriously the fact that signs can symbolize in a number of ways (the first falls into essentially the same problem as substantial nonsense, the second is obliged to hold that propositions are nonsense because of the sense they have).  In other words, I argue that these two conceptions of nonsense cannot accommodate the phenomenon of cross-category equivocation. This will show that the above division is correct and thus that the restrictive claim of the austere conception of nonsense, as we have established it, is in no way threatened, and we do not need, to defend it, to incorporate any new ‘propositions’ or philosophical assumptions about the essence of language (i.e. we only need (I) and (II)). These discussions will also pave the way for the solution of the second paradox, of how to reconcile the austere conception with the claim that the task of philosophy is not to put forward philosophical doctrines, but to make propositions clear.

It is important that we base our criticism of these variant conceptions of nonsense on the Tractarian insight of cross-category equivocation, for no doubt although some proponents of these conceptions would perhaps accept that these accounts are philosophically defective for one reason or other, they would still hold (the degree of ineptness notwithstanding) that they are nevertheless doctrines held by the early Wittgenstein.[2] This reply looses much of its force if the implausibility of these conceptions follows from insights of the Tractatus itself: for these commentators would have to attribute to Wittgenstein not only an implausible philosophical doctrine, but one that is clearly inconsistent with some of his most basic insights, and that arises precisely because of an almost complete disregard of the fact Wittgenstein tells us not to forget as it gives rise to so many confusions in philosophy (3.325).  

 

IV.              A Critique of Hacker’s Logical-Syntax Variant

Philosophy may in no way interfere with the real // actual // use of language…for it cannot give it a foundation either.

Wittgenstein[3]

 

The first of these accounts is given by Hacker in his essay ‘Wittgenstein, Carnap and The New American Wittgensteinians’. Hacker says that Conant mistakenly attributes to him a conception of nonsense akin to the one we have been calling the ‘substantial conception’,[4] and proposes a different account of nonsense that, if valid, is a counter-example to the restrictive claim of the austere conception. Hacker’s account, however, cannot accommodate the fact of cross-category equivocation (and the consequences that follow from it). It is therefore problematic both philosophically and, as will become clear, exegetically.

Hacker suggests that Wittgenstein ‘explained the nonsecality of propositions not by reference to a lack of fit between the meanings of the words but by reference to transgressions of rules of logical syntax’ (Hacker 2003: 8). So Wittgenstein does not hold a substantial view of nonsense; but he does not hold an austere view either, for he holds ‘that philosophical nonsense results from the illicit combination of meaningful words’ (Ibid., 9). But it is not because of a ‘clash’ in the ‘meanings’ or logical categories of symbols that nonsense arises. The rules of logical syntax do not apply to symbols, for ‘a symbol just is a sign used according to the rules for its correct use’, they therefore apply to signs—i.e., are rules for the correct use of signs. Nonsense arises when we employ signs in ways that violate the rules, dictated by logical syntax, for their correct use—again, there is here no notion of a ‘clash’ between symbols, meanings, or anything of the sort (Ibid., 19, 17).

Conant does consider an account of nonsense which holds that nonsense can arise from an impermissible combination of ‘(mere) signs’, and suggests that any attempt to articulate collapses it into austere nonsense (Conant 2002: 400). The same is true of Hacker’s conception of nonsense. More specifically. Hacker’s conception of nonsense, as is clear from the outline above, differs from the austere and substantial conceptions by holding both (a) that there are cases of nonsensical propositions composed of meaningful words, and (b) that in these cases nonsense arises, not because of a clash of ‘meanings’, but because of a violation of the rules of logical syntax for the correct use of signs. But when we try to articulate (a) and (b) in light of the possibility of cross-category equivocation each becomes either false or has to be interpreted in a way that makes it consistent with the austere conception.

