Reading The Tractatus with G. E. M. Anscombe
Abstract and Keywords
Ludwig Wittgenstein intended the Tractatus to reshape our understanding of what philosophy was and what it could accomplish. In the decades after its publication, it was widely admired and deeply influential, but almost totally misunderstood, or so argued Elizabeth Anscombe, in An Introduction to Wittgenstein’s ‘Tractatus’. She wanted her book to inaugurate a radical change in how the Tractatus was read and understood. This chapter looks at her critique of older readings and at her own approach to the Tractatus. I argue that her presentation of the picture theory points us to fundamental connections between Wittgenstein’s thought and that of Gottlob Frege, and that it exemplifies the kind of philosophical activity to which the Tractatus was meant to lead. Working through her arguments and following out ideas implicit in her practice can help us to grasp how Wittgenstein’s book was meant to revolutionize philosophy.
Wittgenstein wrote in the Preface to the Tractatus (TLP) that he believed the book to show that the reason philosophical problems are posed is that ‘the logic of our language is misunderstood’. At the end of the Preface, he said that he took himself to have in essence arrived at a definitive solution of such problems. The book was meant to revolutionize philosophical thinking. In An Introduction to Wittgenstein’s ‘Tractatus’ (1963), Elizabeth Anscombe wrote that almost everything that had been published about the Tractatus ‘has been wildly irrelevant’.1 She meant to inaugurate a rethinking of what sort of book the Tractatus was and how it should be understood. In this chapter, I want to look at these two intentions together: Wittgenstein’s intention of bringing about a revolution in philosophical thinking and Anscombe’s of inaugurating a radical change in how the Tractatus was read. Looking at Anscombe’s intention and at what she does in carrying it out will help us to see what Wittgenstein was hoping to achieve.
In the introduction to her book, Anscombe explains why virtually all that has been said about the Tractatus has been fraught with misunderstanding. But the story she tells has problems, and they are the subject of section 30.1 of this chapter. If, however, we look at Anscombe’s account of the picture theory, we can take it to demonstrate how she thought the Tractatus should be read; and there are lessons to be drawn about what was wrong with the kinds of reading she rejected. The transformation which Anscombe hoped to bring about in how the book was read helped to transform also the study of both Frege and Russell. This latter transformation can help us to see what underlies her claim in the introduction that it was neglect of Frege and over-dependence on Russell that led to the irrelevance of so much of what had been written about the Tractatus. An approach to the Tractatus can be Russellian in two different senses. It can depend upon reading Russell himself as a Humean thinker, as ‘imbued’ with empiricism; or it can take (p. 871) from Russell, or at any rate share with Russell, a form of thinking which goes deeper and is not dependent on his empiricism, however nicely it fits with empiricism. I mean the kind of thinking which is described (by Warren Goldfarb and Peter Hylton) as Russell’s ‘object-based’ approach to metaphysics and meaning, and which can be contrasted with Frege’s ‘judgment-based’ approach.2 Anscombe’s reading of the Tractatus is un-Russellian, in both senses. Section 30.2 is meant to lay out the issues here, first by summarizing Anscombe’s account of the picture theory, and then by showing the importance of the contrast between ‘object-based’ and ‘judgment-based’ accounts of meaning—its importance for understanding what Anscombe was trying to achieve, and how she differs not just from the readers of Wittgenstein whom she plainly did have in her sights, but also from such later readers as David Pears and Norman Malcolm. Her account of the contrast between Russell and Frege, in her argument for the importance of not neglecting Frege, does not reach to the largely implicit understanding of the contrast that is evident in her treatment of the picture theory. I shall not argue for this, but it seems to me that the character of the contrast begins to come out explicitly in Hidé Ishiguro’s ‘Use and Reference of Names’ (1969). Much of the criticism of that essay (for example in Malcolm 1986, ch. 2) can be seen to be directed at a view which is already present in Anscombe’s book. Section 30.3 of my chapter is about how to understand Anscombe’s achievement, not just as a reader of Wittgenstein but as someone engaged in the practice of philosophy as Wittgenstein conceived it. The larger aim of this part of the chapter is to show how Anscombe’s philosophical practice can lead us into further questions about Wittgenstein’s understanding of philosophical activity. Sections 30.4 and 30.5 are about how we can be helped thereby to see the aims and achievements of the Tractatus. One of my aims in this chapter is to suggest a revision in the history laid out by Warren Goldfarb in ‘Das Überwinden: Anti-metaphysical Readings of the Tractatus’ (2011). I shall have some brief words about that in the concluding section, 30.6.
30.1 Anscombe, Russell, Frege
An Introduction to Wittgenstein’s ‘Tractatus’ was published in 1959, and drew on material that Anscombe had developed in lectures over successive years up to 1957–8. In his 1956 study of the history of philosophical analysis, J. O. Urmson presents an interpretation of the Tractatus of exactly the sort Anscombe criticizes.3 He mentions that some philosophers (unnamed, but presumably including Anscombe) have called into question the kind of interpretation of the Tractatus which he presents. He goes on that, whatever the accuracy of that interpretation, it is the ‘received’ view, the view generally accepted in the period going up to the Second World War (pp. ix–x). A central feature of the interpretation he presents is that the Tractatus is read with Russell’s Lectures (p. 872) on Logical Atomism; both works are seen as fundamentally Humean in character. Both Russell and Wittgenstein are seen as empiricists, updating empiricism with the aid of recent developments in logic.
That reading of the Tractatus is the main target of Anscombe’s remark about the irrelevance of most of what had been written about the book. If this irrelevance has any one cause, she says, the cause is neglect of Frege ‘and of the new direction he gave to philosophy’ (IWT, p. 12). She adds that ‘empiricist and idealist preconceptions, such as have been most common in philosophy for a long time, are a thorough impediment to the understanding of either Frege or the Tractatus’ (pp. 12–13). She makes a contrast between Frege and Russell, the point of which is that readers of Wittgenstein have tended to see him as resembling Russell in respects in which he is much closer to Frege. Frege, she notes, is engaged in inquiries that are ‘in no way psychological’; he had no interest in ‘private mental contents’, while Russell, unlike Frege, is concerned with immediate experience and with private mental contents, and introduces those notions into his account of language and his theory of judgment. He is ‘thoroughly imbued with the traditions of British empiricism’; many readers of Wittgenstein share that background with Russell, and it leads them to misunderstandings of Wittgenstein’s concerns in the Tractatus (p. 14). In the following chapter, Anscombe develops further her account of the usual, and (as she sees it) deeply mistaken, reading of the Tractatus. She quotes Karl Popper’s summary of the Tractatus, which ascribes to Wittgenstein a version of the verifiability criterion of meaning. Popper treats Wittgenstein’s Elementarsätze as statements describing directly observable states of affairs; and adds that, for Wittgenstein, ‘every genuine proposition must be a truth-function of and therefore deducible from, observation statements’ (quoted on p. 25). Anscombe mentions that Popper’s account fits with a further feature of the usual reading of the Tractatus, which sees it as combining two independent theories: a ‘picture-theory’ of elementary propositions and a truth-functional account of non-elementary propositions. Before she turns to her own positive discussion of Wittgenstein on elementary propositions, Anscombe argues briefly against Popper’s view of them as observation statements like, for example, ‘Red patch here’, and mentions another respect in which the empiricist tradition may lead to misreadings of the Tractatus. Unlike Russell, and unlike the logical positivists and contemporary British readers of Wittgenstein, Wittgenstein was not concerned with epistemological issues, which he took to be irrelevant to ‘the foundations of logic and the theory of meaning’ (pp. 25–8).4 Anscombe’s account of Wittgenstein’s view of epistemology is meant to connect with her earlier remark about the cause of the misreading of the Tractatus being neglect of Frege and ‘the new direction he gave to philosophy’. That is, for the ‘old direction’ in philosophy, epistemology is central; but the questions with which Frege is concerned, and which we should see as significant for Wittgenstein, are not epistemological.
In explaining the sorts of question with which Wittgenstein was concerned, Anscombe gives as an example the question of the relation to reality of what I say, as for example if (p. 873) I say that Russell is a clever philosopher. The relation cannot be explained in terms of the truth of what I say, since even if my statement had been false, it would still have said something. As Anscombe notes, Wittgenstein was concerned with this problem throughout his life. She mentions that Russell discusses many of the problems that Frege discusses, and indeed this last question, which is meant to exemplify Wittgenstein’s sort of concern, is a central question for Russell. What then of her contrast between a reading of the Tractatus which sees it as Russellian in its approach to philosophical questions and one which sees it as Fregean? The contrast (if it cannot rest on a difference between Russell’s sort of question and Frege’s) has to rest on the difference between Russell’s psychologism and Frege’s anti-psychologism, and between the direction of philosophy as understood in the light of Russell’s empiricism and the new direction given to philosophy by Frege. But it is far from clear that the contrast can be made out in that way. Peter Hylton (1990) has argued in detail that the conception of Russell generally accepted by British philosophers involved a misleading assimilation of his views in the period before the First World War to those of traditional empiricism, and Anscombe herself takes the view of Russell criticized by Hylton. She passes by Russell’s anti-psychologism and the complicated character of his move away from it. Here I want to quote part of a paragraph of Hylton’s about Platonic Atomism, the view developed by Moore and Russell when they gave up the Idealist views which they had earlier accepted. The period about which Hylton is writing includes the period during which Russell wrote Principles of Mathematics:
The anti-psychologism of Platonic Atomism…is complete and thoroughgoing. Platonic Atomism does…imply or suggest a picture of the mind and its capacities, but this picture is very much a by-product of the view. There is no overt concern at all with the nature of thought or the mind or experience, in any sense. It is not that Moore and Russell are concerned to advance a view of these notions which is different from that of the Idealists, it is rather that these notions almost cease to be the subject of explicit philosophical concern. This seems to be because the notions are looked on as psychological, and for this reason of no interest to philosophy.
(1990, p. 108)
Hylton’s remarks suggest that, if we look for a source of Wittgenstein’s anti-psychologism in the Tractatus, Frege’s views are no more obviously the source than are Russell’s, though Russell begins what Hylton calls a ‘turn towards the psychological’ in the years after the publication of Principles. If Russell is the source, it would be Russell in his Platonic Atomist or idealist periods. Kant’s anti-psychologism is itself in the background of Frege’s and in that of idealists like Bradley.5 It is not obvious that Wittgenstein’s anti-psychologism in the Tractatus should be thought of as belonging to a new direction given to philosophy by Frege. Among idealist ‘preconceptions’, one might be said to be that a main thing wrong with empiricism was its psychologizing tendencies.
(p. 874) Anscombe’s brief account of the importance of Frege for an understanding of the Tractatus involves also a problematic contrast between ‘empiricist and idealist preconceptions’ and the ‘new direction’ given to philosophy by Frege. We can see one of the problems if we consider Anscombe’s discussion of the question whether, when I say that Russell is a clever philosopher, I mention both Russell and what I say about him, that he is clever. If I do mention it, what is the connection between the two mentioned things? And, if I do not mention it, what account should be given of the words expressing what I say about Russell? This is one of Anscombe’s examples of the sorts of question which contrast with those which are central for us if we start off with empiricist or idealist preconceptions. But here we can note that the questions mentioned by Anscombe are important for Bradley, and indeed his discussion of them is famous; further, his account of judgment stresses questions about how judgment is related to reality, questions again of exactly the sort which Anscombe is suggesting we need to be struck by if we want to understand the Tractatus. Idealist preconceptions would not stop us seeing the force of such questions. The other part of the contrast between ‘empiricist and idealist preconceptions’ and the ‘new direction’ given to philosophy by Frege does not work much better. Epistemological concerns are supposed to belong to the ‘empiricist and idealist preconceptions’, but epistemological questions are not ignored by Frege. This point is perhaps clearer in the light of material in Frege’s Nachlass that was not available to Anscombe when she wrote the Introduction, but Foundations of Arithmetic would in any case suggest that Frege was deeply interested in the question of the source of our knowledge of arithmetic. The contrast with Russell on the matter of interest in epistemology is indeed complicated.6
Nothing that I have said would cast doubt on Anscombe’s argument that Popper and the logical positivists had misread the Tractatus, and that they were in part responsible for the prevalent misreadings of the book. But I have tried to show problems for her argument that it is neglect of Frege, more than anything, that underlies the irrelevance of most of what had been written about the Tractatus. The features of Frege’s views which she emphasizes can’t bear the weight of the argument. The sorts of question which she suggests we need to think about when we read the Tractatus, questions like that of how a proposition hangs together, and that of how thought and reality are related, are of concern to Bradley and (as Anscombe herself notes) also to Russell; they are not more especially problems that should be associated with Frege; they are in no way out of place in the thinking of those with ‘empiricist or idealist preconceptions’, if that is meant to cover Bradley and Russell. While anti-psychologism is profoundly characteristic of Frege, it can be found within the idealist tradition and in Russell’s idealist and post-idealist views; and the turn away from epistemology is by no means as marked in Frege as (p. 875) Anscombe’s discussion of the period suggests. Russell wasn’t the empiricist Anscombe paints him as being, though his views were becoming more like those of the empiricists during the period in which he was working with Wittgenstein. There are all sorts of problems with the picture Anscombe gives us, of idealists and empiricists on one side, with their preconceptions and their familiar sorts of question, and Frege on the other side, giving a new direction to philosophy, and asking questions much more like those of ancient philosophy than like those that had concerned earlier thinkers. What I want to argue is that, despite the fact that practically everything Anscombe says in sketching why neglect of Frege will lead us astray needs qualification, her intention of following out what she takes to be Fregean in Wittgenstein’s thought leads her right to the heart of the book. But how does she turn out to be right, if her account of the history is, as it stands, unconvincing?
