ON NONSENSE IN THE TRACTATUS LOGICO-PHILOSOPHICUS:
A DEFENSE OF THE AUSTERE CONCEPTION
Guillermo Del Pinal*
Most of the propositions to be found in philosophical works are not false but nonsensical… (4.003)
Philosophy is not a body of doctrine but an activity…The result of philosophy is not a number of ‘philosophical propositions’, but to make propositions clear. (4.111)
…anyone who understands me eventually recognizes [my propositions] as nonsensical. (6.54)
Wittgenstein, Tractatus Logico-Philosophicus
By condemning as nonsensical most of the propositions found in philosophical works, the Tractatus forces us, its readers, to ask,
(1) What is the Tractatus’s notion or theory of ‘nonsense’?
No doubt such a startling claim can be taken to suggest that we should try to find some ‘special’ Tractarian conception or theory of nonsense. But be that as it may, by then condemning as nonsensical its own propositions and by suggesting that philosophy is not a body of doctrine but an activity, the Tractatus forces us to ask, about our answer to (1),
(2) Isn’t that account or theory itself nonsense?
(3) How can we reconcile that account or theory with the idea that philosophy is not a body of doctrine but an activity of clarification, that philosophy does or should not result in a set of ‘philosophical propositions’?
These questions give rise to a series of ‘foundational’ paradoxes that any account of nonsense in the Tractatus has to eventually face, and its capacity or incapacity to solve these paradoxes is a good indication of its philosophical and exegetical merits.
Indeed, ‘traditional’ readings of the Tractatus are strongly criticized, and often rejected, because they fail to solve these foundational paradoxes, both in their more philosophical (2) and in their exegetical (3) variations. To give an example, one common answer of ‘traditional’ readings to (1) is that the Tractatus features a theory of the conditions of sense such that from the recognition that a proposition is neither logical (a tautology or contradiction) nor bipolar (is not capable of being true and capable of being false), we can infer that it is nonsensical. How the paradox then emerges is well known: that all non-logical propositions must be bipolar is neither a bipolar nor a logical proposition, and the same is the case for the propositions that sustain it, and thus it seems that the bipolarity doctrine ‘cuts the branch on which it sits’. In addition, the bipolarity doctrine is sustained by substantial philosophical theses of the essence of language which together form a body of doctrine of the sort that Wittgenstein claims is not the task of philosophy to put forth.
Some commentators take these problems to show that this ‘traditional’ answer to (1) is clearly inadequate; others propose a solution based on a particular interpretation of the saying-showing doctrine. The suggested solution is that although those sentences are indeed nonsense, they somehow manage to convey what they fail to state—they let or help us grasp that language somehow shows that all meaningful propositions must be bipolar, though one cannot properly, or intelligibly, say that. This solution, however, does not really solve either variant of the paradox, and creates new problems. The following are some of the problems, though perhaps those dissatisfied with this solution would not subscribe to all of them. Firstly, it seems that the intelligibility of the solution depends on our being told what supposedly can’t be said: for the solution has essentially this form: it cannot be said that p, but that p is shown in language—but how are we to understand this, such that we can grasp that p is shown, without somehow grasping the sense of p (by the same token, how can we know that whatever it is that is shown is that which we were trying to say)? Secondly, the proposition ‘that all non-logical propositions must be bipolar is shown in language’ is, again, it seems, neither logical nor bipolar, and to go a further step—‘that that all non-logical propositions must be bipolar is shown in language is shown in language’—makes the ‘solution’ unintelligible. Thirdly, the saying-showing doctrine is itself sustained by substantial philosophical theses about the nature of language and reality (e.g. that there are ineffable truths). Fourthly, that a substantial doctrine like the bipolarity condition (as interpreted above) is ‘ineffable’ (supposing that we can make sense of the idea of ineffable doctrines) does not make it less of a philosophical doctrine. These problems suggest that we should ask, about the saying-showing doctrine, ‘isn’t this too one of the theses (assuming this is actually a Tractarian, even if transitional, doctrine) that we are supposed to eventually recognize as nonsense?’, and several commentators have forcefully shown that the answer is, ‘Yes’.
The reading I defend in this essay—the ‘austere conception of nonsense’—gives a different answer to (1). It is this: according to the Tractatus nonsense can only arise because we failed to determine the meaning of the constituent words of a proposition; all that is wrong with many of the propositions of philosophy is that their words lack meanings, not that they have been assigned meanings in ways that violate some special conditions of sense derived from the Tractatus’s insights of and doctrines about the essence of language.
