Wittgenstein on the pictorial roots of referentialism

Title Page Paper: Wittgenstein on the pictorial roots of referentialism [This is an original manuscript / preprint of an article published by Taylor & Francis in Inquiry. An Interdisciplinary Journal of Philosophy on 27 May 2020, available online: https://www.tandfonline.com/doi/full/10.1080/0020174X.2020.1768891] Author: Thomas Oehl Affiliation (English): LMU Munich Faculty of Philosophy, Philosophy of Science and the Study of Religion Chair of Philosophy II Geschwister-Scholl-Platz 1 D-80539 Munich Germany E-mail Address: [email protected] ORCiD: 0000-0002-2720-3540 2 Abstract Abstract [197 words]. In this paper I show that Wittgenstein’s explorations of the roots of the ‘referentialist idea’ sketched in PI #1 reach far deeper than has so far been noticed. I shall reconstruct these philosophical explorations in two major parts: First, some subtleties of the inner structure and wording of #1 are shown which the vast majority of readings, above all Goldfarb’s influential one, have failed to account for, which is partly due to a problem in Anscombe’s translation. Second, I shall reconstruct these explorations by offering a new reading of PI #38 in the light of the clarifications from the first part. In #38, I claim, Wittgenstein diagnoses and scrutinises a certain illegitimate analogy to spatial relations we are prone to draw, unintentionally and unconsciously, when reflecting on the meaning of words. This analogy amounts to a certain relational and pictorial structure of our pre-theoretic assumptions about meaning. I will conclude with some reflections on the question of whether Augustine’s words from #1 are, as Goldfarb and Travis aver, ‘innocent’. From my reading it follows that they are not and that Baker’s and Hacker’s qualification of #1 as the target of Wittgenstein’s criticism is correct in its outline. Keywords Keywords [6]. Wittgenstein ; Opening of the Philosophical Investigations (#1) ; Picture ; Meaning ; Referentialism ; ‘Classical’ vs. ‘New’ Reading (of Wittgenstein) 3 Full Text (Paper) Wittgenstein on the Pictorial Roots of Referentialism Introduction #1 of the Philosophical Investigations is not only the opening of a masterpiece written by Wittgenstein, but also the starting point of a certain critical line of philosophical thought. Unsurprisingly, there has been much discussion about the correct understanding of this paragraph in context.1 There are good reasons for this: Wittgenstein himself makes clear that substantial changes in his philosophy have taken place from his first major work, the Tractatus logico-philosophicus, to the second one, the Philosophical Investigations. Thus, it is quite natural to hold that in the Investigations he critically reflects on views he himself held in the Tractatus – and which are still held by others who share the form of thinking that shapes the Tractatus. The so-called ‘classical’ reading of (the later) Wittgenstein, which is mainly represented by Baker/Hacker (1980/20092), indeed argues that #1 of the PI is the place where Wittgenstein presents the target (or at least one focal target) of his philosophical (self)criticism. This target consists in what can be called the ‘referentialist’ conception of meaning, i.e. the philosophical view that the meaning of a word is the thing (or object) for which it stands. The classical reading argues that in the PI Wittgenstein replaces this view by a new one, a conception of the meaning of a word as the ‘use’ of the word, as he puts it in #43 of the PI. Over the years, another reading has emerged which claims that Wittgenstein neither advances a theory nor criticises theories; rather, he performs a form of therapy that makes us get rid of misunderstanding philosophy in terms of advancing theories and criticizing theories. Thus, 1 The great controversies about it are nicely summed up by McDougall (2017). 2 In discussion with Goldfarb (1983) I shall solely quote the 1980 edition. 4 this so-called ‘new’ reading recommends us readers not to interpret PI #1 and its context in terms of replacing a wrong philosophical conception of meaning by a new – and correct – one. Warren D. Goldfarb is to be counted as one of the earliest proponents of the ‘new’ reading.3 In his influential and thought-provoking paper ‘I Want You to Bring Me a Slab: Remarks on the Opening Sections of the Philosophical Investigations’ (1983), written in the spirit of Cavell’s original approach4, Goldfarb objects to the ‘classical’ reading of the opening of the PI.5 His argument is twofold: (A1) #1 cannot be meant to set out the target of Wittgenstein’s criticism, as Augustine’s words in themselves are (philosophically) innocent6; they are simply reporting what is going on when a little child learns her language. There is nothing wrong with such a report; the report itself does not even claim anything philosophical, let alone take up a philosophical position. (A2) If Wittgenstein really intended to present the target of his philosophical criticism here, he would have paraphrased an actual and serious philosophical theory. But the ‘theory of meaning’ sketched in #1 is no such thing. It has never been argued for by any serious thinker whatsoever. So Goldfarb states: Wittgenstein does not ordinarily set up as his opponent one who expresses anything recognizable as a philosophical theory of naming, meaning, mind, or what have you. (Goldfarb 1983, 266) Goldfarb moreover objects to the view that Wittgenstein in #1 makes explicit implicit presuppositions of (such) philosophical theories in order to shed a critical light on them. He rather takes Wittgenstein to be pursuing something more fundamental: He intends to make us 3 He is flanked by people like Cora Diamond and James Conant. 4 Cf. Cavell 1996. Indeed, many of the motifs Goldfarb (1983) turns on are already present in Cavell’s notes. 5 In that respect Goldfarb is followed by Charles Travis (2006). 6 I adopt this telling expression from Travis (2006, 10), who uses it to characterise Goldfarb’s reading. 5 critically reflect on the (very beginning of any) search for theories as such and the meaning of the words we thereby use (e.g. ‘essence of language’): [T]he slogan [sc. making implicit presupposition explicit – or ‘unmasking temptations’, as Goldfarb puts it] can mislead in suggesting that there are specific theses which play the role of unnoticed premises (in, e.g., Frege’s arguments), which are beguiling, but which can be recognized as incorrect or misguided as soon as they are made explicit. Rather, what Wittgenstein wishes to bring to light operates at a more basic level. For in these sections Wittgenstein is examining what it is to begin looking for a philosophical account of language and meaning. (Goldfarb 1983, 266) Just like his opponents, Baker and Hacker, Goldfarb takes his arguments to emerge from a painstaking exegesis of the first paragraphs of the PI.7 This is one reason for why my line of argument in this paper has to include a partly word-by-word exegesis. What I aim at is this: I shall try to refute (A1) and (A2), showing that both indeed rely on a misunderstanding of the inner structure of #1 that is partly due to a linguistic misrepresentation of the text. Moreover I shall try to show that there are some subtleties in #1 than have not been noticed yet. I shall especially confront the question of why Wittgenstein calls the view we abstract from Augustine’s words a ‘picture’. There are interesting things to be said about this which, to the best of my knowledge, all interpretations of #1 have neglected. So the main aim of this paper is to understand a certain line of thought better, a deep reflection on what I call the ‘pictorial roots of referentialism’, not to rebut Goldfarb and the ‘new reading’. Nevertheless my criticism also intends to challenge it by refuting (A1) and (A2), which are taken to support its spirit, and to show that that there are some views the ‘new’ reading is wrong in alleging to the ‘classical’ one. 7 He speaks about a ‘specifically exegetical form’ his investigation has as well as about a ‘close scrutiny’ of what is going on in the text (Goldfarb 1983, 266). Moreover he promises that his own reading can be ‘flesh[ed] out’ by ‘scrutinis[ing] Wittgenstein’s words’ (267). 