30.2 Anscombe’s Reading of the Tractatus
The heart of Anscombe’s reading of the Tractatus is her account of the ‘picture theory’7 of the proposition in the first six chapters of her book. The view that she rejects is that the ‘whole theory of propositions’ in the Tractatus is ‘a merely external combination of two theories: a “picture theory” of elementary propositions…and the theory of truth-functions as an account of non-elementary propositions’ (pp. 25–6). She had argued in her introductory chapter that, in order to understand Frege or Wittgenstein, it is best not to start with philosophical preconceptions, but rather to be capable of ‘being naively struck’ by questions like the one, mentioned above, of what the relation to reality is of the statement that Russell is a clever philosopher. The two central chapters of Anscombe’s presentation of the picture theory begin with questions of just the sort she had claimed we need to be naively struck by. They are questions that arise from the usual explanations, in logic books, of truth-functional composition. ‘It is usual for us to be told [… that] propositions are whatever can be true or false’; that ‘propositions can be combined in certain ways to form further propositions’; and that ‘in developing the truth-functional calculus, we are not interested in the internal structure’ of the component propositions. One question which may then strike us is whether ‘the property of being true or false, which belongs to the truth-functions, [is] the very same property as the property of being true or false that belongs to the propositions whose internal structure does not interest us’. And, further, if that is so, ‘is it to be regarded as an ultimate fact that propositions combine to form further propositions, much as metals combine to form alloys which still display a good many of the properties of metals?’ I shall quote her comment:
In short, is there not an impression as it were of logical chemistry about these explanations? It is this conception that Wittgenstein opposes in the Tractatus at 6.111: (p. 876) ‘Theories that make a proposition of logic appear substantial are always wrong. It might be thought, for example, that the words “true” and “false” denote two properties among other properties, and then it would look like a remarkable fact that every proposition possesses one of these properties. This now looks no more a matter of course than the proposition “all roses are either red or yellow” would sound, even if it were true’.
Here, interestingly, in the opposition to ‘logical chemistry’, we can see Anscombe picking out a feature of Wittgenstein’s philosophizing that is highly distinctive, and not apparently derived from Russell or Frege or any of the other thinkers whose influence on Wittgenstein can be discerned. (She quite explicitly argues that Frege, for example, in discussing whether every well-formed sentence the names in which are not empty has a truth-value, takes for granted a kind of logical-chemistry view of the nature of concepts.) Anscombe’s own reading of the Tractatus reflects a sense, not just of what sort of questions one needs to be struck by, but also of what constitutes a genuinely satisfying resolution of the puzzlement expressed in the questions. Hence the importance of her treatment of negation. It is not just a pivotal topic for Wittgenstein’s early thought but one through which she can demonstrate what is involved in reaching the kind of clarity at which he aimed. If we think about ordinary pictures, she says, we shall be able to see how the possibility of using a picture to represent that things are so goes with the possibility of using the very same picture to represent that that is how things are not. We can, that is, see in ordinary pictures the possibility of being used in two opposite ways, to say two opposite things. What is central in her account of what a picture is is that ‘the way the elements are connected in the picture is the same as the way [the picture] sets forth the things as being connected’. Hence the possibility of things being connected that way is in the picture itself. It is then, as occurring in such a picture-context, that the elements can have the use of representing this or that thing. We can move from that initial insight to an understanding of negation which does not appeal to some kind of ultimate logical fact. The basic idea is that the possibility of using a picture in two opposite ways, to say this is how things are, and alternatively to say that that is how they aren’t, depends upon correlating elements of the picture with things; and such correlating is something we can do so far as we take some way in which the marks or figures are related to each other to be significant. Only so far as they stand in such significant connections are these items elements of a picture; only in such connections can the picture-elements stand for this or that person or object or whatever it may be. Here is her summary: ‘Only in the connections that make up the picture can the elements of the picture stand for objects’ (p. 67). The picture-character of an ordinary picture is then what makes it possible, once correlations have been made, for there to be a this, such that there are two opposed ways of representing how things are: ‘This is how things are’; ‘this is how things aren’t’, where the this in question is the same. And Anscombe’s account of the picture theory is then that that picture-character that is in ordinary pictures is also in propositions. Only in the connection that makes up the proposition do the expressions in it stand for anything. It is through the significant connection of its parts that it can say that anything is the case; and so far as those significant connections make it possible to represent that this is how things stand, those same connections make it possible to (p. 877) represent that this is not how things stand. If what a picture represents as being so is its sense (TLP 2.221), we can say that a picture’s sense is reversible: it can represent the opposite as being the case. We can see what propositional sense is, if we see propositions to have the reversibility that belongs to pictures, if (that is) we see in propositions the possibility that belongs to pictures of representing that this is how things are, or (the this being the same) that this is how they aren’t. (See TLP 4.05–4.0621, where the point that reality can be compared with propositions is tied to the reversibility of sense of propositions.)
Two chapters after her account of the picture theory, Anscombe summarizes that account, and connects it with Wittgenstein’s remarks about how a symbol can be presented:
we have to remember the central point of the picture theory which we have already explained: ‘Only in the context of a proposition has a name reference’; ‘Only in the context of a proposition has an expression reference.’ This prohibits us from thinking that we can first somehow characterize ‘a’, ‘R’ and ‘b’ as symbolic signs, and then lay it down how we can build propositions out of them. If ‘a’ is a symbolic sign only in the context of a proposition, then the symbol ‘a’ will be properly presented, not by putting it down and saying it is a symbol of such and such a kind, but by representing the whole class of the propositions in which it can occur.
There Anscombe quotes Tractatus 3.3 and Tractatus 3.314, two statements of the ‘context principle’, to give the heart of the picture theory as she had earlier explained it. Here we should pause and ask some questions. The context principle, as it occurs in the Tractatus, certainly appears to mark a connection with Frege’s appeals to the context principle in The Foundations of Arithmetic. But what then is the connection? Warren Goldfarb has argued that it is not clear how far some apparently Fregean features of the Tractatus reflect the influence of Frege’s thought on Wittgenstein.8 It may be that Wittgenstein started with views which were profoundly influenced by Russell, but, in working through the difficulties of those views, came to a position which is close to Frege’s in significant respects. I shall leave open the question whether these ‘Fregean’ features of his thought reflect the direct influence of Frege. I want instead to ask what these features are, and how they are important for Anscombe’s reading of the Tractatus. Goldfarb is very helpful here in laying out differences between Frege’s thought and Russell’s. Goldfarb connects Frege’s commitment to the context principle with what he calls the ‘judgement-based nature’ of Frege’s view. Frege does not think of judgments as put together from parts which have some prior independent logical character. In a remark that has become well known (but which was not available in any of Frege’s published writings when Anscombe wrote her Introduction to Wittgenstein’s ‘Tractatus’), Frege said ‘I do not begin with concepts and put them together to form a thought or a judgment; I come by the parts of a thought by analyzing the thought’ (1979, p. 253). While that remark concerns the parts of a thought, a parallel point holds, on Frege’s view, for propositions: the parts of a proposition which (p. 878) have reference are identifiable only through the logical relations of the proposition to other propositions. Goldfarb notes the sharp contrast with Russell’s approach: For Russell, the primitive parts of propositions ‘subsist in and of themselves’. They are put together into propositions, but are recognizable on their own, independently of their role in propositions. Russell’s account of propositions and their constituents is described by Goldfarb as ‘object-based’, in contrast to Frege’s ‘judgment-based’ view. As a feature of Russell’s thought, it can be found as early as The Principles of Mathematics. Its presence doesn’t indicate that Russell was an empiricist, though such a view of propositions and their parts is indeed found in the writings of empiricists.9
In section 30.1, I argued that, although Anscombe had claimed that neglect of Frege was the main explanation why what had been written about the Tractatus had been for the most part wildly irrelevant, her account of the differences between Frege and Russell left it unclear why neglect of Frege should have so distorted understanding of Wittgenstein. But her presentation of the picture theory and the connection that she makes there with Wittgenstein’s version of the context principle show (I think) why she sees Frege as leading us in the right direction. My suggestion is that Goldfarb’s contrast between Frege’s judgment-based view and Russell’s object-based approach, although it doesn’t correspond to anything Anscombe explicitly says in laying out the contrast between Frege and Russell, lets us see why Anscombe insists on the importance of Frege for an understanding of the Tractatus. Consider the remarks that I quoted in section 30.1, that Russell differs from Frege ‘by introducing the notion of immediate experience, and hence that of private mental contents, into his explanations of meaning and his theory of judgment’; for he is, she says, ‘thoroughly imbued with the traditions of British empiricism’. We should, I think, read those remarks as containing three distinct points, about immediate experience, private contents, and empiricism. If we use ‘experience’ to include what Russell means by ‘acquaintance’, we could then say that Russell introduces into his account of meaning and judgment a notion of immediate experience, but we cannot then go on: ‘hence of private mental contents’. There is no ‘hence’, since one can have a notion of immediate acquaintance (even: of immediate acquaintance conceived on the model of acquaintance with the taste of a pineapple) in which the objects with which one is immediately acquainted need not be private mental entities, but may be such things as the indefinables of logic and universals. Indeed, even sense-data need not be conceived as ‘private mental contents’ on a Russellian view of acquaintance. A notion of immediate acquaintance can play a central role in a philosophical account of meaning and judgment, which may be quite far from empiricism in various ways, or indeed opposed to it.10 So one needs to separate from each other Anscombe’s point that (p. 879) Russell worked with a notion of immediate acquaintance and her characterization of his views as thoroughly empiricist. Anscombe is certainly right that the Tractatus was misunderstood by her contemporaries in large part because they saw it as a working out of a radically empiricist view; they saw Russell as arguing for a very similar kind of empiricism. But we need to focus here on the notion of acquaintance, and in particular on Russell’s principle that one can understand a proposition only if one is acquainted with its constituents. The principle is stated in ‘On Denoting’ and repeated in Problems of Philosophy and ‘Knowledge by Acquaintance and Knowledge by Description’.11 The idea that understanding a proposition depends on acquaintance with its constituents goes with the point mentioned by Goldfarb in characterizing Russell’s views: the primitive parts of propositions subsist on their own, and are recognizable independently of the propositions of which they are parts.
Although I am in the middle of a line of argument here, I shall introduce a digression to indicate where the argument is going. Some years after the publication of Anscombe’s book, Hidé Ishiguro, B. F. McGuinness, and Peter Winch developed readings of the Tractatus which explicitly reject the idea that the connections between names and objects are supposed to be established prior to the use of the names in propositions; on the reading that they reject, Wittgenstein held that the logical form of the object with which a name was correlated determines how the name can be correctly combined with other names in propositions. Their readings depended on taking seriously Wittgenstein’s expression of the context principle at TLP 3.3.12 It thus became incumbent on anyone who wanted to read the Tractatus as committed to the idea that objects have their own independent nature, and that it is first of all through the connection between names and such objects that language has its connection with reality, to explain what sort of force the context principle has, since it at least appears to imply that there is no such thing as a name having meaning prior to its use in propositions. It will be helpful if I summarize here, very briefly, one such line of response, that of David Pears. He argues that it is possible to interpret the context principle so that it is consistent with the idea that contact between names and things is prior to the occurrence of names in propositions; the principle implies merely that the association between a name and the object it names is ‘annulled’ if the name occurs in a context that doesn’t correspond to a genuine possibility for the object.13 Pears’s treatment of the context principle is developed in the course of his criticism of Ishiguro and McGuinness, and is meant as a basis for rejecting their (p. 880) understanding of the context principle and of the relation between names and objects. Anscombe does not share that understanding, but Pears’s account, if correct, would equally constitute an objection to Anscombe’s view.