Let us begin with (a). As I said above, Hacker holds—for various reasons, including picture-theoretic considerations—that a symbol is a sign used according to the rules for its correct use, so in a nonsensical proposition signs do not symbolize, they are empty. It follows that, unless Hacker is contradicting himself or is using the concepts ‘symbol’ and ‘meaning’ in a non-Tractarian and unspecified way, by (a) ‘an illicit combination of meaningful words’ he cannot mean ‘an illicit combination of words symbolizing’; but ought to mean something like ‘an illicit combination of words that ordinarily have meanings’.[5] That indeed seems to follow; but it is not clear whether Hacker would accept it. He says things that suggest this ‘weak’ reading of (a); but he also says things that suggest a stronger reading. An example of the former is when he defends (a) by saying that it is very misleading to talk like Conant and say that a nonsensical proposition consists of meaningless words, when it is obvious that in most cases the words forming nonsensical propositions have a meaning in our language (Hacker 2003: 9, 10, 16). Now, if this is all Hacker means by (a) then proponents of the austere conception would not disagree with him: of course nonsensical strings are often composed of words (signs) that are ordinarily meaningful. The problem, as we have seen, is that in these special, nonsensical propositions we cannot identify signs with their ordinary modes of symbolizing. An example that favors the stronger reading of (a) is this:

Carnap held that in the sentence ‘Caesar is a prime number’ the words have a meaning, although the sentence is meaningless. Indeed, one could hardly argue that ‘we have failed to give a meaning to one of its constituents’, since the proper name ‘Caesar’ signifies the author of the Gallic Wars and the predicate expression ‘is a prime number’ signifies the property of a number of being divisible only by 1 and by itself. (Ibid. 17)

 

Hacker says that the nonsensical string ‘Caesar is a prime number’ is composed of meaningful words; by which he does not mean, for example, that ‘Caesar’ ordinarily has a use—he means that it has a meaning (‘it signifies the author of the Gallic Wars’) in this nonsensical proposition. But if this is supposed to be a statement about the mere word ‘Caesar’ it is clearly incorrect. The mere word ‘Caesar’ signifies in different ways in different propositions—it can stand for some other Italian, or for a political position in Imperial Rome, or, if we want, for a number. The only way we can tell whether it is being used to mean the author of the Gallic Wars or something else is by determining how it is being used, and in this nonsensical context—in this lack of logico-linguistic context—there is nothing that lets us determine that. Indeed, that in this case it appears in ‘Caesar is a prime number’ is, for any sensible person, a good reason for not taking it as signifying the author of the Gallic Wars.[6] On the other hand, we can of course simply stipulate uses for ‘Caesar is’ or for ‘is a prime number’, or for any other way we want to separate the string, but, as we have seen, this involves conditioning the use of the rest of the proposition, bringing in with the symbolizing part the form of the proposition which makes it possible (so, briefly, if we want to use ‘is a prime number’ as it is ordinarily used, i.e. to ascribe a numerical property to a number, then the only thing left to determine is, ‘what number does or can “Caesar” stand for?’).[7] This stronger reading of (a)—which would indeed differentiate Hacker’s view from the austere conception—is quite confused. The only option left is a weak reading of (a); which does not separate Hacker’s reading from the austere conception.[8]

            So interpreted Hacker’s view is already quite close to the austere conception of nonsense. Since Hacker holds that a symbol is a sign or word used according to the rules for its correct use, his view agrees with the austere conception that nonsensical propositions can neither be composed of words that symbolize nor arise because of the incompatibility of the symbols or meanings of its constituent words or phrases. The remaining disagreement, as far as (a) goes, is essentially a verbal disagreement about which way of expressing the phenomenon of nonsense is more accurate or less misleading (in connection with this point this is often what seems to be the problem, see e.g. ibid., 19). However, claim (b) is Hacker’s most important reason for holding that his view of nonsense is neither a version of the austere nor of the substantial conception.