6 Preparing the Argument Exegetically: The Three-Level-Structure of #1 To see whether the ‘classical’ reading has to give way to the ‘new’ one or whether it can be defended against its objections requires a painstaking word-by-word analysis of #1. This is what I shall try to convey first. Notably, #1 of the Philosophical Investigations8 opens with a quotation from Augustine’s Confessiones. Referring to that quotation (‘These words’ (‘[i]n diesen Worten …’)) Wittgenstein says that by these words we get (‘erhalten’) a picture (‘Bild’) of the essence of human language, which he then describes in two crisp sentences: [T]he words in language name objects – sentences are combinations of such names. By referring to these sentences, i.e. the picture (‘In diesem Bild …’), Wittgenstein continues and states that we find the ‘roots’ (‘Wurzeln’) of a certain ‘idea’ (‘Idee’) in it. This idea is then described in three crisp sentences: Every word has a meaning. This meaning is correlated with the word. It is the object for which the word stands. For the sake of simplicity I shall henceforth call this idea ‘referentialism’9. So this is the bare structure of #1 we have to keep in mind, as it matters crucially for Wittgenstein’s concern with ‘referentialism’ as a somewhat natural and widespread philosophical conviction: (Level 1) Quotation (from Augustine’s Confessions) (Level 2) Picture (reference to (Level 1): ‘In diesen Worten erhalten wir’) 8 If not stated otherwise, I use the Anscombe translation of the PI for quotes (= Wittgenstein 1963). I do not use the translation revised by Hacker and Schulte (= Wittgenstein 2009), as it was not yet available to Goldfarb (1983). 9 This telling word is also employed by Schroeder (2006, 128 et seqq.) and Hanna (2010). 7 (Level 3) Idea (reference to (Level 2): ‘In diesem Bild […] finden wir die Wurzeln der Idee’) With the exception of Baker/Hacker and, most recently, Michael Luntley (2015) most interpreters fail to account for this deliberate structure and wording properly.10 But this threelevel-structure fulfils a genuine philosophical function, as we shall come to see by the following scrutiny. Readers who lack an overview of this structure are thus likely to end up in misinterpretations of Wittgenstein’s views. The Two Changes in the Two Transitions We have to look closely at what changes occur in the transitions from one level to the subsequent one each and why they occur. The First Change: Abstraction from Users and Use The first change consists in an abstraction from persons and their use of words. This is obvious from Wittgenstein’s words, as soon as one pauses to reflect: Wörter benennen Gegenstände. Whereas in the quote from Augustine it says: cum ipsi (majores homines) appellabant rem aliquam… (When grown-ups named some object… (Nannten die Erwachsenen irgendeinen Gegenstand...)) There is an important difference between saying that persons name objects (in their stream of life and for certain ends) and saying that names (or words) name objects. This difference has been put tellingly by Peter Strawson: 10 How deliberate this structure and its wording are can also be proven philologically, i.e. by reference to the genesis of #1. Cf. Pichler (2004, 230 et seqq.) on this. 8 ‘[R]eferring’ [.] is not something an expression does; it is something that someone can use an expression to do. (Strawson 1950, 326) This difference as it is displayed in the first transition in #1 is obscured by Goldfarb: The notions Augustine invokes, like “naming some object”, “wishing to point a thing out”, and “state of mind”, can be entirely unexceptionable; after all, we use them all the time. However, Wittgenstein seeks to show that when these notions are used in certain contexts, they come to have a weight that our ordinary understanding of them does not support. (Goldfarb 1983, 268) It is surely right that the abstraction of a phrase from its (ordinary) context and use and the creation of a new ‘philosophical’ context in which it becomes embedded is one (widespread) mechanism of getting into philosophical confusion. It is a mechanism that Wittgenstein generally intends to draw our attention to. But Goldfarb’s very general reference to it is far too little specific for an adequate understanding of what is going on here, in #1; for it is not any random abstraction from the (ordinary) context in general, but a very specific abstraction that makes the first transition problematic: the abstraction from persons who use words and the abstraction of the verb ‘naming’ from the being of which it can be predicated sensibly, i.e. the person. The boldness of Goldfarb’s reading seems to me to reflect a general weakness of the ‘new’ reading: by trying to draw attention to a supposed therapeutic style of thinking (as opposed to discussing particular philosophical views on certain matters), the proponents of the ‘new’ reading fail to account for details and subtleties of the very text that they claim to come closer to. But it would be a misleading simplification to think that an imprecise reading of #1 only occurs among the proponents of the ‘new’ reading (and because of its weaknesses). Eike von Savigny (1994, 33 et seqq.), in his meritorious and quite autonomous commentary, misrepresents the transition in #1, too. He is wrong in claiming, as he does, that both (Level 1) and (Level 2) each overemphasise, or pay attention exclusively to, the relation between a 9 speaker and an object − for there is no speaker on (Level 2) anymore. So von Savigny’s reading obscures the difference between (Level 1) and (Level 2) and thereby gives rise to the following argument: (Premise1) (Level 1) ≈ (Level 2). (Premise2) Wittgenstein in #2 explicitly takes Augustine’s description (= (Level 1)) to be correct for primitive cases of language like the language-game of #2. (Conclusion) Neither (Level 1) nor (Level 2) can be the target of Wittgenstein’s general criticism in Baker’s/Hacker’s sense because the views located on them are each not profoundly mistaken, but only restricted to a small(er) scope.11 Consequently, von Savigny (1994, 37) attacks Baker/Hacker for their claim that (Level 2), the picture, is presented as the main target of Wittgenstein’s fundamental criticism. But the above argument rests on (Premise1), which is false. So the argument collapses. Baker and Hacker are right in emphasising the falsity of (Premise1) and show its relevance with respect to Wittgenstein’s claim of #2: [This paragraph] describes a language (subsequently referred to as a ‘language-game’) for which the description given by Augustine [= sc. (Level 1)] is right. Note that it is not contended that the Augustinian conception of the ‘essence of language’ [= sc. (Level 2)] is right for this proto-language, merely that Augustine’s description is appropriate for it. (Baker/Hacker 1980, 65)12 11 This conclusion in the end resembles Goldfarb’s criticism of the classical reading in (A2). 