One of the virtues of the austere conception is that it is supposed to solve the foundational paradoxes. Yet several commentators strongly disagree:
When it is convenient for their purposes, proponents of the post-modernist interpretation [includes the austere conception of nonsense] have no qualms in quoting and referring to further points Wittgenstein makes in the Tractatus, which they take to be correct rather than plain nonsense…Similar considerations apply to [their] reliance upon…5.473 and 5.4733 in order to sustain the claim that the propositions of the Tractatus are mere nonsense in as much as we have given no meaning to words, otherwise used as formal-concept words…These claims…are surely not bipolar propositions with a sense…Nor are they…tautologies. So they too are nonsense, and cannot legitimately be invoked to support [their] thesis. The merit of [their] interpretation was supposed to be that it saves Wittgenstein from the embarrassment of sawing off the branch upon which he is sitting. But it now turns out that [their] interpretation involves exactly the same embarrassment. (Hacker 2000: 362)
Though this passage is somewhat inflated, it is not hard, initially at least, to sympathize with its general thrust. Take two passages of the Tractatus commonly cited to defend the austere conception of nonsense:
(5) Every possible proposition is legitimately constructed, and, if it has no sense, that can only be because we have failed to give a meaning to some of its constituents. (5.4733, my emphasis)
(6) We cannot give a sign the wrong sense. (5.4732, my emphasis)
These propositions certainly seem like substantial theoretical theses, theses of the sort that form part of a doctrine of ‘philosophical propositions’ and perhaps that we are supposed to eventually recognize as nonsense. Some proponents of the austere conception suggest that these propositions are part of the ‘frame’ of the book, instructions on how to read and understand the method of the Tractatus. But, again, at least initially, this seems very problematic: firstly, (4) and (5) certainly resemble more what we would call, in any ordinary sense of the terms, ‘philosophical doctrines’ than what we would call ‘instructions’, and, secondly, if they are given (somewhat perversely) the status of instructions, then it seems that we have opened up a possible line of defense for those claims austere readers want to reject. In addition, it seems that if we keep statements like (4) and (5) we begin our supposedly ‘doctrine-free’ clarificatory activities with a substantial theoretical framework under our belts.
No doubt statements such as these seem like substantial theses of the nature of language and the world, and they do seem to have some resemblance to the more metaphysical (and perhaps transitional) propositions of the Tractatus. But one cannot decide, a priori as it were, whether a single proposition is ‘philosophical’ or trivial, whether it is nonsensical or perfectly intelligible. Of course not all the sentences of the Tractatus are nonsense; and of course not all assertions are to be classed as ‘philosophical’ or as constituting a ‘philosophical doctrine’. To determine the status of a Tractarian proposition—to determine whether or not it is nonsense, whether or not it is a substantial philosophical ‘claim’—we have first to examine the meaning of that proposition, the kind of work it is doing in, and justification it gets from, the Tractatus. Now, to say that is only to recognize that there is a task open, that there are paradoxes that need to be solved. Indeed several commentators think that proponents of the austere conception have not fully solved these paradoxes. Their worry can be captured thus: ‘the austere conception relies on seemingly strong theses, and thus it is not clear whether in falls, like ‘traditional’ readings, into a ‘foundational’ problem. Proponents of the austere conception must give a perspicuous account of the ‘status’ of its founding claims—i.e., of what claims justify it and why they are not nonsense—and they must make clear how this is not inconsistent (not too much at least) with the idea that philosophy should not put forward philosophical doctrines’.
This is the worry I address in this paper. I argue (sections I and II) that the austere conception can be derived from one fundamental proposition, a proposition that is a consequence of a trivial feature of language. I then discuss (sections III and IV) two accounts of nonsense that threaten the derivation made in section I, and show that they arise through a failure to appreciate that same feature of language. This will show that the austere conception does not have to presuppose ‘substantial’ theses of the essence of language advanced in the Tractatus or elsewhere. I will finish (section V) by reconciling the austere conception with the idea that philosophy is not a set of philosophical doctrines but an activity of clarification. This will provide a clear solution to both paradoxes, and an answer to all the foundations-related objections recently raised against the austere conception, focusing on those raised by P. M. S. Hacker in his essays, ‘Was He Trying to Whistle it?’ and ‘Carnap, Wittgenstein, and the New American Wittgensteinians’.
I. The Presuppositions of The Austere Conception of Nonsense
Logic takes care of itself; all we have to do is look and see how it does.
In this section I give a somewhat extensive interpretation of the austere conception of nonsense with the aim of bringing to light its grounding proposition. No argument will be given here to justify it.
‘Traditional’ readings ascribe to the Tractatus two kinds of nonsense, we shall call them ‘austere’ and ‘substantial’ nonsense. Austere nonsense comprises cases in which a proposition is nonsense because we have failed to give a meaning to its constituent parts. In other words, in cases of austere nonsense we have failed to assign symbols to the constituent signs of a proposition. Substantial nonsense comprises cases in which a proposition is nonsense because we have assigned symbols to its constituent signs in such a way that we violate or do not meet certain conditions of sense. Now, within substantial conceptions, there are different accounts of what a ‘violation of the conditions of sense’ consists of. We shall call these accounts, again following standard usage, the ‘logical-syntax’ view and the ‘bipolarity’ view. According to the logical syntax view, nonsense arises when there is a clash between the logical categories of the symbols assigned to the constituent signs of proposition. According to the bipolarity view, nonsense arises when the result of the symbol assignments is neither a logical nor a bipolar proposition. These are obviously different kinds of violations of the conditions of sense, and although both assume or have to assume that in these kinds of nonsensical propositions the constituent signs do have a meaning, they must be treated separately. In the rest of this section, by ‘substantial nonsense’ I mean only the first of these accounts (I discuss the bipolarity view in section IV).