Anscombe’s account of the picture theory is incompatible with any idea that setting up the connections between names and things is prior to putting the names into significant combinations. As she says, only if significant relations hold among the elements of a picture can we correlate the elements with things, so that the picture-elements stand for the things, and so that their arrangement shows a way in which the things can stand. When she wrote that the Russellian connection leads to misunderstandings of the Tractatus, one of the main things she had in mind was this: that if you read into the Tractatus Russell’s view that the intelligibility of propositions depends upon acquaintance with their constituents, you cannot understand the picture theory. But what would block understanding of the picture theory is not merely Russell’s doctrines about acquaintance and meaning: if Anscombe is right about the picture theory, it is incompatible not just with Russell’s own views and his version of logical atomism but also with any object-based account of meaning. Thus, for example, David Pears takes himself to be disagreeing with any strongly Russellian reading of the Tractatus, because he does not think that the objects of the Tractatus can be identified with sense-data and their properties, but whether an account is object-based has nothing to do with the question what sorts of thing the objects are; and Pears’s account is a paradigmatically object-based account of meaning.14 For Pears, as for Russell, the primitive parts of propositions subsist on their own; and the initial correlation between names and things is prior to the use of the name in propositions.
Consider also Anscombe’s point, quoted above, that a symbolic sign ‘a’ is properly presented, ‘not by putting it down and saying it is a symbol of such and such a kind, but by representing the whole class of propositions in which it can occur’. Here the recognition of some occurrence of the sign ‘a’ as an occurrence of that symbol is dependent on its occurrence in a proposition of the class in question, and there is no question of setting up which propositions it can occur in by considering its correlation with an object, taken to impose restrictions on its use, allowing some combinations of signs and disallowing others. The logical characteristics of what ‘a’ means are plain from its role in the propositions in which it can occur.15 It is a consequence of this view that there is no logical error in using the sign ‘a’ in other sorts of proposition; for in those contexts it would not be the same symbol. There is here a substantial difference from the approach taken later by David Pears, who speaks of the occurrence of a name in a context which doesn’t correspond to a genuine possibility for the object named as ‘annulling’ the (p. 881) connection between name and object (1987, vol. 1, p. 75). It is hard to see how this way of speaking can be connected either to Wittgenstein’s (and Anscombe’s) talk of signs or to Wittgenstein’s (and Anscombe’s) talk of symbols. The name-with-its-connection-annulled isn’t a mere sign, in their sense: for the supposed name, if it is thought of as a name at all, is being thought of as if it had at any rate a shadow of an attachment to an object of a particular kind. Otherwise one could not speak of its connection with some object being annulled. But a mere sign has no logical connection to any particular kind of object. But isn’t that the status of the name after the connection is ‘annulled’, on Pears’s view? The trouble with that reply is that the connection can’t be ‘annulled’ unless it is there to be annulled; and if the connection is there, then what has the connection is a symbol, not a sign; and if it is indeed a symbol, then it is in use in the sort of context of which such symbols are features, and there would then be no question of the connection being ‘annulled’ by the name’s being in the wrong sort of context.
The difficulty in an attempt to explain Pears’s view is that the context principle in the Tractatus, as Anscombe points out, appears to rule out the idea that we can identify a sign as a name of some particular object, and then go on to note that the combinatorial possibilities of that object permit such-and-such propositional occurrences of the name, and rule out others. The context principle is closely tied to the distinction between a mere sign and a symbol; but Pears’s account of what is involved in putting a name into a propositional combination of the ‘wrong’ sort cannot coherently be explained in terms either of the name as a mere sign or of the name as a symbol. The word ‘resolute’ has been given a use in discussions of the Tractatus, in connection with the interpretation of Wittgenstein on sense and nonsense, but I want to suggest that, in the contrast between Anscombe’s treatment of the context principle and that of Pears, we can see another sort of issue of resolution and irresolution. Pears’s interpretation of the Tractatus allows the context principle to rule out the idea that ‘a’ can occur genuinely as a name anywhere except in senseful propositional combinations; but the idea of the name as occurring in the ‘wrong’ sort of context, and thereby having its connection with the object ‘annulled’, employs the idea of a name-object connection that is there, independently of the occurrence of the name in propositions. This connection requires that the name itself not be thought of either as a sign or as a symbol, if a symbol is a symbol only in the context of a proposition. The sign/symbol contrast helps us to be clear about what we want to say and about when we are dithering and not saying anything. There is (I am suggesting) a wiggle or dither in Pears’s account, that operates in what is only apparently a space of possible philosophical conceptions.16 The fundamental difficulty is the attempt to combine an object-based understanding of language, shared with Russell, and the context principle. But a weakened version of the context principle is quite different from the version of the context principle that is reflected in the distinction between sign and symbol. Putting this point another way: there is no room for the Tractatus understanding of what a symbol is if one tries to read into the book an object-based understanding of (p. 882) names. Pears supports his reading of the context principle (as consistent with an object-based understanding of names) by reference to passages in Wittgenstein’s Notebooks from November 1914. These passages are hardly unambiguous, but more important, they were written well before Wittgenstein began to take the context principle seriously. If one takes Wittgenstein’s views to have been evolving during the years before the final version of the Tractatus was written, and in particular, if one takes his treatment of the context principle (and of the relation between the meaning of a name and its occurrence in propositional contexts) to be among the things that changed, it becomes questionable how far remarks from November 1914, or from elsewhere in the Notebooks, can be taken to impose on the context principle in the Tractatus an interpretation weak enough to make it consistent with object-based readings of the Tractatus.17
A strong version of the context principle, like that which Anscombe ascribes to Wittgenstein, has been held by some philosophers of language and some commentators on Wittgenstein to be incompatible with the compositionality of language. So (on this view) if Wittgenstein did hold such a version of the context principle, his account of language is in trouble. It is far from obvious, though, that there is such an incompatibility, and there are good arguments against it (Bronzo 2011). I shall not, however, examine the issues here.
I have argued that there is indeed a fundamentally Russellian way of reading the Tractatus, and that it is common to the interpretations Anscombe criticized and to later readings like that of David Pears. Part of my argument has been that what being ‘Russellian’ in this context amounts to becomes clear only later, in a line of discussion which develops from Anscombe herself, and which includes the writings of (among others) Hidé Ishiguro, Warren Goldfarb, Thomas Ricketts, and Peter Hylton. Anscombe’s interpretation of the Tractatus can be described as Fregean, not in that she claims that Wittgenstein’s approach was in relevant respects derived from Frege (which it might or might not have been) but in that Wittgenstein, as she reads him, came to share with Frege an approach later described as ‘judgement-based’. I have also argued that there is an important feature of Anscombe’s understanding of the central ideas of the Tractatus, a feature which is neither Russellian nor Fregean. I mean her treatment of the idea of ‘logical chemistry’. I quoted her discussion of Tractatus 6.111, where Wittgenstein says that theories that make a proposition of logic appear substantial are always wrong, and where he criticizes any account of logic that makes it look like a queer sort of fact that every proposition is either true or false. Anscombe’s account of the picture theory is meant to make the truth or falsity of propositions fall out of what it is for propositions to be pictures. There is to be no ‘logical chemistry’; and her argument is that, if we think through the analogy with pictures, and take as central the way ordinary pictures can be used to say that something is so, or used to say the opposite, we can see how the logical character of propositions is thereby made ‘extremely intelligible’. What she speaks of as the ‘grounds for being struck even to the point of conviction’ by the account is that it (p. 883) opens up the logical character of propositions, without appeal to ultimate logical facts of any sort. If, at the end of her second chapter on negation, she says that there is surely something right about the picture theory even if it is not correct as it stands, her conviction that there is something right about it comes from being ‘struck even to the point of conviction’ by the ‘extreme intelligibility’ given to the logical character of propositions. The idea is not, I think, that we have antecedently available a conception of philosophical clarity, and that the Tractatus account of propositions provides that sort of clarity. Rather, the making intelligible of the logical character of propositions provides a way of understanding what philosophical clarity can be. My account of Anscombe’s reading of the ‘picture theory’ is not intended to be complete. The most important thing that I have omitted is her discussion of elementary propositions—a discussion which is essential to her claim that the Tractatus makes ‘extremely intelligible’ the logical character of propositions.18 Without going over all that matters in her account, I have tried to show the kind of change she hoped to make in how the Tractatus was understood. In the next three sections, I shall be following out a line of thought that starts from what Anscombe actually does in presenting the picture theory. That will put me in a position to discuss, in section 30.6, how Anscombe’s approach fits into the history laid out by Warren Goldfarb of anti-metaphysical readings of the Tractatus.
30.3 Anscombe and Philosophical Method
Wittgenstein’s view of philosophical method is touched on by Anscombe, but it is not one of her aims [in An Introduction to Wittgenstein’s ‘Tractatus’] to make clear what he thought about method, or in what way the method of the Tractatus is connected with Wittgenstein’s more specific philosophical ideas in the book, or what the importance to the reader should be of his apparently methodological remarks.
(Diamond 2003, p. 172)
That now seems to me a stupid and misleading thing to have said. I was too impressed by a very partial truth, and failed altogether to see the kind of attention to methodology in Anscombe’s own approach. What, after all, does she do in the book—in that part of it that I have been considering? She lays out, makes open to view, a way of using words, the picture-proposition use. She is not simply expounding a theory held by Wittgenstein; she is attempting to put before the reader with the ‘extreme intelligibility’ with which the account can (she thinks) be presented, what it is to say that something is so, on analogy with using a picture to say that this is so, a picture capable of being used also to say that this isn’t so. I mentioned her having quoted Wittgenstein’s criticism of any account of logic that makes it look like a queer sort of fact that every proposition is either true or false. She herself is presenting a use of language, the picture-proposition (p. 884) use, which will not make it look like a queer sort of fact that every proposition is either true or false, but will instead make obvious, open to view, the connection between picturing and the possibility of truth and falsehood, and which will also make it clear, open to view, how such picture-propositions can be combined to form others, which will also be true or false.
In section 30.2, I quoted Anscombe’s statement that, if ‘a’ is a symbolic sign only in the context of a proposition, then the symbol ‘a’ will be properly presented not by putting it down and saying that it is a symbol of such-and-such a kind, but by representing the whole class of the propositions in which it can occur. There are various ways in which we might represent such a class of propositions; but what such a presentation of the class will do is make evident what the entire class has in common. Wherever a class of propositions has a feature in common, it can be presented in some such way; and although the point as Anscombe makes it concerns sub-sentential expressions, it is also applicable to the entire class of picture-propositions. What they have in common can be laid out. They have in common saying that something is so. Anscombe’s account of the picture theory, I am suggesting, can be taken to be a specification of a use of signs. Presenting the use of signs to say that something is so gives one case, indeed a quite special case, of presenting a class of propositions with something in common, of presenting a class so that what the members have in common is open to view. In any such case, the propositions in question are all values of some variable; and making plain what the values of the variable are is the way in which the variable itself is given.
I am suggesting that Anscombe’s presentation of the picture theory can be taken to be a case of making plain what the values of a variable are, where the values of the variable are propositions. This is a case of making plain a use of signs, and is not (in that respect) different in principle from the case Anscombe mentions, of presenting the symbol ‘a’ by representing the whole class of propositions in which it can occur—the class of propositions which have in common the presence in them of that symbol. If (as the Tractatus has it) anything essential to their sense that propositions can have in common with one another is an expression or symbol, then picture-propositionhood is itself a symbol common to picture-propositions, a common formal feature (propositional form), and the class of propositions with that feature (all propositions) can be presented; and this is what Anscombe has done. In fact, Anscombe presents this class of propositions twice over in her book, as does Wittgenstein in the Tractatus. In the Tractatus, the presentation of picturing and of truth-functional construction leads up to TLP 4.5: Symbols constructed in the way described are sayings that this is how things stand. And Wittgenstein also claims that that class of symbols can be given by specifying a formal series the members of which are the symbols in question. Anscombe’s account of the picture theory in the first part of her book gives the class of symbols and what they have in common informally; her chapter on formal concepts and formal series indicates how the symbols can be given as members of a formal series. There are questions about whether Wittgenstein actually succeeds in giving such a variable, and whether, if he does, what its use would be (Sullivan 2004). I have discussed these questions elsewhere (2012), and here I take for granted that the word ‘proposition’ can be used in a logical sense in ordinary talk (p. 885) without philosophical confusion, and that, in such cases, it would (on Wittgenstein’s view) go over in a conceptual notation to a variable.