12 Repeated and even more explicitly emphasised also in Baker/Hacker 2009, 3 fn. 2. This point is sometimes overlooked, e.g. by Hanna (2010, 23), who wrongly takes ‘[t]he Augustinian language-game of the builders in PI 2, 6 and 8’ to be ‘a living picture […] of a Pure Referentialist language’ (by which he means an elaborated version of (Level 3)). Note that using ‘Augustinian’ is Baker/Hacker’s way of indicating that the conception is not simply Augustine’s one, i.e. their way of marking one central aspect of the difference between the levels. Baker/Hacker (1980, 36) also explicitly define their use of the term ‘Augustinian picture’: ‘The family of philosophical accounts of 10 This is because the builders in #2 are persons who use words. Since there are no persons on (Level 2) anymore, the language-game from #2 cannot meet it. The pressing question now is: How and why do we ‘get’ (‘erhalten’) this picture from Augustine’s words? In other words: Are we really inclined to abstract from users and use and, if so, why? We shall revert to this question later. Before that some linguistic reflections are necessary: (i) what is expressed by using the word ‘get’ (‘erhalten’) in the first transition and (ii) why does Wittgenstein interject the phrase ‘it seems to me’ (‘so scheint es mir’)? Ad (i): The German word ‘erhalten’ does neither mean ‘we are imposed/forced’ (this would overemphasise the passive aspect of ‘erhalten’), nor ‘we (sometimes) infer/could infer from it’ (this would overemphasise its active aspect). Moreover Wittgenstein does not use the German Konjunktiv (subjunctive) here. Anscombe translates as follows: These words, it seems to me, give us a particular picture of the essence of human language. (italics mine) This translation may easily give rise to a misunderstanding. It suggests that the words themselves impose the ‘picture of the essence of human language’ on us, by ‘giving’ it to us. But this does not capture the meaning of ‘wir erhalten’ accurately enough. Literally, ‘wir erhalten’ is to be translated as ‘we get’. Anscombe’s unfortunate translation is the seedbed of Goldfarb’s paraphrase, which indeed misrepresents Wittgenstein’s deliberate wording: Wittgenstein begins, as is well known, with a passage from Augustine’s Confessions, and tells us that the passage expresses a particular conception of the essence of language. (Goldfarb 1983, 267; italics mine) meaning that grow out of it [sc. the Bild] are fully blown ‘theories’. This family we will call ‘the Augustinian picture’.’ 11 Firstly, Wittgenstein does not speak about a ‘conception’, but about a ‘picture’ of the essence of (human) language. The importance and implications of the use of ‘picture’ will be discussed in section 3 of this essay. Secondly, he does not ‘tell us’ that the passage ‘expresses’ (nor, as Goldfarb (1983, 268) later suggests, ‘contain[s]’) this picture, but that we ‘get (erhalten)’ it from the passage, as I am trying to show at the moment.13 Goldfarb’s misreading has far-reaching implications. He takes his claim (A2) to be supported by the fact that the reader is ‘shocked’ by Wittgenstein who says that these words express the picture which he then describes (and, generally, express anything like an essence of human language). Thus he writes: Thus, I would take Wittgenstein’s first sentence after the quotation, to the effect that Augustine’s remarks contain a definite picture of the essence of language, to be intended to shock. Many commentators14 would have us meekly acquiesce to this sentence, [.] whereas I suggest that Wittgenstein means to call up amazement. This is giving the essence of language? This is a philosophical conception of meaning? (Goldfarb 1983, 268) I think there are three mistakes here: (1) Goldfarb illegitimately jumps to (Level 3), where the notion of ‘meaning’ is introduced; but he even misrepresents this level, as he talks about a philosophical conception of meaning (instead of a mere ‘idea’ behind it). It would indeed by amazing if (Level 3) were declared to be a philosophical conception of meaning. But this is neither what Wittgenstein says nor what the ‘classical’ reading holds. Baker/Hacker on this: 13 It should be noted that in crucial pieces of the literature the structure of Wittgenstein’s wording here is not scrutinised in detail, but rather summed up in quite condensed paraphrasings: Stern (2002, 435) paraphrases ‘that the opening quotation from Saint Augustine contains a definite picture of the essence of language’; von Savigny (1994, 33) glosses that the picture is found in Augustine (‘das bei Augustinus gefundene Bild von der Sprache’); and McGinn (2013, 39) writes ‘that Wittgenstein uses [the words on (Level 2) and (Level 3)] to sum up Augustine’s view’. 14 In a footnote he explicitly names Kenny, Malcolm and Baker/Hacker. 12 W. does not take Augustine to expound an interesting ‘theory of meaning’. […] What interests W. is Augustine’s pre-theoretical, pre-philosophical picture of the working of language which informs Augustine’s own remarks on language as well as a multitude of sophisticated philosophical analyses of meaning. (Baker/Hacker 1980, 61)15 (2) What Goldfarb says about ‘amazement’ and ‘shock’ is plausible given his reading that Wittgenstein says that Augustine’s words ‘express’ a picture of the essence of human language. If Wittgenstein had said this, an attentive reader might indeed be immediately forced to ask: Why on earth – and how – should these few words express the essence of human language? But Wittgenstein does not say this. And this is why Goldfarb’s reading fails. This failure, however, not only misses the point, rather it turns the actual structure of #1 on its head: Wittgenstein does not intend to cause the reader’s amazement here in order to make us reflect on the very beginning of theorising as such, as Goldfarb avers. Rather, Wittgenstein can be sure that (almost) no reader will be irritated by anything so far.16 (3) Goldfarb further alleges that ‘[m]any commentators’ (namely Kenny, Malcolm and Baker/Hacker) ‘would have us meekly acquiesce to this sentence’. By that, I suspect, he means the following: These commentators fail to realise that there is this shocking absurdity (the claim that these words ‘express’ …) that Goldfarb points out, but hold that there is a kind of logically coherent argument going on instead. But this misrepresents the ‘classical’ reading. First, these commentators can neither realise nor not realise this absurdity, for it is not in the text, as was shown in steps (1) and (2). Second, these commentators – at least Baker/Hacker – indeed do argue that the transitions from one level to another are deeply 15 It is put even more clearly in Baker/Hacker 2009, 49−50. It should be noted that calling the picture ‘Augustine’s’ is problematic, according to my reading. Baker/Hacker seem to be misled by the Anscombe translation (‘give us’) here. 16 I am at one with Lugg (2013) who suggests reading the first paragraphs of the PI ‘straightforwardly’, i.e. without alleging the ‘shock’-strategy pointed out by Goldfarb to it. 13 problematic; they do not take us readers to be justified in accepting them, but take us readers in fact to accept them – or, more precisely, to have accepted them already without noticing. Ad (ii): That Wittgenstein takes many of us readers to have adopted the picture moreover is suggested by the word ‘we (wir)’, whereas in the second crucial phrase, the intersection ‘it seems to me (so scheint es mir)’, he uses the first-person singular pronoun. I take the function of this to be the following17: Wittgenstein naturally contrasts the ‘wir (we)’ of the first crucial phrase (‘erhalten wir’) by using the first-personal singular pronoun, as it is precisely not common and widespread that one (critically) reflects on the picture of (Level 2), let alone the traces in western philosophy where we get (‘erhalten’) it from. It is common and widespread that we have it (which I argued above); but it is an exception that anyone is aware of having it, let alone (critically) reflects on it. Wittgenstein has so far been alone in doing so, being a critical pioneer.18 The Second Change: Introducing the Notion of Meaning What is to be said about the second transition? First, it should be noted that the abstraction of the first change is continued: On (Level 3) we are still told about words, (their) meaning and things, not about language-users that are familiar with meanings or capable of understanding the meaning of the words they use. Secondly, I agree with Eike von Savigny about the basic answer to the question of why the meaning of a word is inserted into the picture the way it is in the second change: It simply (and literally) fits very well into the picture of (Level 2). There is a prepared space for it. (I shall say more about this in section 4). Moreover it serves the purpose of simplicity as well as of ontological parsimony to identify the meaning of the 17 Which seems to me to be far more supported by the text than Goldfarb’s (1983) interlocutor-reading is. 18 This is why Glock (1996, 41) is not exaggerating when he remarks that ‘Wittgenstein was the first to subject this position to sustained criticism’. 