Essential to the austere conception of nonsense is the rejection of the possibility of cases of substantial nonsense. It holds that nonsense can only arise because we failed to assign symbols to the constituent signs of a proposition. This commitment can be separated into two claims: (i) there are cases of austere nonsense, and (ii) these are the only possible cases of nonsense. The first claim is unproblematic: a proposition can only have a meaning if its constituent parts have a meaning, from which it follows that if a word or phrase in a proposition lacks a meaning, the proposition also lacks a meaning. Seen this way, no one will seriously deny—and I don’t think anyone has—claim (i). The important and radical claim of the austere conception is (ii)—mainly, that if a proposition has no sense ‘this can only be because we have failed to give a meaning to some of its constituent parts’ (5.4733, my emphasis). This claim—we shall call it the ‘restrictive claim’. Is clearly a strong claim (or seems like one at least); and at first glance it seems unlikely that it can be defended without presupposing some substantive, theoretical account of the essence of language. It is therefore in defense of this claim that the challenge really lies for proponents of the austere conception of nonsense.
The restrictive claim of the austere conception is based on the principle that there is no such thing as a nonsensical proposition in which its constituent signs can be assigned symbols. For, if true, it follows from this principle that there are no possible cases of substantial nonsense, for if nonsense cannot arise due to the illegitimate use of symbols, then there is no such thing as a nonsensical proposition in which a symbol occurs in the ‘wrong’ logical place. Wittgenstein puts the restrictive principle thus, ‘one cannot give a sign the wrong sense’ (5.4732). The task of uncovering the necessary and sufficient presupposition(s) of the restrictive claim is reduced to the task of uncovering the necessary and sufficient presupposition(s) of the restrictive principle.
These presupposition will begin to become clear if we try to determine what, precisely, 5.4732 might mean. To understand 5.4732 we need at least a rough idea of what is for a word (a sign) to have, or fail to have, a sense or a meaning. Wittgenstein explains this in 5.4731, which is a comment on 5.4732:
Occam’s maxim is, of course, not an arbitrary rule, nor one that is justified by its success in practice: its point is that unnecessary units in a sign-language mean nothing
Signs that serve one purpose are logically equivalent, and signs that serve none are logically meaningless.
The basic point of this passage is that signs have a meaning only if they serve a purpose, only if they have a logico-linguistic use or function. It follows that, as Wittgenstein says in 3.328, ‘if a sign is useless it is meaningless’. Now, to hold that if a sign is useless it is meaningless is not necessarily to hold that use is meaning (though this might be sometimes true). What the claim does entail is that we cannot determine the meaning of a sign without determining its use: ‘in order to recognize the symbol in the sign we must consider the significant use’ (3.326). Determination of use is a necessary condition for determination of meaning. However, it is clear that that passage also suggests that the determination of use is a sufficient condition for the determination of meaning. This follows from statements like, ‘if everything behaves as if a sign had a meaning, then it does have a meaning’, and ‘signs that serve one purpose are logically equivalent’—for if determination of use would not result in determination of meaning, then e.g. that two signs are used in the same way would not necessarily imply that they have the same meaning. These clarifications show that for a word to have, or fail to have, a sense or a meaning, is (or goes hand-in-hand with) for a word to have, or fail to have, a use, a purpose or function in a proposition. We are now in a better position to understand 5.4732. To hold that we cannot give a sign the wrong sense is to hold nothing more than that we cannot give a sign the wrong use.
With this we can redefine the restrictive claim of the austere conception as follows: nonsense can arise only because we have failed to give a use to one or some of the constituent words of a proposition. It will not be hard to see how this follows from the assumption that (a) to determine the meaning or symbol of a sign, we have to determine its use, its role or function in a significant proposition (we are concerned with the nonsecality of propositions). For since we cannot determine the use or function of a sign in a proposition that has no sense, we cannot give a sign a meaning or a symbol in a proposition that is nonsense, and thus nonsense cannot arise because of the way a sign(s) symbolize(s). It follows (since cases of austere nonsense are possible) that nonsense can only arise because we have failed to give a use and thus a meaning to the constituent word/s of a proposition. Clearly, then, (a) is the necessary and sufficient presupposition of the restrictive claim of the austere conception.
This is all quite abstract; it is in need of illustration, and not only for clarification. One can agree with this abstract derivation and still doubt the conclusion that (a) is all the restrictive principle (we cannot give a sign the wrong sense) needs—for it might be the case that the need for more assumptions, to preclude the possibility of cases of substantial nonsense, becomes obvious (as some critics argue) only when dealing with actual examples (examples that might seem like counter-examples to the view we are here defending, that e.g. invite a ‘substantial’ explanation of their nonsecality). I will try to address this worry by discussing some examples, and this will clarify the relation I am here defending between the claim that one cannot determine meaning without determining use and the claim that one cannot give a sign the wrong sense.
The first example I want to consider appears in 5.473:
Logic must look after itself.
If a sign is possible, then it is also capable of signifying. Whatever is possible in logic is also permitted. (The reason why ‘Socrates is identical’ means nothing is that there is no property called ‘identical’. The proposition is nonsensical because we have failed to make an arbitrary determination, and not because the symbol, in itself, would be illegitimate.)
In a certain sense, we cannot make mistakes in logic.