The difference between (b) and the austere conception stems from Hacker’s notion of ‘rules of logical syntax for the correct use of signs’. Unfortunately, he does not explain this notion in any detail—i.e., he does not say more than that they are the rules for the correct use of signs—which leaves many questions open (especially because these discussions refer both to artificial and to natural languages). On what basis do we construct these correct rules—are they simply conventionally established or do they somehow reflect or follow from the ‘meanings’ of the signs? Suppose they are conventionally established. If we consider nothing but signs, it is unclear in what ways our rules can be normative and what relation they would have to our ordinary uses of signs (in natural languages)—of course, we can say that such and such a sign can only be combined in such and such a way or the result will be nonsense, but this seems completely arbitrary and it seems that a mere sign can be used meaningfully while infringing any conventionally established rule. To make the rules normative in the way Hacker wants them to be (to infer nonsense from their infringement), it seems that we  have to appeal to the meaning of signs and then state, given the nature of the word’s meaning, in which combinations the words yield sense and in which nonsense.[9] But here we have to appeal to the concepts ‘sense’ and ‘nonsense’ to construct the rules of syntax; and this use of ‘nonsense’ has not been explained and cannot be explained by saying that it arises from violations of the rules for the correct use of signs, for the notion of a violation of a rule presupposes the notion of nonsense we are here asked to explain. Although it seems we are stuck and cannot make further progress until Hacker clarifies his notion of ‘rules of logical syntax’ (and its relation to natural languages), in these questions we can begin to see that there is a deeper problem with (b), a problem that emerges even if we make no determinate decision about the origin of these ‘rules for the correct use of signs’. 

            The problem is with this proposition: if the constituent signs of a proposition are used in ways contrary to the logico-syntactic rules for their correct use, then we can infer that that sentence is nonsense (ibid., 17). This implies an account of analysis somewhat like this, ‘when examining if a proposition has a sense, examine if its constituent words are being used as they are supposed to, or ordinarily are, used. If that is not the case, then the words are being misused and the proposition is nonsense’ (see ibid., 9, 18, 19). This position only seems plausible if we do not give sufficient weight to the fact that signs can signify differently in different contexts. If we keep the possibility of cross-category equivocation in mind, it will be easy to see that, as Diamond puts it, the fact that ‘a sign is used in contravention of logical syntax by no means enables one to infer that a combination in which it occurred is nonsensical’ (Diamond [a], 4). Diamond gives an example that makes this clear. Imagine a child is showing off, and someone says, ‘he thinks he’s something. He’s so arrogant’. Even if no one had used the word ‘something’ in that way before (say, ‘something’ had only been used to refer to inorganic material objects), it does not follow that the person who makes that remark is either talking nonsense or saying something exceedingly foolish. It would be absurd to reproach her thus, ‘something is a word that stands for a thing not a person, no one thinks he is a thing and not a person because he is arrogant; your statement is therefore absurd’. The context shows that she is using the word ‘something’ in a different way (possibly violating old rules of logical syntax). Although we might not be able to determine the new use of the word, or she might simply be confused about it, before we can make any conclusions about nonsecality we have to attend to the context of use and consider seriously that she might have given it perfectly meaningful use.

             We can easily see how Hacker’s position only seems plausible if we do not give sufficient weight to the possibility of cross-category equivocation by examining uses given to the word ‘use’. Suppose you have a sign A and a set of syntactical rules {R} for its correct use. Here is a list of things that ‘using’ A and ‘misusing’ A could mean:

(1)   ‘using’ A = performing a linguistic act.

(2)   ‘using’ A = performing a physical act in accord with {R}, so that it results in a linguistic act.

(3)   ‘Mis-using’ A = performing a physical act not in accord with {R} (e.g. putting it in a sentence to the right of some other sign B, with rules {T}, which is not allowed by {R} and {T}).

(4)   ‘Mis-using’ A = performing a physical act that does not result in a linguistic act.

Now, if we are to infer nonsense from ‘mis-uses’ in the sense of (3), then no ‘mis-use’ in the sense of (3) can result in a ‘use’ in the sense of (1).  Now, what cross-category equivocation implies—since any sign can symbolize in a number of ways—is that any ‘misuse’ of A, in the sense of (3), i.e., performing a physical act not in accord with R, may result in a ‘use’ in the sense of (1), i.e., in a linguistic use. So we have to add this possibility:

(5)   ‘Mis-using’ A = performing a physical act in violation or not satisfying {R}, but one that may result in a linguistic act.