In the rest of section 30.3, I draw out two consequences of this way of looking at Anscombe’s philosophical method in her account of the picture theory: as a matter of presenting a use of signs, the picture-proposition use. I shall also adduce another example of the method, before turning in sections 30.4 and 30.5 to some questions about it.
1. Connection between this method and the context principle. The idea that any symbolic feature that signs can have in common, essential to their sense, can be given by representing the entire class of propositions that have the feature is based on the context principle. (See TLP 3.3–3.317, and IWT, p. 93.) So, if we view giving the picture-proposition use of signs, their use to say that something is so, as a case of presenting a class of propositions with something in common, and thereby presenting the common feature, we can see the philosophical method in use in such a laying out of a class of propositions as an application of the context principle.
2. Connection with Wittgenstein’s remarks on philosophy. Wittgenstein says, at TLP 4.112, that philosophy is an activity that results in the clarification of propositions. One way in which philosophy can do this is by making plain what expressions have in common, and also by making plain differences between expressions. Later in the Tractatus, Wittgenstein suggests that the activity of philosophy is, properly speaking, appropriate only in response to philosophical confusion, when the activity involves showing the confused person that he hasn’t given meaning to some signs in his propositions. Taking this suggestion seriously would mean that one might, for example, present the symbol ‘a’ to a person who was using the sign ‘a’ without a meaning in his propositions, to show that person the emptiness of his employment of ‘a’. Presenting to that person the class of propositions of which the symbol ‘a’ is the common feature might help him to see that he was not using that symbol at all, since his proposition was not in the class in question. This picture of philosophical activity doesn’t have the implication that there would be no other sort of occasion for presenting the symbol ‘a’ (for example to someone learning a language), but the implication would be that such occasions were not cases of philosophical activity. If this is the way to understand what constitutes a philosophically appropriate sort of case of presenting the use of an expression, it would follow that Wittgenstein’s own presentation of the picture-proposition use, and similarly Anscombe’s presentation of the picture-proposition use, are appropriate if the activity helps to show that we are using some words with no meaning. The ‘picture theory’ then can be taken to be a case (indeed a rather special case, but a case nevertheless) of presenting to people a class of propositions with a common feature, unclarity about which is reflected in their (or, rather, our) use of words with no determinate meaning.
(p. 886) I am suggesting that the Tractatus can be read as a manual for philosophical activity, for philosophical clarification. You clarify by making plain commonalities and differences. I shall not here go into details about making differences plain, but one way of doing so is by making commonalities plain: if you make plain, for example, what is shared by all uses of ‘is’ as an expression for existence, and what is shared by all uses of ‘is’ as the copula, you may thereby make plain the difference between the uses. A shareable feature of propositions is represented by a variable, which can be presented by specifying its values in such a way as to make plain what they all have in common. I am also suggesting, then, that the relevance of Anscombe’s book to the question of the Tractatus understanding of philosophical method is in part that the book provides an example of philosophical activity in presenting the picture-proposition use of signs.
It will be helpful here to look briefly at Peter Sullivan’s account of the picture theory (2001). His approach is in some respects unlike Anscombe’s, but there are interesting similarities which I want to bring out. Like Anscombe, Sullivan takes questions about the relation between thought, the meaningfulness of propositions, and truth to be important in his account of the aims of the picture theory. He uses the metaphor of logical space to explain how a proposition’s meaning something is independent of its truth: its ‘coordinates’ determine a place in logical space. But his examination of the metaphor leads him to the question why we can count on it that what we can think to be so is genuinely a possibility for reality. Why should the possible combinations in which we use the names in our language enable us to represent genuine possibilities of existence for the objects named? It looks, Sullivan notes, as if it would be a kind of leap of faith, or superstition, to think that this was so. That is, it looks as if our capacity genuinely to represent possibilities for reality in our language depends on a kind of magical getting right of the logical character of the names, getting their possibilities to match those of the objects.19 Sullivan responds by explaining the Tractatus conception of pictorial form, using the idea of a kind of transparency in representation. Transparency is exemplified by the use of colour in a naturalistic painting, in contrast with the use of colour to represent which nation (say) has sovereignty over some part of the world, as with maps in which the British Empire was coloured red. In the first or ‘transparent’ case, colour ‘represents nothing other than itself’; ‘a feature of reality has simply been taken up into the system of representation’ (pp. 107–8). It is that notion of transparency, in its most abstract form, which is at the heart of the picture theory. In a transparent representation, the arrangement of the proxies for objects is the arrangement the objects are represented as having. A name then is a name, is a proxy for an object, in the context of a representation with such transparency. This account of the picture theory leads Sullivan to remark that Wittgenstein might well have taken the context principle to be his fundamental thought, since it underlies the idea of pictorial form as common to picture and what is pictured (p. 109). Although Anscombe and Sullivan give somewhat different accounts of (p. 887) the picture theory, both of them take the context principle to be absolutely central to it.20 What I want to emphasize is the role the context principle thus has, for both of them, in their account of how the picture theory makes propositionhood ‘extremely intelligible’. The logical characteristics of propositions, their capacity for truth and falsehood, their relation to reality, can be made clear without appeal to substantial metaphysical facts, which would have to turn out right if our thought is genuinely to be in contact with reality. In somewhat different ways, Sullivan and Anscombe lay out the picture-proposition use of signs. We can take both of them to be engaged in laying out a use of signs in such a way as to achieve philosophical/logical clarity, and thereby to reshape our understanding of what we need in order to solve our philosophical problems. They each start with questions that may seem to demand answers in terms of logical and metaphysical facts; but in response they provide what we might call a perspicuous presentation of a way of using words, and for both of them, the context principle is at the heart of this perspicuous presentation. Obviously, I am pushing a certain way of reading what they are up to. And I’ll push it further: they are responding to the Tractatus by doing what the Tractatus does, and doing philosophy in the sense in which the activity of philosophy is described in the Tractatus.
30.4 Problems About Philosophy
There is an important contrast between two ways of taking talk of presenting a use of signs. If I lay out a use of signs, I might claim that what I have laid out is the use of propositions. Or I can simply lay out a way of using signs, say, the picture-proposition use, and make no such claims. In the first case, it looks as if I will have achieved what I claim to have achieved if the use that I have laid out is indeed the use of propositions. It looks, that is, as if something out there, the way propositions are indeed used, makes my account right or wrong. But what items are we to take to be the ones the use of which makes my account right or wrong? If ‘proposition’ is what Wittgenstein calls a formal concept, then what falls under it are the values of a variable, a variable which can be given by specifying its values. But that is what the account itself purports to do. Underlying the idea that the laying out of the use (say, the picture-proposition use) might be compared with the way we genuinely do use propositions there is an unclarity about what philosophy can accomplish. The problem can be put as a dilemma. If what I have done is simply lay out a use of signs, what is its interest? Unless I make a claim about what sort of symbol it is, the use of which I am laying out, how is what I have done relevant to any philosophical (p. 888) problem? But if I do make such a claim, for example that what I have laid out is the way propositions are used, I use a term, in this case ‘propositions’, to pick out, or try to pick out, a class of symbols, but how can I take myself to have done that? Do I take myself to have, independently of specifying the use of a class of symbols, some way of understanding ‘propositions’? But if my best specification of the use of the class is just precisely what I have given, I don’t have some other specification up my sleeve by which to give content to my claim that I have given the use of that class of symbols. Nor can the difficulty be avoided by saying that the symbols, the use of which I have laid out, are those we would call ‘propositions’. For that certainly isn’t correct, since all sorts of sentences and sentence-constructions, used in a variety of ways, may be called ‘propositions’, and what I have attempted to lay out is a class which can be distinguished from the rest by the logical characteristics of its members, which of course are what I have attempted to lay out. The dilemma then, is this: how can a philosophical presentation of a use be illuminating, if it is not accompanied by such claims? But how can such claims be understood? In sections 30.4 and 30.5 I shall be discussing this problem; section 30.6 contains a brief discussion of a corresponding problem for Frege’s treatment of concepts. In this second half of the chapter, I shall be making problematic my own way of talking in the first half, in which I have unselfconsciously spoken of the Tractatus view of propositions, and of Anscombe as giving an account of the logical features of propositions.21
It has been suggested that we can view (at least some) Tractatus propositions as having a function akin to what Wittgenstein later spoke of as grammatical propositions, and that they are nonsense only in a technical sense. On this view, one use of ‘nonsense’ is simply as a label for propositions that give the characteristics of senseful propositions.22 It might seem that such an approach could resolve the difficulty I have sketched, by leaving room to say that the use that has been laid out is that of propositions, and that saying so is nonsense, but nonsense only in a technical sense. But such an approach cannot actually resolve the difficulty; for its source is a genuine unclarity about what one wants to say if one characterizes a use as that of propositions. In any case, what from the Tractatus point of view corresponds to what Wittgenstein later spoke of as making clear the grammar of some term is specifying the values of a propositional variable. That is how a way of using signs is presented. It is then about that kind of presentation that the question arises whether we can say that the use presented is that of propositions. The suggestion that Tractatus propositions themselves should be taken to have a function analogous to that of grammatical clarifications in Wittgenstein’s later writings seems to depend on not noticing that there already is something else that genuinely does have a comparable function from the point of view of the Tractatus: the specification of the values of a propositional variable. In any case, the supposed parallel between Tractatus propositions and grammatical remarks would hardly resolve the difficulty, since questions parallel (p. 889) to those which arise about presentations of use in the Tractatus can arise about grammatical remarks. There is indeed a further objection to the idea that Tractatus propositions are nonsense only in a technical sense, and are actually in the same business as are Wittgenstein’s grammatical remarks in the later writings. The objection is that saying that a remark is only technically nonsensical hardly makes clear how it is to be understood, if it is not clear that there is any way to arrive at what it means through familiarity with the meaning of its parts. If we are to understand it, surely we must have some way of understanding the words in it, in their context. But do we have any such way of understanding the words in the case of the sort of Tractatus remarks that are in question? A good example is TLP 5, ‘Propositions are truth-functions of elementary propositions’. If that is only ‘technically’ nonsense, the first word must mean something, in its occurrence in that context. It looks as if it is meant to be a formal concept word, but it also appears not to have the use, in that context, of a formal concept word. (See TLP 4.1272: formal concept words don’t have the use of words for functions or classes.) If the remark is supposed to give part of the grammar of ‘propositions’, the problem is that it plainly isn’t meant to give even part of the grammar of all that we might call ‘propositions’, since obviously many sentence-constructions which might get called ‘propositions’ don’t have the use that Wittgenstein is aiming to present, and he certainly didn’t think they did. If he meant to characterize any linguistic items, it was linguistic items used in a certain way. What way? Well, as truth-functions of elementary-propositions. But it is not going to be a grammatical remark to point out that truth-functions of elementary propositions are truth-functions of elementary propositions. The trouble with the idea that the Tractatus remarks are merely ‘technical’ nonsense is that, at the very least, when a sentence is called nonsensical, this should make one worry about whether one might be mistaking a conceptual blur for a meaningful remark. To characterize a sentence like ‘Propositions are truth-functions of elementary propositions’ as ‘grammatical clarification’ may make it appear that the only problem with such a remark is the label ‘nonsensical’, which (as being merely a label) is not a genuine problem. Hence the real problem what it means, where there is indeed such a problem, disappears from view.
Two remarks in the Tractatus can help us with the difficulty I have sketched: TLP 3.317 and 4.126. At 4.126, Wittgenstein says that the sign for the characteristics of a formal concept is a distinctive feature of all symbols whose meanings (Bedeutungen) fall under the concept. I take this remark to imply that the formal concept proposition has, falling under it, symbols with a characteristic kind of Bedeutung. In this context, the word ‘Bedeutung’ is used so that, not just names, but also symbols other than names can be spoken of as having Bedeutung.23 At 3.317, Wittgenstein says that when one gives the values of a propositional variable, and in that way gives the variable, one gives a description of the propositions whose common characteristic the variable is, and such a stipulation of the values of the variable will be concerned only with symbols, not with their meaning (Bedeutung). I am interested in the implication of these remarks for the case in which we are laying out what I have called the picture-proposition use of signs. I take the (p. 890) remarks to imply that we should not add that signs with the picture-proposition use that has been laid out are propositions, for that appears to be a specification of the Bedeutung of the symbols.