Before beginning to interpret this passage let me make a clarificatory point. According to the austere conception of nonsense, nonsense arises because of failure to determine the use of at least some of the constituent words of a proposition. It also holds that all nonsensical sentences are constituted by nothing but meaningless words. These two formulations are similar but not the same. The first formulation is a condensed—I would even say partial—description of why nonsense arises; the second formulation is a description of what a nonsensical string of words is. Each formulation is an answer to a different question: the first answers the question, ‘why does nonsense arise in apparently meaningful expressions?’, and the second answers the question, ‘what is a nonsensical sentence constituted of?’ (I mention this distinction, among other things, because otherwise it might seem that what Wittgenstein says in 5.473, about the string ‘Socrates is identical’, is not consistent with some of the things I said about austere nonsense.)
Wittgenstein says that the reason why (1) ‘Socrates is identical’ is nonsense is that there is no property called ‘identical’. It might seem as if he implicitly assumes that ‘Socrates is’ has a meaning outside a proposition, which is inconsistent with the claim that we have to determine use before determining meaning. However, in this passage Wittgenstein is not referring to what constitutes the nonsensical string of words ‘Socrates is identical’. He is referring to the reason why it is nonsense (5.473: ‘the reason why…’)—we can therefore take his brief remark as a condensed description of how we come about to the recognition that ‘Socrates is identical’ is nonsense. Wittgenstein assumes (quite naturally) that when analyzing (1) the only possible or most plausible way of trying to give it a sense is by taking or trying to see if ‘Socrates is’ symbolizes in (1) in the way it symbolizes in proposition like ‘Socrates is a philosopher’, ‘Socrates is an Athenian’, etc., that is, in the way it symbolizes in propositions that ascribe a property to Socrates. Now, once we take ‘Socrates is’ in that way, the only problem with (1), according to Wittgenstein, is that ‘identical’ does not stand for a property. Wittgenstein never hints at the suggestion that we should consider the possibility that ‘identical’ stands for a symbol of a certain sort of relation (between signs, symbols, or whatever). Indeed, in 5.4733—a comment on 5.473—Wittgenstein says that when ‘identical’ occurs as the sign of identity ‘it symbolizes in an entirely different way’, and, again, does not consider the possibility that ‘identical’ in ‘Socrates is identical’ symbolizes in the way it symbolizes in ‘1 is identical to 3-2’, and even seems to suggest that we cannot (not only that we are not inclined to do so in this case) take it as symbolizing for the sign of identity (again, this is assuming we want to take ‘Socrates is’ in its ordinary way), otherwise his statement would be false. In other words, Wittgenstein talks as if in taking ‘Socrates is’ in its ordinary way, we are thereby (within certain parameters) deciding how we want to take ‘identical’ (there are not two distinct acts of symbol assignments).
The key to understanding this move—i.e. why ‘Socrates is identical’ is nonsense, given the way we want to take ‘Socrates is’, only because ‘identical’ does not stand for a property, and not because certain symbols are incompatible—is to apply, in our analysis, the claim that we cannot determine the meaning of a sign without determining its use. This will also help us understand why this example appears between the claims that ‘logic must look after itself’ and that ‘in a certain sense, we cannot make mistakes in logic’, and why, most importantly, the restrictive principle—‘we cannot give a sign the wrong sense’ (5.4732)—appear as a comment on 5.473.
So we want to use ‘Socrates is’ in (1) in the way we ordinarily use it, such as in propositions like, ‘Socrates is a philosopher’ and ‘Socrates is wise’, were ‘Socrates is’ is used to ascribe a certain property to Socrates. Now, ‘Socrates is’ is obviously not used as a property ascription phrase in places where it does not ascribe a property to Socrates (e.g. ‘Socrates is a nice name’). So: to determine (or stipulate) how ‘Socrates is’ symbolizes in a particular proposition we have to determine (or stipulate) the kind of role or use it has in that context. Accordingly, to try to see if ‘Socrates is’ in (1) is used as it is ordinarily used, requires that we determine if it is functioning as a property ascription phrase, that is, if it is ascribing a property to Socrates. But ‘Socrates is’ cannot ascribe a property to Socrates—cannot therefore be used as a property ascription phrase—unless it is ascribing a property to Socrates, unless, therefore, what goes in ‘-------’ in sentences of the form ‘Socrates is-------’ is a property. This is what it means for ‘Socrates is’ to be used, and hence to symbolize, in its ordinary way, which is the way we want to take the phrase. Now, since in (1) the sign ‘identical’ is in the place of ‘-------’ in a sentences of the form ‘Socrates is-------’, the only question left to ask (the only thing to determine), once we have determined to take the phrase in its ordinary way, is, ‘is ‘identical’ the name of a property?’ (The possibility of ‘identical’ standing for the sign of identity never arise once we take ‘Socrates is’ as a property ascription phrase.) The answer to that question is simple—we have never given a property the name ‘identical’, and so ‘Socrates is’ is not ascribing a property to ‘Socrates’, and therefore is not used as a property ascription phrase. It follows that ‘Socrates is’ has no use (hence no meaning), or we cannot determine its use (which comes to the same thing), and what remains of (1), logically speaking, is nothing but unused, empty words. This shows how smoothly it all follows if we keep (a) in mind through the analysis.