14 word with the thing it names, as on (Level 2).19 So the second transition exemplifies a further widespread mechanism of getting into philosophical confusion: misguided and simplified pictures offer spaces that are apt to be filled by focal philosophical concepts (such as ‘meaning’ or ‘thing’). By filling them into confused pictures, the confusion is not only prolonged, but even magnified. Again, it is worth asking the question of why Wittgenstein uses the phrase ‘we find the roots of the [following] idea’ (‘finden wir die Wurzeln der Idee’) for the second transition. It seems to me that there are three aspects: (i) By using the word ‘find’ Wittgenstein emphasises that it is only by engaging in critical reflection that one comes to see that the picture actually is the root of the following idea. (ii) Calling the picture the ‘root(s)’ of this ‘idea’ is to say that the picture, at least in part, is the source of the misconception which this idea actually is. A root is what keeps a plant nourished and alive without thereby being itself the plant. If the picture disappeared (or stopped bothering philosophers), the (philosophical) idea of meaning would (be likely to) disappear too. (iii) It is a mere ‘idea’ and not a full-fleshed philosophical theory that Wittgenstein sketches on (Level 3). So Wittgenstein is not committed to the view that any serious philosopher offers a theory which is as simple, as bold and as explicit as this idea. That is rightly emphasised by Goldfarb (but has, to my knowledge, never been doubted by Baker/Hacker). But Wittgenstein further claims that lots of philosophers build theories which are actually – consciously or not – elaborations and concretions of this idea that ramifies in these theories. Whether Wittgenstein is right with this diagnosis and, if so, which philosophers belong or do not belong to this group is an open question so far, to be sure. 19 Cf. von Savigny 1994, 35. 15 So I think it is illuminating to distinguish two aspects of the meaning of ‘idea’ here: this ‘idea’ is the kernel of a full-fleshed philosophical theory. This might mean two very different things: ‘He had an idea and later elaborated it in his book’ – or: ‘His book consists of many subtle thoughts, but the main idea behind it is …’. In the first case ‘idea’ is used to signify a focal, far-reaching thought that comes to one’s mind and is fundamental for a deliberate composition of a book etc.; in the second case ‘idea’ indicates what a certain complex of thought finally amounts to, what its inner principle is. This may well be hidden to the author herself as well as to (lots of) the readers.20 In the first case the idea is something present to the thinker’s mind, whereas in the second case it is (or at least can be) something ‘unconscious’, ‘like an invisible force, evident only in its visible effects’, as Baker/Hacker (1980, 46) nicely put it. PI #38: Digging Deeper to the Pictorial (and Relational) Roots of Referentialism Now the open question is this: why do we – often, perhaps even normally – accept these transitions? Why is referentialism rooted so firmly in such a ‘proto-theoretic paradigm’21? Why does the picture, functioning as the ground for these roots, normally never shake, let alone collapse? Why do we rather seem to be attracted by (Level 2) and (Level 3) and their fruits? I think that Wittgenstein provides an answer to this question. This answer can be extracted from #38 of the Philosophical Investigations: 20 Goldfarb (1983) tries to prove his reading by the claim that Frege is not committed to the kind of psychologism that seems to be attacked at the beginning of the PI. This claim he tries to justify by a quote in which Frege declares his anti-psychologist attitude. But that does not prove anything. For as I just said (and as I take Wittgenstein to think), the actual upshot of one’s philosophical thought may well be hidden to oneself and be against one’s intentions. 21 This is a telling expression of Glock’s (1996, 41). 16 Naming [note that the word ‘benennen’ from #1 is used again here!] appears as a queer connexion of a word with an object. – And you really get such a queer connexion when the philosopher tries to bring out the relation between name and thing by staring at an object in front of him and repeating a name or even the word “this” innumerable times. For philosophical problems arise when language goes on holiday. And here we may indeed fancy naming to be some remarkable act of mind […]. Wittgenstein here tells us the story of the birth of the presuppositions of referentialism; more precisely, of the transition from (Level 1) to (Level 2). That is, he digs deeper to the root(s) of referentialism. The insights he thereby gains, I think, matter crucially for an adequate understanding of what really is the target of criticism in the PI and, thus, of the project of (later) Wittgenstein’s philosophy as a whole. But they also matter beyond issues in Wittgenstein, as they offer an explanation for why philosophers are prone to accept referentialism as an often unquestioned standard framework in semantics. This again is not only of interest in (philosophical) semantics, but also if philosophy aims at understanding the roots of common (mis)conceptions of the human mind – something that, I think, is a focal concern of many of the great philosophers, such as Plato, Kant, or Wittgenstein. In that spirit, I shall now try to reconstruct the points Wittgenstein is making in #38: (i) Wittgenstein here describes a philosopher performing a kind of experiment which, qua experiment, is meant to ensure, as it were, the ‘scientific’, ‘formal’ standard of her (precarious) discipline. The philosopher is fascinated by the fact that by using the word ‘tree’ she is able to speak about a real tree she sees ‘in front of’ her, ‘in front of’ her eyes.22 (ii) She now asks: How is it possible that I, being here, can speak about this (the tree), being there ‘in front of’ me? How is it possible to reach from myself to the thing by means of words that belong to me, the utterer (and not to the things)? How can I ever bridge this gap? 22 Interestingly, he only focuses on spatiality, not on temporality. 17 (iii) One decisive move in (ii) can easily be overlooked, which is why it deserves separate notice: I said that the philosopher from #38 holds that the words belong to her. This is an ambiguous phrase. If it is meant to mean that I use words (whereas the tree does not), it is kind of a crude truism. If it is meant to mean that my location in space coincides with the location of the words I use, it is nonsense. For as far as we are concerned with words here (in doing philosophy of language or semantics), words do not have any spatial location. I think that an illegitimate application of spatial categories lies at the ground of many philosophical conceptions that, by the same token, are confused (which is not to say that they are without any merits). Let me just give one example which Wittgenstein himself points to at the end of the above quote (‘act of mind’): words and their meanings are often taken to be ‘in’ the mind, and the mind itself is taken to be where I am (and not outside there), or even ‘(with)in’ me. Here the above ambiguity applies analogously: such a way of talking can have a truistic meaning, but it can also be confusing nonsense. Some German Idealist talking about what is ‘within consciousness’ as well as the ‘internal-external’-distinction in current semantics and epistemology are examples for being affected at least by this ambiguity. Maybe some of that way of talking can be defended; but that can only be achieved my means of critical reflection, not by the straightforward assurance that ‘all this is meant differently’ (remember the Wittgensteinian insight that what we intend to say is not sovereign, as it were, overleaping the words we use in trying to say what we intend to say). So back to Wittgenstein: I think it would be worth rereading his criticism of a certain conception of ‘mental acts’ in the light of the problem of an (illegitimate) application of spatial categories. This would also be an answer to another concern of Goldfarb’s, who is unsatisfied with conceiving of this criticism in terms of (psychologist) mentalism.23 I cannot pursue this here any further. 