Even if what I have said is correct as a bit of Tractatus exposition, it hardly resolves the dilemma about what the philosophical relevance can be of laying out the use of an expression. If Wittgenstein indeed implies that we should not characterize the use which has been laid out as that of propositions, some alternative story has to be told of what the value can be of such an activity. But we should note anyway that the view which I have ascribed to Wittgenstein has an important consequence. If there is no saying that such-and-such use is that of propositions, there is also no saying that, so far as you use words in some other way, then what you are uttering is not genuinely a proposition. So far as laying out a use is nothing but laying out a use, it can exclude nothing. And yet, of course, the Tractatus is usually thought of as excluding something, or some things. How does it exclude anything, if it is in the business of laying out a use, and if laying out a use excludes no other use?
Suppose, then, that I lay out the picture-proposition use of words. For a sentence to have this use is for it to have ‘logical form’, where that means that the connection of its elements sets forth that things are connected in the same way as the elements; hence the connection itself shows the possibility both of things being as they are represented as being and of their not being as they are represented as being. The two possibilities for the things represented are thus internal to this way of using signs. The laying out of such a use doesn’t exclude any other use of signs, but rather helps to bring out a certain kind of confusion, in which one is at one and the same time using, or apparently attempting to use, signs in such a way, and also not using them in that way. The laying out of the picture-proposition use of signs makes clear how the two possibilities—things being as represented, or their not being as represented—are part of that use of signs, part of the proposition’s being a representation in logical space. If you want to give to words the use of expressing a substantial necessary truth (let us say), what you need to do is make clear what it is for what you say to be how things are, how they necessarily are. You need to do that without letting yourself slide into relying on what it is to use words in the picture-proposition way. That use does provide a way of saying how things are, but you are not going to be able to appeal to it. What is excluded by laying out the picture-proposition use is a kind of unconscious slipperiness, in which you take for granted the picture-proposition use and its abrogation at the same time, in which you take yourself to be saying something that really is so, necessarily so, and slippery-slide into a conception of a space in which this is said to be so, a space in which there is an opposite way things might be said to be, but they cannot be that way. This is a space in which there lie certain cases that are excluded, a space in which what is impossible can be thought of as impossible. But, as Wittgenstein points out at TLP 5.61, there is only confusion that lies in this direction. It can unmisleadingly be said that the Tractatus excludes substantial necessary truth only if it doesn’t mean that some way of using words is excluded. Again: no use of words is excluded by laying out a use of words; but laying out a use of words can be meant to sharpen our eyes to the fact that a supposed way of using words is not anything (p. 891) at all. Let me briefly specify more clearly what I mean and what I don’t mean by saying that laying out a use of words does not exclude any way of using words. I don’t mean that, once you have laid out the use of picture-propositions, you cannot say that certain other ways of using words are excluded because they are not genuine propositions, although indeed they are excluded. I mean simply that, if you lay out a use of words, you have not thereby excluded anything. If you lay out the picture-proposition use, put it clearly before us, the only thing that is thereby made clear about other uses is that they are not that use. If there is to be any ‘exclusion’ going on, it is at any rate not done by laying out a use. It is important, in reading the Tractatus, whether one takes it to exclude certain uses of words as not genuine propositions. Such readings, among which would be Anscombe’s, might be labelled ‘exclusionary’. I shall have more to say about such readings in sections 30.5 and 30.6, but shall first turn back to the problem of describing what the Tractatus does.
30.5 More About the Problem of Part 4
Before trying a different approach, I shall restate the problem. One can say that Wittgenstein in the Tractatus was giving the essence of propositions, but what he was giving the essence of can only be made clear by giving that essence itself—that is, by giving a variable the values of which have ‘propositional form’ in common. To think that one might helpfully convey something by saying ‘What he is giving the essence of is propositions’ is nonsense; not nonsense because it doesn’t count as a proper proposition, but nonsense because of the incoherent demands that we make on what the word ‘proposition’ can be thought to do. But that point leads right back to the question how it can be philosophically illuminating to lay out the picture-proposition use, if there is no claim that what is laid out is propositionhood. After discussing this problem again, I shall end this section by showing how philosophical activity, as I have been describing it, is connected with questions about nonsense. Although in this section I shall be wrestling with Anscombe, my aim is to think about what she herself does in presenting the picture theory. I see her philosophical doing as in tension with what she says about what the picture theory was supposed to accomplish, and in tension, in particular, with her exclusionary reading of the theory.
Consider again the philosophical activity of laying out the picture-proposition use. We are invited to have before our minds an ordinary picture: perhaps a realistic picture of a tree, perhaps a schematic diagram of people fencing. We are then led to take that picture differently. The ordinary picture that is used as an example will be one in which it is easy to see that the connection of the picture-elements that represent things is the connection that those things are represented as having. It will be easy to see that the possibility of such a connection of the things is there in the picture itself, in the connection of its picture-elements. By being led to note these features of the ordinary picture, we can be led to take it in as a logical picture, to take it in as having logico-pictorial form. I want (p. 892) to suggest that this transformation to logical taking-in is central in the way the ordinary picture is used in the philosophical activity. This transformation is not a matter of taking in the picture as having a property of which we had been unaware. When we come to see something as having a property of which we had been unaware, we can grasp that other things can also have the property; the grasp of such generality is part of what is involved in recognition of a property. But the generality involved in the transformation of our understanding of an ordinary picture is different. To take in an ordinary picture as a logical picture is to see something generalizable in it, but the generality of ‘logical picture-hood’ is not that of a property. When Wittgenstein says, in the Tractatus, that every picture is also a logical picture (2.182) this doesn’t mean that, for all x, if x is a such-and-such, it is also a thus-and-so. In being led to take in the ordinary picture as a logical picture, we see the logical kind in the particular case; we see in it a logical shareable. To take in the picture as having logico-pictorial form is to take it in as having a logical characteristic that can be present in cases which do not share the particular pictorial form of the simple example from which we started, but which share only the most general logical feature of the example, namely that the way things are represented as being is the way the picture-elements are themselves connected, and the possibility of the things being that way is present in the connection of picture-elements. This identity may be present in only the most abstract sense, in contrast with the ordinary kinds of case used as examples, in which the identity of form is identity in the role of colour or spatial relations in the representation and in what is represented. Consider now what Wittgenstein does in introducing the idea of propositions as themselves pictures. He invites the reader to consider a case in which taking in a proposition as a logical picture will be easy: the case in which the propositional sign is composed of spatial objects rather than written signs (TLP 3.1431). Just as we can be led to take in an ordinary picture as a logical picture, the core of which is its logico-pictorial form, we can be led to take in the spatial-object proposition as a logical picture, led (that is) to take it in as having logico-pictorial form. In both cases we are led to recognize a kind of generality (that of a logical kind) through a transformed taking-in of a simple case. In both cases, what we are supposedly able to see clearly after the philosophical activity is not something of which we can be thought to have been totally unaware beforehand. We could hardly operate with pictures without any awareness of their logical character; we could hardly say what was the case without being able to take in, to some degree, what Wittgenstein means when he says that in a proposition a situation is in a sense constructed by way of experiment. But the point is that such takings-in of the logical features of propositions and pictures are inchoate. What is essentially in common to all such cases is not seen.24 The philosophical activity, focusing at first on simple cases, is meant to open our eyes to what I have called a logical (p. 893) shareable. It can be represented by means of a variable, the values of which are all the symbols that share the logical characteristic in question.
In section 30.3, I suggested that Anscombe’s account of the picture theory lays out a way of using language, the picture-proposition use, and that what she does can also be described as making plain the values of a variable. I’ve been arguing here that, when the picture-proposition use is laid out, the starting point is a transformation of our way of taking in ordinary pictures; we are led to take them in as logical pictures, led to see them as characterized by a logically shareable feature. That’s the starting point, but it is also the point that lets us see the significance of laying out the picture-proposition use. When I take in the ordinary picture as a logical picture, I take it in as exemplifying a logical characteristic. I take in, in this case, picturing. The laying out of the picture-proposition use gives the reach (as it were), the logical generality, of the feature which I originally take in when I conceive an ordinary picture as having an identity of form with what it represents. The significance of ‘laying out the picture-proposition use’ should be tied particularly to TLP 2.1, the first Tractatus remark about picturing. ‘We make to ourselves pictures of facts’. We are meant to take in what we thus do in simple cases; we are meant ultimately to see that in its full generality. Suppose I come to see it so: I take that feature, in its logical generality, to reach through my thought, through my language, through my world. What I have called ‘laying out the use of picture-propositions’ is meant to connect with my self-understanding; it is meant to let me see an essence in my own thought, what it has ‘within’ it, ‘what we see when we look into the thing’ (Philosophical Investigations § 92). It is this reconception of the ordinary, as having within it something hidden, special, and with a unique total generality, that is indeed the target of Wittgenstein’s later thinking.25 In one of the early drafts of the Investigations, Wittgenstein spoke of how we take a ‘clearly intuitive’ case, and treat it as an exemplar of all cases; we take in a single proposition as a picture, and think that we have thereby grasped an all-comprehending essence, lying beneath the surface.26
All kinds of expressions are called propositions in ordinary talk; and Wittgenstein’s remarks about propositions being pictures plainly don’t imply that mathematical propositions, or logical propositions (etc.) are pictures. But note now that it cannot be said that the reason his remarks don’t apply to such cases is that he is making clear what is a genuine proposition, as opposed to those other things. What constantly guides our thought (p. 894) here, and constantly leads us in a wrong direction, is the idea that we have a concept of propositions and Wittgenstein is clarifying it, or trying to show what is involved in it. We constantly think in terms of a concept here, and what genuinely falls under it; but what we have to do with is a formal concept, that is to say, not a concept. As long as we think of the Tractatus as doing something or other with the concept of a proposition, we set ourselves up to miss what he is doing. An essential contrast for the book is that between a property and a logical shareable; and what lacks a particular logical shareable is not thereby shown in any way to be ‘rejected’ or ‘excluded’.
That last point is important when we think about Wittgenstein’s later criticism of the Tractatus. For it is sometimes said, as Anscombe herself says, that what was wrong with the picture-theory is that ‘it is correct only within a restricted area’; the idea is that there are various sorts of expressions that are genuine propositions, but that are excluded from the realm of propositions, because they are not in the ‘restricted area’ to which the picture theory applies. But this criticism depends upon taking the picture theory to be at one and the same time a presentation of a logical shareable (which it is) and a general account of propositionhood or sense (understood in some other way). For the idea is of a ‘larger area’, including both the ‘restricted area’ and what has been left out of it; and if the ‘restricted area’ is that of picture-propositions, i.e., the region characterized by the logical shareable, the larger area must be understood differently. Wittgenstein’s own criticism is quite different; it is that the supposed ‘logical shareable’ was actually part of the form of description of a multitude of very different cases; it was read into them, not discovered in them. And he rejected also the idea that a logical kind was presented by a variable, the values of which shared a logical characteristic.
It might be objected to my claim that laying out a use doesn’t exclude anything that it misses the point, since the Tractatus is engaged not just in laying out the picture-proposition use, but in characterizing all other kinds of cases of proposition-like constructions, apart from tautologies and contradictions, as nonsensical. What is this if not some kind of exclusion? And, indeed, doesn’t the image that I used in explaining Wittgenstein’s aim, that the reader should come to think of logico-pictorial form as reaching, in its generality, right through ‘my thought, my language, my world’ suggest that everything that lacks logico-pictorial form is pushed out into outer darkness? Consider here Anscombe’s way of thinking, as it emerges in her treatment of sentences of the form ‘“p” says that p’. Wittgenstein had said that ‘A believes that p’ is of the same form as ‘“p” says that p’. She says that Wittgenstein thereby gives us ‘“p” says that p’ as a possible form of proposition, and that therefore, if Wittgenstein has not been careless, ‘it must fit his general account of propositions—that is, it must have true-false poles’ (p. 88). That is (on her view), a proposition of this form is plainly not a tautology, and so, if it is not nonsensical, it must be a bipolar proposition, what I have here called a picture-proposition. There is, as she is presenting the case, no room for sentences to ‘fit in’ if they don’t fit in as tautologies or contradictions or bipolar propositions. But we should, I think, follow Michael Kremer (2002) in reading the Tractatus to allow for various kinds of sentences which, like tautologies and contradictions, guide us in inferring non-logical propositions from non-logical propositions. These auxiliaries to inference are without sense, but not nonsensical; (p. 895) far from it. They have an important kind of use, but it is not the use of saying how things are. I believe that there are quite a number of different types of auxiliaries to ordinary talk (Behelfe der Darstellung) recognized in the Tractatus, and there is no reason to think that other sorts of auxiliaries would not also be capable of ‘fitting in’ to the overall picture. Thus, for example, the sentence ‘Aus, bei, mit, nach, seit, von, and zu take the dative’, if it is taken to express a rule (rather than a generalization about German-speakers) can be regarded as an auxiliary to description, in the sense that it guides the construction of propositions. Indeed, one could read TLP 3.343, which says that definitions are rules for translating from one language into another, as introducing a quite broad category of rules of translation, which would include the rule about German prepositions. It would, I think, be wrong to take the Tractatus to imply that such rules are nonsensical. Kremer discusses in detail the Tractatus account of identities and of mathematical equations. He argues that such expressions fit in to the overall Tractatus view in roughly the same kind of way as do tautologies and contradictions, and should be taken to be senseless, rather than nonsensical.27 Other cases, discussed in the Tractatus, of auxiliaries to description include the presentation of an expression by means of a variable whose values are the propositions containing the expression, and again also the stipulation of the values of a variable. Besides those cases, there are the laws of mechanics, which are not logical pictures of the world, but give forms in which descriptions can be cast.28 Wittgenstein’s later discussions of ‘hypotheses’ suggest a similar sort of use of sentences which may look like descriptions but which function as a kind of auxiliary to description. My point here, in opposition to Anscombe, is not that propositions of the form ‘“p” says that p’ should be treated as rules, but only that there is much more room in the Tractatus for miscellaneous uses of language than we might think. It follows that, when a kind of sentence can be seen not to have the use of a picture-proposition, nor to be tautologous or contradictory, it is not thereby cast into outer darkness. It is indeed excluded from the realm of picture-propositions, of sentences that are used as such sentences are, sentences that represent a situation in logical space. But the question what use it has, if any, is open.