This explains both why Wittgenstein never consider the possibility that ‘identical’ stand for the sign of identity in ‘Socrates is identical’ and why he places this example where he places it. This example could easily have been taken by Wittgenstein to illustrate a category clash, instead it is taken to illustrate that, in a certain sense, one cannot make mistakes in logic. Taking ‘Socrates is’ as used in a certain way in ‘Socrates is-----’ is to take whatever goes in ‘------’ as the sort of thing that makes that former use possible (this is the ‘Kantian’ way to put it). So the only question is, ‘does what goes in ‘--------’ stand for the sort of thing that makes that use possible?’ And the only thing that can here go wrong is that we have failed to make that arbitrary determination. There is no place open to give the sign the wrong sense.
I should like to say a little more about the claim that if we take a sign to be used in a certain way, we condition the use and thereby the meaning that the rest of the proposition can have or stand for. For it might seem (as some commentators insist) as if we got rid of the possibility of ‘meaning-clash’ through a bit on question begging, whereas I want to show that if we become clear of what determining a use is, the very notion (or pseudo-notion) of a ‘meaning-clash’ should begin to loose its hold on us. The sort of explanation we need is one that releases us from the temptation to ask ‘why can’t identical stand for the sign of identity in (1)?’ This will also allow me to rephrase some slightly misleading expressions I’ve been forced to use.
To address this worry it will be useful to first see how contexts preclude the use that we can give to whole sentences. This will help us see why assigning a use to parts of a proposition is partly the act of restricting the kinds of uses we can assign the rest of the signs (because it presupposes that we assign those other signs certain sorts of uses), without either begging the question or implicitly relying on the notion of a clash between meanings or logical categories of symbols.
Consider the expression ‘I am here’. There are different uses that we have for this expression, different linguistic acts we perform with it. This is one case: someone enters my house, calls me out, and I reply ‘I am here!’ to guide him by the sounds to my spatial position, or perhaps to let him know that I am in the house and will soon go to where he is. Here is a different case: I’m giving a lecture, everyone in the lecture room is seeing me and aware of my presence, and I suddenly scream ‘I am here!’. Perplexed, someone from the audience asks me, ‘what do you mean?’ Suppose I say, ‘I said that to inform you that I am here’. That is, I want to use the phrase in the same way that I used it in the previous example, to inform someone of my presence. However, in the first case ‘I am here’ had or could be put to such a use because I could use it to tell someone that did not know if I was in my house that I was. The phrase can be informative in that way only because, in certain contexts, it can be used to convey information. In the second case, however, I use the phrase in a context in which everyone who hears the phrase knows were I am; thus to say ‘I am here’ cannot be a way of conveying the kind of information (i.e., new information about my spatial position) that I conveyed or could convey in the first case. I can’t use the phrase to convey information because it is being uttered outside a context in which that phrase can have or be put to such a use, as when the person who hears me lacks information about my spatial position. But this is just another way of saying that the circumstances make it impossible that that phrase might have the use it has in the first case: whatever my intention is, it is impossible to convey to someone new information about something she already knows, for this conveying of information is made possible by the context, not (only) by my linguistic intentions. The uses signs can have are dependent on and limited by the relations they have to a context—and thus certain contexts may preclude certain kinds of uses that a phrase or sentence might have in other contexts.
The case of sentences and the relation between their parts is quite similar. Take the expression, (2) ‘Scott kept a runcible at Abbotsford’, and assume we know two uses for the phrase ‘Scott kept a…at Abbotsford’: (i) the use it has in sentences like, ‘Scott kept a good humor at Abbotsford’, and (ii) the use it has in sentences like, ‘Scott kept a cow at Abbotsford’. In the first cases we use it to say something about Scott’s state (psychological or physiological) at Abbotsford; in the second cases we use it to say that Scott kept some object at Abbotsford. Thus, the use that the phrase ‘Scott kept a…at Abbotsford’ has, in sentences of the form ‘Scott kept a------at Abbotsford’, depends on the context in which it is embedded, and that context is provided by what goes in ‘-------’ (just like the use the expression ‘I am here’ can have depends on the context in which it is uttered). So to assign one use rather than the other to ‘Scott kept a…at Abbotsford’ is to assign it one context rather than the other, i.e., is partly the decision of what sort of thing will go in ‘-------’. If we want to use it in the way it is used in (i) what we are left to determine is what (if any) sort of mood does ‘runcible’ stand for; and if we want to use it in the way it is used in (ii) what we are left to determine is what sort of thing does ‘runcible’ stand for. To decide what use it has is not different from deciding what context we want to embed it in, for the possibility of the use of an expression depends on its contextual embedded-ness (we had forgotten a Kantian question).
What is important to note here is that in assigning a use we are not performing two distinct linguistic acts (where a sort of logical error or the desire to achieve one might creep in): first assigning a certain use to ‘Scott kept a…at Abbotsford’, and then restricting the sorts of uses that we can give to whatever goes in ‘…’. Assigning a use to ‘Scott kept a…at Abbotsford’ is embedding it a certain sort of context, is, that is, assigning a certain sort of use to ‘…’.