23 Cf. Goldfarb 1983. 18 (iv) The next of the steps by which the pre-theoretic assumptions are formed is this: the experiment-philosopher crosses herself out of the scenario (and thereby starts conceiving of it as a static picture instead of a dynamic scenario)24: by ‘dynamic’ I mean that, originally, there is a person who uses words to refer to things, as we saw Strawson putting it. Using something is a process rather than a relation; it is something one does, not something that just ‘is there (or is the case)’. Now a word, if one does not conceive of it as an essential moment of the process of using it for certain ends, becomes something ‘static’, because it itself, trivially, does not and cannot do anything. Again, philosophical jargon leads us astray – precisely the one from (Level 2): ‘words name things.’ Here we again face an ambiguity: if that is a more or less fortunate abbreviation for ‘speakers use (some) words to name things’, it is fine; if it means to describe a static relation between the word itself and the thing, it is, according to Wittgenstein, mistaken; but even if one could give reasons for committing oneself to such a relation, the phrase would stand in need of justification by these very reasons. But if one commits oneself to that relation unconsciously and pre-theoretically, this means that such justification is not given, and that (possible) confusions are not even confronted. (v) But why, and how precisely, does the philosopher cross herself, as the actual user of the words, out of the scenario? First, she repeats the name ‘innumerable times’ (like every experiment, this one is also meant to be repeatable!) – without any actual use (apart from performing this supposedly fruitful ‘experiment’). She thus abstracts it from its proper use, so that there is indeed no framework of usage anymore. The philosopher is not doing anything in a proper sense (not speaking fluently, not looking for a thing, not giving a command, nor anything else). Rather, she allows language to go ‘on holiday’. One can compare this to a child using a hammer to throw it into the air and catch it again, just for fun. There is nothing 24 As already noted above, von Savigny (1994, 84) all along emphasises the ‘social practices’ and, hence, takes #38 to reveal language’s ‘going on holiday’ when and because a second person is absent. On my reading, even the only person is missing, so to speak. 19 problematic about that, as long as the child does not claim that, by doing this, she knows what a hammer actually is. But that is what the philosopher, once she has formed her pre-theoretic assumption, does do, analogously. Second, the philosopher, by doing this kind of ‘experiment’, takes up the position of an attentive, though passive observer. As is characteristic of an experiment (e.g. in the natural sciences), the observer qua observing does not observe herself, but things that are not herself. So playing the role of an observer here is the flipside of abstracting from the user of words (of crossing herself out of the picture) – which is precisely the reason why (such) experiments lead philosophy astray: philosophy is to be done by (self-)reflection, not by observation, as Wittgenstein notably insists. (vi) Now what does the experiment-philosopher observe? Well, the word and its now puzzling connection to the thing ‘it’ names. The question is no longer: How can I bridge the gap by using a word? But rather: How is it possible for this word to bridge the gap? So the philosopher’s question takes this form: How does the word touch the thing it names? How does it (being here) reach up to the thing (over there)? How is the word linked with the thing? How is all this (made) possible? Now, as we have said, the word cannot do that, for it cannot do anything. This is why we need a third factor that functions as a bridge that again links words and things. So the – quite formal, seemingly trivial and innocent – answer to the question is: all this is possible by a certain relation that links words with things. This relation is dubbed ‘reference’. So once you have done such an experiment, ‘you really get such a queer connexion’, as Wittgenstein nicely puts it. It should be noted that in the original (and correct) setting, where the language-user remains present, she herself fulfills the function of ‘linking’ words and things: not by means of a relation called ‘reference’, but by using names in order to refer to things. As her crossing out of herself, the language-user, implies the crossing out of her fulfillment of the function, words and things then stand in need of being 20 ‘linked’ by means of a different link: the reference-relation. The reference-relation thus is introduced to ‘replace’ the language-user who has crossed herself out of the setting.25 (vii) The steps taken so far already imply (or at least strongly suggest) that the referencerelation thus created is a singularetantum. For the philosopher, being conscious of the ‘scientific’ form of her experiment, argues that you can do the same experiment with regard to any random name and any random thing, i.e. for any case of ‘naming’. In all cases the word, though being a different one each, ‘names’ the same way – which moreover seems to be indicated by the uniform notion of ‘naming’ itself. So let us take a closer look at this reference-relation, as it results from the ‘experiment’ described in #38. It is to be represented as follows: 𝑅𝐸𝐹𝐸𝑅𝐸𝑁𝐶𝐸 𝑊𝑜𝑟𝑑/𝑁𝑎𝑚𝑒 → 𝑇ℎ𝑖𝑛𝑔/𝑂𝑏𝑗𝑒𝑐𝑡 This illustration is a widespread and widely accepted one in philosophy (at least in my experience). It looks like an innocent and acceptable philosophical axiom even for those who do not (mean to) find themselves in a scenario like the one described in #38, as it seems to be the mere corollary of some basic (philosophical) truisms: (F1) Words and things are related to each other because (some) words are names of objects (names represent objects). (F2) The relation of a word and a thing is neither a word nor a thing. Hence, it must be a third factor, an onto-semantic one. (It connects the ‘realm of words’ and the ‘realm of things’.) These very simple ‘indisputable facts’26 are actually errors (or misinterpretations of innocent facts): 25 Curiously enough, the idea of a ‘mental act of meaning’, a mental correspondent to the reference-relation, reintroduces the speaker into the picture. This is why it is mentioned in #38 too. 21 Ad (F1): As in lots of cases, it is mistaken to translate ‘x of y’ into ‘x stands in a relation to y’ without further ado. And even if it is not, this translation does not help us a great deal. It rather raises the question of what is meant by ‘relation’ here.27 Ad (F2): (F2) obviously depends on (F1). So rejecting (F1) implies rejecting (F2). But more can be said about (F2). Remember the illuminating quote by Strawson, which nicely catches Wittgenstein’s idea of explaining the meaning of an expression by means of describing its correct use. Names are words that can be used for certain purposes, e.g. for picking out a pupil in my class (by calling his name: ‘Michael!’, or by saying to the class ‘You have been very unfriendly to Michael.’). By such utterances we make clear to/of whom we are speaking. We can use names to pick out the person/thing we intend to speak to/about out of a certain group of people/things. Wittgenstein himself recommends us to compare different (types of) words with different tools in order to recognise their instrumental-character in general and, in a concrete case, their (instrumental) functions in particular: The name is comparable to a pointer we can use to make clear whom we are addressing (and whom not) or whom we are predicating something of (and whom not).28 It is pretty unnatural and confusing to subsume such different uses of that ‘pointer’ under a notion of ‘naming’ that thus is suggested to be uniform. Pointing to that instrumental character has a bearing on the question of why a name and a thing do not stand in a relation to each other, let alone a relation that is a third factor bridging the gap between two realms. An instrument is nothing we can make sense of without its being 26 Baker/Hacker (1980, 46) use that telling and ironic expression in the same context. 