Anscombe’s discussions of what is allegedly excluded by the Tractatus account of propositions are somewhat puzzling, in any case. For consider what she says about ‘Red is a colour’ (p. 82). She says that, for such a case as this, the point is easily made that the sentence cannot express anything that might be false, since there are not two possibilities: that red is and that it isn’t a colour, of which the first happens to be the case. Here she sees the sentence as being excluded from sensefulness because it is not bipolar; but she herself has given, as Wittgenstein’s view, that we present a symbol, not by putting it down and saying that it is a symbol of such-and-such a kind, but by representing the class of (p. 896) propositions in which that symbol occurs. What, then, if we take the symbol ‘red’ that is used in colour-attributions, and think of presenting it through the class of propositions in which it occurs as that symbol? The symbol is a mark of a form and content that propositions can have in common, but ‘Red is a colour’ isn’t one of the propositions with that shareable form and content. It has only the sign in common with those propositions, and the practice of using ‘red’ as a colour-word in those propositions doesn’t settle what meaning, if any, it has in ‘Red is a colour’.29 There is a further question about the formal concept colour. Anscombe’s view appears to be that the reason why formal concepts can’t be presented by a function is that the attempt to do so (as in the attempt to treat being a colour as a property of red) leads us to construct propositions that cannot express anything that might be false. But that is not the problem. That is, the problem isn’t that falling under a formal concept is a matter of having a property that can’t be said to hold of the things in question. It is rather that it honestly and truly isn’t a matter of a property at all, but is seen in a shared feature of a class of senseful propositions. A logical kind isn’t a kind that things necessarily belong to; it’s not a kind that you can’t say things fall into; its difference from kinds, as we think of them, goes deeper than that. The word ‘colour’, in Anscombe’s example, does not have the use of a formal concept word, and it has not been given any alternative meaning as a property word. It is not clear that the word ‘colour’ has any meaning in that context. Anscombe’s illustration of the kind of sentence that is excluded from sensefulness on the Tractatus view was meant to show that the Tractatus view covers only some of the territory of what we take ourselves to be able to say; but the Tractatus does not imply that ‘Red is a colour’ is not capable of being false, and that it is, for that reason, excluded from sense. An argument based on the Tractatus would investigate what use, if any, had been given to the words in ‘Red is a colour’ which were taken over from a context in which they had a different sort of use.30
Most of Wittgenstein’s propositions in the Tractatus use words like ‘object’ and ‘proposition’ which have an unproblematic use in everyday talk as formal concepts. In conceptual notation, he says, the use of these terms would go over to variables. But in the Tractatus, these words are not used as they are in ordinary talk, and the Tractatus remarks would not go over in a conceptual notation to formulae with variables. The contrast is clear in Wittgenstein’s own pair of examples (TLP 4.1272): ‘There are two objects which…’ and ‘There are objects’. The trouble with the latter proposition is not that it (p. 897) is not capable of being false; it is that ‘objects’, in that context, seems not to have any meaning. In ‘There are objects’, ‘objects’ hasn’t got the use it has in ‘There are two objects which…’, which goes over, in conceptual notation, to a formula with quantifiers and variables. The sentence ‘There are objects’ appears to give to ‘objects’ a logical role of the same sort as that of ‘books’ in ‘There are books’. If ‘objects’ in ‘There are objects’ were given an appropriate sort of meaning, as a word for a kind of thing, the sentence would be meaningful. The fact that in other contexts the word is a formal concept word doesn’t carry over to its use in ‘There are objects’, which has only the sign, not the symbol, in common with such other sentential contexts. The difficulty is, of course, that we don’t want to give the word ‘objects’ some other meaning; we want it to carry with it its role as a formal concept word, and to mean, in ‘There are objects’, the logical kind objects. And similarly with Tractatus remarks containing the word ‘propositions’. We read the word as meaning propositions, regardless of the fact that the word does not occur in these remarks as a formal concept word. We don’t read those remarks with any doubt or suspicion about what the word means in them. But the Tractatus view is that the word ‘propositions’ occurs with the meaning that we unthinkingly suppose it to have in those remarks when it is used in a certain way: when it occurs in a sentence which would go over in conceptual notation to a formula containing the variable given in TLP 6, ‘the general form of proposition’. I think it can be shown, for example, that ‘Every proposition uttered by Cheney is false’ is, on the Tractatus view, translatable into a sentence with that variable.31 In that sentence, the word ‘proposition’ has the use of a word for a formal concept, just as, in ‘There are two objects which …’, the word ‘object’ has the use of a word for a formal concept. But the Tractatus remarks containing the word ‘proposition’ do not use the word as a formal concept word. It is used, one might say, as a word pretending to be a formal concept word, and it is not given any other use. The remarks do not convey a content of a special sort which cannot be put into senseful language; they contain words with no meaning and are nonsense. They have a function within the context of the book, a book meant not only to present a use but to lead its readers to take their own thought and language to have in it the logical shareable that the book presents. Their function within the context of the book is described by the metaphor of the ladder that is thrown away. They are helps on the way to recognition of the contrast between logical shareables and kinds of things, a contrast the recognition of which undoes the impression they initially make of conveying a content. The recognition of that contrast is not a matter of taking being a proposition to be something we can’t speak of; it is a matter of getting the point of philosophy as a kind of practice, in which logical shareables are displayed, as in the case of the laying out of the picture-proposition use of words.32
A full discussion of exclusionary readings of the Tractatus would have to look at Wittgenstein’s treatment of the limits of language. Exclusionary readings take Wittgenstein’s talk of limits to involve a conception of what lies outside, what cannot (p. 898) be said, and what sorts of attempted sayings fail because they are attempts to speak of what lies outside. Such readings involve what Peter Sullivan has called a contrastive understanding of the notion of a limit. There are then questions how far, and in what ways, such an understanding might be taken to be undercut by the Tractatus. Peter Sullivan (2011), A. W. Moore (2007), Juliet Floyd (2007), and I (2012) have discussed these questions, but to have gone into them here would have doubled the length of this chapter.
The technique, of sharpening the reader’s eyes to a logical shareable through attention to linguistic patterns and their ties to inference, is Fregean, as is the connection between the recognition of a logical shareable and the use of the context principle. Frege also (in ‘On Concept and Object’) confronts the problem that I have been discussing. He presents a logical shareable, call it ‘Fregean concepthood’, and makes clear that he is not attempting to capture what is usually meant by the term ‘concept’, which, as he notes, is used in various ways (1984, p. 182). He is not suggesting, that is, that all and only Fregean concepts are concepts. At the end of his reply to Benno Kerry, he says that Kerry can use the words ‘concept’ and ‘object’ however he likes, but Frege too has a right to use ‘concept’ and ‘object’ in the way he has laid out (p. 193). But then the question arises, if what he has laid out is his way of using the terms, and if he is not giving an account of concepthood, what is the significance of what he has done? We have, he thinks, only a ‘vague notion’ of what is involved in our own thinking and inferring (1979, p. 253). We can take his logical distinctions up into our understanding, and thereby bring the logical characteristics of our thinking into focus, at the same time separating off what is inessential, and what is the result of psychologistic accretions.
Wittgenstein takes over from Frege, or simply gets on his own, the connection between the idea of a logical shareable and the context principle. He takes over from Frege, or simply gets on his own, the idea that laying out features of our use of words can make ‘extremely intelligible’ what we might otherwise have taken to be a kind of ultimate logical fact, as for example that every proposition has exactly one negation. There is already in Frege a conception of a technique by which thought itself can be clarified, the technique of making plain logical having-in-common, a technique we can see already in Begriffsschrift, in the suggestion that we think of ‘subtracting’ from a proposition a part or parts that can be thought to vary, leaving a part that is logically in common with other propositions. Wittgenstein’s distinction between signs and symbols (where symbols are logical shareables) is an application of the context principle, and a corresponding distinction is clearly at work in Frege, in ‘On Concept and Object’ and elsewhere. But Wittgenstein gives the sign–symbol distinction a particular twist, by applying it to the question what it is for us to be using a word for a logical kind. What is shared by words used for a logical kind, when they are used for a logical kind, is clear if (p. 899) the propositions in which they occur are thought of as they would appear in a conceptual notation, in which sameness of sign invariably does indicate sameness of symbol, unlike the situation in ordinary language. In a conceptual notation, words that in ordinary language are genuinely in use as words for a logical kind go over to an appropriate variable. In this notational point there is reflected the profound difference between words for logical kinds and words for ordinary kinds. Now put together the Fregean point and the Wittgenstein twist.33 We can clarify logical features of our thought by making plain logical having-in-common, as in laying out the picture-proposition use of words. If, on the other hand, we think of ourselves as trying to give a theory of propositionhood (say), we can come to see that, in doing so, we will use the word ‘proposition’, which we want to use as a word for a logical kind, in a way which defeats our purpose. When theorizing about propositionhood, we use the word ‘proposition’ so that it is not the equivalent of a variable, but a word with the grammar of a word for an ordinary kind. Because we want to investigate a logical kind, we are not going to give ‘proposition’, in these investigations, a use as a word for something else; but since it is (in the context of these investigations) not a word for a logical kind, and not a word for anything else, what we say is nonsense. It is noteworthy that in criticizing Wittgenstein’s account of language, we may find ourselves doing exactly the same thing, and thereby missing a fundamental point of the book. We think of the Tractatus as an attempt to convey that thus-and-such is what propositions are, but the point is rather this: If you talk about propositions as you want to, you will not be saying anything at all. You misunderstand what you are after: you want to speak of a logical kind, and you also want to theorize about it in the language of ordinary kinds; and these two aims together will lead you to talk real rubbish.34 The same point can be made about other topics of philosophical investigation. Philosophical investigation is self-defeating when it aims at an understanding of logical kinds, but investigates them in language which is not logical-kind language. You can show understanding of the aim of the book in turning from philosophical theorizing to a form of philosophical activity that can illuminate logical shareables by laying out the ways in which we use language. This is a very different kind of procedure, as comes out especially in the fact that laying out a use does not exclude anything, but can lead us to a different way of conceiving what had appeared problematic. Juliet Floyd (2007, Part 2) makes a related point about Wittgenstein’s understanding of how philosophical problems are posed. Philosophical problems are those ‘whose very formulation contains terms that require interrogation, or reconception, in order to be solved’ (pp. 189–90), and she connects Wittgenstein’s conception in the Tractatus of philosophical problems with his lifelong interest in the contrast between searching when you have a framework (p. 900) for finding an answer and searching when you do not know in advance what will count as a solution.35
Philosophical activity, as Wittgenstein understands it, reshapes desire. We start off wanting to know the essence of propositions, or how thought is connected with reality. So long as we use the language in which the problems present themselves, we will get nowhere; the questions are not the sorts of questions we take them for. The activity of presenting a logical shareable, of putting it into a sharp focus, can put the logic of our thinking before us, and we can recognize in what we thus come to see (although it wasn’t the sort of answer we had been in search of) what we had wanted. If we let the philosophical activity shape our self-understanding in this way, we may give the logical shareable a label. What I have in mind is illustrated by labelling Fregean concepts ‘concepts’ and by labelling picture-propositions ‘propositions’. The word ‘proposition’ can be taken over from its use in our attempts to discover what propositions are. It can be turned into a label for the picture-proposition use, where the choice of such a label reflects seeing our own thought through the logical-organizing lens of the logical shareable, and seeing the importance (the reach through our thought) of that logical shareable. This (I’m suggesting) is not a matter of our taking ourselves to have discovered what propositions are; and speaking of Fregean concepts as concepts isn’t taking ourselves to have discovered what concepts are; we are sharpening our focus on a logical form, and seeing it in, or seeing it ‘into’, our thought. The label itself is no more than a reminder; it points us towards a previous clarification of a logical shareable. Wittgenstein can say ‘Here is the general form of the proposition’, but this is not the discovery of what propositions are, and a fortiori not the discovery of what is excluded from being a proposition. Sentences which are not picture-propositions are not picture-propositions. That’s what they aren’t. It won’t be all that they aren’t (they may also not be heroic couplets, quotations from Hume, or whatever) but there’s nothing that, in not being picture-propositions, they thereby aren’t. I want to pick up a Fregean way of putting these issues from Thomas Ricketts. Frege’s elucidations make use of ‘concept’ and ‘object’ as a contrasting pair of predicates. Once his intended audience has mastered his conceptual notation, the confusion latent in the elucidations becomes manifest, as they try to paraphrase the remarks into the notation; but they find no thought to which ‘No object is a concept’ (for example) corresponds. A master of the notation is free to discard the contrasting use, in Frege’s elucidations, of the predicates ‘concept’ and ‘object’, free to take the elucidations to be so much hand-waving. There isn’t then something left over that is not expressed in the notation.36 What corresponds in the case we have been considering is the contrasting pair, ‘proposition’ ‘not a senseful proposition’, as predicates in the Tractatus. When we have mastered what Wittgenstein is presenting through elucidatory propositions that use such predicates, we are free to (p. 901) drop the predicates, and to take the elucidations to be so much hand-waving. There isn’t then something left over that we haven’t been able to put into words.37
Anscombe wanted her book to change how people read the Tractatus. In section 30.1 of this chapter, I brought out problems in her story about what was wrong with earlier readings. I tried to show that her account of the picture theory points us to fundamental connections between Frege’s approach and Wittgenstein’s. While the sorts of reading she attacked are less popular than they once were, her critique can be taken to be directed also against many later readings—readings which resemble those of her contemporaries in their dependence on an object-based understanding of language and thought, which they read into Wittgenstein. I have also argued that the Tractatus is meant to lead its readers to a different kind of practice of philosophy, and that, although Anscombe doesn’t say much about philosophical activity, what she actually does in her account of the picture theory can be thought of as exemplifying the kind of philosophical activity to which the Tractatus was meant to lead. In sections 30.5 and 30.6, I tried to set against each other some of Anscombe’s own remarks about the Tractatus and ideas to which one can be led by following out what is implicit (I claimed) in her philosophical practice. I tried to show the tension between her remarks about what Wittgenstein’s conception of propositions excludes and her own insistence on the central importance of Frege and the context principle for a reading of the book. For it is just such an approach that helps us to see problems with the word ‘proposition’, as it is used in claims about what Wittgenstein’s theory excludes from propositionhood, helps us to see how the word can cover over a blur in our thought. I should want to claim that, if we follow out ideas implicit in Anscombe’s practice, together with points she makes about ‘logical chemistry’ and about how symbols can be presented, we can get a good idea how Wittgenstein hoped his book would revolutionize philosophy.