The mechanism should now be clear: assigning uses to expressions is assigning the sorts of contexts that makes those uses possible. To talk about use is to talk about a relation to something else: assigning a use to one part of the proposition is assigning certain sorts of uses to the rest (it would be as mistaken to hold that ‘Socrates is’ is or can be used as a property ascription phrase in contexts in which it does not ascribe a property to Socrates as to say that ‘I am here’ is or can be used to convey information in places where it does not convey information). This then is how assigning a use to one part of a proposition necessarily restricts—or rather, is partly the act of restricting—the kinds of uses that we can assign the rest of the proposition. There is here no open hole through which a ‘logical’ error might creep in, so we should no longer be tempted to think that there is something that needs to be taken care of (we should be as tempted to ask, ‘why can’t “identical” be the sign of identity in (1) if “Socrates is” has the use it ordinarily has?’, as to ask, ‘why can’t “I’m here” be used to convey information in situations where everyone knows where I am?’). In a way, the phrase ‘we cannot give a sign the wrong sense’ begins to dissolve on us. It was useful when we were not clear on these issues because it directed us to try to get clear on what precisely assigning uses to signs involved. But now that we are somewhat more clear on that, we no longer have the temptation to think that there can be such a thing as assigning the wrong uses, and hence wrong modes of symbolization, and not because we cannot make mistakes, but because there is no room in which to make those sorts of mistakes. Logic takes care of itself, for us, the moment we begin to see that there is nothing to be taken care of (and so this phrase too begins also to loose its sense) (cf. Mounce 1989: 104-105).
These examples help us sharpen our previous conclusion, mainly, that all the restrictive principle—the principle that one cannot give a sign the wrong sense or use—needs is the assumption that the determinations of meaning presuppose determinations of use, determinations of the role or function of a word or phrase in a proposition. But by the restrictive principle we should no longer understand the holding of the proposition, ‘we cannot give a sign the wrong sense or use’, but rather the process of getting clear on what determining uses involves in such a way that the notion of the possibility of assigning wrong uses (and hence meanings) becomes incoherent. (I will, however, keep using the phrase ‘restrictive principle’ for brevity.)
* Guillermo Del Pinal, Magister Artium en Filosofía; Universidad de Chicago
 Here is an example where the problem is clear:
It cannot be said that 1 is a number, or that a is an object, but these inexpressible necessary truths show themselves in ordinary propositions saying that a is an F or that there is only one F. (Hacker 2001: 326)
We can see clearly the structure of the claim: it cannot be said that X, but that X shows itself in a certain kind of ordinary proposition. Firstly, if we do not understand what X means here, we do not understand what shows itself; secondly, if we do not understand what X means here, we cannot establish that what is shows in a proposition of a certain kind is in any way connected to X (so as to say that it shows what X tries to say).
 See e.g. Diamond 2001 and Conant 2002.
 Critics sometimes take the resolute reading to hold (i) that all the propositions (almost meaning, ‘all the sentences’) of the Tractatus are plain nonsense, and (ii) that the Tractatus holds an austere conception of nonsense. They then point out that to defend (ii) resolute readers rely on sentences in the body of the Tractatus, which clearly conflicts with commitment (i). However, most resolute readers do not hold (i).
 For a related worry, see Sullivan 2002: 54.
 What makes a statement part of a ‘philosophical doctrine’ does not only depend on the statement itself. In a certain historical-context, the statement ‘men are formed of atoms’ was very philosophical, it was (or close to) a metaphysical speculation. Thus one cannot class a statement as ‘metaphysical’, ‘philosophical’, ‘scientific’, ‘trivial’, etc., without first determining its relation to other statements in a historically-embedded theoretic-linguistic framework.
 We can divide the foundations-objections against the austere conception into three groups. Objections of the first kind hold that the austere conception rests on claims or theses advanced in the body of the Tractatus (e.g., 3.3, 4.126-4.1272, and 5.473-3) which are supposed to be recognized as nonsense (see Hacker 2001 and 2002). Objections of the second kind hold that the austere conception rests on substantial (and problematic) philosophical theses either advanced in Tractatus or taken, sometimes without careful justification, from the works of Russell and/or Frege (see Hacker 2003 and Williams [a]). Objections of the third kind are, more than objections, a worry about the sort of foundation the austere conception seems to require (see Hacker 2002: 362 and 2001, Sullivan 2002: 54, Williams [forthcoming], Gustafsson 2003). In this paper I respond to all of these objections.
 This paper is more indebted than I have been able to appropriately cite to essays by J. Conant and C. Diamond, esp. (Conant 2002) and (Diamond 2001).
 N: 11e
 These definitions are taken from Conant 2002 (400). The words ‘sign’ and ‘symbol’ are also used following Conant’s definitions:
Sign: an orthographic unit, that which the perceptible expressions for propositions have in common (a sign design, inscription, icon, etc.)
Symbol: a logical unit, that which meaningful propositions have in common (i.e., an item belonging to a given logical category: proper name, first level function, etc.).
 It might be that some proponents of the austere conception will deny that we have to ‘establish’ the restrictive claim, or that the restrictive claim should be part of the ‘austere conception’ at all (though most proponents of the austere conception of nonsense do, at lest implicitly, take the ‘only’ here as constitutive of their position). I think that the austere conception needs the restrictive claim. For one thing, the restrictive claim is intimately tied to the notion of philosophy as an activity of clarification (but of this later). It is also clear that many of the objections raised against the austere conception are aimed at this strong negative thesis (though perhaps critics are not always explicit about this). That is, the problem is not that there are cases of austere nonsense (this is obvious), but the rejection of other kinds of cases (without assuming stronger theses).