27 It is surely possible to understand the notion of a relation broadly. But that does not matter here, since it is not such a broad, but a very specific (quasi-mathematical) notion of a relation that is essential to the referentialist view. We will revert to this in section 4. 28 This analogy is only meant to illustrate the instrumental character of a name in general; of course there are important differences between a pointer and a name in detail, e.g. one can use a name to talk about somebody even if she is not spatiotemporally present. 22 used. A hammer would not be a hammer without its function, a name would not be a name without its function to address somebody, to talk about somebody, etc.29 As Heidegger, being at one with Wittgenstein in this respect, has rightly emphasised, a process (or use) is prior to its factors in the sense that those factors would not be what they are (and could not be understood properly) if one abstracted them from the framework of usage. Now it can easily be seen how mistaken the ‘relationalist’ framework, as I like to call it, of the picture of (Level 2) is: A name and a person who is addressed by another person who uses this name do not stand in a relation to each other, but are factors of the ‘instrumental’ process of using a name for certain ends (here: a person’s using a name to address another person). A name and the one it ‘names’ (here: addresses) do not stand in a relation to each other qua naming, but are factors of the instrumental process of ‘naming’ (here: addressing)30. So there are no such relations in these frameworks of usage. If philosophical accounts of meaning in a Wittgensteinian sense want to draw on relationalist vocabulary, they have to clarify what that means and show that it really explains anything. At any rate, they cannot take the meaning of ‘relation’ for granted which is taken from a mathematical model.31 Thus, Anthony Kenny’s (2006, 122) talking about a ‘meaning-relation’ is confusing and rather disarms Wittgenstein’s position because it represents it as concerned with the question of how to characterise the ‘meaning-relation’ adequately. But that fails to put the question marks deep enough down, to use a Wittgensteinian phrase. Wittgenstein does not aim to 29 What one can use a name to do cannot sensibly called ‘naming’ in all cases. When the teacher calls up Michael by saying ‘Michael!’ she is addressing him. When she says ‘Michael is a lovely guy’ she talks about him. When she says ‘Michael, come here!’ she is calling him. All this is done by using proper names. 30 Note how funny ‘addressing’ in #38 sounds against the background of what we have just said. There we are told about a philosopher who addresses things by repeating a (proper) name. Von Savigny (1994, 84) rightly and nicely reminds us that you only can address a person, but not a thing. Surprisingly, this has an immediate bearing on the question of whether Augustine’s words from (Level 1) are ‘innocent’ (see section 5 on this): It is precisely this confused talking about ‘addressing things’ that is patent in the quote from the Confessions: ‘Cum ipsi (majores homines) appellabant rem aliquam …’ (italics mine). 31 See fn. 27. 23 characterise the ‘meaning-relation’ more adequately than other philosophers. Rather he aims to get us to realise the fact that there is no meaning-relation. The meaning of a word stands even less in a relation to the word than the thing itself stands in a relation to the name32; nor does meaning itself consist in a uniform relation (‘reference’) that relates words and things. It is not by coincidence that Kenny fails to avoid relationalist vocabulary. He does not pay sufficient attention to the difference between the different levels33 and, in doing so, overlooks what Wittgenstein’s separate discussion of (Level 2) contributes to his criticism: already prior to the introduction of the notion of ‘meaning’ on (Level 2), a profound and far-reaching mistake has happened: the setting of the relationalist framework. Glock (1996, 256) rightly emphasises that ‘there is no such thing as the name-relation’. I add: there is no ‘namerelation’ at all. It is now patent that and why referentialism leads one astray: it makes us search for a relation rather than for frameworks of usage. So the basic misconception, the very idea that there is a relation between words and things called ‘reference’, is fully present already on (Level 2) – understood in the light of #38; and it amounts to a perfect confusion as soon as it is filled in with the concept of ‘meaning’ on (Level 3). And again, even if this were the right conception, the hidden pre-theoretic steps leading to it ought to be confronted at least, and a defence of such a conception would especially be in need of prior clarification and defence. Philosophy does not only consist in saying true things, but also in showing why they are true – even if we are prone to hold them to be true ‘naturally’. As I said above, we now need to scrutinise in more detail (i) the role of spatial relations in getting to the relationalist framework and (ii) the special (‘pictorial’) character of the relation that reference is wrongly taken to be. Both will be the topic of the following section. 32 According to (Level 3), both are two sides of the same coin, to be sure. 33 This is patent from the fact that Kenny (2006, 122) wrongly conceives of Augustine’s words as ‘St Augustine’s naïve theory of language’ (italics mine). 24 Why the Picture and Referentialism Do Lead to the Tractarian Conception – the (quasi)Mathematical Character of the ‘Reference-Relation’ In #38 we saw Wittgenstein digging up the roots of the referentialist misconception: In performing the ‘experiment’ described, we draw a misleading analogy between our spatially being-confronted with things and our referring to these things by using words. Referring is nothing analogous to a spatial relation, not even to the extent of being a relation of this kind at all. The upshot of #38 is a therapeutic hint to this in its best (and a non-mysterious) sense. The reference to #38 helps us to answer the question of why precisely Wittgenstein uses the term ‘picture’ for characterising (Level 2). This question has, to my knowledge, not been raised by the interpreters yet, let alone be answered. It should be noted that ‘picture’ is again used in a very condensed and important remark in PI #115 that summarily points to the root(s) of our fundamental confusion(s) – especially, I take it, to the picture from #134: A picture held us captive. And we could not get outside it, for it lay in our language and language seemed to repeat it to us inexorably. So we can assume that ‘picture’ indicates a very distinct issue of #1. Our interpretation with reference to #38 has shown which: a picture in its literal sense is essentially something relationally structured. What we paradigmatically call (visual) ‘pictures’ (e.g. paintings, photographs, schemes) is always something relationally, if not necessarily spatially, structured.35 It lacks any temporal structure, just like the ‘experiment’ of #38 ignores any temporal aspect of language-use. Spatial structure is a case of relationality. In the genesis of referentialism, as it is described in #38, spatiality is dropped, but its relational form is conserved. Then we have the picture of (Level 2), which is the seedbed for referentialism. 34 For its relation to the ‘picture’ of #1 cf. also Baker/Hacker 1980, 46. 