This implies that there is a possibility of ‘mis-using’ a sign in the sense of (3) which is still ‘using’ it in the sense of (1)—we are still performing a linguistic act. This shows that, given that in any ‘use’ of a sign there is always the possibility of cross-category equivocation, no ‘mis-use’ in the sense of (3) entails a ‘mis-use’ in the sense of (4). Of course, there are cases in which it is easier to recognize, from a ‘mis-use’ in the sense of (3), that there (probably) is a ‘mis-use’ in the sense of (4): such as in de-contextualized examples like ‘Socrates is and’. But, strictly speaking, there is no inference from (3) to (4). It only seems that way because we forget the fact that a sign can symbolize in a number of ways, we forget possibility (5), or do not give it sufficient weight. This is partly because in philosophical discussions of the nonsecality of propositions (such as this one) we use de-contextualized examples, so that the move from (3) to (4) is indeed almost immediate, that is, there is no point in taking into account possibility (5).[10] 

The point is now clear: that a sign does not conform to such rules does not imply that the sentence of which it is a part is necessarily nonsense. We can try to avoid this objection by suggesting that whenever we use a word in a new way we somehow implicitly establish new rules for its correct use. In a certain sense this might be true. But then, as far as their function for recognizing nonsense goes, these ‘rules of syntax’ loose the value they supposedly had for Hacker—for then whenever a word or phrase is used contrary to the rules of its correct use, we still have to examine if it is being given a new use, so we cannot infer from ‘incorrect’ uses that a proposition is nonsense. Now, it is true that the rules for a sign’s correct or ordinary use, which we might say are understood and employed by language users, help us determine the role or use a word might have in a sentence. But if they do not help us, if it is plain that the word is not used as it is ordinarily used, our immediate conclusion should not be, ‘the proposition is nonsense’, but should rather be, ‘perhaps the word is being used differently (and perhaps still close enough that it retains some of its old aspects and thus does not make the choice of the sign completely arbitrary), let us try to determine the new use’.[11] In each unusual case we have to attend to the significant use in a proposition, to the word’s possible logical role. Hacker’s conception of nonsense, once we attempt to articulate it in light of cross-category equivocation, becomes quite implausible.[12]

Using a (natural) language is a normative activity in a narrower sense than Hacker seems to think; there are explicit and implicit rules for the use of words, rules which make linguistic communication possible, but the ‘violation’ of these rules is a regular activity of language users and an essential factor of the complexity, power, and evolution of language. This is the feature of language that we recognize when we insist that we must not loose sight of the fact that signs can symbolize in a number of ways. To attribute to Wittgenstein Hacker’s conception of nonsense is not only to attribute to him a mistaken philosophical doctrine, but also one that is in conflict with, and arises precisely from ignorance of, some of his most basic insights.

 

Continuará en el número de verano.

 

* Guillermo Del Pinal, Magister Artium en Filosofía; Universidad de Chicago

 



[1] PO: 179

[2] See e.g. the following passage from Hacker 2003 (he is specifically referring to criticisms of Conant and Diamond against the saying/showing distinction and several other doctrines that Hacker attributes to the Tractatus, but he is also protesting against the method of criticism and defense that they employ in promoting their reading of the Tractatus):

 

The exegetical task is to make sense of his thinking what he thought, not to make sense of what he thought, since we have it on his own (later) authority that what he thought was confused. (Hacker 2003: 370)

 

Whatever plausibility this statement might have about exegetical methods in general, is irrelevant in this particular case. The criticism of Conant and Diamond against Hacker’s reading show, not only that his position is very weak philosophically (that it would, as Kremer puts it, ‘make of Wittgenstein a bumbling fool’), but also that we can see its inconsistencies precisely by attending to key and undeniable things Wittgenstein said (e.g., the phenomenon of ‘cross-category equivocation’). In Conant’s and Diamond’s works the philosophical critique is not separate from the exegetical critique, which is what makes Hacker’s response, in this case, unpersuasive. There are cases, of course, were the critique is purely philosophical, but as Conant and Diamond insist, the mistakes and ‘logical messes’ in the doctrines attributed to Wittgenstein are ones that it is unthinkable a philosopher with Wittgenstein’s logical rigor and standard of clarity could have simply overlooked.  As they put it, ‘early Wittgenstein was, after all, early Wittgenstein’ (Conant [a]: 51). The critique of substantial conceptions is therefore threefold: (i) it is philosophical invalid, (ii) it is inconsistent with valid principles of the book, (iii) it arises from overlooking the features of language Wittgenstein himself explicitly tells us not to overlook. 