In his study of anti-metaphysical readings of the Tractatus, Warren Goldfarb (2011) includes Anscombe among metaphysical readers, along with Peter Geach, David Pears, Norman Malcolm, and Peter Hacker. He traces the development of anti-metaphysical readings of the book, beginning in 1969 with Hidé Ishiguro’s ‘Use and Reference of Names’. As he mentions, it is the realism of readings like those of Malcolm, Pears, and Hacker that was at first the focus of criticism. Anscombe’s position is very interesting and in some ways anomalous. She would have found unexceptionable the statement of Hacker’s that Goldfarb uses to set out the metaphysical reading. Hacker says, and Anscombe would agree, that Wittgenstein was committed in the Tractatus ‘to a host of claims about logic, language, thought and the logical structure of the world, which cannot be stated in well-formed sentences of language’ (2000, p. 383). But it is not clear how much further Anscombe’s agreement goes—not clear whether she takes (p. 902) the Hacker–Pears–Malcolm view that Wittgenstein’s ‘objects’ are independent of us and prior to language, and impose on language the structure it must have in order for what we say to express genuine possibilities.38 But besides the question how close Anscombe really is to the metaphysical readers, there is the question how close she is to the anti-metaphysical interpreters.39 The starting point of Ishiguro’s essay is the ascription to Wittgenstein of a view that the meaning of a name cannot ‘be secured independently of its use in propositions by some method which links it to an object, as many, including Russell have thought’ (p. 20); but that contrast between Wittgenstein and Russell is already present in Anscombe’s interpretation, and is central in her exposition of the picture theory. I should want to take Anscombe out of the group with which Goldfarb puts her, and treat her as in important ways an inaugurator of the anti-metaphysical readings. I’m suggesting that metaphysical readings like those of Hacker, Pears, and Malcolm combine the ascription to Wittgenstein of a Russellian object-based view of language and thought with the idea, sometimes labelled the ‘ineffabilist view’, that the Tractatus is committed to substantial claims about propositions, objects, facts, and so on, claims which Wittgenstein is supposed to have taken to be correct although not statable in significant language. Although Anscombe accepts such a view, it is (I think) under far greater pressure within the overall context of her interpretation than it is in the writings of Hacker, Pears, and Malcolm. We do not need to appeal to any specifically Tractarian views to see the kind of pressure, for we can simply look to Frege. Frege is, quite explicitly, not telling us about concepts; he is constructing for us an understanding of ‘concept’ as a logical shareable, to be thought of as ‘arising from the decomposition of a judgeable content’ (1980, p. 101). He isn’t getting concepts right; and Wittgenstein isn’t getting propositions wrong (say) by excluding too much; he isn’t getting propositions right by getting just the right things in. In that sense he is not making substantial claims ‘about propositions’, but constructing for us an understanding of ‘proposition’ as a logical shareable. If an expression has the use of a picture-proposition, it doesn’t ‘show’ that it is a proposition. What shows in its use as a picture-proposition is simply that.40
Exclusionary readings and ineffabilist readings are closely linked. The ineffabilist reading ascribes to Wittgenstein a view of what propositions are, a view that cannot be stated but can be communicated despite its unsayability. And the idea that the Tractatus excludes from propositionhood sentences that are not bipolar (or whatever the excluding criterion is supposed to be) depends upon the idea that Wittgenstein held in the Tractatus, but took to be unsayable, the claim that thus-and-such is what genuine propositions are, and that all other putative propositions are excluded from genuine propositionhood.41 Against this, I have argued that Wittgenstein, like Frege, presents logical (p. 903) shareables, and that to do so is to invite a kind of reconceptualization of one’s own practice-of-thinking-and-inferring; one sees the logical shareable reaching through that practice. As a result, features of thought that had appeared philosophically puzzling can come to be seen as unproblematic; this, at any rate, is the hope. Peter Geach had argued that there is a test whether someone has got the differences in logical kind that Frege was trying to convey through his elucidations. The test (which could be carried out by university examiners, say) would be whether she can work properly with a logical notation like Frege’s (1976, p. 70). A test for whether someone has got the point of Wittgenstein’s elucidations would be what she went on to do when she took herself to be engaging in philosophy.
In our thinking, speaking, and inferring—in such doings—logical shareables are displayed. Learning to see these shareables clearly is a kind of philosophical achievement—on a remarkable and interesting and highly original conception of philosophy. An appreciation of the dangers of this conception of philosophy came later for Wittgenstein. When Wittgenstein said that the way we pose philosophical problems reflects misunderstanding of the logic of our language, he did not mean that we go around saying things that are excluded from the realm of sense; he meant that the misrepresentation of logical kinds is deep in our understanding of our problems.42
Anscombe, G. E. M. (1963). An Introduction to Wittgenstein’s ‘Tractatus’, 2nd edn. London: Hutchinson University Library. 1st edn. London: Hutchinson University Library, 1959.Find this resource:
Anscombe, G. E. M. (1989). ‘The Simplicity of the Tractatus’, Critica 21: 3–14.Find this resource:
Bronzo, Silver (2011). ‘Context, Compositionality, and Nonsense in Wittgenstein’s Tractatus’. In Rupert Read and Matthew A. Lavery (eds.), Beyond the Tractatus Wars: The New Wittgenstein Debate. London: Routledge, pp. 84–111.Find this resource:
Crary, Alice (ed.) (2007). Wittgenstein and the Moral Life. Cambridge, MA: MIT Press.Find this resource:
Crary, Alice and Rupert Read (eds.) (2000). The New Wittgenstein. London: Routledge.Find this resource:
Diamond, Cora (2003). ‘Finding One’s Way Into the Tractatus’, Sats 4: 165–82.Find this resource:
Diamond, Cora (2004). ‘Saying and Showing: An Example from Anscombe’. In Barry Stocker (ed.), Post-Analytic Tractatus. Aldershot: Ashgate, pp. 151–66.Find this resource:
Diamond, Cora (2011). ‘The Tractatus and the Limits of Sense’. In Oskari Kuusela and Marie McGinn (eds.), The Oxford Handbook of Wittgenstein. Oxford: Oxford University Press, pp. 240–75.Find this resource:
Diamond, Cora (2012). ‘What Can You Do With the General Form of Proposition?’ In José Zalabardo (ed.), Wittgenstein’s Early Philosophy. Oxford: Oxford University Press, pp. 151–94.Find this resource:
Floyd, Juliet (1995). ‘On Saying What You Really Want to Say: Wittgenstein, Gödel, and the Trisection of the Angle’. In J. Hintikka (ed.), From Dedekind to Gödel: Essays on the Foundations of Mathematics. Dordrecht: Kluwer, pp. 373–426.Find this resource:
Floyd, Juliet (2000). ‘Wittgenstein, Mathematics, Philosophy’. In Crary and Read (2000), pp. 232–61.Find this resource:
Floyd, Juliet (2007). ‘Wittgenstein and the Inexpressible’. In Crary (2007), pp. 177–234.Find this resource:
(p. 904) Frege, Gottlob (1979). ‘Notes for Ludwig Darmstaedter’, tr. P. Long and R. White. In Frege, Posthumous Writings, ed. Hans Hermes et al. Oxford: Blackwell, pp. 253–7. Originally written 1919.Find this resource:
Frege, Gottlob (1980). Letter to Anton Marty, tr. H. Kaal. In Frege, Philosophical and Mathematical Correspondence, ed. Brian McGuinness. Oxford: Blackwell, pp. 99–102. Originally written 1882.Find this resource:
Frege, Gottlob (1984). ‘On Concept and Object’, tr. Peter Geach. In Frege, Collected Papers on Mathematics, Logic, and Philosophy, ed. Brian McGuinness. Oxford: Blackwell, pp. 182–94. Originally published in Vierteljahrsschrift für wissenschaft liche Philosophie 16 (1892): 192–205.Find this resource:
Geach, P. T. (1976). ‘Saying and Showing in Frege and Wittgenstein’. In J. Hintikka (ed.), Essays on Wittgenstein in Honour of G. H. von Wright, Acta Philosophica Fennica 28: 54–70.Find this resource:
Goldfarb, Warren (2002). ‘Wittgenstein’s Understanding of Frege: The Pre-Tractarian Evidence’. In Erich H. Reck (ed.), From Frege to Wittgenstein. New York: Oxford University Press, pp. 185–200.Find this resource:
Goldfarb, Warren (2011). ‘Das Überwinden: Anti-metaphysical Readings of the Tractatus’. In Rupert Read and Matthew A. Lavery (eds.), Beyond the Tractatus Wars: The New Wittgenstein Debate. London: Routledge, pp. 6–21.Find this resource:
Goldfarb, Warren (Unpublished). ‘Objects, Names, and Realism in the Tractatus’.Find this resource:
Griffin, James (1964). Wittgenstein’s Logical Atomism. Oxford: Clarendon Press.Find this resource:
Hacker, P. M. S. (2000). ‘Was He Trying To Whistle It?’ In Crary and Read (2000), pp. 353–88.Find this resource:
Hylton, Peter (1990). Russell, Idealism and the Emergence of Analytic Philosophy. Oxford: Clarendon Press.Find this resource:
Hylton, Peter (2005). ‘Frege and Russell’. In Hylton, Propositions, Functions, and Analysis: Selected Essays on Russell’s Philosophy. Oxford: Clarendon Press, pp. 153–84.Find this resource:
Ishiguro, Hidé (1969). ‘Use and Reference of Names’. In Peter Winch (ed.), Studies in the Philosophy of Wittgenstein. London: Routledge & Kegan Paul, pp. 20–50.Find this resource:
Kremer, Michael (1997). ‘Contextualism and Holism in the Early Wittgenstein: From Prototractatus to Tractatus’, Philosophical Topics 25: 87–120.Find this resource:
Kremer, Michael (2002). ‘Mathematics and Meaning in the Tractatus’, Philosophical Investigations 25: 272–303.Find this resource:
Kremer, Michael (2007). ‘The Cardinal Problem of Philosophy’. In Crary (2007), pp. 143–76.Find this resource:
McGinn, Marie (1999). ‘Between Metaphysics and Nonsense: Elucidation in Wittgenstein’s Tractatus’, Philosophical Quarterly 49: 491–513.Find this resource:
McGinn, Marie (2006). Elucidating the Tractatus. Oxford: Clarendon Press.Find this resource:
McGuinness, B. F. (1981). ‘The So-Called Realism of Wittgenstein’s Tractatus’. In Irving Block (ed.), Perspectives on the Philosophy of Wittgenstein. Oxford: Blackwell, pp. 60–73.Find this resource:
McManus, Denis (2006). The Enchantment of Words. Oxford: Clarendon Press.Find this resource:
Malcolm, Norman (1986). Nothing is Hidden: Wittgenstein’s Criticism of his Early Thought. Oxford: Blackwell.Find this resource:
Mill, John Stuart (1843). A System of Logic. London: John W. Parker.Find this resource:
Moore, A. W. (2007). ‘Wittgenstein and Transcendental Idealism’. In Guy Kahane et al. (eds.), Wittgenstein and His Interpreters. Oxford: Blackwell, pp. 174–99.Find this resource:
Moyal-Sharrock, Daniele (2007). ‘The Good Sense of Nonsense’, Philosophy 82: 147–77.Find this resource:
Pears, David (1987). False Prison: A Study of the Development of Wittgenstein’s Philosophy, 2 vols. Oxford: Clarendon Press.Find this resource:
Ricketts, Thomas (1996). ‘Pictures, Logic, and the Limits of Sense’. In Hans Sluga and David Stern (eds.), The Cambridge Companion to Wittgenstein. Cambridge: Cambridge University Press, pp. 59–99.Find this resource:
(p. 905) Ricketts, Thomas (2010). ‘Concepts, Objects and the Context Principle’. In Michael Potter and Thomas Ricketts (eds.), The Cambridge Companion to Frege. Cambridge: Cambridge University Press, pp. 149–219.Find this resource:
Russell, Bertrand (1932). ‘Knowledge by Acquaintance and Knowledge by Description’. In Russell, Mysticism and Logic. London: Allen & Unwin, pp. 209–32. Originally published in Proceedings of the Aristotelian Society 11 (1910–11): 108–28.Find this resource:
Russell, Bertrand (1956). ‘On Denoting’. In Russell, Logic and Knowledge, ed. R. Marsh. London: Allen & Unwin, pp. 41–56. Originally published in Mind 14 (1905):. 479–93.Find this resource:
Russell, Bertrand (1967). Problems of Philosophy. Oxford: Oxford University Press. Originally published in Home University Library. London: Williams & Norgate, 1912.Find this resource:
Sullivan, Peter (2001). ‘A Version of the Picture Theory’. In Wilhelm Vossenkuhl (ed.), Ludwig Wittgenstein: Tractatus Logico-Philosophicus. Berlin: Akademie Verlag, pp. 89–110.Find this resource:
Sullivan, Peter (2004). ‘“The general propositional form is a variable” (Tractatus 4.53)’, Mind 113: 43–56.Find this resource:
Sullivan, Peter (2011). ‘Synthesizing Without Concepts’. In Rupert Read and Matthew A. Lavery (eds.), Beyond the Tractatus Wars: The New Wittgenstein Debate. London: Routledge, pp. 171–89.Find this resource:
Urmson, J. O. (1956). Philosophical Analysis: Its Development between the Two World Wars. Oxford: Clarendon Press.Find this resource:
Weiner, Joan (1999). Frege. New York: Oxford University Press.Find this resource:
Winch, Peter (1987). ‘Language, Thought and World in Wittgenstein’s Tractatus’. In Winch, Trying to Make Sense. Oxford: Blackwell, pp. 3–17.Find this resource:
Wittgenstein, Ludwig (1953). Philosophical Investigations, ed. G. E. M. Anscombe and R. Rhees, tr. G. E. M. Anscombe. Oxford: Blackwell.Find this resource:
Wittgenstein, Ludwig (1956). Remarks on the Foundations of Mathematics, ed. G. H. von Wright et al., tr. G. E. M. Anscombe. Oxford: Blackwell.Find this resource:
Wittgenstein, Ludwig (1963). Tractatus Logico-Philosophicus, tr. D. F. Pears and B. F. McGuinness. London: Routledge & Kegan Paul.Find this resource:
Wittgenstein, Ludwig (1967). Zettel, ed. G. E. M. Anscombe and G. H. von Wright, tr. G. E. M. Anscombe. Oxford: Blackwell.Find this resource:
Wittgenstein, Ludwig (2000). Wittgenstein’s Nachlass: The Bergen Electronic Edition. Oxford: Oxford University Press.Find this resource:
(1) I use ‘IWT’ as an abbreviation of the title.
(6) For the significance of epistemology for Frege, see, e.g., Weiner 1999, ch. 2. On the development of Russell’s interest in epistemology, see Hylton 2005; for Russell’s pre-1905 view, see Hylton 1990, p. 197, note 33. Hylton argues that Kant and the logical positivists ‘share an interest in knowledge, and a conception of what it is to account for it, which is not to be found in Russell’s work’ in the period during which he wrote Principles. See also pp. 361–2, where Hylton, writing about Russell’s interest in knowledge after 1910, says of Russell’s earlier works that they ‘show no sign at all of any such interest’; also p. 235, where Hylton explains the changing role given to acquaintance as Russell’s views change after 1905.
(7) The inverted commas around ‘picture theory’ are Anscombe’s, p. 19; also pp. 25 and 41.
(10) On these issues, see Hylton 1990, especially pp. 328–33. As Hylton notes, even as late as 1913, the point of acquaintance, for Russell, is that it is to be ‘an unproblematic meeting ground between the mind and what is outside it’ (p. 331). That the notion of acquaintance is not tied to empiricism is also evident in the writings of Gareth Evans and those who are influenced by him, who use the idea of the direct availability of something for thought, an idea developed from Russellian acquaintance.
(13) Pears 1987, vol. 1, pp. 75–6, 102–3. Cf. also Malcolm 1986, pp. 28–31; Malcolm appears to hold both that there is no such thing as a ‘preliminary preparation’ for language in which signs are correlated with objects outside of propositions and that, if I am to construct a proposition using a name for an object, I must know its possible combinations with other objects. The correlations settle for me the propositional contexts in which that name can occur. This latter point appears to involve granting to correlation-making a kind of logical priority close to the kind of ‘preparation’ for language which he had explicitly ruled out three pages earlier. The idea is certainly that the correlations between names and objects allow certain sign-combinations and disallow others.
(15) Anscombe’s view, that logical characteristics of what a symbol means are plain from its role in propositions, is weaker than the views of Winch, Ishiguro, and McGuinness. Anscombe’s view leaves room for a distinction between presenting what kind of thing a symbol means and settling which thing of that kind it means; but Winch, Ishiguro, and McGuinness give accounts of the Tractatus which don’t leave room for that distinction, at any rate in the case of simple names.
(16) A corresponding problem emerges in Malcolm 1986, if one works through Malcolm’s treatment of the context principle and tries to connect it with Wittgenstein’s distinction between signs and symbols.
(20) Anscombe (1989) gives an account of the central ideas of the Tractatus which makes rather more explicit (than does IWT) some points of resemblance between her reading and that of Sullivan. Some of the features of her reading had indeed changed, but the claim that she makes, that Wittgenstein had in the picture theory solved the ancient problem of the relation between thought and reality ‘by the thesis of the identity of the possibility of the structure of a proposition and the possibility of the structure of a fact’ does not seem to me to mark a change, except in explicitness, from her earlier account.
(22) Moyal-Sharrock 2007. In discussing her view I look only at reasons why it does not provide a solution to the difficulty with which I am concerned. There are other problems with Moyal-Sharrock’s account connected with the topics of section 30.4, but I cannot examine these problems here.
(24) See TLP 4.012, where Wittgenstein says that it is obvious that we take in a proposition of the form ‘aRb’ as a picture. We take in the two names, and that they are combined in such-and-such relation in the proposition, and that is indeed why we take the proposition to signify the holding of a relation between a and b. But no more is implied; to see such a proposition as a picture in this minimal sense does not involve awareness of any logical characteristics shared with other symbols.
(25) Marie McGinn 1999 also emphasizes the role of a reconception of the ordinary in Wittgenstein’s method. Her account differs from mine in drawing a distinction between elucidations which make possible the disappearance of philosophical problems through the kind of reconception of the ordinary to which they lead us and remarks which reflect Wittgenstein’s theoretical preconceptions, including centrally the idea of logical form as expressed in a variable. I think that Wittgenstein’s understanding of logical generality (the generality of a variable) is not separable from the kind of reconception of the ordinary at which he aimed in the Tractatus and which he took to be capable of resolving philosophical problems, but I cannot here go into my reasons for disagreeing with McGinn.
(26) Wittgenstein 2000, TS 220 §93/MS 142 §§105–6; cf. also Zettel §444. Wittgenstein also in these remarks says that it is a characteristic of the sort of theory that he accepted that it doesn’t present itself as a theory; one takes oneself merely to have seen what is there in the clear intuitive case.
(27) See TLP 5.5303, where Wittgenstein contrasts two sorts of case of sentences that are not senseful, one sort being nonsense and the other ‘saying nothing’, i.e., being senseless. Here he plainly allows for sentences which do not have sense but are not nonsense, though not tautologies or contradictions. The passage seems inconsistent with readings that ascribe to Wittgenstein the view that the only sorts of sentences that lack sense but are not nonsensical are tautologies and contradictions. See Kremer 2002 for further discussion of TLP 5.5303.
(29) The sign ‘Red’ in any case has a variety of uses: it is a nickname, and people also may say that Red is what it is worse to be than to be dead, or that Red is what Virginia ceased to be in 2008. One might try to rule out such uses of ‘Red’ by saying ‘The colour Red is a colour’, but that spoils the example by making the sentence appear to be totally empty. The example is meant to be a sentence that appears not to be empty, but which also appears not to be capable of being false. See also Wittgenstein (1956, I, § 105) on the use of colour-words as names of colours in contexts like ‘Black is darker than white’. Although Wittgenstein does not there use the contrast between sign and symbol, the point he makes is that there is a question what the use, if any, is of ‘Black is darker than white’, since the words ‘Black’, ‘white’, and ‘darker than’ are plainly not being used as they are in ordinary useful statements.
(33) The ‘Wittgensteinian twist’ is still very Fregean. See Ricketts 2010, Part 5, on the ‘self-stultifying’ character of the attempt to use the predicates ‘concept’ and ‘object’ to make clear the distinction between concepts and objects.
(34) The expression ‘real rubbish’ comes from Anscombe’s discussion (1989, pp. 10–11) of the contrast between ethical nonsense of the sort for which Wittgenstein had great respect and ethical nonsense which he would have liked to see disappear. But I am not using the term in exactly the way she does.
(36) Ricketts 2010, pp. 191–3. I have stayed very close to Ricketts’s wording in this summary of his account of how we are meant to take Frege’s elucidations of ‘concept’ and ‘object’, but I have departed too far from his exact words to use quotation marks.
(37) The case is similar with my use of the contrasting pair of predicates, ‘logical shareable’ and ‘property’. Like the use of ‘concept’ and ‘object’ as a contrasting pair of predicates, such talk can serve a purpose, despite the confusion latent in it. It can help point us to philosophical activity in which differences can be perspicuously presented; and (as with ‘concept’ and ‘object’), there isn’t then an unexpressed something that is left over.
(38) There is one sentence of Anscombe’s which might be read as implying something like the realist reading, on IWT p. 110, where she refers to objects as the ‘original seat’ of form, but her sentence is meant simply to summarize TLP 2.0121. I don’t think it should be taken to indicate agreement with the kind of object-based readings given by Hacker, Pears, and Malcolm.
(42) I am grateful to James Conant, Michael Beaney, and Alice Crary for their comments and suggestions.