35 Which, of course, does not mean that it is exhausted or even sufficiently characterised by that feature. The spatial (and hence relational) structure is a necessary, not a sufficient condition for calling something a ‘picture’. 25 Referentialism thus is rooted in (pictorial) relationalism. The explicitly relationalist version of (Level 2) thus amounts to the following picture: 𝑅𝐸𝐹𝐸𝑅𝐸𝑁𝐶𝐸 𝑊𝑜𝑟𝑑/𝑁𝑎𝑚𝑒 → 𝑇ℎ𝑖𝑛𝑔/𝑂𝑏𝑗𝑒𝑐𝑡 As I said above, my claim that referring (rightly understood as using words for certain ends) is not a relation is in need of further qualification: it means that it is not a relation in the sense that referentialism embraces, namely in a (quasi-)mathematical sense. In order to explicate this sense, we will draw on the Tractatus, which will make us also take a further step in taking issue with Goldfarb’s (A2). The question is: Where exactly (and concretely) is the link between (Level 3) of #1 and false (but elaborate) theories of meaning like those of the TLP? This question should be taken seriously. Otherwise Goldfarb’s objection (A2) somehow is resurrected in a new guise, namely in the suspicion that we are talking about the kernel of elaborate philosophical theories abstractly, but cannot explain concretely how this kernel manifests itself in them. Where do we find the link? Baker and Hacker have written extensively and, I think, convincingly on that issue.36 I take my scrutiny of the roots of referentialism to help to strengthen their reading by the following line of thought. Consider again the relationalist scheme: 𝑅𝐸𝐹𝐸𝑅𝐸𝑁𝐶𝐸 𝑊𝑜𝑟𝑑/𝑁𝑎𝑚𝑒 → 𝑇ℎ𝑖𝑛𝑔/𝑂𝑏𝑗𝑒𝑐𝑡 The reference-arrow points to a ‘thing’. But where or what exactly does it point to? This question resembles the problem of ambiguity of the ostensive definition, which Wittgenstein discusses at length in the PI: In order to explain the meaning of the word ‘banjo’, I can say ‘This is a banjo!’ and point to the banjo. For the person I am talking to it might be unclear 36 Cf. e.g. Baker/Hacker 1980, 57 et seqq.. 26 whether I point to the instrument as a whole, to its head, its style or whatever else. This ambiguity accompanies all ostensive definitions theoretically (cf. PI #28). However, it rarely occurs as a problem in our language-use – simply because there is a context by which it is determined what the speaker is pointing to. But the relationalist scheme of (Level 2) has no such context available. It instead intends to avoid the problem of ambiguity in a quasimathematical, instantaneous, isolated manner. It maintains a relation between a word and a thing, which is realised by the ‘onto-semantic’ factor called ‘reference’, a kind of glue that brings word and things together to ensure meaning. Reference is thus conceived to be a relation, more precisely a function that maps words (names) onto things (objects). This is said quite explicitly on (Level 3): ‘Diese Bedeutung ist dem Wort zugeordnet’ (italics mine). Note that the German word ‘Zuordnung’ is used as a terminus technicus in mathematics to characterise functions. This functional conception seems to be attractive because of its elegance, simplicity and quasi-mathematical purity. This purity implies the quasi-mathematical ontology of the Tractatus having two aspects: (i) The reference-relation is a function only if it maps each word to exactly one thing. This is the solution to the problem of ambiguity. However, it seems strikingly inadequate. ‘Francis’ is both the name of Francis Bacon and the current Pope, as we would say. The relationalist scheme might cope with that by insisting that ‘Francis’ is no proper name, but the abbreviation of the real name being ‘Francis Bacon’ resp. ‘(Pope) Francis, alias José Bergoglio’ – or, as the Tractatus does, by taking such words not to be names at all. All these manoeuvres are transgressions of grammar, however. For we do indeed call ‘Francis’ a name, even a proper paradigm for a proper name. (ii) The reference-relation is a function only if the thing it maps the word onto cannot dissolve or disappear, as is the case in a mathematical function. As every human being called 27 ‘Francis’ will definitely die one day, human beings are no candidates for ‘things’ according to the relationalist scheme, once it is elaborated coherently. The demand for ‘things’ that cannot perish or dissolve directly leads to the substance ontology of the Tractatus.37 Such ‘things’ (or ‘objects’) must be indivisible. That is why the reference-arrow that points to them is free from any ambiguity. There is only one possible target of pointing because there is nothing with respect to this object that could be distinguished anymore (unlike the case with the banjo): ‘Objects are simple. Objects make up the substance of the world. That is why they cannot be composite.’ (TLP 2.02 and 2.021a)38 So Tractarian conceptions are indeed governed by the idea from (Level 3) that is rooted in the picture, i.e. in the relationalist scheme of (Level 2). Such conceptions conceive of the relation to be (quasi-)mathematical. This fits well to the claim of #38 that this relation stems from the illegitimate analogy between our referring to things and spatial relations. For these are themselves of a mathematical form. My claim that referring is not a relation is meant to say that at least no relation of that kind obtains between words and things they refer to. That and how these conceptions are rooted in (Level 3) can only be seen clearly if one has understood the inner structure of #1 and the pictorial form (Level 2) displays. And it is because of a lack of understanding of both that Goldfarb can bring up the objection (A2) against the ‘classical’ reading. Concluding Remarks: How Confusion Works: Are Augustine’s Words Really ‘Innocent’? My reading of #1 was meant to show that there is a sharp difference between Augustine’s words, the picture, and the idea of meaning, which are both results of an illegitimate abstraction from these words. One might be tempted to think that words from which I can 37 This is clearly seen by Baker/Hacker (1980) and rightly emphasised by von Savigny (1994) as well. 38 I use the Pears/McGuinness translation (= Wittgenstein 2001). 28 abstract such a misleading picture are themselves misleading because of that very same token. But this is too fast a conclusion. Not everything which we can abuse is something bad. To answer the question of whether Augustine’s words are innocent or not we have to reflect on their status. First of all, one should not confuse the question of whether Augustine’s words are innocent in the context of PI #1 with the very different question of whether they are innocent in the context of Augustine’s own (philosophical) writings. Concerning the latter, an exegesis of Augustine was needed. Given that, one may convincingly argue that these words are part of an autobiographical narrative and, hence, not a truth-apt claim about language. Moreover one could try to defend Augustine by mentioning the fact that in De magistro he offers a far more elaborate theory about language.39 That might even show that Augustine, on a kind of meta-level, does himself critically reflect on the scenario described in the passage from the Confessiones. All this I do not aim to dispute here. The question to be tackled (and that Goldfarb, I take it, confronts) is this: Are Augustine’s words innocent in PI #1 or, more generally, out of the context of Augustine’s work? One could then argue that it is not the words’ fault, but ours, if we introduce it to a (foreign) philosophical debate by referring to it. But this is too simple. For words themselves do have a suggestive power and are not some tacit material that lacks any tendency before we interpret them. Imagine that the words quoted were a colourful exploration of the varieties of language-uses, roughly according to the list that Wittgenstein gives in PI #23. Then there would be no temptation to think that the essence of language consists in a relation obtaining between words and objects. But Augustine’s words are strikingly reductive in spirit. This is not to say that all incomplete descriptions of linguistic practices are tempting and, together with our own inclination to misunderstanding, lead into confusion. Remember that Wittgenstein says that 39 Cf. von Savigny 1994, 37 as well as Burnyeat 1987 on that point. 