[3] PO: 177

[4] Hacker says that  not only does Conant supply no textual evidence of places were he defines nonsense in accordance with substantial nonsense, but also that it is evident from places were he does define nonsense that he holds a different view.

In a sense, Hacker’s response fails to hit the mark. Conant attributes substantial nonsense to him and other commentators, as is clear from the contexts in which Conant makes those claims, not solely or fundamentally because Hacker claims alliance to that view in his definition of nonsense, but because of the way he uses the concept of nonsense: that is, because of the claims he wants to make about and the conclusions he wants to draw from nonsensical propositions.

Now, even if Conant is right, the fact that Hacker’s refined definition of nonsense is inconsistent with some of his other, earlier theses or uses of the concept does not affect the possibility that this new conception is plausible. So we have to examine Hacker’s case for a ‘middle way’, even if he often uses ‘nonsense’ in other ways. I will argue that Hacker’s ‘new’ conception suffers essentially from the same sort of picture of his old conception, and in both cases the problem arises from not taking seriously the phenomenon of cross-category equivocation.

[5] If the concepts ‘meaning’ and ‘symbol’ are used in the way they are used in the Tractatus, a word cannot have a linguistic meaning if it does not symbolize.

[6] See footnote 29.

[7] Hacker could also mean that ‘Caesar is a prime number’ is nonsense because (assuming) the only meaning we know of ‘Caesar’ is that of the author of the Gallic Wars and the only meaning we know of ‘is prime number’ is that of a certain property of a numbers. There is nothing incorrect in such an explanation; but it is incomplete and might be misleading. To take ‘is a prime number’ as used in ‘1 is a prime number’ is to take ‘Caesar’ as standing for a number. Since the only use of ‘Caesar’ we know of is that of standing for the author of the Gallic Wars, the proposition is meaningless. Only in this sense, which is consistent with the explanation we have given, can we say that ‘Caesar is a prime number’ is meaningless because of the meaning of Caesar. In this sense we can give a description of why nonsense arises that might use talk of ‘meanings’—but not quite in the way the substantial conception uses that notion, and we have to be careful. The ‘pushing out’ of the ordinary meaning of ‘Caesar’ occurs when deciding what use to give to ‘is a prime number’, that is, there are not here two distinct acts.

[8] Hacker’s position on this issue is quite confusing. It might even be the case that this reading is not unfaithful to him. Here is one of the passages where we can see the trouble clearly (Hacker 2003: 19):

 

But of course in these combinations [the signs] are being used in a way contrary to, or in violation of, the rules for their use. So these word sequences are indeed meaningless. If one wishes one can express this by saying that in these contexts no meaning has been assigned to these words or phrases (i.e., the combination is meaningless). But one should not therefore embrace Conant’s conception of ‘austere nonsense’ and say that this meaningless combination of words is composed of meaningless words…

 

Hacker says that if one wishes one can say (i) that in the nonsensical string (‘these combinations’) the constituent words are meaningless, that ‘in these contexts’ no meaning has been assigned to them, and then he says (ii) that one should not, however, embrace Conant’s formulation and say that the nonsensical string is composed of meaningless words. This is either a contradiction, or he is using the word ‘meaning’ in two different ways. One way to reconcile this is by assuming that in (ii) by ‘meaningful’ he means that the words ordinarily have a meaning, which Conant would not deny, or by assuming that in (i) by ‘meaningless’ (words) he means that they do not result in a meaningful combination, which Conant would also not deny if this is because the word(s) have not been assigned symbols.