 One might suspect that taking use as a condition of meaning is inconsistent with the view of meaning of the picture theory, but this is not so. Proper names stand for objects. But proper names can only stand for objects if they enter into a pictorial depiction of reality. For a sign to stand for an object is for it to have a certain kind of use. The kind of use is that of having a certain function—that of standing for a particular object—in a propositional representation of some state of affairs. This is the sort of use that makes the sign a proper name. We can then say: the meaning of names is the objects for which they stand only because the use of the signs for names is to stand for objects in propositional (linguistic) representations of states of affairs.
 One might think that these passages suggest a stronger connection between meaning and use, that e.g. we cannot separate the two as if they were distinct entities, or at least we cannot not do this for most kinds of symbols (e.g. logical connectives). I think this might be right, and I do not want to distance myself from this reading (see Kremer 1997). There are cases (e.g. proper names), however, in which it is useful to keep this distinction (talking about meaning and use as if they were different yet intimately connected). In any case, for our purposes all we need is the claim that only by determining the use of a sign can we determine its meaning. The closer we take the connection between use and meaning the more obvious this claim becomes; and if we hold that meaning should simply be identified with use, or that use is absolutely constitutive of meaning, then the claim that we cannot determine meaning without determining use becomes a tautology, and all that I say here and in what follows will simply become much more obvious.
 Hence the need for the previous claim that if use is determined meaning must be determined. Since nonsense can only arise through failure to determine meaning, if it was not the case that, necessarily, if use is determined meaning is determined, then this re-definition would not work because it could be the case that use is determined but not meaning, in which case the proposition would still be nonsense, though use has been determined.
 It seems to me that some confusion in debates between commentators arises due to failure to distinguish these questions. Suppose we have two nonsensical strings (1) ‘Juwiwalera ist trrttt’ and (2) ‘Caesar is a prime number’. If we are asked, ‘what are (1) and (2) constituted of?’ we would reply (if we hold an austere conception, and assuming we were not able to assign them meanings), ‘nothing but empty words. There is of course a psychological distinction—e.g. we might at first think that (2) is meaningful, or we might associate certain images with (2), whereas (1) might cause no similar reactions to us.’ Now, suppose we are asked, ‘why does nonsense arise in (1) and in (2)?’ Then it seems obvious that we will disappoint the questioner by answering, ‘because the words do not have meanings’. Surely that is part of the answer, the last ‘therefore’ in the argument, so to speak. But although this answer will not disappoint us in the case of (1), it will certainly disappoint us in the case of (2). The words constituting (2) ordinarily have a meaning, and we are therefore expected to explain why they do not stand for their ordinary meanings in this case. This requires a sort of logical explanation (in this sense the distinction is not only psychological, though this of course does not refer to the constituents)—an explanation of the sort I give in the rest of this section—that is not necessary when explaining the nonsecality of (1). (One way we can see the difference is this: (1) does not provide us with an example of a case in which logic takes care of itself, whereas (2) does. Of course the point in the end is to see that there is nothing to be taken care of but we have to first get there.) It is mostly when commentators (critics) take the claim ‘a proposition is nonsense because its constituent words do not have a meaning’ to be a complete answer to the question ‘why does nonsense arise?’, when asked about a proposition whose constituent words ordinarily have a meaning, that they find that answer somewhat absurd, or perplexing, or, at best, incomplete. Hacker e.g. is sometimes willing to concede that as a response to the question ‘what is a nonsensical proposition constituted of?’ one can say that the words in a nonsensical proposition have no meanings, and thus that there is no logical difference between (1) and (2); but as a response to the question ‘why does nonsense arise?’ he seems to think we have to engage in a logical discussion that is different in the case of (2) from the case of (1) (and that requires an appeal to something like the substantial conception, that is, to a clash between rules or meanings). So what we have to show, if we want to satisfy these critics, is why we can give a complete answer to questions of the form ‘why does nonsense arise?’ when asked about cases like (2), without having to appeal to a notion of a clash of the logical category of symbols or anything of the sort. Clearly it will not do simply to say, ‘the reason why it is nonsense is because some sign does not have a meaning’. For then the critic will say, ‘but it does have a meaning! What you have to explain is why it cannot have its ordinary meaning in this proposition.’ This is what I will try to do in the following discussion of ‘Socrates is identical’. (To be clear, the logical distinction between (1) and (2) is not in the propositions themselves--it is not as if one is more ‘meaningful’ than the other--but in the explanation of their nonsecality.)
 In this case we can stipulate a use because (we are assuming) there is no other context restricting the phrase ‘Socrates is’. But this stipulation, as we shall see, is not only one of use, but also of the context that makes that use possible. Or rather, the stipulation of the use is also the stipulation of the context in which that use is possible.
 We could put it a little stronger: to take ‘Socrates is’ as it is ordinarily used, is to take what goes in ‘-------’ in sentences of the form ‘Socrates is-------’ as a property. I will explain this move in more detail in what follows.
Wittgenstein says (3.311-3.12): ‘an expression characterizes a form and a content’; ‘an expression presupposes the form of all the propositions in which it can occur’; ‘it is therefore represented by the general form of the propositions which it characterizes’. The form that the expression characterizes is the form of the propositional context of which it is part. We can represent it, in the case under discussion, like this: ‘Socrates is PROPERTY’ (as Kremer says, this abbreviation is entirely parasitic on the set of sentences like, ‘Socrates is an Athenian’ etc.). We can see this form as determining the kinds of contents we can give to it. So if the symbol goes hand-in-hand with its form: the only question left to ask is, ‘do the (other) signs stand for the symbols of that form?’