29 Augustine’s words (his description of language) fits very primitive language-games (such as the one in PI #2), and that one could (and should) make this restriction explicit (cf. PI #3). Such restriction would rule out the words providing ‘their half’ of the misunderstanding, so to speak. (Which is not to say that our misunderstanding could not flourish nevertheless, for other reasons.) But what if the description comes without this explicit restriction? As I tried to show in section 3, then there is an ongoing temptation to transform it into the picture. This temptation, of course, would never cause any actual problems if there were no readers that succumb to it. Nevertheless, there is the temptation by and in the words: one might even suspect that this temptation is already at work when one gives this description, for there must be some reason why we describe these primitive language-uses instead of others. So the description cannot be regarded as innocent anymore. Is this meant to say that every incomplete description is tempting by its very being incomplete? That seems to be an exaggeration. Is it not a necessary feature of any investigation to focus on (simple or simplified) aspects of a complex field? Are Augustine’s words not ‘innocent’ just because of this? I hesitate to say yes. Indeed, the description does not say that there are no other language-uses and -games. But by exclusively focusing on the case of ‘naming’, it actually tacitly suggests (intentionally or not) that this is how language works and that there is nothing more to it. This is especially true in a suitable context, which (a) is given by the fact that these words are meant to describe a child’s learning of language as a whole and (b) consists in our (i.e. the readers’) being ‘naturally’ prone to perform the ‘experiment’ of #38. So the words are neither innocent nor do they themselves already contain the picture that, in the light of #38, makes up the relationalist seedbed for referentialism. The words are somewhat in between of innocence and confusion; they put out their hands – and if we shake 30 hands, we end up in confusion. This intricate structure cannot easily be expressed by a single verb. This is why Wittgenstein uses the construction ‘In diesen Worten erhalten wir’. In German, one can also say ‘in these words (in diesen Worten) it is said’ or ‘by these words (mit/durch diese Worte) we get (erhalten)’. But both would fail to capture what we have just said, the first suggesting that it is only the words that are confused (and not we), the second that only we are (whereas the words are a kind of neutral medium). So combining both idioms is Wittgenstein’s way of indicating that, though we ‘erhalten’ the picture and are neither given it nor told, it is nevertheless partly grounded in the words. So his wording again turns out to be very deliberate. The same applies to a remark by which Wittgenstein makes clear that not only words themselves, but also our pictorial interpretation of them that I tried to explicate in my reading of #38 is not simply innocent – because of its tempting force. In TS 222:16 Wittgenstein says that pictures force themselves upon us (‘daß sich Bilder uns aufdrängen’).40 Here he emphasises that a picture, though it is not responsible for our misunderstandings, is nothing inert, but has a tempting force that waits to be en-forced by us. So the answer to the question of whether Augustine’s words are innocent can be given: If one considers the role of these words in the context of PI #1, it is mistaken to think that they are ‘innocent’ without any further qualification, as Goldfarb and Travis suggest. This does not affect the fact that Augustine’s words nevertheless are not (philosophically) mistaken in the sense that applies to the picture on (Level 2) and the idea on (Level 3), nor does it mean that the temptation arising from them is the philosophical confusion yet. For such confusion is ours. It is we that succumb to temptations – and end up in confusions, which we then are responsible for. 40 Cf. also Baker/Hacker 1980, 525. 31 If Wittgenstein is right, these confusions are not harmless, as they make us prejudiced with regard to the question of how meaning works and thus, finally, also to the question of what we, as language-using creatures, are. [Acknowledgment:] I am grateful to Peter Hacker for many deep discussions on Wittgenstein’s philosophy as well as for his comments on an earlier draft of this paper. References Baker, Gordon P., and Peter M. S. Hacker. 1980. Wittgenstein: Understanding and Meaning, Volume 1 of an analytical commentary on the Philosophical Investigations. Oxford: Blackwell / Chicago: Chicago University Press. Baker, Gordon P., and Peter M. S. Hacker. 2009. Wittgenstein: Understanding and Meaning, Volume 1 of an analytical commentary on the Philosophical Investigations. Part II – Exegesis §§1−184. Second Edition, extensively revised by P. M. S. Hacker. Oxford: Wiley-Blackwell. Burnyeat, Myles F. 1987. “Wittgenstein and Augustine De Magistro.” Proceedings of the Aristotelian Society Supplement 61: 1−24. Cavell, Stanley. 1996. “Notes and afterthoughts on the opening of Wittgenstein’s Investigations.” In The Cambridge Companion to Wittgenstein, edited by Hans Sluga and David Stern, 261−295. Cambridge: Cambridge University Press. Glock, Hans-Johann. 1996. A Wittgenstein Dictionary. Malden (MA)/Oxford/Carlton: Blackwell. Goldfarb, Warren D. 1983. “I Want You to Bring Me a Slab: Remarks on the Opening Section of the Philosophical Investigations.” Synthese 56: 265−282. Hanna, Robert. 2010. “From referentialism to human action: the Augustinian theory of language.” In Wittgenstein’s Philosophical Investigations. A Critical Guide, edited by Arif Ahmed, 11−29. Cambridge/New York: Cambridge University Press. Kenny, Anthony. 2006. Wittgenstein. Revised Edition. Malden (MA)/Oxford/Carlton: Blackwell. Lugg, Andrew. 2013. “A Sort of Prologue: Philosophical Investigations §§ 1-7.” Philosophical Investigations 36: 20−36. 32 Luntley, Michael. 2015. Wittgenstein. (MA)/Oxford/Chichester: Wiley-Blackwell. Opening Investigations. Malden McDougall, Derek. 2017. “Reading the Opening of the Philosophical Investigations. The Role of the Shopkeeper, the Builders and Augustine’s Child in a ‘Community View’ of Wittgenstein’s Text.” Wittgenstein-Studien 8: 61−80. McGinn, Marie. 2013. The Routledge Guidebook to Wittgenstein’s Philosophical Investigations. Abingdon/New York: Routledge. Pichler, Alois. 2004. Wittgensteins Philosophische Untersuchungen. Vom Buch zum Album. Amsterdam/New York: Rodopi. von Savigny, Eike. 1994. Wittgensteins „Philosophische Untersuchungen“. Ein Kommentar für Leser. Band I. Abschnitte 1 bis 315. 2., völlig überarbeitete und vermehrte Auflage 1994. Frankfurt am Main: Vittorio Klostermann. Schroeder, Severin. 2006. Wittgenstein. The Way Out of the Fly-Bottle. Cambridge/Malden (MA): Polity Press. Stern, David. 2002. “Nestroy, Augustine, and the Opening of the Philosophical Investigations.” In Wittgenstein und die Zukunft der Philosophie. Eine Neubewertung nach 50 Jahren, edited by Rudolf Haller and Klaus Puhl, 425−445. Wien: öbv & hpt Verlagsgesellschaft. Strawson, Peter F. 1950. “On Referring.” Mind 59: 320−344. Travis, Charles. 2006. Thought’s Footing. Oxford: Oxford University Press. Wittgenstein, Ludwig. 1963. Philosophical Investigations. Trans. Gertrude E. M. Anscombe. Oxford: Basil Blackwell. Wittgenstein, Ludwig. 2001. Tractatus Logico-Philosophicus. Trans. D. F. Pears and B. F. McGuinness. London/New York: Routledge. Wittgenstein, Ludwig. 2009. Philosophical Investigations. The German Text, with an English translation by G. E. M. Anscombe, P. M. S. Hacker and Joachim Schulte. Revised 4th edition by P. M. S. Hacker and Joachim Schulte. Malden (MA)/Oxford/Chichester: Blackwell.