[9] In this essay Hacker rejects this possibility (i.e., he rejects a ‘meaning-body’ account of the construction of rules of logical syntax), though in other writings he seems implicitly accept it. He says e.g. that the rules of logical syntax of a name must reflect the actual combinatorial possibilities of that which it names: ‘in the Tractatus the rules of logical syntax governing the combinatorial possibilities of names were thought to mirror the nature of objects and the logical structure of the world’ (Hacker 2002b: 38). It seems (though it depends on the interpretation of Tractarian ‘objects’) that to mirror the combinatorial possibilities of the objects signs name we have to take into account the objects for which signs stand when constructing the rules for their possible combinations.

[10] Nothing I’ve said here should be taken to mean that one should not talk of ‘using a word in violation of the laws of (logical) syntax’ (as a justification for inferring a mistake). This way of talking is often useful: especially because one can be brief in de-contextualized examples, where collapsing (3) and (4) is usually innocuous. (It is also useful, as M. Kremer pointed out to me, in discussions within the context of artificial languages—e.g., in teaching logic and mathematics—where allowing an appeal to (5) would subvert the purpose of the language.) The problem is only in taking these violations or deviations from the ordinary as granting an inference to the nonsecality of propositions. That is to say, there is almost no philosophical gain in not collapsing (3) and (4) when dealing with de-contextualized examples; but this collapse will easily lead one astray when dealing with philosophical texts (i.e. the more complicated the text under analysis, the more weight must be given to possibility (5)). Many of the criticisms against Carnap’s analysis of Heidegger’s propositions in ‘What is Metaphysics?’ can be captured by the claim that, in several cases, he collapsed (3) and (4), and did not seriously consider (5). The same is true, according to some authors, of some of Hacker’s own philosophical analyses. See e.g. Gustafsson 2001. It is indeed important that the mistake I am here attributing to Hacker is also present in his philosophical criticisms, and Gustafsson’s discussion strongly supports this. This also shows that we can learn philosophy from the Tractatus (a point which I would not mention except that Hacker sometimes seems to think otherwise). 

[11] As I just said, this question makes almost no sense if we are examining a proposition such as ‘Caesar is a prime number’, where we have no other context. In this cases, like I said, the inference from ‘mis-use’ in the sense of using (physically) not in accord with how we ordinarily use a sign, to ‘mis-use’ in the sense of failing to meaning anything, seems plausible. In claims like ‘ “Caesar is a prime-number” is nonsense because “Caesar” is not being used appropriately, etc., etc.’ the ‘because’ there does not seem problematic (see Hacker 2003: 9). But, of course, once we get to more complicated examples like the ones we do encounter in philosophical analysis, then this collapse of (3) to (4), this inflated and confident use of the ‘because’, becomes problematic.

[12] There is a brief passage in Hacker’s essay from which we can construct a possible response to this criticism:

 

…one should not therefore embrace Conant’s conception of ‘austere nonsense’ and say that this meaningless combination of words is composed of meaningless words, and hence that all that is wrong with it is that we have given these words no meaning in such contexts. For, among other things, one should also note that if one were to assign to a significant word or phrase a meaning in contexts from which it is excluded, then one would have changed its meaning. So one would, as Wittgenstein noted, be talking of something else. (Hacker 2003: 19-20; see also Hacker 1987: 52-55, my emphasis)

 

By saying that if we stipulate new meanings for the words one would be talking of something else, Hacker is implicitly assuming that we know what we were talking about when we declared that a certain proposition was nonsense (in the relevant cases). E.g. when he says that ‘Caesar is a prime number’ is nonsense he means that, given the ordinary meanings of those words, that string is nonsensical. So he would say that considering possibility (5) is in certain cases irrelevant because we know the meanings of the words (or we know what meanings we are interested in) and what we are concerned is if, given those meanings, the way we are trying to combine them yields a sense. (so e.g. Hacker would not, of course, deny that we can give a sense to ‘Caesar is a prime number’ or any other string through arbitrary stipulations, but would say that that is besides the point) There are, I think, two important issues here: (i) that we can determine the meaning of words, to a certain extent, in isolation of the ‘inept’ context, and (ii) that we can criticize the philosopher, in some cases, by appealing to the ordinary (or proper) rules of the ordinary words he is using (we can call this the ‘if you change your meanings you are talking about something else’ argument). For a detailed criticism of each of these views see footnote 31.