 Only by coming to see that there is nothing going on that needs to be taken care of can we get rid of this question. The metaphysical gain (Reid 1998) of this insight goes hand in hand with the metaphysical gain that we do not have to justify why we supposedly can’t do certain things (once we see that there is nothing there that we ‘can’t’ do). Ultimately the austere conception cannot be divorced from Wittgenstein’s conception of logic (and vice versa). As long as we still see things in such a way that we ask that sort of question, logic seems to be in need of justification (just like as long as we are tempted to ask why a contradiction cannot be true logic is seen as in need of justification). To come to see that logic takes care of itself is to come to see that there is nothing there that needs to be taken care of.
 The discussion of this example is greatly indebted to Hertzberg’s ‘The Sense Is Where You Find It’.
 Merely from these two cases we can begin to see how the meaning of this expression changes if the use changes, which further confirms the strong connection we have drawn between the two. We can see this by specifying the contexts a little; say, in the first case the other person knows I am in the house, so that the expression means something like what ‘come here’ would mean in similar circumstances, and in the second case the other person does not know I am in the house so that the expression means something like what ‘I’m here in the house; I’ll be there soon’ would mean in similar circumstances.
 Of course, it does not follow that I am speaking nonsense; the phrase can have other uses (e.g., ‘put attention to me!’).
 This example is used by Diamond in her essay ‘What Nonsense Might Be’. But she uses the example for a slightly different purposes.
 It is important to note how well the austere conception of nonsense fits some of Wittgenstein’s ideas about logic. I am thinking mainly of the idea that logic does not need a justification because, once we see it in clear light (e.g. that in logical inferences there is strictly speaking no move going on because the sense of the consequent is contained in the sense of the antecedent), we will no longer be tempted to think that there is anything in logical operations that needs a justification. If we see clearly that the ‘laws’ of logic are tautologies, questions of justification loose their sense. A similar idea, I think, underlies the austere conception of nonsense. If we are only clear about the use of our words, we will not be tempted to ask why certain signs can’t be combined with certain other signs. On the other hand, if we hold a substantial conception of nonsense, if we hold that we can be clear about the meaning of our words and still wish to combine them in ‘illogical’ ways, then questions of justification arise (e.g. Carnap reads Heidegger as thinking that he can break the fundamental rules of logic, and thus he reads him as implicitly demanding a justification for why what he is trying to do cannot be done, and, of course, as Conant argues, to tell him that he is breaking the rules of syntax would not convince him of anything, he knows that).
[IMPORTANT. At 4.243 Wittgenstein says that if I know the meaning of X (a German word) and Y(an English word), and they are synonyms, it is impossible for me not to know that they do mean the same. By the same token, if I know the meaning of a word X’ and a word Y’, and they are not synonyms, it is impossible for me not to know that they are not synonyms. It follows that sentences like ‘Socrates is Plato’ (taking ‘is’ as the sign of identity) can never be cases of substantial nonsense (for we would have to know the meaning of those words, yet not know that they are not identical so as to make the assertion). I think this can be extended to all cases. E.g. if I know the (ordinary) meaning of ‘number’ and I know the (ordinary) meaning of ‘Caesar’ it is impossible for me not to know that they are different. It follows that ‘Caesar is a number’ can never be a case of substantial nonsense (for I would have to know the meaning of those words, and not know that they are different). I think that from Wittgenstein’s claims on identity (with a little work) we can show that (most) cases of substantial nonsense can never arise.
 (One way, perhaps somewhat more intuitive, to compare both accounts of nonsense is the following. Suppose someone says that ‘there is a possible world in which 7 is green’, and goes on to explain that by ‘green’ he means ‘a certain sort of property that physical things have in virtue of having an extension in space’, and that by ‘7’ he means ‘a certain non-spatial abstract entity’. If one holds a substantial conception one is at this point almost justified in assuming he is speaking nonsense, that is, he knows what his words mean but wants to combine them in certain illogical ways (like e.g. Heidegger who supposedly knows what ‘nothing’ means but wants to assert things that can only be asserted of objects). If you hold an austere conception then e.g. (this is just one interpretation) that he wants to ascribe a quality of extended things to something that is not extended shows that he does not know the meanings of those words, or is not here using them how we ordinarily use them (for we would say that to know what a spatial entity is, is (partly) to know that it cannot be the same as a non-spatial entity). For on what possible grounds can we assume that someone who holds that a non-extended thing is colored understands the (ordinary) meanings of those words but wants to use them illogically? He can tell us right before the sentence all sorts of explanations, and as a last resort we can assume that he forgets things he says by the time he says others, and that would be a more justified assumption that saying he understands the words but wants to use them illogically. In other words, we can never establish how a person is using his words if we do not sort of understand what he is saying (if it is ‘illogical’ or whatever), for the only thing that justifies our assuming that someone is using words in their ordinary ways is that they are indeed using them in their ordinary ways. (Thus proponents of the substantial conception make a very implausible philological-exegetical assumption.)